Crojfe37 2 by zpandur

2016

Original scientific paper

Evaluation of a New Energy Recycling Hydraulic Lift Cylinder for Forwarders Jussi Manner, Ola Lindroos, Hans Arvidsson, Tomas Nordfjell Abstract In mechanized forestry, much of the work is conducted by use of cranes, and recovering potential energy is a possible method to reduce energy consumption when using cranes for lift work. The objective of this study was to evaluate the capacity of a new »Energy-efficient hydraulic lift cylinder« (EHLC), which has a secondary cylinder built into its piston rod, to store potential energy from lowering the boom in the form of pressurized hydraulic oil in an accumulator and using the stored energy in the next boom lift. The EHLC was mounted on a forwarder, and manipulated to enable its use also as a standard cylinder. We then compared the EHLC and a standard cylinder in terms of function and energy consumption during repetitive boom lifts and lowerings. With the tested settings the EHLC saved up to approximately 9.4% of the energy consumed during the first part of boom lifts and up to 3.2% of the total lift energy. With possible further adjustments, such as optimization of the accumulator size, enlargement of the assisting cylinder diameter, and enhancement of the accumulator pressurization, but most importantly reduction in internal leakage, the current EHLC could have commercial potential. Keywords: weight-balancing, fluid dynamics, fluid mechanics, timber loader, mobile hydraulic lift devices, Boyle’s law, counterweight

1. Introduction Increases in energy costs and environmental concerns have intensified efforts to improve energy efficiency recently, both generally and specifically in engineering research (e.g. European Union 2014, United Nations 2014). Notably, several recent studies have addressed possible methods to improve the productivity of cranes used in harvesters, planting machines and forwarders (Lindroos et al. 2008, Jundén et al. 2013, Laine and Rantala 2013, Ersson et al. 2014, Ortiz Morales et al. 2014). Cranes are used primarily for lift work, in many mobile and stationary applications, which involves relocating objects in such a manner that their potential energy changes. Consequently, recovering potential energy can probably be used to reduce the energy required for the work (e.g. Liang and Virvalo 2001a, Sun and Virvalo 2003, Rydberg 2005, Sun and Virvalo 2005, Virvalo and Sun 2005, Lin et al. 2010, Lin and Wang 2012, Minav et al. 2012, Noréus et al. 2013, Wang et al. 2013). Forwarder cranes are designed to provide large lifts and heights, partly at the Croat. j. for. eng. 37(2016)2

cost of slow horizontal movements (Malmberg 1981, Gerasimov and Siounev 1998, 2000, Virvalo and Sun 2005). Knuckleboom cranes are normally used on forwarders, and consist of a system of hydraulic cylinders and mechanical levers, i.e., a swivelling crane pillar, pivoting mid and outer booms, and an extension boom (Gerasimov and Siounev 2000). While boom is often used as a synonym for crane, that usage is avoided in this paper to avoid confusion between the system and its components (cf. Lindroos et al. 2008). The action of a vertical lift is mainly executed by an angular change in the joint between the crane pillar and mid-boom, which causes the outer and extension booms to rise. However, in this paper the outer and extension booms are treated as rigid parts of the midboom. Thus, boom is used hereafter as a collective term for the mid-, outer and extension booms, with the understanding that lifts are executed via action at the joint between the crane pillar and mid-boom. During forwarder work logs are collected in the forest and carried to roadside landings. Typically, for-

219

J. Manner et al.

Evaluation of a New Energy Recycling Hydraulic Lift Cylinder for Forwarders (219–231)

warder productivities range from 11 to 25 m3 per productive machine hour, and are negatively correlated with transportation distance, while positively correlated with sizes of both logs and loads (Eriksson and Lindroos 2014). Logs are lifted while loading and unloading the forwarder’s bunk. The total time required to forward a load (for which loading and unloading collectively account for about 60%) is typically 45 minutes (Manner et al. 2016). Typically, loading and unloading will require approximately 30 and 20 lifts with a full grapple, respectively (Manner et al. 2013, Manner et al. 2016). The boom on a standard forwarder crane is lifted with a single acting cylinder. Pressurized oil is directed into the cylinder, creating a force that causes the piston rod to lift the boom. To lower the boom, the pressurized oil from the cylinder is released into the non-pressurized reservoir, and no energy is recovered. However, there are several possible methods to save energy, for example, through weight-balancing, a common principle for recovering potential energy during load lowering for cranes and elevators. In a weightbalancing system, some of the potential energy is recovered and stored during load lowering, and then used to assist the next load lift. The potential energy recovery process creates a braking force that reduces the load lowering speed. Thus, the weight-balancing is a trade-off between the additional lift force and braking force. In a fully balanced system, ignoring energy losses through friction, it is theoretically possible to recover almost all of the potential energy, allowing a load to be lifted and lowered with minor energy input. Other examples of the weight-balancing technique for mobile devices are the use of counterweights or coil springs (e.g. Gawlik and Michałowski 2008, Deepak 2012, Lin et al. 2013). The few available energy recovering lift applications for forest machine cranes are typically based on the use of a hydro-pneumatic accumulator tank (accumulator), a common weight-balancing technique (e.g. Liang and Virvalo 2001a, Liang and Virvalo 2001b, Sun and Virvalo 2003, Sun and Virvalo 2005, Virvalo and Sun 2005). An accumulator consists of a vessel and bladder which separates an inert gas (e.g. nitrogen) from hydraulic oil. The flow of pressurized oil into the accumulator charges the accumulator as the sealed inert gas compresses according to Boyle’s law and, similarly, the flow of oil out from the accumulator discharges the accumulator and releases the stored energy (during boom lift and boom lowering, respectively, in crane work). Fast charging and discharging are some of the advantages of accumulators for energy storage (Hui and Junqing 2010, Minav et al.

220

2012, Van de Ven 2013). Moreover, accumulators reduce pressure spikes in the hydraulic system (e.g. Malmberg 1981, Ingvast 1989, Kim et al. 2013, Van de Ven 2013). Experiments show that 21–59% of potential, kinetic or rotational energy can be recovered by using accumulators (Zhang 2011, Ho and Ahn 2012). However, an energy recovery system based on an accumulator also has limitations. Notably, in most systems the accumulator pressure must exceed the pressure in the hydraulic circuit to enable reuse of the stored energy (cf. Einola 2013), but the accumulator pressure might decrease below this pressure due, for example, to internal leakage somewhere, which is likely to occur in all hydraulic systems (Manring 2005). Thus, leakage compensation to re-pressurize the accumulator is required. Another limitation is the low energy storage capacity in relation to their size (e.g. Van de Ven 2013).

2. Materials and methods 2.1 The energy-efficient hydraulic lift cylinder (EHLC) Technical principles and claims for a flawlessly functioning EHLC (Fig. 1) mounted on a forwarder

Fig. 1 Thordab AB’s patented »Energy-efficient hydraulic lift cylinder« (EHLC) with a pressure accumulator tank (accumulator), a movably arranged secondary piston (a) that divides the cylinder system into primary (b) and secondary cylinders (c). The secondary cylinder is built inside the primary piston rod (d) and connected to the accumulator. Load cell1 (p1) measured the pressure in the standard cylinder and primary cylinder of the EHLC. Load cell2 (p2) measured the secondary cylinder pressure and accumulator pressure in the EHLC Croat. j. for. eng. 37(2016)2

Evaluation of a New Energy Recycling Hydraulic Lift Cylinder for Forwarders (219–231)

Fig. 2 Panel 1: a standard cylinder. Panel 2: EHLC. In both panels: a is a movably arranged secondary piston, b is the primary cylinder, c is the secondary cylinder and d is the primary piston rod. The standard piston area (Asp) is the sum of primary piston area (A1) and secondary piston area (A2) crane are described below based on the manufacturer’s information (cf. WIPO Patent WO/2011/075034). In the starting position, the secondary piston (a) is inside the primary piston rod (d), basically forming a standard cylinder (Fig. 2: panel 1). Immediately after the machine has started and during the first boom lift the pressure in the hydraulic circuit of the accumulator will be the same as in the circuit of the standard cylinder. However, during boom lowering the accumulator is further pressurized, the check-valve prevents an outflow and the standard cylinder turns into primary (b) and secondary (c) cylinders (Fig. 1 and Fig. 2: panel 2) Croat. j. for. eng. 37(2016)2

J. Manner et al.

due to the pressure differences. As a result, the secondary piston will move in the direction of the lower pressure at any given time, since both of the secondary piston ends have the same area (Figs. 1–2). Hence, the following lift work will be conducted with assistance from the accumulator, provided that the pressure in the accumulator and secondary cylinder (p2) is higher than in the primary cylinder (p1). If not, the cylinder will function as a conventional cylinder. Given that p2 is higher than p1, the product of p2 and the secondary piston area (A2) create a secondary cylinder force (F2) (see details in Figs. 1–2). As the secondary piston thrusts the primary cylinder head continuously with a F2 that depends on p2, it creates an assisting force during boom lifts and a braking force during boom lowering. During boom lifting, the accumulator discharges and decreases p2, while a boom lowering charges the accumulator and p2 increases (Fig. 3). Thus, F2 decreases the EHLC's need for external energy input. Some losses in p2 are likely to occur due, for instance, to oil leakage from the secondary to the primary cylinder. If p1 exceeds p2 the check-valve opens and the accumulator will be charged from the hydraulic circuit system (Fig. 1). Therefore, a higher pressure must be maintained in the accumulator and secondary cylinder than in the primary cylinder to get an assisting force during a boom lift. Occasional pressure spikes in the hydraulic circuit pass the check-valve and load the accumulator, to 25–30 MPa according to the patent. Thus, occasional pressure spikes that always appear in ordinary hydraulic systems (Manring 2005), and hence occasional short-term check-valve openings are essential contributions to the EHLC's functionality because they maintain a higher p2 than p1 during boom lifts, and also provide leakage compensation (Fig. 3). In addition, the secondary cylinder work (W2) increases with increasing p2, thereby decreasing the need for an external energy input. Thus, check-valve openings caused by occasional pressure spikes should not be confounded with »malfunctions« of the EHLC, i.e. regularly open check-valve. The standard and secondary piston diameters of the studied EHLC cylinder were 124.5 mm and 32.0 mm, respectively (Roger Gustavsson, Thordab AB). This provides a standard piston area (Asp) of 12,174 mm2, A2 of 804 mm2, and a primary piston area (A1) of 11,370 mm2 (Fig. 2). The accumulator volume was 1.0 litre (Roger Gustavsson, Thordab AB). Based on these specifications, the theoretical energy saving is 6.6–6.9%, given that A2/Asp≈0.066, and p2 is in the range of 1.00 to 1.05 times p1 during the entire boom lift.

221

J. Manner et al.

Evaluation of a New Energy Recycling Hydraulic Lift Cylinder for Forwarders (219–231)

Fig. 3 Schematic visualization of flawless work by EHLC. A lift starts by pressurizing the primary cylinder (a), and the boom starts to lift (b) as the primary cylinder pressure ( p1) reaches a required threshold. The secondary cylinder pressure ( p2) decreases (c) until the end of boom lift (d) as the increasing secondary cylinder volume discharges the accumulator (Fig. 1). During the boom lowering (e) p2 increases (f) until the end of boom lowering (g) as the decreasing secondary cylinder volume recharges the accumulator (h). p2 creates an assisting lift force, which is inversely proportional to the lift cylinder length. There are risks of p2 and p1 intersecting without a pressure surplus from the occasional pressure spikes. However, no occasional pressure spike is shown in the figure, only the pressure surplus

2.2 Test and measure procedure An EHLC was mounted on a standard Gremo 1050 F forwarder (engine power 120 kW, hydraulic working pressure 23.5 MPa) equipped with a Cranab FC80 crane (boom mass 890 kg), a standard rotator (56 kg), and a Cranab CR280 grapple (200 kg). The experiment took place between the 11th and 13th of July 2011 in Umeå, Northern Sweden. The mid and extending booms were mechanically blocked in positions that gave a constant crane reach of 5.7 m. During the experiment, the boom was lifted and lowered by actuating the lift cylinder directional control valve using a joystick. Three settings for the valve response to the joystick actuations were used, resulting in different directional control valve opening speeds for the same joystick movement and hence different acceleration and boom speeds, designated slow, medium and fast (Table 1). A boom lift with »slow« valve setting was followed by a boom lowering with »slow« valve setting and so on. When actuated, the joystick was pushed to extremity with a fast movement. At end destinations of the boom (lifted and lowered), the joystick was released back to the neutral position for at least 5 seconds to ensure pressure stabilization in the hydraulic circuit.

222

The boom tip position varied from approximately 1.5–4.5 m above ground during a lift cycle. The operator (a 31 year-old male with no previous experience of work with heavy machinery) tried to keep the lift heights constant throughout the experiment. Initial and final cylinder lengths (Fig. 1) were documented (Table 1), and the possible effects of their variation on the results were reduced statistically (see chapter, »2.3 Statistical analysis«). By closing and opening certain valves in the hydraulic system, the accumulator could be overridden. When it was overridden, the EHLC functioned as a standard hydraulic lift cylinder (standard cylinder) providing a reference cylinder for comparisons of energy use (Fig. 2: panel 1). Pressure observations (p1 and p2) were recorded by Bofors TDS-1 load cells (error ±500 kPa) (Fig. 1). Both of the load cells used were calibrated with a Barnet Instruments dead weight tester once at the beginning of the experiment. The cylinder length sensor used (error ±2.3 mm) was based on a 10 turn 1 kΩ potentiometer. Data from load cells and the cylinder length sensor were recorded 105 times per second using a DEWE 2520 datalogger with an integrated computer running the DEWEsoft 6.5 program. Croat. j. for. eng. 37(2016)2

Evaluation of a New Energy Recycling Hydraulic Lift Cylinder for Forwarders (219–231)

J. Manner et al.

Table 1 Initial and final lift cylinder lengths, stroke lengths, lift times and average piston velocities during boom lifts for each combination of lift cylinder model (cylinder model), payload and directional control valve setting (valve setting). Mean values, with standard deviations in parenthesis Payload kg

Valve setting Slow

Medium Fast Slow

264

Medium Fast Slow

513

Medium Fast

Cylinder model

Observations Initial cylinder length Final cylinder length n mm mm

Lift time s

Piston velocity mm/s

Standard

1093 (1)

1213 (3)

121 (3)

4.6 (0.1)

26.1 (0.3)

EHLC

1091 (1)

1211 (3)

120 (3)

4.6 (0.1)

26.1 (0.3)

Standard

1093 (3)

1223 (5)

130 (4)

1.7 (0.1)

77.4 (2.5)

EHLC

1091 (3)

1219 (8)

128 (8)

1.7 (0.1)

76.7 (2.7)

Standard

1092 (6)

1223 (5)

131 (5)

1.5 (0.1)

85.5 (2.4)

EHLC

1096 (8)

1224 (9)

128 (12)

1.5 (0.1)

84.1 (3.2)

Standard

1113 (1)

1228 (2)

116 (2)

7.7 (0.2)

15.0 (0.2)

EHLC

1113 (1)

1229 (2)

117 (2)

7.9 (0.2)

14.8 (0.2)

Standard

1111 (3)

1234 (10)

123 (12)

1.8 (0.1)

68.8 (3.0)

EHLC

1114 (5)

1233 (6)

119 (9)

1.8 (0.1)

64.7 (1.5)

Standard

1037 (12)

1173 (8)

136 (15)

1.9 (0.2)

71.3 (2.6)

EHLC

1093 (9)

1236 (10)

144 (16)

2.0 (0.2)

72.6 (1.9)

Standard

1095 (1)

1213 (1)

118 (2)

20.6 (0.6)

5.7 (0.1)

EHLC

1094 (3)

1213 (1)

119 (3)

19.3 (1.0)

6.2 (0.2)

Standard

1060 (9)

1220 (8)

160 (13)

3.9 (0.4)

41.6 (2.6)

EHLC

1066 (6)

1229 (9)

162 (9)

3.5 (0.2)

46.2 (1.8)

Standard

1057 (18)

1216 (11)

159 (15)

3.7 (0.8)

44.4 (6.8)

EHLC

1053 (12)

1233 (8)

179 (9)

3.7 (0.3)

49.0 (2.9)

A general model for determining lift work during a given time interval (Wt®t+1) (Eq. 1) was used as a starting point for calculating the work conducted by the two cylinders.

Stroke length mm

Wt®t+1 = 0.5 ´ (pt + pt+1) ´ Apiston ´ (Lt+1 – Lt)

(1)

Where: p the cylinder pressure at time t or t+1, Apiston piston area, L is the cylinder length at time t or t+1. The work conducted during a given boom lift was calculated by dividing the lift into 105 time intervals per second and summing the work for all time intervals. The standard cylinder's lift work during time interval from t to t+1 (Ws, t®t+1) was calculated based on Eq. 1, with p1 used for pressure (i.e. the pressure observations retrieved from load cell1, Fig. 1), and Asp as piston area (Fig. 2: panel 1). Calculation of the functioning of EHLC’s lift work in a technically perfect state (WEHLC, t®t+1) was based on p1 and A1 as piston area (Fig. 2: panel 2). However, the EHLC might not work perfectly; a possible condition being when p1 might exceed p2 during part of the lift. Croat. j. for. eng. 37(2016)2

In this study, the part of a lift where p1<p2 is referred to as the EHLC's »successful lift phase« (Fig. 4: t1®t2), because during this phase the EHLC is theoretically capable of contributing to the lift with recovery energy. Similarly, the »unsuccessful life phase« refers to the part of the lift after which p1>p2 for the first time and the EHLC will not be able to contribute to reductions in energy use. Lift work calculations for the EHLC’s successful lift phase were identical to the calculations for flawless functioning of the EHLC. However, work calculations for EHLC’s unsuccessful lift phase varied depending on whether p1 or p2 was highest. For time intervals with p1>p2 the EHLC was assumed to function as a standard cylinder and work was determined correspondingly (Ws, t®t+1). For time intervals with p1<p2 during the unsuccessful lift phase, the EHLC's work was determined as the sums of WEHLC, t®t+1 and the secondary cylinder lift work (W2, t®t+1), given that regaining the EHLC's functionality was a result of accumulator being loaded from the hydraulic circuit during the lift (i.e. consuming energy), and not by using the recovered potential energy. W2, t®t+1 was calculated according to Eq. 1 with p2 used as pressure and A2 as the piston area (Fig. 2: panel 2).

223

J. Manner et al.

Evaluation of a New Energy Recycling Hydraulic Lift Cylinder for Forwarders (219–231)

The energy savings for the EHLC's successful lift phase were determined as the secondary cylinder proportion of the EHLC's total work during the successful lift phase, which in turn was determined as the sum of the primary and secondary cylinder work.

