Crojfe 39(1)2018 by zpandur

Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten ... (139–152) M. Kühmaier and G. Erber

Croatian Journal of Forest Engineering • Volume 39 • Issue 1

2018

Croat. j. for. eng. 39(2018)1 153

Contents Journal for Theory and Application of Forestry Engineering

Croatian Journal of Forest Engineering is a refeered journal distributed internationally, publishing scientific articles concerning forest engineer ing, both theoretical and empirical. The journal covers all aspects of forest engineering research, ranging from basic to applied subjects. From volume 1 to 25 the journal was published under the title »Meha nizacija šumarstva«. Publishers Forestry Faculty of Zagreb University, »Croatian forests« Ltd. Zagreb Croatian Chamber of Forestry and Wood Technology Engineers Copublishers FORMEC Publishing Council Mario Božić, Krunoslav Jakupčić, Vladimir Jambreković, Tibor Pentek, Tomislav Poršinsky (all from Croatia) Editorial Board Igor Anić, Damir Barčić, Ivan Balenović, Saša Bogdan, Zdenko Bogović, Jura Čavlović, Andreja Ðuka, Boris Hrašovec, Josip Ištvanić, Anamarija Jazbec, Ante P. B. Krpan, Josip Margaletić, Slavko Matić, Milan Oršanić, Zdravko Pandur, Ivica Papa, Renata Pernar, Stjepan Risović, Marijan Šušnjar, Damir Ugarković, Dinko Vusić, Željko Zečić, Marko Zorić International Editorial Board Dalia Abbas (USA), Mauricio Acuna (Australia), Stelian Alexandru Borz (Romania), Raffaele Cavalli (Italy), Woodam Chung (USA), Mehmet Eker (Turkey), Jörn Erler (Germany), Fulvio di Fulvio (Norway), Stefano Grigolato (Italy), Mohammad Reza Ghaffariyan (Australia), Hans Rudolf Heinimann (Switzerland), Dirk Jaeger (Germany), Martin Kühmaier (Austria), Matevž Mihelič (Slovenia), Tadeusz Moskalik (Poland), Ljupčo Nestorovski (Macedonia), Igor Potočnik (Slovenia), Hideo Sakai (Japan), Raffaele Spinelli (Italy), Karl Stampfer (Austria), Jori Uusitalo (Finland), Rien Visser (New Zeland) Editor’s Office P.O. Box 422, HR–10 002 Zagreb, CROATIA Tel. + 385 (0)1 2352417 Fax. + 385 (0)1 2352517 email: crojfe@sumfak.hr Internet: http://www.crojfe.com Editor-in-Chief Tibor Pentek Editor Željko Tomašić Technical Editor Mario Šporčić Junior Editor Ivica Papa Editorial Advisor Dubravko Horvat Technical Editorial Board Andreja Ðuka, Zdravko Pandur, Dinko Vusić

Croatian Journal of Forest Engineering

•

Volume 39

•

Issue 1, 1–152

•

Zagreb, January 2018

Original scientific papers Jennifer Norihiro, Pierre Ackerman, Ben D. Spong, Dirk Längin Productivity Model for CuttoLength Harvester Operation in South African Eucalyptus Pulpwood Plantations

Kalle Kärhä, Asko Poikela, Teijo Palander Productivity and Costs of Harwarder Systems in Industrial Roundwood Thinnings

Arkadiusz Tomczak, Grzegorz Grodzin’ski, Marcin Jakubowski, Tomasz Jelonek, Witold Grzywin’ ski Effects of ShortTerm Storage Method on Moisture Loss and Weight Change in Beech Timber

Vladimir Petković, Igor Potočnik Planning Forest Road Network in Natural Forest Areas: a Case Study in Northern Bosnia and Herzegovina

Rodolfo Picchio, Farzam Tavankar, Rachele Venanzi, Angela Lo Monaco, Mehrdad Nikooy Study of Forest Road Effect on Tree Community and Stand Structure in Three Italian and Iranian Temperate Forests

Aleksey Ilintsev, Elena Nakvasina, Aleksey Aleynikov, Sergey Tretyakov, Sergey Koptev, Alexander Bogdanov MiddleTerm Changes in Topsoils Properties on Skidding Trails and Cutting Strips after LongGradual Cutting: a Case Study in the Boreal Forest of the NorthEast of Russia

Anna Cudzik, Marek Brennensthul, Włodzimierz Białczyk, Jarosław Czarnecki Tractive Performance of Tyres in Forest Conditions – Impact Assessment of Ground and Tyres Parameters

Marco Manzone, Angela Calvo Trailer Overturning during Wood Transportation: an Experimental Investigation of Effects of Trailer Joint Point and Frame Structure

Aldo Calcante, Davide Facchinetti, Domenico Pessina Analysis of Hazardous Emissions of HandOperated Forestry Machines Fuelled with Standard Mix or Alkylate Gasoline

109

Ivan Balenović, Mateo Gašparović, Anita Simic Milas, Alen Berta, Ante Seletković Accuracy Assessment of Digital Terrain Models of Lowland Pedunculate Oak Forests Derived from Airborne Laser Scanning and Photogrammetry

117

Srečko Glodež, Marko Šori, Krešimir Vučković, Stjepan Risović Determination of Service Life of Sintered Powder Metallurgy Gears in Regard to Tooth Bending Fatigue

129

Subject review Martin Kühmaier, Gernot Erber Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten Years (2007–2016) – Part Two: Comminution, Transport & Logistics

Linguistic Advisers Maja ZajšekVrhovac (for English) Articles are abstracted by or indexed in CAB Abstracts, Compendex, GeoBase, Global Health, Paperchem, Science Citation Index Expanded, SCOPUS, VINITI All published scientific papers have been internationally reviewed Two issues of journal are published annually Circulation: 700 Prepress and Print »Laser plus« Ltd., Brijunska 1a, Zagreb Preparation ended 2018–01–31

Jukka Malinen, Janne Taskinen, Timo Tolppa Productivity of CuttoLength Harvesting by Operators’ Age and Experience

Cover photo Mechanical CTL system enters Croatian Forests Ltd. (Photo: T. Poršinsky) Publishing of this journal is cofinanced by funds from Croatian Ministry of Science and Education Subscription: 80 € per year Subscription payment on behalf of: Forestry Faculty of Zagreb University, P.O. Box 422, HR–10 002 Zagreb, CROATIA Swift Code: ZABA HR 2X, Account Number: 2500–03281485, Details of Payment: 2–02–05 Contact: crojfe@sumfak.hr Subscription: 500 HRK/y (Local Payment) Recipient: Faculty of Forestry, University of Zagreb, p.p. 422, HR–10 002 Zagreb Giro account: 2360000–1101340148, Details of Payment: 2–02–05 Contact: crojfe@sumfak.hr

139

Original scientific paper

Productivity Model for Cut-to-Length Harvester Operation in South African Eucalyptus Pulpwood Plantations Jennifer Norihiro, Pierre Ackerman, Ben D. Spong, Dirk LĂ¤ngin Abstract There has been a concerted shift from traditional motor-manual and semi-mechanised timber harvesting systems to mechanised cut-to length (CTL) operations in South Africa. This is particularly true in Eucalyptus pulpwood felling and processing, South Africaâ&#x20AC;&#x2122;s largest commercial wood resources used in the pulp and paper industry. Mechanisation improvements are typically driven by increasing safety regulations, product quality and productivity concerns related to traditional harvesting systems. The objective of this study is to develop productivity models for mechanised Eucalyptus pulpwood CTL felling and processing operations by combining the results of a number of individual studies done over a period of 24 months in the summer rainfall areas of South Africa. The study takes into account species, machine type (purpose built vs. excavator based), silvicultural practices (planted vs. coppiced) and slope. The pooled data revealed general productivity ranges from 5.16 m3 PMH-1 to 27.49 m3 PMH-1. Keywords: cut-to-length, eucalyptus, pulpwood, full-mechanized system, productivity study

1. Introduction Commercial forestry has experienced a global shift toward mechanised harvesting operations (FAO 1997, Nurminen et al. 2006, JirouĹĄek et al. 2007). This change has also occurred in the South African Forest Industry, with the key drivers being forest worker health and product quality. With this transition, there has been an increase in studies dealing with timber harvesting and transport productivity aimed at determining and modelling equipment productivity. These investigations can provide the means to optimise economic gains and volume yields to managers and contractors (Williams and Ackerman 2016). Although a multitude of research related to mechanised harvesting systems have been conducted internationally, little research has been published in related operations in South Africa. In South Africa, Eucalyptus is the predominant genus used for pulpwood and it accounts for 83% of the commercial wood resources for the pulp and paper industry in South African (FES 2011, FSA 2013). Although Eucalyptus is considered the most commonly

planted (18 million ha in 90 countries) and valued hardwood, there remains a global deficiency of published data on mechanised Eucalyptus harvester operations (FAO 2006). As the South African industry has rapidly transitioned to fully mechanised CTL operations, there has been a need to determine the influencing factors that affect harvester productivity within a South African setting. In a review of scientific and peer reviewed publications, domestic and international, a total of 13 articles were found to be related to fully mechanised harvester-based Eucalyptus operations, but they were inconsistent in recording data in one way or another. Although inconsistent, these studies identified and analysed influencing factors that are vital to understanding harvesting productivity. Factors include tree volume (Spinelli et al. 2010), species composition (Nurminen et al. 2010), equipment type (Siren and Aaltio 2003, Spinelli et al. 2010), site characteristics (Puttock et al. 2005, Andersson 2011), silviculture practices (Kellogg and Bettinger 1994, Ramantswana et al. 2013), operator training (Ovaskainen et al. 2004,

Croat. j. for. eng. 39(2018)1 1

J. Norihiro et al.

Productivity Model for Cut-to-Length Harvester Operation in South African Eucalyptus ... (1â&#x20AC;&#x201C;13)

PurfĂźrst and Erler 2012), delimbing and debarking (Hartsough and Cooper 1999). According to Spinelli et al. (2010), tree volume has been identified as the most significant variable to determine harvester productivity and is a reliable predictor of productivity. Additional studies not only verified this, but suggested that production rate is positively correlated to increasing tree volume (Akay et al. 2004, Eriksson and Lindroos 2014). Other projects used diameter at breast height (DBH) as the continual predictor of productivity, which made it difficult to compare with studies that used tree volume (McEwan et al. 2016, Acuna and Kellogg 2009, Hartsough and Copper 1999). Literature also found operator performance as an influencing factor to harvester productivity, but it has been challenging to quantify because training is not standardised globally (Ovaskainen et al. 2004, PurfĂźrst and Erler 2012). The human factor and work shift were considered by Passicot and Murphy (2013), but operations observed consisted of tree volume exceeding the common South African range to be applicable. In addition, productivity was often recorded as m3 PMH-1, but in Hartsough and Nakamura (1990) and Acuna and Kellogg (2009), productivity was recorded as bone dry tonne per scheduled hour (BDT/SH) or tonnes PMH-1 with no information on the machine used. Terrain, more specifically slope, was identified in some of the studies and proven to have a considerable effect on productivity (Davis and Reisinger 1990, Spinelli et al. 2002, Acuna and Kellogg 2009). In Acuna and Kellogg (2009), slope, ranging from gentle to moderate slope, was identified as a significant factor, but productivity was recorded inconsistently when compared to other literature. Despite a few factors within each published paper applicable to a South African context, most were inconsistently recorded and could not be used as a predictor of productivity trends. As a means to address the limited literature, the individual studies performed in South Africa were combined in an attempt to develop general productivity models. The objective of this study is to develop general productivity models for mechanised Eucalyptus pulpwood CTL harvesting (felling and processing) operations by combining the results of five individual and independent productivity studies completed over a period of 24 months in Eucalyptus clearfelling pulpwood stands in the summer rainfall area of South Africa. This study will take into account species, silvicultural practices (planted vs. coppiced), machine type (purpose built vs. excavator based) and slope inherent in the five studies.

2. Material and methods 2.1 Case studies Five individual productivity study sites located in the north-east of South Africa were included in this study. The sites have been sequentially numbered and referred to by this numbering throughout this paper (Fig. 1). These studies covered four different species of Eucalyptus and were all clear-felling pulpwood compartments that were harvested during the dry winter months. Only two components of the harvester operation were considered: felling and processing. The four species harvested included: Eucalyptus grandis x camaldulensis (G x C), Eucalyptus grandis x urophylla (G x U), Eucalyptus smitthii (ES) and Eucalyptus dunnii (ED). Further on in this study, species will be referred to by their acronym. Harvesting sites covered a diverse range of terrain (slope), tree characteristics (species, form, individual tree volume) and harvester machine type (excavator based and purpose built) in order to incorporate site conditions and factors that contribute to productivity trends (Table 1). Even though the five individual studies had varying original objectives, the data was collected using a standardised time-study protocol (Ackerman et al. 2014) that enables comparisons between the studies. The objective of Study 1 was to determine productivity differences between one and three pass debarking and debranching operation in a G x C clones on even terrain. The objective of Study 2 was to determine productivity differences between excavator based and purpose built machines on varying slope terrain in a G x C clone. The objective of Study 3 was to determine productivity differences between three and five pass debranching and debarking in a G x U clone on even terrain. The objective of Study 4 was a pure productivity study of an excavator based harvesting

Fig. 1 Locations of study areas Croat. j. for. eng. 39(2018)1

Productivity Model for Cut-to-Length Harvester Operation in South African Eucalyptus ... (1–13)

J. Norihiro et al.

Table 1 Individual site and stand characteristics of the five studies Site characteristics

Species

Study 1 Eucalyptus grandis x camaldulensis (G x C)

Study 2 Study 21

Study 22

Eucalyptus grandis x camaldulensis (G x C)

Study 3

Study 4

Study 5

Eucalyptus grandis x urophylla (G x U)

Eucalyptus smithii (ES)

Eucalyptus dunii (ED)

DBH, cm Average

15.5

15.3

16.3

21.6

15.9

16.4

Min.

7.0

9.0

7.3

8.6

5.2

8.0

Max.

21.2

27.2

25.3

29.1

35.7

30.5

2.3

2.7

2.9

3.8

4.6

5.0

Age, y

SPH, n ha-1

987

1001

926

1087

1106

826

Average height, m

16.3

19.88

20.03

25.4

17.4

18.5

Average tree volume m3 tree–1

0.12

0.15

0.38

0.14

0.15

Slopea, % (continuous variable)

Level (0–10)

Level – very steep (0–61)

Level (0–10)

Silvilculture

Planted

Planted-Coppice

Planted

Coppice

Carrier type

Excavator

Purpose Built

Excavator

Hitachi Zaxis 200

Timberpro TL-725B

Volvo EC-210bf

Hitachi Zaxis 200

Komatsu PC 200

Waratah H616

Maskiner SP 591-LX

Waratah H616

Maskiner SP 591-LX

Zululand

Melmoth

Kwambo

KZN Midlands

Piet Retief

297

1156

1099

181

1478

177

Machine manufacturer Head Location Sample size a

Slopes are classified using the National Terrain Classification for Forestry (Erasmus 1994)

machine, felling and processing poor form ES on even terrain. The objective of Study 5 was to determine productivity differences between three and five pass debarking and debranching passes operation in ED on even terrain. Debarking and debranching passes are defined as the number of times the harvester head travels along the tree stem debarking and debranching. The last pass will entail cross-cutting in log assortments.

2.2 Time study Different researchers collected time study data at each of the study areas according to the South African Forest Industry Time-study Standard (Ackerman et al. 2014). Field time study observations were recorded using a Trimble GeoXT handheld computer. Time recorded was categorised into one of four elements idenCroat. j. for. eng. 39(2018)1

tified in the standard: fell, process, move and delay (Table 2). All machine operators, although not the same in all studies, were considered trained and capable of operating the harvester in Eucalyptus pulpwood operations consisting of felling, debarking, debranching and crosscutting into assortments. Delay times were recorded regardless of duration. Producti vity results were expressed in productive machine hours (PMH). Individual tree volume (m3) was calculated using the Schumacher and Hall model (Bredenkamp 2012). Individual tree and compartment attributes recorded are reflected in Table 1. In this study, slope is considered as a continuous variable. Continuous slope data were obtained from Digital Terrain Models (DTMs). These models were derived from large-footprint LiDAR data with approximate 1 m resolution.

J. Norihiro et al.

Productivity Model for Cut-to-Length Harvester Operation in South African Eucalyptus ... (1–13)

Table 2 Time study elements breakdown (Ackerman et al. 2014) Time element

Description

Fell

Starts when the operator begins moving the head to a tree, ends when the butt end begins to move through the head

Process

Starts when the butt end begins to move through the head, ends when the head has released the last piece of the tree

Move

Starts when the tracks begin moving, ends when the tracks come to a stop

Delay

Starts when the machine unexpectedly stops working, ends when work begins again

2.3 Experimental design At each of the five study sites, diameter at breast height, measured over bark (DBH), was recorded for every tree using a diameter tape with an accuracy of 0.1 cm. While measuring DBH, each tree was allocated a unique number per study area in order to identify each tree when recording cycle times during the actual harvesting of the samples. Heights of at least 50 representative trees per site, chosen from various locations in the allocated compartment and spanning across the range of DBH available, were measured using a Haglof Vertex laser hypsometer with an accuracy of 0.1 m. The heights and DBH of these representative trees were used to derive a regression, which allowed the heights of the remaining, not measured trees, to be estimated based on the DBH measured for each tree. Every tree was numbered to facilitate the pairing of tree dimensions with felling and processing times to calculate productivity (m3 PMH-1). Numbers were painted on tree stems at an angle to ensure visibility during timing. Prior to harvesting, a randomised block experimental design (RBD) (Clewer and Scarisbrick 2001) was applied to each study area to reduce bias.

2.4 Statistical analysis Basic statistics, correlation analysis and linear regression modelling were performed to determine and clarify variables affecting harvester productivity. Tree volume was used as the continuous predictor for regression models with additional correlation analyses applied to identify the significance of variables, such as species, carrier type, silviculture, slope, and debarking pass on productivity. Where significant factors were identified, additional models were developed. As a secondary analysis, multiple regression analysis was conducted to better fit the dataset. The pooled dataset was categorised according to potential influ-

encing factors, notably species and carrier type, to determine if these factors were significant to harvester productivity, while using tree volume as the predictive variable. To compensate categorical influencing factors with more than two categories, such as species and carrier type, data was grouped and analysed regarding their respective categories. Multiple regression analysis was conducted as a means to capture residuals, and more accurately represent productivity. After each multiple linear regression productivity model was developed, an analysis of covariance (ANCOVA) was conducted in order to verify potential significant differences between the individual linear regression models that make up each of the full multiple linear regression. If the results of the ANCOVA show that the individual linear regressions are non-parallel, then the ANCOVA is rejected and the multiple linear regression model is significant. However, if the test cannot reject that the individual linear regressions are parallel, then significance of the full multiple linear regression is not established. Further testing of intercept equality is conducted in order to establish that the models are not the same. If equal intercept cannot be rejected, the multiple linear regression model developed is not significantly different and a single linear regression model can adequately fit the dataset. However, if intercept equality is rejected, the multiple linear regression productivity model is a better fit for the dataset. All analysis and models were conducted and developed through Excel and STATISTICA 13 (StatSoft, Tulsa, OK, USA).

3. Results All five individual datasets were pooled to produce a mean productivity figure of 14.5 m3 PMH-1 (Table 3). Literature and correlation analysis identified tree volume as the most significant contributor to harvester productivity (p<0.001). The pooled harvester productivity was plotted against tree volume and analysed to develop a single linear regression model. The result of the single regression equation was positively correlated with the dataset (r2=0.64, p<0.001), where the regression equation is y=4.536+63.801x (where x = tree volume) (Fig. 2 and Table 3). The average productivity for each of the individual studies varied between 13.80 and 27.49 m3 PMH-1. Regression models were also developed for each of the different studies (Table 3). The productivity models were developed with »x« equal to tree volume. Croat. j. for. eng. 39(2018)1

Productivity Model for Cut-to-Length Harvester Operation in South African Eucalyptus ... (1–13)

J. Norihiro et al.

ter explain variation of the pooled dataset. All multiple linear regression equations, when significant, were developed considering species, carrier type (excavator based verse purpose-built), silviculture (planted or coppice), slope, debarking and debranching harvester head passes and tree volume. In this analysis, slope and tree volume are continuous variables, while silviculture, harvested head passes, carrier type and species are categorical.

3.2 Species

Fig. 2 Single linear regression model of pooled productivity Table 3 Mean productivity per study Study

Mean productivity m3 PMH–1

Equation

Significance

Overall

14.47 (0.35–69.22)

y=4.536+63.801x

0.64

***

Study 1

17.93 (2.92–43.78) y=5.800+102.784x 0.45

***

Study 2

14.45 (1.90–44.32)

y=4.754+63.611x

0.61

***

Study 3

23.61 (2.46–58.57)

y=3.283+53.041x

0.79

***

Study 4

27.49 (0.35–59.24)

y=1.073+82.817x

0.76

***

Study 5

13.80 (1.56–69.22)

y=1.085+84.778x

0.75

***

x = tree volume, m3; *** refers to significance at p<0.001

Productivity equations were developed by categorising data by species. Along with species, equations of carrier type, silviculture, slope, debarking pass and tree volume were considered. Multiple linear regression models were developed for each species. Models for Eucalyptus smitthii (ES) and Eucalyptus dunnii (ED) were not significant from each other after an ANCOVA test (p=0.48). As the individual models for ES and ED were not significant, both species data were pooled to develop a new combined model (ES+ED). The overall and three species based models, ES+ED; G x C; G x U (Table 4), show a positive relationship with increasing tree volume. Each productivity model was developed with respect to influencing factors. For instance, the influencing factors to ES+ED productivity were silviculture, pass and tree volume, while, G x C productivity was influenced by carrier type, silviculture, slope, pass and tree volume. G x U productivity was only influenced by pass and tree volume. As multiple variables were used to develop these models, predicted values versus observed values were plotted (Fig. 3). Each of the productivity models represent the pooled dataset with r2 greater than 0.60.

3.3 Species and harvester type

3.1 Multiple linear regression Along with single linear regression models, multiple linear regression models were developed to bet-

As suggested by Sirén and Aaltio (2003) and Spinelli et al. (2010), machine differences may have an effect on productivity. Therefore, the pooled dataset was reanalysed and new productivity equations were developed

Table 4 Regression equation by species Equation

Significance

Overall

y=23.684+(0.497)*x1+ (–0.734)*x2+(0.027)*x3+ (–3.963)*x4+(64.430)*x5

0.68

***

ES+ED

y=0.847+(1.189)*x2+(83.087)*x5

0.76

***

GxC

y=21.246+(0.174)*x1+ (–1.906)*x2+(–0.052)*x3+ (–2.633)*x4+(65.652)*x5

0.60

***

GxU

y=3.283+(53.041)*x5

0.78

***

Species

x1 = model type (purpose-built = 1 or excavator = 2); x2 = silviculture (planted = 1 or coppice = 2); x3 = slope (percent); x4 = number of processing passes; x5 = tree volume (m3); *** refers to significance at p<0.001

Croat. j. for. eng. 39(2018)1

J. Norihiro et al.

Productivity Model for Cut-to-Length Harvester Operation in South African Eucalyptus ... (1–13)

Fig. 3 Productivity regression models per species, including predictive values versus observed values Table 5 Regression equation based on harvester machine make per species Machine make

Species

Equation

Significance

Hitachi

ES+ED

y=4.368+(63.286)*x5

0.65

***

Komatsu

ES+ED

y=1.052+(83.114)*x5

0.76

***

TimberPro

GxC

y=10.559+(–2.300)*x2+(–0.094)*x3+(62.286)*x5

0.56

***

Volvo

GxC

y=4.979+(–1.455)*x2+(0.003)*x3+(73.665)*x5

0.64

***

Hitachi

GxC

y=22.427+(–3.196)*x4+(52.717)*x5

0.62

***

Hitachi

GxU

y=20.197+(–2.064)*x4+(40.857)*x5

0.56

***

x1 = Silviculture (planted = 1 or coppice = 2); x3 = Slope (percent); x4 = Number of Processing Passes; x5 = Tree volume (m3); *** refers to significance at p<0.001

Croat. j. for. eng. 39(2018)1

Productivity Model for Cut-to-Length Harvester Operation in South African Eucalyptus ... (1â&#x20AC;&#x201C;13)

J. Norihiro et al.

Fig. 4 Productivity regression models per species and harvester manufacturer and model, including predictive values versus observed values Croat. j. for. eng. 39(2018)1

J. Norihiro et al.

Productivity Model for Cut-to-Length Harvester Operation in South African Eucalyptus ... (1â&#x20AC;&#x201C;13)

with both species and harvester manufacturer as categorical variables. The TimberPro harvester, used at one site, was the only purpose built machine. All the other sites were harvested using excavator based harvesters. Silviculture, slope, debarking passes and tree volume were each tested for significance and included in the appropriate productivity models. Again, each of the multiple linear regression models was positively correlated with increasing tree volume (Table 5). The Hitachi and Komatsu ES+ED productivity was only influenced by tree volume. The TimberPro G x C and Volvo G x C productivity was also influenced by tree volume, but also by silviculture and slope. In the G x C and G x U stands with the Hitachi machine, the productivity was only influenced by pass and tree volume. As previously completed for the species based models, predicted values verses observed values graphs were plotted to demonstrate the accuracy of developed models by plotting the model over the recorded productivity of each carrier make and species (Fig. 4).

4. Discussion When comparing five original studies using multiple linear regressions, the highest productivity was observed in Study 1, while the lowest productivity was recorded in Study 3. Data collected in Study 4 and Study 5 had the second highest productivity when

stem size exceeded 0.19 m3, regardless of poor tree form. However, as tree volume decreased below 0.19Â m3, productivity recorded in Study 2 and Study 3 exceeded the values of Study 4 and Study 5. In Study 2, steep and varying slope may be responsible for the high recorded processing time (Fig. 5) and, hence, lower productivity similar to Acuna and Kellogg (2009). Study 3 had the second highest mean productivity as a result of larger and higher volume trees. While productivity would be expected to be even higher on this site based on most published literature, considerable additional time was required for processing each tree, lowering overall productivity similar to the results found in Nakagawa et al. (2007, 2010).

4.1 General productivity models In previous studies, tree volume was identified as a significant predictor of harvester productivity and, as a result, regression equations were developed based on tree volume (SirĂŠn and Aaltio 2003, Nurminen et al. 2010, Acuna and Kellogg 2009, Strandgard et al. 2013, Standgard et al. 2016). In order to compare the pooled dataset to the literature data, a single linear regression model was developed based on 21 previously published papers. In order to do this, the mean productivity values and the mean tree volume in each publication were plotted and a new single linear regression model was developed. The literature based model was then overlapped with the single linear regression model developed from the pooled dataset (Table 6). Unfortunately, due to the small sample size from literature data, the comparison was limited. Specifically, in this comparison, all productivity data in the combined dataset and the literature models associated with tree volumes greater than 0.5 m3 were removed from the analysis. This process allowed the dataset to stay within an appropriate harvested tree volume range. A typical 10 year old harvested G x C grown on a high site index South African plantation, would have a volume of 0.23 m3 (Kotze et al. 2012), with few ever exceeding this 0.5 m3 limit. Table 6 Regression model equation of literature based data against dataset Regression model

Fig. 5 Individual time consumption per work element per study in centi-minutes

Significance

Current study y=4.0582+67.3274x 0.624

***

4388

Literature

***

y=2.4658+52.6189x 0.623

x = tree volume in m3

Croat. j. for. eng. 39(2018)1

Productivity Model for Cut-to-Length Harvester Operation in South African Eucalyptus ... (1–13)

J. Norihiro et al.

As each species-specific model has its own influencing factors, it is difficult to compare the models. For instance, carrier type and slope appear only in the Overall and G x C models, whereas silviculture, number of passes and tree volume appear in all of the models. Overall productivity estimates can be calculated with the basic data on influencing factors. These estimates are an important component in the management of logging crews and the extended forest products supply chain. 4.2.2 Carrier type

Fig. 6 Combined dataset (CS), published literature (LT) models and data points in respect to tree volume and productivity Fig. 6 plots the literature and combined dataset models in respect to tree volume and productivity. Additionally, all individual data points are plotted to illustrate the spread of data around the models. Overall, both models show clear productivity increases with increasing tree volumes. The mean productivity recorded is 14.47 m3 PMH-1, whereas the productivity recorded for the literature model is 9.91 m3 PMH-1. When compared to the literature through the least squared method, the mean productivity captured by the combined study data was significantly more productive (p<0.001). Although the ANCOVA was ultimately rejected after testing intercept equality, it could not reject that the models may be parallel. (p=0.28). This may imply that the models have similarities, even though productivity is significantly different, or it could be potentially attributed to systematic error related to the removal of data to limit the effect of the large tree sizes in literature models.

4.2 Other influencing factors 4.2.1 Species Similar to Nurminen et al. (2010), this study identified species having a significant effect on productivity (p<0.001). The G x C, ED and ES productivity models (Table 4) have a relatively higher spread of productivity values of less than 30 m3 PMH-1, while the G x U productivity model has a more consistent and regular spread of data with values of less than 45 m3 PMH-1. Croat. j. for. eng. 39(2018)1

In the literature, machine and equipment selection has been considered to make a significant difference on harvester productivity (Sirén and Aaltio 2003, Spinelli et al. 2010). One of the reported potential differences is the influence of harvester head models (Laitila and Väätäinen 2013). This relationship was not confirmed by the current study; it was only able to establish significance for the specific harvester manufacturer and model when tested with a correlation analysis. Furthermore, no published literature was found on productivity based on machine selection between excavator based machines verses purpose-built machines, especially in relation to Eucalyptus CTL harvesting operations. This study compared the two carrier types and confirmed purpose-built machines as being more productive for most tree volumes, but as tree volume decreased so did the margin of significance. Although less common in South Africa because of the high initial investment cost, purpose-built machines specialise in tree felling and processing, which keeps their production rate stable and less affected than excavator based machines by factors such as terrain changes (Martin 2016). 4.2.3 Slope Ground slope of the sites in this study ranged from flat to over 60%. In all studies except Study 2, slope was classified as per Erasmus (1994) as level (0–10%) and, after analysis, it was found to be insignificant to production rate (p=0.07). In contrast, Study 2 had varying slopes ranging from level to very steep. The literature suggests that regardless of tree volume, a steeper slope leads to a decrease in harvester productivity (Spinelli 2002, Acuna and Kellogg 2009, Magagnotti et al. 2011, McEwan et al. 2016). The influence of slope, as stated in the literature, was only significant in Study 2, where there were more data on steeper terrain used in the analysis. At the same time, the less steep terrain had very little influence on productivity in the full tree volume range.

J. Norihiro et al.

Productivity Model for Cut-to-Length Harvester Operation in South African Eucalyptus ... (1–13)

4.2.4 Passes for debarking and delimbing As Eucalyptus trees are typically debarked and delimbed at the stump in CTL operations, these activities are considered in the development of productivity models. Debarking effort is related to the strength of the bark/wood bond; the stronger the bark/ wood adhesion, the greater the impact on debarking productivity (Hartsough and Cooper 1999, van de Merwe 2014). The literature has suggested that climatic conditions can significantly affect the barkwood bond of logs due to varying moisture content and, therefore, the productivity rate of immediate in-field debarking (Öman 2000, Araki 2002, Nuutinene et al. 2010, van de Merwe 2014). Two studies did not have the number of passes included in their models. In Study 2, the main focus of the project was to investigate carrier type interactions with productivity on variable terrain, so little to no data was collected on the number of passes required for debarking and delimbing. Likewise, the focus of Study 4 had limited interest in the number of passes and these data fell out of the model as insignificant (p>0.05). 4.2.5 Independent literature models As previously stated, many studies have shown tree volume to be the most constant variable to determine harvester productivity (Spinelli et al. 2002, Ovaskainen et al. 2004, Jiroušek et al. 2007, Nakagawa et al. 2007, Spinelli et al. 2010, McEwan 2012, Picchio et al. 2012, Seixas and Batista 2012). The strong correlation

between tree volume and productivity is confirmed by the analysis in this study, where tree volume was identified as the most significant predictor of harvester productivity (p<0.001). In the general productivity models discussed in the first part of this section, the literature based model was generated using volume and productivity data points from multiple papers to develop a linear regression model. Three additional published studies fully developed productivity models that allow a further comparison with the combined dataset model. All four of these models are plotted in Fig. 7. The Spinelli et al. (2002) and Strandgard et al. (2016) models focused on developing harvesting productivity models for Eucalyptus with regard to southern Europe and Australia, respectively. Ramantswana et al. (2013) considered harvester productivity effects on differently managed silviculture (coppice verse planted) Eucalyptus plantations. Despite different primary objectives, the models were all based on tree volume as the continuous predictor and thus they were comparable with the combined dataset model. When models were compared, the productivity model developed with the dataset model fits into the existing range and follows the common trend based on literature models (Spinelli et al. 2002, Ramantswana et al. 2013, Strandgard et al. 2016). These regression models not only reveal, but validate the increase in productivity of the harvester as tree volume increases, regardless of the consideration of additional variables (i.e. terrain, silviculture, carrier type). These equations are the start of a potential productivity equation to help local stakeholders and contractors to determine productivity and cost models for future South African operations.

4.3 Limitations

Fig. 7 Harvester productivity (m3 PMH-1) for three independent literature models and the combined dataset model

The main limitations of this study are as follows: Þ as this study consists of a combination of discreet datasets with diverse objectives and variables, not necessarily recorded in all studies, analyses and comparisons were complicated Þ although considered trained, different operators were used over the two-year data collection period of this study. Operator’s efficiency was excluded from analysis Þ weather conditions for each of the studies were not included in this combined dataset. The productivity of different tasks, like debarking, can vary between wet and dry weather, so while these data were assumed to be collected during normal dry conditions, actual daily weather could result in productivity differences. Weather effects were not included in this analysis. Croat. j. for. eng. 39(2018)1

Productivity Model for Cut-to-Length Harvester Operation in South African Eucalyptus ... (1–13)

J. Norihiro et al.

5. Conclusions

6. References

This study developed general productivity models, specific for South Africa, for mechanised Eucalyptus pulpwood CTL harvesting (felling and processing) operations through the combination of the results of five individual studies. The models considered species, silvicultural practices (planted vs. coppiced), carrier type (purpose built vs. excavator based machines), number of passes for debarking and delimbing and slope. When studies were combined, the overall mean productivity from the dataset was 14.47 m3 PMH-1 with a range between 0.35 m3 PMH-1 and 69.22 m3 PMH-1. Through a correlation analysis, tree volume was found to be the most significant predictor of overall productivity, confirming the published results. Based on this result, a single linear regression model was developed with respect to the individual tree volume. To further strengthen the models, the additional influence of species, silvicultural practices, carrier type, number of passes for debarking and delimbing and slope were incorporated into a general productivity model through multiple linear regression analysis. The dataset was then categorised by species, showing that there were productivity differences for each species groups. As each species group used different contributing factors, it was impossible to make significant comparisons between the groups. A new model based on existing data points from published literature and three other published complete productivity models were also compared with the models developed in this study. Similarities between the models confirmed that harvester productivity increases as tree volume increases, regardless of the consideration of additional variables (i.e. slope or silviculture). As the first step in refining a locally relevant productivity model for mechanised CTL systems, these results can help stakeholders and contractors to determine productivity and costs for future operations. This work by no means addresses all aspects of Eucalyptus pulpwood clearfelling productivity, but continued efforts in this field and broadening the database with more and diverse data, will lead to a robust South African specific productivity model.

Ackerman, P., Gleasure, E., Ackerman, S., Shuttleworth, B., 2014: Standards for Time Studies for the South African Forest Industry (Accessed 17 March 2015). Available at: http:// www.forestproductivity.co.za/?page_id=678.

Acknowledgements The authors acknowledge Mondi for providing study areas. We recognise John Rabie and Chad Martin, MSc students at Stellenbosch University, for providing us with data. In addition, thanks to John Eggers and Nonkululeko Ntinga from Mondi for providing support, as well as, Professor Daan Nel at Stellenbosch University for assistance with data analysis.

Acuna, M. A., Kellogg, L. D., 2009: Evaluation of alternative cut-to-length harvesting technology for native forest thinning in Australia. International Journal of Forest Engineering 20(2): 17–25. Akay, A., Erdas, O., Session, J., 2004: Determining productivity of mechanized harvesting machines. Journal of Applied Sciences 4(1): 100–105. Andersson, R., 2011: Productivity of integrated harvesting of pulpwood and energy wood in first commercial thinnings. Maters thesis, Swedish University of Agriculture Sciences Department of Forest Resources Management. Umeå, Sweden. Araki, D., 2002: Fibre recovery and chip quality from debarking and chipping fire-damaged stems. Report, Forest Engineering Research Institute of Canada. Bredenkamp, B. V., 2012: The volume and mass of logs and standing trees. In: Bredenkamp, B., Upfold, S. (eds.), South African Forestry Handbook (5th edn.). Pretoria: Southern African Institute for Forestry: 239–267. Clewer, A. G., Scarisbrick, D. H., 2001: Practical statistics and experimental design for plant and crop science. West Sussex, England: John Wiley and Sons, Ltd. Davis, C. J., Reisinger, T. W., 1990: Evaluating terrain for harvesting equipment selection. Journal of Forest Engineering 2(1): 9–16. Erasmus, D., 1994: National Terrain Classification System for Forestry: Version 1.0. Institute for Commercial Forestry Research. ICFR Bulletin 11/94. Pietermaritzburg, South Africa. Eriksson, M., Lindroos, O., 2014: Productivity of harvesters and forwarders in CTL operations in northern Sweden based on large follow-up datasets. International Journal of Forest Engineering 25(3): 179–200. FAO, 1997: State of the World’s Forests, Rome, Italy: Food and Agriculture Organization of the United Nations. FAO, 2006: Global Forest Resources Assessment 2005: Progress toward sustainable forest management. Forestry Paper No. 147. Rome, Italy: Food and Agriculture Organization of the United Nations. Forestry Economics Services CC, 2011: Report of commercial timber resources and primary roundwood processing South Africa. Department of Agriculture, Forestry and Fisheries, Pretoria. Forestry South Africa, 2013: Abstract of South African forestry facts for the year 2010/2011. Department of Agriculture, Forestry and Fisheries. Johannesburg.

Croat. j. for. eng. 39(2018)1 11

J. Norihiro et al.

Productivity Model for Cut-to-Length Harvester Operation in South African Eucalyptus ... (1–13)

Hartsough, B. R., Cooper, D. J., 1999: Cut-to-length harvesting of short-rotation Eucalyptus. Forest Products Journal 49(10): 69–75. Hartsough, B. R., Nakamura, G., 1990: Harvesting Eucalyptus for fuel chips. California Agriculture 44(1): 7–8. Jiroušek, R., Klvač, R., Skoupý, A., 2007: Productivity and costs of the mechanised cut-to-length wood harvesting system in clear-felling operations. Journal of Forest Science 53(10): 476–482. Kellogg, L., Bettinger, P., 1994: Thinning productivity and cost for a mechanized cut-to-length system in the northwest Pacific coast region of the USA. Journal of Forest Engineering 5(2): 43–54. Kotze, H., Kassier, H. W., Fletcher, Y., Morley, T., 2012: Growth modelling and yield tables. In: Bredenkamp, B. V. and Upfold, S. J. (eds.), South African Forestry Handbook. The South African Institute of Forestry. 5th Edition. Colour Planet. Pinetown. Laitila, J., Vaatainen, K., 2013: The cutting productivity of the excavator based harvester in integrated harvesting of pulpwood and energy wood. Baltic Forestry 19(2): 289–299. Magagnotti, N., Nati, C., Pari, L., Spinelli, R., Visser, R., 2011: Assessing the cost of stump-site debarking in eucalypt plantations. Biosystems engineering 110(4): 443–449. Martin, C., 2016: Assessing the effect of slope on costs and productivity of single-grip purpose-built and excavator based harvesters. MSc. Thesis. Stellenbosch University, South Africa. McEwan, A., 2011: The effect of tree and bundle size on the productivity and costs of Cut-To-Length and multi-stem harvesting systems in Eucalyptus pulpwood. MSc. Thesis. University of Pretoria, South Africa. McEwan, A., Magagnotti, N., Spinelli, R., 2016: The effects of number of stems per stool on cutting productivity in coppice Eucalyptus plantations. Silva Fennica 50(2): 1–14. Nakagawa, M., Hamatsu, J., Saitou, T., Ishida, H., 2007: Effect of tree size on productivity and time required for work elements in selective thinning by a harvester. International Journal Forest Engineering (18): 24–28. Nakagawa, M., Hayashi, N., Narushima, T., 2010: Effect of tree size on time of each work element and processing productivity using an excavator-based single-grip harvester or processor at a landing. Journal for Forest Research 15(4): 226–233. Nurminen, T., Korpunen, H., Uusitalo, J., 2006: Time consumption analysis of the mechanized cut-to-length harvesting system. Silva Fennica 40(2): 335–363. Nuutinen, Y., Väätäinen, K., Asikainen, A., Prinz, R., Heinonen, J., 2010: Operational efficiency and damage to sawlogs by feed rollers of the harvester head. Silva Fennica 44(1): 121–139. Öman M., 2000: Influence of log characteristics on drum debarking of pulpwood. Scandinavian Journal of Forest Research 15(4): 455–463.

Ovaskainen, H., Uusitalo, J., Väätäinen, K., 2004: Characteristics and significance of a harvester operators’ working technique in thinning. International Journal of Forest Engineering 15(2): 67–77. Passicot, P., Murphy, G. E. 2013: Effect of work schedule design on productivity of mechanized harvesting operations in Chile. New Zealand Journal of Forestry Science 43(2): 10 p. Picchio, R., Sirna, A., Sperandio, G., Spina, R., Verani, S., 2012: Mechanized harvesting of Eucalypt Coppice for Biomass Production Using High Mechanization Level. Croatian Journal of Forest Engineering 33(1): 15–24. Puttock, D., Spinelli, R., Hartsough, B. R., 2005: Operational trials of cut-to-length harvesting of poplar in a mixed wood stand. International Journal of Forestry Engineering 16(1): 39–49. Purfürst, F. T., Erler, J., 2012: The Human Influence on Productivity in Harvester Operations. International Journal of Forest of Forest Engineering 22(2): 15–22. Rabie, J., 2014: Analysis of a mechanised cut-to-length harvesting operation working in a poor growth Eucalyptus smithii stand through use of discrete-event simulation in R. MSc. thesis, Stellenbosch University, South Africa. Ramantswana, M., McEwan, A., Steenkamp, J., 2013: A comparison between excavator-based harvester productivity in coppiced and planted Eucalyptus grandis compartments in KwaZulu-Natal, South Africa. Southern Forests: a Journal of Forest Science 75(4): 239–246. Seixas, F., Batista, J. L. F., 2012: Use of Wheeled harvester and excavators in Eucalyptus harvesting in Brazil: In: Proceedings of the 35th Council on Forest Engineering Annual Meeting: Engineering New Solutions for Energy Supply Demand. New Bern, North Carolina, 7 p. Sirén, M., Aaltio, H., 2003: Productivity and Costs of Thinning Harvesters and Harvester-Forwarders. International Journal of Forest Engineering 14(1): 39–48. Spinelli, R., Owende, P. M. O., Ward, S., 2002: Productivity and cost of CTL harvesting of Eucalyptus globulus stands using excavator-based harvesters. Forest Product Journal 52(1): 67–77. Spinelli, R., Hartsough, B., Magagnotti, N., 2010: Productivity Standards for Harvesters and Processors in Italy. Forest Products Journal 60(3): 226–235. StatSoft, 2012: Statistica 13. Tulsa, OK, United States of America. Strandgard, M., Walsh, D., Acuna, M., 2013: Estimating harvester productivity in radiata pine (Pinus radiata) plantations using StanForD stem files. Scandinavian Journal of Forest Research 28(1): 73–80. Strandgard, M., Mitchell, R., Walsh, D., 2013: Productivity and cost of two Eucalyptus nitens harvesting systems when bark is retained on logs. Australian Forests Operations Research Alliance (AFORA): Industry Bulletin 5. Croat. j. for. eng. 39(2018)1

Productivity Model for Cut-to-Length Harvester Operation in South African Eucalyptus ... (1–13)

J. Norihiro et al.

Strandgard, M., Walsh, D., Mitchell, R., 2015: Productivity and cost of whole-tree harvesting without debarking in Eucalyptus nitens plantation in Tasmania, Australia. Southern Forests 77(3): 173–178.

Van der Merwe, J., 2014: The impact of mechanical log surface damage on fibre loss and chip quality when processing Eucalyptus pulpwood using a single-grip harvester. MSc. Thesis, Stellenbosch University, South Africa.

Strandgard, M., Mitchell, R., Acuna, M., 2016: General productivity model for single grip harvesters in Australian eucalyptus plantations. Australian Forestry 79(2): 108–113.

Williams, C., Ackerman, P., 2016: Cost-productivity analysis of South African pine sawtimber mechanised cut-to-length harvesting. Southern Forests 78(4): 267–274.

Authors’ addresses: Jennifer Norihiro e-mail: jennifernorihiro@gmail.com Pierre Ackerman, PhD.* e-mail: packer@sun.ac.za University of Stellenbosch Department of Forest and Wood Science Private Bag X1 7602 Matieland SOUTH AFRICA Dirk Laengin e-mail: Dirk.Laengin@mondigroup.co.za Mondi 380 Old Howick Road 3245 Hilton SOUTH AFRICA

Received: December 12, 2016 Accepted: May 1, 2017 Croat. j. for. eng. 39(2018)1

Ben D. Spong e-mail: ben.spong@mail.wvu.edu West Virginia University Division of Forestry and Natural Resources PO Box 6125 WV 26506 Morgantown USA * Corresponding author

Original scientific paper

Productivity of Cut-to-Length Harvesting by Operators’ Age and Experience Jukka Malinen, Janne Taskinen, Timo Tolppa Abstract In the study, the relationship between operators’ age, experience and mechanized cut-to-length (CTL) harvesting productivity was examined. The data were five-year follow-up data from 28 operators and 38 CTL harvesters collected from southern Finland. Productivities were converted to relative productivities and average productivity models were created. Case specific productivities were compared to modelled values, and productivity ratio models including separate lower and upper quartile models were produced. The relative productivity of operators at the age of 45 years in clear cuttings was 17.8% higher and in thinnings 14.9% higher than that of operators at the age of 25 years. The relative lower quartile productivity increased from operators aged 25 to operators aged 45 years by 38.6% in clear cuttings and 29.4% in thinnings. The relative productivity of operators having experience of 20 years was 23.6% higher in clear cuttings and 16.2% higher in thinnings than that of operators having experience of 3 years. Operators’ experience of 20 years produced 43.1% better lower quartile relative productivity in clear cuttings and 29.1% in thinnings compared to 3 years’ experience. The relative upper quartile productivity was 5.7% higher in clear cuttings for operators aged 45 years than for operators aged 25 years, but otherwise, there was no statistical correlation between upper quartile productivity and age or experience. As a conclusion, CTL harvester operators’ average productivity increases slowly after the initial learning phase up to 15 years of experience. The peak productivity was uncorrelated to age or experience, but the experience raised the bottom productivity values. Keywords: human influence, harvester operator, learning, ageing, experience

1. Introduction Cut-to-length (CTL) harvesters are expensive forest machines, and high investment demands high productivity in order to guarantee investment profitability. Productivity of a CTL harvester depends on numerous variables, stem size being the most recognised (e.g. Jiroušek et al. 2007, Erikson and Lindroos 2014). In addition to stem size, operator effect can be noteworthy (e.g. Kärhä et al. 2004, Purfürst 2010), although other stand properties affect productivity as well (Erikson and Lindroos 2014). Productivity of forest machines has traditionally been measured by manually collected work studies. However, during the last decade, automatic collection of productivity data has become possible (Nuutinen Croat. j. for. eng. 39(2018)1

2013, Erikson and Lindroos 2014, Manner 2015). This kind of data collection enables long term follow up studies, which offers the option to analyse development of productivity at the level of a single machine or operator. As the harvesting operational environment is difficult to change, used machinery and the operator’s capabilities are under scrutiny when aiming to enhance productivity. The operator’s productivity can be improved through education and gained experience (Ovaskainen 2004). In some countries, for example Finland, forest machine operators are systematically trained through vocational education, where students train on basic operations through simulator training and supervised learning at work (Certificate supplement 2015). This education, however, does not

J. Malinen et al.

Productivity of Cut-to-Length Harvesting by Operators’ Age and Experience (15–22)

guarantee professional level productivity from an operator, merely a basis for a lifetime of learning through experience. The learning process of an inexperienced harvester operator is called a learning curve (Purfürst 2010). It describes the level of performance and productivity, which evolves through learning time. Typically, the learning curve of a forest machine operator is described as having the shape of a sigmoidal model (Purfürst 2010). At first, an operator faces the phase of a »slow beginning«, where learning of the basics is occurring. In the second phase, »steep progress«, there is a steep rise which indicates that the rate of improvement is substantial and moves become »automatic«. In the third phase, »plateau«, the learning process becomes slower and the operator has reached the professional level. This, however, does not have to be the maximum level of performance. As the operator gains experience, the collection of available »proven solutions« accumulates and the work becomes easier and more automated. According to Purfürst (2010), most operators begin their career between 50% and 60% of the mean performance and double their performance by the end of learning phase, which lasts on average 8 months. Gellerstedt (2002) noted that, according to machine instructors and trade union representatives, operators’ learning time to get full efficiency takes about five years. In general, job experience improves productivity in several years, but there does come a point where experience does not increase productivity (Skirbekk 2003). Ericsson and Lehman (1996) state that it takes about 10 years to achieve expert competence in tasks where strategic and analytic competence is important. Ageing of operators also has an effect on productivity and work quality. According to Skirbekk’s literature survey (2003), cognitive abilities, reasoning, speed and episodic memory decline significantly before 50 years of age, and continue declining more thereafter. This leads to lower productivity, unless longer experience and higher levels of job knowledge compensates for the declines in mental abilities. However, there are indications that exercising or using speed, reasoning and memory abilities enhance the functional level and soften or halt age-related decline (Skirbekk 2003). An operator’s cognitive performance may vary due to life conditions, shift arrangements and environmental stimuli. For example, lack of quantity and quality of sleep, available light, air quality and cabin conditions, such as vibration and noise, are stress producers that make cognitive tasks more difficult. This, however, does not necessarily have to affect actual perfor-

mance (Ljungberg and Neely 2007). In the study by Nicholls et al. (2004), a 2-shift regime was compared to a 1-shift regime, and over a 24-hour period, the 2-shift regime produced only 22% more wood than the 1-shift regime, although it took 62% more time. In a 2-shift operation, both day shift and afternoon shift operators were slow to reach optimum productivity from the start of the shift. The aim of the study was to examine the relationship between operators’ age, experience and mechanized harvesting productivity in clear cutting and thinning. The used data were long-term data collected from southern Finland by one Finnish forest machine company. The data included 28 operators and 38 different CTL harvesters. Recorded productivities were transformed to relative productivities and average productivity models were created. Case specific productivities were compared to modelled values, and productivity ratio models by age and experience were created.

2. Material and methods 2.1 Harvester data The study data were collected by one Finnish forest machine company. The data included 38 different CTL harvesters (Table 1), from which the production files (prd-files) and time files (dfr-files) were recorded (Skogforsk 2007). The harvesting took place between October 2010 and November 2015, and the data included 802 thinning stands and 582 clear-cutting stands located in southern Finland, in the regions of Pirkanmaa and Kanta-Häme. The total harvested volume was 462,320 m3, of which 181,113 m3 was harvested from thinnings and 281,207 m3 from clear-cuttings. The study data included 28 operators, who were mostly quite experienced in their work. Operators were operating for the same forest machine company and they did not have fixed harvesters, although the Table 1 Harvesters used in the study data collection Manufacturer

Model

Number of machines

John Deere

1070D

John Deere

1170E

John Deere

1270D/E

Komatsu

901TX

Komatsu

911.5

Komatsu

931.1

Croat. j. for. eng. 39(2018)1

Productivity of Cut-to-Length Harvesting by Operators’ Age and Experience (15–22)

J. Malinen et al.

Table 2 Harvester operators in the study data Age class 20–24

25–29

30–34

35–39

40–44

45–50

50–54

Number of operators

Average experience, years

Minimum experience, years

Maximum experience, years

make and type of the machine remained usually the same. Moreover, the forest machine company uses relatively young harvester fleet, and therefore the machinery was mainly renewed at the end of the study. Some of the operators started their employment during the study period and therefore their recorded study span was under five years. The average age of the operators at the end of the recording period was 35 years and the average experience in operating a CTL harvester 14 years (Table 2). Previous experience was collected through operator interviews and the age and experience, accumulated since the beginning of the study, was added to each recorded observation. The collected data files were combined using John Deere Forestry’s TimberOffice 5 software, and further processed in Excel-spreadsheets. Productivities were calculated as gross effective time productivity (E15), which includes delays shorter than 15 minutes. Each productivity recording had to be solely allocated to one harvester operator. If a stand had two operators working in two shifts, the stand included two productivity values. If one stand had more than two operators, the stand was left out from the study.

2.2 Productivity model The study data were divided into two segments, thinning stands and clear-cutting stands. For both segments, a productivity model by average stem size was modelled. The level of productivities in the data were considered confidential, and the data is presented as relative volumes compared to an arbitrarily selected volume, which was given a value of 100. The arbitrarily selected volume for the thinning data was 0.144 m3 and for the clear-cutting data 0.501 m3. The use of scaled values does not affect the results of the study since the aim of the study was to investigate relative productivity of the operators, not the actual level of productivity. For the comparisons of operator’s productivity against modelled average productivity, the operator’s relative productivity (Pr) was calculated by Croat. j. for. eng. 39(2018)1

dividing the stand and operator specific actual productivity ratio (P0) by modelled productivity ratio (Pm). Relative productivities for thinnings and clearcuttings, respectively, were

= Prt

P0t P0t = (1) Pmt –915.68 × Tvolt 2 + 644.96 × Tvolt

and

Prcc = Where: Pr P0 Pm Tvol t cc

P0cc P0cc = (2) Pmcc −175.64 × Tvolcc 2 + 332.54 × Tvolcc relative productivity operators actual productivity ratio, m3%/E15 modelled productivity ratio, m3%/E15 Relative average volume in the stand, m3 thinning clear-cutting.

Both productivity ratio models (Pmt and Pmcc) were forced to go through origin, and the coefficient of determination for the thinning model was 0.589, and 0.436 for the clear-cutting model. Productivity ratios were used in modelling relative productivity by age and by experience. In order to discover the development of operators’ peak performance and also lowermost performance, upper and lower quartiles were calculated by dividing the data into 5 year categories. Age categories started from 20 years and experience categories from 0 years. Within each category, 25% of the best productivities were the upper quartile relative productivities and, respectively, 25% of the lowest productivities were the lower quartile relative productivities. These upper and lower quartile relative productivities were used to model upper and lower quartile relative productivity models. All relative productivity ratio models were second degree polynomial models where the age or the experience of the operator was used as a predictor variable.

J. Malinen et al.

Productivity of Cut-to-Length Harvesting by Operators’ Age and Experience (15–22)

Table 3 Model parameters for relative productivity models Model

Parameter-estimates n

Constant

F-value

Age in clear cuttings q25

147

–0.28965

0.04902

–0.00053

0.374

0.108

43.029

<0.001

All

582

–0.04857

0.05464

–0.00067

0.070

0.226

21.750

<0.001

q75

147

0.35495

0.05251

–0.00070

0.093

0.135

7.419

<0.001

Experience in clear cuttings q25

146

0.48391

0.03601

–0.00092

0.360

0.119

40.214

<0.001

All

582

0.77335

0.03437

–0.00097

0.084

0.224

26.616

<0.001

q75

146

1.24272

0.00864

–0.00027

0.010

0.139

0.731

0.482

Age in thinnings q25

202

0.01938

0.03385

–0.00035

0.204

0.139

25.481

<0.001

All

802

0.50526

0.02417

–0.00024

0.037

0.246

15.406

<0.001

q75

202

1.33130

–0.00051

0.00002

0.003

0.152

0.264

0.768

Experience in thinnings q25

200

0.57411

0.02346

–0.00054

0.195

0.142

23.902

<0.001

All

802

0.88571

0.01905

–0.00044

0.040

0.246

16.723

<0.001

q75

200

1.32584

–0.00068

0.00008

0.006

0.156

0.595

0.553

* n – number of observations; x – predictor variable, R2 – coefficient of determination; SE – standard error; q25 – lower quartile; q75 – upper quartile

3. Results The age and experience of harvester operators were highly correlated (0.947). The youngest age to start operating a CTL harvester was 17 years, the oldest age was 28 years and the average age was 22 years. The experience of operators aged between 30 and 40 years varied between 3 and 18 years. The model parameters for the relative productivities by the harvesting type, and age and experience are presented in Table 3.

3.1 The effect of age and experience on productivity in clear cuttings According to the model, the peak of productivity was in the age range of 40 to 45, after which there was a slight decrease in the productivity (Fig. 1). All operators over 40 were highly experienced with at least 16 years of experience. The relative productivity according to the model at the age of 25 was 0.90, whereas at the age of 45 it was 1.06, 17.8% higher than at the age of 25. The increase is explained by the lack of low

Fig. 1 Relative productivity of CTL harvester operator in clear cuttings by operators’ age Croat. j. for. eng. 39(2018)1

Productivity of Cut-to-Length Harvesting by Operators’ Age and Experience (15–22)

J. Malinen et al.

Fig. 2 Relative productivity of CTL harvester operator in clear cuttings by operators’ experience (years)

Fig. 3 Relative productivity of CTL harvester operator in thinnings by operators’ age

productivity stands, the modelled relative lower quartile productivity increased from the age of 25 to 45 by 38.6%, while the modelled relative upper quartile productivity increased only by 5.7%. The modelled relationship between experience and relative productivity was similar, as expected by the high correlation between age and experience (Fig. 2). Experience of 3 years produced relative productivity of 0.87, and 20 years of experience produced relative productivity of 1.07, 23.6% higher. However, the coefficient of determination and p-value for upper quartile model indicates (Table 3) that there is no statistical correlation between high level productivity and experience among studied trained CTL harvester operators. Conversely, according to the lower quartile model, increased experience improved lower quartile relative productivity by 43.1% from 3 years of experience to 20 years of experience.

model, age had no correlation with relative productivity. However, the productivity in the lower quartile model increased by 29.4% from the operator’s age of 25 to operator’s age of 45, explaining the total increase of productivity by age. The experience of a CTL harvester operator had a similar relationship as age to productivity (Fig. 4). The

3.2 The effect of age and experience on productivity in thinnings The variation in relative productivity was higher in thinnings than in clear cuttings (Table 3). The modelled relative productivity increased quite constantly as the age of operator was higher, although the effect of age was smaller than in clear cuttings (Fig. 3). According to the model, the relative productivity increased from 0.95 to 1.09 (14.9%) between the operator’s age of 25 and 45. According to the coefficient of determination and the p-value for the upper quartile Croat. j. for. eng. 39(2018)1

Fig. 4 Relative productivity of CTL harvester operator in thinnings by operators’ experience (years)

J. Malinen et al.

Productivity of Cut-to-Length Harvesting by Operators’ Age and Experience (15–22)

operators experience level of 3 years produced relative productivity of 0.93, and 20 years of experience produced relative productivity of 1.09, 16.2% higher. Again, the coefficient of determination and p-value for the upper quartile model indicates that there is no statistical correlation between high level productivity and experience among trained CTL harvester operators. Higher experience in the model improved lower quartile relative productivity by 29.1% from 3 years of experience to 20 years of experience.

4. Discussion In the study, the relationship between operators’ ageing, experience and mechanized harvesting productivity was studied by using five-year follow-up data. The advantage of follow-up studies is that operators are unaware of data collection compared to time studies, where motivation and alertness is usually much better. Kuitto et al. (1994) found that productivity of forest transportation in a time study was 22.4% higher than the respective productivity in a follow up study. The highest average relative productivity of individual operator was over 100% better than the smallest average relative productivity. Both the highest and the lowest average relative productivity were accomplished by fairly inexperienced operators, both having less than five years’ experience at the end of the study, highlighting high personal differences between operators. High variance has also been found by Purfürst and Erler (2011). In their study, the best operator worked at a performance level of 125% compared to the relative mean performance level, whereas the worst operator had a mean individual performance of 56%. Lower variances have been reported by Glade (1999): 20–50%, Kärhä et al. (2004): 40% and Väätäinen et al. (2005): 40%. However, in this study, as in the study of Purfüst and Erler (2011), the number of operators was higher than in the studies with lower variance, which could explain some of the difference. According to the relative productivity models, it took 8.8 years’ experience in clear cuttings and 7.2 years’ experience in thinnings to achieve the average productivity level. The respective ages of operators were 31 years for clear cuttings and 29 years for thinnings. These are long periods of experience, but they can be explained by the data in which the majority of the operators were in the age range of 35–39 with an average experience length of 14 years, thus highly experienced but still relatively young for this kind of work requiring high professionalism. However, the shorter time to achieve the average productive level in

thinning than in clear cutting was controversial, as it meant making the generalised assumption that thinning is more demanding work than clear cutting. The result of this study is supported by Häggström (2015), who found out through a visual tracking study that harvester operators looked at their monitor, canopy and falling trees less frequently during the first thinning than during the second thinning and final felling. Although tree selection and strip road planning demand expertise in first thinnings, observation of quality and bucking decisions become more demanding in later thinnings and clear cuttings. The peak in relative productivity models was found at comparatively high experience levels, although previous studies in other occupancies indicate that the productivity-enhancing effect of experience reaches its maximum at ten years (Ericsson and Lehmann 1996). Visual interpretation of the data reveals that the peak values in productivity are achieved at the experience level of 10 years, and these levels decrease slowly as the experience and age increase. At the same time, bottom productivity values are also closing in on average productivity values, leading to the conclusion that there is no clear indication of a drop in productivity due to ageing, although values slowly decrease especially in clear cuttings. This is in line with the findings of Skirbekk (2008), productivity peaks in the age group of 35–44 if the occupancy demands a high level of experience. If the demand for experience falls, the productivity peak shifts toward younger ages. Most typically, harvester operators’ learning and experience studies have concentrated on the initial learning phase, that is, the phase when most of the productivity growth occurs, and lasts up to five years (Gellersted 2002, Purfürst 2010, Purfürst and Erler 2011). This study had only three operators whose experience during the study was under five years. However, the aim of the study was not to investigate the early years of learning, but the development of expertise among experts. Nevertheless, from the three operators, two operators had a learning phase under one year, whereas one operator was still developing after 2.5 years. The information regarding forest machine operators’ productivities and work quality is important for successful forest machine entrepreneurship. According to Purfürst (2010), a harvester operator in the learning phase may cost up to 45,000 Euros in the first eight months in productivity losses, and possibly increase wear and tear or repair costs. Feedback of the individual work performance and possible incentives could enhance motivation, quicken the learning process and improve stability in productivity. Croat. j. for. eng. 39(2018)1

Productivity of Cut-to-Length Harvesting by Operators’ Age and Experience (15–22)

5. Conclusion According to study results, CTL harvester operators’ productivity increases slowly, but steadily, after the initial learning phase up to 15 years of experience, indicating a high demand for experience in this field. High demand for experience keeps the productivity of ageing workers high, despite a possible reduction in cognitive abilities. The effect of ageing diminishes peak productivity, but at the same time the experience raised the bottom productivity values, leading to lower variation in the productivities. The demand for experience was higher in clear cuttings than in thinnings, which was a controversial finding to general estimations. The variability between observations and operators was high, emphasising personal differences affecting productivity between operators regardless of the age and experience.

Acknowledgments The authors are grateful to the operators participating in the study for giving the right to use productivity data.

6. References Certificate supplement, 2015: Vocational upper secondary qualification in forestry, competence area in forest machine operation, forest machine operator. Qualification requirement entered into force (67/001/2014) on 1st August 2015. Finnish National Board of Education. Ericsson, K. A., Lehmann, A. C., 1996: Expert and exceptional performance: Evidence of maximal adaptation to task constraints. Annual Review of Psychology 47(1): 273–305. Eriksson, M., Lindroos, O., 2014: Productivity of harvesters and forwarders in CTL operations in northern Sweden based on large follow-up datasets. International Journal of Forest Engineering 25(3): 179–200. Gellerstedt, S., 2002: Operation of the single-grip harvester: Motor-sensory and cognitive work. International Journal of Forest Engineering 13(2): 35–47. Glade, D., 1999: Single- and double-grip harvesters – Productive measurements in final cutting of shelterwood. Journal of Forest Engineering 10(2): 63–74. Häggström, C., 2015: Human factors in mechanized cut-tolength forest operations. Acta Universitas Agricultureae Sueciae. Doctoral Thesis No. 2015:59. Faculty of Forest Sciences, Swedish University of Agricultural Sciences, Umeå, 77 p.

Croat. j. for. eng. 39(2018)1

J. Malinen et al.

Jiroušek, R., Klvač, R., Skoupý, A., 2007: Productivity and costs of the mechanised cut-to-length wood harvesting system in clear-felling operations. Journal of Forest Science 53(10): 476–482. Kirk, P. M., Byers, J. S., Parker, R. J., Sullman, M. J., 1997: Mechanisation developments within the New Zealand forest industry: The human factors. Journal of Forest Engineering 8(1): 75–80. Kuitto, P. J., Keskinen, S., Lindroos, J., Oijala, T., Rajamäki, J., Räsänen, T., Terävä, J., 1994: Puutavaran koneellinen hakkuu ja metsäkuljetus. Summary: Mechanized cutting and forest haulage. Metsäteho Report 410, 38 p. Kärhä, K., Rönkkö, E., Gumse, S. I., 2004: Productivity and cutting costs of thinning harvesters. International Journal of Forest Engineering 15(2): 43–56. Ljunberg, J. K., Neely, G., 2007: Stress, subjective experience and cognitive performance during exposure to noise and vibration. Journal of Environmental psychology 27(1): 44–54. Manner, J., 2015: Automatic and experimental methods to studying forwarding work. Doctoral Thesis. Faculty of Forest Sciences, Department of Forest Biomaterials and Technology, Umeå, 71 p. Nicholls, A., Bren, L., Humphreys, N., 2004: Harvester productivity and operator fatigue: Working extended hours. International Journal of Forest Engineering 15(2): 57–65. Nuutinen, Y., 2013: Possibilities to use automatic and manual timing in time studies on harvester operations. Dissertationes Forestales 156. School of Forest Sciences, Faculty of Science and Forest, University of Eastern Finland, Joensuu, 68 p. Purfürst, T., Erler, J., 2011: The Human influence on productivity in harvester operations. International Journal of Forest Engineering 22(2): 15–22. Purfürst, T., 2010: Learning curve of harvester operators. Croatian Journal of Forest Engineering 31(2): 89–97. Skirbekk, V., 2008: Age and productivity potential: A new approach based on ability levels and industry-wide task demand. Population and Development Review 34: 191–207. Skirbekk, V., 2003: Age and individual productivity: A literature survey. MPIDR Working Paper WP 2003-028. MaxPlanck-Institut für demografische Forschung, Rostock, 37 p. Skogsforsk, 2007: Standard for forest data and communications – StanForD. http://www.skogforsk.se/contentassets/ b063db555a664ff8b515ce121f4a42d1/stanford_maindoc_070327.pdf Väätäinen, K., Ovaskainen, H., Ranta, P., Ala-forssi, A., 2005: Hakkuukoneenkuljettajan hiljaisen tiedon merkitys hakkuutulokseen työpistetasolla (The importance of harvester operator’s tacit knowledge at the level of work position). Research publications of the Finnish Forest Research Institute 937, 93 p.

J. Malinen et al.

Productivity of Cut-to-Length Harvesting by Operators’ Age and Experience (15–22)

Authors’ addresses: Jukka Malinen, PhD. * e-mail: jukka.malinen@uef.fi University of Eastern Finland School of Forest Sciences Yliopistokatu 7 80101 Joensuu FINLAND

Received: September 23, 2016 Accepted: July 02, 2017

Janne Taskinen e-mail: janne.taskinen@mkpd.fi Timo Tolppa e-mail: timo.tolppa@mkpd.fi Metsäkonepalvelu Oy Konepajantie 12 13300 Hameenlinna FINLAND * Corresponding author Croat. j. for. eng. 39(2018)1

Original scientific paper

Productivity and Costs of Harwarder Systems in Industrial Roundwood Thinnings Kalle Kärhä, Asko Poikela, Teijo Palander Abstract In several studies, the harwarder has proven to be a more cost-effective wood harvesting system than the traditional two-machine (harvester-forwarder) system, especially when the average stem size of the marked stand is relatively small, the removals per hectare/stand low (i.e. the harvesting site small), and the forwarding distance short. One of the strengths of a harwarder is considered to be the lower relocation costs compared to the two-machine system. The time consumption of harwarder relocations have not, however, been reported in the previous harwarder studies. Metsäteho Oy conducted a follow-up study of harwarders in industrial roundwood harvesting, and also investigated the relocations of harwarders. A total of five – three Ponsse Wisent Dual and two Valmet 801 Combi – harwarders were examined in the follow-up study. The amount of harvested industrial roundwood in the study totalled nearly 30,000 m3. The cost calculations showed that the harwarder system is more competitive than the twomachine system when the average stem size of the marked stand is relatively low, i.e. less than 110–170 dm3. Furthermore, harwarders were the most competitive at low-removal harvesting sites. The proportion of the total working time of harwarders used in relocations between harvesting sites was 2.5%, and the effective relocation time was, on the average, 1.3 hours/relocation. The study results underlined that it makes sense to harvest relatively small-removal and small-diameter thinning stands marked for harvesting with a harwarder while, conversely, it is more worthwhile to harvest sites with larger removals and trees using a two-machine harvester-forwarder system, thereby raising the profitability of forest machine business. Keywords: harwarder, harvester-forwarder system, wood harvesting, cost-efficiency, machine relocation, follow-up study

1. Introduction In the 2010s, total roundwood removals amounted to 59.7–70.3 million m3 (over bark) in Finland (Hakkuukertymä metsäkeskusalueittain 2017). Two mechanized harvesting systems are used for industrial roundwood and energy wood: Þ traditional two-machine (harvester and forwarder) system Þ harwarder system (i.e. the same machine can perform both cutting and forest haulage). During the last few years, annually, there has been an average of between 1870–1990 harvesters and 1930–2020 forwarders in use in Finland (Mäki-Simola 2017). At the same time, there have been less than 100 harwarders in wood harvesting operations in Finland. Croat. j. for. eng. 39(2018)1

The active development of harwarders in Finland started in the late 1990s after Lilleberg (1997) demonstrated that the harwarder was a more cost-effective wood harvesting system than a two-machine harvesting system, when the average industrial roundwood stem size in the marked stand was less than 150 dm3. Since then, the productivity and profitability of harwarders in industrial roundwood harvesting, as well as in energy wood harvesting, have been investigated in several studies. These trials have been almost exclusively time studies (e.g. Cederlöf 1997, Hallonborg et al. 1999, 2005, Strömgren 1999, Eriksson and Rytter 2000, Hallonborg and Nordén 2000, Rieppo and Pekkola 2001, Andersson 2002, Bergkvist et al. 2002, 2003, Andersson 2003, Rieppo 2003, Wester and Eliasson 2003, Ljungdahl 2004, Nordén et al. 2005, Laitila and

K. Kärhä et al.

Productivity and Costs of Harwarder Systems in Industrial Roundwood Thinnings (23–33)

Asikainen 2006, Bergkvist 2007, 2008, Johansson 2010, Nordin 2011, Di Fulvio et al. 2012, Zinkevicius and Vitunskas 2013, Spinelli et al. 2014, Jonsson et al. 2016a, 2016b, Manner et al. 2016). Comprehensive, long-term follow-up study data on harwarders has been produced in only two studies: Sirén and Aaltio (2003) in industrial roundwood harvesting, and Kärhä (2006) in energy wood harvesting. In the studies of Strömgren (1999), Hallonborg and Nordén (2000), Rieppo and Pekkola (2001), Bergkvist et al. (2003), Rieppo (2003), Sirén and Aaltio (2003), Talbot et al. (2003), Nordén et al. (2005), Jylhä et al. (2006), Kärhä (2006), Bergkvist (2007), Väätäinen et al. (2007), Johansson (2010), Jonsson et al. (2016a), and (2016b) the harwarder has proven to be a more cost-effective wood harvesting system than the traditional two-machine system, especially when the average stem size of the marked stand is relatively small, the removals per hectare/stand low (i.e. the harvesting site small), and the forwarding distance short. These kind of harvesting conditions are common in Finnish stands, and the share of cuttings from them (e.g. peatland cuttings) for industrial roundwood volume will be increasing in the near future (Palander and Kärhä 2016). One of the main benefits in the use of harwarders is the fact that with harwarder operations one of the normal work elements of wood harvesting with twomachine system (i.e. loading of timber from ground) can be eliminated when direct loading is used in harwarder work (e.g. Di Fulvio and Bergström 2013, Jonsson et al. 2016a). Besides, one of the strengths of a harwarder is considered to be the lower relocation costs compared to the two-machine harvesting system (e.g. Talbot et al. 2003, Asikainen 2004). The time consumption of harwarder relocations has not, however, been reported in the previous harwarder studies. Hence, the aims of this study were to: Þ evaluate the productivity of two harwarder systems by using data on follow-up study Þ estimate the relocation and utilization of harwarder systems by follow-up study data Þ compare the profitability of harwarder systems to a two-machine system also including the relocation and machine utilization effects.

2. Material and methods 2.1 Follow-up study A total of five – three Ponsse Wisent Dual (also in this article Ponsse Dual) and two Valmet 801 Combi (also Valmet Combi) – harwarders were examined in the follow-up study. Ponsse Dual harwarders were equipped with a separate harvester head for wood cut-

ting and a separate timber grapple for forest haulage. Correspondingly, the Valmet 801 Combi harwarders were fitted with a fixed load space and an integrated harvesting grapple for both cutting and forwarding. Eleven different harwarder operators participated in the follow-up study. The operators’ work experience with harwarder work varied from some months to two years and with forest machine work between 1–23 years. The follow-up period started in September, 2004, and continued until May, 2005. The follow-up data was collected by Telmu 100 dataloggers. The duration and reason for each time element was recorded by the dataloggers. The accuracy of data collection was one second (s). The following time elements in the study were used: Þ Effective harwarder work: Þ cutting Þ forwarding. Þ Delays: Þ delays caused by harwarder (e.g. repairs and maintenance in the forest, changing the configuration from harvester to forwarder and vice versa with the Ponsse Dual harwarders) Þ delays due to operator (i.e. eating and personal breaks) Þ delays coming from communication (e.g. telephone calls) Þ harwarder relocations Þ other delays (e.g. larger repairs and maintenance at the workshop). A total of 707 follow-up study days were required to collect the data. The harvesting conditions were obtained from the enterprise resource planning (ERP) systems of the wood procurement organizations for which each harwarder was contracted. The total industrial roundwood harvested with the Ponsse Dual harwarders was close to 25,000 m3 (Table 1). The study material with the Valmet Combi harwarders was smaller, around 5000 m3. The amount of harvested industrial roundwood in the follow-up study totalled nearly 30,000 m3 (Table 1). There were 92 harvesting stands, and data concerning the harvesting conditions was obtained from 70 stands (Table 1). The average size of harvesting stands was 4.0 ha in the follow-up study. The harwarders were primarily used for thinnings in the follow-up study: 14% of the total volume of industrial roundwood harvested came from first thinnings and 43% from later thinnings. Less than one-third of the wood quantity came from final cuttings. The proportion of other/combined cuttings was 11%. Furthermore, harwarders were used principally for real harwarder work, i.e. both cutting and forwarding were done by Croat. j. for. eng. 39(2018)1

Productivity and Costs of Harwarder Systems in Industrial Roundwood Thinnings (23–33)

K. Kärhä et al.

Table 1 Number of stands in follow-up study and data for harvesting by harwarder and on average Stands (with condition data1), No.

Ponsse Wisent Dual

Valmet 801 Combi

Total / Average

64 (55)

28 (15)

92 (70)

Harvesting sites by mode of operation (with condition data ), No. 3

Real harwarder work

35 (33)

20 (15)

55 (48)

Cutting only4

33 (26)

10 (1)

43 (27)

4 (2)

0 (0)

4 (2)

Forest haulage only5

Harvesting sites by cutting method2 (with condition data1), No. First thinnings

18 (13)

9 (6)

27 (19)

Later thinnings

31 (24)

8 (8)

39 (32)

Final cuttings

27 (26)

1 (0)

28 (26)

Other/combined cuttings6

11 (8)

12 (2)

33 (10)

Total roundwood removal, m3

24,935

5023

29,958

Tree species, % Norway spruce

Scots pine

Broadleaf

Total harvesting area, ha

233

132

366

m3/stand

453

335

428

107

Industrial roundwood removal 3

m /ha Number of timber assortments, No.

9.0

3.7

8.4

Average stem size in stand, dm3

247

128

226

Density of removal, trees/ha

488

532

440

Forwarding distance, m

248

229

245

No snow

<20 cm

Thickness of snow, %

21–40 cm

41–60 cm

>60 cm

Undergrowth situation, % Pre-cleared

Non-obstructive

Moderate

High degree of obstructive

Terrain, % Normal

100

More difficult than normal

Difficult

Harvesting stands with obtained harvesting conditions, e.g. average stem size in stand, roundwood removal, area of stand, forwarding distance As the same harvesting stand may consist of several modes of operation and cutting methods, there may be more harvesting sites than study stands 3 Both cutting and forest haulage with harwarder 4 Only cutting with harwarder 5 Only forwarding with harwarder 6 Other cutting methods or different harvesting sites of study stands had to be combined 2

Croat. j. for. eng. 39(2018)1

K. Kärhä et al.

Productivity and Costs of Harwarder Systems in Industrial Roundwood Thinnings (23–33)

a harwarder at the harvesting site (69% of the total volume of industrial roundwood harvested in the follow-up study). Harwarders were also used to balance two-machine (harvester-forwarder) harvesting systems, with the cutting carried out by a harwarder and the forwarding later performed by a forwarder (30% of the total volume of roundwood harvested). There were only a few harvesting sites where the harwarders carried out only forest haulage at the harvesting site (Table 1). Follow-up study data for the Valmet Combi harwarders was focused on thinnings, which was reflected in markedly smaller hectare and stand-specific roundwood removals, average stem sizes, and number of timber assortments compared to the Ponsse Dual harwarders (Table 1). Out of the total harvested timber volume, 44% was Norway spruce (Picea abies (L.) Karst.), 40% Scots pine (Pinus sylvestris L.), and 16% broadleaf (i.e. Betula verrucosa Ehrh., Betula pubescens Ehrh. and Populus tremula L.). Final cuttings were spruce-dominated and first thinnings pine-dominated. With the Ponsse Dual harwarders, half of the harvested volume was spruce, and with the Valmet Combi harwarders, more than two-thirds was pine. The average number of timber assortments per harvesting site was 8.4, ranging from 2–15 between harvesting sites. With the Ponsse Dual harwarders, the number of assortments ranged from 4–15, and with the Valmet Combi harwarders from 2–5. The cutting was initially primarily performed by the Ponsse Dual harwarders included in the follow-up study and, afterwards, the machines were outfitted for forwarding and used to haul the felled timber to roadside landings. In thinnings, the following working method was mainly applied using the Valmet Combi harwarders: The harwarder was driven forward into the stand while, at the same time, the trees along the

strip road were cut and both sides of the strip road were thinned. The felled timber was bunched mainly into piles along the strip road. At the end of the strip road, the harwarder turned around and drove back along the harvested strip road while, at the same time, the bunched logs and poles were loaded. In final cuttings, the Valmet Combi harwarders were driven forward parallel with the edge of the stand, while cutting along one side. Direct loading was not carried out with the Valmet Combi harwarders in the follow-up study in either final cuttings or thinnings.

2.2 Cost calculations and system analysis The operating costs were calculated using the Forest Machine Calculation Program of Metsäteho Oy. Operating costs include both time-dependent costs (capital depreciation, interest expenses, labor costs, insurance fees, administration expenses) and variable operating costs (fuel, repair and service, machine relocations). Cost calculations were prepared for two harwarders, of which the purchase price of the Harwarder system II (Valmet 801 Combi) was 130,000 € (VAT 0%) higher than that of Harwarder system I (Ponsse Wisent Dual), and for the two-machine harvesting system, which consisted of a harvester for thinnings (weight: 16–18 tonnes; e.g. John Deere 1070, Komatsu 901, Ponsse Beaver) and a medium-duty forwarder (carrying capacity: 12 tonnes; e.g. John Deere 1110, Komatsu 845, Ponsse Wisent) (Table 2). For all the machines, the annual operating (E15, including short (<15 min) delays) hours were standardized at 2511 operating hours in the calculations. In the cost calculations, the proportion of thinnings was 40% of the total volume of industrial roundwood harvested. The operating hour costs for the harvester for thinnings were 101 €/E15 hour and for the medium-duty forwarder 72 €/E15 hour (Table 2). The operating hour costs of the Harwarder system I (Ponsse Dual) was 94 €/E15

Table 2 Purchase prices, operating hour productivities and annual outputs used in cost calculations, as well as calculated operating hour costs of machines Machine

Productivity, m3/E15 hour

Purchase price, € (VAT 0%)

Thinnings

Final cuttings

Industrial roundwood,

Operating hour costs,

m /a

€/E15 hour

Harwarder system I (Ponsse Wisent Dual)

370,000

6.1

7.7

17,499

105

II (Valmet 801 Combi)

500,000

6.1

7.7

17,499

Two-machine system Harvester

380,000

9.0

18.0

32,300

101

Forwarder

260,000

11.0

15.0

32,840

Croat. j. for. eng. 39(2018)1

Productivity and Costs of Harwarder Systems in Industrial Roundwood Thinnings (23–33)

hour and of the Harwarder system II (Valmet Combi) 105 €/E15 hour. The wood harvesting costs in thinnings with Harwarder systems I and II were compared to the harvesting costs with the two-machine harvesting system. The effective (E0, excluding delays) hour productivities in thinnings with the two-machine harvesting system in cutting and forest haulage was determined by the time consumption models presented by Kärhä et al. (2006). It was assumed that, when the average stem size in the stand increased from 50 dm3 to 250 dm3, the industrial roundwood removal increased from 38 m3/ha to 84 m3/ha in thinnings (cf. Kärhä and Keskinen 2011). There were 500 Norway spruce undergrowth trees per hectare in the thinning stand, and the average height of the spruce undergrowth trees was 2 m. The average load size was 11.0 m3 in forest haulage with a forwarder (cf. Eriksson and Lindroos 2014). The effective hour (E0) productivities of cutting and forest haulage of two-machine system were converted to operating hour (E15) productivities by coefficients of 1.393 and 1.302, respectively, in the cost calculations.

2.3 Data analysis The variables were analyzed using percentage shares and mean values. The differences between harwarder systems (Ponsse Dual and Valmet Combi) and cutting methods (first thinning, later thinning and final cutting) were analyzed using the Mann-Whitney U-test and Kruskal-Wallis one-way ANOVA test. The operating (E15) hour productivity in real harwarder work was modeled by applying regression analysis with the average stem size in the stand, industrial roundwood removal per hectare, density of removal, share of tree species volume, average forwarding distance, and number of timber assortments as independent variables. The suitability of the models with respect to the data was numerically assessed on the basis of the degree of explanation and statistical significance (p<0.05).

K. Kärhä et al.

Fig. 1 Distribution of effective working time in real harwarder work by cutting method, and on the average in the follow-up study tions 2.5%, and the proportion of communication (i.e. telephone calls, forest visits by forest machine entrepreneur, wood procurement officer, or forest owner) was 1.5% of the total working time. In real harwarder work, based on the entire followup study material (average stem size in marked stand 198 dm3 and average forest haulage distance 239 m), an average of 57% of the effective working time was used for cutting and 43% for forest haulage (Fig. 1).

3. Results 3.1 Total time consumption and productivity In the follow-up study, the technical utilization rate of the harwarders was, on the average, 88.1%, and the operational utilization rate 82.6%. In the real harwarder work, the share of the effective working time was 78.2% of the total working time. Correspondingly, the proportion of machine delays (i.e. repairs and maintenance) was 12.4% of the total working time. The proportion of operator delays (i.e. eating and personal breaks) was 5.2%, the proportion of harwarder relocaCroat. j. for. eng. 39(2018)1

Fig. 2 Distribution of harwarder relocation times out of the number and of time consumption in the follow-up study

K. Kärhä et al.

Productivity and Costs of Harwarder Systems in Industrial Roundwood Thinnings (23–33)

later) thinnings with the harwarders were combined and modeling was conducted based on these data. In the case of thinnings, the productivity per operating hour of real harwarder work was best explained by the average stem size in the marked stand (Table 3, Eq. 1, Fig. 3). The other independent variables (i.e. the share of tree species volume, industrial roundwood removal per hectare, density of removal, average forwarding distance, and number of timber assortments) had no statistically significant impact on the operating hour productivity in thinnings. Table 3 Regression model for operating hour productivity of real harwarder work in thinnings

Fig. 3 Operating hour productivity in thinnings in real harwarder work by harvesting site, and productivity curve as a function of average stem size (Table 3, Eq. 1) With first thinnings (89 dm3 and 280 m), the cutting took an average of 63% and forwarding 37% of the effective working time. With final cuttings (326 dm3 and 179 m), the effective working time was split almost equally between cutting and forest haulage (Fig. 1). In the follow-up study, the effective relocation time of harwarders used in relocations between harvesting sites was, on the average, 1.3 hours/relocation. The majority of harwarder relocations took 0.5–1.0 hour/relocation (Fig. 2). The average relocation time with the Ponsse Dual harwarders was 1.29 hours and with the Valmet Combis 1.48 hours/relocation. In real harwarder work within the follow-up study, the productivity per operating hour in first thinnings was, on the average, 5.1 m3/E15 hour and in later thinnings 6.4 m3/E15 hour. The operating hour productivity in real harwarder work was significantly higher in later thinnings than in first-thinning stands (p<0.05). The larger stem size in later thinnings explained the disparity in the productivity levels (Fig. 3). In the first thinnings, the productivity with the Valmet Combi harwarders was, on the average, 1.5 m3/E15 hour higher than with the Ponsse Duals. Respectively, the productivity in the later thinnings with the Ponsse Dual harwarders was 0.8 m3/E15 hour higher than with the Valmet Combis. There was no significant difference in productivity between the Ponsse Dual and Valmet Combi harwarders in (first and later) thinning stands (p=0.334). The productivity observations in (first and

Coefficient

Estimate of coefficient

Standard error of estimate

t-value

–1.877

3.728

–0.504

1.641

0.796

2.062*

F value = 3.056*; R2 = 0.13*; * p<0.05; ** p<0.01; *** p<0.001

y = a + b ´ LN(x)

(1)

Where: y operating hour productivity, m3/E15 hour x average stem size in the stand, dm3 a constant b coefficient of variable. When the average stem size in the stand was 100 dm3, the productivity of real harwarder work in thinnings was 5.7 m3/E15 hour (Fig. 3, Table 3). The productivity was 6.8 m3/E15 hour when the average stem size was 200 dm3. In the final cutting of the real harwarder work within the follow-up study, the average productivity was 7.7 m3/E15 hour.

3.2 Profitability of harvesting systems The harwarder systems were more competitive than the two-machine system when the average stem size of the marked stand was relatively low, i.e. less than 110–170 dm3 (Fig. 4). In this case, the industrial roundwood removal was typically below 55–70 m3/ha (cf. Kärhä and Keskinen 2011). Furthermore, harwarders were the most competitive in low-removal – i.e. small-sized – stands, particularly at harvesting sites that were below 50 m3. As the stem size in the stand and roundwood removal per hectare/stand increased, the competitiveness of the two-machine harvesting system improved in comparison to that of the harwarder systems (Fig. 4). Croat. j. for. eng. 39(2018)1

Productivity and Costs of Harwarder Systems in Industrial Roundwood Thinnings (23–33)

Fig. 4 Effect of average stem size on relative harvesting costs of thinning wood with harwarder systems I and II and with a twomachine harvesting system; industrial roundwood removal increased from 38 m3/ha (average stem size 50 dm3) to 84 m3/ha (250 dm3) (Kärhä and Keskinen 2011), and the forwarding distance was 250 m; harvesting costs 100 = Harvesting costs with a twomachine harvesting system at an average stem size of 100 dm3

4. Discussion 4.1 Follow-up study The follow-up study material totalled nearly 30,000 m3. The amount of follow-up study material collected was extensive compared to the harwarder follow-up studies carried out earlier: The material in the study of Sirén and Aaltio (2003) was around 16,000 m3 industrial roundwood with the Pika 828 Combi harwarders. Correspondingly, in the follow-up study by Kärhä (2006), the study material was close to 14,000 m3 of small-diameter whole trees from young stands. The amount of study material collected from the Ponsse Wisent Dual harwarders was significantly greater than that from the Valmet 801 Combi har warders. The reasons for that being that the Valmet Combi harwarders were only used in one work-shift, and the study stands were mostly thinnings (cf. Table 1). Nevertheless, the material of this research was quite small compared to the material of very large follow-up study by Eriksson and Lindroos (2014) with two-machine harvesting systems including around 23 million m3 (over bark) of industrial roundwood. In this follow-up study, there were 92 harvesting stands in total, and data concerning harvesting condiCroat. j. for. eng. 39(2018)1

K. Kärhä et al.

tions was obtained from 70 stands. The effect of harvesting conditions on the productivity of real harwarder work could be researched with around 50 harvesting sites. For modeling, it would have been useful if there had been more harvesting sites with harvesting condition data. The following issues restricted the amount of material for examination of real harwarder work: Þ harvesting condition data could not be obtained from all study stands Þ real harwarder work was not conducted in all study stands, but harwarders balanced the twomachine harvesting systems in several stands (Table 1) Þ some follow-up study stands had to be deleted from the final material used for modeling productivity, because there were problems collecting time consumption data in some study stands. Eleven different harwarder operators took part in the study. Their work experience in harwarder work varied from several months to two years. This caused a large amount of variation for operating productivity in thinnings (Fig. 3). Earlier wood harvesting studies have pointed out that there is a significant correlation between the operator’s working experience and his/ her productivity in forest machine work, especially when operating in dense thinning stands. For instance, Sirén (1998), Kärhä et al. (2004) and Ovaskainen (2009) have shown that the differences between operators using the same machines are as high as 35–40%. In thinnings, the operating hour productivity of real harwarder work was, statistically, most significantly explained by the average stem size. The coefficient of determination (R2) of the productivity model (Table 3, Eq. 1) was left relatively low, because of the large variation of productivities by harvesting site (cf. Fig. 3). The productivity of real harwarder work in final cuttings cannot be significantly explained in the study. In the follow-up study by Sirén and Aaltio (2003), the operating hour productivity in thinnings was explained by the average stem size in the stand, roundwood removal, and number of timber assortments. In this study, the number of timber assortments had no significant impact on the productivity of real harwarder work in thinnings. When comparing the productivity models of Sirén and Aaltio (2003) with the productivity models of this study, it is noted that in this study the productivity was 1.2–1.9 m3/E15 hour higher than that of Sirén and Aaltio, when the average stem size in the stand was 50–200 dm3. Development of harwarder

K. Kärhä et al.

Productivity and Costs of Harwarder Systems in Industrial Roundwood Thinnings (23–33)

technology partly explains the higher productivity in this research. Besides, the operators and their working skills may account for the disparity in productivities. According to the results obtained, the harwarder is a more cost-effective wood harvesting system than the traditional two-machine system, when the average stem size in the stand is relatively small and the removals per hectare/stand are low. Hence, the findings are in line with earlier studies (e.g. Lilleberg 1997, Strömgren 1999, Hallonborg and Nordén 2000, Rieppo and Pekkola 2001, Bergkvist et al. 2003, Rieppo 2003, Sirén and Aaltio 2003, Talbot et al. 2003, Nordén et al. 2005, Jylhä et al. 2006, Kärhä 2006, Bergkvist 2007, Väätäinen et al. 2007, Johansson 2010, Jonsson et al. 2016a, 2016b) concerning most suited harvesting conditions for the harwarder system. The study results also emphasized that the Ponsse Dual concept is a more cost-efficient system than the Valmet Combi (cf. Fig. 4). The reason for this being is that the Ponsse Wisent Dual, as well as the Ponsse Buffalo Dual, are clearly inexpensive machines when compared to the Valmet 801 Combi harwarder. Väätäinen et al. (2007) have also underlined the superior competitiveness of Ponsse Dual harwarders when compared to the Valmet 801 Combi with a fixed load space. Correspondingly, Talbot et al. (2003) have found that the Valmet Combi is a more cost-efficient wood harvesting system than the Ponsse Dual concept. In the study, all harwarders were equipped with fixed load space, and therefore the working method of direct loading was not used. Many research reports, for instance by Hallonborg and Nordén (2000), Andersson (2002), Bergkvist et al. (2003), Wester and Eliasson (2003), and Jonsson et al. (2016a), have illustrated that direct loading is more productive working method with the harwarder system in final fellings when separate timber loading from the ground can be avoided. According to the studies, direct loading could improve further the cost-competiveness of harwarders. By developing harwarders and their working methods (e.g. direct loading in thinnings and final cuttings) and organization, it will be possible to enhance further the competitiveness of harwarders (Lindroos 2012, Ringdahl et al. 2012). In the follow-up study, the share of the total working time of harwarders in relocations was 2.5%, and the effective relocation time was, on the average, 1.3 hours/relocation. These findings are new because there is no earlier information on the harwarder relocation. In Finland, the latest comprehensive study on the relocations of harvesters and forwarders was conducted more than 20 years ago (Kuitto et al. 1994). In the beginning of the 1990s, the relocation distance with

the harvesters was, on the average, 28 km, and the relocation time consumption 1.7 hours/relocation (Kuitto et al. 1994). Respectively, with the forwarders, the average relocation time from one harvesting site to another was 1.2 hours and the average relocation distance was 21 km. It can be noticed that the figures of this research are very close to the values of Kuitto and his colleagues (1994). In this study the relocation distances of harwarders were not reported. Kärhä et al. (2007) interviewed 13 harwarder contractors, who mostly owned both harwarders and harvesters and forwarders. Kärhä et al. (2007) reported that harwarder contractors estimated that the average relocation distance is 28 km with their harwarders and 32 km with their two-machine harvesting systems. In the follow-up study by Eriksson and Lindroos (2014), the average share of relocations with harvesters and forwarders were 1.4–1.5% and 0.9–1.1% of the total machine computer uptime, respectively. The bigger size of harvesting sites in Sweden explains probably the smaller proportions of machine relocations.

4.2 Future prospects Currently, the total number of harwarders in use in Finnish forests is less than one hundred, of which more than half are mainly engaged in energy wood harvesting and the remainder in industrial roundwood harvesting. Harwarders have not been as widely adopted as it would be expected in the light of the positive results of harwarder studies. Possible reasons include resistance and prejudice towards harwarders, together with entrenched preferences for traditional harvesting technology. These factors came to light in Metsäteho’s investigation on the increasing use of tracked excavators in harvesting operations in Finland (Bergroth et al. 2006). It is estimated that the number of harwarders will increase in the near future in Finland. This development forecast is based on the following factors: Þ cost effectiveness in wood harvesting is being sought at the level of the stand marked for harvesting. A harwarder has a clear competitive advantage in small-removal thinnings and final cuttings, forest fellings in the archipelago, harvesting of wind-felled trees, and in seed tree and shelterwood fellings (Kärhä et al. 2001, Jylhä et al. 2006). It makes sense to harvest relatively small-removal and small-diameter stands marked for harvesting with a harwarder while, conversely, it is more worthwhile to harvest sites with larger removals and trees using a twomachine harvesting system, thereby raising the profitability of forest machine business Croat. j. for. eng. 39(2018)1

Productivity and Costs of Harwarder Systems in Industrial Roundwood Thinnings (23–33)

Þ the structural change in cuttings is setting new demands on the harvesting fleet. Wood harvesting volumes of thinnings and on peatlands will grow during the next years (e.g. Nuutinen et al. 2000, Korhonen et al. 2007). The harvesting conditions described above (small stem size and low removals) are ideally suited for the harwarder. The use of harwarders also means less driving in stand, which is needed during harvesting operations, thus minimizing strip road rutting (e.g. Palander and Kärhä 2016). In peatland harvesting, however, long forwarding distances may reduce the profitability of harvesting based on a harwarder Þ as a result of changes in the forest machine business field, the size of forest machine contracting businesses is growing and large regional responsibilities in contracting are increasing (e.g. Rekilä and Räsänen 2008). These changes are creating a potential for the use of specialized harvesting fleet. In this respect, the acquisition of a harwarder alongside two-machine harvesting systems may be a profitable alternative in wood harvesting operations.

5. Conclusions It was noted in this study that the productivity of harwarders has increased 1.2–1.9 m3/E15 hour, when the average stem size in the stand was 50–200 dm3. Development of harwarder technology partly explains the higher productivity. Furthermore, the average share of relocations with harwarders was 2.5% of the total machine time with the average relocation time of 1.3 hours/relocation. Nonetheless, harwarders have not been as widely adopted as it would have been expected in the light of the positive results. Actually, the reasons for the relatively slow growth in the use of harwarders have not been documented. Possible reasons include resistance towards harwarders. Moreover, one possible reason may be the fact that harwarder work requires higher professional skills and qualifications for operators than harvester or forwarder. The harwarder operator has to have very good skills in both cutting and forwarding work. In forest industry, cost effectiveness in wood harvesting is being sought at the level of the stand marked for harvesting, as well as from the point of view of the forest machine business. In Finland, the size of forest machine contracting businesses is growing and large regional responsibilities in contracting are increasing. These changes are creating a potential for the use of specialized wood harvesting fleet. In this respect, the Croat. j. for. eng. 39(2018)1

K. Kärhä et al.

optimization of allocation and use of harwarder and two-machine harvesting systems will be a profitable solution in the future wood harvesting operations.

6. References Andersson, J., 2002: Harwarder productivity in final felling – a comparison of three working methods. Sveriges lantbruksuniversitet, Students’ Reports 56. Andersson, J., 2003: Harwarder productivity in thinning – a comparison of two working methods. Sveriges lantbruks universitet, Students’ Reports 59. Asikainen, A., 2004: Integration of Work Tasks and Supply Chains in Wood Harvesting – Cost Savings or Complex Solutions? International Journal of Forest Engineering 15(2): 11–17. Bergkvist, I., 2007: Harwarders in final felling. Skogforsk, Resultat 15/2007. Bergkvist, I., 2008: Direct loading challengers can break the harvester/forwarder dominance. Skogforsk, Resultat 9/2008. Bergkvist, I., Hallonborg, U., Nordén, B., 2002: Valmet 801 Combi i gallring med fast lastutrymme för standardlängder (Valmet 801 Combi with fixed load space in thinnings). Skogforsk, Arbetsrapport 518. Bergkvist, I., Hallonborg, U., Nordén, B., 2003: Valmet 801 Combi i gallring och slutavverkning med vridbart lastut rymme för fallande längder (Valmet 801 Combi with rotating load space in thinnings and final cuttings). Skogforsk, Arbetsrapport 526. Bergroth, J., Palander, T., Kärhä, K., 2006: Excavator-based harvesters in wood cutting operations in Finland. Forestry Studies 45: 74–88. Cederlöf, O., 1997: Time study of a combined harvester-forwarder. Sveriges lantbruksuniversitet, Students’ Reports 1. Di Fulvio, F., Bergström, D., Kons, K., Nordfjell, T., 2012: Productivity and Profitability of Forest Machines in the Harvesting of Normal and Overgrown Willow Plantations. Croatian Journal of Forest Engineering 33(1): 25–37. Di Fulvio, F., Bergström, D., 2013: Analyses of a single-machine system for harvesting pulpwood and/or energy-wood in early thinnings. International Journal of Forest Engineering 24(1): 2–15. Eriksson, M., Lindroos, O., 2014: Productivity of harvesters and forwarders in CTL operations in northern Sweden based on large follow-up datasets. International Journal of Forest Engineering 25(3): 179–200. Eriksson, P., Rytter, L., 2000: Harvesting of wood fuel by harwarder – an alternative to late cleaning in hardwood stands. Skogforsk, Resultat 4/2000. Hallonborg, U., Nordén, B., 2000: Reckoning on harwarders for final felling. Skogforsk, Resultat 21/2000.

K. Kärhä et al.

Productivity and Costs of Harwarder Systems in Industrial Roundwood Thinnings (23–33)

Hallonborg, U., Bucht, S., Olaison, S., 1999: A new approach to thinning: Integrated off-ground handling reduces damage and increases productivity. Skogforsk, Resultat 23/1999.

Kuitto, P. J., Keskinen, S., Lindroos, J., Oijala, T., Rajamäki, J., Räsänen, T., Terävä, J., 1994: Mechanized cutting and forest haulage. Metsäteho, Raportti 410.

Hallonborg, U., Nordén, B., Lundström, H., 2005: Ponsse Dual Buffalo i slutavverkning (Ponsse Dual Buffalo in final cuttings). Skogforsk, Arbetsrapport 586.

Laitila, J., Asikainen, A., 2006: Energy wood logging from early thinnings by harwarder method. Baltic Forestry 12(1): 94–102.

Hakkuukertymä metsäkeskusalueittain (Total roundwood removals by forestry center area), 2017: Natural Resources Institute Finland, Statistics Database.

Lilleberg, R., 1997: Harvester-forwarder for logging in firstthinning stands. Metsäteho, Raportti 28.

Johansson, P., 2010: Forest fuel harwarder in early thinning of lodgepole pine. Sveriges lantbruksuniversitet, Students’ Reports 283. Jonsson, R., Jönsson, P., Lundström, H., 2016a: Performance and cost in final felling for Komatsu X19 harwarder with quick hitch. Arbetsrapport Från Skogforsk 911. Jonsson, R., Jönsson, P., Manner, J., Björheden, R., Lundström, H., 2016b: Performance and costs for the Komatsu X19 harwarder compared to Komatsu 941/895 harvester/ forwarder in large size final felling. Arbetsrapport Från Skogforsk 912. Jylhä, P., Väätäinen, K., Rieppo, K., Asikainen, A., 2006: Aines- ja energiapuun hakkuu ja lähikuljetus korjureilla. Kirjallisuuskatsaus (Cutting and forwarding of industrial roundwood and energy wood with harwarders. Literature review). Working Papers of the Finnish Forest Research Institute 34. Kärhä, K., 2006: Whole-tree harvesting in young stands in Finland. Forestry Studies 45: 118–134. Kärhä, K., Keskinen, S., 2011: Ensiharvennukset metsäteollisuuden raaka-ainelähteenä 2000-luvulla (First thinnings as a raw material source of Finnish forest industries in the 21st century). Metsäteho, Tuloskalvosarja 2/2011. Kärhä, K., Mäkinen, P., Rieppo, K., Sirén, M., 2001: Tarkastelu ja johtopäätökset (Discussion and conclusions). In: Kärhä, K., (Ed.). Alternative harvesting systems in mechanised thinning. Final Report of HARKO Project (1999-2001). TTS Institute, Publications 382: 74–82. Kärhä, K., Rönkkö, E., Gumse, S.-I., 2004: Productivity and Cutting Costs of Thinning Harvesters. International Journal of Forest Engineering 15(2): 43–56. Kärhä, K., Keskinen, S., Kallio, T., Liikkanen, R., Lindroos, J., 2006: Ennakkoraivaus osana ensiharvennuspuun korjuuta (Pre-clearance as a part of the harvesting of first-thinning wood). Metsäteho, Raportti 187. Kärhä, K., Poikela, A., Rieppo, K., Imponen, V., Keskinen, S., Vartiamäki, T., 2007: Korjurit ainespuun korjuussa (Harvesting industrial roundwood with harwarders). Metsäteho, Raportti 200. Korhonen, K. T., Ihalainen, A., Heikkinen, J., Henttonen, H., Pitkänen, J., 2007: Suomen metsävarat metsäkeskuksittain 2004–2006 ja metsävarojen kehitys 1996–2006 (Forest resources by forestry centre in Finland in 2004–2006 and the development of forest resources in 1996–2006). Metsätieteen aikakauskirja 2B/2007: 149–213.

Lindroos, O., 2012: Evaluation of technical and organizational approaches for direct loading of logs in mechanized CTL harvesting. Forest Science 58(4): 326–341. Ljungdahl, S. G., 2004: Harwarder in thinning – a comparative study of three work methods. Sveriges lantbruksuniversitet, Students’ Reports 75. Mäki-Simola, E., 2017: The average forest machinery in commercial roundwood production in Finland, 2010–2016. Natural Resources Institute Finland, Unpublished statistics. Manner, J., Jonsson, R., Jönsson, P., Björheden, R., Lundström, H., 2016: Productivity and logging costs of the harwarder prototype Komatsu X19 and a conventional CTL system. Arbetsrapport Från Skogforsk 916. Nordén, B., Lundström, H., Thor, M., 2005: Kombimaskin jämfört med tvåmaskinsystem. Tidsstudier av Ponsse Dual, Ponsse Beaver och Ponsse Buffalo hos SCA Skog AB (Harwarder compared to two-machine system. Time studies on Ponsse Dual, Ponsse Beaver and Ponsse Buffalo with SCA Skog AB). Skogforsk, Arbetsrapport 606. Nordin, L., 2011: Productivity and profitability in early bioenergy-thinnings – A time study of Vimek 608 BioCombi in stands of Lodgepole pine. Sveriges lantbruksuniversitet, Students’ Reports 315. Nuutinen, T., Hirvelä, H., Hynynen, J., Härkönen, K., Hökkä, H., Korhonen, K. T., Salminen, O., 2000: The role of peatlands in Finnish wood production – an analysis based on large-scale forest scenario modelling. Silva Fennica 34(2): 131–153. Ovaskainen, H., 2009: Timber harvester operators’ working technique in first thinning and the importance of cognitive abilities on work productivity. Dissertationes Forestales 79: 62 Palander, T., Kärhä, K., 2016: Development of Computational Model to Predict Rut Formation using GIS for Planning of Wood Harvesting on Drained Peatlands. International Journal of Advanced Engineering Management and Science 2(12): 2040–2057. Rekilä, M., Räsänen, T., 2008: Laajavastuinen yrittäjyys puunhankinnassa (Wide responsibilities of entrepreneurship in wood harvesting business). Metsäteho, Katsaus 33. Rieppo, K., 2003: Vaihtoehtoista korjuutekniikkaa (Alternative harvesting technology). Metsäteho, Raportti 149. Rieppo, K., Pekkola, P., 2001: Korjureiden käyttömahdollisuuksista (Use of harwarders). Metsäteho, Raportti 121. Ringdahl, O., Hellström, T., Lindroos, O., 2012: Potentials of possible machine systems for directly loading logs in cut-toCroat. j. for. eng. 39(2018)1

Productivity and Costs of Harwarder Systems in Industrial Roundwood Thinnings (23–33)

K. Kärhä et al.

length harvesting. Canadian Journal of Forest Research 42(5): 970–985.

cepts through stand-level simulation. International Journal of Forest Engineering 14(2): 31–43.

Sirén, M., 1998: One-grip harvester operation, it’s silvicultural result and possibilities to predict tree damage. Finnish Forest Research Insitute, Research Papers 694.

Väätäinen, K., Asikainen, A., Sikanen, L., Ala-Fossi, A., 2006: The cost effect of forest machine relocations on logging costs in Finland. Forestry Studies 45: 135–141.

Sirén, M., Aaltio, H., 2003: Productivity of costs of thinning harvesters and harvester-forwarders. International Journal of Forest Engineering 14(1): 39–48.

Väätäinen, K., Liiri, H., Asikainen, A., Sikanen, L., Jylhä, P., Rieppo, K., Nuutinen, Y., Ala-Fossi, A., 2007: Korjureiden ja korjuuketjun simulointi ainespuun korjuussa (Simulation of harwarders and two-machine harvesting system in industrial roundwood harvesting). Working Papers of the Finnish Forest Research Institute 48.

Spinelli, R., Magagnotti, N., Di Fulvio, F., Bergström, D., Danelon, M., Alberti, G., 2014: Comparison of Cost Efficiency of Mechanized Fuel Wood Thinning Systems for Hardwood Plantations on Farmland. Croatian Journal of Forest Engineering 35(2): 111–123. Strömgren, A., 1999: Productivity and economy in thinning and final felling for a combined harwarder-forwarder. Sveriges lantbruksuniversitet, Students’ Reports 23. Talbot, B., Nordfjell, T., Suadicani, K., 2003: Assessing the utility of two integrated harvester-forwarder machine con-

Wester, F., Eliasson, L., 2003: Productivity in final felling and thinning for a combined harvester-forwarder (Harwarder). International Journal of Forest Engineering 14(2): 45–51. Zinkevicius, R., Vitunskas, D., 2013: Assessment of operator effect on machine performance of »Ponsse Buffalo Dual« harwarder. In: Proceedings. 12th International Scientific Conference »Engineering for Rural Development«, Latvia University of Agriculture, Jelgava, May 23-24, 73–78 p.

Authors’ addresses: Kalle Kärhä, PhD. * e-mail: kalle.karha@storaenso.com Stora Enso Wood Supply Finland P.O. Box 309 00101 Helsinki FINLAND Asko Poikela e-mail: asko.poikela@metsateho.fi Metsäteho Oy Vernissakatu 1 01300 Vantaa FINLAND

Received: August 13, 2015 Accepted: July 14, 2017 Croat. j. for. eng. 39(2018)1

Prof. Teijo Palander, PhD. e-mail: teijo.s.palander@uef.fi University of Eastern Finland Faculty of Science and Forestry P.O. Box 111 80101 Joensuu FINLAND * Corresponding author

Original scientific paper

Effects of Short-Term Storage Method on Moisture Loss and Weight Change in Beech Timber Arkadiusz Tomczak, Grzegorz Grodziński, Marcin Jakubowski, Tomasz Jelonek, Witold Grzywiński Abstract Timber harvested from fresh-felled trees has high moisture content and relatively high mass. During storage, the timber dries out and its weight decreases with the loss of moisture. The main objective of this study was to determine how the method and conditions of short-storage in summer affect weight changes and moisture loss in beech timber (not used for fuel). The study was carried out in a stand located in the north-eastern part of the range of beech. The age of the studied stand was 47 years. A total of 60 model trees were selected and divided into two groups. In the first group, 30 whole trees (WT) were left in the stand after felling. In the second group, 30 trees (CT) were delimbed and crosscut, and trunk sections (logs) were obtained (2.5 m). The timber (CT) was stored in a pile, and the weight of each log was measured daily. After 14 days, the trees from the first group, which had been left in the stand (WT), were delimbed and the trunks were cut into 2.5 m sections and weighed. It was assumed that timber intended for mechanical processing is stored in the forest for a short period of time, unlike energy wood. Therefore, the period of storage was not longer than two weeks. A more effective method of drying is to leave whole trees after felling, called transpirational drying. The timber stored in a pile (CT) lost moisture more slowly than the timber from trees that had been left whole after felling (WT). Comparing the weights of the logs stored in a pile, on days after harvesting, a statistically significant difference was found only between the first and the last day. It can be concluded that two weeks is the minimum period of storage in a pile (CT) required to obtain a significant degree of weight change and moisture loss. Keywords: transpirational drying, storage conditions, summer, beech stand

1. Introduction During summer, timber is susceptible to the harmful effects of insect pests and discolouring fungi. Its storage close to the place of felling should, therefore, last for as short a time as possible. In Poland, it is accepted that, in the summer season, timber should be transported as soon as possible after felling, and the maximum interim storage time should not exceed two weeks. Timber harvested from fresh-felled trees has high moisture content and relatively high mass. During storage the timber dries out and its weight decreases with the loss of moisture. This is particularly important in the case of relatively large transportation Croat. j. for. eng. 39(2018)1

distances. The greater weight (moisture) of the timber will cause higher costs and environmental load (CO2 emissions) (Acuna et al. 2012, Busenius et al. 2015, Sosa et al. 2015, Kanzian et al. 2016, Bennamoun et al. 2017). The initial moisture content (MC) depends on tree species, wood structure and season (Wullschleger et al. 1996, Ciganas and Raila 2010, Hultnäs et al. 2013). Drying, in turn, is affected by atmospheric conditions, type of material and time of storage (Kokkola 1993, Filbakk et al. 2011, Erber et al. 2014, Visser et al. 2014, Routa et al. 2015, Anisimov et al. 2017, Erber et al. 2017). Beech has diffuse-porous wood and does not have heartwood. Helińska-Raczkowska (1996) anal-

A. Tomczak et al.

Effects of Short-Term Storage Method on Moisture Loss and Weight Change in Beech Timber (35–43)

ysed MC in birch, a species with similar wood structure to beech. She found that MC decreases as the basic wood density increases. Lower-density wood is more porous and is, therefore, able to store more water. This relationship may explain the high green density of timber obtained from the top parts of trunks of fresh-felled trees. In most trees, the basic density or over-dry density of wood in the bottom of the trunk is greater than in the top part. In pine, oak and birch, the weight of 1 m3 of timber with smaller diameters is higher than in the case of timber with larger diameters (Tomczak et al. 2015, Tomczak and Jelonek 2015, Tomczak et al. 2016). Differences also result from changes in the share of the bark in the volume of timber. The bark on the bottom of the trunk of pine, oak and birch is thick and heavily cracked, while it is usually thin at in the top part of the trunk. Therefore, the share of bark in the total volume is large at the bottom of the trunk and small at the top part. Quite differently, beech has a relatively thin bark along the whole trunk. Dudzińska (2004) reported that the proportion of bark in the volume of trunk in young stands was 3.8% and 14.6% in mature stands. The proportion of bark is an important indicator of quality in the production and combustion of both wood chips and pellets. The bark also reduces the drying out of logs. Röser et al. (2011) carried out experiments in different parts of Europe, comparing the natural drying of wood to be used for bioenergy. The study covered several species of conifers and broadleaved trees. It was found that broad-leaved trees dried out more effectively when debarked. For example, the rate of drying of debarked broad-leaved trees was approximately 30 g/kg per month higher than in the case of logs with bark. Generally, many studies have analysed various methods of preparing and storing timber intended for fuel (Nordfjell and Liss 2000, Nurmi and Hillebrand 2007, Pettersson and Nordfjell 2007, Afzal et al. 2010, Filbakk et al. 2011, Krigstin and Wetzel 2016). Another study of interest with regard to moisture loss is that of Saralecos et al. (2014). Their results showed that moisture loss rate increased as the size of the trunk decreased. They also observed a higher moisture loss rate in the top parts of trunks, which were not delimbed after felling, compared to logs that were delimbed. The aim of this study was to determine how the method and conditions of storage in summer affect weight changes and moisture loss in beech timber, not intended to be used for fuel. Our hypothesis is that the form in which wood is stored has an impact on the moisture loss rate (and consequently timber weight loss). More specifically, trunks with intact crown,

which are subject to transpirational drying, are expected to lose more moisture than delimbed trunks. Moisture loss depends largely on the temperature and relative humidity of the air and on rainfall. Therefore, we analysed how the beech timber reacts to changes in atmospheric conditions in summer. In particular, we investigated how the relation of weight to volume changed over a period of two weeks. We assumed that timber intended for mechanical processing is stored in the forest for only a short period of time. Nevertheless, a significant change in weight and moisture content was expected.

2. Materials and methods 2.1 Experimental setup The study was carried out in a stand located in the north-eastern part of the range of beech (N 53° 20’ 56.526”, E 16° 13’ 26.229”). In the region managed by the Świerczyna Forest District, common beech (Fagus sylvatica L.) is the second dominant species after Scots pine (Pinus sylvestris L.). It forms compact stands with high breeding quality, from which valuable timber is harvested. It also often occurs as an additional species and in the second layer of pine stands. The age of the studied stand was 47 years and it covered an area of 5.03 ha. Within the stand, two homogeneous areas could be identified for the selection of model trees. These were selected based on their diameter at breast height (DBH) out of all trees designated for commercial thinning. Each tree was marked with a number for unique identification. The DBH values for all trees were measured using a caliper, to an accuracy of 0.1 cm. A total of 60 model trees were selected and divided into two groups (Table 1). In the first group, 30 whole trees (WT) were left in the stand after felling. In the second group, 30 trees (CT) were delimbed, and trunk sections (logs) were obtained (2.5 m in length, minimum diameter of 7 cm in bark). The timber (CT) was stored in a pile. At the same location, the weather station to measure temperature, humidity and rainfall were set up. Air temperature was measured to an accuracy of 1°C, humidity – 1%, rainfall – 0.01 l/m2. The experiment was carried out between 6th July and 20th July 2015. The weight of each log was measured once every day, between 4 p.m. and 6 p.m. Each log was weighed using a crane scale (Steinberg Systems SBSKW-300AB), with maximum capacity of 300 kg. Weight was measured to an accuracy of 0.1 kg. After 14 days, the trees from the first group, which had been left in the stand (WT), were delimbed, and the trunks were cut into 2.5 m sections and weighed. Croat. j. for. eng. 39(2018)1

Effects of Short-Term Storage Method on Moisture Loss and Weight Change in Beech Timber (35–43)

A. Tomczak et al.

Table 1 Description of model trees and obtained logs Storage method

Number of trees

Mean DBH cm

Mean height m

Number of logs

Volume of logs with bark, m3

Mean diameter of logs Mean diameter of logs at the top end, cm at the butt end, cm

13.1

17.6

116

4.10

10.6

12.6

13.1

17.7

4.18

11.6

13.8

CT – timber storage in the pile, WT – whole trees storage in the stand after felling

A total of 116 (CT) and 98 (WT) logs were obtained (Table 1). The diameters at the ends of the logs were measured to an accuracy of 0.1 cm. The measurements were performed twice: over bark and under bark. The volume of each log was determined by Smalian’s formula: (1) V = ((d1 + d2) / 2) ´ l Where: V volume; d1 cross-sectional area of the section at the butt end d2 cross-sectional area at the top end l log length.

2.2 Data analysis The data were subjected to statistical analysis for the purpose of comparison of weight changes and moisture loss. In the first step, the Lilliefors test for normality was carried out. The result led to the rejection of the hypothesis that the data was normally distributed. For this reason, the data between independent groups were compared by means of the non-parametric Kolmogorov–Smirnov test (weight between CT and WT). Comparison of data between dependent groups was performed using the Wilcoxon test (MC between start and end of the experiment). For comparing more than two groups, the Kruskal–Wallis test was used. If the null hypothesis was rejected, a post hoc Dunn’s multiple comparison test of mean ranks was performed for all samples. A correlation matrix was developed for the studied features, using Spearman’s rho. A significance level of α=0.05 was employed. All calculations were performed using the Statistica 12PL application (StatSoft Inc.).

3. Results 3.1 Storage conditions According to our measurements, the mean air temperature in the place and time of storage was 18° C, which is similar to the mean summer temperatures recorded by meteorological stations in north-west Poland in July 2015 and in previous years (Wójcik and Croat. j. for. eng. 39(2018)1

Fig. 1 Measurement of temperature, air humidity and precipitation between 6th and 20th of July 2015

Miętus 2014, Owczarek and Filipiak 2016). Rainfall occurred on the 3rd, 4th, 5th, 8th, 9th and 14th day of storage and analogous fluctuation in the air humidity were observed (Fig. 1). During the experiment (14 days), the total precipitation amounted to 22.5 mm. For comparison, in July, the average monthly precipitation in the region was 71.5 mm; Gdańsk meteorological station, eastern part of the region, data from 1880–2008 (Filipiak 2001) and 72.3 mm; meteorological station Szczecin, western part of the region; data from 1861 to 1999 (Kirschenstein 2007). Average air humidity was 69.1%, while it was 76.0% on days with precipitation.

3.2 Weight change and moisture loss Fourteen days after harvesting (14 DAH), the weight of the WT group was lower than CT. The Kolmogorov–Smirnov test showed a significant difference between WT and CT (p-value <0.025). The timber

A. Tomczak et al.

Effects of Short-Term Storage Method on Moisture Loss and Weight Change in Beech Timber (35–43)

Fig. 2 Differences between weight (1 m3) of logs obtained from fresh-felled trees (CT 1 DAH), stored in the pile (CT 14 DAH) and whole trees stored in the stand (WT 14 DAH) stored in a pile (CT) lost moisture more slowly than the timber from trees that had been left intact after felling (WT). The average mass of 1 m3 of logs from fresh-felled trees (CT) was 942 kg. During the storage

period, the mass of 1 m3 of CT logs decreased to 890 kg and that of WT logs to 852 kg. The average green density of CT at 1 DAH was significantly higher than CT at 14 DAH (p-value <0.01). Similarly, the green density of CT at 1 DAH (control) was significantly higher than that of WT at 14 DAH (p-value <0.005) (Fig. 2). The weight of the logs stored in a pile (CT) decreased by 4 kg per m3 and day, on average. The largest reduction was observed between 1 and 2 DAH. Generally, large weight loss was observed at the start of the storage period and following rainfall. After the first 24 hours of storage, the weight of 1 m3 of logs dropped by approximately 10 kg. After rainfall, for example at 6 and 10 DAH, the weight dropped by approximately 5–6 kg per m3. The largest amount of rainfall was recorded at 8 DAH. On that day and the next, the weight increased. The increase at 8 DAH was very small, but at 9 DAH the weight of timber increased by 0.7 kg per 1 m3 (Table 2). Comparing the green density of the logs on particular days of storage, a significant difference was observed only between 1 DAH and 14 DAH (p-value = 0.0054). The fact that the median values were larger than the arithmetic means shows that there were more logs with higher green density, that is, with smaller volume and weight (Table 2). Logs with smaller volume, and hence smaller weight, exhibit

Table 2 Green density and weight changes per 1 m3 logs stored in the pile (CT) (n=116) – grey marked lines present days with rain DAH

Green density kg/m3

Min

Max

Median

Change kg/m3*

Change %*

Change kg/m3**

Change %**

942

146

15.5

620

1315

977

–

932

143

15.3

615

1285

967

–9.86

–1.05

–9.9

–1.0

925

141

15.2

611

1262

961

–6.86

–0.74

–16.7

–1.8

921

140

15.2

609

1255

959

–3.79

–0.41

–20.5

–2.2

918

139

15.2

607

1251

955

–2.91

–0.32

–23.4

–2.5

913

139

15.2

604

1247

949

–5.15

–0.56

–28.6

–3.0

909

138

15.2

602

1239

945

–4.49

–0.49

–33.0

–3.5

909

138

15.2

600

1239

944

+0.05

+0.01

–33.0

–3.5

910

138

15.2

600

1239

945

+0.72

+0.08

–32.3

–3.4

904

137

15.2

598

1232

940

–5.68

–0.62

–38.0

–4.0

899

136

15.2

594

1224

935

–5.00

–0.55

–43.0

–4.6

896

136

15.1

594

1217

932

–2.95

–0.33

–45.9

–4.9

892

135

15.1

593

1205

927

–4.32

–0.48

–50.2

–5.3

890

134

15.1

591

1205

925

–1.81

–0.20

–52.0

–5.5

*changes compared to the previous day; **changes compared to the first day after harvest

Croat. j. for. eng. 39(2018)1

Effects of Short-Term Storage Method on Moisture Loss and Weight Change in Beech Timber (35â&#x20AC;&#x201C;43)

A. Tomczak et al.

Fig. 3 Relationship between the sum of cross-sectional area and log weight depending on time and storage method higher green density. The sections with smaller volume were naturally obtained from the top parts of trees. Considering the variability of wood density (oven dry or basic) along the trunk length (high dry density at the bottom, low dry density at the top part), the moisture content in the upper parts is expected to be higher than at the base. Moisture loss is dependent on many factors. For example, drying out is restricted by bark. In this experiment, the logs were obtained so that the bark was Croat. j. for. eng. 39(2018)1

not damaged. Thus, moisture could leak only through the area of the cross-section (top or bottom). Logs with smaller area of the cross-section lost moisture (weight) more rapidly than those with larger area. For example, in the case of a sum of cross-sectional area at the bottom end of the log of 0.015 m2 (thinner logs), the difference between CT at 1 DAH and CT at 14 DAH was calculated at approximately 1 kg, while in the case of a sum of cross-sectional area of 0.06 m2 (thicker logs), the Âdifference was approximately 2 kg. If it was 1 kg per

A. Tomczak et al.

Effects of Short-Term Storage Method on Moisture Loss and Weight Change in Beech Timber (35â&#x20AC;&#x201C;43)

0.015 m2, it would result in about 67 kg per m2. If it was 2 kg per 0.06 m2, it would result in about 33.3 kg per m2. An even greater weight loss was observed when comparing the control data (CT at 1 DAH) with the data for WT at 14 DAH. Analogous analysis showed that, for logs with the sum of cross-sectional areas at the bottom and top end of 0.06 m2, the difference was around 5 kg. This is about 83.3 kg per m2 (Fig. 3). This is an interesting phenomenon, since the whole trees did not lose moisture through cross-sectional areas. In this case, moisture was lost through transpiration. After a tree is felled, the leaves continue to transpire, but there is no moisture uptake, and so the relative water content in the trunk decreases (Barrs 1968). However, the changes in water content are not uniform in different parts of the trunk.