2.3 Statistical analysis The statistical analysis included only boom lifts, as boom lowerings were excluded. Analysis of covariance (ANCOVA) was used to evaluate effects of three fixed factors (lift cylinder model, payload, and valve setting) on two dependent variables: total work per lift and the initial p1, retrieved from a load cell1 (Fig. 1). The factor cylinder model had two levels ‒ EHLC and standard cylinder, while the factor payload had three levels ‒ objects with the mass 0, 264 and 513 kg ‒ held in the boom tip. Finally, the factor valve setting had three levels: slow, medium and fast (Table 1). In total, this resulted in 18 treatments, which were each replicated several times (15≤n≤57, Table 1). The three-factorial model contained all possible interaction effects between factors. Covariates were used if they significantly contributed to the model, and were considered logical and not risked to be confounded with treatment effects. In this experiment, the continuous variables ‒ initial and final cylinder length, stroke length and lift time ‒ were used as covariates for each boom lift. To avoid a rank

deficiency, the initial and final cylinder lengths were prioritized over the stroke length. The stroke length effect was tested only if the initial or final cylinder length had no effect. In addition to the dependent variables mentioned above, the EHLC was also evaluated separately to address its functionality (i.e. without comparison between cylinder models). For such analyses all the cylinder model-related terms were removed from the three-way ANCOVA, resulting in a two-way ANCOVA with the fixed two-way interaction effect: payload ´ valve setting. The dependent variables analyzed were related to pressure in the primary and secondary cylinders as well as work and time during the successful lift phase. The order between treatments was randomized, but all replicates within a treatment were conducted sequentially. However, the first 10 boom lifts within each treatment were excluded from the analysis to ensure that the system had stabilized in terms of oil pressure and temperature during the data collection. A general linear model (GLM) was used to analyze the analysis of variance (ANOVA) and ANCOVA models. During the GLM procedure, pair-wise differences were analyzed with Tukey’s simultaneous test of means. The normality of residuals was evaluated by the Anderson-Darling test. Differences in initial cylinder pressures within the EHLC were tested for deviation from

Fig. 4 Example of observed pressures and cylinder lengths as a function of time for one whole EHLC boom lift cycle (t1®t7) with the valve setting »medium« and payload of 0 kg. The boom lift, as well as the successful lift phase, starts at time t1. The successful lift phase ends at t2 when p2 exceeds p1 for the first time. The boom lowering, as well as charging of the accumulator, starts at t4. At t5®t6, p2 drops rapidly because the pressure relief valve opens and oil flows to the oil reservoir (Fig. 1). At t7, a new lift starts

224

Croat. j. for. eng. 37(2016)2

Evaluation of a New Energy Recycling Hydraulic Lift Cylinder for Forwarders (219–231)

J. Manner et al.

Table 2 Levels of significance ( p-values) and explained variance (R2 adjusted values) obtained from the analysis of variance (ANOVA) of: effects on the dependent variables listed in the first column of the factors cylinder model (a), payload (b) and valve setting (c); their fixed interaction effects (a ´ b, a ´ c, b ´ c and a ´ b ´ c); and effects of the covariates initial lift cylinder length (d), final lift cylinder length (e), lift time (f) and stroke length (g) p-value Dependent variables

Factor a

a×b

Covariate a×c

b×c a×b×c

Adj. R2 %

Energy consumption per whole lift

<0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001

–

99.7

473

Initial p11)

<0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 Excl.

1) 2

–

83.8

473

Initial p

–

<0.001 <0.001

–

<0.001

–

<0.001

–

99.6

229

Difference between initial p2 and initial p1

–

<0.001 <0.001

–

<0.001

–

Excl.

–

94.4

229

Number of p2 and p1 intersections per lift

Excl.

73.3

229

–

<0.011 <0.001

–

<0.001

–

1) 2

–

<0.001 <0.001

–

<0.001

–

Excl.

–

76.6

162

1) 1

W during the successful lift phase

–

<0.001 <0.001

–

<0.001

–

Excl.

–

78.2

162

Relative energy saving for successful lift phase2)

–

<0.001 0.068

–

<0.001

–

Excl.

–

43.7

162

Successful phase stroke length

–

<0.001 <0.001

–

<0.001

–

Excl.

–

76.0

162

Successful phase proportion of total stroke length

–

<0.001 <0.001

–

<0.001

–

Excl.

0.005

–

Excl.

76.5

162

Successful phase lift time

–

<0.001 <0.001

–

<0.001

–

0.011

–

88.6

162

Successful phase proportion of the total lift time

–

<0.001 <0.001

–

<0.001

–

Excl.

0.002

–

Excl.

78.3

162

W during the successful lift phase

<0.001 Excl.

n£229 when only the EHLC treatments were included, and n=162 when only the EHLC treatments with medium and fast directional control valve settings were included – independent variable was not tested Excl. covariate term was tested, but excluded from the model because it had no effect (p>0.05) or it decreased adj. R2-value 1) p1 primary or standard cylinder pressure (depending on cylinder type) secondary cylinder pressure (only EHLC) p2 W2 secondary cylinder work W1 primary cylinder work 2) W2 during the successful lift phase/(W2 during the successful lift phase+W1 during the successful lift phase)

zero by use of a one-sample t-test. The critical level of significance was set to 5%. Minitab 16 (Minitab Ltd.) was used for all analyses.

3. Results The EHLC's functionality is dependent on the pressure in the secondary cylinder exceeding the pressure in the primary cylinder (p1<p2, Figs. 1–2). However, this was found to never occur during a full lift, but only during various intervals of the first part of the boom lift, which varied from repetition to repetition (Fig. 4: t1®t2). Moreover, p2 and p1 were practically identical during the whole lift when using the valve setting »slow« (no data shown). Croat. j. for. eng. 37(2016)2

For valve settings »medium« and »fast«, the EHLC's p2 level surpassed p1 in the beginning of a lift (Fig. 4). During this »successful lift phase« (Fig. 4: t1®t2), the accumulator contributed recovered energy from the preceding lift. During the next boom lowering, the accumulator was loaded as p1<p2 (Fig. 4: t4®t5). When p2 exceeded approximately 30 MPa, the accumulator stopped charging, indicating that the pressure relief valve was released at that pressure (Fig. 4: t5®t6).

3.1 Comparison of cylinder models for whole lifts The three main factors – the cylinder model, payload, and valve setting – significantly affected both of the two dependent variables initial p1 and the energy

225

J. Manner et al.

Evaluation of a New Energy Recycling Hydraulic Lift Cylinder for Forwarders (219–231)

consumption per full lift (three-way ANCOVA, p<0.001, Table 2: rows 1–2). As expected, most of the discrepancies between the two dependent variables were explained by the payload, while the valve setting and cylinder model only had minor effects (data not shown). The initial p1 was not affected by the only tested covariate ‒ the initial cylinder length (three-way ANCOVA, p=0.774, Table 2: row 2, complete data not shown). On the other hand, the energy consumption per lift was affected by all the recorded covariates: the initial and final cylinder length, and lift time (threeway ANCOVA, p<0.001, Table 2). Overall, the total energy consumption model was improved most by inclusion of the initial cylinder length, followed by the final cylinder length. The lift time had least effect (data not shown).

The EHLC's initial p1 increased significantly (p<0.001) with increasing payload across all the three valve settings (Table 3). However, the standard cylinder's initial p1 increased significantly (p<0.001) with increasing payload only across the valve setting »fast«. In addition, there was a lack of statistically significant differences between the treatments when the payload effects were compared over the valve settings »slow« and »fast«. This resulted in significant (p<0.001) twoand three-way interaction effects (Tables 2–3). As expected, the energy consumption per lift for both cylinder models increased significantly (p<0.001) with increasing payload across all three valve settings (Table 3). The valve settings significantly (p<0.05) affected EHLC's energy consumption with payloads of 0 and 513 kg, but had no significant effect with the 264 kg

Table 3 Pressure and energy consumption for a whole lift with EHLC and standard cylinder for each combination of payload, valve setting and cylinder model. Mean values, with standard deviations in parenthesis Payload kg

Valve setting

Slow

Medium

Fast

Slow

264

Medium

Fast

Slow

513

Medium

Fast

Difference between Number of p2 and p1 Energy consumption Observations initial p2 and initial p12), intersections during per whole lift, J n the whole lift, n kPa

Cylinder model

Initial p11) kPa

Initial p2 kPa

Standard

14,182DE (304)

–

18,992EF (346)

EHLC

13,068E (98)

12,591G (129)

–479F (81)

47.5A (18.9)

19,330E (410)

Standard

13,264E (2617)

–

18,257F (551)

EHLC

10,464 (1670)

ABC

Standard

13,105 (2698)

–

18,400 (712)

EHLC

9791F (777)

21,637D (400)

11,767B (762)

11.7BC (4.4)

18,488EF (1573)

Standard

24,374

–

18,328 (524)

(465)

13,909 (1912)

– F

5.1 (2.3)

– BC

17,669 (1340) F

28,514 (388)

EHLC

17,588 (208)

16,863 (195)

–1053 (93)

11.8 (3.1)

28,186 (465)

Standard

21,647A (2163)

–

27,390C (2423)

EHLC

23.4 (3.4)

27,185 (1663)

14,985 (1154) D

24,516 (180)

9178 (1208)

Standard

15,167 (936)

–

23,064 (3010)

EHLC

15,262D (576)

24,152C (211)

8870C (506)

14.5BC (6.0)

27,172C (3294)

Standard

–

19,859 (145)

36,164 (475)

EHLC

20,043 (115)

19,868 (121)

–227 (33)

18.4 (7.9)

35,642 (841)

Standard

21,142A (1725)

–

32,773A (3214)

EHLC

19,928AB (2417)

24,655A (282)

5119D (2081)

24.4B (11.0)

32,097B (2107)

Standard

20,518 (2047)

–

32,763 (4298)

EHLC

21,662 (1896)

23,974 (244)

– E

2909 (1373)

58.3 (21.9)

32,720 (2269)

Within columns, different superscript letters indicate significant differences (p<0.05). Statistical models are described in Table 2 1) p1 pressure recorded in load cell1 (Fig. 1), i.e. in the primary EHLC cylinder and standard cylinder p2 pressure recorded by load cell2, i.e. in the accumulator circuit and the secondary cylinder 2) Each of the nine EHLC mean values in the column were statistically significantly different from zero (one-sample t-test, p<0.001)

226

Croat. j. for. eng. 37(2016)2

Evaluation of a New Energy Recycling Hydraulic Lift Cylinder for Forwarders (219–231)

J. Manner et al.

Table 4 EHLC's energy savings (mean values with standard deviations in parenthesis) for the successful lift phase Cylinder work 1) Payload Valve kg setting

264

513

W2, J

W1, J

Successful lift phase Successful lift phase Successful lift phase Successful lift Energy Observations proportion of the total lift time saving 2) phase stroke length proportion of the total n lift time, % s stroke length, % mm %

Medium 1033A (423) 9422A (4191) 9.4A (1.0) Fast

34C (15) C

396C (192) C

Medium

26 (14)

318 (140)

Fast

17 (9)

Medium

407B (30)

Fast

340 (29)

191 (87)

8.4B (1.3) B

8.2 (0.9) B

69.6A (31.4)

52.3A (22.0)

0.84A (0.32)

51.9A (17.5)

2.0C (0.9)

0.3C (0.11)

0.19B (0.03)

9.3C (2.2)

1.4 (0.6) C

2.6 (1.1) C

10.0 (3.4)

0.25 (0.06)

8.5 (0.9)

0.8 (0.3)

3.1 (0.5)

0.17 (0.03)

10.2 (1.5)

5326B (365) 7.1C (0.1)

22.1B (1.6)

13.7B (1.3)

0.98A (0.07)

33.7B (1.3)

4496 (386) 7.0 (0.0)

29.1 (6.6)

17.7 (4.2)

0.98 (0.12)

31.9 (1.7)

Within columns, different superscript letters indicate significant differences (p<0.05). Statistical models are described in Table 2 1) W2 secondary cylinder work, and W1 = primary cylinder work 2) W2 during the successful lift phase/(W2 during the successful lift phase+W1 during the successful lift phase)×100

payload. In contrast, the standard cylinder's lift energy consumption was significantly affected by the valve settings with the 264 kg payload, but not with the 0 and 513 kg payloads. This resulted in significant (p<0.001) two- and three-way interaction effects (Tables 2–3).

3.2 Evaluation of the EHLC's successful and unsuccessful lift phases In addition to the initial p1 and energy consumption per full lift, the EHLC's successful lift phase was further analyzed with supplemental dependent variables (Table 2: number of observations in the range of 162–229). The two main factors, payload and valve setting, significantly (p<0.001) affected all the supplemental dependent variables, when data for all three valve settings were included in the analyses (Table 2). The initial p2, and number of intersections of p2 and p1 per lift, were also significantly affected by the covariate initial cylinder length (p<0.001) (Table 2). The significant (p<0.001) interaction effects showed that the factors effects varied between the compared treatments, and occasionally, the interaction effect was the result of a lack of differences between the compared treatments (Tables 2–3). With the valve setting »slow«, the initial p1 was significantly higher than p2 (Table 3), thus the possibilities of energy recovery were eliminated even before the boom was lifted. Consequently, only data obtained with the valve settings »medium« and »fast« were further analyzed. With these settings, the initial p2 was substantially higher than the initial p1, which enabled energy recovery (Table 3). The payload and valve setCroat. j. for. eng. 37(2016)2

ting significantly (p<0.05) affected all except one dependent variable in EHLC's successful lift phase (Table 4). The exception, which fell just outside the set level for significance, was that the relative energy savings for EHLC's successful lift phase was not affected by the valve setting (p=0.068) (Table 2). In addition, the valve settings had a significant (p<0.05) effect on the dependent variables with the 0 kg payload, but not with the 264 or 513 kg payloads (Table 4). This resulted in significant two-way interaction effects (p<0.001, Table 2). The EHLC's successful lift phase corresponded to 9.3–10.2% of the total lift time when lifting a payload of 0 kg with the valve setting »fast«, or when lifting a payload of 264 kg with either »fast« or »medium« valve settings. However, given that the successful lift phase covered only an acceleration phase, the successful lift phase proportion of the total stroke length was only 0.3–3.1%. Thus, W2 was only 17–34 J (Table 4). The EHLC's successful lift phase constituted 31.9–33.7% of the total lift time when lifting a payload of 513 kg with either valve settings, corresponding to 13.7–17.7% of the total stroke length and W2 of 340–407 J (Table 4). When using the valve setting »medium« and lifting a payload of 0 kg, the successful lift phase was 51.9% of the total lift time. This corresponded to 52.3% of the total stroke length and resulted in a W2 of 1033 J (Table 4).

3.3 Energy savings When evaluating a full boom lift, the EHLC functioned best for payloads of 513 kg and 0 kg with the valve setting »medium«. Under these conditions lifting with the EHLC consumed significantly less energy

227

J. Manner et al.

Evaluation of a New Energy Recycling Hydraulic Lift Cylinder for Forwarders (219–231)

than lifting with the standard cylinder (2.1–3.2%, p<0.001) (Table 3). However, when lifting 264 kg with the valve setting »fast« the EHLC did not function flawlessly at all, and consumed significantly more energy for a full lift than the standard cylinder (17.8%, p<0.001) (Table 3). During the successful lift phase, the EHLC saved approximately 7.0–9.4% of the energy consumed (p<0.001). The relative energy savings were also significantly higher with a lower payload (Table 4).

4. Discussion 4.1 Comparison with previous studies During the successful lift phase EHLC saved up to 9.4% of the energy consumed. However, on average the successful lift phase accounted for only a small proportion of the whole boom lift, and sometimes during the following unsuccessful lift phase the EHLC even increased energy requirements. In addition, the savings during the successful lifting phase were in the lower end of ranges of savings (ca. 5–65%) reported in previous evaluations of other solutions for reducing energy requirements of mobile lifting devices (e.g. Liang and Virvalo 2001a, Sun and Virvalo 2003, Rydberg 2005, Sun and Virvalo 2005, Virvalo and Sun 2005, Lin et al. 2010, Lin and Wang 2012, Minav et al. 2012, Noréus et al. 2013, Wang et al. 2013). Furthermore, the EHLC did not give any energy recovery at all with the valve setting »slow«, presumably because of internal oil leakage between the primary and secondary cylinders. The boom lifting and lowering times were at least twice as long with the »slow« valve setting than with the settings »medium« or »fast«, and inevitably internal leakage and thus the pressure decrease will be larger if the duration is longer (assuming all other variables remain constant). An example of leakage from the secondary cylinder is shown in Fig. 4, where p2 is decreasing during the time interval t6®t7 and the only possible reason for the decrease (after the pressure relief valve has closed) is leakage. The finding that degrees of leakage are correlated with its duration is consistent with previous reports (Wang et al. 2013). An apparent problem with the tested EHLC is that the accumulator gas volume seems to be too small in relation to the secondary cylinder oil flow volume during the boom liftings and lowerings. This issue can be seen in Fig. 4, where the EHLC's pressure curve decreases sharply as a function of cylinder length (t1®t2), then increases rapidly until the pressure relief valve opens (t4®t5). According to Boyle’s law, increasing the accumulator gas volume could solve this

228

problem as it would stabilize p2 or reduce its peak-topeak pressure amplitude. The main problem here is not too low maximum p2, which is already regulated by the pressure relief valve, as described, but that p2 decreases too rapidly.

4.2 Strengths and weaknesses of the study The study was conducted in an experimental setting, with standardized work procedures and high data recording frequencies, facilitating isolation of the effects of cylinder type under various work conditions, and potentially, the identification of useful general procedures for assessing forwarder cranes energy requirements, forces and functional parameters during lift work. Considering the EHLC as an early prototype, we have focussed mainly on work phases where the EHLC made a successful contribution when analyzing the data. The forwarder used was a standard forwarder equipped with a load sensing system in the hydraulic system, so the hydraulic oil flow and pressure delivered from the hydraulic pump depended on the power required at the moment (see e.g. Scherer et al. 2013). The load sensing system was not overridden during the experiment. Thus, effects from the load sensing system could have been confounded with the tested factor effects, which may have affected the ANOVA results. However, to our knowledge any effects of the load sensing system should be negligible for the technical evaluation of the EHLC. When determining the energy consumption of the EHLC for a whole boom lift, it was assumed that the secondary piston thrusts either the primary or the secondary cylinder head. However, this assumption was not entirely valid because the secondary piston could also move in the direction of lower pressure. Nevertheless, this error sources only applies when the successful lift phase is complete and should not therefore impact the analysis of the EHLC's successful lift phase.

4.3 Improvements and future studies The results of this study indicate several possible improvements for the EHLC. First, the internal leakage problem has to be resolved. Second, in addition to enlarging the accumulator, its pressurization should be optimized so that the EHLC works flawlessly throughout the whole lifting phase. This suggestion is prompted by the observation that the generated pressure spikes do not appear to be sufficient to maintain pressurizations long enough to reduce energy needs within the current accumulator system. Consequently, it is essential to redesign the system to provide a notably higher initial accumulator pressure. During the Croat. j. for. eng. 37(2016)2

Evaluation of a New Energy Recycling Hydraulic Lift Cylinder for Forwarders (219–231)

»unsuccessful lift phase« (after the first p2 and p1 intersection), the hydraulic pump pressurized the primary cylinder as well as the accumulator and secondary cylinder (Fig. 4: t2®t3). However, flawless functioning of the EHLC depends on the actual boom length and load, since if p2 is too high and the boom too short energy will actually be needed to lower the boom. This reflects a general trade-off for all weight-balancing systems, and requires optimization according to the given crane dimensions and the loads lifted and lowered. Third, A2 should be enlarged. Overall, the experiment should be replicated with a larger accumulator, a refined accumulator pressurization system, and possibly a larger A2.

4.4 Potential energy savings for all forwarder work The current EHLC can save up to 3.2% of energy for a full boom lift under optimal conditions. However, if its technical weaknesses can be resolved, the savings should be at least 6.6%, as explained in the Introduction. Moreover, if some additional technical improvements are made, additional energy savings could be achieved. For instance, if the accumulator pressurization system is redesigned to ensure that p2 is consistently at least 20% higher than p1, the energy savings for a whole boom lift should theoretically be at least 7.9%. In addition, increasing A2 would increase the energy recycling potential of the EHLC. For example, increasing secondary piston diameter by 20% would increase the possible energy savings from 6.6% to 9.5%. Therefore, the EHLC's energy recycling potential could be enhanced by increasing both secondary piston diameter and p2 by 20%. With these two improvements, the EHLC could theoretically recycle at least 11.4% of potential energy, ignoring possible leakage. However, energy-efficient lifting devices have not yet acquired any of the market shares for forwarder cranes. So far, the EHLC's market consists primarily of hydraulic lift devices, where the full engine power is used for lift work, which is not the case for forwarders. Currently minor energy savings can be gained from energy-efficient lift cylinders for a standard forwarder, as driving the machine requires substantially more force and power than the boom lifting (cf. Löfgren 1999, Edlund et al. 2013, Table 3). Thus, the research priority should be placed on decreasing power requirements during the driving phase, i.e. improving powertrain efficiency, as in a few recent studies (e.g. Edlund et al. 2013, Swedish Energy Agency 2014). Overall, combining energy-efficient lifting devices with a hybrid powertrain could be interesting for fuCroat. j. for. eng. 37(2016)2

J. Manner et al.

ture studies. With an energy-efficient lifting device reducing the energy needed for the crane work, more power from the combustion engine could be directed to loading the battery throughout the crane work. This electric energy, stored during the crane work, would then be available for use during driving, when the power input requirement is highest. Thus, an energy-efficient lift device could improve the battery loading efficiency of a hybrid system during crane work. Such technology could enable the use of less powerful, i.e. less fuel consuming, combustion engines in forwarders.

Acknowledgement This study was funded by Stora Enso Skog AB and the Forest Industrial Research School on Technology (FIRST). We thank Sees-editing Ltd for revising the English.

5. References Deepak, S.R., 2012: Static balancing of rigid-body linkages and compliant mechanisms, Doctoral thesis. Indian Institute of Science. The faculty of engineering. Department of mechanical engineering. Bangalore, India. 168 p. Edlund, J., Bergsten, U., Arvidsson, H., 2013: A forest machine bogie with a bearing capacity dependent contact area: acceleration and angular orientation when passing obstacles and drawbar pull force and free rolling resistance on firm ground. Silva Fennica 47(3): article id 1017, 7 p. Einola, K., 2013: Prestudy on Power Management of a CutTo-Length Forest Harvester with a Hydraulic Hybrid System. In: The 13th Scandinavian International Conference on Fluid Power, Linköping, Sweden, 71–83. 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. DOI: 10.1080/14942119.2014.974309. Ersson, B.T., Jundén, L., Lindh, M., Bergsten, U., 2014: Simulated productivity of conceptual, multi-headed tree planting devices. International Journal of Forest Engineering 25(3): 201–213. DOI:10.1080/14942119.2014.972677. European Union, 2014: Energy Efficiency Directive. http:// ec.europa.eu/energy/efficiency/eed/eed_en.htm. Accessed 9 March 2015. Gawlik, A., Michałowski, S., 2008: Concept of active counterweight system for medium size excavators. In: 5th FPNIPhD Symposium, Cracow, Poland, 462–465. Gerasimov, Y.Y., Siounev, V.S., 1998: Forest machinery crane compound scheme synthesis: »Skeleton« part. Journal of Forest Engineering 9(1): 25–30. DOI: 10.1080/08435243.1998.10702708. Gerasimov, Y.Y., Siounev, V.S., 2000: Forest machinery crane compound scheme synthesis: Optimization of hydraulic cyl-

229

J. Manner et al.

Evaluation of a New Energy Recycling Hydraulic Lift Cylinder for Forwarders (219–231)

inder operating mechanisms. Journal of Forest Engineering 11(1): 73–79. DOI: 10.1080/08435243.2000.10702746.

Malmberg, C.A., 1981: Terrängmaskinen. Del 2 (The off-road vehicle. Volume 2). 461 p.

Ho, T.H., Ahn, K.K., 2012: Design and control of a closedloop hydraulic energy-regenerative system. Automation in Construction 22: 444–458.

Manner, J., Nordfjell, T., Lindroos, O., 2013: Effects of the number of assortments and log concentration on time consumption for forwarding. Silva Fennica 47 (4): 19 p. article id 1030. DOI: http://dx.doi.org/10.14214/sf.1030

Hui, S., Junqing, J., 2010: Research on the system configuration and energy control strategy for parallel hydraulic hybrid loader. Automation in Construction 19:213–220. Ingvast, H., 1989: Hydraulikens grunder (Basic hydraulics) Utgåva 2. Institutet för tillämpad hydraulik (ITH). Örnsköldsvik, Sweden, 243 p. Jundén, L., Bergström, D., Servin, M., Bergsten, U., 2013: Simulation of boom-corridor thinning using a double-crane system and different levels of automation. International Journal of Forest Engineering 24(1): 16–23. DOI: 10.1080/14942119.2013.798131. Kim, J., Yoon, G.H., Noh, J, Lee, J., Kim, K., Park, H., Hwang, J., Lee, Y., 2013: Development of optimal diaphragm-based pulsation damper structure for high-pressure GDI pump systems through design of experiments. Mechatronics 23(3): 369–380.

Manner, J., Palmroth, L., Nordfjell, T., Lindroos, O., 2016: Load level forwarding work element analysis based on automatic follow-up data. Silva Fennica 50(3): 19 p. article id 1546. DOI: http://dx.doi.org/10.14214/sf.1546. Manring, N.D., 2005: Hydraulic control systems. John Wiley & Sons Inc. New York, USA, 464 p. Minav, T.A., Virtanen, A., Laurila, L., Pyrhönen. J., 2012: Storage of energy recovered from an industrial forklift. Automation in Construction 22: 506–515. Noréus, O., Hägglund, M., Emlén, M., 2013: Measurements and Simulations to Evaluate Strategies for Improved Energy Efficiency of a Reach Stacker, Forwarder and Wheel Loader. In: The 13th Scandinavian International Conference on Fluid Power, Linköping, Sweden, 319–325.

Laine, T., Rantala, J., 2013: Mechanized tree planting with an excavator-mounted M-Planter planting device. International Journal of Forest Engineering 24(3): 183–193. DOI: 10.1080/14942119.2013.844884.

Ortiz Morales, D., Westerberg, S., La Hera, P.X., Mettin, U., Freidovich, L., Shiriaev, A.S., 2014: Increasing the level of automation in the forestry logging process with crane trajectory planning and control. Journal of Field Robotics 31(3) :343–363. DOI: 10.1002/rob.21496.

Liang, X., Virvalo, T., 2001a: Development and research of an energy saving drive in a hydraulic crane. In: 7th Scandinavian International Conference on Fluid Power, Linköping, Sweden, 151–161.

Rydberg, K.E., 2005: Hydraulic accumulators as key components in energy efficient mobile systems. In: 6th International Conference on Fluid Power Transmission and Control, Hangzhou, China, 124–129.

Liang, X., Virvalo, T., 2001b: Energy reutilization and balance analysis in a hydraulic crane. In: The 5th International Conference on Fluid Power Transmission and Control, Hangzhou, China, 306–310.

Scherer, M., Geimer M., Weiss B., 2013: Contribution on Control Strategies of Flow-On-Demand Hydraulic Circuits. In: 13th Scandinavian International Conference on Fluid Power, SICFP2013, Linköping, Sweden, 531–540.

Lin, P.Y., Shieh, W.B., Chen, D.Z., 2013: A theoretical study of weight-balanced mechanisms for design of spring assistive mobile arm support (MAS). Mechanism and Machine Theory 61: 156–167.

Sun, W., Virvalo, T., 2003: Accumulator-pump-motor as energy saving in hydraulic boom. In: 8th Scandinavian International Conference on Fluid Power, Tampere, Finland, 297– 309.

Lin, T., Wang, Q., Hu, B., Gong, W., 2010: Research on the energy regeneration systems for hybrid hydraulic excavators. Automation in Construction 19(8): 1016–1026.

Sun, W., Virvalo, T., 2005: Simulation study on a hydraulicaccumulator-balancing energy-saving system in hydraulic boom. In: 50th National Conference on Fluid Power, Las Vegas, USA, 371–381.

Lin, T., Wang, Q., 2012: Hydraulic accumulator-motor-generator energy regeneration system for a hybrid hydraulic excavator. Chinese Journal of Mechanical Engineering 25(6): 1121–1129. DOI: 10.3901/CJME.2012.06.1121. Lindroos, O., Bergström, D., Johansson, P., Nordfjell, T., 2008: Cutting corners with a new crane concept. International Journal of Forest Engineering 19(2): 21–27. DOI: 10.1080/14942119.2008.10702564. Löfgren, B., Granlund, P., Brunberg, T., 1999: Test av tre stora skotare ‒ dragkraft, bromsar, rullmotstånd och bränsle förbrukning. Skogforsk Resultat, nr. 2, 1999. Stiftelsen Skogbrukets Forskningsinstitut, Uppsala, Sweden, 4 p.

230

Swedish Energy Agency, 2014: Elforest Technologies. http:// www.energimyndigheten.se/en/Innovations-R--D/BusinessDevelopment-and-Commercialisation/Invest-in-CleanTechEnergy/El-forest. Accessed 9 March 2015. United Nations, 2014: UNFCCC and the Kyoto Protocol. http://www.un.org/wcm/content/site/climatechange/pages/ gateway/the-negotiations/the-un-climate-change-convention-and-the-kyoto-protocol. Accessed 9 March 2015. Van de Ven, J., 2013: Constant pressure hydraulic energy storage through a variable area piston hydraulic accumulator. Applied Energy 105: 262–270. Croat. j. for. eng. 37(2016)2

Evaluation of a New Energy Recycling Hydraulic Lift Cylinder for Forwarders (219–231) Virvalo, T., Sun, W., 2005: Improving energy utilization in hydraulic booms – what it is all about. In: 6th International Conference on Fluid Power Transmission and Control, Hangzhou, China, 55–65. Wang, T., Wang, Q., Lin, T., 2013: Improvement of boom control performance for hybrid hydraulic excavator with potential energy recovery. Automation in Construction 30: 161–169.

J. Manner et al.

WIPO Patent Application WO/2011/075034, 2011: World Intellectual Property Indicators. 9 p. Zhang, L., 2011: An energy-saving oil drilling rig for recovering potential energy and decreasing motor power. Energy Conversion and Management 52(1):359–365.

Personal communication: Roger Gustavsson e-mail: roger.gustavsson@thordab CEO Thordab AB Arnäsvall SWEDEN

Authors’ address: Researcher, Jussi Manner, PhD. * e-mail: jussi.manner@skogforsk.se Skogforsk Uppsala Science Park 751 83 Uppsala SWEDEN Assoc. prof. Ola Lindroos, PhD. e-mail: ola.lindroos@slu.se Swedish University of Agricultural Sciences Department of Forest Biomaterials and Technology 901 83 Umeå SWEDEN Hans Arvidsson, senior test engineer e-mail: hans.arvidsson@smp.sp.se Swedish Machinery Testing Institute 904 03 Umeå SWEDEN

Received: May 12, 2015 Accepted: December 24, 2015 Croat. j. for. eng. 37(2016)2

Prof. Tomas Nordfjell, PhD. e-mail: tomas.nordfjell@slu.se Swedish University of Agricultural Sciences Department of Forest Biomaterials and Technology 901 83 Umeå SWEDEN * Corresponding author

231

Original scientific paper

Effects of Forwarder Operation on Soil Physical Characteristics: a Case Study in the Italian Alps Martina Cambi, Stefano Grigolato, Francesco Neri, Rodolfo Picchio, Enrico Marchi Abstract One of the most important issues in ground based wood extraction in forestry is to minimize the soil damage caused by heavy forestry machines. Generally, harvesting effects include changes in vegetation nutrient availability, soil microclimate/structure and litter quantity/ quality. Several studies were carried out on the impacts of heavy machines on the soil. However, only few studies took into consideration the effect of bogie tracks on the soil. The research focuses on the influence of forwarder machines equipped with bogie tracks on the soil compaction through changes of physical soil parameters and precisely bulk density, porosity, shear and penetration resistance. The study was carried out in a conifer stand of Larix decidua Mill. and Picea abies L. in a forest of North-eastern Italy during logging with forwarder. In this site, 3 tracks were identified, 2 concerned loaded forwarder passages and 1 control (no passages). The tracks were: (i) track A with a slope of 31% with 2 passages and track B with a slope of 3% having 10 passages. Soil samples were collected on all tracks in order to determine the influence of forwarder passes on soil physical properties. The results showed a different impact of logging operations on the soil of different tracks. Keywords: forwarder, bogie track, soil compaction

1. Introduction One of the most important issues of the forest sector is to minimize the ground damage caused by heavy forestry machines during forest operations (Edlund et al. 2013). Generally, harvesting effects include changes in vegetation, nutrient availability, soil microclimate and structure and litter quantity and quality (Keenan and Kimmins 1993, Jurgensen et al. 1997). However, adequately managed forest ecosystems are suggested to be highly resilient in the long-term perspective (Sanchez et al. 2006). Forest operations, such as forwarding and skidding, have a high potential for soil compaction (Jamshidi et al. 2008, Cambi et al. 2015, 2016). Harvesting induced soil compaction depends on several factors (Berli et al. 2004, Arthur et al. 2013, Berli et al. 2004, Han et al. 2006, Magagnotti et al. 2012, Sakai et al. 2008) and among them one of the most important is the means of propulsion (Alakukku et al. 2003, Cambi et al. 2016, Eliasson 2005, Marchi et al. Croat. j. for. eng. 37(2016)2

2014, Picchio et al. 2011, Picchio et al. 2012a, 2012b). The means of propulsion influence the size of the contact area that changes continuously due to accelerating/braking. Superimposition of different types of stress from neighbouring contact areas (e.g., tandem tires, pendulum axles, bogies) may occur, leading to stress paths specific for any axle or wheel arrangement (Alakukku et al. 2003, Savelli et al. 2010). Due to their lower contact area, wheeled vehicles generally disturb soil more dramatically than tracked ones (Johnson et al. 1991, Jansson and Johansson 1998). Bogie tracks, in spite of increasing the mass on the trailer by 10–12%, may reduce rut depth by up to 40% when compared to rather wide and soft tires, which is probably due to a reduction in the relative rolling resistance coefficient (Bygdén et al. 2004). In addition, bogies of forestry vehicles are used to increase the traction and stability of the machine and to increase the ride comfort of the operator (Meyer et al. 1977, Mac Donald 1999, Pijuan et al. 2012, Alakukku et al. 2003, Bygdén and Wäster-

233

M. Cambi et al.

Effects of Forwarder Operation on Soil Physical Characteristics: a Case Study in the Italian Alps (233–239)

lund 2007). The use of bogie tracks may reduce both the depth of the ruts and the soil compaction (i. e. Ansorge and Godwin 2007, Sakay et al. 2008, Syunev et al. 2009, Gerasimov and Katarov 2010, Uusitalo et al. 2011). Rut formation in the soil is a sign that the load is higher than the bearing capacity in the upper soil layer (Bygdén et al. 2004). Some energy applied to tracks to deform the soil (Yong et al. 1984), thus reducing the rut formation, would generally mean reducing rolling resistance energy (Muro 1982). This means a smaller risk of negative effects on forest growth (Bygdén et al. 2004). A key point to be made here is that the resulting rolling resistance, as the coefficient obtained, is the towing force required to move the device divided by the vertical load (Wong 1989 and Okello et al. 1994). The benefits relative to the use of bogie tracks established from these studies are: Þ reduction in relative rolling resistance coefficient Þ reduction of rutting Þ reduction of surface cone resistance in ruts of about 10% Recently, Gerasimov and Katarov (2010) have investigated the influence of a bogie track on soil compaction. Edlund et al. (2013) considered a new design of tracked bogie called the long-tracked-bogie (LTB) to combine the features of wheels and tracks in a single bogie design; the LTB showed about 3 and 4% higher pull force at 75 and 90% slip, respectively, than the conventional bogie. In addition to the results shown, LTB appears to pass wider ditches/cavities, more smoothly with a lower pitch angle, than a conventional bogie (Edlund et al. 2013). The aim of our research was to investigate the influence of a bogie track on physical soil parameters through soil compaction, as well as the impact of slope terrain and forwarder traffic. Soil compaction was evaluated by analyzing the changes in bulk density, porosity, shear and penetration resistance.

The study site is an even-aged conifer stand of Larix decidua (Mill. 1768) and Picea abies (L.) H. Karst and located between 1550 and 1600 m a.s.l. The area was selected because it is a focus point due to the presence of Italian and foreign logging companies that use this type of forest machines. The tests were conducted in June 2014. In this area 3 trails (of 18 m in length) were identified – 2 affected by passes and 1 control (no passes). The trails were: Þ trail A with the slope of 31% and affected by 2 passes Þ trail B with the slope of 3% of and affected by 10 passes Table 1 shows in details slope percentages of trails. The soil in the test area was clay and the moisture was 19% during the test. Each of the identified trails (Fig. 1) were divided into three plots (approximately Table 1 Slope percentage Plot

Length

Longitudinal slope

Lateral slope

–30

–31

–32

–2

Unit Trail A

Trail B

Control

2. Methods and data 2.1 Site description and soil sampling Experiments were carried out in the Dolomite area of the Northern-east Alps in Italy close to Borca di Cadore (46°26’14’’28 N; 12°13’14’’16 E). The climate of this area is alpine and the mean annual precipitation in the last five years is 1150 mm. The mean annual temperature in the same period was 8°C with the minimum temperatures below 0°C in the coldest months and maximum temperatures over 30°C in the warmest months.

234

Fig. 1 Sample scheme of a trail Croat. j. for. eng. 37(2016)2

Effects of Forwarder Operation on Soil Physical Characteristics: a Case Study in the Italian Alps (233–239)

M. Cambi et al.

6.30 m), where 30 soil samples were sampled (15 within the left track and 15 within the right track) for a total of 90 soil samples of tracks (30 soil samples x 3 plots). Soil samples were collected by a rigid metallic cylinder (8.5 cm high and 5.0 cm inner diameter) after litter removal. Considering all tracks, bulk density and porosity values were calculated for 90 soil samples to analyze the soil compaction. The samples were placed in plastic bins and transported to the laboratory, where the BD and soil porosity (n) were determined for each soil sample collected by the following formulas:

Dimensions of forwarder 1110D – 8 wheel

Value

Unit

Length

10,310

Width

2880

(2)

Transport Height

3700

Where: Dp is the particle density measured by a pycnometer on the same soil samples used to determine the bulk density

Ground Clearance

605

Wheelbase

5400

Load Stake Width

2760

Length of Wood Bunk

4580

Approach Angle

Bulk density = Dry

Weight of cylinder (158.96 cm3) (1) Volume

 Dp − Db   ×100 n=   Dp 

Close to each sampling point, penetration resistance and shear resistance were measured in triplicate using a TONS/FT2 penetrometer and a GEONOR 72412 scissometer, respectively. These samples were made of 30 measures per track (10 measures x 3 plots). Sta tistical analyses of data were done using STATISTICA 7.1 Software. All data were checked for normality (Kolmogorov–Smirnov test) and homogeneity of variance (Levene test) and then MANOVA analysis and factorial ANOVA were applied to check differences between treatments. Post-hoc testing was conducted with Tukey HSD test method.