4. Discussion The trunks of living trees contain a large quantity of water, which is essential for the organism to function. Timber from fresh-felled trees, therefore, has a high moisture content and high green density. Moisture loss leads to changes in the weight and green density of wood. The rate of change in moisture depends on many factors. Moisture loss is largely dependent on the atmospheric conditions at the place and time of storage (Persson et al. 2002, HultnĂ¤s et al. 2013, Erber et al. 2014, Routa et al. 2015, Erber et al. 2016, Anisimov et al. 2017). Particularly significant is the air humidity, which increases at times of rainfall (Afzal et al. 2010). Wood is a hygroscopic material, capable of absorbing and releasing water in the form of vapour. In heavy rainfall, moisture loss stops completely (the weight remains unchanged), or the weight of logs may increase, as occurred on days eight and nine of the experiment. In spite of the rainfall and periods of elevated air humidity, the weight of both the logs stored in a pile and those stored in the form of whole trees showed a decrease. There is only a low level of light (semi-darkness) inside a beech stand, even in summer, and particularly in younger stands. Such conditions reduce the drying of timber, and it would certainly affect the final results. However, such conditions are common, and hence reflect the actual potential of moisture loss during a short period of storage. Garret (1985) evaluated the average moisture loss for many species of conifers and broad-leaved trees over a wide range of diameters and reported the decrease of 4% of the moisture content within two weeks in the summer period. In the present study, the logs stored in a pile (CT) lost approximately 6% of their mass, while those left in the stand (WT) lost ap-

proximately 10%. The differences were statistically significant (p-value <0.05). It can be concluded that a more effective method of drying is to leave whole trees after felling. This phenomenon, called transpirational drying, is a basic method of drying out timber and method for increasing the energy content of woody feedstocks (Cutshall et al. 2013, Civitarese et al. 2015). A review of a number of studies on transpirational drying was published by Stokes et al. (1993). They show that drying by this route takes place especially over the first two weeks. Comparing the weights of the trunk sections stored in a pile on days after harvesting, a statistically significant difference was found only between the first and the last day. It can be concluded that two weeks is the minimum period of storage required to obtain a significant degree of weight change and moisture loss. Nonetheless, the specific atmospheric conditions during the period of our experiment must be taken into account. Beech is diffuse-porous wood. The weight differences obtained between the first and last day of storage were statistically significant. A similar result was obtained by Patterson and Post (1980) for paper birch, another diffuse-porous species (Betula papyrifera Marsh). However, they found no significant difference for northern red oak (Quercus rubra L.), a ringporous species. The authors noted an increasing gradient in the moisture content from bole to crown. Similarly, in our experiment, logs with smaller volume and weight were found to have higher green density. These logs were obtained from the top part of trees, where wood is less dense than at the base of the trunk. Low-density wood is more porous, and therefore has a greater capacity to hold water, which results in a higher moisture content and in a higher green density in this part of the trunk. In our study, logs cut from the top parts of trees lost weight (moisture) more rapidly than those with larger diameters (taken from the bottom of the trunk). A review of the literature shows that many authors have segregated logs on the basis of diameter (Garret 1985, Stokes et al. 1987, Spinelli et al. 2011, Assirelli et al. 2013, Saralecos et al. 2014), as this is one of the factors affecting drying (Kokkola 1993, Visser et al. 2014, Tomczak et al. 2016, Anisimov et al. 2017) or not affecting drying (Manzone 2015). Logs of smaller diameter have smaller cross-sectional areas, through which moisture may be lost. This is an interesting phenomenon, since the undelimbed beech trees did not take up water after felling, but continued to lose it in the same way as living trees, through evaporation from the surface of the leaves. The evaporated water is replaced by water stored in the wood, causing a rapid outflow of moisture. Croat. j. for. eng. 39(2018)1

Effects of Short-Term Storage Method on Moisture Loss and Weight Change in Beech Timber (35–43)

The drying of wood prior to use has many advantages. These include lower transport costs, greater energy value of fuel, and increased efficiency of heating systems. In Poland, transport costs are becoming relatively higher, as timber harvesting is increasing. In the past decade, the total quantity of timber harvested has increased by several million cubic metres. There is also social pressure to use natural and renewable energy sources. The development of principles and methods for the storage of timber will eventually lead to a reduction in economic losses and increased efficiency in the supply chain of bioenergy (Erber et al. 2016, Krigstin and Wetzel 2016, Bennamoun et al. 2017).

5. Conclusions To leave trees intact in the stand after felling is a more effective method of drying, known as transpirational drying. The timber stored in a pile (CT) lost moisture more slowly than the timber from trees that had been left whole after felling (WT). Comparing the weights of the logs stored in a pile on days after harvesting, a statistically significant difference was only found between the first and the last day. It can be concluded that two weeks is the minimum period of storage required to obtain a significant degree of weight change and moisture loss.

Acknowledgments This research was funded by the Polish Ministry of Science and Higher Education through subsidies for the Faculty of Forestry Poznań University of Life Sciences. The authors would like to thank the staff of the Świerczyna Forest District for their help with the experiment and to an anonymous reviewer for commenting and improving scientific quality of this paper.

A. Tomczak et al.

Assirelli, A., Civitarese, V., Fanigliulo, R., Pari, L., Pochi, D., Santangelo, E., Spinelli, R., 2013: Effect of piece size and tree part on chipper performance. Biomass and Bioenergy 54: 77–82. Barrs, H. D., 1968: Determination of water deficits in plant tissue. In: Water deficits and plant growth, ed. T. T. Kozlowski, New York, Academic Press, V 1: 235–368. Bennamount, L., Afzal, M. T., Chauhan, S., 2017: Assessment of moisture effect in simulating forest biomass supply chain strategy: case study on New Brunswick, Canada. Croatian Journal of Forest Engineering 38(1): 19–31. Busenius, M., Engler, B., Smaltschinski, T., Opferkuch, M., 2015: Consequences of increasing payloads on carbon emissions – an example from the Bavaria State Forest Enterprise (BaySF). Forestry Letters 108(8): 7–14. Ciganas, N., Raila, A., 2010: Analysis of heating value variations in stored wood. Proceedings of the Conference Engineering for rural development, 27–28 May, University of Agriculture, Jelgava, Latvia, 186–191. Civitarese, V., Spinelli, R., Barontini, M., Gallucci, F., Santangelo, E., Acampora, A., Scarfone, A., Del Giudice, A., Pari, L., 2015: Open-air drying of cut and windrowed shortrotation poplar stems. Bioenergy Research 8(4): 1614–1620. Cutshall, J., Greene, J. B., Dale, W., Baker, S. A., 2013: Transpirational drying effects on energy and ash content from whole-tree southern pine plantation chipping operations. Southern Journal of Applied Forestry 37(3): 133–139. Dong-Wook, K., Murphy, G., 2013: Forecasting air-drying rates of small Douglas-fir and hybrid poplar stacked logs in Oregon, USA. International Journal of Forest Engineering 24(2): 137–141. Dudzińska, M., 2004: Percent bark volume in Carpathian beech stands. Sylwan 148(3): 3–13. Erber, G., Holzleitner, F., Kastner, M., Stampfer, K., 2017: Impact of different time interval bases on the accuracy of meteorological data based drying models for oak (Quercus L.) logs stored in piles for energy purposes. Croatian Journal of Forest Engineering 38(1): 1–9.

6. References

Erber, G., Kanzian, C., Stampfer, K., 2016: Modelling natural drying of European beech (Fagus sylvatica L.) logs for energy based on meteorological data. Scandinavian Journal of Forest Research 31(3): 294–301.

Acuna, M., Anttila, P., Sikanen, L., Prinz, R., Asikainen, A., 2012: Predicting and controlling moisture content to optimise forest biomass logistics. Croatian Journal of Forest Engineering 33(2): 225–238.

Erber, G., Routa, J., Kolström, M., Kanzian, C., Sikanen, L., Stampfer, K., 2014: Comparing two different approaches in modeling small diameter energy wood drying in logwood piles. Croatian Journal of Forest Engineering 35(1): 15–22.

Afzal, M., Bedane, A., Sokhasanj, S., Mahmood, W., 2010: Storage of comminuted and un-comminuted forest biomass and its effect on fuel quality. BioResources 5(1): 55–69.

Filbakk, T., Høibø, O. A., Dibdiakova, J., Nurmi, J., 2011: Modelling moisture content and dry matter loss during storage of logging residues for energy. Scandinavian Journal of Forest Research 26(3): 267–277.

Anisimov, P., Onuchin, E., Vishnievskaja, M., 2017: Modeling pine and birch whole tree drying in bunches in the cutting area. Croatian Journal of Forest Engineering 38(1): 11–17. Croat. j. for. eng. 39(2018)1

Filipiak, J., 2011: Variability of precipitation totals in Gadńsk in the period 1880–2008. Prace i Studia Geograficzne 47: 119–128.

A. Tomczak et al.

Effects of Short-Term Storage Method on Moisture Loss and Weight Change in Beech Timber (35–43)

Garret, L. D., 1985: Delayed processing of felled trees to reduce moisture content. Forest Products Journal 35(3): 55–59.

treatments during seasonal storage of wood for bioenergy in different European locations. Biomass and Bioenergy 35(10): 4238–4247.

Helińska-Raczkowska L., 1996: Zmienność wilgotności i gęstości drewna w świeżo ściętych pniach brzozy (Betula pendula Roth.). Folia Forestalia Polonica, Series B, Issue 27: 23–30.

Saralecos, J. D., Keefe, R. F., Tinkham, W. T., Brooks, R. H., Smith, A., Johnson, L. R., 2014: Effects of harvesting systems and bole moisture loss on weight scaling of Douglas-Fir saw logs (Pseudotsuga menziesii var. glauca Franco). Forests 5(9): 2289–2306.

Hultnäs, M., Nylinder, M., Ågren, A., 2013: Predicting the green density as a means to achieve the volume of Norway spruce. Scandinavian Journal of Forest Research 28(3): 257– 265. Kanzian, C., Kühmaier, M., Erber, G., 2016: Effects of moisture content on supply costs and CO2 emissions for an optimized energy wood supply network. Croatian Journal of Forest Engineering 37(1): 51–60. Kirschenstein, M., 2007: Long-term changes in the participation in Szczecin. In: Fluctuations of climate in different spatial and temporal scales, ed. K. Piotrowicz and R. Twardosz, Institute of Geography and Spatial Management, Jagiellonian University in Cracow, Poland, 375–382. Kokkola, J., 1993: Drying of pulpwood in northern Finland. Silva Fennica 27(4): 283–293. Krigstin, S., Wetzel, S., 2016: A review of mechanisms responsible for changes to stored woody biomass fuels. Fuel 175: 75–86. Nordfjell, T., Liss, J. E., 2000: Compressing and drying of bunched trees from a commercial thinning. Scandinavian Journal of Forest Research 15(2): 284–290. Nurmi, J., Hillebrand, K., 2007: The characteristics of wholetree fuel stocks from silvicultural cleanings and thinnings. Biomass and Bioenergy 31(6): 381–392. Manzone, M., 2015: Energy and moisture losses during poplar and black locust logwood storage. Fuel Processing Technology 138: 194–201. Owczarek, M., Filipiak, J., 2016: Contemporary changes of thermal conditions in Poland, 1951–2015. Bulletin of Geography. Physical Geography Series 10(1): 31–50. Patterson III, W. A., Post, I. L., 1980: Delayed bucking and bolewood moisture content. Journal of Forestry 78(7): 407– 408. Persson, E., Filipsson, J., Elowson, T., 2001: Roadside storage of Norway spruce. Paperi ja Puu – Paper and Timber 84(3): 174–178. Pettersson, M., Nordfjell, T., 2007: Fuel quality changes during seasonal storage of compacted logging residues and young trees. Biomass and Bioenergy 31(11): 782–792. Routa, J., Kolström, M., Ruotsalainen, J., Sikanen, L., 2015: Validation of prediction models for estimating the moisture content of small diameter stem wood. Croatian Journal of Forest Engineering 36(2): 283–291. Röser, D., Mola-Yudego, B., Sikanen, L., Prinz, R., Gritten, D., Emer, B., Väätäinen, K., Erkkilä, A., 2011: Natural drying

Sosa, A., Acuna, M., McDonnell, K., Devlin, G., 2015: Managing the moisture content of wood biomass for the optimisation of Ireland’s transport supply strategy to bioenergy markets and competing industries. Energy 86: 354–368. Spinelli, R., Magagnotti, N., Paletto, G., Preti, C., 2011: Determining the impact of some wood characteristics on the performance of a mobile chipper. Silva Fennica 45(1): 85–95. Stokes, B. J., McDonald, T. P. Kelley, T., 1993: Transpirational drying and costs for transporting woody biomass – a preliminary review. In: Proceedings of IEA/BA Task IX, Activity 6: Transport and Handling,16–25 May, New Brunswick, Aberdeen University, Canada, 76–91. Stokes, B. J., Watson, W. F., Miller, D. E., 1987: Transpirational drying of energy wood. ASAE Paper No. 87-1530, American Society of Agricultural Engineers, St. Joseph, MI, USA, 13 p. Tomczak, A., Jakubowski, M., Jelonek, T., Wąsik, R., Grzywiński, W., 2016: Mass and density of pine pulpwood harvested in selected stands from the Forest Experimental Station in Murowana Goślina. Acta Sci. Pol. Silv. Colendar. Ratio Ind. Lignar. 15(2): 105–112. Tomczak, A., Jelonek, T., 2015: Mass and density of birch pulpwood harvested from stands in different types of forest habitats. Forestry Letters 108(8): 27–31. Tomczak, A., Jelonek, T., Jakubowski, M., Wąsik, R., Jaszczak, A., 2015: Weight and green density of oak pulpwood harvested from the selected stands of Łąck Forest Inspectorate. Ann. WULS – SGGW, For. And Wood Technol. 91: 172–178. Tomczak, A., Wesołowski, P., Jelonek, T., Jakubowski, M., 2016: Weight loss and green density changes of Scots pine pulpwood harvested and stored during the summer. Sylwan 160(8): 619–626. Wullschleger, S. D., Hanson, P. J., Todd, D. E., 1996: Measuring stem water content in four deciduous hardwoods with a time-domain reflectometer. Tree Physiology 16(10): 809–815. Wójcik, R., Miętus, M., 2014: Some features of long-term variability in air temperature in Poland (1951–2010). Przegląd Geograficzny 86(3): 339–364. Visser, R., Berkett, H., Spinelli, R., 2014: Determining the effect of storage conditions on the natural drying of radiata pine logs for use energy. New Zealand Journal of Forest Science 44(3): 1–8. Croat. j. for. eng. 39(2018)1

Effects of Short-Term Storage Method on Moisture Loss and Weight Change in Beech Timber (35–43)

A. Tomczak et al.

Author’s addresses:

Received: September 12, 2016 Accepted: May 17, 2017 Croat. j. for. eng. 39(2018)1

Arkadiusz Tomczak, PhD. * e-mail: arkadiusz.tomczak@up.poznan.pl Grzegorz Grodziński. Eng. e-mail: grodzinski.grzegorz@gmail.com Marcin Jakubowski, PhD. e-mail: marcin.jakubowski@up.poznan.pl Tomasz Jelonek, PhD. e-mail: tomasz.jelonek@up.poznan.pl Assoc. prof. Witold Grzywiński, PhD. e-mail: witold.grzywinski@up.poznan.pl Department of Forest Utilisation Faculty of Forestry Poznań University of Life Sciences Wojska Polskiego 71A St. 60-625 Poznań POLAND * Corresponding author

Original scientific paper

Planning Forest Road Network in Natural Forest Areas: a Case Study in Northern Bosnia and Herzegovina Vladimir Petković, Igor Potočnik Abstract Natural forests are one of the three types of forest management in terms of origin. These forests are of seed origin and they regenerate naturally. Therefore, natural forests are the most important forest category from the point of view of timber production, as well as its quality and biodiversity. The natural forests accessibility and overall forest accessibility are insufficient for sustainable forest management. This is the reason for dealing with planning of forest roads, actually planning of forest accessibility and designing of forest roads in this forest category. This task requires quantity and quality analysis of the current forest road network, determination of optimal density of forest roads, determination of suitability of forest area for the construction of forest roads and designing of forest roads in the end. Planning of forest roads is carried out at strategic level. Analytic Hierarchy Process (AHP) allows the selection and evaluation of influential factors related to planning of forest roads. The tools of Geographic Information System (GIS) allow a complete spatial and statistical analysis and management of data collected from the forest management plans or data surveyed in the field and obtained by means of »Digital Terrain Model« (DTM) and AHP method. Planning of forest roads will be done in the Management Unit (MU) »Prosara«, located in the northern part of Bosnia and Herzegovina (BIH). The current density of forest roads is 7.3 m/ha in natural forests of this management unit. The optimal density of forest roads should be 17 m/ha. The length of new forest roads designed in the MU »Prosara« is 21 km, and forest accessibility has increased to 13.5 m/ha. Keywords: forest road network, planning, forest road density, AHP, Bosnia and Herzegovina

1. Introduction and research problem Sustainable forest management is based on forest accessibility whose aim is to achieve an optimal density of forest roads. Overall forest accessibility in Bosnia and Herzegovina (BIH) is 10 m/ha (Sokolović and Bajrić 2013, Bureau of Statistics of Republika Srpska (RS) 2015). It is insufficient for normal and intensive forest management in accordance with IUFRO 1995 recommendations. The second problem of forest management is that there are no studies on forest accessibility for each forest management area and forest management unit, and there are no data about optimal density of forest roads for these areas. They contain lists of compartments and length of forest roads that should be designed in the forest management plans. Croat. j. for. eng. 39(2018)1

Therefore, it has been decided to deal with planning of forest accessibility, the actual accessibility of natural forests. Planning of forest accessibility is the first phase in planning of forest roads, while the second phase is designing of forest roads (Potočnik 2004). Planning of forest accessibility includes determination of optimal forest accessibility, which is expressed by an optimal density of forest roads or optimal road spacing. The optimal density of forest roads should ensure more rational, more complete and more successful forest management with minimal impact on the environment (Pentek et al. 2005). Planning of forest accessibility at the strategic level requires slope classification of the terrain, by means of specific harvesting system, analyzing current forest accessibility and

V. Petković and I. Potočnik Planning Forest Road Network in Natural Forest Areas: a Case Study in Northern Bosnia ... (45–56)

a nalyzing timber supply concerning the relief areas (Đuka et al. 2015). In accordance with the above mentioned, Krč and Beguš (2013) asked two questions: where should the new forest roads be constructed, and how many kilometers of forest roads should to be constructed. The answers to these two questions should provide enough information for planning of forest roads at the strategic level, i.e. regional level. Strategic plan of forest management should take into account production of timber, protection of the environment and social function of the forest area. One of its objectives is to determine the location and type of forest roads in the observed forest area (Fannin and Lorbach 2007). Forest roads should be constructed in the areas suitable for the construction of forest roads. These areas are determined by evaluation of the terrain, site conditions and current forest road network i.e. current forest accessibility. Chung and Session (2001) developed forest road design method using genetic algorithm and simulated annealing method for optimization of forest road location. Multi-criteria evaluation was proposed and applied by Pičman (1994), Pentek (2002), Pentek et al. (2004), Lotfalian et al. (2008), Abdi et al. (2009), Sokolović et al. (2009), Samani et al. (2010), Enache et al. (2013), Hribernik (2013), Pellegrini et al. (2013), Hayati et al. (2013), Lepoglavec (2014), Tampekis et al. (2015), Jež (2016). We have decided to observe natural forests because they account for 43% of the total public forest area. Public forests account for 78%, and private for 22% of the total forest area, which is 2.6 million ha or 46% of the total area. It is the most important forest category

from the point of view of timber production and its quality and biodiversity. Natural forests produce 2.6 million m3 of timber from the total amount of 2.9 million m3 per year. Accessibility of natural forests is 11 m/ha (Sokolović and Bajrić 2013, Bureau of Statistics RS, 2015). Natural forests (high forests) are the type of forest management in terms of origin. They regenerate naturally from seeds (Bunuševac 1951, Govedar 2011); further to the above, it can be concluded that natural forests represent a bank of seeds of domestic forest trees species. The research objectives of this paper are related to planning of forest road network in natural forests, as follows: Þ quantity and quality analysis of current primary forest accessibility, where density of forest roads and skidding distance will be analyzed Þ it has been assumed that current density of forest roads in natural forests is less than optimal density of forest roads, and insufficient for normal and intensive forest management. The average current real skidding distance and its costs are higher than the targeted average real skidding distance and its costs, and the optimal forest roads density could be achieved only over several decades Þ designing of forest roads – optimal length of forest roads should be distributed in the management unit area, which is insufficiently accessible and suitable for the construction of forest roads. Forest roads should be designed in accordance with applicable laws and regulations.

Fig. 1 Location of research area in BIH

Croat. j. for. eng. 39(2018)1

Planning Forest Road Network in Natural Forest Areas: a Case Study in Northern Bosnia ... (45–56) V. Petković and I. Potočnik

2. Materials and Methods

2.3 Determining optimal density of forest roads

2.1 Research area The research was done in the natural forests of the Management Unit (the MU) »Prosara«, located in the mountain Prosara, in the northern part of BIH between 17°00’00˝ E to 17°07’30˝ E, and 45°13’00˝ N to 45°17’00˝ N (Fig. 1). It is managed by the Forest Administration »Gradiška«, Forest Office »Gornji Podgradci«. The forests and forest lands cover 3980 ha, and natural forests cover 3470 ha or 87% of the overall forest area in the MU. Forest vegetation grows on silicate soil, which is deep and consists of mixed forests of sessile oak with hornbeam (Querco-Carpinetum illyiricum), and above them, there are pure beech forests (Fagetum illyricum). Those are typical forests of the hilly area of the northern and central parts of the country (Stupar and Čarni 2017). The total length of forest roads in the MU is 27.83 km, and in the natural forests this length is 25.25 km. Accessibility to the overall and natural forests is equal in this MU, and it amounts to 7.3 m/ha (Public Forestry Company (PFC) »Šume RS« 2012).

2.2 Analysis of current primary forest accessibility The first task in the research is to collect the necessary data. Data are collected from the forest management documentation such as the Forest Management Plan or operational projects. These data are the following: list of compartments and their area, management classes, growing stock, growing increment, cut volume, etc. Forest road network is recorded by GARMIN GPSMap 62st. These data are necessary for spatial and statistical analysis with ArcGIS 10 software. The results of these analyses are the current length of forest roads, current forest accessibility, and skidding distance. Determination of the current forest accessibility is based on the influence of forest roads according to the rules established by Pičman (2007). An average geometrical skidding distance is determined by means of the method measuring distance from the center of gravity of the compartments to the nearest forest road. Real average skidding distance is obtained by multiplying the geometrical skidding distance and skidding factor (Eq. 1). Skidding factor depends on the relief area, slope and presence of obstacles on the surface.

S= SdG × kG dS

(1)

Where: SdS real average current skidding distance, m SdG geometrical average current skidding distance, m kG skidding factor (Pentek et al. 2005). Croat. j. for. eng. 39(2018)1

After the analysis of the current forest accessibility, planning of forest accessibility i.e. determination of optimal density of forest roads in natural forests of the MU has been carried out. Optimal density of forest roads is calculated by FAO (1998) formula (Eq. 2). c=

100,000 × hV (2) 4R

Where: c optimal density of forest roads, m/ha 100,000 constant R costs of construction and maintenance of forest roads in the period of depreciation, €/km h costs of timber skidding for 100 m of skidder LKT 81 productivity, €/m3 V discounted allowable cut volume of timber in lifespan of forest road, m3/ha. Skidding costs consist of skid trails construction costs and unit skidding costs. The skid trails construction costs depend on the costs of excavation, which are prescribed in the Description and Unit Costs of Forest Road Construction Works (2009) for certain type of soil and volume of excavation. They are divided with discounted allowable cut volume of roundwood in lifespan of forest road. Discounted volume of roundwood is calculated on the basis of an average allowable cut volume of roundwood in the natural forests, which is reduced due to delay in forest road lifespan, which is 50 years (Naghdi and Limaei 2009, Hribernik 2013) and capital interest of 4.7% (Republika Srpska Investment-Development Bank (IRBRS), 2017). The unit skidding costs are calculated as the ratio of costs of a skidder and its productivity on 100 m of skidding distance. The total roading costs consist of the costs of forest road construction and the costs of their maintenance. The costs of forest road construction are calculated according to prices of work prescribed in the Description and Unit Costs of Forest Road Construction Works (2009) and the amount of work. They consist of the costs of excavation and filling, pavement construction and finishing, construction of drainage and planning of forest road project (Naghdi and Limaei 2009). The amount of excavation is determined on the basis of the carriageway depth and terrain inclination (0–75%). The carriageway depth is the depth where all truck wheels are on hard surface (Potočnik 2005). After that, the annual maintenance costs are calculated as the sum of costs of profiling of forest road pavement by the grader and cleaning of drainage, as well as the

V. Petković and I. Potočnik Planning Forest Road Network in Natural Forest Areas: a Case Study in Northern Bosnia ... (45–56)

costs of repairing the pavement in 5 to 10 years. Total maintenance costs per km are determined on the basis of annual maintenance costs, which depend on capital interest of 4.7%, and period of forest road depreciation (50 years). The total roading costs are the sum of construction costs and maintenance costs of forest road per km. The annual construction costs of forest roads are calculated on the basis of depreciation and capital interest of forest roads. The annual roading costs are the sum of annual construction and maintenance costs. On the basis of optimal density of forest roads, the targeted real average skidding distance is calculated by formula of Rebula (1980) (Eq. 3).

Sdos =

kS c

× 10,000

(3)

Where: SdOS targeted real average skidding distance, m kS theoretical total skidding factor c optimal density of forest roads, m/ha. Optimal forest accessibility is observed from the point of view of relative forest accessibility and efficiency coefficient of the forest road network, except optimal density of forest roads. Relative forest accessibility (OR) is the ratio of the accessible forest area and the total forest area (Eq. 4).

O= R

× 100,%

(4)

Where: PO accessible forest area, ha PU total forest area ha. The accessible forest area is obtained by creating a buffer zone around the forest roads, whose width is equal to the targeted double geometrical average skidding distance. The areas outside the buffer zone are insufficiently accessible (Pentek et al. 2005). The efficiency coefficient of the forest road network (kU) is the ratio of a forest area accessible by a single forest road and multiple forest roads (Eq. 5).

 PN kU =  1 − PO 

  × 100,% (5) 

Where: PN surface of inefficiency of bordered areas (multiple accessible area), ha PO accessible area for the chosen targeted double geometrical skidding distance (single accessible area), ha (Pentek et al. 2005). When the planning phase of forest accessibility is completed, the phase of designing of forest roads starts. This economic length of forest roads per ha

should be laid in the area of natural forests, which are insufficiently accessible and suitable for construction of forest roads.

2.4 Designing of forest roads Multi-criteria analysis of relief and site factors will determine the areas suitable for construction of forest roads. The selected factors are the following: slope of terrain, depth of soil, normal and current growing stock. Terrain slope is one of the most important factors during designing of the forest road in terms of suitability of the forest road and terrain after construction. The raster of terrain slope will be obtained from DTM with resolution 5×5 m. Soil depth is easy measurable in the field and, in combination with the cross slope, it can have great influence on forest road construction. Shallow soils on mildly sloped terrains and deep soils even on steeper terrains demand less construction costs, that is, soil works (Sokolović et al. 2009). Normal and current growing stock are indicators of potential and current site production. The rasters of soil depth, normal and current growing stock are obtained from the attribute table with Arctool Feature to raster. Original values of these factors are standardized by Eq. 6 to values 0 to 1.

Xi =

( Ri − Rmin )

( Rmax − Rmin )

× Xmax (6)

Where: xi standardized value Ri basic value Rmin lower value of basic scale Rmax upper value of basic scale Xmax upper value of standardized scale. The standardized rasters are entered into Arctool Weighted Sum and multiplied by the weights of influential factors, which are summed. The results of this analysis are two maps of suitability. The first map is the map of optimal suitability obtained by summing the terrain slope, depth of soil and the normal growing stock rasters. The second map is the map of the actual suitability of the natural forest area for the construction of forest roads and it is the result of summing the terrain slope, depth of soil and the current growing stock rasters. These rasters are divided into three classes: 0–0.33 unsuitable area, 0.33–0.66 medium suitable area and 0.66–1 highly suitable area. The weights of influential factors are obtained by AHP method based on the opinion and experience of fourteen forest engineering scientists. This method has been developed by Tomas Saaty as a tool of decisionmaking analysis. It is created as an auxiliary tool of Croat. j. for. eng. 39(2018)1

Planning Forest Road Network in Natural Forest Areas: a Case Study in Northern Bosnia ... (45–56) V. Petković and I. Potočnik

Fig. 2 Process of sending of planned forest roads into GPS device engineers in multi-criteria decision-making. Decisionmaking group consists of engineers or scientists, who evaluate the importance of the criteria for a given problem. Their average grades are standardized by Eq. 6, and after that, a pair-wise comparison matrix is created, based on which weights of the criteria are calculated. Evaluation of importance of the criteria for the problem is a result of experience and opinion of decision-makers, and it is necessary to calculate consistency ratio CR in order to determine the accuracy of the evaluation. CR is the ratio of consistency index CI and RI random index. The AHP can be implemented in three simple consecutive steps: evaluation of importance of the criteria, computing the vector of criteria weights and consistency calculation (Saaty1980, Saaty 2008, Pellegrini 2013, Hribernik 2013, Lepoglavec 2014, Jež 2016). During designing of forest roads, protection of soil and protection of quality and stability of forest roads should be taken into account. Forest roads will be designed in the area whose slope is up to 60% (Keller and Sherar 2003), and the maximum longitudinal slope of forest roads is up to 8% from the poibt of view of risk of erosion and type of soil (PFC »Šume RS« 2002b, Potočnik 2004). It is the most important constructive element of forest roads in terms of maintenance of forest roads. Designing of forest roads is carried out by plotting of a zero line on contour map. Plotting of a zero line is carried out by using ArcGIS Toolbar Editor, Edge Snapping tool Length. The raster of natural forest area with the plotted zero lines was imported in software GlobalMapper and it was saved in kmz or kml format. The raster was imported into BaseCamp software and later it was sent to GPS device (Fig. 2) to be set up in the field.

3. Results and Discussion 3.1 Analysis of current primary forest accessibility The total length of the forest roads surveyed is 34.93 km, but 31.02 km or 88% were taken into account Croat. j. for. eng. 39(2018)1

for calculating the overall forest accessibility by Pičman’s method (2007), where the location of the forest road regarding the forest area is taken into consideration. According to these data, the density of forest roads is 7.8 m/ha in Prosara, and 9.04 m/ha in its natural forests. The natural forests accessibility is 10 m/ha in BIH (Sokolović and Bajrić 2013, Bureau of Statistics of RS 2015). The densities of forest roads in the natural forests of the MU and in the natural forests of BIH do not allow for intensive forest management (IUFRO 1995). The overall forest accessibility in BIH is 2 to 6 times lower than forest accessibility in other countries in the region. For example, the forest accessibility is 7 to 30 m/ha in Croatia (Pentek et al. 2007), and around 25 m/ha in Slovenia (Krč and Beguš 2013). An average density of forest roads is 45 m/ha in Austria (Ghaffarian et al. 2009). Forest accessibility depends on the relief of the region where the forest area is located. According to that, FAO (1998) recommended the density of 7 to 10 m/ha of forest roads in hilly relief region. Minimum required current density of forest roads is 7 m/ha in lowland region and 12 m/ha in hilly region of Croatia. Current density of forest roads is 8.85 m/ha in lowland- relief region and 11.26 m/ha in hilly relief region. Planned density of forest roads should be 15 m/ha in low-land region and 25 m/ha in hilly region up to 2020 (Pentek et al. 2011). According to Bertović (1999), the elevation raster was obtained by spatial analysis of DTM and it is classified into two relief regions, lowland and hilly relief regions. The first region accounts for 30%, and the second for 60% of the MU. According to that, current density of forest roads is sufficient. However, current density of forest roads in the natural forests of the MU »Prosara« should be increased when taking into account the quality of forest management and especially the results and recommendations of the above mentioned research. Current average geometrical skidding distance is 510 m. It was determined by the method of distance from the center of gravity of compartments to the nearest forest roads (Fig. 3). Skidding factor for this

V. Petković and I. Potočnik Planning Forest Road Network in Natural Forest Areas: a Case Study in Northern Bosnia ... (45–56)

Fig. 3 Determination of average skidding distance management unit is 1.5 (Petković et al. 2015) and real average skidding distance is 765 m. Skidding costs, according to the cost of the skidder LKT 81 per work day and its productivity for the real average skidding distance, are 10.62 €/m3.

3.2 Determining of the optimal density of forest roads Forest accessibility will be expressed by density of forest roads, which depends on transportation costs and allowable cut volume of roundwood. Transportation costs consist of skidding and roading costs and their calculation was stated above. The skidding costs consist of construction costs and unit costs of skidding. Construction costs of skid trails were obtained by multiplying the costs of excavation and the amount of excavation. The unit costs of excavation are 2.44 €/m3 (PFC »Šume RS« 2009), considering the type of soil that belongs to the terrain category 3 (Jeličić 1983). The volume of excavation on terrain category 3 is 0.5 m3/m (Jeličić 1979). An average allowable cut volume of roundwood is 39.54 m3/ha in the

natural forests and according to that, discounted roundwood volume is 92.1 m3/ha. Costs of construction of skid trails are 0.01 €/m3. The unit skidding costs, taking into consideration the cost of the skidder LKT 81 of 324.52 €/WD1, as regulated by the Decision of PFC »Šume RS« No. 01-2166/12 from 2012, its productivity for the skidding distance of 100 m, the average tree volume of 0.5 m3 and winching distance of 30 m, are 58.58 m3/WD (PFC »Šume RS« 2002a) or 5.54 €/m3. The total skidding costs are 5.55 €/m3. The skidding costs are 8.93 $/m3 or 8.24 €/m3 for Timberjack skidder and 8 $/m3 or 7.38 €/m3 for Clark skidder for the skidding distance of 500 m or 9 m/ha of forest road density (Naghdi and Limaei 2009). According to the same skidding distance and unit costs of skidding, the costs of skidding of LKT 81 is 8.82 €/m3. When comparing the skidding costs, it can be seen that the Clark skidder is more suitable for skidding than the other two skidders.

WD-work day-8 hours work shift

Croat. j. for. eng. 39(2018)1

Planning Forest Road Network in Natural Forest Areas: a Case Study in Northern Bosnia ... (45–56) V. Petković and I. Potočnik

The calculation of the total roading costs was described above, and according to the unit costs of the construction of forest roads, and prices of work, which are prescribed in the Description and Unit Costs of Forest Road Construction Works (2009), and the amount of work, these costs are 31,737 €/km. Enache et al. (2011) have calculated that the overall average road construction costs are 33.52 €/m or 33,520 €/km. The annual maintenance costs are 720 €/km, and total maintenance costs are 13,248 €/km, depending on the lifespan of forest road and capital interest. According to that, the total roading costs are 44,985 €/km. The annual construction costs of forest road are 1725 €/km, and hence, the total annual roading costs are 2445 €/km. The annual roading costs were 3647 $/km or 3428 €/km in 2009 (Naghdi and Limaei 2009). Building of forest road retaining walls was not planned, and if the possible differences in the cost of fuel and labor were added here, the difference between the roading or road construction costs could be justified. Optimal density of forest roads in natural forests, calculated based on transportation costs and cut vol-

ume of timber by FAO (1998) formula, is 17 m/ha. According to the given density, the targeted real average skidding distance should be 353 m, which is shorter by 412 m than the current skidding distance. The skidding costs for the real average skidding distance are 8.44 €/m3, or 2.18 €/m3 lower than the current skidding costs. The average targeted geometrical skidding distance will be 235 m if the skidding factor is 1.5. This optimal density allows normal forest management in accordance with IUFRO 1995. The average optimal density is higher than the current density of forest roads by 10 m/ha. Achieving optimum density of forest roads leads to the reduction of skidding distance and skidding costs, and it is the aim of planning of forest accessibility (Jež 2016). The accessible natural forest area is 1974.36 ha for the targeted double average geometrical skidding distance (235 m). The relative natural forest accessibility is 57%, and it means that natural forests of this MU are poorly accessible according to Pentek et al. (2005) (Fig. 4).

Fig. 4 Efficiency coefficient of forest road network Croat. j. for. eng. 39(2018)1

V. Petković and I. Potočnik Planning Forest Road Network in Natural Forest Areas: a Case Study in Northern Bosnia ... (45–56)

Multiple accessible natural forest area is PN = 931.26 ha, single accessible natural forest area is PO = 1974.36 ha, and the efficiency coefficient of the forest road network is 53% (Fig. 4). However, whether this optimal density of forest roads would be achieved, depends on the terrain and site conditions of a certain area, i.e. the MU. In order to achieve the optimal density of forest roads, it is necessary to design forest roads in the MU.

3.3 Designing of forest roads The forest roads are designed as zero lines on the map of optimal suitability of natural forest areas, which are medium and highly suitable for the construction of forest roads and insufficiently accessible (Fig. 5). The results of AHP analysis are weights for the terrain slope (0.42), for soil depth (0.19) and for growing stock (0.40). The weights were obtained by AHP method in other studies, and they are: 0.2–0.4 for slope, 0.15–0.2 for soil and 0.02–0.1 for timber volume (Abdi et al. 2009, Samani et al. 2010). Lepoglavec (2014) uses the same weights (0.25) for each influential factor, i.e. the ratio between 1 and the number of influential factors. When comparing the weights of influential factor for determining suitability of natural forests area, which were obtained in this study and in other studies, it can be seen that weights of terrain slope and soil are similar, while the weight of timber volume is bigger in our study than in some other studies. The results of spatial analysis show that 24 ha or 1% of the total area of natural forests are unsuitable for construction of forest roads, 2147 ha or 63% of the total area are medium suitable and 1246 ha or 36% of the total area are highly suitable for construction of forest roads (Fig. 4). The aim of designing forest roads is the reduction of skidding distance and skidding costs with the increase of density of forest roads, i.e. the forest accessibility. Enache et al. (2013) developed a decision support tool for evaluating different forest road options before designing a suitable variant based on stakeholder’s preference and multicriteria analysis. The main process refers to locating a new road, productivity assessment, calculation of extraction costs and evaluation of impact on the environment and the utility analysis of forest road alternatives. The multiple attribute utility theory (MAUT) was proven as an appropriate tool for evaluating forest road alternatives for its simplicity and its practicality in the development of decision support tools in the sector of forest engineering. In practice, GIS tools and AHP analysis are valuable tools for determining maintenance activities. This

approach uses pair-wise comparison process, the evaluation of erosion risk and the evaluation of social value of roads based on stakeholders preferences, so as to determine the function of forest road in the analyzed area (Pellegrini et al. 2013). Important criteria for forest road planning were selected in 3 rounds by Delphi method. After that, the selected criteria were used for developing the pairwise comparison questionnaire, which was sent to the participants in order to obtain the weight of each criterion. Their answers were analyzed by the Expert Choice software, and finally the weights of all the criteria were obtained (Hayati et al. 2013). Multi-criteria analysis was also used in the paper of Tampekis et al. (2015), who evaluated the type of criteria, such as intensity and absorption, in order to determine spatial distribution of optimal forest roads density in the forest area. The intensity criteria evaluation means the environmental impacts, which are caused by forest roads in the forest area. The evaluation of absorption criteria refers to the ability of the environment to absorb the impacts caused by forest road construction. This method refers to non-productive forests. Criteria evaluation and selection of alternatives from the point of view of productive, protective and social function of forests are a thread that connects the papers. Keeping in mind soil protection from erosion and the costs of forest road construction, the forest road will be designed in the area with the slope up to 60%. Terrain slope raster is obtained by spatial analysis of DTM, and the slope ranges between 0 and 121% in the natural forests of the MU, while the average slope is 27% or 15°. Statistical analysis showed that the slope of 3342 ha or 97.4% of the total natural forest area of the MU was up to 60%, and the slope of 86.6 ha or 2.6% of the total natural forests area of the MU was over 60%. Therefore, roundwood will have to be extracted by skidder LKT 81. The maximum slope gradient acceptable for skid trails construction varies between 45% and 60% in BIH. The roundwood was produced by motor-manual assortment method, which is the most common harvesting method in BIH. However, skidder works under its capacity in the assortment method (Marčeta 2014). 21 km of zero lines have been plotted (Fig. 5) for new forest roads, and the total length of forest roads, which make the natural forests of the MU »Prosara« accessible, is 46.25 km, while the new density of forest roads is 13.5 m/ha. This new density of forest roads does not allow normal forest management, almost in Croat. j. for. eng. 39(2018)1

Planning Forest Road Network in Natural Forest Areas: a Case Study in Northern Bosnia ... (45–56) V. Petković and I. Potočnik

the same way as the current density of forest roads. It is lower than the calculated economical optimum density of forest roads, and the reason for this is probably the relief. Some of the planned zero lines are connected to public roads, which shows that modern forest management should also take into account the quality of life of inhabitants in the forest area, and this represents a social moment of forest management. The average geometrical skidding distance is 271.5 m, and the real average skidding distance is 407 m. It is 358 m shorter than the real current skidding distance. The skidding costs should be 8.77 €/m3 on the basis of this skidding distance, and it is 1.85 €/m3 lower than the current skidding costs. The construction of 21 km of these new roads should cost 666,480 €, and this was calculated on the basis of the estimated costs of forest road construction, which are 31,737 €/km. The total saving of skidding costs could amount to 253,321 € in 10 years, or 25,332 € per year and this has been calculated on the basis of the saving of skidding

costs as a result of the shortening of skidding distance, which is 1.85 €/m3 and total cut volume of roundwood of 137,216 m3. The total savings of skidding costs could justify 38% of construction costs of 21 km of new forest roads. By these savings of skidding costs, 8 km could be constructed in 10 years, which means that 21 km of forest roads could be constructed in 26 years.

4. Conclusions The results of earlier researches show that GIS, GPS, RoadEng are very useful tools in planning forest accessibility and forest management. Forest accessibility depends on many factors, such as: terrain conditions, site potentials, transportation costs, demands and supply of timber, social demands, etc. Planning of forest roads consists of the planning of forest accessibility and designing of forest roads. Planning of forest accessibility requires determination of optimal density of forest roads or optimal spacing of

Fig. 5 Proposed forest roads in suitable and insufficiently accessible areas of natural forests Croat. j. for. eng. 39(2018)1

V. Petković and I. Potočnik Planning Forest Road Network in Natural Forest Areas: a Case Study in Northern Bosnia ... (45–56)

forest roads. Forest accessibility was considered from the point of view of timber production and transportation costs. The second phase of the planning of forest roads is their designing in the field. The forest roads were represented by zero lines at the strategic level of forest road planning. The zero lines have been laid on the contour map into the areas which are suitable for construction of forest roads with horizontal and vertical alignments according to the Manual for Design of Forest Truck Roads (2002b), and considering protection and social function of the forest environment. Determination of suitability of the forest area for the construction of forest roads should be considered from the point of view of terrain conditions and growing stock of timber. The assumptions stated below have been confirmed by the analysis of accessibility of the current forest road network and calculation of optimal density of forest roads: Þ the current density of forest roads is lower than optimal Þ the real average current skidding distance is bigger than the real average targeted skidding distance, meaning that the costs of skidding for the real average targeted skidding distance are lower than the costs of skidding for real average current skidding distance Þ density of the current and proposed forest roads designed on the contour map can be achieved in 26 years on the basis of skidding cost savings, which can justify 38% of the necessary construction costs of forest roads. Based on the results, it can be concluded that: Þ it is necessary to establish a cadastre of forest roads with respective database Þ forest accessibility should be expressed by density of forest roads, skidding distance, relative forest accessibility and the efficiency coefficient of the forest road network Þ the approach of planning forest accessibility at strategic level has been obtained by determining the suitability of the forest area for the construction of forest roads with ArcGIS 10 software. The selected factors affecting the determination of an area suitability for the construction were evaluated by AHP method Þ forest roads have been designed as zero lines plotted on the contour map and laid on the field. This approach enables spatial distribution of the economic length of forest roads for the natural

forests into areas which are suitable for construction of forest roads and in accordance with applicable standards Þ the findings of the research should renew the practice of making the Study of Forest Accessibility for management units.

5. References Abdi, E., Majnounian, B., Darvishsefat, A., Mashayekhi, Z., Sessions, J., 2009: A GIS-MCE based model for forest road planning. Journal of Forest science 55(4): 171–176. Bertović, S., 1999: Reljef i njegova prostorna raščlamba. Šumarski list 123(11–12): 543–563. Bunuševac, T., 1951: Gajenje šuma I, Univezitet u Beogradu, Srbija. Bureau of Statistics of Republika Srpska, 2015: Forestry, Statistical bulletin, No. 15, Banja Luka, Republika Srpska, Bosnia and Herzegovinia. Chung, W., Session, J., 2001: Designing of forest road network using Heuristic optimization techniques, Council on Forest Engineering Conference Proceedings: »Appalachian Hardwoods: Managing Change«, July 15–18, Snowshoe. Đuka, A., Prošinsky, T., Vusić, D., 2015: DTM models to enhance planning of timber harvesting. Glasnik Šumarskog fakulteta, Univerzitet u Beogradu, 35–44. Enache, A., Ciobanu D.,V., Kühmaier, M., Stampfer, K., 2013: An Integrative Decision Support Tool for Assessing Forest Road Options in a Mountainous Region in Romania. Croatian Journal of Forest Engineering 34(1): 43–60. Enache, A., Stampfer, K., Ciobanu V., Brânzea, O., Duţă, C., 2011: Forest road network planning with state of the art tools in a private forest district from lower Austria. Bulletin of the Transilvania University of Braşov Series II-Forestry. Wood industry. Agricultural food engineering 4(53): 33–40. Fannin, R.J., Lorbach, J., 2007: Guide to forest road engineering in mountainous terrain. Food and Agriculture Organization of the United Nations-FAO, Rome, 2007. Food and Agriculture Organization of the United Nations– FAO, 1998: A Manual for the Planning, Design and Construction of Forest Roads in Steep Terrain, FAO, 130 p. Ghaffarian, M.R., Stampfer, K., Sessions, J., 2009: Comparison of three methods to determine optimal road spacing for forwarder-type logging operations. Journal of Forest Science 55(9): 423–431. Govedar, Z., 2011: Gajenje šuma-ekološke osnove. Univezitet u Banjoj Luci, Šumarski fakultet, 299 p. Hayati, E., Majnounian, B., Abdi, E., Sessions, J., Makhdoum, M., 2013: An expert-based approach to forest road network planning by combining Delphi and spatial multi-criteria evaluation. Environmental Monitoring Assessment 185(2): 1767–1776. Croat. j. for. eng. 39(2018)1

Planning Forest Road Network in Natural Forest Areas: a Case Study in Northern Bosnia ... (45–56) V. Petković and I. Potočnik Hribernik, B., 2013: Model zgoščevanja omrežja gozdnih cest v večnamenskem gozdu, Doktorska disertacija, Univerza v Ljubljani, Biotehniška fakulteta, Oddelek za gozdarstvo, Ljubljana, 177 p.

Pentek, T., 2002: Računalni modeli optimizacije mreže šumskih cesta s obzirom na dominantne utjecajne čimbenike. Doktorska disertacija, Sveučilište u Zagrebu, Šumarski fakultet, Zagreb, 271 p.

IUFRO, 1995: Forest work study. Nomenclature. Test Edition valid 1995–2000. International Union of Forestry Research Organisations WP 3.04.02. Swedish University of Agricultural Sciences, Department of Operational Efficiency, Garpenberg, 16 p.

Pentek, T., Pičman D., Nevečerel H., Lepoglavec K., Papa I., Potočnik I., 2011: Primarno otvaranje šuma različitih reljefnih područja Republike Hrvatske. Croatian Journal of Forest Engineering 32(2011): 401–416.

Jeličić, V., 1983: Šumske ceste i putevi. SIZ Odgoja i usmjerenog obrazovanja SRH. Zagreb. Jeličić, V., 1979: Otvaranje sječina sekundarnom mrežom šumskih puteva u borovim i hrastovim šumama. Radovi Šumarskog fakulteta i Instituta za šumarstvo u Sarajevu, knjiga 22., sveska 5–6, Sarajevo. Jež, B., 2016: Model gostitve cestnega omrežja na primeru gozdnogospodarske enote Slivnica, Magistarsko delo, Univerza v Ljubljani, Ljubljana, Slovenija. Keller, G., Sherar, J., 2003: Low-volume roads engineering. The Best Management Practices Field Guide, USDA, Forest Service. Krč, J., Beguš, J., 2013: Planning of forest openning with forest roads. Croatian Journal of Forest Engineering 34(2): 217–228. Lepoglavec, K., 2014: Optimizacija primarne i sekundarne šumske prometne infrastrukture nagnutih terena. Doktorski rad, Sveučilište u Zagrebu, Šumarski fakultet, Zagreb, 266 p. Lotfalian, M., Kooch, Y., Sarikhani, N., 2008: Effective factors in determination optimal density of forest road network. Asian Journal of Scientific Research 1(4): 470–475. Marčeta, D., 2015: Comparison of technologies of wood biomass utilization in beech stands. Doctoral dissertation, University of Ljubljana, Biotechnical faculty, Ljubljana, Slovenia, 148 p. Naghdi, R., Limaei, S.M., 2009: Optimal Forest Road Density Based on Skidding and Road Construction Costs in Iranian Caspian Forests, Caspian Journal Environmental Science 7 (2): 79–86. Pellegrini, M., Grigolato, S., Cavalli, R., 2013: Spatial MultiCriteria Decision Process to Define Maintenance Priorities of Forest Road Network: an Application in the Italian Alpine Region. Croatian Journal of Forest Engineering 34(1): 31–42. Pentek, T., Nevečerel, H., Pičman, D., Prošinsky, T., 2007: Forest road network in the Republic of Croatia – Status and perspectives. Croatian Journal of Forest Engineering 28(1): 93–106. Pentek, T., Pičman, D., Potočnik, I., Dvorščak, P., Nevečerel, H., 2005: Analysis of an existing forest road network. Croatian Journal of Forest Engineering 26(1): 39–50. Pentek, T., Pičman, D., Prošinsky, T., 2004: Planning of forest roads in Croatian mountainous forest by the use of modern technologies. International scientific conference »Forest engineering: New techniques, technologies and the enviroment«. Croat. j. for. eng. 39(2018)1

Petković, V., Marčeta, D., Španjić, S., Kosović, M., 2015: Određivanje srednje transportne distance privlačenja primjenom GIS-a u nizijsko-brdskim uslovima. Glasnik Šumarskog fakulteta Univerziteta u Banjoj Luci 23: 5–14. Pičman, D., 1994: Utjecaj konfiguracije terena i hidrografskih prilika na ekonomsku opravdanost izgradnje optimalne mreže šumskih prometnica. Glasnik za šumske pokuse 31: 231–316. Pičman, D., 2007: Šumske prometnice, Šumarski fakultet u Zagrebu, Hrvatska, 460 p. Potočnik, I., 2004: Šumske komunikacije, Skripta, Šumarski fakultet u Banjoj Luci, Bosna i Hercegovina. Potočnik, I., 2005: Depth of carriageway and cut slopes on forest roads. International Scientific Conference »Ecological, Ergonomic and Economical optimization of Forest Utilization in Sustainable Forest Management«, Department of forest an wood utilization Faculty of Forestry The Hugo Kollataj, Agricultural University of Krakow, Poland. Public Forestry Company »Šume Republike Srpske«, 2002b: Manual for Design of Forest Truck Roads, Banja Luka. Public Forestry Company »Šume Republike Srpske«, 2012: Forest Management Plan for Forest management area »Posavsko«, Banja Luka. Public Forestry Company »Šume Republike Srpske«, 2002a: The Unique Standards for Forest Works, Sokolac. Public Forestry Company »Šume Republike Srpske«, 2009: Description and Unit Costs of Forest Road Construction Works, Banja Luka. Rafiei, A.A., Lotfalian, M., Hosseini, S.A., Parsakhoo, A., 2009: Determining the optimum road density for ground skidding system in Dalak Khely forest-Hyrcanian zone. World applied sciences journal 7(3): 263–270. Rebula, E., 1980: Optimal openness of forests. Mehanizacija šumarstva 5(3-4): 107–119. Republika Srpska Investment-Development Bank (IRBRS), 2017: http://www.irbrs.org/azuro3/azuro/uploads/Struktura_kamatnih_stopa_IRBRS_do_30.09.2017.pdf. Saaty, T.L., 1980: The Analytic Hierarchy Process, McGrawHill, New York. Saaty, T.L., 2008: Decision making with the analytic hierarchy process. International Journal Services Sciences 1(1): 83–98. Samani, K.M., Hosseiny, S.A., Lotfalian M., Najafi A., 2010: Planning road network in mountain forests using GIS and

V. Petković and I. Potočnik Planning Forest Road Network in Natural Forest Areas: a Case Study in Northern Bosnia ... (45–56) Analytic Hierarchical Process (AHP). Caspian Journal Environmental Sciences 8 (2): 51–162. Sokolović, Dž., Bajrić, M., 2013: Šumska prometna infrastruktura u Federaciji Bosne i Hercegovine. Nova mehanizacija šumarstva 34(1): 39–50. Sokolović, Dž., Lojo, A., Bajrić, M., Halilović, V., 2009: Uticajni faktori na izbor područja pogodnih za gradnju čumskih kamionskih puteva. Univerzitet u Sarajevu, Šumarski fakultet, 43–57.

Stupar, V., Čarni, A., 2017: Ecological, floristic and functional analysis of zonal forest vegetation in Bosnia and Herzegovina. Acta Botanica Croatica 76(1): 15–26. Tampekis, S., Sakellariou, S., Samara, F., Sfougaris, A., Jaeger, D., Christopoulou, O., 2015: Mapping the optimal forest road network based on the multicriteria evaluation technique: the case study of Mediterranean Island of Thassos in Greece. Environmental Monitoring Assessment 187(11): 687.

Authors’ addresses: Vladimir Petković, MSc.* e-mail: vladimir.petkovic@sfbl.org University of Banja Luka Faculty of Forestry Bulevar Vojvode Stepe Stepanovića 75 78000 Banja Luka BOSNIA AND HERZEGOVINA

Received: August 16, 2016 Accepted: July 12, 2017

Prof. Igor Potočnik, PhD. e-mail: igor.potocnik@bf.uni-lj.si University of Ljubljana Biotechnical Faculty Jamnikarjeva 101 1000 Ljubljana SLOVENIA * Corresponding author Croat. j. for. eng. 39(2018)1

Original scientific paper

Study of Forest Road Effect on Tree Community and Stand Structure in Three Italian and Iranian Temperate Forests Rodolfo Picchio, Farzam Tavankar, Rachele Venanzi, Angela Lo Monaco, Mehrdad Nikooy Abstract Roads are built in forests for two main reasons, but always in function of management of forest ecosystems, and these reasons are to provide access to the forest area for transportation mobility and wood extraction. This creates a relatively even network in the forest. This topic has received much attention in recent years due to its function and effect on forested rural landscapes and the related environment. Forest road network is important for various types of functional use, such as the interface between forested lands and roads. The aim of this study is to assess the effects of road existence and use on the occurrence of tree dieback and on the composition of the tree community in three forest areas (two in Italy and one in Iran). The effort to determine the dynamics of the effects caused by road use was done by examining the changes in stand structure and abundance of species. As demonstrated by the results, the edges (20 m) of the forest road network are a fine mosaic composed of different trees (qualitative and quantitative), coupled with the moderate presence of dead trees. In the three areas, from the road edges to the interior forest, a similar taxonomic composition of forest community was found. The first main difference was related to the abundance of less shadow tolerant species along the road. The second main difference was related to the tree biodiversity indices that are higher along the road. The main similarities are in the structure of live and dead trees. Keywords: forest road, logging, tree species diversity, SIV, tree dieback intensity index

1. Introduction The basic infrastructure of forestry operations are forest roads. Roads are built in forests for two main reasons: to provide access to the forest area for transpor tation mobility and to provide extraction roads. This creates a relatively even network in the forest (Heinimann 1998, Ryan et al. 2004, Cerdà 2007, Grigolato et al. 2013). It should also be considered that the road network is used for transportation purposes and is, therefore, a facility for inhabitants. Forest roads are the most costly structures in forestry. However, the construction activities involved have a heavy environmental impact on the adjacent ecosystems. These activities include earth movement that can disturb the watershed (Demir et al. 2007). Croat. j. for. eng. 39(2018)1

Much attention has been given to forest road networks in recent years, by observing the functional uses and effects that they have on forested rural landscapes and the related environment (Grigolato et al. 2013, Gumus et al. 2008). The primary purpose of forest road networks is to make logging operations and forest transportation more efficient. Therefore, the forest road network is important for various types of functional use, such as the interface between forest lands and roads. Forest road network planning requires the commercial assessment of production costs, which imply silvicultural treatments. A forest road network, which is well designed and properly used, allows to use low-impact extraction and transport systems (Magagnotti et al.

R. Picchio et al. Study of Forest Road Effect on Tree Community and Stand Structure in Three Italian and Iranian ... (57–70)

2012, Picchio et al. 2011 and 2012, Marchi et al. 2014 and 2016). Nevertheless, the undesirable effects caused by roads in forest areas affect both biotic and abiotic environmental components. There are several papers written about the negative effects caused by forest roads (Coffin 2007). For example, extensive consideration has been given to the analysis of the forest road network as a source of accelerated soil erosion and the production of sediment (Bilby et al. 1989, Pellegrini et al. 2013) and also by GIS models used in Skaugset et al. (2011). These alter the surrounding habitat in a number of ways that eventually influence the quality and suitability of the roadside areas for trees and animals. The creation of a forest edge causes a dramatic increase in the amount of solar radiation reaching the forest understory. Solar radiation, soil moisture content and soil temperature in gaps created by forest roads are significantly greater than in adjacent closed canopy plots (Osma et al. 2010, Li et al. 2010). Generally, roadsides are more disturbed, drier and warmer (Forman and Alexander 1998). Soil moisture levels are reduced within 20 meters of a forest edge, due to an increase in evapotranspiration rates as trees are exposed to higher temperatures and solar radiation (Kapos 1989). Variations in light, temperature and moisture levels have a significant impact on plant biodiversity, tree regeneration and quality of edge trees. Most of the trees that are situated along the newly created edges do not have the ability to adapt to the increased stress caused by higher temperatures and stronger winds (Laurance et al. 2007). Fast growing species, such as Alnus, Rubus, and others, are found around forest roads (Parendes and Jones 2000). Extensive tree mortality often occurs at the forest edges, with trees suffering physical damage caused by strong winds (Murcia 1995). Many other trees remain as standing dead trees, killed by excessive desiccation resulting from drier conditions and increased light exposure (Olupot 2009, Tavankar et al. 2017). Forest roads increase tree mortality via microclimatic changes (Kapos 1989, Williams-Linera 1990), mechanical damages (Chen et al. 1992) and an increased infestation by pathogens (Dickie and Reich 2005). A recent study of Natura 2000 (European Environment Agency 2015) about European forest areas, as described in Sitzia et al. (2016), underlined that more than 60% of the assessments of forest and woodland habitats and species listed in the Habitats Directive revealed a bad or inadequate conservation status and that roads were among the most important threats to these habitats and species. However, forest roads are important for an economically viable forest manage-

ment (European Commission 2015) and for the implementation of frequent and careful conservation management activities. Moreover, the forest road network provides access necessary for water protection, wildlife habitat improvement, fire control and many recreational activities (Akay and Sessions 2004, Grigolato et al. 2013). The edge effect of forest roads on the tree community is an important issue regarding road ecology. Roads can affect species in at least three ways: by reducing the habitat available, by affecting movement patterns, and by extending the edge conditions into the forest. The main objective of this study is to verify if, in productive and managed forests, the viability could represent not only an environmental risk, but also a possibility of increasing the biodiversity at landscape level. In fact, a fine mosaic composed of different trees (qualitative and quantitative), coupled with the presence of scattered old or dead trees provided favorable conditions also for a variety of invertebrates. Considering both environmental aspects and engineering applications for forest production and protection, the research is well within the scope of ecological engineering (Mitsch and Jørgensen 2003). In this study, the effects were assessed of forest roads on tree community composition, structure and snag presence in three temperate forest areas, two in Italy and one in Iran. However, this work does not provide data extrapolated for any typology of forest, but aims to identify a methodology and some indexes that can be used for the development of subsequent studies on larger geographical scales. This study specifically addresses the following questions: Þ Does the forest road affect the tree species diversity? Þ Are there any significant differences between tree species diversity along the roads and inside the forest? Þ Are there any significant differences between forest structure along the roads and inside the forest, in terms of live and death trees (snag)?

2. Material and methods 2.1 Study areas The first study area is the Iranian Caspian forests located in northern Iran and on the south coast of the Caspian Sea. The Caspian forests (also called the Hyrcanian forests) cover about 1.9 million ha and are rich in biological diversity, with endemic and endangered Croat. j. for. eng. 39(2018)1

Study of Forest Road Effect on Tree Community and Stand Structure in Three Italian and Iranian ... (57–70) R. Picchio et al.

Fig. 1 Localization of the Iranian studied area, highlighted with a circle species and rich in many ecological niches (Heshmati 2007, Tavankar et al. 2014). The research was carried out in district n° 2 of the Nav forest area (latitude 37°41’30” to 37°45’21” N, longitude 48°33’44” to 48°51’33” E) in watershed n° 7 in the Guilan province, north of Iran, Fig. 1. The average rainfall ranged from 920 to 1,100 mm per year, with the heaviest precipitation in the summer and fall. The average daily temperature ranges from a few degrees below 0°C in December, January, and February, and up to +25 °C during the summer. The original vegetation of this area is an uneven aged mixed forest dominated by Fagus orientalis Lipsky and Carpinus betulus L., with the companion species Alnus subcordata C.A. May, Acer platanoides L., Acer cappadocicum Gled., Ulmus glabra Huds., and Tilia begonifolia Steven (Tavankar et al. 2014). The Caspian forests are managed as a mixed uneven aged high forest with single and group selective cutting regime. The soil of the study site is classified as brown forest (Alfisols) and well-drained. The texture of the soil ranges from clay loam to loamy. All roads in this forest are 5.5 m wide and are unpaved. Croat. j. for. eng. 39(2018)1

They were constructed in the 1970s to implement shelterwood systems. The total length of the forest roads in the study area (district n°. 2) is 8.09 km and the density of the forest road is 22.9 m ha–1. The last selective logging of the area was performed in 2008 with cable-skidder logging, and the roads have not been used since then. There was no grazing in this area, because it was protected by barbed wire. The second study area is the Amiata Mountain forests, an isolated relief of volcanic origin located in Tuscany, central Italy. This forest area is covered by about 2751 ha of beech (Fagus sylvatica L.) forest. Due to the frequency of pure stands, their good productivity, the presence of different silvicultural typologies (high forest, coppice stands converted into high forest and aged coppices), it represents an interesting laboratory for the analysis of different management choices. The research was carried out in »Macchia faggeta« forest propriety (Abbadia San Salvatore, latitude 42°89’55” to 42°89’64” N, longitude 11°63’73” to 11°64’12” E), designated with the letter »A« in Fig. 2. The average rainfall ranged from 900 to 1600 mm per year, with the

R. Picchio et al. Study of Forest Road Effect on Tree Community and Stand Structure in Three Italian and Iranian ... (57–70)

heaviest precipitation in the autumn and winter. The dry period is limited to 18 days only. The mean annual temperature is 10°C, the hottest month (August) with 18.5°C, the coldest month (January) with 1.8°C. The main vegetation of this area is an even aged mixed forest dominated by Fagus sylvatica L. and Abies alba Mill., with the companion species Acer pseudoplatanus L., Acer opalus Mill., Carpinus betulus L., Ulmus glabra Huds., Quercus cerris L., Ilex aquifolium L., Sorbus torminalis L., Fraxinus excelsior L., Tilia plathyphyllos Scop. and Prunus avium L. The Amiata forests are managed mainly (about 80%) as a mixed even aged high forest with shelterwood cutting regime. The parent material is a trachyte lava with a high silicate content and poor in basis; the slope is generally gentle but a few outcrops are present; there is no erosion and no landslides. Brown soils of good physical structure are prevalent. The texture of the soil ranges from clay loam to loamy. The main forest roads in this area are about 4 m wide and are paved with rocks. They were constructed in the 1900s. The total length of the forest roads in the study area is about 11.25 km and the density of the forest road is 18.1 m ha-1. The last logging of the area was partly performed in 2010 and the rest in

Fig. 2 Localization of the Italian studied areas, highlighted with two circles, A: Macchia Faggeta, B: Pramosio

2013 with cable-skidder logging and forwarding, and the roads have not been used since then. There was no grazing in this area. The third study area is in the Carnia Alps, the Pramosio forests located in Friuli Venezia Giulia, north Italy. This forest area is covered by about 495 ha of high forests, mainly composed of Abies alba Mill., Picea abies (L.) H.Karst. and Fagus sylvatica L. Due to high frequency of high forest pure stands and their good productivity, it represents an interesting laboratory for the analysis of management choices. The research was carried out in »Pramosio« forest regional propriety (Paluzza, latitude 46°56’84” to 46°58’09” N, longitude 13°01’80” to 13°02’65” E), shown with the letter »B« in Fig. 2. The average rainfall ranged from 1400 to 1900 mm per year, with the heaviest precipitation in the autumn and winter. The mean annual temperature is 8°C, the hottest month (August) with 17.0°C, the coldest month (January) with –1.5°C. The main vegetation of this area is an even aged mixed forest dominated by Abies alba Mill., Picea abies (L.) H.Karst. and Fagus sylvatica L., with the companion species Acer pseudoplatanus L., Acer opalus Mill., Ulmus glabra Huds., Sorbus aria L., Fraxinus excelsior L., Tilia plathyphyllos Scop. and Larix decidua Mill. The Pramosio forests are managed mainly (about 90%) as a mixed even aged high forest with shelterwood cutting regime. The Late Ordovician (Caradoc) to Early Carboniferous successions of the Carnic Alps, along the Italian–Austrian border in the easternmost part of the Southern Alps, consist predominantly of carbonate rocks representing shallow to open sea environments. Carbonate sedimentation persisted through into the earliest Viséan, resuming during the Late Carboniferous. Carbonate build-ups, characteristic of the Middle Devonian, persisted into the Frasnian, but ceased during early Famennian transgressive tectono-eustatic events when sediments representing open marine environments, the late Frasnian–earliest Viséan Pramosio Limestone, accumulated. This unit consists of 0.5–3 cm layers of light grey, beige to pink biomicritic limestones (wackestones and packstones) interbedded with thin (mm–cm) calcisiltitic levels. The Pramosio Limestone was previously referred to in various ways, such as the »Calcari climenie«, clymenid limestone, clymenid and goniatitid-bearing pelagic limestone. They are considered to equate with the combined Pal and Kronhof Limestones of Austrian colleagues and the »Calcari a climenie« auctorum outcropping in southwestern Sardinia. The texture of the soil ranges from clay loam to loamy. The main forest roads in this area are about 4 m wide and are paved with rocks. They were constructed in the 1900s. The total length of the forest roads in the study area is about Croat. j. for. eng. 39(2018)1

Study of Forest Road Effect on Tree Community and Stand Structure in Three Italian and Iranian ... (57–70) R. Picchio et al.

4 km and the density of the forest road is 22.5 m ha–1. The logging in the area was performed yearly, with cable-skidder logging, cable yarder and forwarding. There was no grazing in this area.

2.2 Data collection and analysis For each studied area, four transects, each of approximately 1 km in length, were established parallel to the road. Two transects were established adjacent to the road edge (cut and fill slopes) as road stands, and two transects were established at a 50 m distance from the road edges as internal stands, as reported in others studies (Olander et al. 1998, Watkins et al. 2003, Prasad 2009, Li et al. 2010, Avon et al. 2010, Bergès et al. 2013). On each transect, 10 square sample plots (20×20 m) were established randomly. In each sample plot, all live and dead trees (snag) were counted, species were identified, diameter at breast height (DBH) and height were measured. The tree dieback intensity index (DTI) was computed as the ratio of dead trees and live trees in each plot. Species importance value (SIV) for each species was calculated as reported by other authors (Ganesh et al. 1996, Krebs 1999, Pourbabaei and Abedi 2013, Pourbabaei et al. 2013, Tavankar 2015): SIV = Relative density (RDe) + Relative frequency (RF) + Relative dominance (RDo) (1) Where: RDe number of individuals of a species x100/total number of individuals of all species number of plots containing a species x100/ RF sum of frequencies of all species. The Basal area was considered for dominancy and relative dominance (RDo) calculated by: RDo basal area of a species ×100/total basal area of all species The species diversity index was computed using the Shannon–Wiener information function (Hill 1973, Ozcelik et al. 2008, Tarvirdizadeh et al. 2014) as:

∑

H' = − (ni / n) × Ln (ni / n) (2) Where: ni denotes the SIV of a species n the sum of total SIV of all species. After checking for normality (KolmogorovSmirnov test) and homogeneity of variance (Levene’s test), the average values of stand characteristics (tree density, basal area, stand volume) were compared in two plot positions (road and internal) using paired Croat. j. for. eng. 39(2018)1

samples t test. The average biodiversity index of the two positions of the plots was also compared using paired samples t test. Regression analysis was applied to test the relation between tree density and DBH and to test the relationship between tree height and DBH. Chi-square nonparametric test was used to compare the dead tree index (DTI) in two positions and tree species. Number of trees (live and dead) in the plots (road and interior) were analysed by ANOVA test. Post-hoc test was conducted with Tukey HSD test method. SPSS 19.0 software was used for statistical analysis; the results of the analysis were also presented using descriptive statistics.

3. Results 3.1 Tree species composition The average number of tree species along the road and internal plots is shown in Fig. 3. The graph illustrates the highest variation of the number of trees in the two positions, along the road and in the internal stand. For the three areas studied, the genus Fagus showed a higher density, even if along the road its density decreased. Caspian forest Among the seven tree species that were observed in the plots, the Fagus orientalis showed higher density (105 trees/ha) in the internal stands, while the Alnus subcordata showed higher density (112.5 trees/ha) in the stand along the road. The density of Fagus orientalis, Carpinus betulus, Tilia begonifolia and Ulmus glabra was higher in the internal stand than in the road edge stand, while the density of Acer velutinum, Acer cappadocicum and Alnus subcordata was higher in the road edge stand than in the internal stands. Amiata forest Among the eleven tree species that were observed in the plots, the Fagus sylvatica showed higher density (120 trees/ha) in the internal stands and in the stand along the road (80 trees/ha). The density of Fagus sylvatica, Abies alba and Ulmus glabra in the internal stand was higher than the road edge stand, while the density of Acer pseudoplatanus, Acer opalus, Carpinus betulus, Quercus cerris, Sorbus torminalis, Fraxinus excelsior, Tilia plathyphyllos and Prunus avium was higher in the road edge stand than in the internal stands. Pramosio forest Among the eleven tree species that were observed in the plots, the Picea abies showed higher density

R. Picchio et al. Study of Forest Road Effect on Tree Community and Stand Structure in Three Italian and Iranian ... (57–70)

Fig. 4 Tree frequency values of DBH classes in road and internal plots for the three areas (87.5 trees/ha) in the internal stands, while the Acer pseudoplatanus showed higher density (52.5 trees/ha) in the stand along the road. The density of Fagus sylvatica, Abies alba, Picea abies and Ulmus glabra in the internal stand was higher than the road edge stand, while the density of Acer pseudoplatanus, Acer opalus, Larix decidua, Sorbus aria, Fraxinus excelsior, Tilia plathyphyllos and Prunus avium was higher in the road edge stand than in the internal stands. Tree frequency values of DBH classes in the road and internal plots are shown in Fig. 4. The tree number decreased with increasing DBH in the road plots of the three areas and in internal plots of the Caspian forest, while in the internal plots of Amiata and Pramosio forests, the number of trees in terms of DBH showed a Gaussian distribution, typical of even-aged forests. The regression analyses were applied to live trees to test the relationship between tree frequency and DBH in the road and internal stands; the results shown in Table 1 are statistically significant (p<0.01). The average volume of trees in road and internal plots and ANOVA results are shown in Table 2. The statistical test showed that the average volume of trees in the road stands was significantly (p<0.05) higher than in internal stands for the Caspian forest, while it was significantly (p<0.05) lower than in internal stands for the Amiata and Pramosio forests. Fig. 3 Number of tree species in road and interior plots for the three areas

The average number of trees (live and dead) in the two plot positions (road and internal) is based on 1 Croat. j. for. eng. 39(2018)1

Study of Forest Road Effect on Tree Community and Stand Structure in Three Italian and Iranian ... (57–70) R. Picchio et al.

Table 1 Results of non linear regression analyses applied to test the relationship between live tree frequency (y) and BDH (x) in the road and internal stands Study area

Regression

R2 Adj.

p-value

Caspian forest

y=–0.848ln(x)+4.1468

0.971

<0.01

0.969

<0.01

Tree position

Interior

Amiata forest

y=–0.00003x +0.0023x +0.0451x–0.8554 2

Pramosio forest

y=–0.0017x +0.188x–2.4791

0.926

<0.01

Caspian forest

y=–2.046ln(x)+9.3564

0.893

<0.01

Amiata forest

y=–2.106ln(x)+9.0564

0.817

<0.05

Pramosio forest

y=–2.156ln(x)+9.564

0.789

<0.05

Road

Table 2 Mean volume of trees in road and interior plots and ANOVA and Tukey test results Position

Road

Interior

Study area

Volume m ha–1 ± SE

Caspian forest

361.75±13.25 a

Amiata forest

311.8±10.12 b

Pramosio forest

395.9±19.21 c

Caspian forest

222.75±52.25 d

Amiata forest

345.2±9.78 e

Pramosio forest

405.1±14.51 c

p-value

<0.05

hectare, and ANOVA results are shown in Table 3. The results of ANOVA indicated that the density of live trees in the stands along the road was significantly higher than in the internal stands in all the three areas. The results of ANOVA also indicated that the density of dead trees in the road stands was significantly higher than in the internal stands. The tree death index obtained (DTI as ratio of dead trees to live trees) for each species in the plots along the road and in the internal plots is shown in Fig. 5. The species were the same as the living tree species reported in Fig. 3.

Table 3 Mean number of trees (live and dead) near the road and in the forest interior for each area; ANOVA and Tukey test (different letters show significant statistic differences) were done for the total number of trees to compare the studied areas, while the two positions (road and interior) were compared separately for tree typology Position

Road

Study area

Total number of trees n ha–1±SE

Caspian forest

532.5±20.5 a

Amiata forest

495.4±23.8 b

Pramosio forest

452.5±18.1 c

p-value

<0.01

Interior

Caspian forest

240.0±12.2 d

Amiata forest

345.7±18.1 e

Pramosio forest

402.5±15.8 f

Croat. j. for. eng. 39(2018)1

Tree typology

Number of trees n ha–1±SE

Live trees

352.5±32.5 a

Dead trees

180.0±17.5 b

Live trees

410.0±28.1 a

Dead trees

70.0±27.5 b

Live trees

387.5±15.3 a

Dead trees

65.0±22.4 b

Live trees

182.5±20.0 a

Dead trees

57.5±7.5 b

Live trees

285.2±12.2 a

Dead trees

60.5±25.4 b

Live trees

357.5±12.3 a

Dead trees

45.0±10.1 b

p-value

<0.01

R. Picchio et al. Study of Forest Road Effect on Tree Community and Stand Structure in Three Italian and Iranian ... (57–70)

Caspian forest The DTI in the plots along the road was 0.51, and it was significantly higher (X2=5.94, p<0.05) than the value of the tree death index of the internal plots (0.32). The index of tree death of all tree species in the »road plots« was higher than in the internal plots. The maximum value of the index of tree death (0.82) was obtained in Tillia begonifolia and the minimum (0.22) was obtained in Carpinus betulus. In the »road« plots, the DTI in Fagus orientalis was 0.62, while in the internal plots, it was 0.23 (X2=9.69, p<0.01). Except Carpinus betulus, the index of tree death was significantly higher in the road plots than in the internal plots. Amiata forest The DTI in the plots along the road was 0.17, and it was not significantly higher (X2=2.83, p>0.05) than the value of the tree death index of the internal plots (0.21). The DTI in Fagus sylvatica, Abies alba and Sorbus torminalis was higher in the »road plots« than in the internal plots, while for the other eight species, the DTI in the internal plots was higher than in the road plot. The maximum value of the index of tree death (0.88) was obtained in Ulmus glabra and the minimum (0.01) was obtained in Carpinus betulus. in the »road« plots, the DTI in Fagus sylvatica was 0.31, while in the internal plots it was 0.25 (X2=7.13, p>0.05). Pramosio forest The DTI in the plots along the road was 0.17, and it was not significantly higher (X2=2.47, p>0.05) than the value of the tree death index of the internal plots (0.13). The DTI in Acer opalus, Acer pseudoplatanus and Prunus avium was higher in the »internal plots« than in the road plots, while for the other eight species the DTI in the internal plots was similar to that in the road plots. The maximum value of the index of tree death (0.70) was obtained in Acer opalus and the minimum (0.01) was obtained in Prunus avium, Fraxinus excelsior and Sorbus aria. In the »road« plots, the DTI in Fagus sylvatica was 0.10, while in the internal plots it was 0.20 (X2=8.95, p<0.05).

Fig. 5 Tree dieback intensity index of tree species in road and interior plots for the three areas

The variation of tree death index in DBH classes considering the road and internal plots is shown in Fig. 5. DTI values decrease with increasing of DBH in the road and internal stands. The values of tree death in road plots were higher than in internal plots in all DBH classes. For the two areas (Pramosio and Amiata) with even-aged management, the DTI values were higher for the DBH class <50 cm and they were similar for interior and road plots. For the Caspian forest with uneven-aged management, the DTI values were highCroat. j. for. eng. 39(2018)1

Study of Forest Road Effect on Tree Community and Stand Structure in Three Italian and Iranian ... (57–70) R. Picchio et al.

Fig. 6 Variation of DTI (adimensional index) in DBH classes considering road and internal plots for the three areas er for the DBH class <60 cm in road plots, while the DTI values were under the value of 0.4 for all the DBH classes in the interior plots. Species Importance Value (SIV) of the tree species in the road and internal stands is shown in Fig. 7 for the three areas, where the SIV of the secondary species is slightly higher in the road plots. Caspian forest The SIV of Fagus orientalis was the highest in both the road and internal plots. The SIV of Fagus orientalis, Carpinus betulus and Ulmus glabra was higher in the internal plots than in the road plots. However, the SIV of Acer velutinum, Acer cappadocicum and Alnus subcordata was higher in the road plots than in the internal plots. The SIV of Tilia begonifolia was almost equal along the road and in internal plots. The mean value of tree species diversity, shown in Fig. 8 and Table 4 (Shannon-Wiener index), in the road and internal plots was 1.42±0.18 and 1.01±0.29, respectively. The evenness value, shown in Table 4, in the road and internal plots was 0.81±0.02 and 0.64±0.05, respectively. Amiata forest The SIV of Fagus sylvatica was the highest in the interior plots, while the SIV of Fagus sylvatica and Prunus avium was the highest in the road plots. The SIV of Acer pseudoplatanus, Acer opalus, Carpinus betulus, Quercus cerris, Ulmus glabra, Sorbus torminalis, Fraxinus excelsior, Tilia plathiphyllus and Prunus avium was Croat. j. for. eng. 39(2018)1

Fig. 7 SIV values along the road and in interior plots for the three areas

R. Picchio et al. Study of Forest Road Effect on Tree Community and Stand Structure in Three Italian and Iranian ... (57–70)

Fig. 8 Shannon-Wiener index values along the road and in interior plots for the three areas (in the graph, the indicators represent the mean values and the bars the standard deviations) igher in the road plots than in the interior plots. The h SIV of Abies alba was almost equal along the road and in internal plots. The mean value of tree species diversity, shown in Fig. 8 and Table 4 (Shannon-Wiener index), in the road and internal plots was 2.31±0.21 and 1.29±0.11, respectively. The evenness value, shown in Table 4, in the road and internal plots was 0.96±0.08 and 0.54±0.01, respectively. Pramosio forest The SIV of Fagus sylvatica was the highest in the interior plots, while the SIV of Picea abies was the highest in the road plots. The SIV of Picea abies, Acer pseudoplatanus, Acer opalus, Larix decidua, Ulmus glabra, Sorbus aria, Fraxinus excelsior, Tilia plathiphyllus and Prunus avium was higher in the road plots than in interior plots. The SIV of Fagus sylvatica and Abies alba was lower in the road plots than in interior plots. The mean value of tree species diversity, shown in Fig. 8 and Table 4 (Shannon-Wiener index), in the road and internal plots was 2.29±0.27 and 1.87±0.25, respectively. The evenness value, shown in Table 4, in the road and internal plots was 0.95±0.11 and 0.78±0.07, respectively.

4. Discussion 4.1 Structure The results indicated a similar taxonomic composition of forest community between the plots along the road and the internal ones. These results accord with

the results obtained by Tehrani et al. (2015) in the Patom district (Iran). The species observed were the same near the road and in the internal plots of the forest but the frequencies of the tree species were different. The road effect on the abundance of the species was evident. It is interesting to note that exotic species were not found. Watkins et al. (2003) examined the roadside edge effect on plant distribution and noticed a lower canopy cover and more light and, as a result, a different species composition from that of the internal area. Delgado et al. (2007) have detected short and abrupt changes from the road edge to the internal area of forests in temperature, light and canopy traits. The clearing made for the road in the study areas made it possible for the pioneer tree species (Alnus subcordata, Acer velutinum, Acer cappadocicum, Acer pseudoplatanus, Acer opalus, Larix decidua, Sorbus aria and Prunus avium) to reach a higher number than that found in the internal forest. The road produced a greater abundance of these less tolerant shadow tree species due to decreased competition, providing more resources for some species, which were less abundant in the internal zone. For the Caspian forest Alnus subcordata and for the Amiata (Apennines) and Pramosio (Alps) forests, Acer pseudoplatanus, A. opalus and Prunus avium are the most abundant species in the stand near the road. This was probably due to their capability of fast growing and growing in mineral soils. Caspian forest The structure of the two plot positions was also different. The analysis of the tree density distribution in DBH classes (Fig. 4) showed a similar uneven aged structure, a legacy of the previous period of the road construction. The stand along the road showed larger tree diameters than those in the internal area. However, the internal area exhibited a higher complexity due to the number of trees with a diameter larger than 80 cm. The tree heights were greater in the internal area for trees over 25 cm DBH and the average volume of the trees was significantly lower in the internal plots. Amiata forest The structure of the two plot positions was also different. The analysis of the tree density distribution in DBH classes (Fig. 4) showed an uneven aged structure only for the road plots, a legacy of the previous period of the road construction or the effect of a major light exposure, while it showed an even aged structure only for the interior plots. The stand along the road showed lower tree diameters than those in the interior area. However, the internal area exhibited a higher composition of big trees, a diameter larger than 70–80 cm. The Croat. j. for. eng. 39(2018)1

Study of Forest Road Effect on Tree Community and Stand Structure in Three Italian and Iranian ... (57–70) R. Picchio et al.

tree heights were greater in the internal area for trees over 25 cm DBH and the average volume of the trees was significantly higher in the internal plots. Pramosio forest The structure of the two plot positions was also different. The analysis of the tree density distribution in DBH classes (Fig. 4) showed an uneven aged structure only for the road plots, a legacy of the previous period of the road construction or the effect of a major light exposure, while it showed an even aged structure only for the interior plots. The stand along the road showed lower tree diameters than those in the interior area. However, the internal area exhibited a higher com position of big trees, a diameter larger than 70–80 cm. The tree heights were greater in the internal area for trees over 25 cm DBH and the average volume of the trees was significantly higher in the internal plots. The snags are an important component in providing wildlife habitat and can be assumed as an attribute of forest biodiversity and naturality, even in managed forests (Tavankar et al. 2014). Snags species in plots along the road and in internal plots are the same as those observed in living trees, but the dead trees were significantly higher in plots near the road than in internal stands. The results indicated that the tree death was three times greater in the stands near forest roads (<20 m) than in internal stands, in Caspian forest, while this increase ranged from 16% to 44% in Amiata and Pramosio forests, respectively. The greater tree mortality in association with forest roads and clearings compared to internal areas has also been reported in a variety of forest ecosystems (Temperate forests, Chen et al. 1992, Boreal forests, Esseen 1994, Tropical rainforests, Ferreira and Laurance 1997, Tropical dry forests, Prasad 2009). In Puerto Rico, roads in tropical forests affected the composition of adjacent vegetation over a distance of 10 m (Olander et al. 1998). Watkins et al. (2003) studied the effects of forest roads on understory plants in a managed hardwood landscape in the Chequamegon National Forest, Wisconsin (U.S.A.). They reported that roads appeared to be associated with a disturbance corridor that affected site variables up to 15 m into the hardwood stands. The dead tree index (DTI) for species in the internal plots were similar, indicating a homogeneous pressure on different species. The disturbance effect on the area near the road was denoted both by more abundance of dead trees in the plots along the road and by the variation of the dead tree index of the species. Caspian forest The higher dead tree index per plot was found along the road for Tilia begonifolia. A high DTI was exCroat. j. for. eng. 39(2018)1

hibited in the »rare« species, such as U. glabra, in Alnus subcordata and in Fagus orientalis. Only Carpinus betulus showed a low DTI, with hardly any significant statistical difference in comparison with internal plots. The difference in dead tree index and in snag presence could be assigned to the effect of the road on community along the roads that have been influenced by microclimatic characteristics due to the linear clearing in the forest. The snag distribution by diameter classes showed that on average the young trees (DBH<30 cm) were more sensitive to the construction of forest roads than the older trees. The negative effects of forest roads on trees can be a very important issue in forests managed for biodiversity conservation and tourism. Amiata forest A higher dead tree index per plot was found in the interior plots for Ulmus glabra. A high DTI was exhibited also for Abies alba along the road. A high DTI in the interior plots was exhibited in the heliophilous species, such as Acer pseudoplatanus and A. opalus. The difference in dead tree index and in snag presence could be assigned to the effect of the road on community along the roads that have been influenced by microclimatic characteristics due to the linear clearing in the forest, but also by logging activities, mainly thinnings in the interior plots. The snag distribution by diameter classes showed that, on average, small trees (DBH<30 cm) were more frequent along the road, while big trees (DBH>50 cm) were more frequent in the interior plots. Pramosio forest A higher dead tree index per plot was found in the interior plots for Acer opalus. A high DTI was also exhibited for Acer pseudoplatanus and Ulmus glabra in the interior plots. For the U. Glabra, a medium-high DTI was also exhibited along the road. The difference in dead tree index and in snag presence could be assigned to the effect of road on community along the roads that have been influenced by microclimatic characteristics due to the linear clearing in the forest, but also by logging activities, mainly thinnings in the interior plots. The snag distribution by diameter classes showed that, on average, small trees (DBH<30 cm) were more frequent along the road, while big trees (DBH>50 cm) were more frequent in the interior plots.

4.2 Species diversity Between the two plot positions, there was no difference in tree species richness, as indicated by other authors (Tehrani et al. 2015, Watkins et al. 2003). However, the tree community of the plots along the road showed not only different densities of tree species, but also different Species Importance Value. The rare spe-

R. Picchio et al. Study of Forest Road Effect on Tree Community and Stand Structure in Three Italian and Iranian ... (57–70)

cies showed similar SIV along and in the internal plots. The differences found in the abundance of species indicate that there was no disturbance. The species importance value (SIV) of pioneer tree species was higher in the plots along the road than in the internal stands. Watkins et al. (2003) observed that a road side edge has a different species composition due to the changed microclimate conditions. Najafi et al. (2012) reported that the density of tree regeneration changed significantly over a 7.5 m distance. The researchers also concluded that the construction of forest roads may cause the presence of light-demanding species, such as Acer sp. (Maple), which were found to be present closer to the road edges, while the density of shade tolerant species, such as Fagus spp. (Beech), increased with the distance from the road. Bergès et. al. (2013) sampled 30 pairs of 2000 m2 plots, one on the road and the other 30 m inside the forest, on two road surfacing materials (limestone gravel and bare soil) and three stand ages (young, middle-aged and mature) in Scots and Corsican pine stands in a large managed forest in northern France. They reported that the species richness of all plant groups was always higher on roads compared to forest stands, and that the forest plot communities were nearly completely nested within the roadside plot communities. Avon et al. (2010) studied the effect of forest road distance on plant understory diversity at 20 sites in young and adult oak stands in a French lowland forest with a long history of management and road construction. They reported that the main road effect extended less than 5 m into the forest stand. Li et al. (2010) reported that the effect distance reached up to 20–34 m, regardless of the road grade in Great Hingan Mountains in China. Shannon and evenness indices are two models to measure species diversity. They account for the degree of homogeneity in species abundance and can be used to measure the human effect on ecological systems. The tree species diversity, tested by Shannon-Wiener and evenness indices, showed different situations, confirming that there were statistically significant differences in all the three areas (Table 4 and Fig. 8) between road and interior plots. The highest diversity was found in road plots in the Amiata forest, while the lowest diversity was found in interior plots in the Caspian and Amiata forests. On average, the highest diversities found in the road plots, evenness from 0.81 to 0.96, showed very similar values to those of more complex forest ecosystems. The forest roads did not affect negatively the species diversity. A similar result was found by Tehrani et al. (2015).

5. Conclusion In this study, the effects of forest roads on forest tree species composition, structure and snag presence were assessed in three forest areas, two in Italy (Apennines and Alps) and one in Iran. The main objective of this study was to verify if, in productive and managed forests, the forest road network could represent not only an environmental risk, but also a possibility of increasing the biodiversity at landscape level. As demonstrated by the results, the edges of the forest road network are a fine mosaic composed of different trees (qualitative and quantitative), coupled with the moderate presence of dead trees. The answers to the specific questions posed in this study are detailed below: Þ the forest road affects the tree species diversity and there are significant differences between tree species diversity along the roads and inside the forest Þ there are significant differences between forest structure along the roads and inside the forest, mainly in terms of live trees and in part in terms of death trees Þ in the three areas, from the road edges to the interior forest, a similar taxonomic composition of forest community was found. The first main difference was related to the abundance of less shadow tolerant species along the road. The second main difference was related to the biodiversity indices, which are higher along the road, probably as a result of the »edge effect« (Brockerhoff et al. 2008). The main similarities in structure of live and dead trees are the result of the previous period of road construction Þ this methodology and these indexes could be used to develop further studies on larger geographical scales and to consider all the ecosystem aspects linked to the forest road network for a valid forest road planning.

6. References Akay, A.E., Sessions, J., 2004: Roading and transport operations, in: Burley, J., Evans, J., Youngquist, J., Encyclopedia of Forest Sciences (Eds), Elsevier Academic Press, Amsterdam, The Netherlands, 259–269. Avon, C., Bergès, L., Dumas, Y., Dupouey, J.L., 2010: Does the effect of forest roads extend a few meters or more into the adjacent forest? A study on understory plant diversity in managed oak stands. For. Ecol. Manag. 259(8): 1546–1555. Bergès, L., Chevalier, R., Avon, C., 2013: Influence of forest road, road-surfacing material and stand age on floristic diversity and composition in a nutrient-poor environment. Appl. Veg. Sci. 16(3): 470–479.

Croat. j. for. eng. 39(2018)1

Study of Forest Road Effect on Tree Community and Stand Structure in Three Italian and Iranian ... (57–70) R. Picchio et al. Bilby, R.E., Sullivan, K., Duncan, S.H., 1989: The generation and fate of road-surface sediment in forested watersheds in southwestern Washington. Forest Sci. 35(2): 453–468. Brockerhoff, E.G., Jactel, H., Parrotta, J.A., Quine, C.P., Sayer, J., 2008: Plantation forests and biodiversity: oxymoron or opportunity? Biodivers Conserv 17(5): 925–951. Cerdà, A., 2007: Soil water erosion on road embankments in eastern Spain. Sci. Total Environ. 378(1): 151–155. Chen, J., Franklin, J.F., Spies, T.A., 1992: Vegetation responses to edge environments in old-growth Douglas fir forests. Ecol Appl 2(4): 387–396. Coffin, A.W., 2007: From roadkill to road ecology: a review of the ecological effects of roads. J. Trans. Geography 15(5): 396– 406. Delgado, J.D., Arroyo, N.L., Arévalo, J.R., Fernández-Palacios, J.M., 2007: Edge effects of roads on temperature, light, canopy cover, and canopy height in laurel and pine forests (Tenerife, Canary Islands). Landscape Urban Plan 81(4): 328–340. Demir, M., Makineci, E., Yilmaz, E., 2007: Investigation of timber harvesting impacts on herbaceous cover, forest floor and surface soil properties on skid road in an oak (Quercus petrea L.) stand. Build. Environ. 42(3): 1194–1199. Dickie, A., Reich, P.B., 2005: Ectomycorrhizal fungal communities at forest edges. J. Ecol. 93(2): 244–255. Esseen, P., 1994: Tree mortality patterns after experimental fragmentation of an old-growth conifer forest. Biol. Conserv. 68(1): 19–28. European Commission, 2015: Natura 2000 and Forests. Part I–II, Technical Report, 2015–088. Luxembourg: Office for Official Publications of the European Communities European Environment Agency, 2015: State of Nature in the EU. Results from Reporting Under the Nature Directives 2007/2012. Vol. 2, EEA Technical Report. Luxembourg: European Environment Agency Ferreira, L.V., Laurance, W.F., 1997: Effects of forest fragmentation on mortality and damage of selected trees in central Amazonia. Conserv. Biol. 11(3): 797–801. Forman, R.T.T., Alexander, L.E., 1998: Roads and their major ecological effect. Annu. Rev. Ecol. Syst. 29(1): 207–231. Ganesh, T., Ganesan, R., Devy, M.S., Davidar, P., Bawa, K,. 1996: Assessment of Plant biodiversity at a mid elevation evergreen forest of Kalakad-Mundanthurai Tiger Reserve, Western Ghats, India. Curr. Sci. India 71: 379–392. Grigolato, S., Pellegrini, M., Cavalli, R,. 2013: Temporal analysis of the traffic loads on forest road networks. IForest 6(5): 255–261. Gumus, S., Acar, H.H., Toksoy, D., 2008: Functional forest road network planning by consideration of environmental impact assessment for wood harvesting. Environ. Monit. Assess. 142(1): 109–116. Heinimann, H.R., 1998: A computer model to differentiate skidder and cable yarder based road network concepts on steep slopes. J. For. Res. 3(1): 1–9. Heshmati, G.A., 2007: Vegetation characteristics of four ecological zones of Iran. J. Plant. Prod. 1(2): 215–224. Hill, M.O., 1973: Diversity and evenness: a unifying notation and its consequences. Ecology 54(2): 427–473.

Croat. j. for. eng. 39(2018)1

Kapos, V., 1989. Effects of isolation on the water status of forest patches in the Brazilian Amazon. J. Trop. Ecol. 5(2): 173–185. Krebs, C.J., 1999: Ecological Methodology. Harper and Row, New York. Laurance, W.F., Nascimento, H.E.M., Laurance, S.G., Andrade, A., Ewers, R.M., Harms, K.E., Luizão, R.C.C., Ribeiro, J.E., 2007: Habitat fragmentation, variable edge effects, and the landscapedivergence hypothesis. Plos ONE 2(10): e1017. Li, Y.H., Hu, Y.M., Chang, Y., Li, X.Z., Bu, R.C., Hu, C.H., Wang, C.L., 2010: Effect zone of forest road on plant species diversity in Great Hing’an Mountians. Chinese J. Appl. Ecol. 21(5): 1112– 1119. Magagnotti, N., Spinelli, R., Güldner, O., Erler, J., 2012: Site impact after motor-manual and mechanised thinning in Mediterranean pine plantations. Biosystems Engineering 113(2): 140– 147. Marchi, E., Picchio, R., Mederski, P.S., Vusić, D., Perugini, M., Venanzi, R., 2016: Impact of silvicultural treatment and forest operation on soil and regeneration in Mediterranean Turkey oak (Quercus cerris L.) coppice with standards. Ecol. Eng. 95: 475-484. Marchi, E., Picchio, R., Spinelli, R., Verani, S., Venanzi, R., Certini, G., 2014: Environmental impact assessment of different logging methods in pine forests thinning. Ecol. Eng. 70: 429–436. Mitsch, W., Jørgensen, S., 2003: Ecological engineering: a field whose time has come. Ecol. Eng. 20(5): 363–377. Murcia, C., 1995: Edge effects in fragmented forests: implications for conservation. Trends Ecol. Evol. 10(2): 58–62. Najafi, A., Torabi, M., Nowbakht, A.A., Moafi, M., Eslami, A., Sotoudeh Foumani, B., 2012: Effect of forest roads on adjacent tree regeneration in a mountainous forest. Ann. Biol. Res. 3(4): 1700–1703. Olander, L.P., Scatena, F.N., Silver, W.L., 1998: Impacts of disturbance initiated by road construction in a subtropical cloud forest in the Luquillo experimental forest, Puerto Rico. Forest Ecol. Manag. 109(1): 33–49. Olupot, W., 2009: A variable edge effect on trees of Bwindi Impenetrable National Park, Uganda, and its bearing on measurement parameters. Biol. Conserv. 142(4): 789–797. Osma, E., Özyiğit, İ.İ., Altay, V., Serin, M., 2010: Urban vascular flora and ecological characteristics of Kadıköy district, Istanbul, Turkey. Maejo Int. J. Sci. Technol. 4(1): 64–87. Ozcelik, R.A., Gul, A.U., Merganic, J., Merganicova, K., 2008: »Tree species diversity and its relationship to stand parameters and geomorphology features in the eastern Black sea region forests of turkey«. J. Envir. Biol. 29(3): 291–298. Parendes, L.A., Jones, J.A., 2000: Role of light availability and dispersal in exotic plant invasion along roads and streams in the H.J. Andrews Experimental Forest, Oregon. Conserv. Biol. 14(1): 64–75. Pellegrini, M., Grigolato, S., Cavalli, R., 2013: Spatial multi-criteria decision process to define maintenance priorities of forest road network: An application in the Italian alpine region. Croat. J. For. Eng. 34(1): 31–42. Picchio, R., Neri, F., Petrini, E., Verani, S., Marchi, E., Certini, G., 2012: Machinery-induced soil compaction in thinning of conifer stands. For. Ecol. Manag. 285: 38–43.

R. Picchio et al. Study of Forest Road Effect on Tree Community and Stand Structure in Three Italian and Iranian ... (57–70) Picchio, R., Spina, R., Maesano, M., Carbone, F., Lo Monaco, A., Marchi, E., 2011: Stumpage value in the short wood system for the conversion into high forest of a oak coppice. For. Studies China 13(4): 252–262. Pourbabaei, H., Abedi, R., 2013: Plant species groups in Chestnut (Castanea sativa Mill.) sites, Hyrcanian forests of Iran. Ecol. Balkanica 5(1): 37–47. Pourbabaei, H., Haddadi-Moghaddam, H., Begyom-Faghir, M., Abedi, T., 2013: The influence of gap size on plant species diversity and composition in beech (Fagus orientalis) forests, Ramsar, Mazandaran Province, North of Iran. Biodiversitas 14(2): 89–94. Prasad, A.E., 2009: Tree community change in a tropical dry forest: the role of roads and exotic plant invasion. Environ. Conserv. 36(3): 201–207. Ryan, T., Phillips, H., Ramsay, J., Dempsey, J., 2004: Forest Road Manual. Guidelines for the design, construction and management of forest roads. COFORD: Dublin Sitzia, T., Campagnaro, T., Grigolato, S., 2016: Ecological risk and accessibility analysis to assess the impact of roads under Habitats Directive. J. Env. Plan. Manag. 59(12): 2251–2271. Skaugset, A., Surfleet, C., Meadows, M., Amann, J., 2011: Evaluation of erosion prediction models for forest roads. Transport Res. Rec. 2203: 3–12.

Tarvirdizadeh, H., Nikooy, M., Pourbabaei, H., Naghdi, R., 2014: Effects of road construction on biodiversity and composition of herbaceous species cover, asalem forest, northern Iran. Forestry Ideas 20(2): 157–169. Tavankar, F., 2015: Structure of natural Juniperus excelsa stands in Northwest of Iran. Biodiversitas 16(2): 161–167. Tavankar, F., Nikooy, M., Picchio, R., Venanzi, R., Lo Monaco, A., 2017: Long-term effects of single-tree selection cutting management on coarse woody debris in natural mixed beech stands in the caspian forest (Iran). IForest 10(3): 652–658. Tavankar, F., Picchio, R., Lo Monaco, A., Bonyad, A., 2014: Forest management and snag characteristics in Northern Iran lowland forests. J. For. Sci. 60(10): 431–441. Tehrani, F.B., Majnounian, B., Abdi, E., Zahedi Amiri, G., 2015: Impacts of forest road on plant species diversity in a Hyrcanian Forest, Iran. Croat. J. For. Eng. 36(1): 63–71. Watkins, R.Z., Chen, J., Pickens, J., Brosofske, K.D., 2003: Effects of Forest Roads on Understory Plants in a Managed Hardwood Landscape. Conserv Biol 17(2): 411–419. Williams-Linera, G., 1990: Vegetation structure and environmental conditions of forest edges in Panama. J. Ecol. 78(2): 356–373.

Authors’ addresses: Prof. Rodolfo Picchio, PhD.* e-mail: r.picchio@unitus.it Rachele Venanzi, MSc. e-mail: venanzi@unitus.it Prof. Angela Lo Monaco e-mail: lomonaco@unitus.it University of Tuscia Department of Agriculture and Forest Sciences (DAFNE) Via S. Camillo de Lellis 01100 Viterbo ITALY Assist. prof. Farzam Tavankar, PhD. e-mail: tavankar@aukh.ac.ir Islamic Azad University Khalkhal Branch Department of Forestry Khalkhal IRAN

Received: October 20, 2016 Accepted: January 16, 2017

Assoc. prof. Mehrdad Nikooy, PhD. e-mail: nikooy@guilan.ac.ir University of Guilan Faculty of Natural Resources Department of Forestry Somesara IRAN * Corresponding author Croat. j. for. eng. 39(2018)1

Original scientific paper

Middle-Term Сhanges in Topsoils Properties on Skidding Trails and Cutting Strips after Long-Gradual Cutting: a Case Study in the Boreal Forest of the North-East of Russia Aleksey Ilintsev, Elena Nakvasina, Aleksey Aleynikov, Sergey Tretyakov, Sergey Koptev, Alexander Bogdanov Abstract In this work, physical and chemical properties of the upper horizons of podzolic light loamy soil were investigated 21–23 years after forest cutting. This was after the first shift of long-term, gradual felling was carried out by tree-length logging in wintertime in mixed conifer stands of the Middle Taiga of the Arkhangelsk Region in Russia. The increased density of the forest litter composition was observed. This was especially the case on skidding trails. On the forest floor of skidding trails subjected to a greater stress caused by timber skidding, lower total porosity and aeration porosity was observed, in comparison with the cutting strip and natural forest. It was established that timber skidding during wintertime does not affect the density of podzolic horizon composition. An inverse pattern was observed here: the total porosity and the aeration porosity became higher and were close to the optimum values for plant growth (54.16–52.99% and 15.72–19.97%). In the podzolic horizon on skid roads, comparison to the natural forest showed a significant reduction of phosphorus mobile forms and an increase in the amount of absorbed bases, which is the result of grassy vegetation overgrowth and natural birch regeneration. On skidding trails and cutting strips, the organic matter content and total nitrogen significantly increased, which is related to a change of light intensity, the composition of living ground cover and vigorous decompositions of the organic horizon and woody residues. In cutting areas, a system mosaic of soil cover developed, which differed according to favourable conditions for tree species regeneration, compared to the control stands. Keywords: boreal forest, forest soil disturbance, skidding trails, cutting strips, long-gradual cuttings

1. Introduction Different systems of forest cuttings, carried out at a modern technical level, have a significant impact on forest ecosystems (Dymov and Milanovskii 2014, Cambi et al. 2015a, Puettmann et al. 2015). The consequences of soil and forest floor infringements are not only synchronous, but also have a long-term nature, which reveals itself over several decades after logging (Modry and Hubeny 2003, Rozhkov and Karpachevskii 2006), and in some cases have irreversCroat. j. for. eng. 39(2018)1

ible consequences (Hartmann et al. 2014, Klaes et al. 2016). The main share of the workload falls on forest soils that are susceptible to improper forest management and, in particular, to large-scale harvesting (Cambi et al. 2015a). The bigger the cutting area, the greater are its consequences for the surrounding forest communities. This can cause changes in the microclimate, composition, abundance, and ecology of plants and animals. It is known that the effect of selective cutting is much weaker than that of clear cutting (Pobedinskii 2013).

A. Ilintsev et al.

Middle-Term Сhanges in Topsoils Properties on Skidding Trails and Cutting Strips ... (71–83)

Logging is mostly carried out by three main systems: Full-Tree (FT), Tree-Length (TL) and Cut-ToLength (CTL) (Karvinen et al. 2006). Each system has its own specific features, which depend on natural and production conditions, technology used and the share of manual operations in the overall process (Gerasimov and Sokolov 2014). In the study region, the proportion of CTL–system is constantly increasing every year, as it has the best efficiency and less impact on the forest environment (Goltsev et al. 2011, Derbin and Derbin 2016). The same trends are observed in Europe and Scandinavia (Leinonen 2004). In the study area, TL–logging is also a traditional forest harvesting system. This technology is used with chain saws or feller-buncher machines and skidding tractors (skidders). In Russia, 26% of harvested wood is transported to the intermediate warehouses in stems. Meanwhile, in the United States and Canada, the TL –system is continuously improved, its effectiveness increases, and opportunities for further development are visible (Sukhanov 2012). According to various estimates, in the United States, the TL–system ranges from 15 to 85% (Hartsough et al. 1997, Leinonen 2004), in Canada – 85% (Sukhanov 2012). A wide range of domestic and foreign equipment, wheeled and tracked vehicles, such as harvesters, forwarders, skidders, appears at the logging sites. The degree of impact on soils depends on the type of technology used (Picchio et al. 2012, Marchi et al. 2014, Cambi et al. 2015b). As a rule, stems or logs are delivered to the intermediate warehouses by skidding or forwarding, meaning that vehicles travel on skidding trails. Thus, as a result of heavy machinery movement (harvesters, forwarders, tractors), areas such as skidding trails and wood loading areas are subjected to repeated impact, while the cutting strips are affected to a lesser extent. In recent years, these vehicles are becoming more powerful and economical, but they are also having negative impacts on the soil (Vossbrink and Horn 2004, Horn et al. 2007). The skidding stems stripped forest floor and sometimes organogenic horizon form the compacted and mixed sites. When harvesting, the soil morphology is disturbed in the same as in the case of natural fall-outs of the forests (Karpachevsky 1981). Changing conditions of soil formation during logging activities affect, in varying degrees, the physical, chemical and biological soil properties (Standish et al. 1988, Worrell and Hampson 1997, Powers et al. 2005, Zetterberg et al. 2013, Osman 2013) and composition of the soil cover. Soil properties often largely depend on silvicultural treatments and logging operations, and this may imply soil compaction and consequent restrictions to tree growth and

natural regeneration. (Venanzi et al. 2016, Marchi et al. 2016). Soil variations, related to forest harvesting operations, can lead to changes in biogeochemical cycles that affect soil ecosystems (Cambi et al. 2017a). The physical and also morphological properties of the soils are seriously violated on skidding trails and loading sites. A large-scale study in different regions of Russia showed a significant change of water and soil physical properties on clear cutting areas with heavy loam and clay soils (Pobedinskii 2013). Substantial changes of the physical properties of the soils are observed in cuttings areas to a depth of 50–60 cm, and on wood loading sites up to 90 cm (Dymov and Lapteva 2006); changes of the chemical properties are observed up to 30–40 cm (Fedorets and Bahmet 2003). The generally accepted criterion for assessing the impact of logging equipment on soil is the change in its density and associated air, thermal and water regimes of the soil, which affect the soil organisms and plants and have a negative impact on soil properties and forest productivity (Kozlowski 1999, McNabb et al. 2001, Ares et al. 2005, Agherkakli et al. 2010, Cambi et al. 2015a, Cambi et al. 2017b). Many authors (Brais 2001, Akay et al. 2007, Bagheri et al. 2011) noted that compaction, caused by forestry machines, is one of the main causes of soil degradation. The greatest increase in soil compaction occurs at a depth of 10 cm, followed by 20 cm and 30 cm soil layers, respectively (Akay et al. 2007). After logging, there are significant changes in the nutrients structure in forest floor and upper mineral soil horizons (Fedorets and Bahmet 2003). Clearings of coniferous in boreal forests activated the formation process of podzolic, but after the growth of soft-wooded broadleaved species (birch and aspen) over time an advanced stage of the turf process was observed. The chemical structure of litter also changes due to different species composition of vegetation, dominated by meadow and weeds that appear in forest stands exposed to infringement or in the stand »windows«. These differences determine the character of the litter decomposition, the microflora structure and its activity. Thus, various types of logging can affect forest soils by altering their properties. It was assumed that: Þ natural processes of recovery of the upper soil horizon properties will be found two decades after long-term gradual felling carried out in wintertime, but not restored to a natural state of untouched soils Þ these processes will be different in cutting strips and in skidding trails Þ the impact on skidding roads will be more considerable than in cutting strips Croat. j. for. eng. 39(2018)1

Middle-Term Сhanges in Topsoils Properties on Skidding Trails and Cutting Strips ... (71–83)

Þ the effect on mineral soil horizon will be minimal, both in cutting strips and in skidding roads, as the felling was carried out with a steady snow cover and frozen soil.

2. Material and methods 2.1 Site description and logging method The research was carried out in the North of the European part of Russia on the territory of the Arkhangelsk Region in forest department of North Arctic Fed-

A. Ilintsev et al.

eral University. The object of the study is located in the central part of the Arkhangelsk Region, in Plesetsk administrative district and belongs to the Middle Taiga according to the forest zonation (Kurnaev 1973). The geographical position of the territory is from 62°55’ to 63°10’ North latitude and from 40°15’ to 40°40’ East longitude from Greenwich (Fig. 1). The climate of the research area is temperate – continental, formed in a small amount of solar radiation in wintertime, under the influence of the Nordic Seas and the intense western removal of moist air masses from the Atlantic ocean (in summer – cold, winter –

Fig. 1 Location of study area and experimental Long-Gradual Cuttings Croat. j. for. eng. 39(2018)1

A. Ilintsev et al.

Middle-Term Сhanges in Topsoils Properties on Skidding Trails and Cutting Strips ... (71–83)

warm), and also influenced by local physical-geographical characteristics of the territory. A feature of the climate is a frequent change of air masses of different origin. The average annual air temperature is 0.4°C. The average temperature of the warmest month (July) is +16.1°C, and of the coldest (January) –14.1°C. The annual rainfall is in the range of 380–690 mm, which contributes to the excessive soils moisture, but the natural karst drainage of the study area ensures removal of excess moisture. Winter precipitation occurs mostly in solid form. There is a snow cover from October to April. The average height of the snow cover during winter in the study area is 75–85 cm. The average relative humidity varies during the year from 67% to 87%. Such a high humidity is due to the relative proximity of the seas, numerous rivers, lakes and especially wetlands. The relative proximity of the seas,

numerous rivers, lakes and especially swamps contribute to such high humidity. Average annual wind speeds vary from 3–5 m s-1 to 7–8 m s-1. The territory of the study area is characterized by widespread Retisols (94%), less – Gleyic Retisols (5%) and Gleysols (1%), varying in genesis and production values. The upper genetic horizons have fine-textured composition (sandy, loamy, sandy, sandy loams), and the bottom is composed of loams and clay loams. Coniferous pine (Pinus sylvestris L.) and spruce (Picea abies /L./ H.Karst) forests, and occasionally larch (Larix sibirica Ledeb.) and fir (Abies sibirica Ldb.), dominate the study area. Deciduous forests are represented by birch (Betula pendula Roth.), aspen (Populus tremula L.), alder (Alnus incana L.) and willow (Salix caprea L.). Deciduous forests are represented by birch, aspen, alder and willow. The share of coniferous tree

Table 1 Brief characteristics of experimental objects (smooth surface, no slope data) Indicator

Long-term gradual cutting

Stand untouched by felling

cutting strip

skidding trail

57P28S9B6L

55P28S12L5B

–

150

155

–

24/31

24/28

–

Growing stock, m3 ha–1

539

364

–

–1

968

536

–

94S3P3B

62S30B8P sin. L.

79B17S3P1L

Average age of undergrowth, years

Average height, m

1.3

1.5

2.2

1013±102

3747±598

12160±1310

Families

Types

Species

Dominant family (number of species)

Poaceae (3), Orchidaceae (3), Pyrolaceae (2), Ericaceae (2), Scrophulariaceae (2)

Asteraceae (3), Orchidaceae (3), Poaceae (2), Ericaceae (2), Scrophulariaceae (2), Rosaceae (2)

Poaceae (5), Asteraceae (3), Scrophulariaceae (2), Ericaceae (2)

Herbs proportion with turf life form, %

10.7

9.1

21.7

Forest stands Stand composition*, % Average age of stand, years Average height, m / Average diameter, cm

Stand density, unit ha

Undergrowth Undergrowth composition*, %

–1

Undergrowth density, unit ha

Grass-shrub layer Projective cover of herb-dwarf shrub layer, % Taxa number, unit

*p – pine, s – spruce, b – birch, L – larch

Croat. j. for. eng. 39(2018)1

Middle-Term Сhanges in Topsoils Properties on Skidding Trails and Cutting Strips ... (71–83)

species in the study region is 82.4%, the share of softwood is 17.6%. The experimental long-term, gradual felling was carried out in 1993 and 1995 in uneven age (from 65 to 202 years) mixed pine-spruce stands growing on Haplic Abruptic Retisols (Loamic) (IUSS Working Group 2015). Logging was done in wintertime, with frozen soil and a steady snow cover. A stand untouched by felling (control) adjoins the western side of the longterm gradual felling and represents original characteristics of harvested stands (Table 1). The cutting area is designed according to mid-size cutting strips technology; the width of skidding trails does not exceed 5 m, and the width of cutting strips is 30 m. The layout of the trails is perpendicular. The large pine, spruce and birch trees were selected in cutting strips according to the target diameter, and larch trees were cut only on skidding trails. The intensity of the long-term, gradual felling in 1993 was 30% of the growing stock, and long-term, gradual felling in 1995 was 40%. Tree felling was performed with the apex on the trail route at an angle of 60° using chain saws. After that, branches and treetops were cut. Cleaning of felling areas was performed by laying slash residues on skidding trails. Skidding of tree tops was carried out by using cable skidder LTT-55A (mean ground pressure of 50 kPa and 70 kW engine power) with an empty mass of 5800 kg; it moved strictly on skidding trails.

A. Ilintsev et al.

Where: M moisture content, % WD dry soil (forest floor) weight, g WW moist soil (forest floor) weight, g. The composition density was calculated according to the following formula:

Db =

WD V

Where: Db bulk density, g cm-3 V volume of cylinder or volume of frame template, g cm-3. The density of the solid phase (particle density) was obtained by pycnometer method. The composition density and density of the soil solid phase were used to calculate the total porosity, which was determined by the following formula:



ϕ=  1 − 

Db   × 100 Dd 

Aeration porosity (the share of large pores occupied by air) calculated by the following formula:

ϕa =ϕ − M × Db

Samples of forest floor and soil podzolic horizon were collected in the natural stand, in the cutting strips and skidding trails of stands, after long-gradual cuttings. In 2016, 200 samples of forest floor and soil podzolic horizon were collected – 40 in the control, 80 in cutting strips and 80 in skidding trails. For determining the density of the forest floor, sampling was performed by using frame templates (area of 100 and 144 cm2). Samples of the podzolic horizon were collect ed using rigid metallic cylinders (volume of 52.78 cm3), after removing the forest floor. All samples were weighed on an analytical balance (»moist weight«) (Nakvasina et al. 2007).

Where: ja is the aeration porosity, %.

In the laboratory, the samples were dried at 105°C for 24 h to constant weight (»dry weight«). Physical characteristics were determined by laboratory analyses. Field soil moisture was determined by the following formula:

( WW − WD )

Croat. j. for. eng. 39(2018)1

(1)

(3)

Where: j total porosity, % Dd particle density, g cm-3.

2.2 Soil sampling

2.3 Physical analysis

(2)

(4)

2.4 Chemical analysis To study the chemical characteristics, podzolic horizon was investigated as the first mineral horizon under trees. In this horizon, the main part of sucking roots is located, so changes in the chemical characteristics have a significant impact on further growth and development of plants. Chemical properties were determined for 5 randomly selected samples for each research subject. In total, 25 specimens were studied. The following parameters were evaluated: the amount of phosphorus (P2O5) mobile forms and potassium (K2O), sum of absorbed bases, amount of soil organic matter (C), acidity (pH), hydrological acidity, total nitrogen (N) amount. The methods, generally accepted in Russia and confirmed by state standards, were used. Acidity (pH) was determined by a potassium chloride (KCl) solution in the concentration of 1 mol dm-3 at a 1:2.5 ratio of soil to solution, and pH potentiometric was determined using a pH meter with glass electrodes. Hydrolytic acidity was determined (CH3COONa) in

A. Ilintsev et al.

Middle-Term Сhanges in Topsoils Properties on Skidding Trails and Cutting Strips ... (71–83)

the concentration of 1 mol dm-3 at a 1:2.5 ratio of soil to solution with aliquots of extracts titrated with 0.01 M NaOH. Mobile compounds of phosphorus (P2O5) and potassium (K2O) were removed from the soil by the solution of hydrochloric acid (HCl) (extracting solution) in the molar concentration of 0.2 mol dm-3. Phosphorus (P2O5) mobile compounds were then determined quantitatively by a photoelectric colorimeter, while potassium (K2O) was determined by a flame photometer. The amount of absorbed bases is determined in accordance with Kappen method. Soil organic matter was determined by Tyurin method using a photoelectrocolorimeter. Total nitrogen was determined according to the semi-micro-Kjeldahl method.

2.5 Statistical analysis Statistical analysis of data normality of physical and chemical properties was preceded by KolmogorovSmirnov and Sapiro-Wilk tests. The homogeneity of the samples was established by using distributions kurtosis. The data obtained were analyzed by using the STATISTICA® ver. 6.1 (StatSoft Russia). To establish the differences between two mean values, the independent t-test at a 0.05 significance level was applied.

3. Results 3.1 Physical properties of forest floor and podzolic horizon Test results of distribution normality showed that the analyzed distributions do not differ from normal.

The Kolmogorov-Smirnov criterion varies from 0.10 to 0.15 (p>0.20), and the Sapiro-Wilk criterion varies from 0.95 to 0.98 (p>0.05–0.80). The kurtosis of curves distribution ranges from –1.00 to 2.20 and does not exceed the empirical limit (–2). This suggests that the statistical sets are homogeneous. The capacity of the forest floor in the control of the undisturbed stand corresponds to the average data capacity of the forest floor for the middle taiga of Arkhangelsk Region (Table 2). The capacity of podzolic horizon significantly differs from the average horizon size of podzol soils type for midle taiga (Tab. 3). However, this difference is due to the soil genesis and variability within a stands soil cover, which corresponds to distribution limits (3.0 and 12.0), especially on the minimum size of the podzolic horizon often associated with the confinement to large trees. 21–23 years after the first shift of long-gradual felling in wintertime, the capacity of the forest floor on skidding trails is significantly different from the control values (t0.05=3.5), and in the cutting strips, the average power corresponds to the natural stands (t0.05=0.86). At the same time, increased density is noted on skidding trails compared to the control stand (t0.05=–3.48), despite the fact that the logging was carried out during winter and skidding trails were fortified by wood residues. An increased density is also observed in the cutting strips (t0.05=–2.14), but t-test coefficient is on the border of the confidence interval. The sealing effect of the equipment on the podzolic horizon in the cutting strips and skidding trails was not identified. The change in the total porosity and aeration porosity of

Table 2 Physical characteristics of forest floor and podzolic horizon at the experimental sites (mean ± standard deviation); The average data for mixed pine-spruce forests growing on podzolic soils are given, according to materials for 25 soil profiles (Sklyarov and Sharova, 1970)1; Statistically significant differences at the level of 0.05 between the experimental objects after independent t-test are marked with an asterisk Experimental site

Horizon

Thickness, cm

Bulk density, g cm–3

Particle density, g cm–3

Total porosity, %

Aeration porosity, %

Natural stands Average data1

Control area

5.24±0.28

–

6.44±0.40

–

4.91±0.21

0.071±0.003

1.433

95.05±0,24

77.80±1.67

4.85±0.31*

1.240±0.030

2.443

49.23±1,22

13.74±1.60

Long-gradual cutting 1993 and 1995 Cutting strip

Skidding trails

4.98±0.17

0.080±0.003*

1.366

94.03±0.25*

77.50±1.01

6.05±0.34

1.161±0.021*

2.533

54.16±0.87*

19.97±1.14*

4.17±0.16*

0.118±0.010*

1.544

92.29±0.62*

64.3±1.57*

5.85±0.34

1.176±0.021

2.503

52.99±0.86*

15.72±0.99

Croat. j. for. eng. 39(2018)1

Middle-Term Сhanges in Topsoils Properties on Skidding Trails and Cutting Strips ... (71–83)

Thickness

Cutting strip Skidding trails Cutting strip

Bulk density Skidding trails Cutting strip Total porosity Skidding trails Cutting strip Aeration porosity Skidding trails

t-value

p-value

Control area

Horizon

site

Parameters

Experimental

Table 3 Results of t-test for independent samples (physical charac teristics)

0.91

0.369413

2.95

0.005185*

0.86

0.390818

0.72

0.473653

3.50

0.000857*

1.08

0.286345

–2.14

0.042101*

2.13

0.037457*

–3.48

0.000966*

1.70

0.095379

2.55

0.013411*

–3.28

0.001775*

3.05

0.003434*

–2.51

0.014771*

0.14

0.886149

–3.16

0.002488*

4.36

0.000066*

–1.10

0.275953

*p values less than 0.05

the upper soil horizons during the formation of cenoses (after cutting) is related to changes of life forms in ground cover and a large amount of birch trees (Table 1). Soft birch litter changes the humification processes in forest litter, compared to natural stands, where litter consists mainly of mosses and conifer litter. The in-

A. Ilintsev et al.

crease in total porosity in the podzolic horizon is rather associated with the appearance of grass having a fibrous root system. The proportion of such grasses is particularly high in skidding trails (21.7%). On skidding trails, significantly lower values of the total porosity and aeration porosity of the forest floor are observed compared to the background values (Table 3). Packing of forest floor, when cutting and skidding, are not compensated by disintegrating effect of the herbs and soft-leaved breeds litter. In the podzolic horizon, inverse pattern is observed: the total porosity and aeration porosity become higher and are close to the optimum values for plant growth. The technology impact on the horizon overlaps by buffer horizon (forest floor).

3.2 Chemical properties of podzolic horizons Data analysis for distribution normality showed that the studied distributions do not differ from normal, the Kolmogorov-Smirnov criterion varies from 0.15 to 0.24 (p>0.20), and the Sapiro-Wilk criterion varies from 0.91 to 0.98 (p>0.05–0.8). The kurtosis distribution curves vary from 1.96 to 0.30 and the curves do not break up into two separate curves indicating the population homogeneity. In the podzolic horizon (Table 4 and Table 5), 21–23 years after long-term gradual felling, the migration of mobile forms of phosphorus (t0.05 = 7.62) is increasing in skidding trails, while maintaining the amount of mobile potassium (t0.05 = –1.08) and environmental reac tion (pH=3.1). However, the increase in the concentration of hydrogen ions is reflected in the hydrolytic acidity increase (t0.05 = –6.48). At the same time, in the podzolic horizon on skidding trails and cutting strips, there is a significant increase in the quantity of organic matter (t0.05= –7.64 and –5.72) and total nitrogen (t0.05= –2.97 and –2.59), associated with the illumination

Table 4 Chemical characteristics of podzolic horizon at the experimental sites (mean ± standard deviation); The average data for mixed pine-spruce forests growing on podzolic soils are given, according to materials for 25 soil profiles (Sklyarov and Sharova, 1970)1; Statistically significant differences at the level of 0.05 between the experimental objects after independent t-test are marked with an asterisk Experimental site

Contents

Total base absorption

Organic C %

pH (KCl)

Hydrolytic acidity

Total N

mg-eq per 100 g

C/N ratio

P2O5, mg 100 g–1

K2O, mg 100 g–1

mg-eq per 100 g

–

0.76±0.06

1.49±0.07

3.6

6.69±0.09

0.051±0.005

29.21

15.06±0.21

2.76±0.11

0.54±0.07*

1.61±0.01

3.0

8.80±0.35*

0.057±0.001

28.05

Natural stands Average data1 Control area

Long-gradual cutting 1993 and 1995 Cutting strip

15.53±2.19

2.97±0.12

0.91±0.05

2.11±0.04*

3.0

10.60±0.94*

0.072±0.003*

29.31

Skidding trails

9.33±0.51*

2.97±0.12

1.42±0.12*

2.30±0.02*

3.1

10.44±0.13*

0,075±0.001*

30.67

Croat. j. for. eng. 39(2018)1

A. Ilintsev et al.

Middle-Term Сhanges in Topsoils Properties on Skidding Trails and Cutting Strips ... (71–83)

Table 5 Results of t-test for independent samples (chemical characteristics) Parameters P2O5 K2O

Total base absorption

Organic C

Hydrolytic acidity

Total N

Experimental site

Horizon

t-value

p-value

Cutting strip

–0.15

0.884068

Skidding trails

7.62

0.000004*

Cutting strip

–1.08

0.299011

Skidding trails

–1.08

0.299011

Control area

2.38

0.018268*

Cutting strip

–1.56

0.128598

Skidding trails

–5.51

0.000004*

Control area

–0.82

0.419425

Cutting strip

–5.72

0.000002*

Skidding trails

–7.64

0.000000*

Control area

–2.58

0.015371*

Cutting strip

–4.81

0.000032*

Skidding trails

–6.48

0.000000*

Control area

–0.58

0.565015

Cutting strip

–2.59

0.014319*

Skidding trails

–2.97

0.005494*

*p values less than 0.05

change, composition of ground cover and vigorous decomposition of the organic horizon and woody residues. The C/N ratio in the soil on skidding trails is increased, reaching 30.69 vs. 28.05–29.21 in natural stands, which generally means reducing the enrichment of humic substances with nitrogen in this microsite. Changing of soil formation conditions related to the illumination and changes in vegetation leads to an increase in the amount of absorbed bases and soil saturation level in the podzolic horizon, which is especially evident on skidding trails. After 21–23 years, the sum of exchange bases is 3 times higher in podzolic horizon on skidding trails than the background values in natural stands, and the level of saturation with bases – 2 times higher, reaching values not typical for the natural soil (1.42 mmol per 100 g of soil and 12%, respectively).