2.2 Description of forest vehicle used The forwarder used in this study was a John Deere JD1110 D of 17.5 tons (displacement 4140 cm3 and 121 kW). The forwarder was equipped with 8 wheels

Nokian Forest Rider 700/50x26.5 tires inflated to 550 kPa and passed over the trail in one direction at the average speed of 4 km/h. The ground pressure data were: 47 kPa for front empty and loaded and 32 kPa for rear empty and 90 kPa for rear loaded. Descriptions of machinery are shown in Table 2. Table 3 shows the estimated traction data for the forwarder used according to Wästerlund (1992). Table 2 Description of the forwarder used

3. Results The control bulk density value was 0.83 g cm-3; the post-treatment bulk density increased slightly upwards to 0.87 g cm-3 in trail A and to 0.91 g cm-3 in trail B. The control porosity value instead was 68.8% and post-passes decreased to 67.2% in trail A and to 65.5% in trail B. Regarding resistance forces of the soil (i) the shear resistance value increased from 24.02 to 43.62 kPa in trail A and to 38.65 kPa in trail B. Also (ii) the penetration resistance value increased from 1.84 to 2.25 MPa in trail A and to 3.72 in trail B. The percentage variations for all parameters are shown in Table 4.

Table 3 Estimated traction data for the forwarder used Type

Unit

Mass

Forwarder 1110D Front empty

Front loaded

Rear empty

Rear loaded

10,500

10,483.4

7000

13,896.6

Wheels

Real mean ground pressure

kPa

Contact area

26,058

1/3 of contact area

17,372

8,686

Max. ground pressure

kPa

75.2

51.2

144

Croat. j. for. eng. 37(2016)2

235

M. Cambi et al.

Effects of Forwarder Operation on Soil Physical Characteristics: a Case Study in the Italian Alps (233–239)

Table 4 Percentage variations of physical parameters Trail

Passages

Bulk density

Porosity

Shear resistance

Penetration resistance

Litter + duff layer

–3

Unit

g cm

kPa

MPa

Control

0.83

68.8

24.02

1.84

3.5

+5%

–3%

+82%

+22%

–29%

+10%

–4%

+19%

+102%

–43%

Table 5 MANOVA analysis (Wilks value: 0.0312; df: 6, 530; p-value <0.001) for the four variables tested in three treatments. For the HSD results, different letters show difference between treatments Shear resistance

Penetration resistance

Bulk density

Porosity

Mean Std. dev. Std. err. HSD Mean Std. dev. Std. err. HSD Mean Std. dev. Std. err. HSD Mean Std. dev. Std. err. HSD Unit Total Diff. stat

kPa

270 35.430

kPa

MPa

g cm–3 g cm–3

g cm–3

8.900

0.542

2.603

0.916

0.056

0.870

0.012

67.2

7.6

0.5

0.202

Trail A

90 43.617

4.235

0.446

2.250

0.454

0.048

0.869

0.215

0.023 a, b

67.2

8.1

0.9

a, b

Trail B

90 38.650

2.494

0.263

3.722

0.506

0.053

0.914

0.251

0.027

65.5

9.5

1.0

Control

90 24.022

2.323

0.245

1.837

0.298

0.031

0.826

0.096

0.010 a, c

68.8

3.6

0.4

a, c

The results of MANOVA analysis, made for comparing the two trails and the control, show a significant statistical difference between all the variables tested (Tab. 5). In particular, regarding the analysis of the first 8.5 cm of soil, a significant statistical difference in bulk density and porosity was recorded between the control and trail B, while for the trail A the high SD value does not permit to consider the average value statistically different from both the control and the trail B. Focusing the analysis on the first 4 cm of soil, the shear and penetration resistance data highlighted a significant impact in both paths compared to the control It can be shown and highlighted how these parameters behave after the forwarder passes on the slope. In fact, the greater slope (trail A) was more influential on shear resistance and the minor slope (trail B) was influential on penetration resistance. A specific factorial ANOVA was applied to compare the two trafficked trails checking differences between the two forwarder tracks (right or left side). The results showed a statistical difference between the two trails only for shear and penetration resistance, while a different situation was revealed for the two tracks (Tab. 6). In the trail A, the low number of passes showed a statistically significant impact only for the shear and penetration resistance, with clear

236

differences between the tracks due to the trail transversal slope. In the trail B, the highest number of passes showed a statistically significant impact for all the variables studied, with low statistical differences between tracks.

4. Discussion and conclusions In the last decade, several studies have evaluated the effects of bogie tracks on forest sites (Bygdén et al. 2003, Sakay et al. 2008, Gerasimov and Katarov 2010), Table 6 Factorial ANOVA analysis (Wilks value: 0,646; df: 3, 174; p-value <0.001) for the four variables tested in two treatments. The underlined values show statistically significant differences between the two tires. R= right; L= left Trail

Shear resistance

Penetration resistance

Bulk density

Porosity

kPa

MPa

g cm–3

Track

46.8

40.4

2.4

2.1

0.89

0.85

38.9

38.4

3.5

3.9

0.85

0.98

Croat. j. for. eng. 37(2016)2

Effects of Forwarder Operation on Soil Physical Characteristics: a Case Study in the Italian Alps (233–239)

making comparison with wheel treatment, and the last study also considered a new design of tracked bogie (Edlund ed al. 2013). This paper has given an overview of the possible effects on the ground that can occur in the alpine environment, following timber extraction by a forwarder equipped with bogie tracks, considering two trail slopes, several passages and the differences between the two tracks due to transversal slope. These two aspects have not yet been analyzed for bogie track until now. The results obtained show a clear difference in soil physical parameters before and after trafficking. In particular, the four soil properties monitored in this research (bulk density, porosity, shear and penetration resistance) were suitable for assessing the impact of bogie tracks on the physical quality of soil. In trial A – with greater slope and minor forwarder passages – bulk density, porosity and penetration resistance did not change significantly, while shear resistance increased substantially. In trail B – with lower slope and major number of forwarder passages – all soil properties underwent significant changes, suggesting much higher deterioration of soil. With the increasing number of passes, the compacted zone deepened and partly collapsed, and there was a lateral bulging of the soil, as also verified by Gerasimov and Katarov (2010). On the trail with lower slope, the high value of the vertical force component was more stressed in the soil and this considerably affected the penetration resistance. With the increase of the slope, the resultant force (sum of the vertical and horizontal components) had a major effect on the shear resistance. The results of this study help to increase the knowledge about the impact of a high level of mechanization on forest soil. Other studies have been made on this topic with medium to large forwarders (20 – 38 t) used for wood transport (Jacobsen and Greacen 1985, Jansson and Johansson 1998, Eliasson 2005). These results can also concur for better assessing when different levels of mechanization can apply. In effect, as also found in other studies (Picchio et al. 2012, Marchi et al. 2014) that compared medium and high level of mechanization, high level of mechanization affected bulk density and porosity with similar percentages. It should be noted that, speaking of these specific variables, only the number of passes greatly affects the soil proprieties. The differences or rather the innovative aspects highlighted by the results of this study concern both parameters – penetration and shear resistance. The percentage values of impact found differ from those reported in the cited references and strictly connected to trail slope and number of passes. However, this was a preliminary investigation because only one Croat. j. for. eng. 37(2016)2

M. Cambi et al.

forest site was analyzed. In addition, there is no much information on any study of this topic for the Italian Alps. So, it is not possible to determine clearly the level and use of mechanization. The topic requires further studies on the effects of soil compaction due to bogie tracks. In particular, it would be interesting to have more trails to compare them with trails of the same slope and number of forwarder steps. Further research should focus on investigations: Þ to assess the effect on physical variations over time (in terms of fertility and soil degradation) Þ to evaluate the variations in term of soil microbial components after the forwarder passage Further information and data on these aspects could give a better overall picture of the impact on the environment.

5. References Alakukku, L., Weisskopf, P., Chamen, W.C., Tijink, F.G, van der Linden, J., Pires, S., Sommer, C., Spoor, G., 2003: Prevention strategies for field traffic-induced subsoil compaction: a review. Soil and Tillage Research 73(1): 145–160. Ansorge, D., Godwin, R.J., 2007: The effect of tyres and a rubber track at high axle loads on soil compaction, Part 1: Single axle-studies. Biosystems Engineering 98(1): 115–126. Arthur, E., Schjønning, 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. Berli, M., Kulli, B., Attinger, W., Keller, M., Leuenberger, J., Flühler, H., Springman, S.M., Schulin, R., 2004: Compaction of agricultural and forest subsoils by tracked heavy construction machinery. Soil and Tillage Research 75(1): 37–52. Bygdén, G., Eliasson, L., Wästerlund, I., 2004: Rut depth. soil compaction and rolling resistance when using bogie tracks. Journal of Terramechanics 40(3): 179–190. Bygdén, G., Wästerlund, I., 2007: Rutting and soil disturbance minimized by planning and using bogie tracks. Forestry Studies 46: 5–12. Cambi, M., Certini, G., Fabiano, F., Foderi, C., Laschi, A., Picchio, R., 2016: Impact of wheeled and tracked tractors on soil physical properties in a mixed conifer stand. IForest 9: 89–94. Cambi, M., Certini, G., Neri, F., Marchi, E., 2015: Impact of heavy traffic on forest soils: A review. Forest Ecology and Management 338: 124–138. Edlund, J., Keramati, E., Servin, M., 2013: A long-tracked bogie design for forestry machines on soft and rough terrain. Journal of terramechanics 50(2): 73–83. Eliasson, L., 2005: Effects of forwarder tyre pressure on rut formation and soil compaction. Silva Fennica 39(4): 549–557.

237

M. Cambi et al.

Effects of Forwarder Operation on Soil Physical Characteristics: a Case Study in the Italian Alps (233–239)

Gerasimov, Y., Katarov, V., 2010: Effect of bogie track and slash reinforcement on sinkage and soil compaction in soft terrains. Croatian Journal forestry Engeenering 31(1): 35–45. Han, H.S., Page-Dumroese, D.S., Han, S.K., Tirocke, J., 2006: Effect of slash, machine passes, and soil moisture on penetration resistance in a cut-to-length harvesting. International Journal of Impact Engeenering 17(2): 11–24 Keenan, R.J., Kimmins, J.P., 1993: The ecological effects of clear-cutting. Environmental Reviews 1(2): 121–144. 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. Johnson, C.E., Johnson, A.H., Huntington, T.G., Siccama, T.G., 1991: Whole-tree clearcutting effects on soil horizons and organic-matter pools. Soil Science Society of America Journal 55(2): 497–502. Jurgensen, M.F., Harvey, A.E., Graham, R.T., Page-Dumroese, D.S., Tonn, J.R., Larsen, M.J., Jain, T.B., 1997: Impacts of timber harvesting on soil organic matter, nitrogen, productivity and health of inland Northwest Forests. Forest Science 43(2): 234–251. Yong, R.N., Fattah, E.A., Skiadas, N., 1984: Vehicle traction mechanics. Developments in agricultural engineering 3. Amsterdam: Elsevier. Magagnotti, N., Spinelli, R., Güldner, O., Erler, J., 2012: Site impact after motor-manual and mechanised thinning in Mediterranean pine plantations. Ecological Engineering 113(2): 140–147. 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. Muro, T., 1982: Tyre/wheels and tracks state-of-the-art report. Journal of Terramechanics 19(1): 55–69. Okello, J.A., Dwyer, M.J., Cottrell, F.B., 1994: The tractive performance of rubber tracks and a tractor driving wheel tyres as influenced by design parameters. Journal of Agricultural Engineering Research 59 (1): 33–43. Picchio, R., Magagnotti, N., Sirna, A., Spinelli, R., 2012a: Improved winching technique to reduce logging damage. Ecological Engineering 47: 83–86.

238

Picchio, R., Neri, F., Petrini, E., Verani, S., Marchi, E., Certini, G., 2012b: Machinery–induced soil compaction in thinning two pine stands in central Italy. Forest Ecology and Management 285: 38–43. 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. Forestry Studies in China 13(4): 252–262. Pijuan, J., Comellas, M., Nogués, M., Roca, J., Potau, X., 2012: Active bogies and chassis levelling for a vehicle operating in rough terrain. Journal of Terramechanics 49(3): 161–171. Sanchez, F.G., Tiarks, A.E., Kranabetter, J.M., Page-Dumroese, D.S., Powers, R.F., Sanborn, P.T., Chapman, W.K., 2006: Effects of organic matter removal and soil compaction on fifth year mineral soil carbon and nitrogen contents for sites across the United States and Canada. Canadian Journal Forest Research 36(3): 564–575. Savelli, S., Cavalli, R., Baldini, S., Picchio, R., 2010: Small scale mechanization of thinning in artificial coniferous plantation. Croatian Journal of Forest Engineering 31(1): 11–21. Sakai, H., Nordfjell, T., Suadicani, K., Talbot, B., Bøllehuus, E., 2008: Soil compaction on forest soils from different kinds of tires and tracks and possibility of accurate estimate. Croatian Journal of Forest Engineering 29(1): 15–27. Syunev, V., Sokolov, A., Konovalov, A., Katarov, V., Seliverstov, A., Gerasimov, Y., Karvinen, S., Välkky, E., 2009: Comparison of wood harvesting methods in the Republic of Karelia. Working Papers of the Finnish Forest Research Institute 120: 117 p. Uusitalo, J., Ala-Ilomäki, J., Salomäki, M., Niemistö, P., 2011. New solutions in management and harvesting of peatland forests. Presentation at Oscar Seminar Soil and Machine Workshop, Hyytiälä, Finland. Wästerlund, I., 1992: Extent and causes of site damage due to forestry traffic. Scandinavian Journal of Forest Research 7(1–4): 135–142. Wong, J.Y., 1986: Computer aided analysis of the effects of design parameters on the performance of tracked vehicles. Journal of Terramechanics 23(2): 95–124. Wong, J.K., 1989: Terramechanics and off-road vehicles. Amsterdam, Elsevier: 251 p.

Croat. j. for. eng. 37(2016)2

Effects of Forwarder Operation on Soil Physical Characteristics: a Case Study in the Italian Alps (233–239)

M. Cambi et al.

Authors’ address: Assoc. prof. Enrico Marchi, PhD. e-mail: enrico.marchi@unifi.it Martina Cambi, PhD. e-mail: martina.cambi@unifi.it Francesco Neri, PhD. e-mail: francesco.neri@unifi.it University of Florence Department of Agricultural, Food and Forestry Systems Via San Bonaventura 13 50145. Florence ITALY Assoc. prof. Stefano Grigolato, PhD. e-mail: stefano.grigolato@unipd.it University of Padova Department of Land and Agriculture and Forestry Systems Viale dell’Università 16 35020 Legnaro ITALY

Received: March 30, 2015 Accepted: November 26, 2015 Croat. j. for. eng. 37(2016)2

Assoc. prof. Rodolfo Picchio, PhD. * e-mail: r.picchio@unitus.it University of Tuscia Department of Agriculture and Forestry Sciences (DAFNE) Via San Camillo de Lellis 01100 Viterbo ITALY * Corresponding author

239

Original scientific paper

Analysis of Productivity and Cost of Forwarding Bundles of Eucalyptus Logging Residues on Steep Terrain Sandra Sánchez-García, Elena Canga, Eduardo Tolosana, Juan Majada Abstract The objective of this study was to evaluate the productivity and costs of two Spanish forwarders, models Dingo AD-8468 and AD-2452, in the hauling of bundles of residues after Eucalyptus globulus clear cuts on steep terrain in Northern Spain. In addition, various models to predict time consumption for the main work elements and productivity were fitted including several independent variables previously selected using stepwise regression. Finally, the models explain between 83% and 97% of variability. Since the equations are based on simple variables (depending on each individual equation this was either velocity empty and loaded, slope loading, distance empty/loading/loaded or load per cycle), they will be a helpful and easy to use tool to assist in forest management planning. Productivity was 6.75 odt/PMH for the Dingo AD-8468 forwarder and 11.56 odt/PMH for the Dingo AD-2452. Cost per tonne for the Dingo AD-8468 was 6.77 €/odt compared to 3.94 €/odt for the Dingo AD-2452. Keywords: forwarder, time study, bundles, Eucalyptus globulus, displacements, GPS, steep slope

1. Introduction As noted by the European Union, the welfare of its citizens, the competitiveness of industry and the overall functioning of society depend on safe, secure, sustainable and affordable energy. EU energy policy objectives for 2020 are consequently ambitious, and indeed, the commitment continues beyond this date, aiming to reduce member state CO2 emissions by almost 40% by 2050 (CEC 2011) and to resolve obstacles and difficulties in developing actions to achieve these goals. A key problem is the high dependence on energy from outside the EU. One of the priorities established by the European Parliament in May 2013 was, therefore, to increase the diversification of the EU’s energy supply, particularly renewable energy options, and develop local energy resources to ensure energy security, i.e. a drive towards member states making better use of their own energy resources (EP 2013). Within the broad field of renewable energy, in recent years biomass has taken on great importance because of the potentially critical role it is assumed to Croat. j. for. eng. 37(2016)2

play in mitigating the effects of climate change (Viana et al. 2010). Also, according to recent reports from the Commission of the European Communities (CEC 2005, 2008), biomass has many advantages over conventional energy, in particular, its resilience to shortterm climate changes, the fact that it promotes and strengthens regional economic structures and provides alternative sources of income for farmers. However, there are complex issues to consider, such as the sustainability of production practices and the efficiency of bioenergy systems (IPCC 2014). Currently, the supply chain costs of forest biomass (extraction, pretreatment, forwarding and trucking) are considerable obstacles to the development of its bioenergy use, given that they are higher than for other fuels like petroleum or natural gas. To counter this situation would require guaranteed biomass sales, equipment adapted to the needs of forest biomass processing, and changes in operational procedures that can minimize these costs and integrate conventional timber harvesting systems and forest biomass harvesting systems.

241

S. Sánchez-García et al.

Analysis of Productivity and Cost of Forwarding Bundles of Eucalyptus ... (241–249)

The main factors that influence the total supply chain cost of forest biomass are the type of logging, the size of the area to be harvested, amount of resources, slope, infrastructure (i.e. harvesting and permanent tracks), impediments in the terrain (e.g. rocky outcrops) and the transport (forwarding and trucking) distance between the supply and demand point. Furthermore, due to these factors, there is not one single optimum harvesting system for all regions or all conditions. It is, therefore, essential to carry out studies of different operational systems, considering various conditions and types of forest and focus on the optimization of productivity and costs so as to make this type of forest biomass profitable. In the Iberian Peninsula, where currently Eucalyptus spp. stands cover approximately 1,125,000 ha (ENCE 2009), cost and productivity studies are essential to evaluate the viability and improve the efficiency of the harvesting of logging residues for energy purposes. Such harvesting not only provides a means for mobilizing a biomass resource, which would otherwise be »wasted« and would provide no income to the forest owner, but it is also an important intervention, both to reduce forest fire hazard and to improve access to the forest. In the north of Spain, Eucalyptus globulus stands are typically managed as coppice and harvested in clear cuts with the cut-to-length system (CTL), the felling being done manually and the processing mechanically, resulting in large amounts of residues, including bark left on the ground. Once the residue material had been collected together in the forest, the first difficulty encountered in its manipulation is its low density, which complicates and raises the price of its mobilization. For this reason, collection technology is often based on chipping the residues on-site to reduce size, or in creating compact bundles on-site to increase the density of the units for transportation. Bundling of residues was launched commercially in the study area at the beginning of 2007, and currently there are about 15 units producing bundles for fuel, mainly working with eucalypt logging residue collection and providing biomass to a single large power plant, whose consumption of such residues grew from 340,000 oven dry tonnes (odt) to a current level of 420,000 t after a recent expansion. Bundling has a great advantage over chipping in that it simplifies the logistics and storage of the biomass, especially when supplying fuel on a large scale to power plants (Johansson et al. 2006). These advantages initially made the bundling system a positive choice in Nordic countries (Kärhä and Vartiamäki 2006, Gustavsson et al. 2011, Eliasson 2011, Laitila et al. 2013), and also in Southern European countries

242

(Agudo 2010, Spinelli et al. 2011, Sánchez-García et al. 2015). However, bundling system costs must be analyzed in-depth, since, following such evaluations in Nordic countries, the tendency has changed to using chip systems rather than bundling. Nonetheless, in the area of Spain under study in this work, both professional opinion and technical studies (Sánchez-García et al. 2015) consider bundling to be an economically competitive technique in local Eucalyptus stands, which generally present some very particular conditions for forestry, e.g. steep terrain, small sized forest plots, and long haulage distances, making the collection of forest residues more difficult and expensive than in other scenarios. One advantage is that forwarding of bundles can be carried out with the same machinery that was in use in harvesting timber. The aim of this study was to determine the productivity and costs of two Spanish forwarder models (a Dingo AD-2452 and a Dingo AD-8468) in the hauling of bundles of residues from Eucalyptus globulus logging operations to a landing point. These specific forwarders, characterized by being small and light, have been designed and built by a factory in the north of Spain (www.dingoma.es) specifically to work under difficult conditions commonly found in this area, such as steep slope and narrow tracks. The study involves a time and cost study of each forwarder, and includes the fitting of equations to predict work element times and productivity as a function of different independent variables such as slope, distances and velocity.