4. Discussion The analyzed physical properties (composition density, solid phase density, total porosity, aeration porosity) are the most illustrative for assessing the impact of logging equipment on forest soils (Cambi et al. 2015a). For example, the main reason for poor plant growth is the nutrients unavailability, as by increasing the composition density, pores and humidity are reduces, which affect the plants root system (Šušnjar et al. 2006). Recent studies (Cambi et al. 2017b) showed

that limited access and acquisition of nutrients and water due to the shorter length of main root likely played a key role for the growth and physiological responses to soil compaction in Q. robur seedlings. Reduction of soil compaction and total porosity are the inevitable consequences in skidding places, and can vary in intensity and distribution as the result of interaction between machine and local factors during timber harvesting. The impact level depends on many factors, such as the number of skidder passes, skidding track tilt, places, characteristics, logging equipment, skid trails location and harvesting season (Laffan et al. 2001, Demir et al. 2007, Najafi et al. 2009, Solgi and Najafi 2014). Thus, soil compaction can lead to mass reduction of roots in the upper soil horizons (Karpechko 2008), which is visible as density is increased by 15% or more (Page-Dumroese et al. 1998). However, when logging in wintertime, when the soil is frozen and covered with snow, the impact of logging machines on soil and roots is considerably lower. We found saving seals of the forest floor two decades after cutting, which affects the reduction of total porosity in cutting strips and skidding trails, which is in agreement with the opinion of several authors (Ares et al. 2005, Ampoorter et al. 2007, Picchio et al. 2012). It may be due to the change in the number of macropores (Seixas and McDonald 1997, Ampoorter et al. 2007). Croat. j. for. eng. 39(2018)1

Middle-Term Сhanges in Topsoils Properties on Skidding Trails and Cutting Strips ... (71–83)

The impact of logging equipment that lasted for 21–23 years, after the cuttings, is not included into the density composition of podzolic horizon, because there was no direct exposure to skidding equipment on the roads in wintertime. However, the density of podzolic horizon composition at the studied sites is less than 1.4 g cm-3, which indicates the disposition of the horizon to compaction (Powers et al. 2005). Similar results were obtained when logging in cold weather (–20°С) with stable snow cover in the beech forests of the Czech Republic (Modry and Hubeny 2003). Frozen ground and snow cover provide the best carrying capacity (Ballard 2000). Trail works in wintertime on frozen soil are more effective and cause less damage to the soil surface (Šušnjar et al. 2006). The aeration porosity of podzolic horizon in all studied sites is low (13.74–19.97%). It is a dynamic component of the soil and changes depending on external factors. In our experience, first of all, a low proportion of pores saturated with air is related to rain, which passed before samples were collecting. The average aeration porosity for loamy mineral soil is 25% (Osman 2013). When the aeration porosity is reduced to 15%, the reduction in root growth is noted (Richards and Cockroft 1974). The trees roots typically operate at oxygen levels over 10% (Kozlowski 1985). Logging can lead to the disturbance of upper soil horizons, with subsequent increase in mineralization and certain nutrients leaching (phosphorus, potassium, etc.), which usually lasts for 2–5 years. Later, there may be changes in the soils nutrient regime, as this process is reversible (Worrell and Hampson 1997) and related to changes in site vegetation. 20 years after logging on skidding trails, a significant reduction of phosphorus mobile forms has been observed in podzolic horizon, amounting to 9.33 mg 100 g-1 compared to 15.53 mg 100 g–1 in the control. At the same time, the amount of mobile potassium remains quite stable at 2.97 mg 100 g–1 (2.76 mg 100 g–1 in the control). Other authors (Naghdi et al. 2016) similarly show that, on skidding trails, after the summer works depending on the number of skidder passes and slope, the reduction of phosphorus ranges from 11.5–44.5% and of potassium from 8.7–51.4% of the undisturbed area. The mobility of phosphorus is caused by acidic environment (pH=3.0) (Lambers et al. 2011). Based on the acidity degree, the studied sites are highly acidic (pH=3.0–3.1). Most forest soils have a pH of 3 to 7 (Osman 2013). Our research has shown that the environmental reaction in stands continues after long-term gradual felling. Similar results have been obtained for winter logging (Modry and Hubeny 2003). Croat. j. for. eng. 39(2018)1

A. Ilintsev et al.

The increase in the amount of soil organic matter is observed in the podzolic horizon both in cutting strips and skidding trails (2.11–2.30%), compared to the control stand (1.61%). This is related to changes in the species composition of lower storeys vegetation, increasing in grass and leafy litter, soil moisture and forest floor decomposition. As noted (Aragon et al. 2000, Arthur et al. 2013), soil organic matter acts as a link in forest soils, at least in the upper layer of the soil. Change and movement of organic matter in the soil profile occur in cutting areas (Dymov and Lapteva 2006, Dymov and Milanovskii 2014), which may have negative consequences for soil structure and result in susceptibility to compaction (Cambi et al. 2015a). Mobility of humic substances contributes to the eluviations process. However, some studies show that, 17 years after wood harvesting, depletion of C and N were observed in cutting place compared to the control area (McLaughlin and Phillips 2006). The amount of organic matter in forest soils is usually 1–5% of the dry weight and it decreases with depth. Under natural conditions, the organic matter in the soil is stable, but when the balance of forest ecosystems is infringed, this indicator can change (Osman 2013). It is known that organic matter is closely related to gross nitrogen in soils. In cutting strips and on skidding trails, 21–23 years after long-gradual felling, a larger amount of total nitrogen has been observed in the podzolic horizon (0.072–0.075%), compared to the control (0.057%). The source of total nitrogen in the soil is mainly the forest floor material (Fedorets and Bahmet 2003). The main reason is the increase of mosses and herbaceous protective cover. So, on skidding trails, there are some kinds of herbs that are not marked in natural plantings (Trifolium repens, Ranunculus acris, Lathyrus pratensis, Deschampsia cespitosa, Calamagrostis epigeios, Chamaenerion angustifolium), and there is an increase in birch litter, which contained more nitrogen than the needles (Marschner 2012, Osman 2013). Approximately one-third of the total nitrogen consumed by plants over the growing period is returned with litter (Fedorets and Bahmet 2003). C:N relationship indicates the nitrogen amount in the humus. The optimal value is 10 (Fedorets and Bahmet 2003). Changes of soil properties and vegetation composition have led to C:N ratio conversion. Its increase is observed in the skidding trails (30.67) compared to natural stands (29.21), which suggests that mineralization is slower, and therefore a small amount of mineral nitrogen is produced. An inverse relationship was established (Bolat et al. 2015) on the skid road and tractor road in oak forests of Turkey, where there was a lower C:N after a summer harvesting performed to reduce the vegetation cover.

A. Ilintsev et al.

Middle-Term Сhanges in Topsoils Properties on Skidding Trails and Cutting Strips ... (71–83)

Skidding trails, with clear cutting of trees, and wood residuals stowage have undergone significant changes in the formation of new stand, compared to the cutting strips. Therefore, the processes of vegetation formation and evolution of soil formation differ markedly. As a result, a system of sites with different vegetation and soil properties is formed on the cutting area. Irregular stands are created (system diversity of vegetation and forest floor), which persist for a long period (at least 20 years after logging).

5. Conclusions Our research shows that long-term gradual cutting by tree-length system in wintertime in mixed conifer stands on podzolic soils in the middle taiga, on skidding trails and cutting strips, results in changes in physical and chemical properties of upper soil horizons, which persist for more than two decades. Significant differences compared to native stand and cutting strip have been observed in skidding trails, subjected to the repeated passage of heavy equipment transporting trees. The site with trees completely cut down is overgrown usually by birch and by grasses that change the processes of soil formation. The forest floor in the cutting strips and skidding trails maintains the increased density, and the reduced total and aeration porosity. Podzol horizon lying under the forest floor is much less affected. The effect on the podzolic horizon is contrary to the effect on the forest floor: total and aeration porosity becomes higher and close to the optimum values for the growth of plants. The concentration of hydrogen ions increases, which is reflected in the raising of hydrolytic acidity, the amount of organic matter and total nitrogen. In the podzolic horizon on the cutting strips, migration of phosphorus mobile forms is enhanced, while the mobile potassium and environmental reaction maintain the same values. The amount of absorbed bases and soil saturation level are increased to values not characteristic for natural coniferous stands of the middle taiga. The vegetation diversity and soil cover, formed as a result of logging, create a system mosaic of sites that should be considered when monitoring and implementing economic activities in the stands, after longgradual cuttings. Further research on the change and recovery dynamics of physical and chemical characteristics of forest soils over time after logging is needed to obtain a more complete understanding of the soil and vegetation self-regeneration processes in cutting areas.

Acknowledgements The authors would like to acknowledge the financial support of the Russian Foundation for Basic Research (RFBR) for this work according to the research projects №16-34-50130 and №17-44-290127. We thank anonymous reviewers for constructive comments on the manuscript.

6. References Agherkakli, B., Najafi A., Sadeghi, S., 2010: Ground based operation effects on soil disturbance by steel tracked skidder in a steep slope of forest. Journal of Forest Science 56(6): 278–284. Akay, A. E., Yuksel, A., Reis, M., Tutus, A., 2007: The impacts of ground-based logging equipment on forest soil. Polish Journal of Environmental Studies 16(3): 371–376. Ampoorter, E., Goris, R., Cornelis, W. M., Verheyen, K., 2007: Impact of mechanized logging on compaction status of sandy forest soils. Forest Ecology and Management 241(1–3): 162–174. Aragon A., Garcia, M. G., Filgueira, R. R., Pachepsky, Y. A., 2000: Maximum compactibility of Argentine soils from the Proctor test: The relationship with organic matter and water content. Soil and Tillage Research 56(3–4): 197–204. Ares, A., Terry, T., Miller, R., Anderson, H., Flaming, B., 2005: Ground-based forest harvesting effects on soil physical properties and Douglas-fir growth. Soil Science Society of America Journal 69(6): 1822–1832. Arthur, E., Schjonning. P., Moldrup, P., Tuller, M., de Jonge L. W., 2013: Density and permeability of a loess soil: longterm organic matter effect and the response to compressive stress. Geoderma 193: 236–245. Bagheri, I., Kalhori, S. B., Akef, M., Khormali, F., 2011: Effect of compaction on physical and micromorphological properties of forest soils. American Journal of Plant Science 3(1): 159–163. Ballard, T. M., 2000: Impacts of forest management on northern forest soils. Forest Ecology and Management 133(1–2): 37–42. Bolat, I., Melemez, K., Davut, O., 2015: The influence of skidding operations on forest soil properties and soil compaction in Bartin, Turkey. European Journal of Forest Engineering 1(1): 1–8. Brais, S., 2001: Persistence of soil compaction and effects on seedling growth in northwestern Quebec. Soil Science Society of America Journal 65(4): 1263–1271. Cambi, M., Certini, G., Fabiano, F., Foderi, C., Laschi, A., Picchio, R., 2015b: Impact of wheeled and tracked tractors on soil physical properties in a mixed conifer stand. iForest 9(1): 89–94. Cambi, M., Certini, G., Neri, F., Marchi, E., 2015a: Impact of heavy traffic on forest soils: A review. Forest Ecology and Management 338: 124–138. Croat. j. for. eng. 39(2018)1

Middle-Term Сhanges in Topsoils Properties on Skidding Trails and Cutting Strips ... (71–83) Cambi, M., Hoshika, Y., Mariotti, B., Paoletti, E., Picchio, R., Venanzi, R., Marchi, E., 2017b: Compaction by a forest machine affects soil quality and Quercus robur L. seedling performance in an experimental field. Forest Ecology and Management 384: 406–414. Cambi, M., Paffetti, D., Vettori, C., Picchio, R., Venanzi, R., Marchi, E., 2017a: Assessment of the impact of forest harvesting operations on the physical parameters and microbiological components on a Mediterranean sandy soil in an Italian stone pine stand. European Journal of Forest Research 136(2): 205–215. Demir, M., Makineci, E., Yilmaz, E., 2007: Investigation of timber harvesting impacts on herbaceous cover, forest floor and surface soil properties on skid road in an oak (Quercus petrea L.) stand. Building and Environment 42(3): 1194–1199. Derbin, V. M., Derbin, M. V., 2016: Cut-to-Length Method in Selective Felling. Forestry journal 5: 123–131. Dymov, A. A., Lapteva, E. M., 2006: Changes in podzolic soils on bilayered deposits under the influence of felling. Russian Journal of Forest Science 3: 42–49. Dymov, A. A., Milanovskii, E. Y., 2014: Changes in the organic matter of taiga soils during the natural reafforestation after cutting in the middle taiga of the Komi Republic. Eurasian Soil Science 46(12): 1164–1171. Fedorets, N., Bahmet, O., 2003: Ecological settings of carbohydrate and nitrogen transformations in forest soils. Petrozavodsk: Karelian research centre Russian academy of sciences forest research institute, 240 p. Forest biodiversity: lessons from history for conservation, 2004: (Honnay O et al. eds). CABI Publishing, 320 р. Gerasimov, Y., Sokolov, A., 2014: Ergonomic evaluation and comparison of wood harvesting systems in Northwest Russia. Applied Ergonomics 45(2): 318–338. Goltsev, V., Tolonen, T., Syunev, V., Dahlin, B., Gerasimov, Y., 2011: Wood harvesting and logistics in Russia – focus on research and business opportunities. Final report of the research project: 157 p. (online): Retrieved on 2nd February 2017 from http://www.metla.fi/julkaisut/workingpapers/2011/ mwp210.pdf Hartmann, M., Niklaus, P. A., Zimmermann, S., Schmutz, S., Kremer, J., Abarenkov, K., Lüscher, P., Widmer, F., Frey, B., 2014: Resistance and resilience of the forest soil microbiome to logging-associated compaction. Multidisciplinary Journal of Microbial Ecology 8(1): 226–244. Hartsough, B., Drews, E., McNeel. J., Durston, T., Stokes, B., 1997: Comparison of mechanized systems for thinning ponderosa pine and mixed conifer stands. Forest Products Journal 47(11/12): 59–68.

A. Ilintsev et al.

Karpachevsky, L. O., 1981: Les i lesnye pochvy (Forest and forest soils). Moscow: Forest industry, 264 p. Karpechko, A. Y., 2008: Changes in density and root mass in soils the influence of harvesting machines in spruce forests of southern Karelia. Russian Journal of Forest Science 5: 66–70. Karvinen, S., Välkky, E., Torniainen, T., Gerasimov, Y., 2006: Northwest Russian forestry in a nutshell. Working Papers of the Finnish Forest Research Institute 30: 98 p. (online): Retrieved on 2nd February 2017 http://www.metla.fi/ julkaisut/workingpapers/2006/mwp030.pdf Klaes, B., Struck, J., Schneider, R., Schu, G., 2016: Middleterm effects after timber harvesting with heavy machinery on a fine-textured forest soil. European Journal of Forest Research 135(6): 1083–1095. Kozlowski, T. T., 1985: Soil aeration, flooding and tree growth. Journal of Arboriculture 11(3): 85–96. Kozlowski, T. T., 1999: Soil compaction and growth of woody plants. Scandinavian Journal of Forest Research 14(6): 596–619. Kurnaev, S. F., 1973: Lesorastitel’noe rajonirovanie SSSR. (Forest growth zoning of the USSR). Moscow: Science, 203 p. Laffan, M., Jordan, G., Duhig, N., 2001: Impacts on soils from cable-logging steep slopes in northeastern Tasmania, Australia. Forest Ecology and Management 144(1): 91–99. Lambers, H., Finnegan, P. M., Laliberté, E., Pearse, S. J., Ryan, M. H., Shane M. W., Veneklaas, E. J., 2011: Phosphorus nutrition of proteaceae in severely phosphorus-impoverished soils: are there lessons to be learned for future crops? Plant Physiology 156(3): 1058–1066. Leinonen, A., 2004: Harvesting technology of forest residues for fuel in the USA and Finland. Espoo. VTT Tiedotteita – Research Notes 2229, 132 p. Marchi, E., Picchio, R., Mederski, P. S., Vusić, D., Perugini, M., Venanzi, R., 2016: Impact of silvicultural treatment and forest operation on soil and regeneration in Mediterranean Turkey oak (Quercus cerris L.) coppice with standards. Ecological Engineering 95: 475–484. Marchi, E., Picchio, R., Spinelli, R., Verani, S., Venanzi, R., Certini, G., 2014: Environmental impact assessment of different logging methods in pine forests thinning. Ecological Engineering 70: 429–436. Marschner, P., 2012: Marschner’s mineral nutrition of higher plants. Third edition. Academic Press, London, 651 p.

Horn, R., Vossbrink, J., Peth, S., Becker, S., 2007: Impact of modern forest vehicles on soil physical properties. Forest Ecology and Management 248(1–2): 56–63.

McLaughlin, J. W., Phillips, S. A., 2006: Soil carbon, nitrogen, and base cation cycling 17 years after whole-tree harvesting in a low-elevation red spruce (Picea rubens)-balsam fir (Abies balsamea) forested watershed in central Maine, USA. Forest Ecology and Management 222(1/3): 234–253.

IUSS Working Group WRB, 2015: World reference base for soil resources. International soil classification system for naming soils and creating legends for soil maps. World Soil Resources Reports no. 106, FAO, Rome, Italy, 181 p.

McNabb, D., Startsev, A., Nguyen, H., 2001: Soil wetness and traffic level effects on bulk density and air-filled porosity of compacted boreal forest soils. Soil Science Society of America Journal 65(4): 1238–1247.

Croat. j. for. eng. 39(2018)1

A. Ilintsev et al.

Middle-Term Сhanges in Topsoils Properties on Skidding Trails and Cutting Strips ... (71–83)

Modry, M., Hubeny, D., 2003: Impact of skidder and highlead system logging on forest soils and advanced regeneration. Journal of Forest Science 49(6): 273–280. Naghdi, R., Solgi, A., Labelle, E. R., Zenner, E. K., 2016: Influence of ground-based skidding on physical and chemical properties of forest soils and their effects on maple seedling growth. European Journal of Forest Research 135(5): 949– 962. Najafi, A., Solgi, A., Sadeghi, S. H., 2009: Soil disturbance following four wheel rubber skidder logging on the steep trail in the north mountainous forest of Iran. Soil and Tillage Research 103(1): 165–169. Nakvasina, E. N., Seryj, V. S., Semenov, B. A., 2007: Polevoj praktikum po pochvovedeniju. (Field Workshop on soil science). Arkhangelsk, Arkhangelsk State Technical University, 127 p. Osman, K. T., Forest Soils: Properties and Management, 2013, Springer International Publishing Switzerland, 221 p. Page-Dumroese, D. S., Harvey, A. E., Jurgensen, M. F., Amaranthus, M. P., 1998: Impacts of soil compacition and tree stump removal on soil properties and outplanted seedlings in northern Idaho, USA. Canadian Journal of Forest Research 78(1): 29–34. Picchio, R., Neri, F., Petrini, E., Verani, S., Marchi, E., Certini, G., 2012: Machinery-induced soil compaction in thinning two pine stands in central Italy. Forest Ecology and Management 285(1): 38–43. Pobedinskii, A. V., 2013: Vodoohrannaja i pochvozashhitnaja rol’ lesov: vtoroe izdanie (Water and soil protection role of forests: Second Edition). Pushkino: ALL-Russian Research Institute for Silviculture and Mechanization of Forestry, 208 p. Powers, R. F., Scott, D. A., Sanchez, F. G., Voldseth, R. A., Page-Dumroese, D., Elioff, J. D., Stone, D. M., 2005: The North American long-term soil productivity experiment: findings from the first decade of research. Forest Ecology and Management 220(1–3): 31–50. Puettmann, K. J., Wilson, S. Mc. G., Baker, S. C., Donoso, P. J., Drossler, L., Amente, G., Harvey, B. D., Knoke, T., Lu, Y., Nocentini, S., Putz, F. E., Yoshida, T., Bauhus, J., 2015: Silvicultural alternatives to conventional even-aged forest management – what limits global adoption? Forest Ecosystems 2(8): 1–16.

Richards, D., Cockroft, B., 1974: Soil physical properties and root concentrations in an irrigated apple orchard. Australian Journal of Experimental Agriculture and Animal Husbandry 14(66): 103–107. Rozhkov, V. A., Karpachevskii, L. O., 2006: The forest cover of Russia and soil conservation. Eurasian Soil Science 39(10): 1041–1048. Seixas, F., McDonald, T., 1997: Soil compaction effects of forwarding and its relationship with 6- and 8-wheel drive machines. Forest Products Journal 47(11/12): 46–52. Sklyarov, G. A., Sharova, A. S., 1970: Pochvy lesov Evropejskogo Severa (The soils of the North European forests). Moscow, Science, 268 p. Solgi, A., Najafi, A., 2014: The impacts of ground-based logging equipment on forest soil. Journal of Forest Science 6(1): 28–34. Standish, J. T., Commandeur, P. R., Smith, R. B., 1988: Impacts of forest harvesting on physical properties of soils with reference to increased biomass recovery: a review. Canadian Forestry Service, Inf. Rep. BC-X-301, Pacific Forestry Centre, Victoria, BC, 24 p. Sukhanov, V. S., 2006: Destiny of development of deep processing of wood – in hands of lumberers. Moscow State Forest University bulletin – Lesnoy vestnik 8: 51–55. Šušnjar, M., Horvat, D., Šešelj, J., 2006: Soil compaction in timber skidding in winter conditions. Croatian Journal of Forest Engineering 27(1): 3–15. Venanzi, R., Picchio, R., Piovesan, G., 2016: Silvicultural and logging impact on soil characteristics in Chestnut (Castanea sativa Mill.) Mediterranean coppice. Ecological Engineering 92: 82–89. Vossbrink. J., Horn, R., 2004: Modern forestry vehicles and their impact on soil physical properties. European Journal of Forest Research 123(4): 259–267. Worrell, R., Hampson, A., 1997: The influence of some forest operations on the sustainable management of forest soils – a review. Forestry 70(1): 61–85. Zetterberg, T., Olsson, B. A., Löfgren, S., von Brömssen, C., Brandtberg, P-O., 2013: The effect of harvest intensity on long-term calcium dynamics in soil and soil solution at three coniferous sites in Sweden. Forest Ecology and Management 302: 280–294.

Croat. j. for. eng. 39(2018)1

Middle-Term Сhanges in Topsoils Properties on Skidding Trails and Cutting Strips ... (71–83)

A. Ilintsev et al.

Authors’ addresses: Aleksey Ilintsev * e-mail: a.ilintsev@narfu.ru Northern Research Institute of Forestry Laboratory of Taiga Ecosystems and Biodiversity Nikitov Str. 13 163062 Arkhangelsk RUSSIA Prof. Elena Nakvasina, PhD. e-mail: e.nakvasina@narfu.ru Prof. Sergey Tretyakov, PhD. e-mail: s.v.tretyakov@narfu.ru Assoc. prof. Sergey Koptev, PhD. e-mail: s.koptev@narfu.ru Northern (Arctic) Federal University named after M.V. Lomonosov Department of Silviculture and Forest Management Severnaya Dvina Emb. 17 163002 Arkhangelsk RUSSIA Aleksey Aleynikov, PhD. e-mail: aaacastor@gmail.com Center for Problems of Ecology and Productivity of Forests, Russian Academy of Sciences Laboratory of Structural and Functional Organisation of Forest Ecosystems Profsoyuznaya Str. 84/32 117997 Moscow RUSSIA

Received: January 27, 2017 Accepted: March 10, 2017 Croat. j. for. eng. 39(2018)1

Alexander Bogdanov, PhD. e-mail: aleksandr_bogd@mail.ru Northern Research Institute of Forestry Laboratory of Forest Management Nikitov Str. 13 163062 Arkhangelsk RUSSIA * Corresponding author

Original scientific paper

Tractive Performance of Tyres in Forest Conditions – Impact Assessment of Ground and Tyres Parameters Anna Cudzik†, Marek Brennensthul, Włodzimierz Białczyk, Jarosław Czarnecki Abstract This article deals with the assessment of traction properties of tyres on forest grounds. The research was carried out on skid trails located in pine stands. The tested grounds were different due to the cover of the soil and its mechanical properties. The study also deals with the evaluation of ways to improve traction by reducing the inflation pressure and using the tyre chain. The research was carried out using a specialized traction test stand for two tyres (9.5–24 and 400/55–22.5) different in width and tread pattern. The studies showed significant effect of ground conditions on traction. As a result of changes in the ground conditions, the values of drawbar force, rolling resistance and tractive efficiency were altered by 25%, 23% and 6%, respectively. The higher values of the drawbar force and tractive efficiency on all tested trails were obtained for 400/55–22.5 tyre. Both the use of tyre chains and the reduction of inflation pressure resulted in the increase in drawbar force and tractive efficiency. A better way to improve traction properties was the reduction of the tyre inflation pressure, which caused the increase in drawbar force and tractive efficiency. The use of tyre chains caused an increase in drawbar force over the entire slip range, while an increase in tractive efficiency has only been shown for the slip larger than 15%. Keywords: forest ground, tyre, drawbar force, rolling resistance, traction efficiency, inflation pressure, tyre chains

1. Introduction In forestry operations, the use of ground-based machinery for logging is common practice around the world. A wide range of equipment, such as skidders, forwarders and tractors, is used (Seixas and McDonald 1997, Jansson and Johansson 1998, Agherkakli et al. 2010, Picchio et al. 2011). Traction abilities of any off-road vehicle depend on several key factors such as: soil strength, vertical wheel/track load, contact area between wheels and the soil surface (Zoz and Grisso 2003, Molari et al. 2012, Battiato and Diserens 2013). Tractive performance is affected both by the soil normal strength and its shear strength. Generally, normal strength has the most effect on motion resistance, while shear strength has most effect on wheel slip (Zoz and Grisso 2003). Soil moisture content, soil texture and soil cover affect meCroat. j. for. eng. 39(2018)1

chanical behaviour and soil strength (Schreiber and Kutzbach 2008). Studies indicate that about 20–55% of the power delivered to the vehicle drive wheels is wasted in the tyre-soil interaction, because of the wheel slip and the rolling resistance (Muhsin 2010, Šmerda and Čupera 2010, Taghavifar et al. 2014). On the one hand, the slip of vehicle drive wheels causes energy losses in the process of interaction between the tyres and the topsoil; on the other hand, the slip may increase the risk for wheel rutting in the forest (Olsen and Wästerlund 1989), cause damage to ground vegetation and superficial roots (Greacen and Sand 1980) and reduce the growth of nearby standing forest trees (Wästerlund 1990). To reduce the slip of vehicle wheels, various ways are pointed out in terramechanics. One way is to increase the vertical load of drive wheels. As a result of

A. Cudzik et al.

Tractive Performance of Tyres in Forest Conditions – Impact Assessment of Ground ... (85–96)

increase in vertical load in tillage works, drawbar force can be increased by up to 15% depending on the value of ballast weights (Serrano et al. 2007, Serrano et al. 2009, Muhsin 2010). Increase in dynamic load of forwarder equipped with Trelleborg Twin 421 Mark II 600/55–26.5 tyres also resulted in significant increase in tractive efficiency at 5% slip (Jun et al. 2004). This method to improve traction abilities has a very important drawback – compacting of the soil and damaging its structure even to great depths, which can reduce soil productivity (Grečenko and Prikner 2014, Barbosa and Magalhaes 2015). Unfortunately, increase in vertical load of vehicle drive wheels may also lead to increased rolling resistance and consequently decrease the tractive efficiency, and increase tyre wear, fuel consumption and greenhouse gas emissions (Patel and Mani 2011, Lacour et al. 2014, Damanauskas et al. 2015, Damanauskas and Janulevičius 2015). The other way to reduce the slip is to increase the contact area between tyres and ground. Currently, tractor performance researchers recommend the reduction of inflation pressure in the tyres (Šmerda and Čupera 2010, Battiato and Diserens 2013). It means that vehicle weight is spread across a larger area, wheels »sink« less into the soil, ruts are not so deep and the rolling resistance is reduced (Nam et al. 2010). Kurjenluomar et al. (2009) reported that the reduction of tyre inflation pressure resulted in reduced rolling resistance and rut depth only on soft soil, when the soil strength was low, while in hard soil conditions the effect on rolling resistance was quite the opposite. Depending on the tractor, tyre size and type, the drawbar force may be increased by up to 8% (Sumer and Sabanci 2005, Elwaleed et al. 2006, Taghavifar and Mardani 2013). Important factors influencing the traction performance are tyre parameters, such as radius, width and tread pattern. Tyre diameter has a significant effect on the traction force. The larger tyre width increases traction capability due to increasing flexibility of the tyre and assists in the development of uniformity of pressure application, but it can also produce more motion resistance (Nkakini and Fubara-Manuel 2014). Forest soils, in general, are susceptible to compaction as they are loose, with high organic-matter content and are generally low in bulk density, high in porosity, and low in strength (Froehlich et al. 1985, Horn et al. 2007, Jourgholami and Majnounian 2011). These conditions are unfavourable for forest vehicle traffic. One way to improve the traction abilities of wheeled forest machines is the use of tyre chains. Chains may improve traction by increasing the soil shear area through better penetration to the soil sur-

face, without using wide tyres or tracks to increase the contact area. Only two publications were found that quantitatively documented the effect of forestry tyre chains on traction performance. Vechinski et al. (1999) reported that traction performance of tyres with chains depends on the type and cover of soil. The improved performance of the tyre with chains on clay soils with pine straw and sod cover was due to the penetration into the surface cover by chains. The use of tyre chains caused an increase in net traction of up to 11%. Tractive efficiency increased (by 7%) when chains were added to new tyres, but decreased (by 5–8%) when they were used on worn tyre. Stoilov (2007) showed that, on deformable forest roads, the use of tyres equipped with chains caused slight decrease (2%) of net traction due to higher motion resistance and enlarged windage between the tyres and chains. The majority of previous studies about the traction performance concerned farm tractors and agricultural soils. In regards to forest conditions, the most of research relates to the impact of machinery and vehicles on the forest soils. Only a few publications concern the issues related to tyre traction performance on forest grounds. The issue of traction performance of an offroad vehicle is very important, because of high energy losses and soil disturbance as a result interaction between drive wheels of vehicle and soil. In this context, the aim of the present study was to evaluate: Þ the effect of soil conditions on the traction abilities of drive wheel Þ the influence of tyre dimensions on traction properties Þ the effect of tyre inflation pressure and the tyre chains use on the traction performance.

2. Materials and methods 2.1 Study site The study was carried out in Poland (Lower Silesia Province) in Forest District Oława, Forest Sub-district Chrząstawa Wielka. Research on traction performance of tyres was conducted in lowland pine forest stands of different age on three selected skid trails (ground forest roads). All analysed skid trails were used with low intensity, because of young age of stands and limited wood harvesting of these stands. These skid trails also fulfil the function of territorial forestry division – as a border between forest stands. The skid trails were constructed on typical acidic brown soils. The analysis of soil particle size distribution indicated that all analysed soils are sands. Detailed description of the soil parameters of skid trails is presented in Table 1. Croat. j. for. eng. 39(2018)1

Tractive Performance of Tyres in Forest Conditions – Impact Assessment of Ground ... (85–96)

Table 1 Study site description Skid trail 1 Location

Skid trail 2

Skid trail 3

Pine forest

Landform, slope, °

Flat terrain 0 – 1° Bare soil and pine litter mix

Surface cover

Grass and pine litter mix

Grass

Soil particular size at a depth of 0–0.2 m, % Sand, 0.05–0.2 mm

Silt, 0.002–0.05 mm

Clay, <0.002 mm

Specific density of the soil, g·cm

1.60

1.57

1.54

Soil moisture content, % vol. *

7.5

8.3

9.1

Maximum shearing stress, kPa *

126

140

Soil penetration resistance, MPa *

2.26

3.20

3.40

Gravel, 0.2–2 mm –3

* at a depth 0–0.1 m; cone base area of 1 cm2, cone angle of 60°

A. Cudzik et al.

carried out for two inflation pressure values of 200 and 80 kPa. The study concerning the effect of tyre chains on traction properties was carried out for the wheel fitted with 9.5–24 tyre. The »traktor doppelspur 3334d« chains type was used. The vertical load of tested tyres was 6110 N. The field trials were conducted using the mobile stand mounted on the three-point linkage of Massey Ferguson 235 tractor. The tested wheel was powered by tractor PTO, using a reduction gear. During the study, the test stand was moved by tractor. The data from all measurement devices were recorded by a data recording system. In the final phase of measurement, the test stand was stopped, using the basic tractor brake (for the full range slip of the tested wheel). The measuring equipment consisted of: inductive dynamometer for the measurement of the drawbar force, located between the external (immobile) and internal (moving) part of the frame, and the inductive torquemeter for the measurement of the drive torque. The actual and theoretical distance was measured using two rotational encoders MOK40, mounted on the extra wheel and on the shaft with the tested wheel, respectively. The overall view of research stand is pre-

Table 2 Technical data of tyres Tyres designation manufacturer

Tread type

Overall diameter mm

Section width mm

Height of tread lugs mm

Maximum load capacity kg

Nominal inflation pressure kPa

9.5–24 Mitas

TD02 Universal tread pattern

1000

241

1120

250

400/55–22.5 Trelleborg

T404T Twin

1000

400

1120

250

2.2. Equipment, test procedures and calculations Traction properties were analysed for tyres with the same maximum load capacity, nominal inflation pressure and overall diameter, but with treads of different width and design. The main parameters of the tested tyres are presented in Table 2. These types of tyres are commonly used in machinery and tractors working in forests. Evaluation of traction performance improvement by reduction of tyre inflation pressure or use of tyre chains was conducted on skid trails with the worst traffic conditions (the lowest soil penetration resistance and the lowest shearing strength). The effect of changes in tyre inflation pressure on the traction parameters was assessed for 400/55–22.5 tyre. Tests were Croat. j. for. eng. 39(2018)1

sented in Fig. 1. Detailed specifications of measurement devices are presented in Table 3. All tests (for both tyres in all conditions) were carried out for the ride in one direction, with 5 repetitions. Table 3 Specifications of measurement devices Instrumentation

Measurements

Range

Accuracy

Inductive dynamometer

Pulling force

0–20 kN

± 20 N

Inductive torquemeter

Torque

0–3000 Nm

± 3 Nm

Rotational encoder MOK40

Angle of the wheel rotation

0–360 ° ´ n*

0.36 °

* the number of rotation

A. Cudzik et al.

Tractive Performance of Tyres in Forest Conditions – Impact Assessment of Ground ... (85–96)

Fig. 1 The stand for testing traction properties of tyres

From the obtained results, the mean values were calculated. Based on the measured parameters, slip of the wheels, traction force, rolling resistance and tractive efficiency values were calculated using Eq. 1–4:

 SR d=  1 − ST  TF =

  × 100 

MO rD

RR= TF − DF m=

DF TF

× (100 − d )

Where: Δ wheel slip, % sR actual distance of the wheel, m sT theoretical distance of the wheel, m

TF MO rD RR DF Η

traction force, N wheel torque, Nm dynamic radius of the wheel, m rolling resistance, N drawbar force, N tractive efficiency, %

(1)

The values of the wheel dynamic radius (rD) were determined based on the measured distance covered by the wheel during five full rotations.

(2)

2.3. Statistical analysis

(3) (4)

The statistical analysis was done using Statistica 12.5 software. To evaluate the impact of factors on traction parameters, the analysis of variance (ANOVA) was used, with the significance level (a) of 0.05. Before carrying out the ANOVA tests, the normal distribution and homogeneity of variance were verified (using Shapiro-Wilk and Levene tests, respectively). Moreover, the post-hoc tests (homogenous groups tests – LSD Fisher) were done – these tests should point out the differences between the factor levels. Croat. j. for. eng. 39(2018)1

Tractive Performance of Tyres in Forest Conditions – Impact Assessment of Ground ... (85–96)

A. Cudzik et al.

Fig. 2 Drawbar force as a function of slip a) 9.5–24 tyre, b) 400/55–22.5 tyre

3. Results The analysed traction parameters were presented for the slip in the range 0–30%. At larger wheel slip, the operation of the vehicle is neither economically nor ecologically justified, due to high energy losses and damage to the soil. The drawbar forces, as a function of slip for 9.5–24 and 400/55–22.5 tyres, at three skid trails are shown in

Fig. 2. For both tyres, the highest values of drawbar force were obtained on skid trail 3. The lowest values of drawbar force were observed on skid trail 1. Differences in cover type and soil compactness of skid trails contributed to the significant differences in drawbar force values obtained by the tested tyres. For 9.5–24 tyre, the average drawbar force values measured on skid trail 2 and skid trail 3 were greater than on skid trail 1 by 8% and 14%, respectively. The drawbar force

Fig. 3 Rolling resistance as a function of slip a) 9.5–24 tyre, b) 400/55–22.5 tyre Croat. j. for. eng. 39(2018)1

A. Cudzik et al.

Tractive Performance of Tyres in Forest Conditions – Impact Assessment of Ground ... (85–96)

Fig. 4 Tractive efficiency as a function of slip a) 9.5–24 tyre, b) 400/55–22.5 tyre measured for 400/55–22.5 tyre on skid trails 2 and 3 was higher by 15% and 25%, respectively, than on the skid trail 1. On all analysed skid trails, the higher drawbar force was obtained for a 400/55–22.5 tyre. Differences in the average values of drawbar force obtained by the tested tyres occurred at slip in the range of 7–18%, the largest were observed on skid trail 3. Greater increase in drawbar force along with the increase of slip was observed for 400/55–22.5 tyre. Fig. 3 shows the relationship between rolling resistance and wheel slip. The greatest motion resistance for both tested tyres was found on skid trail 1, and the lowest on skid trail 3. The rolling resistance values of 9.5–24 tyre on skid trails 2 and 3 were lower by 20% and 23%, respectively, compared to the values obtained on skid trail 1. For 400/55–22.5 tyre on skid trails 2 and 3, the average rolling resistance values were lower than those obtained on skid trail 1 by 11% and 22%, respectively. Greater rolling resistance was found for 400/55–22.5 tyre; the value of this parameter was higher than for 9.5–24 tyre by 13–26%. Tractive efficiency is a very important parameter characterizing the process of generating driving force. It illustrates the losses of energy delivered to the drive wheel. The relationship between the tractive efficiency and slip is shown in Fig. 4. The lowest values of tractive efficiency for both tested tyres were obtained on skid trail 1. Soil mechanical parameters within skid trails 2 and 3 showed higher traction efficiency, with

relative differences of 10% and 17% for 9.5–24 tyre, and 9% and 15% for 400/55–22.5 tyre, respectively. The maximum tractive efficiency of 58–65% (depending on skid trail), for 9.5–24 tyre, was reached at slip of up to 3%, in the case of 400/55–22.5 tyre, the maximum values of 52–60% were reached at slip of 3–6%. For 9.5–24 tyre, the highest differences in tractive efficiency values obtained on individual trails were observed at slip range of 0–5% and 17–30%. Significant differences in the tractive efficiency values obtained by 400/55–22.5 tyre on individual routes were visible throughout the whole slip range. Furthermore, in the case of the 400/55–22.5 tyre, less drop in tractive efficiency values due to increase in wheel slip was observed. On all tested trails, higher tractive efficiency of 4–6% (relative difference in mean values) was obtained for 9.5–24 tyre. Table 4 shows the results of statistical analysis regarding the impact of ground conditions (trails) on traction parameters, separately for 9.5–24 and 400/55–22.5 tyre. Presented p-values represent the level of probability of basic hypothesis acceptance (when the p-value is smaller than significance level a = 0.05, the factor is significant). Based on results of statistical analysis, it can be stated that in all cases the trail type was a significant factor for all analysed parameters. The post-hoc tests showed that the skid trail 1 was classified as a separate group, while the trails 2 and 3 were considered as one group.

Croat. j. for. eng. 39(2018)1

Tractive Performance of Tyres in Forest Conditions – Impact Assessment of Ground ... (85–96)

A. Cudzik et al.

Table 4 Results of statistical analysis; factor: ground conditions; SD – standard deviation, p-value – probability level, A, B – homogeneous groups Analysed parameter

Tyre

9.5–24 Drawbar force 400/55–22.5

Ground conditions

Mean

±SD

Skid trail 1

1295A

32.4

Skid trail 2

26.2

Skid trail 3

1481

33.3

Skid trail 1

1392A

37.9

Skid trail 2

45.4

44.8

Skid trail 3

9.5–24 Rolling resistance

9.5–24 Tractive efficiency 400/55–22.5

1602 1748

Skid trail 1

1137

34.9

Skid trail 2

910B

19.3

Skid trail 3

875B

26.6

32.9

Skid trail 2

1101

31.5

Skid trail 3

1000B

38.6

Skid trail 1

45,9A

0.57

Skid trail 2

52.8

0.55

Skid trail 3

55.8B

0.59

Skid trail 1

46.6

0,66

Skid trail 2

50.1B

0.57

Skid trail 3

0.59

Skid trail 1 400/55–22.5

1400

1284

53.1

p-value

0.042879

0.038489

0.035530

0.037977

0.001174

0.005006

Analysis of traction properties of tyres performed on three skid trails showed that the skid trail 1 was the worst for the forest vehicles traffic. For that reason, the assessment of ways to improve the traction abilities was carried out on this skid trail. The dependence between the drawbar force, rolling resistance and slip for 9.5–24 tyre with and without chain is shown in Fig 5. The use of the tyre chain contributed to a comparable increase in drawbar force, and the average increase in values was of 8%. Fitting the tyre with chain caused a slight increase in rolling resistance values at the slip in the range of 0–20%. For the higher slip values, the rolling resistance of tyre with chain was lower than for tyre without chain. The effect of the use of tyre chain on the wheel tractive efficiency is shown in Fig. 6. At slip in the range of 2–9% for tyre with chain, a slight decrease in tractive efficiency was observed. At slip of 10–15%, the values of the tractive efficiency of the tyre with and without chain did not differ. At slip over 15%, the Croat. j. for. eng. 39(2018)1

Fig. 5 Comparison of drawbar force and rolling resistance of 9.5–24 tyre – without and with chain higher traction efficiency was achieved by the wheel with chain, and the observed difference in the values of this parameter increased as the slip increased. It has also been observed that tractive efficiency values of the tyre with chain varied significantly less than the values of the tyre without chain due to the increase in wheel slip.

Fig. 6 Comparison of tractive efficiency of tyre with and without chain

A. Cudzik et al.

Tractive Performance of Tyres in Forest Conditions – Impact Assessment of Ground ... (85–96)

Table 5 Results of statistical analysis; factor: use of tyre chain, SD – standard deviation, p-value – probability level, A, B – homogeneous groups Analysed Slip parameter range <10 Drawbar force

10–20

>20

<10 Rolling 10–20 resistance >20

<10 Tractive 10–20 efficiency >20

Tyre

Mean

±SD

9.5–24

1079A

33.7

9.5–24 with chain

1192

25.5

9.5–24

1360A

38.8

9.5–24 with chain

1504B

35.0

9.5–24

1565A

24.8

9.5–24 with chain

1610

28.0

9.5–24

905A

28.0

9.5–24 with chain

980B

28.1

9.5–24

1143A

29.9

9.5–24 with chain

1171

39.2

9.5–24

1516A

46.9

9.5–24 with chain

1326B

39.2

9.5–24

51.4A

0.62

9.5–24 with chain

49.4B

0.54

9.5–24

46.0A

0.54

9.5–24 with chain

47.1

0.58

9.5–24

36.6A

0.54

9.5–24 with chain

42.5B

0.58

p-value 0.001740

0.001509

0.053979

0.008499

0.291183

0.000588

0.001972

0.032505

0.000006

The effect of the use of tyre with chain on traction parameters of 9.5–24 tyre is shown in Table 5. The analysis was done separately for three ranges of wheel slip. For the drawbar force, the use of chain was statistically insignificant only at the wheel slip higher than 20%. In the case of rolling resistance, the significant impact of the use of chain occurred at the wheel slip of up to 10% and over 20%. However, the significant impact of the use of tyre chain on tractive efficiency values was found over the whole slip range. The effect of the change in tyre inflation pressure on changes in drawbar force and rolling resistance is shown in Fig. 7. Reduction of tyre inflation pressure contributed to the increase in drawbar force (average) of 13%, wherein the greatest differences were observed for the slip of 5–20%. At 80 kPa, the tyre rolling resistance values were higher by 10% than at 200 kPa, wherein the dif-

Fig. 7 Comparison of drawbar force and rolling resistance values of 400/55–22.5 tyre at different inflation pressure ference in the values of this parameter increased with the increase of slip. The greater increase in drawbar force and rolling resistance with an increase in slip was observed for lower inflation pressure. Effect of changes in tyre inflation pressure on tractive efficiency is shown in Fig. 8. Reduction of the tyre inflation pressure resulted in a slight increase in tractive efficiency (the average relative increase in tractive efficiency was 6%). The highest increase in tractive efficiency due to the reduction of the tyre inflation pressure was found at slip lower than 10%. It was observed that the change in tyre inflation pressure did not result in changes of tractive efficiency as a function of slip. Maximum tractive efficiency (52% at 200 kPa and 54% at 80 kPa) was achieved at slip lower than 5%. The increase in slip resulted in a decrease in tractive efficiency. Table 6 shows the results of statistical analysis for 400/55–22.5 tyre, with the inflation pressure as a factor. Three slip ranges were analysed separately. In accordance with the obtained results, the change of inflation pressure had significant impact on all analysed parameters. This relationship was observed for all ranges of wheel slip.

4. Discussion The results of the study show that values of all analysed traction parameters depend on the soil conCroat. j. for. eng. 39(2018)1

Tractive Performance of Tyres in Forest Conditions – Impact Assessment of Ground ... (85–96)

Fig. 8 Comparison of tractive efficiency of 400/55–22.5 tyre at different inflation pressure ditions, determined by soil cover, soil compactness and shear strength. The highest values of drawbar force and tractive efficiency for both tested tyres were obtained on skid trail 3, where the soil was characterized by greater resistance to shearing by tyre treads and their limited depth impact. This resulted in lower energy losses associated with rolling resistance and consequently higher tractive efficiency. The most susceptible to the deep impact of the tyres was trail 1 covered by bare soil with a local pine litter, which was reflected in higher values of rolling resistance and lower tractive efficiency. The presented results are in line with the results of Vehinski et al. (1998, 1999), who examined the drawbar force and tractive efficiency of tyres in forest conditions on different types of soil and various cover. They also obtained the greatest values of analysed traction parameters on soils with grass cover and with greater compactness. The present study has shown that the rate of increase of the drawbar force, along with the increase of slip, was mainly determined by tyres parameters, which was confirmed by comparable waveforms of all trails. The greatest increase in drawbar force was observed at slip range of 0–5% and 20–30% for 9.5–24 tyre. A greater and more intensive increase in drawbar force with the increase of slip was observed for 400/55–22.5 tyre. Hittenbeck (2013) obtained similar results in the study on relationship between the slip and drawbar force of forwarder equipped with: worn tyres, new tyres, tyres with reduced inflation pressure Croat. j. for. eng. 39(2018)1

A. Cudzik et al.

and combination of tracks and chains. This experiment was conducted on loess soil with moisture content of 32.1%. He reported that, with increasing wheel slip, the values of the drawbar force rise immediately up to a level of about 30% slip. Above that, the increase of the drawbar force is still clear but less rapid. Battiato and Diserens (2017) reported that the influence of slip on rolling resistance and tractive efficiency is controlled by many factors, of which the most considerable are the soil deformation parameters under normal and horizontal stress, tyre width and tyre stiffness. The statement was confirmed in our study. According to the presented results, the greatest rolling resistance values for both tested tyres was observed at skid trail 1 – located on soil of smaller compactness and the lowest shear strength. Due to higher width, greater rolling resistance was found for 400/55–22.5 tyre. The distribution of the tractive efficiency, as a function of slip, is due to the fact that at low slip (<4–6%), a small increase in slip occurs and consequently soil shear displacement results in a great increase in traction force, this resulting in an overall rise in traction efficiency. At slip greater than 4–6%, an increase in slip and in soil shear displacement results in a slight increase in traction force with reduction in traction efficiency. Battiato and Diserens (2017) also showed that tractive efficiency sharply rises at low slip and reaches a peak with slip ranging between 6–12%. Beyond this peak, it decreases progressively with slip. The present study has shown that the tyre equipped with chain caused an increase of about 8% in drawbar force for the whole slip range. The increase in rolling resistance was found only at slip of 0–20%. These results are in line with the results of Stoilov (2007), who showed that the use of chains for LKT skidder 16.9–30 tyres caused an increase in drawbar force and rolling resistance. Vehinski et al. (1999) reported that, on bare soil, adding chains to the new tyre increased the average drawbar force and tractive efficiency, by 10% and 7%, respectively. They also showed that the use of chains on worn tyres resulted in an increase in drawbar force, but a decrease in tractive efficiency. Our study has shown that reducing the inflation pressure in 400/55–22.5 tyre resulted in an increase both in drawbar force (13%) and rolling resistance (10%). The changes in the values of both traction parameters were the result of the increase of soil-tyre contact area. Under conditions of low humidity and relatively high soil compactness, the wide tyre did not penetrate into the ground. The reduction of tyre inflation pressure caused an increase in the tyre-soil contact surface, but mainly due to tyre deformation. The manufac-

A. Cudzik et al.

Tractive Performance of Tyres in Forest Conditions – Impact Assessment of Ground ... (85–96)

Table 6 Results of statistical analysis; factor: inflation pressure, SD – standard deviation, p-value – probability level, A, B – homogeneous groups Analysed parameter

Slip range <10

Drawbar force

10–20

>20

<10 Rolling 10–20 resistance >20

<10 Tractive efficiency

10–20

>20

Tyre inflation pressure, kPa

Mean

±SD

200

1081A

42.0

1289

34.0

200

1459A

37.8

1738B

30.7

200

1792A

34.1

2197

31.3

200

991A

31.9

1084B

27.9

200

1395A

29.5

1589

33.6

200

1653A

37.4

2082

28.2

200

49.6A

0.60

52.6B

0.37

200

45.6A

0.72

47.1

0.33

200

39.9A

0.66

41.2B

0.46

p-value 0.000249

0.000026

0.000002

0.004853

0.000130

0.000002

0.000797

0.011742

0.019644

turer states that 400/55–22.5 tyre is characterized by high flexibility on both hard and soft ground. Our results are partially compliant with the results of the traction test of the skidder LKT 81 T carried out by Stoilov (2007) on brown forest soil with a moisture content of 20.3%. The author showed that the reduction of tyre inflation pressure from 230 to 190 kPa caused an increase in the drawbar force value of 6.8% and a drop in the rolling resistance value of 7.69%. Kurjenluomar et al. (2009) showed that, on firm soil, lower tyre inflation pressure resulted in higher rolling resistance values, which is in agreement with evidence presented in this study. Our results also confirm the results of Gharibkhani et al. (2012), who reported that reduction in inflation pressure increases the rolling resistance of tyre in their experiment conducted on hard soil. Jun et al. (2004) showed that the reduction in inflation pressure from 240 to 100 kPa in the Trelleborg Twin 421 Mark II 600/55–26.5 tyre resulted in an increase in drawbar force of 23% measured at 5%

slip. In our study, the reduction of inflation pressure from 200 to 80 kPa in the Trelleborg 400/55–22.5 tyre caused an increase in drawbar force of 21% at 5% slip. The present study has also shown that the reduction of tyre inflation pressure contributed to better tractive efficiency; the average relative increase of this parameter was 6%. Similar results have been obtained by Jun et al. (2004), who showed that the reduction of the inflation pressure in 600/55–26.5 tyre resulted in an increase in tractive efficiency of 9% at 5% slip. In our study, the relative increase in traction efficiency at 5% slip was lower and amounted to 5%.

5. Conclusions The soil conditions had a significant influence on the values of all traction parameters. The highest values of drawbar force and tractive efficiency and the smallest rolling resistance values for both tyres were obtained on skid trail 3. The highest motion resistance and the lowest values of drawbar force and tractive efficiency were found on skid trail 1. Differences in the analysed traction parameters between the best and the worst substrate due to traffic conditions amounted to even 25% for drawbar force, 23% for rolling resistance and 17% for tractive efficiency. The 400/55–22.5 tyre is better for forestry applications, as higher drawbar forces were achieved with it than with 9.5–24 tyre. This tyre had a slightly lower tractive efficiency than the 9.5–24 tyre, but was characterized by a smaller decrease in the value of this parameter as a result of increase in wheel slip. It means less energy loss in the wheel-soil system due to changes in traffic conditions. The lower tractive efficiency of the 400/55–22.5 tyre is due to greater rolling resistance resulting from greater width and deformability of this tyre. In practice, greater tyre width means lower ground pressure and smaller susceptibility to rut formation. The better way to improve traction properties on the tested skid trail was to reduce the tyre inflation pressure, resulting in an increase of 13% and 6%, respectively, both in drawbar force and tractive efficiency. The use of tyre chain proved to be most advantageous for wheel slip of more than 15%, due to the increase in drawbar force and tractive efficiency of about 15% and 6%, respectively.

6. References Agherkakli, B., Najafi, A., Sadeghi, S., 2010: Ground based operation effects on soil disturbance by steel tracked skidder in a steep slope of forest. Journal of Forest Science 56(6): 278–284. Croat. j. for. eng. 39(2018)1

Tractive Performance of Tyres in Forest Conditions – Impact Assessment of Ground ... (85–96) Barbosa, L. A. P., Magalhaes, P. S. G., 2015: Tyre tread pattern design trigger on the stress distribution over rigid surfaces and soil compaction. Journal of Terramechanics 58(1): 27–38. Battiato, A., Diserens, E., 2013: Influence of tyre inflation pressure and wheel load on the traction performance of a 65 kW MFWD tractor on a cohesive soil. Journal of Agricultural Science 5(8): 197–215. Battiato, A., Diserens, E., 2017: Tractor traction performance simulation on differently textured soils and validation: A basic study to make traction and energy requirements accessible to the practice. Soil & Tillage Research 166(1): 18–32.

A. Cudzik et al.

Kurjenluomar, J., Alakukku, L., Ahokas, J., 2009: Rolling resistance and rut formation by implement tyres on tilled clay soil. Journal of Terramechanics 46(6): 267–275. Lacour, S., Burgun, C., Perilhon, C., Descombes, G., Doyen, V., 2014: A model to assess tractor operational efficiency from bench test data. Journal of Terramechanics 54(1): 1–18. Molari, G., Bellentani, L., Guarnieri, A., Walker, M., Sedoni, E., 2012: Performance of an agricultural tractor fitted with rubber tracks. Biosystems Engineering 111(1): 57–63. Muhsin, S. J., 2010: Studying the power losses of two and four-wheel drive tractors (2WD and 4WD) of Massey Ferguson (2680). Journal of Basrah Researches (Sciences) 36(6): 59–66.

Damanauskas, V., Janulevičius, A., 2015: Differences in tractor performance parameters between single-wheel 4WD and dual-wheel 2WD driving systems. Journal of Terramechanics 60(1): 63–73.

Nam, J. S., Park, Y. J., Kim, K. U., 2010: Determination of rating cone index using wheel sinkage and slip. Journal of Terramechanics 47(4): 243–248.

Damanauskas, V., Janulevičius, A., Pupinis, G., 2015: Influence of extra weight and tyre pressure on fuel consumption at normal tractor slippage. Journal of Agricultural Science 7(2): 55–67.

Nkakini, S. O., Fubara-Manuel, I., 2014: Effects of soil moisture and tillage speeds on tractive force of disc ploughing in loamy sand soil. European International Journal of Science and Technology 3(4): 157–164.

Elwaleed, A. K., Yahya, A., Zohadie, M., Ahmad, D., Kheiralla, A. F., 2006: Effect of inflation pressure on motion resistance ratio of a highlug agricultural tyre. Journal of Terramechanics 43(2): 69–84.

Olsen, H. J., Wästerlund, I., 1989: Terrain and vehicle research with reference to forestry. Swedish University of Agricultural Sciences. Technical report 149, SLU, Department of operational efficiency, Garpenberg: 61 p.

Froehlich, H. A, Miles, D. W. R, Robbins, R. W., 1985: Soil bulk density recovery on compacted skid trails in central Idaho. Soil Science Society of America Journal 49(4): 1015– 1017.

Patel, S. K., Mani, I., 2011: Effect of multiple passes of tractor with varying normal load on subsoil compaction. Journal of Terramechanics 48(4): 277–284.

Gharibkhani, M., Mardani, A., Vesali, F., 2012: Determination of wheel–soil rolling resistance of agricultural tyre. Australian Journal of Agricultural Engineering 3(1): 6–11. Greacen, E. L., Sand, R., 1980: Compaction of forest soils. A review. Australian Journal of Soil Research 18(2): 163–189. Grečenko, A., Prikner, P., 2014. Tyre rating based on soil compaction capacity. Journal of Terramechanics 52(1): 77–92. Hittenbeck, J., 2013: Estimation of trafficable grades from traction performance of a forwarder. Croatian journal of forest engineering 34(1): 71–81. Horn, R., Vossbrink, J., Peth, S., Becker, S., 2007: Impact of modern forest vehicles on soil physical properties. Forest Ecology and Management 248(1–2): 56–63. Jansson, K., Johansson, J., 1998: Soil changes after traffic with a tracked and a wheeled forest machine: a case study on a silt loam in Sweden. Forestry 71(1): 57–66. Jourgholami, M., Majnounian, B., 2011: Effects of wheeled cable skidder on rut formation in skid trail – a case study in Hyrcanian forest. Journal of Forestry Research 22(3): 465−469. Jun, H., Way, T. R., Löfgren, B., Landström, M., Bailey, A. C., Burt, E. C., McDonald, T. P., 2004: Dynamic load and inflation pressure effects on contact pressures of a forestry forwarder tyre. Journal of Terramechanics 41(4): 209–222. Croat. j. for. eng. 39(2018)1

Picchio, R., Neri, F., Maesano, M., Savelli, S., Sirna, A., Blasi, S., Baldini, S., Marchi, E., 2011: Growth effects of thinning damage in a Corsican pine (Pinus laricio Poiret) stand in central Italy. Forest Ecology and Management 262(2): 237–243. Seixas, F., McDonald, T., 1997: Soil compaction effects of forwarding and its relationship with 6- and 8-wheel drive machines. Forest Products Journal 47(11/12): 46–52. Serrano, J. M., Peca, J. O., Marques da Silva, J., Pinheiro, A., Carvalho, M., 2007: Tractor energy requirements in disc harrow systems. Biosystems Engineering 98(3): 286–296. Serrano, J. M., Peca, J. O, Silva, R., Marquez, L., 2009: The effect of liquid and tyre inflation pressure on tractor performance. Biosystems Engineering 102(1) 51–62. Schreiber, M., Kutzbach, H. D., 2008: Influence of soil and tyre parameters on traction. Research in Agricultural Engineering 54(2): 43–49. Šmerda, T., Čupera, J., 2010: Tyre inflation and its influence on drawbar characteristics and performance – Energetic indicators of a tractor set. Journal of Terramechanics 47(11/12): 395–400. Stoilov, S., 2007: Improvement of wheel skidder tractive performance by the tyre inflation pressure and tyre chains. Croatian Journal of Forest Engineering 28(2): 137–144. Sümer, S. K., Sabanci, A., 2005: Effects of different tyre configurations on tractor performance. Turkish Journal of Agriculture and Forestry 29(6): 461–468.

A. Cudzik et al.

Tractive Performance of Tyres in Forest Conditions – Impact Assessment of Ground ... (85–96)

Taghavifar, H., Mardani, A., 2013: Investigating the velocity, inflation pressure and vertical load on rolling resistance of a radial ply tyre. Journal of Terramechanics 50(2): 99–106. Taghavifar, H., Mardani, A., Karim-Maslak, H., 2014: Multicriteria optimization model to investigate the energy waste of off-road vehicles utilizing soil bin facility. Energy 73(1): 762–770. Vechinski, C. R., Johnson, C. E., Raper R. L., 1998: Evaluation of an empirical traction equation for forestry tyres. Journal of Terramechanics 35(1): 55–67.

Vechinski, C. R., Johnson, C. E., Raper, R. L., McDonald, T. P., 1999: Forestry tyre tractive performance: new, worn, and with chains. Applied Engineering in Agriculture 15(4): 263–266. Wästerlund, I., 1990: Soil strength in forestry measured with a new kind of test rig. In: Proceedings of the 10th International ISTVS Conference of the, Kobe, Japan; School of Civil Engineering, Kyoto University: 73–82. Zoz, F. M., Grisso, R. D., 2003: Traction and tractor performance. ASAE distinguished lecture series (Tractor Design No. 27), ASAE Publication No. 913C0403. St. Joseph, Michigan, USA, ASAE. 46 p.

Authors’ addresses:

Received: April 8, 2017 Accepted: July 7, 2017

Anna Cudzik, PhD. † Marek Brennensthul, PhD. * e-mail: marek.brennensthul@upwr.edu.pl Prof. Włodzimierz Białczyk, PhD. e-mail: wlodzimierz.bialczyk@upwr.edu.pl Jarosław Czarnecki, PhD. e-mail: jaroslaw.czarnecki@upwr.edu.pl Institute of Agricultural Engineering Faculty of Life Sciences and Technology Wrocław University of Environmental and Life Sciences Chełmońskiego 37a 51-630 Wrocław POLAND *Corresponding author Croat. j. for. eng. 39(2018)1

Original scientific paper

Trailer Overturning during Wood Transportation: an Experimental Investigation of Effects of Trailer Joint Point and Frame Structure Marco Manzone, Angela Calvo Abstract Trailers may increase the risk of tractor overturn during wood transportation in dangerous conditions. In this work, tests were carried to simulate a trailer rollover using three two wheel tractors and a crawler tractor and three trailers (two single-axle and one two-axle), all of their combinations moving downhill along the path on a short dirt road. The trailers were always loaded with the same load of logs cut at a length of about 1.5 m and put transversely to the longitudinal axis of the trailer. During each test, the following parameters were measured: the lateral dragging of the rear wheels/crawler of the tractor, the ground detachment of the rear upstream wheel/crawler and both the longitudinal and transversal strains (released over the tractor hooking system) produced by the trailer overturn. The study highlighted that the biaxle trailer structure with a turntable steering had the best performances compared to the single-axle in terms of safety during trailer overturning. Independently of the trailer type considered in this work, a tied load is more dangerous than a load restrained only by steel struts, because during the overturn the load forms a single unit with the trailer mass, which increases the transversal and longitudinal strain. Keywords: trailer structure, forestry, rollover, safety

1. Introduction In the European Union (EU 28) in 2010 there were 12.2 million farms (with many small family farms) with around 5% of European workers involved (Eurostat 2014): this rate is not entirely correct because in agriculture there are many seasonal, part-time and irregular workers. In the period 2008–2014 in agriculture, forestry and fishing both the fatal accidents and the fatal accident rate (cases for 100,000 workers) diminished, but they were always high (Table 1). It is, furthermore, difficult to imagine a constant decreasing trend, because the fatal accidents are very variable in time; for example, it was observed by the Eurostat data that the fatal accidents decreased from 591 to 484 from 2008 to 2009, but in 2010 they increased again to 583. The most common severe accidents in agriculture involve machineries and vehicles (also during repair Croat. j. for. eng. 39(2018)1

and maintenance works), animals and falling from height (Eurostat 2014). The tractor is the main cause of occupational fatalities in agriculture (Lee et al. 1996): in the US in 1998, 32% of fatal injuries in agriculture were machinery-related accidents (Myers 2002) and many fatalities are due to tractor rollover (Erlich et al. 1993, Bernik and Jerončič 2008, HSE 2015, INAIL 2015). Among all the agro-forestry tasks, tree and forestry works are considered high risk activities, with the sector having high fatal and injury rates (HSE 2015, INAIL 2015): manual and mechanical logging are among the most hazardous operations because operators work with potentially dangerous machines and use vehicles running through rough and sloped terrains (Blombäck et al. 2003). They are, moreover, exposed to the effects of bad weather and tasks are physically demanding: the long and repetitive nature of the work causes a range of health problems, including severe back pain

M. Manzone and A. Calvo

Trailer Overturning during Wood Transportation: an Experimental Investigation ... (97–108)

Table 1 Fatal accidents at work in some economic sectors in 2008 and 2014 (EU 28), source Eurostat data Fatal accidents 2008

Fatal accident rate 2008

Fatal accidents 2014

Fatal accident rate 2014

Absolute variation (fatal accidents)

Absolute variation (fatal accident rate)

Agriculture, forestry and fishing

591

8.06

536

5.81

–55

–2.25

Construction

1258

7.44

782

6.08

–476

–1.36

Transportation

711

6.75

622

5.84

–89

–0.91

Manufacturing

837

2.37

574

1.78

–263

–0.59

Human health and social activities

0.23

0.3

0.07

(HSE 2015). Also, in this case, the highest risk is due to tractor overturning: high tractor center of gravity and stability loss causes accidents which are often fatal, as observed by several authors (Maybryer 1952, Knapp 1968, Cole et al. 2006, Myers et al. 2009). A high number of accidents in hill and mountain areas are due to tractor overturning in sloped fields (Springfeldt 1996). The expected tractor overturn may be sideway or rearward (Kim and Rehkugler 1987). There are two types of stability: Þ static stability – when a tractor is not moving Þ dynamic stability – during tractor movement. Accidents usually occur when the tractor is moving. The key factors affecting the dynamic stability (Spencer and Gilfillan 1976) are both exogenous (environment dependent: slope, washboards, stones, rough terrain, potholes, ground obstacles) and endogenous (driver dependent: forward speed, driving style, slip, tiredness). Hunter (1991) established that more than 55% of the total tractor rollover accidents were caused by exceeding tractor limitations due to steep slopes, high speeds, and rough terrain. Moreover, the overturn risk could increase in presence of additional masses fitted on the tractor such as ballast, towed implements, and trailers (Yisa et al. 1998): the safety of larger vehicles is a matter of concern because they usually have large bodies with a high center of gravity and high loading capacity (Chou and Chu 2014). Many studies have been carried out on the tractortrailer stability and many simulations and mathematical models have been studied concerning both the static and dynamic stability, with focus on automotive sector (Chisholm 1979a, Blythe 2007, Mai et al. 2008, Barbieri et al. 2014). Moreover, the studies carried out on agricultural tractor stability were based on laboratory tests (Karkee et al. 2011, Guzzomi 2012, Ahmadi, 2013, Baker et al. 2013, Mazzetto et al. 2013, Previati et al. 2014), while in this case, the study was performed in the field.