2. Material and methods 2.1 Data collection The productivity and cost study of forwarding of eucalyptus bundles was carried out in 2 zones, and distributed in a total of 4 different stands (Table 1) choTable 1 Description of study sites Zone

623,091 – 623,478 Coordinates xmin – xmax 733,586 – 733,768 UTM ymin – ymax 4,814,643 – 4,814,858 4,832,966 – 4,833,262 Altitude, m Stand

Max

250

190

Min

Total area, ha

0.38

0.58

0.22

4.99

Age, years

44.3

41.8

15.0

Croat. j. for. eng. 37(2016)2

Analysis of Productivity and Cost of Forwarding Bundles of Eucalyptus ... (241–249)

sen to be as similar and thus as comparable as possible. Operators were skilled and had similar qualifications and experience in order to minimize operator effects in the study. In the clear cuts, the collection of residues was made by two bundler models: a Monra Enfo 2000 (zone 1) with a cutting device using shears and a John Deere 1490D (zone 2) with chainsaw cutting device. The bundles were forwarded to the roadside landing with the two Spanish forwarder models; a Dingo AD8468 (zone 1) and a Dingo AD-2452 (zone 2). At the landing, bundles were stacked side by side and then transported on standard timber trucks to the end-use facility. For the productivity analysis, four detailed time studies were performed. Each work cycle was divided into work elements (Table 3) and to avoid later mistakes, the work elements were clearly and concisely defined, setting the start and finish points. Data acquisition was conducted using the specific time study software UMT® (LAUBRASS Inc 2007). The time spent in each work element of the forwarder work was recorded on a Trimble Nomad handheld computer. A total of 14 hours and 24 minutes (26 cycles) were timed in the four different stands. In addition, certain parameters known to have a great influence on cycle time were also recorded (number of grabs and bundles per cycle, harvesting area, slope, disposal of residues, etc) and a GPS model XH Trimble Explorer was mounted on top of each forwarder to georeference its position on a continuous basis (every second), in order to obtain the slope, distance and velocity travelled as independent variables for each displacement (either empty, loading or loaded).

S. Sánchez-García et al.

Table 2 Specifications of forwarders Zone

Dingo AD-8468

Dingo AD-2452

Configuration

6x6

Load size, tonnes

8.5

13.5

Forwarder model

Engine model

DEUTZ BF6L 914 DEUTZ TCD 2012 L06

Power rating (Diesel), kW

Max. velocity km/h of displacement

40.0

Max. velocity of 1000 rpm work, km/h 2500 rpm Operative range of crane, m Box size

141

0.8 20.0 7.5

9.1

Length, mm

3600

4100

Width, mm

2100

2500

Oven dry tonnes (odt) were calculated using data from a previous study of bundler productivity performed in the same stands, that is, an average weight of bundles of 169 kg and 187 kg for zone 1 and zone 2, respectively.

2.2 Study of productivity and cost: model adjustment The timing data were reviewed to eliminate errors and outliers (Olsen et al. 1998) and then, time study data and additional data (influential parameters) were

Table 3 Description of work elements for forwarders hauling bundles of eucalyptus residues Work element

Description

Moving empty

Begins when work starts or after unloading of bundles at landing, the forwarder has to return to the work zone unloaded

Begins once the forwarder is at the side of the bundles to be loaded, displacement stops and crane arm begins to move or seat begins to turn in order to begin loading. It includes the time spent after the forwarder finishes loading the bundles from one pile and moves to the next pile, until the forwarder is fully loaded

Moving loaded

Once the box of the forwarder is full, it begins to move with the load to the landing

Unloading

At landing, the forwarder uses the crane to unload the bundles from its box. This activity includes small displacements required at landing in order to complete the unloading

Complementary work times

Action involving the crane and/or the machine, other than loading, unloading and displacement such as: handling bundles (at landing, stand or in the forwarder box), planning or accessing forest road

Refuel time

The portion of the service time used to refuel the machine; such as transporting to refuel, refuelling, etc.

Delays

Mechanical, operator or other delays

Others

All work elements not covered by the above categories

Croat. j. for. eng. 37(2016)2

243

S. Sánchez-García et al.

Analysis of Productivity and Cost of Forwarding Bundles of Eucalyptus ... (241–249)

Table 4 Specifications of forwarders Data

Dingo AD-8468 Dingo AD-2452

Investment, Euro

168,000

186,000

Service life, years

Maintenance and repair cost, €/h

6.01

6.65

Fuel cost, €/h

13.22

20.98

Lubricant cost, €/h

4.36

6.92

2000

Driver cost, €/h Total yearly utilization, SMH/year

combined into a single data set using proc SQL from SAS/STAT® (SAS Institute Inc 2004), and grouped by cycle number. The same procedure was then carried out to combine the GPS data with this timing data, using time (hours, minutes, seconds) as the common variable such that the accurate position of the forwarders was obtained at all times during the timings. The paths were analyzed using ArcGis 9.2, (ESRI 2006) filtering errors (for GPS precision or point clouds caused by a machine stop) and including points on the forest roads, where no points were recorded due to GPS signal loss. The point shape was transformed into polylines with XTools Pro tool of ArcGIS 9.2 (ESRI 2006), joining points included in the same work element and the same cycle. In this way, the slope, distance and velocity travelled by the machine (empty, loading or loaded) in each cycle were calculated. Productivity for each forwarder was estimated per hour by dividing the tonnes of residues extracted (odt, oven dry tonnes and gt, green tonnes) or number of

bundles, by productive hour (PMH, Productive Machine Hour). Machine costs were estimated with the method described by Miyata (1980) and employing the utilization rates according to Spinelli et al. (2004). The main cost assumptions, obtained directly from the company, are presented in Table 4. Different models were fitted by regression analysis to predict time consumption of the main work elements, using the influential parameters measured as independent variables. A dummy variable was used to take into account the model of forwarder as an independent variable. To select the variables to include in each model, step-by-step regression was used, implementing the stepwise command of the REG procedure of SAS/STAT® (SAS Institute Inc, 2004). The best models were selected using goodness-of-fit statistics (R2 and RMSE) and graphical analysis, as well as taking into account the simplicity of the selected model. In addition, a model that combined the models of work element times was developed to predict productivity.

3. Results and discussion 3.1 Study of productivity and cost Considering both forwarders, Table 5 shows the descriptive statistics of work element times in hauling of bundles and the percentage of time spent in each work element time in relation to total time. For 93% of cycle time, forwarders were involved in main work times, specifically moving empty, loading, moving loaded and unloading. The mean time consumption per cycle was 33 minutes and 13 seconds.

Table 5 Descriptive statistics of work element times, hh:mm:ss and percentage of total time,% Work element times

N cyles

Avg.

Min

Max

Std. Dev.

Moving empty

0:07:06

0:02:37

0:16:39

0:04:16

21.36

0:09:44

0:05:35

0:15:56

0:03:24

29.28

Moving loaded

0:07:46

0:02:29

0:14:22

0:03:32

23.36

Unloading

0:06:01

0:03:12

0:09:18

0:02:29

19.05

Complementary work times

0:01:18

0:00:06

0:04:05

0:01:23

2.86

Refueling

0:10:22

0:00:00

1.42

Delays

0:04:43

0:01:09

0:13:08

0:05:42

2.58

Others

0:00:39

0:00:00

0.09

Cycle time

0:33:13

0:23:34

0:48:27

0:06:49

100.00

244

Croat. j. for. eng. 37(2016)2

Analysis of Productivity and Cost of Forwarding Bundles of Eucalyptus ... (241–249)

S. Sánchez-García et al.

Table 6 Descriptive statistics of load per cycle, distances travelled, slope and velocity Descriptive parameters by model machine

Dingo AD-8468

Dingo AD-2452

N cyles

Avg.

Min

Max

Std. Dev.

Load/cycle, odt

3.51

2.70

4.22

0.47

Load/cycle, gt

4.95

3.82

5.96

0.66

Distance empty, m

1467.31

940.69

2316.52

555.07

Distance loaded, m

1129.02

258.21

1543.77

403.41

Distance between load piles, m

64.21

19.37

146.74

44.52

Slope empty, %

15.18

9.28

23.94

6.14

Slope loaded, %

14.12

11.14

17.74

2.18

Slope between load piles, %

14.42

3.35

37.74

10.17

Velocity empty, km/h

10.50

6.40

15.67

3.25

Velocity loaded, km/h

6.73

1.07

8.92

2.16

Velocity between load piles, km/h

2.00

1.25

3.27

0.64

Load/cycle, odt

5.92

5.23

6.73

0.46

Load/cycle, gt

11.53

10.20

13.11

0.89

Distance empty, m

334.85

133.67

480.04

118.70

Distance loaded, m

385.23

257.64

517.91

75.51

Distance between load piles, m

117.24

61.17

274.44

66.67

Slope empty, %

15.04

8.52

20.94

4.192

Slope loaded, %

15.15

9.70

23.67

4.07

Slope between load piles, %

15.21

8.38

22.35

4.08

Velocity empty, km/h

4.66

2.57

6.06

1.03

Velocity loaded, km/h

5.12

4.53

6.23

0.53

Velocity between load piles, km/h

1.79

1.33

2.42

0.28

Descriptive statistics of main influential variables for both machines are presented in Table 6. The main results of the time study are shown in Table 7. The differences between empty and loaded moving times are due to the different topography and forwarding distances involved in the two work zones. These times were higher in zone 1, where the forwarder (Dingo AD-8468) worked on zig-zagging tracks with greater slopes (see Table 1) making manoeuvring more difficult, and distances being longer (see Table 6). In contrast, in zone 2, the forwarder (Dingo AD2452) worked within the forest due to good conditions and low degree of slope. Time consumption in moving between loadings in zone 1 (Dingo AD-8468) was lower because the bundles were grouped and located only at the sides of the Croat. j. for. eng. 37(2016)2

track, whereas in zone 2 the bundles were grouped but distributed across the whole area due to the more suitable topography. This is in accordance with McNeel and Rutherford (1994), who reported that this time parameter is strongly influenced by the distribution of bundles, and is lower when bundles are in groups of an optimum size. Regarding loading and unloading times, they are longer or the Dingo AD-2458 (around 40%) due to the larger box size, which allows a greater load per cycle according to the study of Jiroušek et al. (2007) and its larger crane allows it to have a greater working radius, meaning that an increased number of bundles can be collected from the same point (Laitila et al. 2009). In this study, the productivity for the Dingo AD2452 forwarder working in zone 2 was 11.56 odt/PMH

245

S. Sánchez-García et al.

Analysis of Productivity and Cost of Forwarding Bundles of Eucalyptus ... (241–249)

Table 7 Main results of time studies between forwarder models

and for the Dingo AD-8468 forwarder working in zone 1, it was 6.75 odt/PMH. The difference is due to the lower load size and more difficult topography in the latter scenario. This is in accordance with the work by Jiroušek et al. (2007), where the productivity of a forwarder increased with improvement in harvesting area conditions, since this implies better track conditions and better accessibility to collect the bundles. The number of bundles extracted per cycle was very close to the results obtained by other authors (see Table 7). Eriksson (2008) recorded 25 bundles extracted per cycle with a size of forwarder similar to the Dingo AD-2452, and Laitila et al. (2009) obtained 24 bundles per cycle (minimum 8 bundles, maximum 31) using a similar machine. Also, Kärhä et al. (2010) obtained 22 bundles per cycle, with a smaller forwarder that had similar technical characteristics to the Dingo AD-8468. Assuming a utilization rate of 70%, the hourly cost of the Dingo AD-8468 and Dingo AD-2452 was 39.99 €/SMH and 44.98 €/SMH, respectively. These results were very closer to those obtained with similar forwarders by Spinelli et al. (2004), i.e. 38.6 €/SMH and 57.4 €/SMH, but slightly below those obtained by Agudo (2010), i.e. 62.91 €/SMH, for a forwarder with similar characteristics to the Dingo AD-2452. Cost per oven dry tonne varied between 3.94 to 6.77 €/odt (see Table 7).

Dingo AD-8468

Dingo AD-2452

Moving empty, min

8.94

4.43

Loading, min

7.20

13.38

Moving loaded, min

9.94

4.61

Unloading, min

4.36

8.21

Main work time/cycle, min

31.11

30.63

Complementary work times, min

1.59

0.46

Delay times, min

5.90

1.15

Other times, min

0.66

–

Cycle time, min

34.79

30.96

Productive time (PMH), min

32.58

30.83

Non-productive time, min

7.18

1.15

gt/PMH

9.52

22.51

Odt/PMH

6.75

11.56

Bundles/PMH

39.94

61.78

Bundles/cycle

€/PMH

47.11

52.86

€/SMH

39.99

44.98

3.2 Model adjustment

€/odt

6.77

3.94

€/gt

4.79

2.02

Different models to predict main work times (moving empty and loading, loading and unloading) were evaluated using linear regression and selecting the independent variables by step-by-step regression. The equations finally proposed and goodness-of-fit statistics of each model are presented in Table 8. All parameters were significant at the 5% level.

Variables

PMH – Productive Machine Hour odt – oven dry tonnes € – euros gt – green tonnes

Table 8 Equations finally proposed and goodness-of-fit statistics of each model Work element times

Equation

RMSE

Moving empty, min

Eq. 1

tmoving empty = 4.633 − 0.5274 × Vempty + 0.00677 × Dempty

0.77

0.97

Loading, min

Eq. 2

tloading = 11.15 − 0.06635 × Sloading + 0.02381× Dloading

1.51

0.83

Moving loaded, min

Eq. 3

tmoving loaded = 10.53 − 1.844 × Vloaded + 0.01066 × Dloaded

1.15

0.91

Unloading, min

Eq. 4

tunloading = 1.569 + 0.5726 × Lcycle

0.58

0.92

RMSE – Root Mean Square Error – Coefficient of Determination R2 t – time (minutes) Vempty/loaded – velocity when the machine is empty or loaded (km/h) Dempty/loading/loaded – distance travelled when the machine is empty, loading, or loaded (meters) Sloading – slope (percentage) – load hauled per cycle (green tonnes) Lcycle

246

Croat. j. for. eng. 37(2016)2

Analysis of Productivity and Cost of Forwarding Bundles of Eucalyptus ... (241–249)

S. Sánchez-García et al.

Total hourly cost of the Dingo AD-8468 was 3 9.99 €/SMH and 44.98 €/SMH of the Dingo AD-2452. Equations were developed to predict different work element times as a function of independent variables (Vempty/loaded, velocity; Dempty/loading/loaded, distance travelled; Sloading, slope; Lcycle, load hauled per cycle) depending on the equation, which represented between 83% and 96% of total variability. A productivity equation was developed based on main work element times as the sum of individual times employing the model fitted. These equation can also be used to calculate costs (€/gt) using the costs in €/SMH and the specific utilization rate.

Fig. 1 Plot of predicted values against observed values for the productivity equation (Eq. 5)

Taking into account that in this type of equation the common value of R2 is around 0.5 (Olsen et al. 1998), the time consumption models explained a high percentage of total variability (between 83 and 97%). Main work time could be calculated as the sum of individual times estimated employing the models fitted (Equations 1–4 in Table 8), so it is possible to develop a productivity model following Equation 5. Productivity =

60 ⋅ Lcycle

 gt  MH  tmoving empty + tloading + tmoving loaded + tunloading  PM

(5)

Where: Lcycle load hauled per cycle (green tonnes) t time in minutes. Plot of the values predicted (gt/PMH) from the productivity equation (Eq. 5) against the observed values are shown in Fig. 1.

4. Conclusions Productivity and costs of forwarding bundles of residues on steep terrain were calculated for two different models of forwarder, a Dingo AD-8468 and a Dingo AD-2452. Productivity in oven dry tonnes per productive time was 6.75 odt/PMH and 11.56 odt/ PMH for the Dingo AD-8468 and Dingo AD-2452, respectively. Croat. j. for. eng. 37(2016)2

Under difficult working conditions, such as in the study area (steep terrain, limited infrastructure, long forwarding distance), these results will be of great practical help in terms of improving logging planning, and consequently for performing and achieving cost competitiveness of the system of eucalyptus logging residues collection. Since specific terrain conditions of forest harvesting operations have such a significant effect on this type of machine (forwarder), these equations should be extended to include various other scenarios.

Acknowledgements The authors are grateful to the Severo Ochoa Asturian Fellowship (subsidized by the Government of the Principality of Asturias) awarded to SSG, and to the Singular Strategic Proyect ECOCOMBOS (subprojects LOGIFOR and MEMAP); Ministry of Education and Science. Also special thanks to the companies Forestal Hosintra and Edelmiro López Rodríquez that worked with us on this study. Also thanks to Alejandro Vivas for his technical support and Ronnie Lendrum for her essential technical English assistance.