In the agroforestry sector, there are many studies regarding the tractor stability (Davis 1974, Chisholm 1979b, Song 1989, Ahmadi 2011, Gravalos et al. 2011, Franceschetti et al. 2014, Li et al. 2014, Li et al. 2016), but there are only very few studies concerning the tractor-trailer stability analysis in forestry (Melemez et al. 2013, Manzone and Balsari 2014, Manzone 2015). Bietresato et al. (2015) proposed a methodological approach for the evaluation of an agricultural wheeled tractor equipped with different implements while operating on sloping hillsides. The combination of the slope and uneven ground are limiting factors for a safe use of tractor in steep terrain and the use of a trailer may worsen the situation, especially during transport of heavy and unstable loads, such as logs (Pereira et al. 2011). Moreover, in some cases, the use of trailers may be dangerous because in presence of a little traction, the trailer can push the tractor off the road because of its small mass compared to the gross mass of the trailer (Lindroos and Wasterlund 2014). A solution to this problem is the use of trailers equipped with motor axles. In this case, the gross mass of the trailer improves the traction of the combined vehicles (i.e. tractor plus trailer). Recently, at the University of Turin, an innovative electronic control system for motorized axles has been developed (Manzone and Balsari 2015, Manzone 2015), able to synchronize the forward speed of the trailer to that of the tractor, independently of the tractor type used. Nevertheless, the frame structure of the trailer and its articulation point with the tractor may have a fundamental role in the trailer traction and in the convoy stability, especially when driving on sloped terrain. Wood transportation using low-powered tractors with little trailers (single-axle or two-axle) is a common practice in the Italian alpine West Regions, characterized by steep and rough terrains. In this area, small scale logging companies with obsolete tractors and equipment are spread (Spinelli et al. 2013). The Croat. j. for. eng. 39(2018)1

Trailer Overturning during Wood Transportation: an Experimental Investigation ... (97–108)

target of these companies, however, is to guarantee wood regeneration and to optimize the environment resiliency, ensuring the operator safety. For these reasons, the aim of this research was to analyze the potential strains on the tractor caused by the trailer overturning. The trailers rollover tests were carried out using a wooden wedge during the convoy moving downhill. In detail, the effects of the trailer joint point (longitudinal and transversal strains, tractor rear wheel detachment from the ground) on the trailer overturning during wood transportation were analyzed, using different types of combination of tractors and regular trailers.

M. Manzone and A. Calvo

Table 3 Trailers characteristics Single-axle (A)

Single-axle (B)

Two-axle (C)

Trailer mass, kg

235

240

330

Flatbed width, m

1.50

Flatbed height, m

0.75

Flatbed length, m

3.00

Wheel dimension

195 R14

Hooking height, m

0.35

0.75

Centre of mass*, m

1.27

1.28

2.33

2. Materials and methods 2.1 Machines Tests were carried out with two 2WD and one crawler tractors (named respectively #1, #2 and #3, Table 2). To avoid the tractor mass influence on the system (tractor and trailer) stability, all the tractors used in the tests had a similar mass of about 1.5 t each, driver mass included (64 kg). The differences between the wheeled tractors were the wheel dimensions and total width (Table 2). All tractors were equipped with ROPS and seat belts. Different tractor types were chosen to analyze the trailer strain over the tractor structure. Three trailers were used: two single-axle trailers (hereafter A and B) and one two-axle trailer (hereafter C) with steering turntable. They had a load floor 3 m long and 1.5 m wide, while the height of the load plat-

Table 2 Tractors characteristics

Fig. 1 Tractor coupling »fork« system form was about 750 mm. The trailers had the same wheel track and their tires had the same dimension (195 R14). The single-axle trailers were different in the hooking height of the towing eye: 0.35 and 0.75 m (half width of the load floor), respectively. The two-axle trailer was coupled to the tractor at the height of 0.35 m (Table 3).

Same Puledro (1)

Fiat 312R (2)

Itma Nike 320 (3)

Power, kW

23.68

22.52

25.51

Mass, kg

1433

1462

1514

Wheels

Tracks

–

Rear wheel type

280/85–24

280/85–28

–

2.2 Environment characteristics

Front wheel type

5.00–15

5.50–16

–

Wheelbase, m

1.50

1.77

–

Overall width, m

1.38

1.81

1.23

Tests were carried out on a short dirt road (about 15 m), which connects a municipal road with a private dirt road. The path was not traced along the line of maximum slope, but transversal to the hillside. The average slopes of the path were about 30% longitudinal and 20% transversal. The path with these charac-

Propulsion system Driving wheel

* Does not include the mass of the driver (64kg) ** Measure calculated in correspondence with the propulsion wheels

Croat. j. for. eng. 39(2018)1

The trailers had the rotating towing eye and were hauled to the tractor with a »fork« system (Fig. 1).

M. Manzone and A. Calvo

Trailer Overturning during Wood Transportation: an Experimental Investigation ... (97–108)

teristics was chosen because there was a flat area that could simplify the maneuvers of the trailer re-overturn. Furthermore, this horizontal plane was useful for removing wood and consequently for reloading the trailer with the overturned logs.

2.3 Trailer rollover simulation The trailer rollover simulation occurred with the convoy moving downhill along the path and placing an artificial obstacle in front of the trailer wheels. The use of an artificial obstacle was necessary to make the test repeatable. The obstacle was made of a wooden wedge (100 mm height, 300 mm length and 200 mm width). To ensure the tractor driver safety during the trailer rollover, the tipping point of the trailer was identified in a point where the tractor was already in the flat area (at the basis of the identified path). Furthermore, to improve the test safety, a rope anchorage was placed between the base of a tree upstream the track and the frame of the tractor rollover protection system (ROPS). The first end of the rope was fixed to the ROPS using a knot, while the second end was rolled around the base of the trunk where a second operator kept this end of the rope in his hands. In this way, the second operator let the rope slide around the trunk during the tractor forwarding. In emergency situations (e.g. overturning of the trailer), the second operator promptly intervened and prevented the tipover of the tractor, stopping the rope sliding (technique commonly used to control the fall of large branches during pruning in tree-climbing) (Fig. 2).

Fig. 2 Scheme of safety anchorage used during tests

100

During all the tests, the trailers were loaded with the same logs of about 1.5 m length and placed transversely to the load platform. The trailer gross masses (wooden and trailer weight) were equal to the tractor mass (1500 kg). This choice was the result of a survey carried out in some Italian forestry yards where, in extreme sloped conditions (with a high risk of trailer overturning), the trailer gross mass does not exceed the tractor mass. Logs had a regular shape (cylinder) with an external diameter between 120 mm and 260 mm. Each log was numbered with a numeric code and, each time the load was applied, logs were placed on each trailer in the same identified position. This precaution was taken so that the same load distribution and the same weight on the towing eye were provided during all the performed tests: 543.7±1.2 kg for the two single-axes (A and B) and 9.4 kg for the biaxle (C). In the case of trailer C, the precise number is ascribable to the unique drawbar weight. Tests were carried out both with the load held in place only by steel supports fixed at the two ends of the load floor (front and rear) and with the load tied by two ropes placed diagonally to the longitudinal axis of the trailer. The average forward speed was always around 3 km h–1 and three overturning repetitions were performed for each trailer and for each tractor type (27 tests).

2.4 Measurements and instruments During each test, measurements were made of: Þ lateral deviation (side slipping) of rear wheels/ crawler of the tractor Þ detachment of rear upstream wheel/crawler from the ground Þ longitudinal and transversal strains (released over the tractor hooking system) produced by the trailer overturn. The lateral deviation (side slipping) of the tractor wheels/crawler was measured using a graduate steel ruler (1 mm precision). This measurement was performed starting from the rubber tire crampons until the end of the sideslip on the ground. The detachment height of the wheel/crawler from the ground was evaluated using a measurement device (fixed on the mudguard) made with a plastic graduate strip (1 mm scale) rolled on a reel without return spring. The end of the strip was linked to a steel support (1 kg mass), sliding on the ground by a small rope linked to another support fixed to the tractor frame. Before the start of the test, the strip was stretched: this condition was maintained until the Croat. j. for. eng. 39(2018)1

Trailer Overturning during Wood Transportation: an Experimental Investigation ... (97â&#x20AC;&#x201C;108)

M. Manzone and A. Calvo

strain). In all the test conditions, each device was fixed by two bolts in the centre of the ROPS top (Fig. 4). Positive measurements were clockwise. In detail, front longitudinal measured strains were positive, negative at rear. Transversal strains were positive at left, negative at right. The goniometer enabled the measurement of small displacements in linear unit (mm) (Timoshenko and Gere 1976): for this reason, in data elaboration, only linear measurements were considered.

2.5 Data processing

Fig. 3 Scheme of the system used to measure the detachment height of the wheel/crawler from the ground trailer overturning. At this occurrence, the strip extended in function of the wheel/crawler detachment: the strip length difference was the detachment height of the wheel/crawler from the ground (Fig. 3). The transversal and longitudinal strains were calculated using a specific device based on mechanic pendulum. It was built connecting a mechanic pendulum (100 mm length) to a goniometer (120 mm diameter) by a centre hinge (Fig. 4): a weight (20 g mass) was joined at the unrestrained extremity of the pendulum. To measure the maximum transversal and longitudinal strains, two metallic pointers (free to move) were added to the same hinge (Fig. 4): at the beginning of each test, these pointers were aligned to the pendulum (point zero). Two mechanics pendulum were used during the tests: the first positioned orthogonally to the tractor forward speed (to measure the transversal strain), the latter parallel (to measure the longitudinal

Fig. 4 Mechanic pendulum system used in the test Croat. j. for. eng. 39(2018)1

Data processing was performed using Microsoft Excel and IBM SPSS V. 22.0 Statistic package. The ANOVA and post-hoc Tukey tests were performed in order to evaluate possible differences between the tested trailers, tractors and loose or tied loads (Keppel and Wickens 2004). Tukey test was chosen for its high power for this data distribution (Tukey 1949). Tests differences were evaluated considering a=0.05.

3. Results 3.1 Lateral deviation and ground detachment height of the tractor wheels/crawler The highest lateral deviation of the tractor propulsion system was produced by the single-axle trailer with the lowest hooking point (A). It always caused

Fig. 5 Box & whisker graph of lateral deviations measured on the tested convoy configurations during trailer overturning

101

M. Manzone and A. Calvo

Trailer Overturning during Wood Transportation: an Experimental Investigation ... (97â&#x20AC;&#x201C;108)

Table 4 Lateral deviation and ground detachment measured on the tractor wheel/crawler during trailer overturning Load

Tractor

Loose

Tied

Trailer

Wheel / crawler lateral deviation, mm Mean

Min

Wheel/crawler ground detachment, mm

Max

Mean

Min

Max

11.5

3.8

4.5

4.6

4.2

0.0

5.9

5.7

3.2

0.0

5.3

0.0

2.1

3.0

1.0

0.0

2.0

0.0

6.5

4.6

2.5

0.0

5.5

0.0

the exception of tractor 2 with the loose load, the biaxle trailer (C) had the lowest lateral deviation values (never higher than 32 mm). If the load was tied, the same tractor (#2) presented higher lateral deviation data (about twice, independently of the trailer type). Tractor #2 and #3 showed the lower lateral deviation data of the wheel (or crawler). The single-axle trailer with the low hook (A) always caused the highest detachment values of the rear wheel (or crawlers) from the ground (Fig. 6), with values around 30 mm in the wheeled tractors with the loose load (about 40 mm with the tied load): the single-axle trailer with the high hook (B) caused only a slight detachment value of 8 mm in tractor #1. The bi-axle trailer never caused any wheel or crawler detachment from the ground, independently of the tractor and load type.

3.2 Longitudinal strain Fig. 6 Box & whisker graph of the rear wheel (or crawler) ground detachment measured on the tested convoy configurations during trailer overturning the lifting (of some millimeters) of the tractor rear axle in all the tractor overturning tests (Table 4). Lateral deviations were in a range between 7 and 68 mm (Fig. 5): the trailer A showed the highest lateral deviations in the wheeled tractors from 14 to 68 mm (in the crawler tractor the single-axle trailers A and B had about the same values, around 18 mm) while, with

102

Similar results for the strain measured on the tractors were found in the single-axle trailer with the high hook (B) and in the bi-axle trailer with steering turntable (C): here the longitudinal strain (the pointer movement along the goniometer during the trailer rollover) produced swinging intervals lower than 15Â mm (Table 5). Different values were obtained for the single-axle trailer with the low hook (A): in this case the longitudinal strain was higher and varied in a wider interval between 6 and 20 mm. Also in the case of the total longitudinal strain, the trailer with the low hook produced the highest values (Fig. 7). Differently from the tests discussed in paraCroat. j. for. eng. 39(2018)1

Trailer Overturning during Wood Transportation: an Experimental Investigation ... (97–108)

M. Manzone and A. Calvo

Table 5 Longitudinal strains measured on tractors during trailer overturning Load

Tractor

Loose

Tied

Trailer

Front, mm

Rear, mm

Mean

Min

Max

Mean

Min

Max

11.0

1.00

–13.0

–12

–14

1.00

6.7

1.15

–7.0

–6

–8

1.00

6.3

1.15

–8.3

–7

–9

1.15

11.7

2.08

–7.3

–6

–8

1.15

4.7

0.58

–5.7

–5

–6

0.58

5.7

1.53

–7.3

–6

–8

1.15

6.7

0.58

–6.3

–6

–7

0.58

4.3

0.58

–3.3

–3

–4

0.58

4.7

0.58

–4.3

–4

–5

0.58

15.0

2.65

–17.7

–16

–20

2.08

11.7

1.15

–12.0

–10

–14

2.00

11.3

1.15

–10.7

–10

–11

0.58

3.3 Transversal strain The ransversal strain was generally higher than the longitudinal one (Table 5 and Table 6): also in this case the single-axle trailer with low hook (A) caused the highest boosts on the tractor (Fig. 8). The trailer with the low hook (A) generated transversal strains always 30 % higher than the same trailer type with the high

Fig. 7 Box & whisker graph of the rear wheel (or crawler) ground detachment measured on the tested convoy configurations during trailer overturning graph 3.1, the bi-axle trailer (C) always produced higher total longitudinal strains than the single-axle trailer with high hook (B) for loose load. The highest longitudinal strains were observed with the tied load, independently of the trailer type. The crawler tractor best absorbed the strains caused by the trailer overturning (Fig. 7). Croat. j. for. eng. 39(2018)1

Fig. 8 Box & whisker graph of transversal strains measured on the tested convoy configurations during trailer overturning

103

M. Manzone and A. Calvo

Trailer Overturning during Wood Transportation: an Experimental Investigation ... (97–108)

Table 6 Transversal strains measured on tractors during trailer overturning Load

Tractor

Loose

Tied

Trailer

Left, mm Mean

Min

Right, mm Max

Mean

Min

Max

15.0

1.00

–10.3

–11

–9

1.15

11.0

1.00

–8.7

–9

–8

0.58

7.7

1.53

–7.7

–9

–7

1.15

11.0

1.00

–13.3

–15

–11

2.08

7.0

1.00

–3.3

–4

–3

0.58

4.7

1.15

–4.0

–5

–3

1.00

11.7

0.58

–10.0

–11

–9

1.00

7.0

2.08

–4.3

–5

–4

0.58

6.0

2.08

–7.0

–8

–6

1.00

20.3

3.79

–21.3

–24

–20

2.31

14.3

2.08

–13.7

–15

–13

1.15

14.3

1.53

–14.3

–16

–12

2.08

hook (B): there are even peaks of 100% when the tractor #2 was used, with the loose or tied load (Fig. 8). For this parameter, too, the highest values (between 26 and 43 mm) were obtained when operating with the tied load (Fig. 8).

3.4 Influence of tractor and trailer structure on measured parameters Results highlighted that the tractor structure had an important role in the trailer overturning. The ANOVA procedure evidenced tractor similarities in only one parameter, the ground rear wheel/

crawler detachment, mainly influenced by the trailer type (Table 7). In absolute terms, even though all the tractors had the same mass, the crawler and the bigger wheeled tractors guaranteed higher resistance forces to the strains provided by different trailers (Table 8). In detail, tractor #2 and #3 showed similar values in lateral deviation and in both longitudinal and transversal strain data. Some significant differences among the three tractor types were observed in the ground detachment of the rear wheels or crawlers (Table 8). Statistical analysis showed similar results for all types of tested trailers only in lateral deviation tests

Table 7 ANOVA statistical analysis of tractors SS Ground detachment

Lateral deviation

Longitudinal strain

Transversal strain

104

Significance

Among groups

156.911

78.456

0.529

0.596

Inside groups

3560.136

148.339

–

Total

3717.047

–

Among groups

3834.741

1917.370

20.875

<0.0001

Inside groups

2204.444

91.852

–

Total

6039.185

–

Among groups

259.556

129.778

7.765

0.003

Inside groups

401.111

16.713

–

Total

660.667

–

Among groups

154.741

77.370

2.269

0.025

Inside groups

818.444

34.102

–

Total

973.185

–

Croat. j. for. eng. 39(2018)1

Trailer Overturning during Wood Transportation: an Experimental Investigation ... (97–108)

M. Manzone and A. Calvo

Table 8 Tukey test for different parameters of tractors tested Tractor code

Lateral deviation, mm

14.44

–

4.67

9.89

–

14.78

–

18.33

–

9.78

14.33

15.33

–

41.44

9.78

–

17.44

–

20.11

0.670

1.000

0.651

0.074

0.259

0.150

0.230

Significance

Ground detachment, mm

Longitudinal strain, mm

Transversal strain, mm

Subset for a=0.05

Table 9 ANOVA statistical analysis of trailers

Ground_detach

Lateral_dev

Long_strain

Trans_strain

Among groups

3146.156

1573.078

66.131

<0.0001

Inside groups

570.891

23.787

–

Total

3717.047

–

Among groups

983.630

491.815

2.335

0.118

Inside groups

5055.556

210.648

–

Total

6039.185

–

Among groups

324.222

162.111

11.564

<0.0001

Inside groups

336.444

14.019

–

Total

660.667

–

Among groups

682.741

341.370

28.208

<0.0001

Inside groups

290.444

12.102

–

Total

973.185

–

Table 10 Tukey test for different parameters of trailers tested Trailer code

Lateral deviation, mm

B A Significance

Ground detachment, mm

Longitudinal strain, mm

17.22

0.01

–

12.44

–

12.33

–

25.00

0.89

–

10.56

–

14.11

–

32.00

–

23.33

–

18.67

–

23.78

0.099

0.922

1.000

0.541

1.000

0.533

1.000

(Table 9): for the other parameters, the p value was always lower than 0.0001. The analysis of statistical data showed that the single axle trailer with high hook (B) and the bi-axle trailer with steering turntable (C) produced the same statistical results (Table 10).

Transversal strain, mm

4. Discussion

sence of ground detachment of the rear wheels (or crawlers), also produced the lowest strain (longitudinal and transversal) to the tractor during its overturning. The higher absolute values observed in longitudinal solicitations for the bi-axle trailer are only attributable to the higher trailer tare and, consequently, to the higher resistance force exercised by the trailer during the overturning (rear longitudinal strain values of trailer C are presented in Table 5).

The study highlighted that trailers with an »articulate« drawbar (turntable steering), apart from giving lower lateral deviation data and resulting in the ab-

The best performance of the bi-axle trailer (C) could be explained with the higher number of junction joints (4 swivel joints) of the steering turntable system

Croat. j. for. eng. 39(2018)1

105

M. Manzone and A. Calvo

Trailer Overturning during Wood Transportation: an Experimental Investigation ... (97–108)

Fig. 9 Swivel joints in different drawbar types compared to a drawbar fixed to the trailer frame (2 swivel joints, Fig. 9). In fact, the articulate drawbar (Trailer C) is able to compensate different types of critical points during the trailer overturning (support points, coupling point, etc.) (Fig. 10a). In contrast, a drawbar fixed to the trailer frame (Trailer A) could generate an overturning force on the coupling point of the tractor, mostly, if this latter showed a height lower than half of the load floor width of the trailer (Fig. 10b). This can cause a higher instability of the tractor, especially if it has no brakes on front axle and, as a consequence, there is a higher possibility of overturning. On the contrary, if the height of coupling point of the tractor is equal to half of the trailer width (Trailer B), no overturning forces are generated on the tractor itself (Fig. 10c). Unfortunately, these results are in contrast with the »guidelines« usually adopted in forestry activities, where trailers with fixed drawbar are preferred to

trailers with an articulate drawbar (Manzone 2015). In fact, the use of the first trailer type guarantees a higher traction force because part of the trailer load is discharged on the tractor and because it shows a greater simplicity in maneuvering due to minor articulation points of the convoy structure. It was moreover observed that the loose load guaranteed a better safety during the trailer overturn: the tied load is dangerous because during the trailer overturning its mass is added to the trailer mass and causes higher strains to the tractor. Concerning the tractor structure, the wide system propulsion tracks guaranteed the best absorption of the trailer strain caused by its overturning: in fact, the crawler tractor had the best performances, as opposed to the tractor with the smallest wheels.f Other Authors (Marinello et al. 2013) simulated and tested a double steering trailer prototype composed o two single-axle trailers, which is a good solution for the transport of high length logs (more than 8–10 m long) on narrow and steep forest roads in order to reduce the curve radius and to improve the convoy maneuverability. Nevertheless, on the basis of the results of this study, this prototype may produce the same strains as the single-axle during the trailer overturning (although it seems to have a bi-axle structure) because the first module of the prototype has a fixed single-axle drawbar.

5. Conclusions The study highlighted that the bi-axle trailer structure with a turntable steering had the best performances compared to the single-axle trailer in terms of safety during trailer overturning. These results are not entirely in line with forestry practices that usually suggest the use of a single-axle trailer because it discharges part of the load on the coupled tractor (Spinelli et al. 2013). Independently of the trailer type considered in this work, a tied load is more dangerous than a load restrained only by steel struts, because during the overturn the load forms a single unit with the trailer mass, which increases the transversal and longitudinal strain.

6.References Ahmadi, I., 2011: Dynamics of tractor lateral overturn on slopes under the influence of position disturbances (model development). Journal of Terramechanics 48(5): 339–346.

Fig. 10 Scheme of trailer overturning on flat ground in different configurations

106

Ahmadi, I., 2013: Development of a tractor dynamic stability index calculator utilizing some tractor specifications. Turkish Journal of Agriculture and Forestry 37(2): 203–211. Croat. j. for. eng. 39(2018)1

Trailer Overturning during Wood Transportation: an Experimental Investigation ... (97–108) Baker, V., Guzzomi, A. L., 2013: A model and comparison of 4-wheel-drive fixed-chassis tractor rollover during phase I. Biosystems Engineering 116(2): 179–189. Barbieri, F. A. A., Lima, V. D. A., Garbin, L., Boaretto, J., 2014: Rollover study of a heavy truck combination with two different semi-trailer suspension configurations. SAE Technical Papers Volume 3, 8th SAE Brasil International Suspension and Trailer Colloquium and Engineering Exhibition, 7–9 May, BRASILCOLL, Rio Grande do Sul, Brasil, Code 107159: 1–12. Bernik, R., Jerončič, R., 2008: The Research of the Number of Accidents with the Agriculture and Forestry Tractors in the Europe and the Main Reasons for those Accidents. Strojniški vestnik (Journal of Mechanical Engineering) 54(7–8): 557– 564. Bietresato, M., Carabin, G., Vidoni, R., Mazzetto, F., Gasparetto, A., 2015: A Parametric Approach for Evaluating the Stability of Agricultural Tractors Using Implements during Side-Slope Activities. Contemporary Engineering Sciences 8(28): 1289–1309. Blombäck, P., Poschen, P., Lövgren, M., 2003: Employment Trends and Prospects in the European Forest Sector, (ILO). Geneva Timber and Forest Discussion Papers, United Nations 1–45 p. Blythe, W., 2007: Tractor-trailer Response to High Crosswinds; Static Computations and Dynamic Simulations. SAE Technical Papers 2007-01-4255: 1–18. Chisholm, C. J., 1979a: A mathematical model of tractor overturning and impact behaviour. Journal of Agricultural Engineering Research 24(4): 375–394. Chisholm, C. J., 1979b: Experimental validation of a tractor overturning simulation. Journal of Agricultural Engineering Research 24(4): 395–415. Chou, T., Chu, T. W., 2014: An improvement in rollover detection of articulated vehicles using the grey system theory. Vehicle System Dynamics 52(5): 679–703. Cole, H. P., Myers, M. L., Westneat, S. C., 2006: Frequency and severity of injuries to operators during overturns of farm tractors. Journal of Agricultural Safety and Health 12(2): 127–138. Davis, D. C., Rehkugler, G. E., 1974: Agricultural wheel tractor overturns, Part II: mathematical model verification by scale model study. Transactions of the ASAE 17(3): 484–492. Erlich, S. M., Driscoll, T. R., Harrison, J. E., Frommer, M. S., Leigh, J., 1993: Work-related agricultural fatalities in Australia, 1982–1984. Scandinavian Journal of Work, Environment & Health 19(3): 162–167. Eurostat, 2014: Agriculture, forestry and fishery statistics, Statistical books. Luxembourg: Publications Office of the European Union, 1–199 p. Franceschetti, B., Lenain, R., Rondelli, V., 2014: Comparison between a rollover tractor dynamic model and actual lateral tests. Biosystems Engineering 127(1): 79–91. Croat. j. for. eng. 39(2018)1

M. Manzone and A. Calvo

Gravalos, I., Gialamas, T., Loutridis, S., Moshou, D., Kateris, D., Xyradakis, P., Tsiropoulos, Z., 2011: An experimental study on the impact of the rear track width on the stability of agricultural tractors using a test bench. Journal of Terramechanics 48(4): 319–323. Guzzomi, A. L., 2012: A revised kineto-static model for phase I tractor rollover. Biosystems Engineering 113(1): 65–75. HSE (Health and Safety Executive), 2015: Health and Safety in Agriculture, Forestry and Fishing in Great Britain 2014– 2015, www.hse.gov.uk/statistics/ Hunter, A. G. M., 1991: Stability of agricultural machinery on slopes. Progress in Agricultural Physics and Engineering, ed. J. Matthews. Wallingford, Oxon, U.K., CAB International, 83–118 p. INAIL, 2015: Report annuale sugli infortuni mortali e con feriti gravi verificatisi nel 2014 nel settore agricolo e forestale. http://sicurezzasullavoro.inail.it/PortalePrevenzioneWeb/ wcm/idc/groups/prevenzione/documents/document/ ucm_184751.pdf Karkee, M., Steward, B., 2011: Parameter estimation and validation of a tractor and single axle towed implement dynamic system model. Computers and Electronics in Agriculture 77(2): 135–146. Keppel, G., Wickens, T. D., 2004: Design and analysis: A researchers’ handbook (4th Edition). Pearson, Prentice Hall, Upper Saddle River, NJ, USA, 1-611 p. Kim, K. U., Rehkugler, G. E., 1987: A review of tractor dynamics and stability. Transactions of the ASAE 30(3): 615– 623. Knapp, L. W., 1968: The farm tractor: overturn and power take-off accident problem. Bulletin No 11, Institute of Agricultural Medicine, University of Iowa, USA, 1–51 p. Lee, T. Y., Gerberich, S. G., Gibson, R. W., Carr, W. P., Shutske, J., Renier, C. M., 1996: A population-based study of tractor-related injuries: Regional Rural Injury Study-I (RRISI). Journal of Occupational Environmental Medicine 38(8): 782–793. Li, Z., Mitsuoka, M., Inoue, E., Okayasu, T., Hirai, Y., 2014: Dynamic analysis of agricultural wheel tractor driving on uneven surface under the influences of speed and slope angle. Journal of the Faculty of Agriculture, Kyushu University 59(2): 339–343. Li, Z., Mitsuoka, M., Inoue, E., Okayasu, T., Hirai, Y., Zhu, Z., 2016: Parameter sensitivity for tractor lateral stability against Phase I overturn on random road surfaces. Biosystems Engineering 150(1): 10–23. Lindroos, O., Wasterlund, I., 2014: Theoretical potentials of forwarder trailers with and without axle load restrictions. Croatian Journal of Forest Engineering 35(2): 211–219. Mai, L., Zong, C. F., Gao, Y., Huang, C., 2008: Simulation and analysis of roll stability for heavy tractor semi-trailers. Journal of Jilin University (Engineering and Technology Edition) 38(2): 5–10.

107

M. Manzone and A. Calvo

Trailer Overturning during Wood Transportation: an Experimental Investigation ... (97–108)

Manzone, M., 2015: Performance of an electronic control system for hydraulically driven forestry tandem trailers. Biosystems Engineering 130(1): 106–110. Manzone, M., Balsari, P., 2015: Electronic control of the motor axles of the forestry trailers. Croatian Journal of Forest Engineering 36(1): 131–136. Marinello, F., Grigolato, S., Sartori, L., Cavalli, R., 2013: Analysis of a double steering forest trailer for long wood transportation. Journal of Agricultural Engineering 154(2–3): 10–15. Maybrier, O., 1952: Safety guard for a tractor operator. US Patent 2,729,462. Filed on July 28. Issued on January 3, 1956, 1–5 p. Mazzetto, F., Bietresato, M., Vidoni, R., 2013: Development of a dynamic stability simulator for articulated and conventional tractors useful for real–time safety devices. Applied Mechanics and Materials 394(1): 546–555. Melemez, K., Di Gironimo, G., Esposito, G., Lanzotti, A., 2013: Concept design in virtual reality of a forestry trailer using a QFD–TRIZ based approach. Turkish Journal of Agriculture and Forestry 37(6): 789–801.

Pereira, D., Fiedler, N. C., Soares de Souza Limam, J., De Oliveira Bauer, M., Rezende, A. V., Missiaggia, A. A., Pavesi Simão, J. B., 2011: Lateral stability limits of farm tractors for forest plantations in steep areas. Scientia Forestalis 39(92): 433–439. Previati, G., Gobbi, M., Mastinu, G., 2014: Mathematical models for farm tractor rollover prediction. International Journal of Vehicle Design 64(2/3/4): 280–303. Song, A., Huang, B. K., Bowen, H. D., 1989: Simulating a powered model wheel–tractor on soft ground. Transactions of the ASAE 32(1): 2–11. Spencer, H. B., Gilfillan, G., 1976: An approach to the assessment of tractor stability on rough sloping ground. Journal of Agricultural Engineering Research 21(2): 169–176. Spinelli, R., Magagnotti, N., Facchinetti, D., 2013: A survey of logging enterprises in the Italian Alps: firm size and type, annual production, total workforce and machine fleet. International Journal of Forest Engineering 24(2): 109–120. Springfeldt, B., 1996: Rollover of tractors: international experiences. Safety Science 24(2): 95–110. Timoshenko, S. P., Gere, J., 1976: Mechanics of Materials. Nelson Thornes Ltd, UK, 1–832 p.

Myers, M. L., 2002: Tractor risk abatement and control as a coherent strategy. Journal of Agricultural Safety and Health 8(2): 185–198.

Tukey, J., 1949: Comparing Individual Means in the Analysis of Variance. Biometrics 5(2): 99–114.

Myers, M. L., Cole, H. P., Westneat, S. C., 2009: Injury severity related to overturn characteristics of tractors. Journal of safety research 40(2): 165–170.

Yisa, M. G., Terao, H., Noguchi, N., Kubota, M., 1998: Stability criteria for tractor–implement operation on slopes. Journal of Terramechanics 35(1): 1–19.

Authors’ addresses:

Received: June 28, 2016 Accepted: October 3, 2017

108

Marco Manzone, PhD. * e-mail: marco.manzone@unito.it Angela Calvo, PhD. e-mail: angela.calvo@ unito.it Department of Agricultural, Forest, and Food Science University of Turin Largo Braccini 2, 10095 Grugliasco, Turin ITALY *Corresponding author Croat. j. for. eng. 39(2018)1

Original scientific paper

Analysis of Hazardous Emissions of Hand-Operated Forestry Machines Fuelled with Standard Mix or Alkylate Gasoline Aldo Calcante, Davide Facchinetti, Domenico Pessina Abstract In addition to safety, small hand-operated forestry machines can be criticised for affecting the operators’ health, especially because of high levels of exhaust gas emissions, noise and vibrations. In this study, gas emissions, noise and hand-arm vibrations (HAV) levels have been measured on chainsaws, hedge cutters and blowers fuelled with two different types of fuel: a commercial RON 95 gasoline with the addition of 2% of synthetic oil suitable for two-stroke engines and, as an alternative, a specific advanced mixture available on the market, based on alkylate gasoline. For two different running conditions, i.e. with the engine at idle speed and when executing a typical working routine (maximum speed with load), tests were carried out for: Þ gas emissions, using a gas analyser, for measuring the volatile organic compounds (VOC) values Þ noise, using a sound level meter, to record the levels at both of the operator’s ears Þ HAV, using a tri-axial accelerometer fixed on the handgrip(s) of the machines. The results demonstrated that, when using the alkylate fuel, the VOC emissions were reduced, in the considered machines, from 23 to over 77%, while for noise and HAV, the differences in level were not statistically significant. The present study confirms that the reduction in the amount of emissions can be remarkably improved by adopting advanced fuels that lead to a more efficient combustion process. Keywords: alkylate fuel, blower, chainsaw, hedge cutter, exhaust gas emission, noise, hand-arm vibrations

1. Introduction In forestry, but also in green maintenance and even more in hobby-oriented fields, the use of small handoperated machines is widespread. This category of machines is widely diffused at a worldwide level; in the Italian market alone, in 2014 more than 700,000 small hand-operated machines were purchased for professional and hobby use (Comagarden/Morgan 2015). The internal combustion engine of specialised machines for forestry and green maintenance (i.e. chain saws, lawn mowers, hedge cutters, blowers, etc.) is typically a two-stroke spark ignition, fuelled by gasoline with synthetic oil added to provide suitable enCroat. j. for. eng. 39(2018)1

gine lubrication. This simplifies manufacturing, as it is necessary to provide reduced mass, a small size and a relatively low production cost. For example, the engines of chain saws typically have a very high specific power, up to 45 W/cm³, but at the same time, they have a reduced mass, even less than 115 g/cm³ (Wòjcik 2002). The two-stroke internal combustion engine offers a heat efficiency that is lower than a four-stroke internal combustion engine, due to the incomplete combustion of the air-fuel mixture into the cylinder. The main consequences are higher fuel consumption and higher emission of exhaust gas, with damaging effects on the environment and human health (Bünger et al. 1997, Magnusson et al. 2000, Skoupi et al. 2010).

109

A. Calcante et al.

Analysis of Hazardous Emissions of Hand-Operated Forestry Machines ... (109–116)

In detail, the representative pollutants emitted are carbon monoxide (CO), nitrogen oxides (NOx), unburned hydrocarbons (HC), sulphur compounds (SO2 and H2S) and particulate matter (PM) (Wòjcik and Skarżyński 2006, Volckens et al. 2007, Nilsson et al. 2010). Moreover, the exhaust emission includes other gases (not dangerous for the operator but sometimes harmful for the climate), such as nitrogen (N2), oxygen (O2) and carbon dioxide (CO2) (Wòjcik and Skarżyński 2006). In the statement »Trend of Emissions in 2002«, the US Environmental Protection Agency (EPA) (EPA 2005) underlined that the emissions of HC and CO from this type of engine represented approximately 10.5% and 4.8%, respectively, of total emissions in the USA. In order to reduce harmful compounds in the exhaust gas of small hand-operated machines, with Directives 97/42/CE and 99/38/CE, the European Commission established the limits for toxic pollutants by introducing the European Emission Standards and the contextual duty of the manufacturers to install catalytic converters for these particular categories of machines. Besides the accomplishment of the EU Directive, in order to reduce the pollution due to the exhaust gas, alternative fuels specifically developed for hand-operated small machines are available on the market. These fuels are a mix of alkylate gasoline and synthetic oil, produced with the goal of reducing drastically the impact on operators’ health and environment. Alkylate gasoline is a fuel characterised by highly branched alkanes and only very small amounts of arenes and alkenes, derived from the residual gas of petroleum refining. This fuel is practically lacking in sulphur, benzene, aromatics and ethanol. Alkylate gasoline is also a desirable fuel due to its high octane number in comparison with unleaded gasoline (Ostermark 1996, Ostermark and Petersson 1993). Nowadays, the main limitation to the wide diffusion of this alternative fuel is its high cost (about 2–2.5 times the price of normal fuel, based on unleaded gasoline). To reduce the health risk for the operator, besides the use of alkylate gasoline, it is important to evaluate carefully the performance of small hand-operated forestry machines from further points of view. Apart from personal protective devices and systems created to reduce injuries, other factors affecting the onset of occupational diseases are noise pollution and vibrations transmitted to the hand and arm of the operator (Davis 1978, Groves and Lyons 2013). In fact, these machines typically generate noise levels that can be dangerous for the operator’s hearing (Fonseca et al. 2015). In particular, considering chain saws, Tunay and Melemez (2008) executed an audiometric test on 114 forest loggers to ascertain whether

110

they were experiencing hearing loss, finding a dramatic loss (hearing threshold) reaching the range of 40–50 dB. Moreover, concerning hand-arm vibrations (HAV), small hand-operated forestry machines are characterised by high acceleration levels, often exceeding the limits specified by the national and international standards. Unfortunately, technological progress has not improved this factor in recent years (Monarca et al. 2003). The aim of this work was to study the behaviour of some important parameters influencing the health of operators arising from the use of hand-operated forestry machines (chain saws, hedge cutters and blowers). These machines were fuelled with a commercial RON 95 gasoline (EN 228 2013) with addition of 2% of synthetic oil suitable for two-stroke engines and, as an alternative, a specific advanced mixture available on the market, based on alkylate gasoline (Swedish Standard SS 155461 2008). In particular, field tests were carried out to evaluate the influence of this alternative fuel on pollutant emissions, in terms of VOC (Volatile Organic Compounds, particularly dangerous for the operator’s health and environment), due to fuel combustion. Furthermore, alkylate gasoline leads to a more efficient combustion process due to a higher Motor Octane Number (MON) than conventional gasoline (Stratiev 2011). In fact, conventional gasoline has MON usually equal to 85, whilst the MON value of alkylate gasoline is higher than 90; so it was interesting – as a secondary hypothesis – to verify if the use of alkylate fuel could influence the noise and HAV levels emitted by hand-operated machines. With this aim, field tests were also carried out to measure the noise at the operator’s ear and the HAV affecting the operator’s hand-arm area. All the field tests (VOC, noise and HAV) were carried out considering different engine loads, to simulate the real working conditions in which these machines are typically used.

2. Materials and method The research involved eleven machines: six chain saws, two hedge cutters and three blowers, all equipped with a two-stroke engine, except for one blower equipped with a four-stroke engine, alternatively fuelled with conventional gasoline and alkylate gasoline. The investigated machines show different technical characteristics in terms of displacement, engine power, emission stage, and year of production (Table 1). The characteristics of the two fuels used during the tests are reported in Table 2. Croat. j. for. eng. 39(2018)1

Analysis of Hazardous Emissions of Hand-Operated Forestry Machines ... (109–116)

Machine ID

Manufacturing year

Strokes n

Emission stage

Displacement cm3

Engine power kW

Table 1 Main technical characteristics of machines tested

CS1

1997

n.a. *

40.2

1.9

CS2

2004

Euro 1

30.1

1.2

CS3

2010

Euro 1

55.5

2.8

CS4

2012

Euro 1

45.6

2.2

CS5

2014

Euro 2

31.8

1.5

CS6

2015

Euro 2

35.2

1.8

HC1

2010

Euro 1

41.6

2.0

HC2

2013

Euro 2

45.0

2.2

2011

Euro 1

64.8

2.8

2012

Euro 2

36.3

1.4

2014

Euro 2

27.2

0.8

Category

Chain saw

Hedge cutter

Blower

* n.a. – Not applicable

Table 2 Characteristics of conventional and alkylate gasoline used in tests Fuel characteristic

Conventional gasoline

Alkylate gasoline

Octane number RON

>95

Octane number MON

>90

Sulphur, ppm

<10

Aromatic content, % vol.

<0.5

Benzene content, % vol.

<0.05

Olefins content, % vol.

<0.1

Concerning gas emissions, only the VOC levels were measured, while the noise and HAV values were respectively obtained in accordance with the ISO 22868:2011 and ISO 22867:2011 Standards. The tests involved the use of the following instrumentation: Þ for gas emission: a gas analyser MultiRAE Plus model PGM 50 (Rae Systems, San Jose, USA) Þ for the noise: a noise level meter Cesva mod. SC–30 (CESVA Instruments s.l.u, Barcelona, Spain) equipped with Cesva C–130 ½” condenser microphone, to measure the overall level (in dB(A)), with a sampling rate of 1 s

A. Calcante et al.

Þ for HAV: a four-channel analyser Quest mod. HAVPro (Ashtead Technologies, Letchworth Garden City, UK), equipped with a tri-axial accelerometer of make Dytran, model 3023A2 (Dytran Instruments, Chatsworth, USA), with a sensitivity of approximately 11 mV/g on each axis. According to the UNI EN ISO 5349-1 (2004), the acceleration values on the x, y and z axes were recorded in real time, as well as the RMS value (the square root of the sum of the single squared values) calculated over each single trial duration of 1 min. From the operating point of view, the tests were carried out by the same operator, outdoors in an experimental field. The hand-operated machines were set in both idle and working conditions. In particular, the working conditions considered the maximum engine speed under load or a typical operating condition of use. In detail, the working conditions were as follows: Þ for chain saws: complete cutting of a trunk section of a tree (Robinia Pseudoacacia L.), Ø approximately 400 mm Þ for hedge cutters: to simulate the load produced by the cutting of vegetation, the length of the cutting nylon string was doubled with respect to the one normally used Þ for blowers: the air flow was oriented towards the soil, maintaining the output approximately 0.2 m away from the soil surface. The duration of each trial for measurement of VOC, noise and HAV, was exactly 1 minute in all operating conditions apart from the cutting with chain saws of the trunk of Robinia Pseudoacacia L. In this case, as the single task was completed in a few seconds (depending on the chain saw engine power and the sharpening of the chain), the duration of 1 minute was obtained by consecutive cuttings. Before each test with alkylate and 2% mix gasoline, the engine was warmed for around 5 minutes to burn off all the residues of the fuel previously used. For executing the VOC emission measurements, the probe sensor was located in proximity to the output of the muffler (Fig. 1A). It was not possible to insert the probe sensor directly inside the muffler because in this condition the pump of the analyser, used to capture a given amount of gas to perform the analysis, caused an immediate overflow of the pressure sensor. However, this arrangement did not influence the test results; in fact, the goal was to compare the gas emission composition and amounts in two different and real working conditions, and not to obtain absolute

Croat. j. for. eng. 39(2018)1 111

A. Calcante et al.

Analysis of Hazardous Emissions of Hand-Operated Forestry Machines ... (109–116)

values. The gas analyser was set for a sampling rate recording of 1 Hz; as each trial lasted approximately 1 min, the VOC values (expressed as mg/g) are the average of about 60 data. For the noise measurements, in all the investigated working conditions, the sensor of the noise level meter (the microphone) was placed or held by another operator very close to both ears of the operator using the hand-operated machine (1 minute near to the right ear, 1 minute near to the left ear) (Fig. 1B). According to ISO 22868:2011, the A-weighted sound pressure was measured for each trial, in terms of Equivalent Level, in dB(A). For HAV, the tri-axial accelerometer was fixed to the machines in accordance with ISO 22867:2011. Specifically, it was rigidly fixed with cable ties to the handgrip(s) of the machines. In particular, for chain saws and edge cutters, the HAV values were measured on both the right (driving) and left (transporting/grasping) handgrips, while for blowers, just the single handgrip (drive/grasping) was taken into account (Fig. 1C). The recorded data were then stored and elaborated using a commercial spreadsheet (Microsoft Excel 2013). The statistical analysis, aimed at studying the possible influence of alkylate gasoline on VOC emission, noise levels and HAV, has been carried out using XLSTAT software v. 7.5.2 (Addinsoft, Paris, FR).

3. Results and discussion Hereinafter, the results of VOC emissions, noise and HAV measured during the experimental tests car-

ried out on the selected machines (six chain saws, two hedge cutters and three blowers) are described. Each value (VOC, noise, HAV) is the average, provided by each instrument, of about 60 raw data recorded during a single experimental test.

3.1 Gas emissions In general, when fuelled with alkylate gasoline, the machines had a remarkable reduction in VOC emissions (Table 3). This behaviour is confirmed by the Mann-Whitney test of two populations – VOC values obtained using conventional gasoline and alkylate gasoline. This non-parametric test was applied because the measured values of VOC did not follow a normal distribution according to the Shapiro-Wilk test. The result of the Mann-Whitney test rejected the null hypothesis (p=0.010, a>0.05); therefore, the observed differences can be considered significant. Specifically, in all conditions investigated, when using the alkylate gasoline, the VOC levels were lower than VOC levels recorded using the 2% mix fuel, with a decrease ranging from 23 to over 77%. This is in accordance with the results obtained by other studies conducted on small two-stroke engines (Tsai et al. 2003, Alander et al. 2005, Neri et al. 2016). Moreover, the blower B1 showed the lowest values of VOC; in fact, this is the only model equipped with a four-stroke engine. On the contrary, the most polluting machine is the chain saw CS1, which produced a VOC amount exceeding the gas-analyser’s full scale (2000 mg/g) in all conditions.

Fig. 1 A) VOC emission measurement on a blower, placing the sensor probe as close as possible to the gas output; B) operator’s ear noise measurement; C) detail of the tri-axial accelerometer fixed on the chain saw driving handgrip

112

Croat. j. for. eng. 39(2018)1

Analysis of Hazardous Emissions of Hand-Operated Forestry Machines ... (109–116)

Table 3 Test results for gas emissions Fuel type Category

Machine ID

Engine speed

2 % Mix

Alkylate fuel

Mean difference %

VOC, mg/g CS1

CS2

CS3 Chain saw CS4

CS5

CS6

HC1 Hedge cutter HC2

Blower

idle

>2000*

1030

max

>2000*

1371

idle

1600

730

max

1350

450

idle

1831

750

max

1870

795

idle

470

314

max

800

650

idle

460

200

max

485

290

idle

1950

350

max

1200

650

idle

1000

540

max

680

550

idle

1116

185

max

>2000*

570

idle

120

max

120

idle

380

250

max

410

362

idle

205

max

200

–40*

–61

–58

–26

–48

–64

–33

–77*

–56

–23

–59

* Value exceeding the instrument’s full scale

Furthermore, the behaviour of the tested machines is not homogeneous: some models fuelled with 2% mix (CS2, HC1 and B3) emitted a higher quantity of VOC at the engine idling speed in respect to the working speed, while the rest showed the opposite results. Only the chain saw CS2 (the oldest Euro 1 machine) showed VOC levels higher at engine idling compared to the maximum speed, for both fuels. This behaviour is not unusual for this engine type, due to the less efficient removal of the unburned fuel from the cylinder, resulting from the slower piston movement at idle speed than working speed (Wòjcik and Croat. j. for. eng. 39(2018)1

A. Calcante et al.

Skarżyński 2006). This can cause high emission of VOC especially with 2% mix, which, having a low octane number, tends to burn worse than alkylate fuel. In any case, in most of considered forestry machines, using either 2% mix or alkylate fuel, the values of VOC measured at working conditions are quite similar to those measured at idle speed. It can be concluded that leaving the hand operated forestry machine engine working on idle running is undesirable for the health of operators and, if moving from one workplace to another is necessary, it is recommended to switch off the machine (Wòjcik and Skarżyński 2006). Concerning the chain saws, the decrease of VOC emissions is mainly due to the use of alkylate gasoline and to the improvement of the engines, passing from Euro 1 to Euro 2. In fact, the most modern machines (CS5 and CS6) produce low VOC levels both when idling and at the max engine speed. On the contrary, when the same machines were fuelled with 2% mix, this trend was not observed. Even in this case, the chain saw CS6 (Euro 2, manufactured in 2015) showed an emission level similar to CS2 (Euro 1), manufactured 10 years earlier. This confirms the evident improvement, in terms of VOC emissions, when passing from Euro 1 to Euro 2. The chain saw CS1, produced in 1994 (before the introduction of the emission regulations), dramatically reduced the VOC emissions passing from the traditional mix to the alkylate fuel, from over 2000 to around 1000 μg/g, in all the running conditions. As an old machine, designed to work with a traditional oilgasoline mixture and not subject to anti-pollution regulation, it can reduce its VOC emissions by using alkylate fuel.

3.2 Noise In general, the average noise level at the right ear is higher than that measured at the left ear (Table 4), because the tested machines were mainly operated by a right-handed worker. The average difference ranges from approximately 2 (for blowers) to 6 dB(A) (for hedge cutters). Furthermore, the use of alkylate fuel seems to reduce the noise levels in respect of the 2% mix. A statistical analysis was then conducted to ascertain if this difference is significant. The measured values of noise did not follow a normal distribution according to the Shapiro-Wilk test, so the Mann-Whitney test was applied to verify the differences between the two data populations. The result between medians did not reject the null hypothesis (p=0.957, a>0.05). As a consequence, the observed differences cannot be considered significant.

113

A. Calcante et al.

Analysis of Hazardous Emissions of Hand-Operated Forestry Machines ... (109–116)

In any case, except for those obtained at idle speed on most machines, all the levels greatly exceeded the limits provided by the national standards (i.e. for Italy Ministero del Lavoro D. L. 81/2008), and, therefore, the use of a hearing protector (i.e. earmuff or earplug) should be provided.

hedge trimmers, only right hand for blowers) were lower when using machines fuelled with alkylate gasoline (Table 4). The average decrease was of 11.4% (9.8% for the right hand, 13.6% for the left hand), with a maximum of 40% recorded for the B3 blower handgrip at maximum engine speed. Similarly to the noise, the distribution of the studied variables was not normal according to the ShapiroWilk test. Again, the application of the Mann-Whitney test to the HAV dataset showed that the observed differences were not statistically significant (p=0.222, a>0.05). Therefore, the two fuels did not significantly influence the HAV.

3.3 Hand-arm vibrations For chain saws and hedge blowers, HAV measured at the right (rear) handle exceeded those of the left (front) handle, according to Rottensteiner and Stampfer (2013). Similarly to noise, also in this case the average levels (right and left hand for chain saws and

Table 4 Test results for noise at the operator’s ears and HAV at the front (left hand) and rear (right hand) handgrips, except for blowers, which are operated only with the right hand. Each value is the average of 60 data recorded during each experimental test Hand-arm vibrations (HAV), m/s2

Noise, dB(A) Category

Machine ID

Engine speed

CS1 CS2 CS3 Chain saw

CS4 CS5 CS6

Hedge cutter

HC2

Blower

B2 B3

Alkylate fuel

2% Mix

Left ear

Right ear

Left ear

Right hand

Left hand

Right hand

Left hand

idle

88.7

87.0

86.8

86.5

4.3

2.9

4.9

3.5

max

107.5

108.3

107.1

107.0

3.9

3.8

5.2

4.0

idle

83.6

78.5

82.0

78.4

3.8

2.1

4.1

2.7

max

106.2

102.7

106.6

101.5

9.6

7.8

10.1

8.6

idle

82.6

77.1

83.2

79.9

3.2

2.8

3.3

3.2

max

107.2

104.7

106.1

103.8

1.5

3.5

1.7

3.9

idle

79.5

78.7

80.4

77.0

6.0

3.6

7.6

4.8

max

109.5

107.3

109.8

105.5

4.7

7.3

4.8

8.6

idle

82.9

80.6

85.5

84.0

2.6

2.0

3.3

3.0

max

104.8

103.3

104.6

103.0

3.7

3.6

4.0

3.7

idle

81.8

79.2

80.2

74.9

4.8

3.6

4.9

3.7

max

103.8

102.7

103.0

98.8

6.1

3.2

6.7

4.1

85.4*

82.3*

85.4*

81.6*

4.5**

3.9**

5.1**

4.5**

idle

76.6

71.3

78.5

72.9

2.2

1.2

2.3

1.6

max

99.3

90.7

100.3

90.9

10.0

7.8

10.3

8.2

idle

85.1

78.9

85.0

79.0

4.7

3.0

5.2

3.7

max

98.1

95.0

99.5

96.6

8.0

6.2

8.9

7.2

82.0*

76.6*

83.6*

77.9*

6.2**

4.6**

6.7**

5.2**

Average B1

2% Mix

Right ear

Average HC1

Alkylate fuel

idle

73.8

73.3

76.5

76.6

1.6

–

1.8

–

max

97.6

98.1

96.8

97.6

1.3

–

1.6

–

idle

76.8

72.2

76.6

73.4

2.3

–

2.5

–

max

98.4

98.0

98.1

97.0

2.1

–

2.3

–

min

71.8

70.5

72.3

71.3

2.3

–

2.0

–

max

84.9

83.2

85.0

82.7

1.5

–

2.5

–

76.6*

74.7*

77.5*

76.1*

1.9**

–

2.1**

–

Average * Logarithmic average; ** Arithmetic average

114

Croat. j. for. eng. 39(2018)1

Analysis of Hazardous Emissions of Hand-Operated Forestry Machines ... (109–116)

If the obtained results are theoretically compared to the 8-hour limits provided by the 44/2002/EC Directive (2.5 m/s² action value, 5.0 m/s² limit value), a different behaviour for the eleven machines could be observed. For many chain saws (in the working condition) and hedge cutters (at max engine speed), the levels often exceed 9–10 m/s². Thus, to respect the limit value of 5 m/s², a reduction in the working time must be considered. On the contrary, the blowers showed HAV levels always below the action values of 2.5 m/s² (mainly due to the absence of rotating working parts), thus also respecting the lowest value provided by the Directive.

4. Conclusions In the last two decades, the standards finalised to reduce the exhaust gas emissions of on-road and offroad vehicles have dramatically affected the technological progress of internal combustion engines (Punov et al. 2017), including those fitted on forestry and agricultural machinery. Manufacturers have made a great effort to ensure the compliance of their production with the rigorous requirements provided. Although with a certain delay in respect to medium-high power engines, those fitted on small hand-operated machines used in the agriculture and forestry sectors must also now respect the limits provided by the relevant standards. The present survey confirms that the reduction in the amount of emissions can be remarkably improved not only with the technological progress of the engine design, but above all by adopting advanced fuels that lead to a more efficient combustion process. In fact, the experimental results demonstrated that alkylate fuel is able to reduce VOC pollution more than 77% compared to the use of the traditional 2% mix. On the other hand, the use of alkylate gasoline has no effect on noise and HAV, even though this fuel has a higher octane number. In any case, a generalised evolution from twostroke to four-stroke engines could lead to a further, considerable reduction in the pollutant gas emissions (Jüttner et al. 1995, Tsai et al. 2000, Tsai et al. 2003). Along these lines, the performance of hand-operated machines will also be improved, in order also to assure better working conditions for the operators.

5. References Alander, T., Antikainene, E., Raunemaa, T., Elonen, E., Rautiola, A., Torkkell, K., 2005: Particle emissions from a small two-stroke engine: effects of fuel, lubricating oil, and exhaust Croat. j. for. eng. 39(2018)1

A. Calcante et al.

aftertreatment on particle characteristics. Aerosol Science and Technology 39(2): 151–161. Bünger, J., Bombosch, F., Mesecke, U., Hallier, E. 1997: Monitoring and analysis of occupational exposure to chain saw exhausts. American Industrial Hygiene Association Journal 58(10): 747–751. Comagarden/Morgan, 2015: Risultati rilevazione trimestrale, andamento del mercato – 4° trimestre. Federunacoma, Roma, Italy. Davis, G., 1978: Noise and vibration hazards in chainsaw operations: A Review. Australian Forestry 41(3): 153–159. EPA, 2005: Average annual emissions tables: 1970–2002 – All criteria pollutants. Available at: <http://www.epa.gov/ttn/ chief/trends/index.html> [Accessed May 2015.] European Committee for Standardization, 2013: European Standard EN 228 – Automotive fuels – Unleaded petrol – Requirements and test methods. Brussels, Belgium. European Commission, 2002: 44/2002/EC Directive – Vibrations. OJL 177. Brussels, Belgium. Fonseca, A., Aghazadeh, F., de Hoop, C., Ikuma, L., Al-Qaisi, S., 2015: Effect of noise emitted by forestry equipment on workers’ hearing capacity. International Journal of Industrial Ergonomics 46: 105–112. Groves, K.W., Lyons, R.G., 2013: Occupational hazards for chainsaw operators. Journal of Australian Forestry 32(4): 205–210. ISO, 2011: Forestry and gardening machinery – Vibration test code for portable hand-held machines with internal combustion engine – Vibration at the handles. Norm ISO 22867:2011. International Organization for Standardization Publ., Geneva, Switzerland. ISO, 2011: Forestry and gardening machinery – Noise test code for portable hand-held machines with internal combustion engine – Engineering method. Norm ISO 22868:2011. International Organization for Standardization Publ., Geneva, Switzerland. Jüttner, F., Backhaus, D., Matthias, U., Essers, U., Greiner, R., Mahr, B., 1995: Emissions of two-stroke outboard engines – I. Quantification of gases and VOC. Water Research 29(8): 1976–1982. Magnusson, R., Nilsson, C., Andersson, K., Andersson, B., Rannug, U., Ostman, C., 2000: Effect of gasoline and lubricant on emissions and mutagenicity of particles and semivolatiles in chain saw exhaust. Environmental Science & Technology 34(14): 2918–2924. Ministero del Lavoro, 2008: D. L. n. 81 in materia di Tutela della salute e della sicurezza nei luoghi di lavoro. Gazzetta Ufficiale 101. Monarca, D., Cecchini, M., Vassalini, G., 2003: Vibrazioni trasmesse al Sistema mano-braccio dai principali modelli di motoseghe disponibili sul mercato italiano. Rivista di Ingegneria Agraria 1: 53–64.

115

A. Calcante et al.

Analysis of Hazardous Emissions of Hand-Operated Forestry Machines ... (109–116)

Neri, F., Foderi, C., Laschi, A., Fabiano, M., Cambi, F., Sciarra, G., Aprea, M.C., Cenni, A., Marchi, E., 2016: Determining exhaust fumes exposure in chainsaw operations. Environmental Pollution 218: 1162–1169. Nilsson, C.A., Lindahl R., Norström A., 2010: Occupational exposure to chain saw exhausts in logging operations. American Industrial Hygiene Association Journal 748(2): 99–105. Ostermark, U., 1996: Better use of cleaner petrol. Journal of Cleaner Production 4(2): 105–110. Ostermark, U., Petersson, G., 1993: Volatile hydrocarbons in exhaust from alkylate-based petrol. Chemosphere 27(9): 1719–1728. Punov, P., Evtimov, T., Chiriac, R., Clenci, A., Danel, Q., Descombes, G., 2017: Progress in high performance, low emissions, and exergy recovery in internal combustion engines. International Journal of Energy Research 41(9): 1229–1241. Rottensteiner, C., Stampfer, K., 2013: Evaluation of operator vibration exposure to chainsaws equipped with a Kesper safety bar. Scandinavian Journal of Forest Research 28(2): 193–200. Skoupi, A., Klvac, R., Hosseini, S., 2010: Changes in the external speed characteristics of chainsaw engines with the use of mineral and vegetable oils. Croatian Journal of Forestry Engineering 31(2): 149–155. Stratiev, D., 2011: Effect of gasoline hydrocarbon composition on the properties of gasoline/bioethanol blend. Oxidation Communications 34(1): 183–192.

Swedish Standard Institute, 2008: Motor fuels - Swedish standard SS 155461:2008 – Special gasoline for powered implements. Stockholm, Sweden. Tsai, J.H., Hsu, Y.C., Weng, H.C., Lin, W.Y., Jeng, F.T., 2000: Air pollutant emission factors from new and in-use motorcycles. Atmospheric Environment 34(28): 4747–4754. Tsai, J.H., Chiang, H.L., Hsu, Y.C., Weng, H.C., Yang, C.Y., 2003: The speciation of volatile compounds (VOCs) from motorcycle engine exhaust at different driving modes. Atmospheric Environment 37(18): 2485–2496. Tunay, M., Melemez, K., 2008: Noise induced hearing loss of forest workers in Turkey. Pakistan Journal of Biological Sciences 11(17): 2144–2148. UNI, 2004: EN ISO 5349-1:2004 – Mechanical vibration – Measurement and evaluation of human exposure to hand – transmitted vibration – Part 1: General requirements. UNI Ente Italiano di Normazione, Milano, Italy. Volckens, J., Braddock, J., Snow, R.F., Crews, W., 2007: Emissions profile from new and in-use handheld, 2-stroke engines. Atmospheric Environment 41(3): 640–649. Wòjcik, K., 2002: The analysis of basic parameter of chain saw on Euromarket. Przegląd Techniki Rolniczej i Leśnej 10: 18–20. Wòjcik, K., Skarżyński, J.G., 2006: Emission and composition of exhaust gases by new chain saws produced by Husqvarna and Stihl. Acta Scientiarum Polonorum Silvarum Colendarum Ratio et Industria Lignaria 5(2): 147–157.