5. References Agudo, R., 2010: Empacado discontinuo a pie de tocón de residuos selvícolas: gestión integral de biomasa forestal. Tesis doctoral. (Discontinuous bundling of forest residues at stump site: integral management of biomass). Doctoral Thesis. Universidad de Córdoba. Servicio de Publicaciones de la Universidad de Córdoba, Campus de Rabanales. Ctra. Nacional IV, km. 396, 14071 Córdoba. http://helvia.uco.es/ xmlui/handle/10396/3519 (Accessed 2 May 2015). CEC 2005: Commission of the European Communities. Communication from the Commission. Biomass action plan. COM (2005) 628 final. Brussels, 7.12.2005. http://www.ebbeu.org/legis/Biomass_action_plan_en_07122005.pdf (Accessed 15 April 2015). CEC 2008: Commission of the European Communities. Communication from the Commission to the European Par-

247

S. Sánchez-García et al.

Analysis of Productivity and Cost of Forwarding Bundles of Eucalyptus ... (241–249)

liament, the Council, the European Economic and Social Committee and the Committee of the Regions. 20 20 by 2020 – Europe’s climate change opportunity. COM (2008) 30 final. Brussels. 23.01.2008. http://www.europarl.europa.eu/RegData/docs_autres_institutions/commission_europeenne/ com/2008/0030/COM_COM%282008%290030_EN.pdf. (Accessed 22 June 2015). CEC 2011: Commission of the European Communities. Communication from the Commission to the European Parliament, the Council, the European Economic and the Social Committee and the Committee of the Regions. Energy Roadmap 2050. COM(2011) 885 final. Brussels, 15.12.2011. http:// eurlex.europa.eu/LexUriServ/LexUriServ.do?uri=COM: 2011:0885:FIN:EN:PDF (Accessed 15 April 2015). Eliasson, L., 2011: Follow-up of the John Deere logging residue bundler. In: Efficient forest fuel supply systems. Composite report from a four year R&D programme 2007–2010 (Thorsén, Å., Björheden, R., Eliasson, L., eds). Skogforsk. http://www.skogforsk.se/contentassets/13f65170eaa5477b84 2f4d2f3de7b282/ess-2007-2010-eng-low.pdf (Accessed 27 May 2015). ENCE 2009: La Industria del eucalipto en España. Seminario: La Industria Forestal Española. (The Eucalyptus Industry in Spain. Seminar: The Spanish Forestry Industry) 5º Congreso Forestal Español. 22 September, Ávila, Spain. http:// www.congresoforestal.es/fichero.php?t=12225&i= 469&m=2185.pdf [in Spanish]. (Accessed 15 December 2014). EP 2013: European Parliament. Current challenges and opportunities for renewable energy in the European internal energy market. 21.05.2013. P7_TA(2013)0201. http://www. europarl.europa.eu/RegData/seance_pleniere/textes_adoptes/definitif/2013/05-21/0201/P7_TA(2013)0201_1_EN.pdf (Accessed 12 April 2015). Eriksson, L., 2008: Forest-fuel systems comparative analyses in a life cycle perspective. Doctoral Thesis. Mid University Sweden, 76 p. ESRI 2006. Environmental Systems Research Institute, Inc. (ESRI). ArcGIS software versión 9.2. 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. IPCC 2014: Summary for Policymakers. In: Climate Change: Mitigation of Climate Change. Contribution of Working. Group III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (Edenhofer, O., PichsMadruga, R., Sokona, Y., Farahani, E., Kadner, S., Seyboth, K., Adler, A., Baum, I., Brunner, S., Eickemeier, P., Kriemann, B., Savolainen, J., Schlomer, S., von Stechow, C., Zwickel, T., Minx, J.C. eds). Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. Jiroušek, R., Klvač, R., Skoupý, A., 2007: Productivity and costs of the mechanised cut-to-length wood harvesting sys-

248

tem in clear-felling operations. Journal of Forest Science 53(10): 476–482. Johansson, J., Liss, J.E., Gullberg, T., Bjorheden, R.B., 2006: Transport and handling of forest energy bundles. Advantages and problems. Biomass and Bioenergy 30(4): 334–341. Kärhä, K., Vartiamäki, T., 2006: Productivity and costs of slash bundling in Nordic conditions. Biomass and Bioenergy 30(12): 1043–1052. Kärhä, K., Jylhä, P., Laitila, J., 2010: Integrated procurement of pulpwood and energy wood from early thinnings using whole-tree bundling. Biomass and Bioenergy 35(8): 3389– 3396. 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. Laitila, J., Kilponen, M., Nuutinen, Y., 2013: Productivity and cost-efficiency of bundling logging residues at roadside landing. Croatian Journal of Forest Engineering 34(2): 175– 187. LAUBRASS Inc. 2007. UMT PLUS® Software. User’s Guide. Umt Manager and StatUmt programs, Version 16.7. 197 p. McNeel, J.F., Rutherford, D., 1994: Modeling harvester-forwarder system performance in a selection harvest. Journal of forest engineering 6(1): 7–14. Miyata, E.S., 1980: Determining fixed and operating costs of logging equipment. USDA Forest Service, North Central Forest Experiment Station. General Technical Report NC-55. St. Paul, Minnesota. 16 p. Olsen, E., Hossain, M., Miller, M., 1998: Statistical Comparison Of Methods Used In Harvesting Work Studies. Research Contribution 23. Forest Research Laboratory. Oregon State University, 45 p. Sánchez-García S., Eliasson L., Tolosana E., Majada J., Canga E., 2015: Evaluation of technological improvements in bundling units for the collection of eucalyptus logging residues in Northern Spain. Forest Systems 24(2): e030, 8 pages. SAS Institute Inc. 2004. SAS/STAT®. 9.1. User’s Guide. SAS Institute Inc., Cary, NC. Spinelli, R., Magagnotti, N., Picchi, G., 2011: A supply chain evaluation of slash bundling under the conditions of mountain forestry. Biomass and Bioenergy 36: 339–345. http:// dx.doi.org/10.1016/j.biombioe.2011.11.001 (Accessed 12 February 2015) Spinelli, R., Owende, P.M.O., Ward, S.M., Tornero, M., 2004: Comparison of short-wood forwarding systems used in Iberia. Silva Fennica 38(1): 85–94. Viana, H., Cohen W.B., Lopes, D., Aranha, J., 2010: Assessment of forest biomass for use as energy. GIS-based analysis of geographical availability and locations of woodfired power plants in Portugal. Applied Energy 87(8): 2251–2560.

Croat. j. for. eng. 37(2016)2

Analysis of Productivity and Cost of Forwarding Bundles of Eucalyptus ... (241–249)

S. Sánchez-García et al.

Authors’ address: Sandra Sánchez-García, PhD. student* e-mail: ssanchez@cetemas.es Elena Canga, PhD. e-mail: ecanga@cetemas.es Juan Majada, PhD. e-mail: jmajada@cetemas.es CETEMAS, Forest and Wood Technology Research Centre Sustainable Forest Management Area Finca Experimental »La Mata« s/n 33820 Grado, Asturias SPAIN

Received: October 26, 2015 Accepted: March 30, 2016 Croat. j. for. eng. 37(2016)2

Prof. Eduardo Tolosana, PhD. e-mail: eduardo.tolosana@upm.es E.T.S.I. Montes, Technic University of Madrid Ciudad Universitaria s/n 28040 Madrid SPAIN * Corresponding author

249

Original scientific paper

Analysis of Accuracy of Evaluating the Structure of a Harvester Operator’s Workday by Work Sampling Grzegorz Szewczyk, Janusz M. Sowa, Jiri Dvořák, Krzysztof Kamiński, Dariusz Kulak, Arkadiusz Stańczykiewicz Abstract The study covered an analysis of the accuracy level of measuring time within a working shift using the method of regular snapshot observation at a harvester operator’s worksite in Scots pine stands. A conformance level of the analyzed methods was evaluated through assessing the accuracy of rectilinear fitting of time structures, established using the photography of work day method and snapshot observations. The accuracy of snapshot measurements performed in 3-minute intervals was determined as high, exceeding 95%. Increasing the time interval between observations to 10 or 15 minutes resulted in higher estimation error in snapshot observation time, ranging between 5 and 10% for late thinned and clear-cut stands. The accuracy of evaluating proportions of specific work times within a working shift, in regular snapshot observations, was correlated with work cyclicality. The strongest work cycle in thinned stands consisted of 43 activities, with total duration of 13 minutes, whereas in clearcut stands it comprised 45 activities, with total duration of 15 minutes. One of the advantages of the described method, apart from its lower labour intensity as compared to working day photography, was the possibility to assess labour time and breaks as well as estimate the share of downtime. Keywords: work measurement, time series, snapshot observation, timber harvesting

1. Introduction Work measurement is a two-stage examination, including measuring time during work and determining a number of work products after completing the production process. Most methodological errors are made in the first stage, i.e. at determining the time of work. Field research of technology of timber harvesting and skidding are normally conducted within the operational work time, i.e. for activities directly related to the work being done, and then an assumption of their proportions in the entire working shift is made (Backhaus 1990, Sowa et al. 2006, Zečić et al. 2005, Nurminen et al. 2006, Spinelli and Visser 2008). The share of main work time within a working shift is usually estimated based on a labour-intensive time study, i.e. measurement of recurrent work sequences (Szewczyk et al. 2014). Croat. j. for. eng. 37(2016)2

Evaluating the proportions of preparatory and complementary work times is even more troublesome, due to their irregular occurrence. Therefore, they are estimated based on observations of the entire working shift. This method is known as photography of work day (Monkielewicz and Czereyski 1971, Samset 1990, Häberle 1992, Kärhä et al. 2004, Moskalik 2004, Ovaskainen et al. 2004, Picchio et al. 2012, Sowa et. al. 2007, Szewczyk 2010, 2011a). Changeable and complex production processes, typical of forest works, require a suitable measuring method. In this respect, methodologies involving an estimation of frequency of certain activities may appear extremely interesting when many worksites are subjected to investigation, in particular, at early stages of labour research (Szewczyk 2014b). The method of snapshot observations was introduced into studies on work in industry as early as in

251

G. Szewczyk et al.

Analysis of Accuracy of Evaluating the Structure of a Harvester Operator’s Workday ... (251–259)

1930s. The grounds for this methodology, called »snapreading«, were established in 1927 by the British statistician, Tippett (1935). Work analyses performed on similar basis were described as »ratio-delay« (Morrow 1957) or »work sampling« (Brisley 2001), while in Poland they were popularised as »snapshot observations«, at the end of 1950s (Wołk 1960). Work measurement by means of snapshot observations consists in recording events taking place in a constant or varying time interval. The frequency of particular activities encountered at a given worksite translates into a proportion (share) of their durations within the entire time of observation (Wołk and Strzelecki 1993, Szewczyk 2014a). An essential problem recognised for the method in question is determining the time interval, within which the measurements should be taken. This paper presents an analysis of the possibility to extend the time interval between snapshot observations in work sampling at timber harvesting. Such methods have already been applied in forestry sciences, though they still require thorough testing under varied conditions of logging operations (Miyata et al. 1981). The research aimed to assess the possibility of increasing the time interval between snapshot observations recorded at a harvester operator’s worksite and determining their accuracy in various time intervals. The reference point was the time structure of a working shift, obtained using the photography of work day method. In the initial stage of the research, it was as-

sumed that differences would occur in the accuracy of evaluating time consumption for various time intervals in snapshot observations, and that a factor would be indentified related to the cyclicality of work activities and determining the measurement accuracy.

2. Material and Methods Trial plots were located within the Regional Directorate of the State Forests in Radom, Staszów Forest District (21˚10’E, 50˚33’N), and the Regional Directorate of the State Forests in Łódź, Kolumna Forest District (19˚12’E, 51˚36’N). At these plots, studies of efficiency of mechanised timber harvesting were carried out in 2009–2011, and based on them primary assumptions for methodology of work sampling with the use of snapshot observations were developed (Szewczyk 2014b). The research presented in this paper covered additional, supplementary work measurements. The selected taxation features of the investigated stands were gathered in Table 1. Wood extraction was mechanised, employing mid-class harvesters Ponsse Ergo (Jiroušek et al. 2007, Dvořák et al. 2011). There were three operators aged 30 to 40, with more than one-year experience. In the course of works performed in the CTL system, the following assortments were bucked: longwood with a length ranging from 8 to 12 m, logged timber

Table 1 Selected taxation features of the investigated stands Forest district

Cutting category*

Species composition

Age years

Stocking index

DBH cm

Large timber in total m3/ha

Total removal of large timber m3/ha

Pine 10

0.8

300

Pine 10

112

Oak locally

0.9

336

Birch locally

Pine 10

361

Pine 8

280

Oak 2

CC Staszów LT LT

LT Kolumna

Pine 9 Birch 1

323

Pine 10

355

Pine 10

0.9

318

Pine 10

0.9

300

*CC – clear cutting; LT – late thinning; ET – early thinning

252

Croat. j. for. eng. 37(2016)2

Analysis of Accuracy of Evaluating the Structure of a Harvester Operator’s Workday ... (251–259)

G. Szewczyk et al.

Table 2 Flow chart presenting classification of harvester work times. Numerical codes referring to the applied classification were used for constructing time series Main work time

Pulling out a crane arm, positioning, cutting, felling Pulling a tree onto a machine, delimbing, bucking Time of relocation around the workplace

WP Workplace time

2 Activity code

Complementary work time

Time of preparing a worksite – removing branches and fragments of logs hindering access to the tree being cut Time of arranging logs, sorting bucked assortments

Maintenance time

Repair time

Non-working time

Time of changing a cutting chain, refuelling, technological adjustments Time of unblocking a harvester head, removing branches Times of deleting faults in hydraulic system Rest time Meal time

with a length of 4 m and middle-sized wood in a form of 2.5 meter rollers. In fragments of the stands, uniform in terms of their taxation features (DBH, height, species composition), work time study was conducted using the photography of workday method (Szewczyk 2011b, 2012, Nurminen et al. 2006, Dvořák and Walczyk 2013). Time measurement was taken automatically by means of PSION Workabout device with »Timing« software, developed especially for chronometric analyses (Sowa et al. 2007, Sowa and Szewczyk 2013). The duration of specific work activities were recorded (accurate to 1 second) and then classified according to the IUFRO standards (Björheden 1991, Dvořák et al. 2011, Szewczyk et al. 2014) (Table 2). For selecting a homogenous trial sample, the significance of differences in mean time of relocation around the workplace area was tested. From the chronometric sequences, chosen in the above-described manner and obtained by means of the photography of work day method, time categories, theoretically observed in snapshot observations, were sampled. Testing was performed for 3-, 5-, 10-, 15-, 20- and 25-minute intervals between subsequent observations. A simple correlation between percentages of specific time categories from the above-mentioned database was computed and used as the measure of conformance of time structures, determined with the use of both methods, photography of work day and snapshot observations. Since the research covered an analysis of a linear model without an absolute term, a directly proportional relationship between both of the variables was put to the test. Two measures of Croat. j. for. eng. 37(2016)2

fitting accuracy were assumed: a coefficient of regression and a coefficient of determination, with error rate at 5%. The accuracy of work time measurement, assessed with the use of snapshot observations in the given time interval, was correlated with the cyclicality of work activities. This cyclicality was determined through displaying the measurement data as time series, i.e. a sequence of observations for a certain variable in a constant time interval (Box and Jenkins 1976). For converting the chronometric sequences recorded at the investigated worksites into time series, particular observations were encoded using the notation presented in Table 2. Such constructed time series contained the names of respective activities expressed as numerals and playing the role of an observed variable, while the succession of activities (work sequencing), typical of a certain worksite, became an ordering variable (Szewczyk 2011a, Szewczyk et al. 2014, Szewczyk 2014a). The difficulty in adjusting the method of time series analysis to the investigated phenomenon consisted in the fact that particular cases (observed work activities) did not occur in a stable time interval. For displaying the results of time study, conducted at the selected work sites, in a form of time series, successive observations were encoded according to the scheme presented in Table 2. Thus, the occurrence of sequences of work activities was considered as time series, where individual activities, with their names encoded as numbers, were an observable variable, while the succession of activities typical of the analyzed work sites (work sequencing) constituted an ordering vari-

253

G. Szewczyk et al.

Analysis of Accuracy of Evaluating the Structure of a Harvester Operator’s Workday ... (251–259)

able (Szewczyk 2011a, Szewczyk et al. 2014, Szewczyk 2014a). Since the examined time series revealed random fluctuations, their moving average (of equal weights) was smoothed. Time series, in contrast to random samples, are characterised by a non-accidental order of observations. This attribute was used in the present study. The simplest examination of sequencing of certain observations is based on identifying regular fluctuations, i.e. with fixed lengths (expressed by a number of observations), determined as seasonal fluctuations. This simplified approach was assumed by Szewczyk in his studies on establishing the length of work cycles of forest machines. With regard to work sites characterised by a great variability of work (e.g. at timber harvesting or skidding), the length of stable and recurrent fragments of a workday was difficult to determine. In fact, the structure of work time succession at work sites related to timber harvesting and skidding indicated an occurrence of two overlapping work cycles with various lengths. For detecting a variable of this cyclical structure of the investigated time series, methodology of the single-band Fourier analysis was employed (Kot et al. 2007, StatSoft Inc. 2009). Lengths of recurrent work cycles were estimated based on the entire measurement database. The operation of cutting a tree was assumed as the beginning of each cycle, whereas its total duration was a sum of durations of successive activities, the number of which determined the total cycle length. The beginning of next cycle was marked with another cutting operation. During the field research, nine working shifts were measured, three per each category of operation. The measurement database contained 10,193 records documenting durations of the distinguished activity categories, while the measurements were taken continuously within more than 86 hours. An analysis of homogeneity of the research sample was performed based on the variance analysis of the time of relocation between successive worksites. The obtained chronometric sequences did not reveal statistically significant differences in respect of the mean value of relocation time; therefore further study was carried out using the entire measurement sample. The tested database for snapshot observations enclosed 731 records for early thinned stands, 589 for late thinned ones and 652 for clear-cut stands.

3. Results The lowest share of operational work times was recorded in early thinned stands (0.63), where work

254

Fig. 1 Time structure of a harvester working shift conditions were far more difficult when compared with those of late thinnings. The graph in Fig. 1 presents the time structure of a harvester working shift. With regard to stands of all categories, work times within the operational work time, i.e. main work times (40.2–52.6%) and ancillary work times (15.9–37.7%), had the highest share. The fact that over 40% of all activities were those directly related to work proved that the manipulation areas were very well prepared and the training level of operators was suitable. The highest share of ancillary work times was recorded in late thinned stands (37.7%), which was justified by longer time of relocation between subsequent work sites. The share of times of repairs and technical maintenance, in the technological variants under scrutiny, accounted for ca. 10%, while the share of times for rest and physiological needs amounted to 13%. A slightly higher share of complementary work times was recorded in early thinned stands, which may be explained by harder work conditions. Activity categories in the analyzed time intervals were sampled, with the use of EXCEL spreadsheets, from the entire database obtained by means of the snapshot observation method. Sampling was conducted using the LOOKUP logical function (vector form), which searches for given data in a one-row or onecolumn range (vector), and returns the value in the same position in another one-line or one-column range. Croat. j. for. eng. 37(2016)2

Analysis of Accuracy of Evaluating the Structure of a Harvester Operator’s Workday ... (251–259)

G. Szewczyk et al.

Table 3 presents a comparison between databases created using the photography of work day method and those based on the selected variants of snapshot observations.

Table 4 contains the accuracy parameters of the time structure assessment obtained using the snapshot observation method. The performed analyses demonstrated that the time structures, determined upon application of both of the investigated methods in 3-, 5-, 10- and 15-minute intervals, were very strongly correlated (R2ε≤0.70; 0.99≥) (Stanisz 2007). Moreover, the relationship between the time structures for these time intervals was nearly directly proportional (Rε≤0.86; 1.00≥). Thus, the accuracy of the snapshot observation method for the above-mentioned time intervals was proved to be considerable. In respect of other investigated time intervals, the estimation error was higher. Fourier spectral analysis revealed an occurrence of overlapping cycles of work activities within the investigated working shifts, differing in terms of their

Fig. 2 Fragment of a periodgram of harvester work in early thinning. The highest peak of the periodgram indicates the strongest work cycles

Fig. 3 Fragment of a periodgram of harvester work in late thinning. The highest peak of the periodgram indicates the strongest work cycles

Table 3 Number of observations in the tested variants Early thinned stands

Late thinned stands

Clear cutting stands

Photography of work day, pc.