Authors’ addresses:

Received: August 16, 2016 Accepted: July 12, 2017

116

Assist. prof. Aldo Calcante, PhD. e-mail: aldo.calcante@unimi.it Assist. prof. Davide Facchinetti, PhD. e-mail: davide.facchinetti@unimi.it Prof. Domenico Pessina* e-mail: domenico.pessina@unimi.it University of Milan Department of Agricultural and Environmental Sciences – Production, Landscape, Agroenergy Via Giovanni Celoria 2 20122 Milano ITALY * Corresponding author Croat. j. for. eng. 39(2018)1

Original scientific paper

Accuracy Assessment of Digital Terrain Models of Lowland Pedunculate Oak Forests Derived from Airborne Laser Scanning and Photogrammetry Ivan Balenović, Mateo Gašparović, Anita Simic Milas, Alen Berta, Ante Seletković Abstract Digital terrain models (DTMs) present important data source for different applications in environmental disciplines including forestry. At regional level, DTMs are commonly created using airborne digital photogrammetry or airborne laser scanning (ALS) technology. This study aims to evaluate the vertical accuracy of DTMs of different spatial resolutions derived from high-density ALS data and existing photogrammetric (PHM) data in the dense lowland even-aged pedunculate oak forests located in the Pokupsko basin in Central Croatia. As expected, the assessment of DTMs’ vertical accuracy using 22 ground checkpoints shows higher accuracy for ALS-derived than for PHM-derived DTMs. Concerning the different resolutions of ALS-derived (0.5 m, 1 m, 2 m, 5 m) and PHM-derived DTMs (0.5 m, 1 m, 2 m, 5 m, 8 m) compared in this research, the ALS-derived DTM with the finest resolution of 0.5 m shows the highest accuracy. The root mean square error (RMSE) and mean error (ME) values for ALS-derived DTMs range from 0.14 m to 0.15 m and from 0.09 to 0.12 m, respectively, and the values decrease with decreasing spatial resolution. For the PHM-derived DTMs, the RMSE and ME values are almost identical regardless of resolution and they are 0.35 m and 0.17 m, respectively. The findings suggest that the 8 m spatial resolution is optimal for a given photogrammetric data, and no finer than 8 m spatial resolution is required. This research also reveals that the national digital photogrammetric data in the study area contain certain errors (outliers) specific to the terrain type, which could considerably affect the DTM accuracy. Thus, preliminary evaluation of photogrammetric data should be done to eliminate possible outliers prior to the DTM generation in lowland forests with flat terrain. In the absence of ALS data, the finding in this research could be of interests to countries, which still rely on similar photogrammetric data for DTM generation. Keywords: DTM, ALS, LiDAR, stereo-photogrammetry, aerial images, even-aged forest stands, Central Croatia

1. Introduction Digital terrain models (DTMs) provide three-dimensional information of the Earth’s bare surface excluding vegetation and man-made features (Li et al. 2005). DTMs present the important data source for different applications in environmental disciplines such as hydrology, geology, agronomy, and forestry. In forestry, DTMs can be used to support forestry operations (Stereńczak and Moskalik 2014), disaster risk Croat. j. for. eng. 39(2018)1

analysis (Ristić et al. 2017), assessment of forest structure variables (Rahlf et al. 2015, Puliti et al. 2016, Balenović et al. 2017), etc. In particular, DTMs are used in combination with the digital surface models (DSMs) to obtain canopy height models (CHMs) or normalized point clouds, which are further used to estimate various forest variables at tree level (Rahlf et al. 2015) or area level (Puliti et al. 2016, Balenović et al. 2017). A classical field survey by means of Global Navigation Satellite System (GNSS) receivers or total stations

117

I. Balenović et al.

Accuracy Assessment of Digital Terrain Models of Lowland Pedunculate Oak Forests ... (117–128)

presents the most accurate method for acquiring data for DTM generation (Höhle and Potuckova 2011). However, for large areas, field surveys are expensive and time-consuming and, therefore, less efficient than the remote sensing methods. At regional level, DTMs are nowadays commonly created using airborne digital photogrammetry or airborne laser scanning (ALS) technology (Höhle and Höhle 2009). Currently, the most efficient remote sensing technology, in terms of accuracy, is ALS based on light detection and ranging (LiDAR) (Stereńczak et al. 2011). Namely, both technologies perform well in open areas, but as the laser beam can penetrate through the forest canopy and reach the ground, ALS is especially effective in forested and vegetated areas. National ALS campaigns have been performed or are still being performed in more and more countries worldwide, where ALS data now present the main source for DTM generation. However, a number of countries in Europe (e.g. Croatia, Greece, Hungary, Slovakia, etc.) and worldwide do not have a good ALS data coverage, and photogrammetrically derived data present the national standard for DTM generation. The accuracy of DTMs, especially of ALS-derived DTMs, over different forest areas has been the focus of several studies. For example, Reutebach et al. (2003) investigated the accuracy of ALS-derived DTM of 1.52 m resolution over different stand densities of mountainous pine forests in western Washington, USA. The root mean square error (RMSE), mean error (ME) and standard deviation (SD) between DTM and ground checkpoints was 0.32 m, 0.22 m and 0.24 m, respectively. They found no significant differences in their results for different stand densities. Somewhat lower accuracy (RMSE≈0.55 m), but similar overestimation pattern (ME=0.20 m) for ALS-derived DTM, was obtained by Su and Bark (2006), whose research was conducted in aspen forests of different ages in Alberta, Canada. More recent research by Stereńczak et al. (2013) was focused on assessing the accuracy of ALS-derived DTMs of different spatial resolutions (0.5 m, 1 m, 2 m, 3 m, 4 m, 5 m, 10 m) in pine forest stands at flat and steep terrain in western Poland. The RMSEs ranged between 0.09 m and 0.18 m for the flat terrain and 0.39 m and 1.34 m for steep terrain, whereas the MEs ranged between –0.12 m and 0.09 m for flat terrain, and –0.26 m and –1.16 m for steep terrain. Stereńczak et al. (2013) concluded that DTM had higher accuracy on flat terrains and that the DTMs’ accuracy in general increases with the increase of spatial resolution. Furthermore, Stereńczak et al. (2016) evaluated the influence of various factors (forest structure, slope, off-nadir angle, understory vegetation and filtration and interpolation algorithm) on ALS-derived

118

DTM accuracy in mixed mountainous forests in southwestern Poland. The RMSE and ME ranged from 0.19 m to 0.23 m, and from 0.13 m to 0.16 m, respectively. They identified the slope and particularly undergrowth vegetation as the most important factors influencing DTM accuracy. A direct relationship between the slope increments and DTM error has been recently confirmed by Aryal et al. (2017) within the study conducted in heterogeneous forest sites of Bavarian Forest National Park (southeastern Germany). Moreover, they obtained higher DTM errors for forest stands dominated by deciduous trees rather than for coniferous, mixed, and deadwood-dominated stands. From the above-mentioned studies, it is evident that DTM accuracy is affected by various factors, such as ALS data characteristics, forest and terrain characteristics. Thus, further research, which will include and encompass different ALS, forest and terrain characteristics, is needed. Contrary to ALS-derived DTM, the accuracy of photogrammetrically derived DTM has not attracted a lot of attention, especially within recent research. One of the rare exceptions is the research of Gil et al. (2013), which compared ALS and photogrammetrically derived DTMs in pine forests of Tenerife Island (Spain). The aim of this study is to evaluate the vertical accuracy of DTMs of different spatial resolutions (0.5 m, 1 m, 2 m, 5 m, 8 m) derived from high-density ALS point clouds and existing photogrammetric data in dense lowland even-aged pedunculate oak (Quercus robur L.) forests. To the best of authors’ knowledge, the DTM’s accuracy in lowland pedunculated oak forests has not been the subject of similar studies. The results of this research could be of great interest to countries that still do not have ALS data, but have similar photogrammetric data for DTM generation.

2. Materials and methods 2.1 Study area The study area, a part of the management unit Jastrebarski lugovi (991.50 ha), covers a portion of the forested area in the Pokupsko basin, located in Central Croatia, approximately 35 km southwest of Zagreb (Fig. 1). The management unit Jastrebarski lugovi is mainly covered with dense lowland even-aged pedunculate oak (Quercus robur L.) forests aged between 0 to 160 years, and even-aged narrow-leaved ash (Fraxinus angustifolia Vahl.) forests aged between 0 to 80 years. Oak and ash even-aged forest stands cover ≈77% and ≈20% of the total forest area, respectively. The oak stands are commonly mixed with other tree species, such as Carpinus betulus L., Alnus glutinosa (L.) Geartn., Croat. j. for. eng. 39(2018)1

Accuracy Assessment of Digital Terrain Models of Lowland Pedunculate Oak Forests ... (117–128)

and Fraxinus angustifolia Vahl., while the ash stands are more homogenous and rarely mixed with tree species, such as Alnus glutinosa (L.) Geartn., and Quercus robur L. Two understory species, Corylus avellana L. and Crataegus monogyna Jacq., are common in the entire area. The forest stands are actively managed for sustained timber based on the 140-year (oak stands) and 80-year (ash stands) rotation cycles, having two or three regeneration fellings during the last 10 years of the rotation. The study area is mostly flat, with ground elevations ranging from 105 to 118 m a.s.l. Soils are hydromorphic on clay parent material (Mayer 1996), and according to the World Reference Base for Soil Resources (WRB 2006), they are classified as luvic stagnosol. According to Köppen classification, the climate of the area is warm temperate with the mean annual temperature of 10.6°C and precipitation of 962 mm·y−1 (data from national Meteorological and Hydrological Service for the nearest meteorological station for the period 1981–2010).

2.2 Field reference data The ground checkpoints recorded in the field by geodetic methods (e.g. total station, GNSS receivers) are

Fig. 1 Location of the study area (Jastrebarski lugovi) situated within the Pokupsko basin in Croatia with reference data (GNSS measured locations – white dots) for DTMs accuracy assessment Croat. j. for. eng. 39(2018)1

I. Balenović et al.

the most common reference data used for the accuracy assessment of DTM (Höhle and Höhle 2009). In order to obtain reliable accuracy measures, the ground checkpoints should have at least three times higher accuracy than DTM (Maune 2007). Besides high accuracy, the number of checkpoints should be sufficiently large, and they should be randomly distributed over the target area (Höhle and Höhle 2009). According to the American Society for Photogrammetry and Remote Sensing, for testing the vertical accuracy of DTM a minimum of 20 checkpoints should be collected in each of the major land cover categories of the area (ASPRS 2004). Prior to the field work, the selection of potential ground checkpoints and pre-definition of their locations were done in the GIS environment (Global Mapper version 19 software; Blue Marble Geographics, Hallowell, Maine, USA). For that purpose, a regular grid with 100 m grid size was overlaid over the selected part of the study area. Four adjacent points were grouped in a quadrat and then numbered in the counter clockwise direction. From each quadrat, one point with the lowest numeration was selected as a ground checkpoint, given that the predefined condition of the minimum distance between the point and the forest edge and/or forest sub-compartment border was satisfied. The minimum distance was set to be 20 m, 30 m, and 50 m in the 21–60 year-old, 61–100 year-old, and 101–160 year-old forest stands, respectively. Namely, this approach was used because the recorded points would be used as the centers of the sample plots in the consequent research. In total, 114 points were selected as potential ground checkpoints. The positioning (x, y, z) coordinates of the selected points were recorded using the GNSS receiver Stonex S9IIIN connected with the Croatian Positioning System (CROPOS), a network of reference stations which transmit corrections in real time directly to the GNSS receiver by mobile Internet. In this way, it is possible to obtain both horizontal and vertical positional accuracy from 2 to 5 cm (CROPOS – Users’ Manual). To further increase the accuracy of the GNSS measurements, the data was collected with antenna receiver of 4 m in height during leaf-off conditions between 8 and 15 March 2017. The maximum time for measuring the positions was 30 minutes per location. During this time, an attempt was made to record the positions in FIXED receiver mode, which provided more accurate and more reliable results than the FLOAT mode (Brach and Zasada 2014). However, if the measurement with the FIXED mode were not realized during the 30 minutes, the point would be recorded with the FLOAT mode. In order to obtain highly accurate and reliable reference data for the vertical accuracy assessment of

119

I. Balenović et al.

Accuracy Assessment of Digital Terrain Models of Lowland Pedunculate Oak Forests ... (117–128)

DTMs, only points recorded with FIXED receiver mode, and with vertical precision (standard deviation calculated by the receiver) ≤10 cm, were selected as ground checkpoints. Out of 114 recorded points, 22 points met these criteria. Due to limitations caused by dense forest conditions, the number of the selected ground checkpoints was not excessive, but it was relatively large under given conditions and also randomly distributed over the study area (Fig. 1).

2.3 Digital terrain models (DTMs) data and processing 2.3.1 ALS data and ALS-derived DTM The ALS data were provided by the company Hrvatske vode, Zagreb, Croatia, a legal entity for water management, whereas the acquisition and processing were conducted by the Zavod za fotogrametriju d.d. and Mensuras d.d. under leaf-on conditions in several surveys between 29 June and 25 August 2016. The data were collected with an Optech ALTM Gemini 167 laser scanner mounted on the Pilatus P6 aircraft. The average ﬂying height was 720 m above ground level with an average flying speed of 51 m∙s-1. The laser pulse repetition frequency was 125 Hz, and field of view was ±25°. Resulting point densities for the study area, considering all returns and »last only«, were 13.64 points∙m-2 and 9.71 points∙m-2, respectively. According to the data provider, the horizontal accuracy of recorded points was 15 cm, and vertical accuracy was 8.3 cm. The stated accuracies were based on a much larger area than the one considered in this study, where forested and non-forested areas were included. The point data were classified into ASPRS Standard LiDAR Point Classes (ASPRS 2008) using TerraSolid (version 11) software (Terrasolid Ltd. 2012). In total, approximately 7% of all returns over the study area were classified as »ground«, resulting with an average ground point density of 0.91 points∙m-2. Classification of the ground was based on the progressive triangulated irregular network (TIN) densification algorithm developed by Axelsson (2000). This procedure for LiDAR data classification for the purposes of defining ground data was used in several studies (Gobakken et al. 2014, Guan et al. 2014, SánchezLopera and Lerma 2014, Torresan et al. 2014). From the classified ground returns, regular grids of various spacing (0.5 m, 1 m, 2 m, and 5 m) were generated and provided to us in .las format. A raster DTMs with 0.5 m (ALS-DTM0.5), 1 m (ALSDTM1), 2 m (ALS-DTM2) and 5 m (ALS-DTM5) resolution (pixel size) were generated from regular grids using the triangulated irregular network (TIN) and linear interpolation techniques of Global Mapper software.

120

2.3.2 Photogrammetric data and PHM-derived DTM In this research, the digital terrain data used to generate photogrammetric DTMs (PHM-derived DTMs) were provided by the Croatian State Geodetic Administration (CSGA). These data present the Croatian national standard and, currently, they are the only available DTM data for most of Croatia, with the exception of several smaller areas surveyed with ALS, as shown in this study. According to the rules of the CSGA (Product Specifications 301D150), the digital terrain data consisted of several three-dimensional vector data such as breaklines, formlines, spot heights and mass points. Breaklines are line strings (features) that describe the change of a slope declination, mostly ruptures as e.g. mountain ridges, incisions, dams, shores, and they are recognized as terrain fracture. Formlines are line strings that describe smooth terrain physical properties (e.g. the deepest line in a shallow ditch, the highest lines along the mountain ridge). Spot height presents the highest or lowest point of the characteristic landscape (e.g. peaks or sinks). Mass points are height points measured in irregular distribution that fill the area surrounded by breaklines and formlines. The digital terrain data were collected by aerial stereo photogrammetry using digital aerial images of GSD≤30 cm (GSD – ground sample distance), supported with vectorization of existing maps and field data collection (especially for areas which were not visible from aerial images). Data density was dependent on terrain type, slope and surface roughness. The average distance between points in breaklines and formlines was recommended to be 25 meters. The average distance between mass points was considered to be 70–90 meters for open areas, whereas for flat areas or areas covered by vegetation, the distance was not more than 120 meters. For the flat terrain, which corresponds to the terrain type of Pokupsko basin, the required number of points in breaklines and formlines was 400–800 points∙km-2, while the required number of mass points and spot heights was 100–150 points∙km-2. The required absolute accuracy of digital terrain data (including both horizontal and vertical accuracy) validated with ground control points was <±1 m of the standard deviation for the well-defined details and <±2 m for not well-defined details. For this study area, the average number of points in breaklines and formlines was 455 points∙km-2, while the average number of mass points and spot heights was 141 points∙km-2, which corresponds to the recommendations of Product Specification 301D150 of CSGA. Croat. j. for. eng. 39(2018)1

Accuracy Assessment of Digital Terrain Models of Lowland Pedunculate Oak Forests ... (117–128)

I. Balenović et al.

Table 1 Results of Shapiro-Wilk test for normality of errors distribution Parameter

Digital terrain models ALS-DTM0.5

ALS-DTM1

ALS-DTM2

ALS-DTM5

PHM-DTM0.5

PHM-DTM1

PHM-DTM2

PHM-DTM5

PHM-DTM8

0.92

0.97

0.98

0.97

0.95

0.11

0.77

0.97

0.80

0.28

0.29

0.30

Similar to ALS-DTMs generation, a photogrammetric raster DTMs with 0.5 m (PHM-DTM0.5), 1 m (PHMDTM1), 2 m (PHM-DTM2) and 5 m (PHM-DTM5) reso-

lution were generated from the national digital terrain data using the triangulated irregular network (TIN) and linear interpolation techniques. Additionally, a raster

Fig. 2 Normal Q-Q plots Croat. j. for. eng. 39(2018)1

121

I. Balenović et al.

Accuracy Assessment of Digital Terrain Models of Lowland Pedunculate Oak Forests ... (117–128)

DTM with 8 m resolution (PHM-DTM8) was generated based on optimal grid spacing (for used data) automatically determined by the Global Mapper software.

2.4 Accuracy assessment The vertical accuracy assessments of the ALS-derived and PHM-derived DTMs of different resolutions were conducted by comparing elevations of the ground checkpoints and elevations of the planimetrically corresponding points extracted from raster DTMs. Since the different accuracy measures should be applied to DTMs with normal and those with nonnormal error distribution (Höhle and Höhle 2009), the normality of the error distribution was tested for each DTM. Vertical errors between DTMs and checkpoints elevations were calculated, and the normality of errors distribution was analyzed using: Þ Shapiro-Wilk test (Shapiro and Wilk 1965) (Table 1) Þ normal Q-Q plots (Fig. 2). All the statistical analyses were performed using the STATISTICA software (version 11; Hill and Levicki 2007). Both tests, Shapiro-Wilk and normal Q-Q plots, revealed that vertical errors were normally distributed for each resolution of both ALS-derived and PHMderived DTMs. Namely, p values of Shapiro-Wilk test were greater than 0.05 (Table 1) for each DTM, whereas only slight deviations of points from straight lines were observed in each Q-Q plot (Fig. 2). Therefore, as suggested by Höhle and Höhle (2009), the following measures were used for DTMs accuracy assessment with reference data (ground checkpoints): vertical error at point i (∆zi), root mean square error (RMSE), mean error (ME), and standard deviation error (SD):

= Dzi zDTMI − zGCPi

SD =

1 n

RMSE =

ME = 1 n−1

1 n

∑Dz i=1

∑ (D z i=1

2 i

(1) (2)

(3)

− ME)2 (4)

Where: zDTMi the elevation of DTM calculated from point i, zGCPi the elevation of ground checkpoints measured in the field on point i, and n is the number of points.

122

To further evaluate the accuracy of PHM-derived DTM, the comparison between ALS- and PHM-derived DTMs of the highest accuracy (according to the evaluation with ground checkpoints) was conducted using the difference raster model (Gil et al. 2013). The difference raster model was generated for the entire study area by subtracting ALS- from PHM-derived DTM. In this case, the elevations from ALS-derived DTM were considered as reference data and the abovementioned accuracy measures (∆zi, RMSE, ME, SD) were calculated.

3. Results and discussion As expected, the assessment of DTMs vertical accuracy in dense lowland pedunculated oak forests, conducted using 22 ground checkpoints, shows higher accuracy for ALS-derived DTMs than for PHMderived DTMs (Fig. 3). The RMSE values for ALS-derived DTMs range from 0.14 m to 0.15 m, while for PHM-derived DTMs they amount to 0.35 m for all resolutions. This suggests that the ALS data provide more accurate DTM values and are more suitable remote sensing data for dense forested areas (Höhle and Potuckova 2011, White et al. 2013). Furthermore, the ME values are positive for both ALS- and PHM-derived DTMs indicating that elevations extracted from DTMs on average overestimate elevations of the ground checkpoints. The overestimations for ALS-derived DTMs range from 0.09 m to 0.12 m, whereas PHMderived DTMs produce slightly greater overestimations with the value of 0.17 m for all resolutions. This is in agreement with other similar studies (Reutebuch et al. 2003, Su and Bork 2006, Gil et al. 2013, Stereńczak et al. 2016), which reported that due to near ground vegetation (e.g. grass, litter, woody debris) in the forest, ALS-derived DTMs had a tendency to overestimate the real reference terrain elevations. Among all evaluated DTMs, the ALS-derived DTM (ALS-DTM0.5) with the highest resolution (0.5 m) shows the highest accuracy (RMSE=0.14 m, ME=0.09 m, SD=0.10 m). All other ALS-derived DTMs with resolutions of 1 m, 2 m and 5 m produce just slightly larger errors (Fig. 3). Although the differences between the errors are small, it is evident that the accuracy of DTM slightly decreases with the decrease of resolution, which is in accordance with findings of Stereńczak et al. (2011, 2013). Differences between the results for all PHM-derived DTM resolutions are almost negligible (Fig. 3). The results (errors) differ in the third decimal place only. However, among all PHM-derived DTMs, the DTM with the coarser spatial resolution of 8 m Croat. j. for. eng. 39(2018)1

Accuracy Assessment of Digital Terrain Models of Lowland Pedunculate Oak Forests ... (117–128)

Fig. 3 Vertical accuracy (RMSE, ME, SD) of ALS- and PHM-derived DTMs of different resolutions assessed with ground checkpoints Croat. j. for. eng. 39(2018)1

I. Balenović et al.

(PHM-DTM 8) exhibits the highest accuracy (RMSE=0.35 m, ME=0.17 m, SD=0.31 m) This suggests that this spatial resolution can be considered as optimal for given source data and that there is no need for finer spatial resolution for a given photogrammetric data. Moreover, keeping the coarser spatial resolution especially for larger areas, the processing of raster data is faster. The vertical errors (∆zi; differences between DTM and checkpoint elevations) of the most accurate PHMand ALS-derived DTMs at each ground checkpoint are shown in Fig. 4. At each checkpoint, the errors of PHM-DTM8 are considerably larger ranging from –0.43 m to 0.69 m than errors of ALS-DTM0.5, which range from –0.06 m to 0.25 m. Furthermore, it can be seen that elevation values of PHM-DTM8 vary noticeably from the underestimation to the overestimation, whereas ALS-DTM0.5 elevation values mostly overestimate checkpoints elevations. Only elevations of 5 checkpoints are slightly underestimated by ALSDTM0.5. The common overestimation of the ALS-derived DTMs in forest areas caused by near ground vegetation has been already discussed above. Another possible reason for the overestimation may arise from field measurements of ground checkpoints. Namely, due to weather and terrain conditions (e.g. moist soil) during the field measurements, some sinking of the posts used to hold the GNSS antenna receiver is possible, causing changes in the level and inclination of posts and ultimately the overestimation of ALS-derived values. Finally, as already mentioned, the vertical precision (SD) of the selected ground checkpoints ≤10 cm, may also slightly influence the differences. The obtained accuracies for ALS-derived DTMs in this study are in agreement with the other studies (Kraus and Pfeifer 1998, Reutebuch et al. 2003, Su and Bork 2006, Spaete et al. 2011, Stereńczak et al. 2011, 2013, 2016, Gil et al. 2013), although the direct comparison of the results is not fully justified due to a number of differences between studies regarding the used ALS data (e.g. point density, scanning angle, acquisition period), forest/vegetation type and structure (e.g. species composition, canopy and stem density, presence and density of low vegetation) and terrain characteristics (e.g. slope). For further comparison between the PHM- and ALS-derived DTMs, a difference raster model of 0.5 m resolution was generated by subtracting ALS-DTM0.5 from PHM-DTM8 (Difference Model A, Fig. 5a) to detect the areas of greater elevation differences between the PHM-DTM8 and ALS-DTM0.5 (Gil et al. 2013). Further analysis revealed that the discrepancies were caused by photogrammetric data such as mass or

123

I. Balenović et al.

Accuracy Assessment of Digital Terrain Models of Lowland Pedunculate Oak Forests ... (117–128)

Fig. 4 Vertical errors (D zi; differences between DTM and checkpoint elevations) of ALS-DTM0.5 and PHM-DTM8 for each of 22 ground checkpoints height points used for PHM-DTM8 generation, which elevations considerably differed from elevations of all surrounding points (Fig. 5c). Therefore, all points for which elevation differed for more than 1 m from elevations of four surrounding points were removed, and new, improved PHM-DTM8, as well as new Difference Model B, were generated (Fig. 5b, d). Out of 21 removed points (outliers), only one has greater negative elevation values compared to the elevations of surrounding points. The improvements of PHM-DTM8 by eliminating 21 points can be observed even visually in Difference Model B (Fig. 5b, d). To validate the outliers, additional field observations were conducted, and the absence of the outlier was confirmed. The removed points present gross errors (outliers) that most likely occurred during photogrammetric measurements (restitution). Based on the authors’ field experience, the threshold of 1 m for removing outliers is applicable in this study. Descriptive statistics calculated for the entire area (54,687,600 pixels) for both difference models are presented in Table 3. Compared to Difference Model A, both the RMSE and SD values for Difference Model B are decreased by 0.02 m. On the other hand, ME is increased by 0.01 m because of 20 removed points (outliers) with positive values and only one point with a negative value. Opposite to the validation with ground checkpoints, where PHM-DTM8 produce larger ME value than ALS-DTM0.5 indicating that PHM elevations on average overestimate ALS elevations by

124

0.06 m, the comparison based on the Difference Model B for the entire research area shows that PHM elevations underestimate ALS elevations by –0.15 m on average. It should be noted that the ground checkpoints are placed in the north part of the research area (Fig. 1) where, according to Difference Model B (Fig. 5b), PHM values mostly overestimate ALS values, while PHM values mostly underestimate ALS values in the remaining area. Similarly, Gil et al. (2013) reported a tendency of PHM-derived DTM to underestimate the terrain elevations compared to ALS-derived DTM over the different conditions (e.g. flat and open terrain, hilly area covered by dense pine forest). Due to a large area (a large number of pixels used in calculations), the improvement of the PHM-DTM8 by eliminating outliers seems to be negligible in terms of RMSE and SD, but Figs. 5c, 5d, and 5e, show that the improvement is considerable. For the exemplary area, vertical differences between ALS-DTM0.5 and PHMDTM8 (i.e. vertical error of PHM-DTM8) decrease up to 4 m (Fig. 5e).

4. Conclusions This research first provided comparative accuracy assessment of ALS- and PHM-derived DTMs in dense lowland even-aged pedunculated oak forests. In accordance with recent studies, the results of this research confirmed that ALS was highly suitable remote sensing technology for accurate terrain modelling in dense forested areas. More precisely, the comparative Croat. j. for. eng. 39(2018)1

Accuracy Assessment of Digital Terrain Models of Lowland Pedunculate Oak Forests ... (117â&#x20AC;&#x201C;128)

I. BalenoviÄ&#x2021; et al.

Fig. 5 (a) Difference Model A of 0.5 m resolution generated by subtracting ALS-DTM0.5 from PHM-DTM8; (b) Difference Model B of 0.5 m resolution generated by subtracting ALS-DTM0.5 from improved PHM-DTM8; (c) the exemplary area of Difference Model A; (d) the exemplary area of Difference Model B; (e) vertical profile of Differences models A and B across the exemplary area Croat. j. for. eng. 39(2018)1

125

I. Balenović et al.

Accuracy Assessment of Digital Terrain Models of Lowland Pedunculate Oak Forests ... (117–128)

Table 2 Descriptive statistics for Difference Model A and B Difference Model

RMSE, m

ME, m

SD, m

Max., m

Min., m

0.51

–0.14

0.49

4.25

–3.68

0.49

–0.15

0.47

3.10

–3.68

accuracy assessment confirmed that ALS-derived DTM had higher accuracy then PHM-derived DTM. Concerning different resolutions of ALS-derived DTMs (0.5 m, 1 m, 2 m, 5 m) and PHM-derived DTMs (0.5 m, 1 m, 2 m, 5 m, 8 m) compared in this research, the ALS-derived DTM with the finest resolution of 0.5 m showed the highest accuracy. While the accuracy of ALS-derived DTMs slightly decreased with the decrease of its resolution, the differences between the obtained results for all PHM-derived DTM resolutions were almost negligible. Among all PHM-based DTMs, the DTM with the coarser spatial resolution (8 m), which is defined as optimal for given source data, showed the highest accuracy. This suggested that, for a given photogrammetric data, no finer than 8 m spatial resolution was needed. Regardless of the spatial resolution, both ALS- and PHM-derived DTMs on average overestimated elevations of the ground checkpoints, which was in agreement with findings of other recent studies based on ALS. Due to dense near ground vegetation present in the forest, ALS-derived DTMs tended to overestimate the real terrain elevations. Additionally, this research revealed that the national digital photogrammetric data for forest areas contained specific errors (outliers), which considerably affected the DTM accuracy. Visual assessment of the difference raster model enabled detection of the areas of greater elevation differences between the PHM- and ALS-derived DTMs. The elimination of the outliers from photogrammetric data was successfully managed with the improved PHM-derived DTM. The occurrence of outliers is not uncommon for PHM-derived DTMs and their detection presents a challenge for scientists. Thus, in the absence of ALS data, the photogrammetric terrain data could be used for DTM generation in lowland forests with flat terrain but with the greatest caution.

Acknowledgments This research has been supported by the Croatian Science Foundation under the Project IP-2016-06-7686 »Retrieval of Information from Different Optical 3D Remote Sensing Sources for Use in Forest Inventory (3D-FORINVENT)«. The authors wish to thank Hrvatske vode, Zagreb, Croatia for providing ALS data.

126

5. References Aryal, R.R., Latifi, H., Heurich, M., Hahn, M., 2017: Impact of Slope, Aspect, and Habitat-Type on LiDAR-Derived Digital Terrain Models in a Near Natural, Heterogeneous Temperate Forest. PFG - Journal of Photogrammetry, Remote Sensing and Geoinformation Science 85(4): 1–13. ASPRS, 2004: ASPRS Guidelines Vertical Accuracy Reporting for Lidar Data, version 1.0. The American Society for Photogrammetry and Remote Sensing, Bethesda, Maryland, USA. ASPRS, 2008: LAS Specification, version 1.2. The American Society for Photogrammetry and Remote Sensing, Bethesda, Maryland, USA. Axelsson, P., 2000: DEM generation from laser scanner data using adaptive TIN models. International Archives of Photogrammetry and Remote Sensing 33(B4/1; PART 4): 111– 118. Balenović, I., Simic Milas, A., Marjanović, H., 2017: A Comparison of Stand-Level Volume Estimates from Image-Based Canopy Height Models of Different Spatial Resolutions. Remote Sensing 9(3): 205. Brach, M., Zasada, M., 2014: The Effect of Mounting Height on GNSS Receiver Positioning Accuracy in Forest Conditions. Croatian Journal of Forest Engineering 35(2): 245–253. Gil, A.L., Núñez-Casillas, L., Isenburg, M., Benito, A.A., Bello, J.J.R., Arbelo, M., 2013: A comparison between LiDAR and photogrammetry digital terrain models in a forest area on Tenerife Island. Canadian Journal of Remote Sensing 39(5): 396–409. Gobakken, T., Bollandsås, O.M., Næsset, E., 2015: Comparing biophysical forest characteristics estimated from photogrammetric matching of aerial images and airborne laser scanning data. Scandinavian Journal of Forest Research 30(1): 73–86. Guan, H., Li, J., Yu, Y., Zhong, L., Ji, Z., 2014: DEM generation from lidar data in wooded mountain areas by crosssection-plane analysis. International Journal of Remote Sensing 35(3): 927–948. Hill, T., Lewicki, P., 2007: STATISTICS: Methods and Applications. StatSoft, Inc.: Tulsa, OK, USA, 800 p. Höhle, J., Höhle, M., 2009: Accuracy assessment of digital elevation models by means of robust statistical methods. ISPRS Journal of Photogrammetry and Remote Sensing 64(5): 398–406. Croat. j. for. eng. 39(2018)1

Accuracy Assessment of Digital Terrain Models of Lowland Pedunculate Oak Forests ... (117–128)

I. Balenović et al.

Höhle, J., Potuckova, M., 2011: Assessment of the quality of digital terrain models. European Spatial Data Research, Frankfurt. Report No. 60, 91 p.

Shapiro, S.S., Wilk, M.B., 1965: An analysis of variance test for normality (complete samples). Biometrika 52(3–4): 591– 611.

Kraus, K., Pfeifer, N., 1998: Determination of terrain models in wooded areas with airborne laser scanner data. ISPRS Journal of Photogrammetry and Remote Sensing 53(4): 193– 203.

Spaete, L.P., Glenn, N.F., Derryberry, D.R., Sankey, T.T., Mitchell, J.J., Hardegree S.P., 2011: Vegetation and slope effects on accuracy of a LiDAR-derived DEM in the sagebrush steppe. Remote Sensing Letters 2(4): 317–326.

Li, Z., Zhu, Q., Gold, C., 2005: Digital Terrain Modeling – Principles and Methodology. CRC Press, 340 p.

Stereńczak, K., Kozak, J., 2011: Evaluation of digital terrain models generated from airborne laser scanning data under forest conditions. Scandinavian Journal of Forest Research 26(4): 374–384.

Maune, D.F., 2007: Digital Elevation Model Technologies and Applications: The DEM User Manual. 2nd ed., American Society for Photogrammetry and Remote Sensing, 655 p. Mayer, B., 1996: Hydropedological relations in the region of lowland forests of the Pokupsko basin. In: Lowland forests of the Pokupsko basin (Mayer, B., ed.), Radovi Šumarskog instituta Jastrebarsko, 37–89. Puliti, S., Gobakken, T., Ørka, H.O., Næsset, E., 2016: Assessing 3D point clouds from aerial photographs for speciesspecific forest inventories. Scandinavian Journal of Forest Research 32(1): 68–79. Rahlf, J., Breidenbach, J., Solberg, S., Astrup, R., 2015: Forest Parameter Prediction Using an Image-Based Point Cloud: A Comparison of Semi-ITC with ABA. Forests 6(11): 4059– 4071.

Stereńczak, K., Zasada M., Brach, M., 2013: Influence of terrain slope, model pixel size and stand structure on accuracy of DTM generated under pine stands from LIDAR data. Baltic Forestry 19(2): 252–262. Stereńczak, K., Moskalik, T., 2014: Use of LIDAR-based digital terrain model and single tree segmentation data for optimal forest skid trail network. iForest 8(5): 661–667. Stereńczak, K., Ciesielski, M., Balazy, R., ZawiłaNiedźwiecki, T., 2016: Comparison of various algorithms for DTM interpolation from LIDAR data in dense mountain forests. European Journal of Remote Sensing 49(1): 599–621. Su, J., Bork, E., 2006: Influence of Vegetation, Slope, and Lidar Sampling Angle on DEM Accuracy. Photogrammetric Engineering Remote Sensing 72(11): 1265–1274.

Reutebuch, S.E., McGaughey, R.J., Andersen, H.E., Carson, W.W., 2003: Accuracy of a high-resolution lidar terrain model under a conifer forest canopy. Canadian Journal of Remote Sensing 29(5): 527–535.

Terrasolid Ltd., 2012: Terrascan. http://www.terrasolid.fi/en/ products/terrascan.

Ristić, R., Polovina, S., Malušević, I., Radić, R., Milčanović, V., Ristić, M., 2017: Disaster Risk Reduction Based on a GIS Case Study of the Čađavica River Watershed. South-east European Forestry 8(2): 99–106. doi: https://doi.org/10.15177/ seefor.17–12.

Torresan, C., Strunk, J., Zald, H.S., Zhiqiang, Y., Cohen, W.B., 2014: Comparing statistical techniques to classify the structure of mountain forest stands using CHM-derived metrics in Trento province (Italy). European Journal of Remote Sensing 47(1): 75–94.

Sánchez-Lopera, J., Lerma, J.L., 2014: Classification of lidar bare-earth points, buildings, vegetation, and small objects based on region growing and angular classifier. International Journal of Remote Sensing 35(19): 6955–6972.

White, J.C., Stepper, C., Tompalski, P., Coops, N.C., Wulder, M.A., 2015: Comparing ALS and Image-Based Point Cloud Metrics and Modelled Forest Inventory Attributes in a Complex Coastal Forest Environment. Forests 6(10): 3704–3732.

Croat. j. for. eng. 39(2018)1

127

I. Balenović et al.

Accuracy Assessment of Digital Terrain Models of Lowland Pedunculate Oak Forests ... (117–128)

Authors’ addresses: Ivan Balenović, PhD. e-mail: ivanb@sumins.hr Division for Forest Management and Forestry Economics Croatian Forest Research Institute Trnjanska cesta 35 10 000 Zagreb CROATIA Mateo Gašparović, PhD. * e-mail: mgasparovic@geof.hr Chair of Photogrammetry and Remote Sensing Faculty of Geodesy, University of Zagreb Kačićeva 26 10 000 Zagreb CROATIA Assist. prof. Anita Simic Milas, PhD. e-mail: asimic@bgsu.edu School of Earth, Environment and Society Bowling Green State University 190 Overman Hall Bowling Green, OH USA Alen Berta, PhD. e-mail: aberta@oikon.hr Department of Natural Resources Management Oikon Ltd. Institute of Applied Ecology Trg senjskih uskoka 1–2 10 000 Zagreb CROATIA

Received: October 06, 2017 Accepted: November 10, 2017

128

Assoc. prof. Ante Seletković, PhD. e-mail: aseletkovic@sumfak.hr Department of Forest Inventory and Management Faculty of Forestry University of Zagreb Svetošimunska 25 10 000 Zagreb CROATIA * Corresponding author Croat. j. for. eng. 39(2018)1

Original scientific paper

Determination of Service Life of Sintered Powder Metallurgy Gears in Regard to Tooth Bending Fatigue Srečko Glodež, Marko Šori, Krešimir Vučković, Stjepan Risović Abstract The aim of this study is to check the possibility of replacing the pinion gear made of structural steel with the one made of sintered material. The pinion is part of the gear pair mounted in front of the gearbox of the skidder Ecotrac 55V to increase the speed and lower torque. In larger series, powder metallurgy (PM) gears are used as a cost-effective alternative for wrought metal gears in a number of industries including the one producing forest products. The present paper discusses the computational and experimental approach for determining the service life of sintered PM gears in regard to tooth bending fatigue. The proposed computational model is based on the stress-life approach, where the stress field in a gear tooth root is determined numerically using finite element method. The needed material data have been taken from the authors’ previous work. Due to the scattering nature of fatigue, the statistic approach has also been considered by presentation of computational results. The experimental procedure was done on a custom made back-to-back gear testing rig. The comparison between computational and experimental results has shown that the proposed computational approach is an appropriate calculation method for estimating the service life of sintered gears regarding tooth root strength. Namely, it has been shown that, in case of proper heat treatment of tested gears, tooth breakage occurred in the interval with 95% probability of failure, which has been determined using the proposed computational model. Keywords: skidder, service life, sintered powder metallurgy gears, tooth bending fatigue, stresslife approach, back-to-back test rig

1. Introduction There has been an increased awareness of the need to develop forestry techniques and technologies, which is often expressed through the need of providing convenient technical, environmental and ergonomic features as well as achieving higher profits. Therefore, the development of machines and procedures is an important prerequisite for improving a forestry company’s performance in complex terrain or when it comes to specific ways of management in naturally renewable forests. According to Beuk et al. (2007), there are four mayor criteria in terms of suitability of technical and technological solutions: environmental suitability, efficiency, safety, ergonomic suitability. As a result, it is necessary to design new vehicles for off road travel in forest thinning operations, which Croat. j. for. eng. 39(2018)1

are suitable for all cut-to-length, half-tree length and tree-length methods. The first such vehicle for wood skidding from thinnings was the articulated forest tractor IWAFUJI, originated in Japan. Afterwards, in 1987, some of the Croatian local forest units encouraged the structural improvement of domestic equipment and vehicles, especially for thinning operations. The result was a prototype of a new machine in the group of the so called small techniques for wood transportation, middle category skidder Ecotrac 33V. In 1996, Horvat and Sever conducted a research on two agricultural adapted tractors (Torpedo TD 75A and Steyr 9080a) and two middle-size skidders (LPKT 40 and Ecotrac 33V), all of which were intended for wood relocation from thinnings. According to research results, morphologic features of the domestic skidder Ecotrac 33V are much more suitable for such opera-

129

S. Glodež et al.

Determination of Service Life of Sintered Powder Metallurgy Gears in Regard to Tooth ... (129–137)

tions, especially its form index with lower length, width and height than the other researched vehicles. From 1995 to 2007, during a comprehensive study of Ecotrac 33V done by the Croatian Forests Ltd, several technical-technological problems (engine power, break system, winch), maintenance problems, economic profitability and overall working condition problems have been considered. Desiring to keep the basic dimensions of the skidder tractor Ecotrac 33V and build a new skidder capable of working in forest thinning operations and occasionally in main felling, the development started of the new ECOTRAC 55V tractor (see Fig. 1) equipped with forest winch 2x35 kN and 4x4 drive. The skidder was powered by an aircooled tricylinder diesel motor DEUTZ, with a working volume of 3236 cm3, compression factor 20:1, nominal power 40 kW at 3400 min-1 and maximum torque of 207 Nm at 1600 min-1. Power transfer was performed by a mechanical transmission: drive motor – coupling – gearbox – transfer box – front and back differentials with individual blockades – final (planetary) reducers mounted on the tractor wheels. In order to obtain the mobility features of the skidder (Đuka et al. 2016) throughout the thinning operation (length, width, height), while also adapting those required for main felling, such as motor power, a gear pair has been mounted in front of the gearbox to increase speed while lowering torque. The gear pair was added in front of the gearbox due to the limitations of the main gearbox dimensions; as a result, the gearbox was unable to work with greater torque than specified. Therefore, the final reduction had to be carried out through a robust planetary transmission. The aim of this study is to check the possibility of replacing the pinion gear made of structural steel with

Fig. 1 Skidder ECOTRAC 55V

130

the one made of sintered material, which is less expensive because no additional grinding and thermal processing is required. Large series production demands a cost-effective process with low waste of material and low energy consumption. If massively produced parts also have a complex geometry and tight tolerances, Powder Metallurgy (PM) should be considered for the production process. Conventional PM production process (automatic die compaction and sintering) has already proven its advantages (low cost, low waste, high accuracy, high strength, etc.) and shown its shortcomings (dimensional change during sintering, uneven porosity), which can now be kept under control with various methods, including alloying elements and compaction tool design and movement (Danninger at al. 2001, Khoei and Bakhshiani 2004). The main characteristic of sintered PM components is the porosity, which is usually not available directly, but in a form of density, which is the main factor for mechanical properties of a final product (Cipolloni at al. 2014). Density is unfortunately affected by numerous factors, which are not only the obvious theoretical density of the base powder and compaction pressure, but also the amount and type of alloying elements, amount and type of lube, sintering atmosphere, time and temperature, cooling rate after sintering, etc (L’Esperance at al. 1996, Khraisat and Nyborg 2004, Bolzoni at al. 2013). PM components are widely used in motor vehicle industry as oil pump gears and housings, various small component brackets, synchronizer gears and engine gears. The next step is PM power transmission gears (Flodin at al. 2011). The idea of PM gears dates back to 1985, when researchers in Soviet Union investigated tooth strength of sintered planetary pinions for automobile differential (Dorofeev and Baidala 1985). However, due to low Young`s module and low surface hardness, which led to significant wear, the use of PM gears was limited to low load applications. Investigations of RCF (rolling cycle fatigue) on sintered steel led to selective surface densification of gear flanks, which improved surface hardness and therefore reduced wear (Lawcock 2006, Koide at al. 2008, Bengtsson at al. 2001, Bengtsson at al. 2001a). However, surface densification of gear flanks occurs after sintering, thus causing an additional procedure, which raises the cost of the final product. This, consequently, reduces the economic advantage over conventional gear production with milling or hobbing. Recent development in powder metallurgy production technology increased the strength of final PM products to an interesting level Croat. j. for. eng. 39(2018)1

Determination of Service Life of Sintered Powder Metallurgy Gears in Regard to Tooth ... (129–137)

for PM gears applications without additional procedures (Dlapka at al. 2012, Straffelini at al. 2014). Usually, when considering tooth breakage, gear drives are dimensioned according to standard ISO 6336:2006 or in accordance with other procedures, which cannot be used for sintered PM gears. However, some information on sintered gear bending load capacity can be found in AGMA 930-A05:2005 information sheet, which contains a complex geometrical analysis to determine the geometry factor for bending strength. Furthermore, the needed parameters of fatigue strength are not given and should be found in available literature or determined experimentally. This paper uses stress life approach to evaluate the fatigue life of sintered gears in terms of the bending stress field in the tooth root. The computational results are then compared with experimental data, obtained on the back-to-back gear-testing rig.

2. Material and geometry Powder manufacturers are already producing steel metal powders for sintered gears (Höganäs 2015). Based on experience in compaction and sintering, one of such powders was chosen to produce PM gears. It contains 1.47% Cu, 1.69% Ni, 0.50% Mo, 0.29% C and 0.58% of lube with Fe balance. Apparent density of this powder is 3.15 g/cm3 and Hall flow rate is 29 s / 50 g. This powder was compacted into gear specimens and sintered at 1120 °C for 30 minutes in a controlled 10/90 hydrogen and nitrogen atmosphere. After sintering, PM gears were machined to reduce tooth width and to produce a key slot in the hole (see Fig. 2). Sintered PM gears were then austenitized at 915 °C, oil quenched and tempered for one hour at 175 °C. Final density of specimen gears was 7.03 g/cm3.

S. Glodež et al.

Table 1 Mechanical properties of tested gear (Šori at al. 2014, Šori at al. 2014a) Property

Value

Young`s module E, GPa

142

Poisson`s ratio v (Hirose at al. 2004)

0.27

Ultimate tensile strength Rm, MPa

842

Elongation at break A

0.86

Fatigue strength coefficient sf`, MPa

875

Fatigue strength exponent b

–0.153

Mechanical properties of sintered gears (see Table 1) were considered the same as those determined in previous mechanical testing of this sintered steel described in our previous work (Šori at al. 2014, Šori at al. 2014a). Although final density of sintered gears was slightly lower than in previous experiments, everything else remained the same as in the mentioned experiments. Tested sintered pinion with 9 teeth and 5 mm tooth width was paired with wrought carburized tempered steel gear with 31 teeth and 20 mm tooth width at centre distance of 80 mm. Normal module was 4 mm, pressure angle at normal section was 25° and helix angle was zero. Nominal rotational speed of a tested pinion was 800 rpm.

3. Computational estimation of service life Computational estimated service life of a sintered gear with chosen geometry and known material parameters is determined in relation to the maximum tensile

Fig. 2 Sintered gear after compaction and sintering (left), after machining (middle) and after heat treatment (right) Croat. j. for. eng. 39(2018)1

131

S. Glodež et al.

Determination of Service Life of Sintered Powder Metallurgy Gears in Regard to Tooth ... (129–137)

stress field in the tooth root, from which the number of stress cycles is calculated according to the modified Basquin`s exponential law (Stephens at al. 2001)

σ a = σ f, ⋅ ( 2 ⋅ N ) b

(1)

which gives the correlation between amplitude stress σa and the number of stress cycles until failure N. Fatigue strength coefficient σf` = 875 MPa and fatigue strength exponent b = −0.153 are material parameters and have been determined previously using rectangular test specimens, which have been produced in the same way as treated sintered gears (compaction, sintering and additional heat treatment). Detailed experimental procedure is described in Šori at al. 2014a. Due to pulsating load regime in the tooth root, amplitude stress σa is equal to the half of the maximum Principal stress:

s a=

1 ⋅s 2 max

(2)

The maximum stress in tooth root σmax can be calculated according to ISO 6336:2006, but in case of PM gears, the tooth root is not dependant on gear tooth tools and can be optimized to the maximum performance (Liu 2011). Some calculation programs, e.g. KISSSoft 2015, also enable extensive profile corrections, but describing the correct tooth root form with various factors can be more difficult than numerical simulation with the existing 3D model, which is used in compaction die preparation. Therefore, the finite element method (FEM) analysis was used to calculate the tooth stress for a given torque load. In the FEM approach seen in Fig. 3, the gear in question was modelled with one third geometry as Instance 1 and the pairing gear with only one tooth as Instance 2. The contact between them was set at the outermost single contact point. As only the position of contact point is important to tooth root stress field distribution,

the pairing tooth of a large gear was substituted with one pinion tooth, but appropriately wider and stiffer. The surfaces of shaft holes and cut result surfaces were tied to each reference point so that all relative displacements and rotations were constrained. At initial step, all degrees of freedom of the reference points were restricted. To apply the load, the reference point of gear pinion instance was rotated in 20 increments for a fraction of a degree to initiate contact between both instances, which caused a reaction moment in the reference point that was considered as load torque for the pinion, which caused stress field in the tooth root. Due to contact, this approach enabled significantly better simulation stability than boundary condition with directly applied torque. Both sections were modelled with linearly elastic material defined with appropriate Young`s modulus E and Poisson`s ratio ν. Contact between the sections was defined as tangentially frictionless and in normal direction as »hard contact«. Both sections were meshed with half a million tetrahedral C3D4H type elements. For more details about numerical approach, see Glodež at al. 2014. In case of HCF (high cycle fatigue) regime, stress levels are much lower than Yield strength and thus linear material model is sufficiently suitable for such calculation. This assumption makes it possible to find direct linear correlation between load torque T and maximum Principal stress σmax

132

(3)

where κ is a correlation coefficient, which is dependent on gear geometry and linear material properties (Young`s module E and Poisson`s ratio ν), and was determined to be 8.4408 m1mm2 in our case. If Eq. 2 is inserted in Eq. 1 and σmax substituted according to Eq. 3, direct correlation between load torque T and estimated number of stress cycles until failure Nest can be written as in Eq. 4, which can also be written as an equation of a straight line in Log(T) – Log(Nest) plane, as in Eq. 5.

Fig. 3 3D model (left) and mesh (right) of treated gear pair

s max= k ⋅ T

1 b k ⋅ T = s f, ⋅ ( 2 ⋅ N est ) 2

= T

2 ⋅ s f, b ( 2 ⋅ N est ) k

(4) (5)

Furthermore, when estimating the number of stress cycles until failure, the scatter of dynamic tests should be considered. Usually at least 10 fatigue tests are conducted at the same stress level to determine the scatter and to evaluate stress life with certain probability (Glodež at al. 2013). In such approach, material parameters σf` and b can vary in dependence to stress level. Due to low number of data-points in previous mateCroat. j. for. eng. 39(2018)1

Determination of Service Life of Sintered Powder Metallurgy Gears in Regard to Tooth ... (129–137)

S. Glodež et al.

Fig. 4 Definition of sf`i for data-point Ti with coordinates (Ni, sai)

Fig. 5 Area of estimated number of stress cycles Nest with 95% probability of failure

rial testing (Šori at al. 2014A), fatigue strength exponent b was assumed to be constant at all stress levels. However, to estimate the area in S-N diagram with 95% probability of failure, fatigue strength coefficient σf`i for each data point Ti with coordinates (Ni, σai) was calculated in a way schematically shown in Fig. 4. Standard deviation of fatigue strength coefficient δσf was then calculated on a set of points σf`i, making it possible to write Eq. 6, which is Eq. 5 extended with ±2·δσf.

for our case. Under the lower one, the probability of failure is less than 2.5% and over the upper one, the probability of failure is more than 97.5%. Effectively, they enclose an area with 95% probability of failure.

T =

2 ⋅ ( sf′ + 2 ⋅ dof )

⋅ ( 2 ⋅ Nest ) b

(6)

Eq. 6, therefore, defines two curves in T – Nest diagram, which are drawn in Fig. 5 also with median curve

4. Experimental evaluation of service life Evaluation of service life was done on a custom made backtoback gear testing rig shown in Fig. 6 with a mechanical torque loop between test gear pair in housing (1) and support gear pair in housing (2). Backtoback configuration enables large testing moments in comparison to the power of a driving motor (3), which only provides rotational movement and covers the

Fig. 6 Custom made gear-testing rig Croat. j. for. eng. 39(2018)1

133

Determination of Service Life of Sintered Powder Metallurgy Gears in Regard to Tooth ... (129–137)

Test pinion

Support gear

Centre distance a, mm

Normal module m, mm

Pressure angle at normal section an, °

Helix angle b, °

Number of teeth z

CuNiMo sintered steel

42CrMo4

800

232

Material Nominal rotational speed n, rpm Lubrication

Immersed in gearbox oil: ISO VG 220

losses caused by friction between the gears, in bearings and between gaskets and shafts. Basic parameters of the testing rig are given in Table 2. Setting the loading torque begins with constraining the shaft rotation near the coupling (7) at position (b). Torque is then applied at position (a), next to support bearing (9). With loading torque applied, coupling (7) is engaged and the mechanical loop is established. The loading torque of the closed loop is monitored with a 200 Nm torque transducer (4), coupled between the shafts with two bellow couplings (6). A smaller 20 Nm torque transducer (5) is used to monitor the exact input torque at the motor (3), which is connected with a bellow coupling (6) to the motor (3) and with a safety bellow coupling (8) to the testing rig. Data acquisition system (DAQ) connects the two torque transducers (4 and 5) to the computer, where measured data are recorded at 50 Hz. This data acquisition rate provides 3.75 measurements per pinion revolution and prevents data overflow at the runout tests – 106 pinion revolutions. Data acquisition starts automatically, when the load at smaller 20 Nm torque transducer (5) is more than 2 Nm and also stops automatically, when the torque is lower than 0.5 Nm. If gear breakage occurs, test gears in housing (1) jam and safety coupling (8) disengages. According to the operation period estimated in the previous section, such loading torque was applied that the expected number of cycles until failure was between 104 and 106. Therefore, the loading torques varied between 22 and 40 Nm – see Table 3. However, the exact setting of the desired torque during the complete test was not possible due to the way of torque application, which does not permit configuration of loading torque during operation of the testing rig. In cases without tooth flank wear, this would not have posed a problem, but with wear of tooth flanks, the loading

134

5. Results and discussion Results of sintered gear testing are shown in Table 3, where rotational speed, average torque, time and number of cycles are given for each tested sintered gear marked from K01 up to K20. The average torque was calculated from all data points from the beginning of the test until tooth breakage. The time between these two events is also given. The number of cycles was then calculated as a product of rotational speed and time. In the last column, some remarks are given where necessary. The first four test gears achieved significantly lower operation time as expected and were, therefore, Table 3 Gear testing results Number of cycles

Time min

Average torque Nm

Width b, mm

Rotational speed rpm

Parameter

torque drops. The majority of the tests were done at 800 rpm, which is 33% faster than the fatigue tests on specimens in previous tests. Two tests were also done for higher rotational speed at 1200 rpm for a quick speed dependence evaluation discussed in the following section and are clearly marked in the table.

K01

800

34.3

24.946

19 957

K02

800

28.0

83.179

66 543

K03

800

28.9

61.223

48 978

K04

800

27.0

61.898

49 518

K05

800

22.4

1293.1

1 034 480

K06

800

34.3

142.55

114 040

K07

800

39.5

56.885

45 508

K08

800

31.5

286.94

229 552

K09

800

28.4

400.80

320 640

K10

800

26.2

569.79

455 832

Remarks

Table 2 Basic parameters of the testing rig

Specimen

S. Glodež et al.

Inappropriate heat treatment Runout

K11

1200

25.0

838.38

1 006 056

K12

800

29.1

294.36

235 488

Runout

K13

1200

29.6

809.26

971 112

K14

800

38.3

54.105

43 284

K15

800

31.6

149.77

119 816

K16

800

31.3

185.71

148 568

K17

800

25.8

313.68

250 944

K18

800

25.2

288.08

230 464

K19

800

22.2

1258.8

1 007 040

Runout

K20

800

22.6

12504

10 003 200

Runout

Shaft failure

Croat. j. for. eng. 39(2018)1

Determination of Service Life of Sintered Powder Metallurgy Gears in Regard to Tooth ... (129–137)

subjected to metallographic investigation, where inappropriate heat treatment was discovered, which was caused by an electrical issue in one of the heaters in the furnace for additional heat treatment. Subsequent test gears were then properly heat-treated and tested as seen in Table 3. All the results, except K20, are plotted in Fig. 7. Regular tests are marked with black dots and are well within 95% probability of estimation, which is the grey area in the diagram. The first four tests with inappropriate heat treatment are marked with black circles and are on the edge of the estimation area. Black cross indicates an irregular test, where shaft failure occurred, but the gear remained intact. All four marks near 106 indicate runouts, but the two triangles are the tests ran at 1200 rpm and one of them is well outside the estimation area. This suggests a speed dependence of fatigue tests. The calculated estimation is based on fatigue tests that were done at 10 Hz, which would result in rotational speed of 600 rpm. The majority of tests were, however, done at 800 rpm, which is 33% more than fatigue tests. On the other side, the tests at 1200 rpm have the double gear meshing frequency of the fatigue tests, which reduces the time of contact and therefore increases the number of stress cycles until failure. This, off course, does not affect the service time, since rotational speed is higher. Examination of failed test gears showed that all of them had similar damage. Typical damage of a tested gear is shown in Fig. 8, presenting a photo of the test gear K08. Apart from obvious tooth breakage, severe pitting of the working tooth flank can be seen. Although

S. Glodež et al.

Fig. 8 Typical damage after test – test gear K08 sintered gears were additionally hardened, the porosity of the material kept hardness relatively low if compared to wrought steel gears, but this can be improved by surface rolling (Sonsino at al. 1992). The hardness of tested gears was measured to be from 310 HV to 340 HV. Significant wear is also visible at tooth tip, which is a result of high sliding velocities and possibly also a consequence of a low Young’s module, which could cause meshing interference. Another area of extreme wear is seen at tooth root on a neighbouring tooth. As it is only seen at the neighbouring tooth, this wear can be contributed to unusual meshing conditions right after tooth breakage and before jamming of gears.

6. Conclusions

Fig. 7 Gear tests versus calculated estimation of operation period Croat. j. for. eng. 39(2018)1

The aim of research presented in this paper was to compare the computational and experimental approach for determining the service life of sintered PM gears in regard to tooth bending fatigue in order to check the possibility of replacing the steel gear used in skidder ECOTRAC 55V. The proposed computational model is based on the stress-life approach, where the needed material data was overviewed and referenced to our previous work. The stress field in a gear tooth root was determined numerically using FEM. The experimental determination of service life was done on a custom made backtoback gear testing rig. The standardized procedure (AGMA 2005) for determining the load capacity of sintered gears requires the calculation of many influential factors that are mainly dependent on gear geometry. FEM analysis may be used to replace this procedure and determine stress field numerically. Usually, dies for sinter-press

135

S. Glodež et al.

Determination of Service Life of Sintered Powder Metallurgy Gears in Regard to Tooth ... (129–137)

technology are made by wire-EDM and at least 2D contour exists, which can be used for quick preparation of a FEM model. Due to the scattering nature of fatigue, not just a single value is obtained for a particular load, but an interval with arbitrary probability, which is expanded for all torque loads as an area plotted in a load to service time diagram. Due to small number of data points in fatigue tests, standard deviation was calculated based on fatigue strength coefficient σf` assuming that fatigue strength exponent b remains constant at all load levels. To validate the proposed approach, the expected service life was calculated with 95% probability of failure with material parameters from our material characterization work (Šori at al. 2014a, Šori at al. 2014b). Sintered pinions were then produced and tested on a custom build back-to-back gear-testing rig until tooth breakage. It was shown that, in case of proper heat treatment, tooth breakage occurred well within the calculated estimation. Therefore, it can be assumed that this approach is an appropriate calculation method for estimating the service time of sintered gears regarding tooth root strength. The computational model was successfully confirmed with experimental tests on a custom-made gear-testing rig. However, during the evaluation, some further damage, other than gear tooth breakage, was also observed. Presumably, due to relatively low hardness of tooth flanks, severe pitting occurred on the working flanks of sintered gears. Furthermore, some wear was spotted at gear-tooth tips, which may also be a result of low hardness or even improper meshing of gears due to lower Young`s module of the sintered gear. Our future work will, therefore, focus on wear estimation of sintered PM gears. The aim will be to improve the surface hardness with the existing method and to implement sinter-hardening procedure, where sintered parts are hardened right after sintering, thus reducing production time and cost. Because of great loads on the gearbox of skidder Ecotrac 55V, and considering the obtained results of low tooth root strength and observed tooth flank damage, it can be concluded that the pinion gear made of sintered material is not a suitable replacement for the one made of structural steel.

Acknowledgements This study was funded by the Slovenian Research Agency in the scope of the Training of Young Researchers programme (grant number: 1000-11-310171) and by Ministry of Science and Education of the Republic of Croatia in the scope of Croatian-Slovenian Scientific and Technological Cooperation.

136

7. References AGMA, 930-A05: 2005: Calculated Bending Load Capacity of Powder Metallurgy (P/M) External Spur Gears. American Gear Manufacturers Association. Alexandria, Virginia, U.S.A: 78 p. Bengtsson, S., Fordén, L., Dizdar, S., Johansson, P., 2001: Surface densified P/M transmission gear. PM01-25: Paper Presented at 2001 International Conference on‘ Power Transmission Components. Advances in High Performance Powder Metallurgy Applications’ Ypsilanti, Michigan, USA, October 16–17. Bengtsson, S., Fordén, L., Johansson, P., Dizdar, S., 2001a: Rolling contact fatigue tests of selectively densified materials. SAE Technical Paper 2001-01-3285: 10 p. Beuk, D., Tomašić, Ž., Horvat, D., 2007: Status and development of forest harvesting mechanisation in Croatian state forestry. Croatian Journal of Forest Engineering 28(1): 63–82. Bolzoni, L., Ruiz-Navas, E. M., Gordo, E., 2013: Influence of sintering parameters on the properties of powder metallurgy Ti-3Al-2.5V alloy. Materials Characterization 84: 48–57. Cipolloni, G., Menapace, C., Cristofolini, I., Molinari, A., 2014: A quantitative characterisation of porosity in a Cr-Mo sintered steel using image analysis. Materials Characterization 94: 58–68. Danninger, H., Frauendienst, G., Streb, K.-D., Ratzi, R., 2001: Dissolution of different graphite grades during sintering of PM steels. Materials Chemistry and Physics 67(1): 72–77. Dlapka, M., Danninger, H., Gierl, C., Klammer, E., Weiss, B., Khatibi, G., Betzwar-Kotas, A., 2012: Fatigue Behaviour and Wear Resistance of Sinter-Hardening Steels. International Journal of Powder Metallurgy 48(5): 49–60. Dorofeev, Y. G., Baidala, É. S., 1985: Static strength of the teeth of powder metallurgy planetary pinions for the Zhiguli automobile differential. Soviet Powder Metallurgy and Metal Ceramics 24(8): 644–477. Đuka, A., Pentek. T., Horvat. D., Poršinsky, T., 2016: Modelling of Downhill Timber Skidding: Bigger Load – Bigger Slope. Croatian Journal of Forest Engineering 37(1): 139–150. Flodin, A., Brecher, C., Gorgels, C., Rothlingshofer, T., Henser, J., 2011: Designing Powder Metal Gears. Gear Solutions 9(101): 26–35. Glodež, S., Šori, M., Kramberger, J., 2013: Prediction of micro-crack initiation in high strength steels using Weibull distribution. Engineering Fracture Mechanics 108: 263–274. Glodež, S., Šori, M., Verlak, T., 2014: A Computational Model for Bending Fatigue Analyses of Sintered Gears. Strojniški vestnik – Journal of Mechanical Engineering 60(10): 649–655. Hirose, N., Asami, J., Fujiki, A., Oouchi, K., 2004: Poisson’s Ratio of Sintered Materials for Structural Machine Parts. Journal of the Japan Society of Powder and Powder Metallurgy 51(7): 515–521. Croat. j. for. eng. 39(2018)1

Determination of Service Life of Sintered Powder Metallurgy Gears in Regard to Tooth ... (129–137)

S. Glodež et al.

Höganäs, http://www.hoganas.com/. Accessed 6 October 2015.

properties of sinter hardening alloys. Advances in Powder Metallurgy and Particulate Materials 3: 11 p.