4229

3129

2703

Observation time interval, min

399

200

132

Snapshot observations, min Share of snapshot observations,%

Table 4 Accuracy parameters of the time structure assessment obtained using the snapshot observation method for the selected observation time intervals Cutting category, working shift (R2/R) Observation time interval, min

ET1

ET2

ET3

LT1

LT2

LT3

CC1

CC2

CC3

0.99/0.99

0.99/1.00

0.94/0.87

0.99/0.98

0.96/0.93

0.99/0.97

0.98/0.96

0.99/0.98

0.96/0.97

0.98/1.03

0.98/0.98

0.83/0.86

0.97/0.96

0.96/0.94

0.74/0.97

0.99/0.95

0.96/0.91

0.93/0.92

0.95/0.98

0.70/1.00

0.93/0.90

0.96/0.94

0.95/0.92

0.73/0.96

0.96/0.90

0.98/1.00

Croat. j. for. eng. 37(2016)2

255

G. Szewczyk et al.

Analysis of Accuracy of Evaluating the Structure of a Harvester Operator’s Workday ... (251–259)

4. Discussion

Fig. 4 Fragment of a periodgram of harvester work in clear-cutting. The highest peak of the periodgram indicates the strongest work cycles lengths. The highest peaks of periodograms, presented in Figures 2–4, were typical of cycles having the strongest impact on the observed general variability (Table 5). Lengths of work cycles and their durations within the analyzed groups of stands were presented in Table 5. Due to the skewness of distribution of work activity durations, a median value was given as a cycle duration. With regard to all the investigated stands, the lengths of work cycles were similar, enclosing 43–45 successive activities. Duration of one cycle in thinned stands counted 13 minutes. Whereas, in clear-cut stands, due to longer times of relocation and wood processing, the work cycles were by ca. 15% longer (15 minutes). High accuracy of snapshot observations, presented in Table 4, recorded in longer, 15-minute intervals may be explained by similar sequencing of work.

Assessing the labour consumption of production processes is based on the proportions of operational times (main work time, ancillary work time), as well as complementary and preparatory times (maintenance time, repair time, non-working time) (Jabłoński 2006, Dvořák et al. 2011). In forestry, due to the great variability of work environment, such studies are usually conducted using the photography of work day method, consisting in measuring duration of all activities encountered during a working shift. This method, though it provides an abundant research material, has many drawbacks. Among others, it requires the observers to be perfectly acquainted with the employed logging technology. Moreover, the measurements involved are extremely time-consuming, which makes the entire investigation very costly. These are the factors due to which the databases obtained in the above-mentioned manner are usually unsatisfactory and insufficient to draw any undisputable conclusions in terms of the general variability of the analyzed phenomena. With regard to harvesters, it would be possible to perform the work time analyses based on the data recorded by their computer systems (Dvořák et al. 2011), though a decrease in precision of such obtained data is likely to occur (Purfürst and Erler 2011). The method of snapshot observations is an important research tool for analyzing the duration of work activities, in particular, while determining the proportions of non-operational times (SW – supportive work time, NT – non-working time, NW – non-workplace time) within a working shift. Miyata et al. (1981) noticed that the method in question is much more useful than the photography of work day since it enables to observe a few worksites at the same time, which considerably shortens the time of taking measurements. Moreover, it is less tiresome, thus the obtained results are believed to be less burdened with measurement errors. One of the greatest advantages of snapshot observations is the possibility to reduce the size of databases (Miyata et al. 1981, Szewczyk 2014b).

Table 5 Characteristics of a number of activities constituting a work cycle and total duration of the cycle in the investigated stands Cycle length Cutting method

Number of observations

Cycle length (number of observations) determined in Fourier analysis

Number of cycles, based on which a cycle duration was determined

Cycle duration (median) min

Early thinning

4229

Late thinning

3129

118

Clear-cutting

2703

256

Croat. j. for. eng. 37(2016)2

Analysis of Accuracy of Evaluating the Structure of a Harvester Operator’s Workday ... (251–259)

An essential problem arising upon applying the snapshot observation method in work measurement is choosing an appropriate time interval, in which the observations are recorded. This is extremely significant while investigating a highly changeable work environment, which is usually encountered at operations of timber harvesting and skidding. The measurement frequency determines the size of the research sample directly. Therefore the simplest solution would be to establish the minimum number of observations, and based on this, to assess the constant or variable time interval of the measurement. This approach was adopted by Miyata et al. (1981) in their studies on harvester operation, making the minimum number of snapshot observations dependent on the standard deviation and the assumed confidence level, both calculated for the primarily taken research sample. The above-mentioned researchers assessed the accuracy of their measurements based on the confidence interval of binominal distribution. A serious drawback in the presented approach was the necessity of assuming one feature arbitrarily (e.g. non-working times) as a reference point for the entire variability, determined by the characteristic structure of all productive and non-productive times. The snapshot observations, even those recorded at low frequencies and under conditions of great variability of work, typical of logging operations, allow determining the time structure of a working shift very precisely. The results obtained in the course of the studies presented here, at frequency of measurements up to 10 or 15 minutes, correspond to the data recorded for forest machines published by Miyata et al. (1981). One of the most essential factors that lowers the accuracy of snapshot observations is the conformance of time interval, in which an observation is recorded, and the length of work cycle, in particular, at high recurrence of work operations, which is often encountered, e.g. in industry, on production lines (Wołk 1960, Wołk and Strzelecki 1993, Szewczyk 2014b). The conducted studies proved that the time interval of slightly shorter or longer duration than the one of a work cycle reflects the shares of all activities within a working shift very accurately. Theoretically, one may expect that in such a case sampling would always indicate the very same work activity, making the measurements completely unreliable. However, since work cycles have varied lengths and sequences of various work activity structures overlap one another, the synchronisation between cycle lengths and snapshot observations is practically impossible. Croat. j. for. eng. 37(2016)2

G. Szewczyk et al.

5. Conclusions Accuracy of evaluating the time structure of a work day employing the method of regular snapshot observations was assessed with the use of coefficients of determination and regression, calculated for a linear dependence between the time structures, determined upon application of the photography of work day method and snapshot observations. The research covered testing an accuracy of the data obtained in snapshot observations recorded in 3-, 5-, 10-, 15-, 20- and 25-minute intervals. In early thinned stands, high accuracy (max. 5% margin error) of regular snapshot observations was proved for time intervals up to 10 minutes. Assuming the lower limit for values of coefficients of determination and regression at the level of 0.90 (max. 10% margin error) allows to apply 10- or 15-minute intervals also in late thinned and clear-cut stands. The accuracy of evaluating the proportions of work times within a working shift using the regular snapshot observation method depends, among others, on the ratio of the work cycle length to the time interval between subsequent observations. High accuracy of estimation can be reached providing that the length of work cycle and the duration of time interval are similar, and the measurements are taken at a worksite characterised by great variability of work and overlapping cycles with varied lengths. Among the advantages of snapshot observations, except for the lower labour intensity of the research than the one required by the photography of work day method, a possibility to assess the shares of complementary and preparatory work times should be named. Furthermore, quick sampling of time within a working shift allows to evaluate a few worksites at the same time.

Acknowledgement This work was supported by the Ministry of Science and Higher Education of Poland under Grant No DS ZULiD/3412/2014

6. References Backhaus, G., 1990: Die Allgemeinen Zeiten im forstlichen Arbeitsstudium (Time classification in the work study of forestry). Forsttechnische Informationen 1: 1–5. Björheden, R., 1991: Basic time concepts for international comparison of time study reports. Journal of Forest Engineering 2(2): 33–39. Box, G.E.P., Jenkins, G.M., 1976: Time series analysis: forecasting and control. San Francisco, Holden-Day, 18 p.

257

G. Szewczyk et al.

Analysis of Accuracy of Evaluating the Structure of a Harvester Operator’s Workday ... (251–259)

Brisley, C.L., 2001: Work sampling and group timing technique. In: Zandin, K.B., (ed.) Maynard ś Industrial Engineering Handbook, 5th edition. New York, McGraw-Hill. Dvořák, J., Bystrický, R., Hošková, P., Hrib, M., Jarkovská, M., Kováč, J., Krilek, J., Natov, P., Natovová, L., 2011: The use of harvester technology in production forests. Folia Forestalia Bohemica, Kostelec nad Černými lesy, 156 p. Dvořák, J., Walczyk, J., 2013: Wydajność pozyskania drewna przy pomocy harwesterów i pilarki spalinowej (Productivity of timber harvest with the use of a harvester and a chainsaw). Sylwan 157(3): 171–176. Häberle, S., 1992: IUFRO Symposium ‘Time Study – Measurement and Terminology’. Forst und Holz 47(15): 471. Jabłoński, K., 2006: Optymalizacja sposobów realizacji procesu technologicznego pozyskiwania i zrywki drewna na określonym obszarze (Optimization models for the technological proces of wood harvesting and extraction in a given forest area). Wydawnictwo Akademii Rolniczej im. Augusta Cieszkowskiego w Poznaniu, 126 p. 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. Kärhä, K., Rönkkö, E., Gumse, S., 2004: Productivity and cutting costs of thinning harvesters. International Journal of Forest Engineering 15(2): 43–56. Kot, M.S., Jakubowski, J., Sokołowski, A., 2007: Statystyka. Statistic. Statsoft Polska Sp. z o.o. Warszawa, 520 p. Miyata, E.S., Steinhilb, H.M., Winsauer, S.A., 1981: Using work sampling to analyze logging operations. USDA Forest Service. North Central Forest Experiment Station. Research paper NC–213, 8 p. Monkielewicz, L., Czereyski, K., 1971: Analiza metod ustalania technicznych norm pracy przy pozyskaniu i transporcie drewna (An analysis of methods for establishing technical standards of work at wood extraction and transportation). Prace Instytutu Badawczego Leśnictwa, Warszawa, 77 p. Morrow, R.L., 1957: Motion economy and work measurement. The Ronald Press Company, New York, 265–319. Moskalik, T., 2004: Model maszynowego pozyskiwania drewna w zrównoważonym leśnictwie polskim (Model of fully mechanized timber harvesting in sustainable Polish forestry). Wydawnictwo SGGW, Warszawa, 120 p. Nurminen, T., Korpunen, H., Uusitalo, J., 2006: Time consumption analysis of the mechanized cut-to-length harvesting system. Silva Fennica 40(2): 335–363. Ovaskainen, H., Uusitalo, J., Väätäinen, K., 2004: Characteristics and significance of a harvester operator’s working technique in thinnings. International Journal of Forest Engineering 15(2): 67–77. Picchio, R., Sirna, A., Sperandio, G., Spina, R., Verani, S., 2012: Mechanized harvesting of eucalypt coppice for bio-

258

mass production using high mechanization level. Croatian Journal of Forest Engineering 33(1): 15–24. Purfürst, F.T., Erler, J., 2011: The human influence on productivity in harvester operations. International Journal of Forest Engineering 22(2): 15–22. Samset, I., 1990: Some observations on time and performance studies in forestry. Meddeleser fra Norsk Institut for Skogforskning, 80 p. Sowa, J.M., Leszczyński, K., Szewczyk, G., 2006: Human energy expenditure in late thinning performed in mountain spruce stands. Acta Scientiarium Polonorum. Silvarum Colendarum Lignaria 5(1): 73–80. Sowa, J.M., Szewczyk, G., 2013: Time consumption of skidding in mature stands performed by winches powered by farm tractor. Croatian Journal of Forest Engineering 34(2): 255–265. Sowa, J.M., Kulak, D., Szewczyk, G., 2007: Costs and efficiency of timber harvesting by NIAB 5-15 processor mounted on a farm tractor. Croatian Journal of Forest Engineering 28(2): 177–184. Spinelli, R., Visser, R., 2008: Analyzing and estimating delays in harvester operations. International Journal of Forest Engineering 19(1): 36–41. Stanisz, A., 2007: Przystępny kurs statystyki. Tom 2, Modele liniowe i nieliniowe. A comprehensible course on statistics. Volume 2, Linear and non-linear models. StatSoft Polska Sp. z o.o., Kraków, 868 p. StatSoft Inc., 2009: STATISTICA (data analysis software system), version 9.0, https://www.statsoft.com/. [Cited 10 September 2015]. Szewczyk, G., 2010: Czasochłonność zrywki konnej w drzewostanach trzebieżowych (Time consumption of horse skidding operation). Sylwan 154(1): 52–63. Szewczyk, G., 2011a: Variability of the harvester operation time in thinning and windblow areas. Proceedings of conference Technology and ergonomics in the service of modern forestry. Publishing House of the University of Agriculture in Krakow, 183–196. Szewczyk, G., 2011b: Czasochłonność zrywki drewna wyciągarkami zagregowanymi z pilarkami spalinowymi w drzewostanach trzebieżowych (Time consumption of skidding operations using winches aggregated with chainsaws in thinned stands). Sylwan 155(6): 401–412. Szewczyk, G., 2012: Sekwencyjność operacji zrywki w drzewostanach trzebieżowych i poklęskowych (Sequence of skidding operations in thinnings and snowblow areas). Nauka Przyroda Technologie 6(3): 1–10. Szewczyk, G., 2014a: Model strukturalny dynamiki zmienności pracy na wybranych stanowiskach roboczych w pozyskiwaniu i zrywce drewna (Structural model of work variability dynamics at selected work sites in timber harvesting and skidding). Zeszyty Naukowe Uniwersytetu Rolniczego im. Hugona Kołłątaja w Krakowie, 174 p. Croat. j. for. eng. 37(2016)2

Analysis of Accuracy of Evaluating the Structure of a Harvester Operator’s Workday ... (251–259) Szewczyk, G., 2014b: Obserwacje migawkowe w pomiarach pracy przy pozyskiwaniu i zrywce drewna (Snapshot observations in work measurement during timber harvesting and skidding). Sylwan 158(11): 803–810. Szewczyk, G., Sowa, J.M., Grzebieniowski, W., Kormanek, M., Kulak, D., Stańczykiewicz, A., 2014: Sequencing of harvester work during standard cuttings and in areas with windbreaks. Silva Fennica 48(4): 1–16. doi: 10.14214/sf.1159. Tippett, L.H.C., 1935: A snap-reading method of making time-studies of machines and operatives in factory surveys. The Journal of the Textile Institute Transactions 26(2): 51–70.

G. Szewczyk et al.

Wołk, R., 1960: Techniczne normowanie czasów obróbki. Cz. 1. Podstawowe zasady normowania (Technical standardization of work times at wood processing. Part 1. Basic rules of work time standardization). Państwowe Wydawnictwo Techniczne, Warszawa, 364 p. Wołk, R., Strzelecki, J.S., 1993: Badanie metod i normowanie pracy (Studies on methods and work time standardization). Wydawnictwa Politechniki Warszawskiej, Warszawa. Zečić, Ž., Krpan, A.P.B., Vukušić, S., 2005: Productivity of C Holder 870 F tractor with double drum winch Igland 4002 in thinning beech stands. Croatian Journal of Forest Engineering 26(2): 49–57.

Authors’ address: Assist. prof. Grzegorz Szewczyk, PhD.* e-mail: rlszewcz@cyf-kr.edu.pl Prof. Janusz M. Sowa, PhD. e-mail: rlsowa@cyf-kr.edu.pl Agricultural University of Cracow Faculty of Forestry Al. 29 Listopada 46 31-425 Kraków POLAND Assist. prof. Jiri Dvořák, PhD. e-mail: dvorakj@fld.czu.cz Czech University of Life Sciences Prague Faculty of Forestry and Wood Sciences Kamycka 1176 165 21 Prague 6 – Suchdol CZECH REPUBLIC

Received: September 10, 2015 Accepted: January 11, 2016 Croat. j. for. eng. 37(2016)2

Krzysztof Kamiński, MSc. e-mail: krzysztof.kaminski@radom.lasy.gov.pl Dariusz Kulak, PhD. e-mail: rlkulak@cyf-kr.edu.pl Arkadiusz Stańczykiewicz, PhD. e-mail: rlstancz@cyf-kr.edu.pl Agricultural University of Cracow Faculty of Forestry Al. 29 Listopada 46 31-425 Kraków POLAND * Corresponding author

259

Original scientific paper

Comparison of Cable Skidding Productivity and Cost: Pre-Choking Mainline Versus Tagline Systems Pierre Ackerman, Reino Pulkki, Benedict Odhiambo Abstract This study quantifies the operational efficiency and cost of pre-choking main and tagline systems for tree-length extraction using a cable skidder. The study was done by comparing productivity and costs of the two systems in a semi-mechanised tree-length harvesting operation. Study data was collected using time studies and work sampling for choking and dechoking operations, and GNSS tracking for recording and analysing machine in-field travel time and skidding distance. Operating costs were estimated using South African Harvesting and Transport Costing Model. Average productivity of the tagline system (46 m3 PMH-1) exceeded that of the mainline system (34 m3 PMH-1) by 35%. The extraction cost of the tagline system (US$1.10 m-3) was 26% lower than the cost of using the mainline system (US$1.50 m-3). Keywords: cable skidding, mainline system, tagline system, productivity, cost, extraction, time study

1. Introduction One of the most common methods of primary transportation (extraction) for pine sawlog production in South Africa is ground-based cable skidding (Ackerman et al. 2014). Ground-based primary transport from stump to roadside landing (i.e. extraction) of tree-lengths or tree sections using specialised primary transport equipment, such as an articulated skidder, is impacted by the terrain conditions normally encountered. These include slope, low bearing capacity soils and surface obstacles such as rocks, depressions, stumps and felling debris (Kluenderet al. 1997, FESA 1999). Grapple and cable skidder are the two types of articulated skidders most commonly used. At present cable skidders are more prevalent than grapple skidders in South Africa (Ackerman et al. 2014). They are mostly used in larger timber as their productivity is severely compromised when extracting smaller dimension trees or tree parts (de Wet 2000). As the name suggests, a cable skidder uses a winch to draw the trees to the machine and then skid them to a roadside landing. Globally, cable skidding is the only method currently being used post motor-manual felling since stems are often scattered throughout the comCroat. j. for. eng. 37(2016)2

partment, making this method of extraction necessary for the efficient gathering of logs (Rummer 2002). Further, skidding is also required for extraction in difficult and steep terrain (Bejhou et al. 2008). Cable skidders use either mainline or tagline systems (i.e. the winching method attaches chokers to a mainline or taglines) to gather and draw trees or tree sections to the machine, although tagline systems are not common in South African extraction systems. The choking system, most often used in South ÂAfrica, is the mainline system, utilizing a single set of choker chains or cable chokers. This system is less productive compared to pre-choking, which uses two (or even three) sets of chain or cable chokers, or a tagline choking system (Bromley 1969, APA 1988, De La Borde 1992, MacDonald 1999). The mainline system comprises a wire rope that forms the main line to which the individual tree-lengths are attached by shorter wire ropes or chain chokers (Fig. 1). The mainline wire rope, typically used in South African operations, is a 19 mm diameter IWRC wire rope (depending on treesize) that is 50 m or longer in length and fitted with 4â&#x20AC;&#x201C;6 sliders (which slide on the main-line). Each slider can accommodate a choker attached to a tree-length

261

P. Ackerman et al.

Comparison of Cable Skidding Productivity and Cost: Pre-Choking Mainline Versus ... (261â&#x20AC;&#x201C;268)

Fig. 1 Diagram of mainline rigging system: the main line is a wire rope to which tree-lengths are attached by shorter wire ropes or chain chokers

Fig. 2 Diagram of tagline rigging component: sliders are set up along the tagline similarly to the mainline system, but the end of each tagline is fitted with a hook for easy attachment when winching

or tree section. The number of sliders fitted to the mainline depends on the skidder power, tree size, wire rope diameter and terrain. Choker chains are mainly used in South Africa. They are made of a 1.8 to 2.0 m length of 10 mm or 12Â mm diameter Herc-Alloy chain (depending on the application) with rings or hooks at one end, which are set around either the thin or butt end of a tree-length or tree section. During load hook-up, the winch brake is released and the mainline is pulled from the winch drum to individual or bunched tree-lengths identified for the next extraction cycle. The tree-lengths are then attached to the mainline by slotting the chain ends into the sliders mentioned above. They are then winched to the skidder and skidded to the roadside landing. At the landing, a choker setter releases the load from the choker. The mainline, along with the chokers, are winched back to the skidder to be returned to the field for the next cycle (APA 1988). The use of two sets of choker chains allows prechoking and can significantly increase productivity of operations by reducing the terminal cycle times. Prechoking involves choker setters setting the load using one set of choker chains infield while the skidder extracts the previous load to the roadside landing. The

skidder returns infield with the chokers that have just been off-loaded, and the set already pre-choked is attached to the mainline for the next cycle. The result of this is that the skidder spends less time waiting to pick up the load compared to when a single set of choker chains is used (APA 1988). The tagline choking system, commonly making use of two sets of chain chokers, has been reported to be more productive than the mainline system (Bromley 1969, APA 1988, De La Borde 1992, MacDonald 1999). The tagline system of extraction involves the use of a tagline to assemble tree sections for extraction (Fig. 2). A tagline is approximately 15 to 20 m in length and of the same dimension as the mainline. The setup of sliders is exactly the same as the mainline system outlined above. The end of each tagline is fitted with a hook or a loggerhead grab for easy attachment to the mainline before winching. Three taglines are used in the operation as follows: at any one time, one tagline is infield being pre-choked, the second is travelling loaded with the skidder to the landing, and the third is being dechoked at the landing after which it is returned infield (De La Borde 1992). When the skidder returns infield, the empty tagline is off-loaded and the mainline pulled from the winch

262

Croat. j. for. eng. 37(2016)2

Comparison of Cable Skidding Productivity and Cost: Pre-Choking Mainline Versus ... (261–268)

drum to the loaded tagline. The loaded tagline is attached to the winch mainline and winched back to the skidder for extraction to the roadside landing. At the roadside, a choker setter unhooks the mainline from the load once it has dropped to the ground and attaches the empty tagline (from the previous load) to return infield (Bromley 1969, De La Borde 1992). This tagline system significantly minimises the waiting times both at the roadside landing and infield (de Wet 2000, Lusso 2003). The objective of this study was to compare the efficiency of a mainline system with two sets of choker chains to a tagline choking system with three sets of choker chains in terms of productivity (m3 PMH-1) and cost (US$ m-3). Productivity in terms of Productive Machine Hours (m3 PMH-1) and cost (US$ m-3) of prechoking in a mainline system with two sets of choker chains was compared to a tagline system with three sets of choker chains.