Horvat, D., Sever, S., 1996: Neke tehničke značajke traktora za privlačenje drva u prorjedama sastojina brdsko-planinskog područja. Šumarski list, znanstveno-stručno i staleško glasilo Hrvatskoga šumarskog društva 120(3–4): 157–162.

Lawcock, R., 2006: Rolling-contact fatigue of surface-densified gears. International Journal of Powder Metallurgy 42(1): 17–29.

ISO, 6336: 2006: Calculation of load capacity of spur and helical gears. International Standard Organization. Geneva, Switzerland. Khoei, A. R., Bakhshiani, A., 2004: A hypoelasto-plastic finite strain simulation of powder compaction processes with density-dependent endochronic model. International Journal of Solids and Structures 41(22–23): 6081–6110. Khraisat, W., Nyborg, L., 2004: Effect of carbon and phosphorus addition on sintered density and effect of carbon removal on mechanical properties of high density sintered steel. Materials Science and Technology 20(6): 705–710. KissSoft A. G., 2015: KissSoft release 03/2015 User manual. Bubikon, Switzerland: 1209 p. Koide, T., Ishizuka, I., Takemasu, T., Miyachika, K., Oda, S., 2008: Load bearing capacity of surface-rolled sintered metal gears. Int. J. of Automation Technology 2(5): 334–340. L’Esperance, G., Duchesne, E., De Rege, A., 1996. Effects of materials and process parameters on the microstructure and

Liu, H. R., 2011: The Profile Calculation and the Best Fillet of Powder Metallurgical Gears. Materials Science Forum 694: 851–854. Sonsino, C. M., Müller, F., Mueller, R., 1992: The improvement of fatigue behaviour of sintered steels by surface rolling. International Journal of Fatigue 14(1): 3–13. Stephens, R. I., Fatemi, A., Stephens, R. R., Fuchs, H. O., 2001: Metal fatigue in engineering. New York: John Wiley & Sons Inc: 496 p. Straffelini, G., Benedetti, M., Fontanari, V., 2014: Damage evolution in sinter-hardening powder-metallurgy steels during tensile and fatigue loading. Materials & Design 61: 101–108. Šori, M., Šuštaršič, B., Glodež, S., 2014: Fatigue properties of sintered DIN SINT-D30 powder metal before and after heat treatment. Materiali in Tehnologije 48(6): 837–840. Šori, M., Verlak, T., Glodež, S., 2014a: Heat treatment effects on static and dynamic mechanical properties of sintered SINT D30 powder metal. Key Engineering Materials (592– 593): 643–646.

Authors’ addresses: Prof. Srečko Glodež, PhD. e-mail: srecko.glodez@um.si Marko Šori, PhD. e-mail: marko.sori@um.si University of Maribor Faculty of Mechanical Engineering Smetanova 17 2000 Maribor SLOVENIA Asst. prof. Krešimir Vučković, PhD. * e-mail: kresimir.vuckovic@fsb.hr University of Zagreb Faculty of Mechanical Engineering and Naval Architecture Ivana Lučića 5 10000 Zagreb CROATIA

Received: December 6, 2017 Accepted: December 18, 2017 Croat. j. for. eng. 39(2018)1

Prof. Stjepan Risović, PhD. e-mail: risovic@sumfak.hr University of Zagreb Faculty of Forestry Svetošimunska cesta 25 10000 Zagreb CROATIA * Corresponding author

137

S. GlodeĹž et al.

138

Determination of Service Life of Sintered Powder Metallurgy Gears in Regard to Tooth ... (129â&#x20AC;&#x201C;137)

Croat. j. for. eng. 39(2018)1

Subject review

Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten Years (2007–2016) – Part Two: Comminution, Transport & Logistics Martin Kühmaier, Gernot Erber Abstract Within the fuel wood supply chain, comminution and transport have been identified as processes with the highest costs, energy consumption and emissions. The coordination of comminution and transport aimed at avoiding operational delays is also complex. Nevertheless, the use of forest biomass helps to reduce the effects of climate change and produces an additional income, especially in rural areas. About 20 years ago, at the beginning of the industrial forest fuel utilisation, the focus of the research was on developing and analysing adequate supply chains and machines. Nowadays, as state-of-the-art systems have been established, the focus is on improving the efficiency of the processes and the quality of the products. This paper provides a review of research trends of the last ten years focusing on comminution and transport of forest biomass in Europe. Comminution should become more efficient by analysing the effects of wood characteristics on chipper performance and product quality, by tailoring chipper configuration according to those findings and by introducing mechanical devices for improving the quality of chips. Transport processes have the potential to become more efficient if the configuration of trucks is adapted according to operational and legal requirements, and when considering moisture content management. Finally, economic and environmental assessment of supply chains was made by several studies. Future research is expected to focus on customizing the product quality according to user’s requirements and on optimising the coordination of chipper and truck by simulation and automatization tools. Keywords: forest biomass, chipping, environmental assessment, wood energy

1. Introduction The generation of energy from biomass plays an important role in current international strategies to mitigate climate change and to enhance energy security. The European Union (EU) has committed to produce 27% of its energy from renewable sources by 2030 (COM/2014/015). Forest biomass is a key renewable energy source to help countries meet their longterm renewable energy targets. However, a large proportion of the available wood biomass is not utilised due to difficult operating conditions, low efficiency and high supply chain costs (Ghaffariyan et al. 2017). Croat. j. for. eng. 39(2018)1

Therefore, the key objectives of the research are to improve the efficiency and reduce the supply chain cost. Several overviews of state-of-the-art technologies and efficient biomass harvesting have been compiled recently. Stampfer and Kanzian (2006) focused on the current and development possibilities of comminution and transport in Austria, outlining the challenges and opportunities in mountainous regions. Routa et al. (2013) investigated the driving forces behind the current technical solutions of forest energy procurement systems in Finland and Sweden and presented some perspectives on possible future developments.

139

M. Kühmaier and G. Erber Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten ... (139–152)

To access and collect the papers relevant for this review, an extensive literature search was conducted. »Scopus«, »Web of Science« and »Google Scholar« are the most used search engines. A combination of the following key words was applied so that at least one word from each of the search terms in boxes (logical OR operator) and at least one term from each box should appear (logical AND operator) either in the title or the abstract of the paper: »comminution«, »chipping«, »transport«, »fuelwood«, »energy wood« and »supply chain«. Except for some highly relevant papers from other continents, only studies performed or demonstrated in Europe were included in the analysis. This brute-force search resulted in a gross list of about 105 papers, of which 7 were not relevant for the focus of this paper, leaving 98 papers for the review. The remaining papers were summarised and classified into comminution and transport processes and

140

3. Results 3.1 Comminution Biomass supply is very challenging in European mountain forests because of steep terrain, limited space, narrow roads and extreme weather conditions. Despite the challenging operating conditions, chipping contractors in Italy are able to achieve a high machine use and product output. Chipping contractors adopt different operational strategies to achieve their production targets (Spinelli and Magagnotti 2014a). 3.1.1 Allocation of the comminution process within the supply chain The efficiency of the comminution process within the fuel supply chain varies greatly depending on the production site. Of the various supply chain designs considered, chipping at the landing seems to be often the most suitable option (Röser et al. 2011). Roadside Table 1 Supply processes and research trends. As many publications included more than one research trend, publications can be counted several times and the sum of the numbers in the right column does not represent the total number of papers Topic

Comminution

2. Material and methods

12 research trends, each section containing novel knowledge gained during the last decade. Many papers fitted to more than one supply process or research trend and were, therefore, counted multiple times (Table 1). It showed that comminution was an extensively studied supply process during the last 10 years (Fig. 1).

Transport

íaz-Yáñez et al. (2013) compiled a general overview D of current procurement methods of forest chips in Europe. Ghaffariyan et al. (2017) provided a state-of-the art overview of best practice examples of forest biomass harvesting technologies and supply chains used in North America, Europe and the Southern Hemisphere. Eriksson et al. (2013) were providing an overview of state-of-the-art in woody biomass comminution and sorting in Northern Europe. Wolfsmayr and Rauch (2014) compiled a review of the primary forest fuel supply chain, focusing on transportation of primary forest fuel to heat and/or power plants. Gold and Seuring (2011) presented a review of articles published from 2000 to 2009, covering the interface of bio-energy production and logistics and supply chain management issues. De Meyer at al. (2014) gave an overview of the optimisation methods and models focussing on decisions regarding the design and management of the upstream segment of the biomass-for-bioenergy supply chain. Eskandarpour et al. (2015) did the same but focusing on sustainability. However, no paper covers comprehensively all relevant research trends in the field of forest fuel supply chains. Therefore, this paper, as part two of a series of two, aims to cover the last two steps in the forest fuel supply chain, namely comminution and transport. Part one, dealing with harvesting and storage, was analysed by Erber and Kühmaier (2017). Papers published between the years 2007–2016 will be classified according to key supply processes and research trends. Finally, the need for future research will be identified to push forward both industrial and academic development.

Research trends

Publications

Allocation of the comminution process within the supply chain

Effects of wood characteristics on chipper performance

Effects of wood characteristics on product quality

Evaluating the effects of chipper design

Evaluating the effects of knife configuration

Evaluating the effects of screens and sieves

Operator effects and impacts on human health

Selecting suitable transportation modes

Improving the efficiency of fuel wood transportation

Coordination of supply processes

Economic assessment of supply chains

Environmental assessment of supply chains

Multi tree handling in fuel wood harvesting

Croat. j. for. eng. 39(2018)1

Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten ... (139–152) M. Kühmaier and G. Erber

Fig. 1 Number of publications by topic and year. As many publications included more than one research trend, publications can be counted several times for different research trends and supply processes chipping is very common in Europe and usually more productive than terrain chipping, and it allowes reducing harvesting costs (Kärhä 2011, Marchi et al. 2011). At the roadside, space is limited because operations are executed on narrow roads and small landings. Since wood is often directly chipped into transport vehicles, waiting and delay times will occur. The organization of the supply chain, especially scheduling vehicles, changing transport units and transferring the chipper to the next pile is a crucial point (Spinelli and Visser 2009). Croat. j. for. eng. 39(2018)1

At the terminal or at the plant, however, there is more space for changing transport units and for chipping without coordinating transport vehicles. Therefore, higher productivities can be achieved (Ranta and Rinne 2006). Depending on the assortments to be comminuted, the efficiency at the terminal could be increased by up to 43% compared to chipping at the forest landing. This also leads to a reduction in chipping costs between 0.11 and 1.02 €/m3 loose (Kühmaier et al. 2016). In an Irish case study, whole tree terrain chipping was the lowest cost method of woodchip

141

M. Kühmaier and G. Erber Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten ... (139–152)

production in conifer first thinning but no brash was left on the extraction racks for machines to operate on (Kent et al. 2011). The majority of all stump wood chips consumed were comminuted at the plant, and with only around one fifth comminuted at terminals (Kärhä 2011, Röser et al. 2011). 3.1.2 Effects of wood characteristics on chipper performance The characteristics of the raw material have an effect on chipper productivity (e.g. Mizaras et al. 2011, Kováč et al. 2011, Röser et al. 2012, Mola-Yudego et al. 2015). Pochi et al. (2015) and Assirelli et al. (2013) analysed the chipping performance of logs and tops. Chipping stems required more power and torque than chipping residues. Fuel consumption was not affected by tree part, storage or their combination. According to a study of Nuutinen et al. (2015), the productivity of forest residues grinding was the highest. The productivity of whole tree grinding was second highest and the lowest productivity was observed in stump grinding. Nuutinen et al. (2016) have shown that, when comminuting bundles, the productivity was 1.5–3.2 times higher than for unbundled forest residues. Spinelli et al. (2011b) determined that species and moisture content have a secondary effect on chipper productivity and fuel consumption, which are primarily controlled by piece size. Eisenlauer and Teipel (2016) have shown that the energy demand during chipping increases with a higher moisture content and lower particle sizes of the final products. At the physical level, cutting along rather than across the fibre direction generally requires less force, and cleavage close to the surface requires less force because the pressure on the knife from the surrounding wood is lower. Abdallah et al. (2014) developed a chipping test bench to measure cutting forces during the wood chipping process, which implied the use of oversized chipper motors. Energy consumption can also be reduced by exploiting movement in the direction of the knife edge (Eriksson et al. 2013). 3.1.3 Effects of wood characteristics on product quality The quality of wood chips is dependent on the raw material processed (e.g. Patterson et al. 2011, Kuptz and Hartmann 2015, Nuutinen et al. 2016). Nati et al. (2010) have shown that tree species (poplar or pine) and tree part (residues or logs) have a significant impact on chip size distribution. The requirements for high fuel quality are best met when wood chips are produced from round wood because they contain a smaller proportion of oversize particles and a higher proportion of accepts. In contrast, wood chips from

142

forest residues can be considered more suitable for medium or larger CHP and heating plants as they do not require high quality chips (Kuptz and Hartmann 2015). The effect of feedstock type has generally a strong effect on energy efficiency and product quality (Spinelli and Magagnotti 2013, Spinelli et al. 2015, Kons et al. 2015). For the same large mesh screen, poplar chips tend to be larger than pine chips and to contain a higher proportion of oversize particles. On the contrary, pine chips tend to be smaller and to contain a higher proportion of fines. The average size of beech chips is significantly larger than that of poplar chips, possibly due to the higher strength of beech wood (Spinelli and Magagnotti 2013). Krajnc and Dolšak (2014) found that softwood biomass has rougher structures and hardwood biomass finer structures, which is more suitable for larger systems. When residues were chipped, a larger share of smaller chip fractions, a lower ash content, and a slightly higher moisture content were found (Vangansbeke et al. 2015). Spinelli et al. (2011b) came to the conclusion that moisture content has a significant effect on the particle size distribution. 3.1.4 Evaluating the effects of chipper configuration Productivity and energy use are more likely to be determined by machine types (Nuutinen et al. 2016, Yoshida et al. 2016), machine configuration and machine-level factors (Kuptz and Hartmann 2015), such as the speed of rotation, feeding rate, available power, and conversion efficiency. Using more efficient power sources and reducing the power required during interruptions may be at least as important as improving the performance of the physical comminution process (Eriksson et al. 2013). 3.1.4.1 Chipper design Long-term productivity varies with machine type on which the chipper is mounted: tractor-powered units are less productive than larger independentengine chippers (Spinelli and Magagnotti 2014a, Laitila and Routa 2015). The productivity of drum chippers is higher than that of disc chippers (Spinelli and Magagnotti 2013, Facello et al. 2013b). Spinelli et al. (2015) tested two alternative drum chipper designs on different feedstock types and under different knife wear conditions. The closed drum full-length knife design was more efficient than the open drum staggered-knife design, when negotiating branches, especially when knives were dull. Under these conditions, productivity was higher, fuel use lower and product quality better for the closed drum design. The grapple load of the crane had a large effect on the overall productivity of the operations (Röser et al. Croat. j. for. eng. 39(2018)1

Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten ... (139–152) M. Kühmaier and G. Erber

2012). A smaller crane will have positive secondary effects on the operation since it should reduce the fuel consumption and the equipment stress (Laitila and Routa 2015). Picchi and Eliasson (2015) evaluated a container handling chipper truck (CCT) to study the utilisation and to determine if the choice of grapple on the CCT or the preceding forwarder influenced chipping productivity. A standard residue grapple was the better choice for the CCT, while residues forwarded with an asymmetrical grapple increased chipping efficiency. Latest chipper models feature new in-feed and evacuation systems that can be adjusted on the fly to match variable work conditions. A study by Spinelli et al. (2016a) verified the effects of the two systems on productivity, diesel fuel consumption and chip quality. The feedstock type has a dominant effect on all the studied parameters, whereas in-feed mode has no effect on any of them. In contrast, blower setting has a significant effect and offers a strong potential for increased wood fuel processing efficiency. In particular, decreasing blower speed, when full ejection power is not necessary, allows reducing diesel fuel consumption while increasing chip integrity. Spinelli et al. (2012b) analysed a chipper prototype fitted with innovative tubular blades, mounted on a flywheel, designed to produce high quality chips when processing delimbed logs. The machine was as efficient as most conventional disc or drum chippers in the same size class, but offered a much better chip quality. Chips were free from any particles longer than 45 mm, and with a very limited content of fine particles. A tractor-powered drum chipper was designed to reduce the gap between industrial chippers and small-scale chippers (Spinelli and Magagnotti 2013). Using hybrid systems may yield higher energy efficiency compared to direct diesel-powered comminution systems. In order to design hybrid chippers, a series of data on load variations is required, in order to estimate the amount of energy that needs to be stored, and the peak power required. Di Fulvio et al. (2015) have studied the effects of wood properties on the specific power and energy demand and time consumption of a 30 kW electric chipper. 3.1.4.2 Knife configuration Depending on feedstock type, the knife configuration is 50% more productive than the hammer configuration and requires 24% less energy. Fuel consumption is 52% higher for the hammer configuration (Spinelli et al. 2012a). Hammer mills are not suitable for the comminution of raw material with high water content. On the other hand, the comminution of wood with low moisture content with hammer mills proCroat. j. for. eng. 39(2018)1

duces chips with smaller particles sizes, using the same processing energy as knive mills (Eisenlauer and Teipel 2016). Knife angle, moving speed and drum spinning speed have an influence on the average grain size, the dust share and the form constancy (Krajnc and Dolšak 2014). In case of use of a larger knife-edge angle, a reduction of energy consumption can be achieved due to the increased compressive loading parallel to the wood fibres. Isaksson et al. (2013) defined a chip damage parameter D of spruce, which is relevant for cracking parallel to the fibres. D is defined and its dependence on chip length and edge angle of the chipping knife is analysed numerically by means of finite element analyses. Wearing chipper knives causes a significant reduction of chipping productivity and a remarkable increase of fuel consumption (e.g. Nati et al. 2010, Facello et al. 2013a, Spinelli and Magagnotti 2014a, Spinelli et al. 2014a, Kuptz and Hartmann 2015). Dry sharpening with a grinder mitigated this effect, but it could not replace proper wet sharpening. Increasing the frequency of wet sharpening sessions determined a moderate increase of knife depreciation cost, but it could drastically enhance machine performance and reduce biomass processing cost (Spinelli et al. 2014a). Increasing knife sharpness requires more downtime, and therefore an optimum time interval between knife replacements has to be found (Eriksson et al. 2013). Costs of severe damage caused to conventional knife set-ups, following accidental introduction of metal contaminants inside the chipper, can amount to over 30,000 €. Disposable micro-knives may avoid such severe damage and offer savings of about 30% of the knife-related cost of a conventional knife set, or about 18 Euro cents per tonne (Spinelli and Magagnotti 2014b). Cut length setting and piece breaker option are relevant drivers of chip size, and they are manipulated with the main purpose of managing particle size distribution. A study of Facello et al. (2013a, 2013c) showed that the proportion of small chips increased dramatically with the shortest cut length setting (7 mm). Installing a piece breaker allowed maximizing the incidence of small chips, which reached 70% of the total mass when the piece breaker was used in combination with the shortest cut length setting. Reducing cut length determined a substantial decrease of productivity (ca. 30%), and an even higher increase of specific fuel consumption (ca. 50%). All strategies to reduce chip size also resulted in increasing the incidence of fines. Power and energy consumption are lower when processing with a larger cutterhead diameter (Kuljich et al. 2015). These parameters were also greater when cutting frozen logs compared with unfrozen logs.

143

M. Kühmaier and G. Erber Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten ... (139–152)

Experiments made with a model helical chipper have shown that the infeed angle has a huge impact on chip quality parameters. Furthermore, power requirement and energy consumed are decreasing with increasing infeed angle. The results seem to be extraordinary since increasing infeed angles lead to decreasing chip sizes being produced (Wegener et al. 2015).

3.1.4.3 Screens and sieves For optimal combustion, the fuel should have a low content of fine particles (Kons et al. 2015), which can be achieved by screening or sieving. A newly designed mobile screening device achieved an average productivity of 1.9 odt/h, corresponding to screening costs of 28.5 €/odt (Spinelli et al. 2011a). This figure was lower than the price increase obtained by upgrading the industrial chips to residential user standards. The economic value of such screening depends heavily on the costs of the refining process and the value/utility of the separated fine particles. Laitila and Nuutinen (2015) evaluated the significance of screening to guarantee sufficient quality when processing stump fuel. On the other hand, Kons et al. (2015) and Eliasson et al. (2015a) found that the sieve size had no significant effects on particle size distribution and ash content. The replacement of the standard wide mesh screen with a narrower screen causes decreasing productivity and increasing fuel consumption (Nati et al. 2010). Röser et al. (2012) revealed that there are significant differences in the chipping productivity in Austria and Finland, which are largely based on the use of different sieve sizes. The use of narrower 80 mm × 80 mm sieves on Scots pine material does not seem to offer any benefit compared to 100 mm × 100 mm from the chip quality point of view (Laitila and Routa 2015). Nati et al. (2015) tested the use of a trommel screen originally designed for compost materials to reject oversize particles from hog fuel. The study consisted in screening material previously comminuted by a convertible crusher, designed to use both hammers and knives. Trommel screen productivity varied between 4.2 t/h, and 5.2 t/h of oven dry material. Screening hog fuel derived from pallets was respectively 30% and 40% less productive than screening fuel derived from logs and residues. 3.1.5 Operator effect and impacts on human health »Operator effect« has a strong impact on operational chipper performance, due to individual differences in technique, motor skills, work-planning capacity, decision-making abilities and general experience (Ovaskainen et al. 2004, Röser et al. 2012, Mola-Yudego et al. 2015).

144

Comminution is often performed with large, powerful machines, capable of generating much noise. In turn, high noise levels may have negative impacts on the health and comfort of workers, and of the people living in the surroundings of a wood fuel yard (Kühmaier et al. 2014). A study of Spinelli et al. (2016a) demonstrated that the chipper generated more noise than the grinder, due to its better ability to process wood and to transmit more energy into it. Since the chipper was equipped with less working tools and turned slower than the grinder, it generated its noise peaks at lower frequency bands. Nuutinen et al. (2015) and Poje et al. (2015) recorded an average noise level of a grinder of about 82 dB(A), which is more than the limit level of 80 dB(A). The main source of noise was the powerful diesel engine, followed by the chipper drum: they generated the highest noise levels in the 100–200 Hz and the 20–50 Hz frequency ranges, respectively. Tractor-trailer chippers have higher noise levels than truck-mounted chippers (Rottensteiner et al. 2013). During chipping, machine operators are exposed to whole-body vibration bearing a risk to health (Rottensteiner et al. 2013). Truck-mounted chippers have higher vibration values than tractor-trailer chippers. The highest vibration levels were recorded while driving on the forest road and the second highest during chipping. Chipping hardwood produced higher vibration magnitudes than chipping softwood. Nevertheless, the exposure limit values set by the EU were usually not exceeded. The International Agency for Research on Cancer (IARC) has classified hardwood dust as a human carcinogen (IARC 1995). Magagnotti et al. (2013) determined the exposure of chipper operators to inhalable wood dust. Exposure to dust varied widely with wood conditions and machine productivity, and only occasionally exceeded the occupational exposure limit of 5 mg/m 3. Operators working inside a cab were three times less exposed than operators working outside.

3.2 Transportation and logistics 3.2.1 Selecting suitable transportation modes Manzone and Balsari (2015) compared tractor + trailer and trucks in terms of working time, working rate, fuel consumption, energy costs and economic costs. In Northern Scotland, forest chips can be delivered starting from approximately 20 €/MWh within a 50 km transportation distance when chipping is at roadside. If the transportation distance is 100 km, wood chips could be delivered at approximately 23 €/MWh (Röser et al. 2011). Time studies related to on-road transportation were made by Laitila et al. (2009); they Croat. j. for. eng. 39(2018)1

Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten ... (139–152) M. Kühmaier and G. Erber

compared the terminal handling time. The mean loading and unloading time of bundles per truck load was 46% higher compared to that of conventional 5 m pulpwood. Vainio et al. (2009) also developed transportation costs to a very detailed level. Fuelwood is usually transported by trucks but there is also a growing medium or long distance transportation of energy wood by railways or waterways (Tahvanainen and Anttila 2011, Wolfsmayr et al. 2016). According to a study of Tahvanainen and Anttila (2011), for distances shorter than 60 km, truck transportation of loose residues and end-facility comminution was the most costcompetitive chain. Over longer distances, roadside chipping with chip truck transportation was the most cost-efficient option. When the transportation distance went from 135 to 165 km, train-based transportation offered the lowest costs. The most cost-competitive alternative for long-distance transport included a combination of roadside chipping, truck transportation to the terminal and train transportation to the plant. A MILP model was developed by Rauch and Gronalt (2011), comprising decisions on modes of transportation and spatial arrangement of terminals. An increase of energy costs results in a procurement cost increase. While domestic waterways become more important because of the energy cost increase, rail only does so. One way to decrease procurement costs would be to reduce the share of empty trips with truck and trailer. Routing influences the modal split considerably, and the truck transport share increases from 86% to 97%, accordingly. Increasing forest fuel imports by large CHPs lowers domestic competition and enables smaller plants to cut their procurement costs. Sukhanov et al. (2013) came to the conclusion that in Northwest Russia the collection of logging residues for chipping is cost-effective if the distance to the customer is less than 100 km. The use of logs for the production of forest chips is economically more feasible, compared with the use of residues. In this case, forest chips can be transported up to 150 km (Gerasimov and Karjalainen 2013). The cost-efficiency of waterway transportation operations related to forest chips in Finland’s Lake Saimaa region was studied by Karttunen et al. (2012) using practical demonstrations and discrete-event simulation. The waterway supply chain of forest chips was cost-competitive to road transport by truck after 100–150 km. 3.2.2 Improving the efficiency of fuelwood transportation The factors influencing transporting efficiency and cost are payload, loading and unloading time, transporting distance, hourly costs and operational delays, such as waiting and auxiliary times. Transportation of Croat. j. for. eng. 39(2018)1

Table 2 Comparison of maximum vehicle gross weights in European countries Weight limits, tonnes

Countries (ISO code)

AZ, BY

AT, BA, BG, CH, DE, ES, FR, HR, LT, LV, MD ME, MK, MT, PL, PT, RO, RS, SI, SK, TR, UA

AL, BE, EE, GB, GE, IE, IS, IT, LU, RU

FI, NO, SE

logs is usually the most efficient transport system, causing less cost than transporting bundles and residues (Bergström and Di Fulvio 2014). Maximum permitted weights of truck transport differ in Europe with a range of 31 to 60 tonnes (Table 2) IRU (2017), ITF (2017). The presented limits can be higher for restricted road networks, the first or last leg of transport, intermodal/combined/container transport or for the transport of special goods (e.g wood). In Finland, the laws providing the physical dimensions of freight transport vehicles were changed and the new legislation enables higher gross weights as well as larger load capacities. Laitila et al. (2016) dealt with the determination of the most cost-effective vehicle type. The 69-tonne truck-trailer was a feasible choice when the payload was not limited by the bulk weight of the forest industry by-products. With heavier forest industry by-products, such as sawmill wood chips and bark, the 76-tonne truck-trailer was the most feasible choice. The results showed clearly that the transporting costs associated with using the new type truck-trailers were lower than those for conventional 60-tonne truck-trailers in all assortments. Increased use of forest biomass for energy and rising transportation costs are forcing biomass suppliers towards better moisture content management (Pasila 2013, Routa et al. 2016, Sosa et al. 2015). Kühmaier et al. (2016) found that the transport of wood chips causes a cost advantage of 0.32–0.49 €/m3 loose if the moisture content can be reduced from 55 to 35%. Erber et al. (2016) has shown that moisture content management increased truckload volume utilisation by 25%

145

M. Kühmaier and G. Erber Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten ... (139–152)

and transported calorific value by 48%. The number of truck trips for transportation was reduced by 20%. Stump parts are bulky and it is impossible to achieve full tonnages on trucks and trailers even though the load space is completely full. Grinding the stumps at the landing and sieving of the produced chips has the potential to increase load weights and reduce both the amount of contaminants and the transport costs. Transport payloads increase substantially, but according to a case study of Anerud et al. (2016), a transport distance of 110 km is needed before the coarse grinding system provided lower cost than the standard system with transports of stump parts and grinding at the heating plant. 3.2.3 Coordination of supply processes Forest fuel has to be converted into chips before delivery to the customer and the demand for forest fuel varies over the year depending on temperature. To balance the chipping and transportation capacities over time, it is important to manage inventory levels at terminals. An optimisation model developed by Flisberg et al. (2012) provides decision support for questions regarding the choice of technology for chipping, where to perform the chipping operations, and the allocation of different assortments to heating plants. It is important to maximise the proportion of effective work time in relation to scheduled work time. Currently, the effective work time is often less than 50 per cent of scheduled work time, due to chip transports using the chipper, waiting for chip trucks, and other delays. Increased chipper utilisation requires greater coordination between the chipper and the chip trucks transporting the produced chips to the customer. Eliasson et al. (2015b) have simulated supply systems to examine how transport distance, number of trucks, shift scheduling and chip buffers affect the system costs for a high-performance chipper system. Spinelli et al. (2014b) compared chipping operations of residues at the roadside landing or at the yarder pad, the latter inaccessible to heavy road vehicles, by carefully reflecting interaction delays between individual units along the chain. Chipping at the pad with a chipper and two shuttles was the best compromise solution of low supply cost and fuel consumption. Machine activity based controlling offers a new way to increase efficiency and productivity. A study of Holzleitner et al. (2013) aims to monitor the forest fuel supply processes via fleet management equipment. Large data sets were automatically and efficiently gathered with little effort by drivers and operators. Data management was conducted in a pre-configured database that contained pre-defined reports.

146

3.2.4 Economic assessment of supply chains Several studies have analysed the efficiency and costs of supply chains for different operating conditions. Valente et al. (2014) made a comparative analysis of Norwegian and Italian supply chains. In Norway, the supply chain is more mechanised than in Italy, which explains the higher productivity. Laitila (2008) calculated the procurement costs of six whole tree chips supply chains. The harvesting system based on the harvester with an accumulating felling head was the cheapest, while the harwarder system was the most expensive. The procurement costs of fuel chips made from delimbed stemwood and whole trees for another case study in Finland (Laitila et al. 2010) were 49.1 and 41.8 €/m3, respectively. Belbo and Talbot (2014) and Gustavsson et al. (2011) analysed widely applied supply chains producing forest fuel from whole trees from energy thinnings. Results showed that the most expensive chain (roadside bundling, roadside storage, terminal storage and delivery using timber truck) was 23% more costly than the cheapest chain (roadside chipping and direct transport to conversion plant with container truck). In economic terms, the transport of forest residues by truck and trailer presents the highest cost followed by chipping and processing of trees. These three operations are responsible for approximately 80% of the total costs (Ferreira et al. 2014). Cost-supply curves can be used to support the planning of heating plant investments (Anttila et al. 2011). Bergström and Di Fulvio (2014) analysed the effect of future harvesting and handling technologies on the cost and energy efficiency of supply chains. If boom-corridor thinning technologies, optimised bundle-harvesters and load-compression devices are developed, costs are reduced on average by 12–27%, and 11–30% less energy is required when compared with current systems. Linear programming models have been developed to study the logistics and determine the best setup for bioenergy chains (e.g. Kanzian et al. 2009, VelazquezMarti and Fernandez-Gonzalez 2010, Van Dyken et al. 2010, Kanzian et al. 2013). The focus of the work of Van Dyken et al. (2010) was to represent the relationship between moisture and energy content of different kinds of biomass and to handle long-term processes in the optimisation like passive drying effects. A spatial decision support tool based on LP (Linear Programming) developed by Sosa et al. (2015) uses drying curves to assess the moisture content, weight and energy content of biomass. The model helps solve the best spatial allocation for the demanded wood products by optimising the number of trucks required to Croat. j. for. eng. 39(2018)1

Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten ... (139–152) M. Kühmaier and G. Erber

satisfy the demand at power and mill plants whilst still applying the volume and payload weight constraints. Gronalt and Rauch (2007) evaluated different supply lines for the woody biomass from forest to plants by calculating the system cost for a number of alternative configurations. Especially, they compared central chipping against a local approach. A partial equilibrium model for the forest chips market in Finland was developed and demonstrated by Kallio et al. (2011). Since the supply of stumps and logging residues is tied to roundwood harvests, reaching the target seems unrealistic without investments in the new production capacity of the forest industry. 3.2.5 Environmental assessment of supply chains Kühmaier and Stampfer (2012) developed a computer-based decision support tool to assist stakeholders in identifying the most suitable fuelwood supply chain. The tool considers a number of criteria, such as energy efficiency, nutrient balance, stability and vitality of the remaining stand and soil, contribution margin, supply guarantee, employment rate and working safety. Kanzian et al. (2013, 2016) have formulated a multi-criteria optimisation problem (MOP), whereby the profit must be maximised and the CO2 emissions have to be minimised. In an effort to minimise CO2 emissions, 30% of the woody biomass should be delivered chipped from the terminals and more than 50% chipped directly from forest. By changing the weight to maximise the profit, CO2 emissions will only increase by 4.5%, whereas the profit more than doubles. Murphy et al. (2016) developed an optimisation model, which ensures minimal GHG emissions. Based on forest growth simulations, a set of realistic forest biomass supply chains for Bavarian forestry conditions were modelled (Klein et al. 2016). Total GHG emissions are estimated for the Bavarian forestry sector indicating a share of 0.41% in the total GHG emissions of Bavaria. Most decisive parameters are forest road maintenance, biomass harvesting, forwarding and biomass transport. Tree species, age class, wood assortment and site quality also notably influence GHG emissions. Life cycle assessment (LCA) was used in several studies (Whittaker et al. 2011, Murphy et al. 2016, Klein et al. 2016, De la Fuente et al. 2017) to evaluate biomass-to-energy systems to reduce environmental impacts during production and transportation. Typically used impact categories are global warming, acidification, eutrophication, and energy demand. Mechanised forest harvesting generates more greenhouse gas (GHG) emissions than motor-manual harvesting. The main sources of GHG emissions are truck transportation and chipping (Galezia 2013, Ferreira et al. Croat. j. for. eng. 39(2018)1

2014, De la Fuente et al. 2017). The lower bulk density of wood chips means that transportation energy requirements and GHG emissions are higher compared with round wood logs and brash bales, suggesting that chipping should occur near the end-user (Whittaker et al. 2011). CO2 emissions for supply chains have been calculated by several authors (e.g. Eriksson 2008, Whittaker et al. 2011, Valente et al. 2014, De la Fuente et al. 2017). Although forest roads were constructed, their relative contributions to the overall energy requirements and GHG emissions are small (Whittaker et al. 2011). The ratio between the extra fossil energy input to harvest biomass on the one hand and the possible energy output from the wood chips on the other hand varied between 0.7% and 1.3% (Vangansbeke et al. 2015).

4. Discussion and conclusion Strategies and measures to increase the efficiency of fuelwood supply systems have been a topic of research ever since the development of industrial forest fuelwood supply systems started. The focus and trends have undergone changes over time but the challenges remained more or less the same. The coordination of fuelwood supply processes is complex, the costs for harvesting are high and the revenues from forest biomass are low. In mountainous areas, limited space on forest roads and strong limitation in available transport routes are additional challenges. When introducing industrial fuelwood supply, the major research topics were the assessment of ecological risks, the coordination of individual processes and improvement of transport efficiency. Stampfer and Kanzian (2006) already mentioned that the position of the chipper within the work chain influences the efficiency of the supply chain and determines the type of biomass that will be transported. Direct chipping of the material into trucks results in operational delays in the order of 20%. If chipping and transportation is carried out independently, then additional costs occur for chip loading. Röser et al. (2012) emphasised that more consideration has to be given to the close interlinkage between the chipper, crane and grapple in the future. The optimal positioning of the comminution process within the supply chain is still an important research question to reduce delays caused by a lack of coordination between expensive machines. It was proven by several authors that the characteristics of the raw material have a strong effect on chipper performance and product quality. The most significant parameters are tree parts (logs, residues, bundles etc.),

147

M. Kühmaier and G. Erber Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten ... (139–152)

volume of the material to be chipped, tree species and moisture content. One of the latest research trends was the improvement of chipper design with a special focus on knifes as well as screens and sieves configuration. The machine type on which the chipper is mounted, the grapple load of the crane as well as in-feed and evacuation systems have an effect on the overall productivity. Di Fulvio et al. (2015) have tested hybrid systems, which store excess energy from a diesel engine during periods of low loading for use during peak loading times. Knife configuration is more productive than a hammer configuration. The impact of knife wear, knife angle, moving speed, drum spinning speed on productivity and product quality has been assessed by several studies during the last 5 years. It was probably the most investigated research trend in the recent past. For optimal combustion, the fuel should have a low content of fine particles (Kons et al. 2015), which can be achieved by screening or sieving. The use of narrower screens improves the product quality but decreases productivity and increases fuel consumption. Therefore, recommendations for optimal screen sizes should be defined for different areas of application and quality requirements. Noise and vibration have been identified as the most relevant parameters influencing the human health during chipping operations. The information produced in these studies is important because it raises awareness about the higher noise impact of a chipper, so that appropriate countermeasures can be taken (additional distance, noise-attenuation and shielding measures, etc.) (Spinelli et al. 2016b). Due to spatial distribution, low mass density, low energy density and low bulk density, the transportation of forest fuel is crucial for economic and ecological efficiency (Wolfsmayr and Rauch 2014). The question of efficient production of wood chips is tightly connected to the reduction of transportation costs. In the past, this problem was overcome by keeping transportation distances very short. With the recent boom in bio-energy and construction of larger bio-energy power plants, the total woody fuel needs within a region have increased, and the required supply region has become larger. Transportation distances and costs have increased accordingly (Asikainen et al. 2001). Transportation by railways or waterways (Tahvanainen and Anttila 2011, Wolfsmayr et al. 2016) has become a valuable alternative. Intermodal transport, however, has not been studied in the past and, therefore, future research requirements have been identified (Wolfsmayr and Rauch 2014). The best possible utilisation of the vehicle capacity will become a key factor for remaining competitive. The capacity for trucks can be im-

148

proved through higher payloads, increased on-truck load density, reduced loading and unloading time, smaller transporting distance, as well decreased hourly costs and operational delays (Stampfer and Kanzian 2006). Operational delays can be minimised by balancing the chipping and transportation capacities and inventory levels at landings and terminals. Linear programming models have been developed to study the logistics and determine the best setup for forest fuel chains (e.g. Kanzian et al. 2009, Velazquez-Marti and Fernandez-Gonzalez 2010, Van Dyken et al. 2010, Kanzian et al. 2013). In recent years, the assessment of ecological impact is becoming more and more important. Life cycle assessment (LCA) was used in several studies (Whittaker et al. 2011, Murphy et al. 2016, Klein et al. 2016, De la Fuente et al. 2017) to evaluate biomass-toenergy systems to reduce environmental impacts. It is expected that this topic will also become one of the main future research trends because of recently defined climate protection targets. The focus is not only climate change but also water and soil protection, which will probably become more important.

5. References Abdallah, R., Auchet, S., Méausoone, P. J., 2014: A dynamic measurement of a disc chipper cutting forces. Biomass and Bioenergy 64: 269–275. Anerud, E., von Hofsten, H., Eliasson, L., 2016: An alternative supply system for stump biomass – coarse grinding combined with sieving of the produced hog fuel. International Journal of Forest Engineering 27(2): 109–114. Anttila, P., Asikainen, A., Laitila, J., Broto, M., Campanero, I., Lizarralde, I., Rodriguez, F., 2011: Potential and supply costs of wood chips from forests in Soria, Spain. Forest Systems 20(2): 245–254. Asikainen, A., Ranta, T., Laitila, J., 2001: Large-scale forest fuel procurement. In: Pelkonen, P., Hakkila, P., Karjalainen, T., Schlamadinger, B. Woody Biomass as an Energy Source – Challenges in Europe, European Forest Institute (EFI): 73–78. Assirelli, A., Civitarese, V., Fanigliulo, R., Pari, L., Pochi, D., Santangelo, E., Spinelli, R., 2013: Effect of piece size and tree part on chipper performance. Biomass and Bioenergy 54: 77–82. Belbo, H., Talbot, B., 2014: Systems analysis of ten supply chains for whole tree chips. Forests 5(9): 2084–2105. Bergström, D., Di Fulvio, F., 2014: Comparison of the cost and energy efficiencies of present and future biomass supply systems for young dense forests. Scandinavian Journal of Forest Research 29(8): 793–812. COM/2014/015: Communication from the commission to the European parliament, the council, the European economic Croat. j. for. eng. 39(2018)1

Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten ... (139–152) M. Kühmaier and G. Erber and social committee and the committee of the regions. A policy framework for climate and energy in the period from 2020 to 2030.

Facello, A., Cavallo, E., Spinelli, R., 2013b: Chipping machines: Disc and drum energy requirements. Journal of Agricultural Engineering 44(2): 378–380.

De la Fuente, T., González-García, S., Athanassiadis, D., Nordfjell, T., 2017: Fuel consumption and GHG emissions of forest biomass supply chains in Northern Sweden: a comparison analysis between integrated and conventional supply chains. Scandinavian Journal of Forest Research 32(7): 568–581.

Facello, A., Cavallo, E., Magagnotti, N., Paletto, G., Spinelli, R., 2013c: The effect of chipper cut length on wood fuel processing performance. Fuel Processing Technology 116: 228– 233.

De Meyer, A., Cattrysse, D., Rasinmäki, J., Van Orshoven, J., 2014: Methods to optimise the design and management of biomass-for-bioenergy supply chains: A review. Renewable and Sustainable Energy Reviews 31: 657–670. Díaz-Yáñez, O., Mola-Yudego, B., Anttila, P., Röser, D., Asikainen, A., 2013: Forest chips for energy in Europe: Current procurement methods and potentials. Renewable and Sustainable Energy Reviews 21: 562–571. Di Fulvio, F., Eriksson, G., Bergström, D., 2015: Effects of wood properties and chipping length on the operational efficiency of a 30 kw electric disc chipper. Croatian Journal of Forest Engineering 36(1): 85–100. Eisenlauer, M., Teipel, U., 2016: Comminution of woody renewable resources. Chemie-Ingenieur-Technik 88(7): 948– 957. Eliasson, L., von Hofsten, H., Johannesson, T., Spinelli, R., Thierfelder, T., 2015a: Effects of sieve size on chipper productivity, fuel consumption and chip size distribution for open drum chippers. Croatian Journal of Forest Engineering 36(1): 11–18. Eliasson, L., Eriksson, A., Mohtashami, S., 2015b: Analysis of factors affecting productivity and costs for a high-performance chip supply system. Applied Energy 185: 497–505. Erber, G., Kanzian, C., Stampfer, K., 2016: Modelling natural drying of European beech (Fagus sylvatica L.) logs for energy based on meteorological data. Scandinavian Journal of Forest Research 31(3): 294–301. Erber, G., Kühmaier, M., 2017: Research Trends in European Forest Fuel Supply Chains: a Review of the Last Ten Years (2007–2017) – Part one: Harvesting and Storage. Croatian Journal of Forest Engineering 38(2): 269–278. Eriksson, L. N., 2008: Comparative analyses of forest fuels in a life cycle perspective with a focus on transport systems. Resources, Conservation and Recycling 52(10): 1190–1197. Eriksson, G., Bergström, D., Nordfjell, T., 2013: The state of the art in woody biomass comminution and sorting in Northern Europe. International Journal of Forest Engineering 24(3): 194–215. Eskandarpour, M., Dejax, P., Miemczyk, J., Péton, O., 2015: Sustainable supply chain network design: An optimizationoriented review. Omega 54: 11–32. Facello, A., Cavallo, E., Magagnotti, N., Paletto, G., Spinelli, R., 2013a: The effect of knife wear on chip quality and processing cost of chestnut and locust fuel wood. Biomass and Bioenergy 59: 468–476. Croat. j. for. eng. 39(2018)1

Ferreira, J. V., Viana, H., Esteves, B., Cruz Lopes, L. P., Domingos, I., 2014: Life cycle assessment of residual forestry biomass chips at a power plant: A Portuguese case study. International Journal of Energy and Environmental Engineering 5(2–3): 1–7. Flisberg, P., Frisk, M., Rönnqvist, M., 2012: FuelOpt: A decision support system for forest fuel logistics. Journal of the Operational Research Society 63(11): 1600–1612. Gałezia, T., 2013: Analysis of selected methods effectiveness of wood biomass harvesting for energy purposes. Silva Balcanica 14(1): 94–101. Gerasimov, Y., Karjalainen, T., 2013: Energy wood resources availability and delivery cost in Northwest Russia. Scandinavian Journal of Forest Research 28(7): 689–700. Ghaffariyan, M. R., Brown, M., Acuna, M., Sessions, J., Gallagher, T., Kühmaier, M., Spinelli, R., Visser, R., Devlin, G., Eliasson, L., Laitila, J., Laina, R., Iwarsson Wide, M., Egnell, G., 2017: An international review of the most productive and cost effective forest biomass recovery technologies and supply chains. Renewable and Sustainable Energy Reviews 74: 145–158. Gold, S., Seuring, S., 2011: Supply chain and logistics issues of bio-energy production. Journal of Cleaner Production 19(1): 32–42. Gronalt, M., Rauch, P., 2007: Designing a regional forest fuel supply network. Biomass and Bioenergy 31(6): 393–402. Gustavsson, L., Eriksson, L., Sathre, R., 2011: Costs and CO2 benefits of recovering, refining and transporting logging residues for fossil fuel replacement. Applied Energy 88(1): 192–197. Holzleitner, F., Kanzian, C., Höller, N., 2013: Monitoring the chipping and transportation of wood fuels with a fleet management system. Silva Fennica 47(1): 1–11. IARC 1995: IARC monographs on the evaluation of carcinogenic risks to human. Wood dust and formaldehyde. Vol. 62. Lyon: IARC. IRU, 2017: Weights and dimensions good transport. Retrieved from www.iru.org/resources/tools-apps/infocentre [15/12/2017]. Isaksson, P., Gradin, P. A., Hellström, L. M., 2013: A numerical and experimental study regarding the influence of some process parameters on the damage state in wood chips. Holzforschung 67(6): 691–696. ITF, 2017: Permissible Maximum Weights of Lorries in Europe. Retrieved from www.itf-oecd.org/node/19217 [15/12/2017]. Kallio, A. M. I., Anttila, P., McCormick, M., Asikainen, A., 2011: Are the Finnish targets for the energy use of forest

149

M. Kühmaier and G. Erber Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten ... (139–152) chips realistic-Assessment with a spatial market model. Journal of Forest Economics 17(2): 110–126.

ity – a German case study. International Journal of Forest Engineering 26(1): 60–70.

Kanzian, C., Holzleitner, F., Stampfer, K., Ashton, S., 2009: Regional energy wood logistics - Optimizing local fuel supply. Silva Fennica 43(1): 113–128.

Laitila, J., 2008: Harvesting technology and the cost of fuel chips from early thinnings. Silva Fennica 42(2): 267–283.

Kanzian, C., Kühmaier, M., Zazgornik, J., Stampfer, K., 2013: Design of forest energy supply networks using multi-objective optimization. Biomass and Bioenergy 58: 294–302.

Laitila, J., Nuutinen, Y., 2015: Efficiency of integrated grinding and screening of stump wood for fuel at roadside landing with a low-speed double-shaft grinder and a star screen. Croatian Journal of Forest Engineering 36(1): 19–32.

Kanzian, C., Kühmaier, M., Erber, G., 2016: Effects of moisture content on supply costs and CO2 emissions for an optimized energy wood supply network. Croatian Journal of Forest Engineering 37(1): 51–60.

Laitila, J., Routa, J., 2015: Performance of a small and a medium sized professional chippers and the impact of storage time on scots pine (Pinus sylvestris) stem wood chips characteristics. Silva Fennica 49(5): 1–19.

Kärhä, K., 2011: Industrial supply chains and production machinery of forest chips in Finland. Biomass and Bioenergy 35(8): 3404–3413. Karttunen, K., Väätäinen, K., Asikainen, A., Ranta, T., 2012: The operational efficiency of waterway transport of forest chips on Finland’s lake Saimaa. Silva Fennica 46(3): 395–413. Kent, T., Kofman, P., Coates, E., 2011: Harvesting wood for energy. Cost-effective woodfuel supply chains in Irish forestry. COFORD Report: 98 p. Klein, D., Wolf, C., Schulz, C., Weber-Blaschke, G., 2016: Environmental impacts of various biomass supply chains for the provision of raw wood in Bavaria, Germany, with focus on climate change. Science of the Total Environment 539: 45–60. Kons, K., Bergström, D., Di Fulvio, F., 2015: Effects of sieve size and assortment on wood fuel quality during chipping operations. International Journal of Forest Engineering 26(2): 114–123. Kováč, J., Krilek, J., Mikleš, M., 2011: Energy consumption of a chipper coupled to a universal wheel skidder in the process of chipping wood. Journal of Forest Science 57(1): 34–40. Krajnc, M., Dolšak, B., 2014: The influence of drum chipper configuration on the quality of wood chips. Biomass and Bioenergy 64: 133–139. Kühmaier, M., Kanzian, C., Stampfer, K., 2014: Identification of potential energy wood terminal locations using a spatial multicriteria decision analysis. Biomass and Bioenergy 66: 337–347. Kühmaier, M., Stampfer, K., 2012: Development of a multicriteria decision support tool for energy wood supply management. Croatian Journal of Forest Engineering 33(2): 181–198.

Laitila, J., Niemistö, P., Väätäinen, K., 2016: Productivity of multi-tree cutting in thinnings and clear cuttings of young downy birch (Betula pubescens) dominated stands in the integrated harvesting of pulpwood and energy wood. Baltic Forestry 22(2): 116–131. Laitila, J., Heikkilä, J., Anttila, P., 2010: Harvesting alternatives, accumulation and procurement cost of small-diameter thinning wood for fuel in Central Finland. Silva Fennica 44(3): 465–480. Laitila, J., Kärhä, K., Jylhä, P., 2009: Time consumption models and parameters for off- and on-road transportation of whole-tree bundles. Baltic Forestry 15(1): 105–114. Magagnotti, N., Nannicini, C., Sciarra, G., Spinelli, R., Volpi, D., 2013: Determining the exposure of chipper operators to inhalable wood dust. Annals of Occupational Hygiene 57(6): 784–792. Manzone, M., Balsari, P., 2015: The energy consumption and economic costs of different vehicles used in transporting woodchips. Fuel 139: 511–515. Marchi, E., Magagnotti, N., Berretti, L., Neri, F., Spinelli, R., 2011: Comparing terrain and roadside chipping in mediterranean pine salvage cuts. Croatian Journal of Forest Engineering 32(2): 587–598. Mizaras, S., Sadauskiene, L., Mizaraite, D., 2011: Cost and profitability of biofuel chipping in Alnus incana stands in Lithuania. Scandinavian Journal of Forest Research 26(2): 154–160. Mola-Yudego, B., Picchi, G., Röser, D., Spinelli, R., 2015: Assessing chipper productivity and operator effects in forest biomass operations. Silva Fennica 49(5): 1–14. Murphy, F., Sosa, A., McDonnell, K., Devlin, G., 2016: Life cycle assessment of biomass-to-energy systems in Ireland modelled with biomass supply chain optimisation based on greenhouse gas emission reduction. Energy 109: 1040–1055.

Kühmaier, M., Erber, G., Kanzian, C., Holzleitner, F., Stampfer, K., 2016: Comparison of costs of different terminal layouts for fuel wood storage. Renewable Energy 87: 544–551.

Nati, C., Magagnotti, N., Spinelli, R., 2015: The improvement of hog fuel by removing fines, using a trommel screen. Biomass and Bioenergy 75: 155–160.

Kuljich, S., Hernández, R. E., Blais, C., 2015: Effects of cutterhead diameter and log infeed position on energy requirements of a chipper-canter. Wood and Fiber Science 47(4): 399–409.

Nati, C., Spinelli, R., Fabbri, P., 2010: Wood chips size distribution in relation to blade wear and screen use. Biomass and Bioenergy 34(5): 583–587.

Kuptz, D., Hartmann, H., 2015: The effect of raw material and machine setting on chipping performance and fuel qual-

150

Nuutinen, Y., Petty, A., Bergström, D., Rytkönen, M., Di Fulvio, F., Tiihonen, I., Lauren, A., Dahlin, B., 2016: Quality and productivity in comminution of small-diameter tree bunCroat. j. for. eng. 39(2018)1

Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten ... (139–152) M. Kühmaier and G. Erber dles. International Journal of Forest Engineering 27(3): 179– 187.

tion of Ireland’s transport supply strategy to bioenergy markets and competing industries. Energy 86: 354–368.

Nuutinen, Y., Laitila, J., Rytkönen, E., 2015: Grinding of stumps, logging residues and small diameter wood using a CBI 5800 grinder with a truck as a base machine. Baltic Forestry 20(1): 176–188.

Spinelli, R., Visser, R., 2009: Analyzing and estimating delays in wood chipping operations. Biomass and Bioenergy 33(3): 429–433.

Ovaskainen, H., Uusitalo, J., Väätäinen, K., 2004: Characteristics and significance of a harvester operators’ working technique in thinnings. International Journal of Forest Engineering 15(2): 67–77. Pasila, A., 2013: Changes, challenges and opportunities in the wood energy supply chain. Agronomy Research 11(2): 529–532. Patterson, D. W., Hartley, J. I., Pelkki, M. H., 2011: Size, moisture content, and British thermal unit value of processed in-woods residues: Five Case Studies. Forest Products Journal 61(4): 316–320. Picchi, G., Eliasson, L., 2015: Chip truck utilization for a container handling chipper truck when chipping logging residues and the effect of two grapple types on chipping efficiency. International Journal of Forest Engineering 26(3): 203–211. Pochi, D., Civitarese, V., Fanigliulo, R., Spinelli, R., Pari, L., 2015: Effect of poplar fuel wood storage on chipping performance. Fuel Processing Technology 134: 116–121.

Spinelli, R., Ivorra, L., Magagnotti, N., Picchi, G., 2011a: Performance of a mobile mechanical screen to improve the commercial quality of wood chips for energy. Bioresource Technology 102(15): 7366–7370. Spinelli, R., Magagnotti, N., Paletto, G., Preti, C., 2011b: Determining the impact of some wood characteristics on the performance of a mobile chipper. Silva Fennica 45(1): 85–95. Spinelli, R., Cavallo, E., Facello, A., Magagnotti, N., Nati, C., Paletto, G., 2012a: Performance and energy efficiency of alternative comminution principles: Chipping versus grinding. Scandinavian Journal of Forest Research 27(4): 393–400. Spinelli, R., Cavallo, E., Facello, A., 2012b: A new comminution device for high-quality chip production. Fuel Processing Technology 99: 69–74. Spinelli, R., Magagnotti, N., 2013: Performance of a smallscale chipper for professional rural contractors. Forest Science and Practice 15(3): 206–213. Spinelli, R., Magagnotti, N., 2014a: Determining long-term chipper usage, productivity and fuel consumption. Biomass and Bioenergy 66: 442–449.

Poje, A., Spinelli, R., Magagnotti, N., Mihelic, M., 2015: Exposure to noise in wood chipping operations under the conditions of agro-forestry. International Journal of Industrial Ergonomics 50: 151–157.

Spinelli, R., Magagnotti, N., 2014b: Using disposable chipper knives to decrease wood fuel processing cost. Fuel Processing Technology 126: 415–419.

Ranta, T., Rinne, S., 2006: The profitability of transporting uncomminuted raw materials in Finland. Biomass and Bioenergy 30(3): 231–237.

Spinelli, R., Glushkov, S., Markov, I., 2014a: Managing chipper knife wear to increase chip quality and reduce chipping cost. Biomass and Bioenergy 62: 117–122.

Rauch, P., Gronalt, M., 2011: The effects of rising energy costs and transportation mode mix on forest fuel procurement costs. Biomass and Bioenergy 35(1): 690–699. Röser, D., Sikanen, L., Asikainen, A., Parikka, H., Väätäinen, K., 2011: Productivity and cost of mechanized energy wood harvesting in Northern Scotland. Biomass and Bioenergy 35(11): 4570–4580. Röser, D., Mola-Yudego, B., Prinz, R., Emer, B., Sikanen, L., 2012: Chipping operations and efficiency in different operational environments. Silva Fennica 46(2): 275–286. Rottensteiner, C., Tsioras, P., Neumayer, H., Stampfer, K., 2013: Vibration and noise assessment of tractor-trailer and truck-mounted chippers. Silva Fennica 47(5): 1–14. Routa, J., Asikainen, A., Björheden, R., Laitila, J., Röser, D., 2013: Forest energy procurement: State of the art in Finland and Sweden. Wiley Interdisciplinary Reviews: Energy and Environment 2(6): 602–613. Routa, J., Kolström, M., Ruotsalainen, J., Sikanen, L., 2016: Validation of prediction models for estimating the moisture content of logging residues during storage. Biomass and Bioenergy 94: 85–93. Sosa, A., Acuna, M., McDonnell, K., Devlin, G., 2015: Managing the moisture content of wood biomass for the optimisaCroat. j. for. eng. 39(2018)1

Spinelli, R., Di Gironimo, G., Esposito, G., Magagnotti, N., 2014b: Alternative supply chains for logging residues under access constraints. Scandinavian Journal of Forest Research 29(3): 266–274. Spinelli, R., Cavallo, E., Eliasson, L., Facello, A., Magagnotti, N., 2015: The effect of drum design on chipper performance. Renewable Energy 81: 57–61. Spinelli, R., Eliasson, L., Magagnotti, N., 2016a: Increasing wood fuel processing efficiency by fine-tuning chipper settings. Fuel Processing Technology 151: 126–130. Spinelli, R., Magagnotti, N., Deboli, R., Preti, C., 2016b: Noise emissions in wood chipping yards: Options compared. Science of the Total Environment 563–564: 145–151. Stampfer, K., Kanzian, C., 2006: Current state and development possibilities of wood chip supply chains in Austria. Croatian Journal of Forest Engineering 27(2): 135–145. Sukhanov, Y., Seliverstov, A., Gerasimov, Y., 2013: Efficiency of forest chip supply systems in Northwest Russia. Advanced Materials Research 740: 799–804. Tahvanainen, T., Anttila, P., 2011: Supply chain cost analysis of long-distance transportation of energy wood in Finland. Biomass and Bioenergy 35(8): 3360–3375.

151

M. Kühmaier and G. Erber Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten ... (139–152) Vainio, P., Tokola, T., Palander, T., Kangas, A., 2009: A GISbased stand management system for estimating local energy wood supplies. Biomass and Bioenergy 33(9): 1278–1288. Valente, C., Spinelli, R., Hillring, B. G., Solberg, B., 2014: Mountain forest wood fuel supply chains: Comparative studies between Norway and Italy. Biomass and Bioenergy 71: 370–380. Van Dyken, S., Bakken, B. H., Skjelbred, H. I., 2010: Linear mixed-integer models for biomass supply chains with transport, storage and processing. Energy 35(3): 1338–1350. Vangansbeke, P., Osselaere, J., Van Dael, M., De Frenne, P., Gruwez, R., Pelkmans, L., Gorissen, L., Verheyen, K., 2015: Logging operations in pine stands in Belgium with additional harvest of woody biomass: Yield, economics, and energy balance. Canadian Journal of Forest Research 45(8): 987–997. Velazquez-Marti, B., Fernandez-Gonzalez, E., 2010: Mathematical algorithms to locate factories to transform biomass in bioenergy focused on logistic network construction. Renewable Energy 35(9): 2136–2142.

Wegener, J. K., Frerichs, L., Kemper, S., Sümening, F., 2015: Wood chipping with conical helical blades – Practical experiments concerning the impact of the infeed angle on the power requirement of a helical chipper. Biomass and Bioenergy 80: 173–178. Whittaker, C., Mortimer, N., Murphy, R., Matthews, R., 2011: Energy and greenhouse gas balance of the use of forest residues for bioenergy production in the UK. Biomass and Bioenergy 35(11): 4581–4594. Wolfsmayr, U. J., Rauch, P., 2014: The primary forest fuel supply chain: A literature review. Biomass and Bioenergy 60: 203–221. Wolfsmayr, U. J., Rauch, P., Gronalt, M., Merenda, R., Longo, F., 2016: Evaluating primary forest fuel rail terminals with discrete event simulation: A case study from Austria. Annals of Forest Research 59(1): 145–164. Yoshida, M., Berg, S., Sakurai, R., Sakai, H., 2016: Evaluation of chipping productivity with five different mobile chippers at different forest sites by a stochastic model. Croatian Journal of Forest Engineering 37(2): 309–318.

Authors’ addresses:

Received: March 07, 2017 Accepted: December 15, 2017

152

Martin Kühmaier, PhD. * e-mail: martin.kuehmaier@boku.ac.at Gernot Erber, PhD. e-mail: gernot.erber@boku.ac.at University of Natural Resources and Life Sciences Vienna Department of Forest and Soil Sciences Institute of Forest Engineering Peter Jordan Strasse 82 1190 Vienna AUSTRIA * Corresponding author Croat. j. for. eng. 39(2018)1

CONTENTS

Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten ... (139–152) M. Kühmaier and G. Erber

Original scientific papers Jennifer Norihiro, Pierre Ackerman, Ben D. Spong, Dirk Längin Productivity Model for Cut-to-Length Harvester Operation in South African Eucalyptus Pulpwood Plantations

Jukka Malinen, Janne Taskinen, Timo Tolppa Productivity of Cut-to-Length Harvesting by Operators’ Age and Experience

Kalle Kärhä, Asko Poikela, Teijo Palander Productivity and Costs of Harwarder Systems in Industrial Roundwood Thinnings

Arkadiusz Tomczak, Grzegorz Grodzin’ski, Marcin Jakubowski, Tomasz Jelonek, Witold Grzywin’ski Effects of Short-Term Storage Method on Moisture Loss and Weight Change in Beech Timber

Vladimir Petković, Igor Potočnik Planning Forest Road Network in Natural Forest Areas: a Case Study in Northern Bosnia and Herzegovina

Rodolfo Picchio, Farzam Tavankar, Rachele Venanzi, Angela Lo Monaco, Mehrdad Nikooy Study of Forest Road Effect on Tree Community and Stand Structure in Three Italian and Iranian Temperate Forests

Aleksey Ilintsev, Elena Nakvasina, Aleksey Aleynikov, Sergey Tretyakov, Sergey Koptev, Alexander Bogdanov Middle-Term Changes in Topsoils Properties on Skidding Trails and Cutting Strips after Long-Gradual Cutting: a Case Study in the Boreal Forest of the North-East of Russia

Anna Cudzik, Marek Brennensthul, Włodzimierz Białczyk, Jarosław Czarnecki Tractive Performance of Tyres in Forest Conditions – Impact Assessment of Ground and Tyres Parameters

Marco Manzone, Angela Calvo Trailer Overturning during Wood Transportation: an Experimental Investigation of Effects of Trailer Joint Point and Frame Structure

Aldo Calcante, Davide Facchinetti, Domenico Pessina Analysis of Hazardous Emissions of Hand-Operated Forestry Machines Fuelled with Standard Mix or Alkylate Gasoline

109

117

Srečko Glodež, Marko Šori, Krešimir Vučković, Stjepan Risović Determination of Service Life of Sintered Powder Metallurgy Gears in Regard to Tooth Bending Fatigue

129

Subject review Martin Kühmaier, Gernot Erber Research Trends in European Forest Fuel Supply Chains: A Review of the Last Ten Years (2007–2016) – Part Two: Comminution, Transport & Logistics

139

Croat. j. for. eng. 39(2018)1 153