2. Materials and Methods The study was conducted on three Pinus radiata compartments harvested by Cape Pine Investment Holdings Ltd. located near the town of Grabouw in the Western Cape Province of South Africa. The stand and site conditions for each compartment are shown in Table 1. The terrain conditions provided in Table 1 are based on the classification system by Erasmus (1994). Compartments M7a and M7b were adjacent to each other but separated by a stream. Their terrain conditions were similar, as well as stand conditions, having been established at the same time and subjected to the same silvicultural treatments as shown in the compartment records.

P. Ackerman et al.

Time study, work sampling and Global Navigation Satellite System (GNSS) tracking were used to obtain information about each system. Time study was done using stop watches to record time consumption for choker-setting and dechoking. Work sampling was used to systematically and critically examine the methods applied in executing the various tasks, thus providing detailed time-based information on each work element. GNSS tracking, due to its ability to perform autonomous time studies, monitor and track mobile machines (Spruce et al. 1993, McDonald 1999, Reutebuch et al. 1999, Veal et al. 2000, Veal et al. 2001, Robert 2002, Ronald et al. 2006), was used to gather complimentary data to the time study and work sampling. Travel times and speeds were also extracted from the GNSS data. The GNSS system comprised of a GPS device (FM LOC GPS) was installed on the skidder to record operational data, which was then analysed using FDO Fleet Manager Professional Version 8.3 software. Through GNSS tracking, detailed summaries of machine system performance over long periods of time alongside spatial detail of machine travel including distances, speeds and travel times could be recorded and matched with time study data. Choking time, dechoking time, travel loaded and travel empty data were combined into a work cycle and used in calculating productivity per productive machine hour (PMH). The three compartments were each divided into three strips. Each of the strips had two predetermined designated skid trails located parallel to each other 30 m apart (Fig. 3). The position of each skid trail was marked and cleared of trees and other vegetation to create a uniform running surface, free of obstacles.

Table 1 Summary of stand and site conditions in the compartments Stand parameters

Compartment M6

Compartment M7a

Compartment M7b

Area, ha

7.5

10.7

9.1

Age, years

-1

425

400

Average tree volume, m

0.87

0.99

Volume/ha

370 m3 ha-1

Stand density, stem ha

396 m3 ha-1

Good in dry state

Moderate in moist state

Poor in wet state

Ground roughness

Slightly uneven

Slope condition

Gentle slope –10%

Gentle slope +10%

Ground condition

Croat. j. for. eng. 37(2016)2

263

P. Ackerman et al.

Comparison of Cable Skidding Productivity and Cost: Pre-Choking Mainline Versus ... (261â&#x20AC;&#x201C;268)

Fig. 3 Graphical representation of the harvesting plan within one compartment Slash material was left on the surface of skid trails to improve ground stability, ensure suitable traction and limit potential damage to the soil. Roadside landings were located on the lower side of the three compartments to utilise gravity assisted skidding. The operations commenced sequentially from strip 1 to 2 and finally to strip 3 in each of the three compartments. For each strip, extraction to the landing began only after the entire strip had been felled. Directional felling facilitated the tree-length extraction phase (Bromley 1969, Conway 1979, Spiers 1986, Andersson and Young 1998, MacDonald 1999). The skidder was confined to the pre-marked skid trails by winching the tree-lengths from both sides of the skid trail to the skid trail before returning to the roadside landing. Each of the two predetermined skid trails in each strip was randomly allocated either a mainline or tagline winching system: i.e., each system being applied in one of the skid trails in each strip. The mainline system was comprised of two sets of six choker chains (total 12). The two sets of choker chains facilitated pre-choking of tree-lengths infield, while the skidder was hauling the previous load to the landing. The tagline system was comprised of three taglines each having four choker chains (a total of 12 choker chains). Man-power requirements remained the same for both the mainline and tagline systems in the study and consist of two chocker-setters and one dechoker person at road-

264

side landing. Generally, this is the arrangement of manpower in cable skidder operations country wide and the number of chocker setters will only increase if slope increases above 25 to 30% when up-hill pull by the crew of the wire rope is required (Ackerman et al. 2014). In this case, the number of choker-setters will increase from two to three. In each skid trail, 20 work cycles were studied. This was determined from a pilot study on the skidder daily average work routine per shift using Eq. 1, George (1992). PQ sp = (1)

Where: ÎŁp 5% standard error of proportion (the confidence level is 95%) P 14.5% non-work time Q 85.3% work time N number of cycles per skid trail Individual work elements, comprising a work cycle, were identified in Table 2. The time consumption of each element was recorded. Load size (m3) was derived from pre-determined tree-length volumes multiplied by the number of tree-lengths extracted per load. The GNSS tracking data was extracted from the GPS device at the end of the work shift and each cycle Croat. j. for. eng. 37(2016)2

Comparison of Cable Skidding Productivity and Cost: Pre-Choking Mainline Versus ... (261–268)

P. Ackerman et al.

Table 2 Elements comprising a work cycle and their break points Work elements

Work element defined

Travel unloaded

From when the skidder starts to travel back infield at the landing to when the skidder operator releases the winch break to drop the chokers to the ground at the stump site

Choking

From when the skidder operator releases the winch break to drop the chain chokers to the ground to when it starts to move after its complete load has been winched

Travel loaded

From when the loaded skidder starts to move towards the landing to when it drops the load at the landing surface (once the winch has been released)

De-choking

From when the load makes contact with the landing surface to when the skidder starts to travel back infield

data matched to its respective time study data using real time recordings. The costs of the skidder and material (mainlines and chain chokers) were calculated using the South African Harvesting and Transport Systems and Costing Model (Hogg et al. 2009). Costs were then converted from ZAR to USD using an exchange rate of 0.073. The data was analysed using Statistica Version 8. The Shapiro-Wilk test was used to test the normality distribution of residuals at 95% level of significance. The residuals were not normally distributed, and were then subjected to log, square root and exponential transformation in an attempt to normalise them. These transformation attempts were unsuccessful and the original data was analysed by non-parametric techniques: i.e., Kruskal-Wallis test and Bootstrapping. The

Kruskal-Wallis test is used to compare three or more samples and is applicable in situations where the assumptions of ANOVA are violated (Siegel and Castellan 1988). When data are not normally distributed and transformation of the data is unsuccessful, non-parametric bootstrap multiple comparison tests are often used for statistical inference; H0: P(X<Y) = P(X>Y) against H0 P(X<Y) ≠ P(X>Y) at a = 0.05 (Reiczigelet al. 2005). The bootstrap methods replace inaccurate approximations to biases, variances and other measures of uncertainty and have proved to work better than traditional methods in solving non-parametric problems (Davison and Hinkley 1997). Kruskal-Wallis test was used to test the differences between groups, specifically the differences between the compartments

Table 3 Comparison of cycle elements between tagline and mainline systems Statistical comparison

Mainline 6 choker chains

Tagline 4 choker chains

Sig

Choking time, min

3.47

2.13

118.445

0.0001***

Dechoking time, min

2.05

1.24

94.860

0.0001***

Travel empty time, min

0.95

0.78

0.959

0.328 ns

Travel loaded time, min

1.50

0.78

21.727

0.0001***

Travel empty distance, m

77.23

70.26

1.166

0.281 ns

Travel loaded distance, m

66.05

61.76

2.321

0.128 ns

Travel empty speed, ms-1

1.35

1.50

3.614

0.060ns

Travel loaded speed, m ms-1

0.73

1.32

15.714

0.0001***

Load per cycle, m

4.51

3.46

86.791

0.0001***

Cycle time, min

7.97

4.93

105.485

0.0001***

*** – very highly significant ns – not significant)

Croat. j. for. eng. 37(2016)2

265

P. Ackerman et al.

Comparison of Cable Skidding Productivity and Cost: Pre-Choking Mainline Versus ... (261â&#x20AC;&#x201C;268)

(M6, M7a and M7b) in terms of tree sizes and stocking (m3 ha-1). Bootstrap test was used to determine significant differences in cycle elements between the two systems.

3. Results of the study The three compartments were adjacent and their stand and site conditions were homogeneous. There was no significant difference in tree size (P=0.89) or stand density (P=0.99) in the three compartments. The compartments did not, therefore, differ in terms of stems per ha or wood volume per ha. Similarities in stand and site conditions in all three compartments (Table 1) permit the pooling of data from the three compartments to analyse the differences between tagline and mainline systems against the independent variables. The statistical comparisons of productive work cycle elements for the mainline and tagline systems are presented in Table 3. The costs of using the mainline and tagline systems are presented in Table 4. To account for the difference in the number of choker chains between the main and tagline, the cycle element times for travel empty and loaded, and load size of the mainline system were used to recalculate the productivity and cost for a tagline system using three sets of six chokers (Table 5). The cost per PMH of the skidder equipped with the tagline system increased slightly due to the additional six chokers in the system. To account for the larger load size with six tagline chokers, the loaded travel speed of the mainline system with six chokers was used. Table 4 Machine productivity and costs when using mainline (2 sets of 6 choker chains) and tagline systems (3 sets of 4 choker chains) System

Productivity m3 PMH-1

Cost US$ m-3

Cost US$ PMH-1

Mainline system

34.0

1.50

50.77

Tagline system

42.1

1.21

51.08

Table 5 Machine productivity and costs when using mainline and tagline systems with 6 choker chains per load Productivity m3 PMH-1

Cost US$ m-3

Cost US$ PMH-1

Mainline system

34.0

1.50

50.77

Tagline system

46.5

1.10

51.12

System

266

Fig. 4 Mainline and tagline cable skidding productivities modelled over 50 m, 150 m and 250 m average extraction distances Based on the mean travel speeds, choking times, dechoking times and load sizes (Table 3), productivity of each system was modelled (Fig. 4) over three average extraction distances (50 m, 150 m and 250 m).

4. Discussion The objective of the study was to compare productivity (m3 PMH-1) and cost (US$ m-3) of mainline and tagline systems used in softwood sawlog tree-length extraction operations. In this study, the tagline system was introduced on a trial basis. The mainline system, contrarily, was already in use. Results show that productivity of the tagline system using four choker chains per load (42 m3 PMH-1) exceeds that of the mainline system (34 m3 PMH-1), even though there was a significant difference in load size extracted per cycle in the two systems: i.e., six chokers as opposed to four chokers for the main and tagline systems, respectively. However, taglines are not commonly used in South Africa. One reason may be that practitioners are not aware of the potential benefits of increased productivity and reduced costs associated with the correct use of tagline systems (MacDonald 1999), particularly in smaller dimension timber, where it becomes difficult to attain optimal load sizes with more traditional choking systems (De La Borde 1992). Other reasons may be the increased degree of complexity and supervision required. However, the applicability of taglines goes beyond that of only smaller dimension timber, as demonstrated in this study. And although the study Croat. j. for. eng. 37(2016)2

Comparison of Cable Skidding Productivity and Cost: Pre-Choking Mainline Versus ... (261–268)

was conducted with unequal numbers of chokers between the mainline and tagline systems, there was a significant improvement with the use of the tagline system. However, with the simulation of equal numbers of chokers, a further improvement of productivity of 4.4 m3 PMH-1 was achieved (Table 5). The main differences between mainline and tagline systems occur during the terminal (choking and dechoking) phases of timber extraction (de Wet 2000, Lusso 2003), as demonstrated in Table 5. Taglines allow choking to take place prior to the return of the skidder to the compartment, while at the landing, where the entire tagline is removed for the dechoking process, the skidder can return to the stump site without delays. Pre-choking as well as quick attachment and release of taglines results in shorter cycle times compared to the mainline system. Choking and dechoking operations are prolonged using the mainline system. Shorter cycle times in taglines increases machine utilization, resulting in more cycles per PMH compared to the mainline system (Table 5) as Bromley (1969), APA (1988), De La Borde (1992) and MacDonald (1999) have shown. The basic objective in industrial forest harvesting operations is to maximize productivity while minimizing costs (FAO 1998). Tagline system hardware is more expensive than mainline systems. The difference in costs between the two systems is related to the extra costs incurred in acquiring the taglines (a set of three taglines). In this study, the difference was US$ 2274.32 as opposed to US$ 1503.87 for a simulated mainline system comprising the winch line cable and two sets of six choker chains. The high costs of implementing the tagline systems are however offset by improved productivity. The unit production cost of operating the skidder using the tagline system (US$ 1.10 m-3) was 26% less than the cost of operating the skidder using the mainline system (US$ 1.50 m-3), making tagline systems more cost efficient. It is interesting to note that the tagline productivity remains greater than that of mainline even over extraction distances of up to 250 m, but the difference between the two curves, however, diminishes with increasing distance (Fig. 4). This is due to the longer extraction distances off-setting terminal times. It is unfeasible to see where they equal each other as skidders operate best in the range of about 100 to 150 meters maximum extraction distance (MacDonald 1999).

5. Conclusions Tagline system is more productive than the mainline system due to its shorter choking and dechoking times. Shorter terminal times result in shorter cycle Croat. j. for. eng. 37(2016)2

P. Ackerman et al.

times, which directly result in higher productivity of the tagline system. The higher cost of using the tagline system is offset by the high productivity resulting from the system. The tagline system is, therefore, more productive (46 m3 PMH-1) and cost efficient (US$ 1.10 m-3) compared to the mainline system (34 m3 PMH-1 and (US$ 1.50 m-3) when using a cable skidder in semi-mechanised tree-length harvesting operations. Tagline systems are, however, more complex in use and require more operational awareness to maintain improved efficiencies. The results of this study will hopefully encourage the use of the more efficient choking systems within pine sawtimber treelength extraction operations in South Africa.

6. References Ackerman, P., Pulkki, R., Gleasure, E., 2014: Modelling of wander ratios, travel speeds and productivity of cable and grapple skidders in softwood sawtimber operations in South Africa. Southern Forests 76(2): 101–110. Andersson, B., Young, G., 1998: Harvesting coastal secondgrowth forests: summary of harvesting systems performance. FERIC Technical report 120 p. APA, 1988: Timber harvesting, Fourth edition, American Pulp Wood Association. The Interstate Printers & Publishers, Illinois. Behjou, F.K., Majnounjan, B., Namjranian, M., Dvořák, J., 2008: Time study and skidding capacity of the wheeled skidder Timberjack 450C in Caspian forests. Journal of Forest Science 54(4): 183–188. Bromley, W.S., 1969: Pulpwood production, 2nd edition, Interstate Printers & Publishers, Illinois. Conway, S., 1979: Timber Cutting practices, Miller Freeman Inc., San Francisco. Davison, A.C., Hinkley, D.V., 1997: Bootstrap methods and their application. Cambridge University Press, New York. De La Borde, R.M., 1992: Timber harvesting manual, Institute for Commercial Forestry Research, South Africa. De Wet, P., 2000: Ground based extraction. In: South African Forestry Handbook (Owen, D.L., ed.), South African Institute of Forestry, South Africa, 316–322. Erasmus, D., 1994: National terrain classification systems for forestry. Institute for Commercial Forestry Research, Bulletin Series 684/3/3, South Africa. FAO, 1998: Conventional versus environmentally sound harvesting: impacts on non-coniferous tropical veneer log and saw log supplies. Unasylva 49(193): 23–30. FESA, 1999: Guidelines for forest engineering practices in South Africa,Forest Engineering South Africa, South Africa. George, T. 1992: Harvesting forest products, Stobart Davies ltd, Greece.

267

P. Ackerman et al.

Comparison of Cable Skidding Productivity and Cost: Pre-Choking Mainline Versus ... (261–268)

Hogg, G.A., Krieg, B., Ackerman, P.A., Längin, D.W., 2010: Harvesting system and equipment costing. In: Ground based harvesting handbook (Längin, D.W. & Ackerman P.A., ed.), FESA Publication, South Africa, 88 p. Reiczigel, J., Zakariás, I., Rózsa, L., 2005: A bootstrap test of stochastic equality of two populations.The American Statistician 59(2): 156–161. Kluender, R., Lortz, D., McCoy, W., Stokes, B., Klepac, J., 1997: Productivity of rubber – tired skidders in southern pine forests. USDA Forest Products Journal 47(11–12): 53–58. Lusso, C., 2003: Ground based extraction within forest harvesting systems. MSc thesis, University of Natal, Pietermaritzburg, South Africa. MacDonald, A.J., 1999: Harvesting systems and equipment in British Columbia. FERIC Hand book No. HB – 12, British Columbia. Reutebuch, S.E., Fridley, J.L., Robinson, L.R., 1999: Integrating real-time forestry machine activity with GPS positional data. ASAE Technical Paper No. 99-5037. Robert, H.D., 2002: The feasibility of using GPS technology for continuous time studies of rubber-tired grapple skidders. MSc Thesis, Louisiana State University, USA. Ronald, C., Mauricio, M., Oscar, M., 2006: Evaluation of Forestry Machinery Performance in Harvesting Operations using GPS Technology. In: Proceedings of the International

Precision Forestry Symposium. Stellenbosch University, South Africa, 5–10 March. Rummer, B., 2002: Forest operations technology. In: Southern Forest Resource Assessment General Technical Report SRS53. USDA-Forest Service Southern Research Station, Asheville, USA, 341–353. Siegel, S., Castellan, N.J., 1988: Nonparametric statistics for the behavioral sciences, McGraw-Hill International, Singapore. Spiers, J.J.K., 1986: Logging operations guidelines, LIRO, Wellington. Spruce, M.D., Taylor, S.E., Wilhoit, J.H., Stokes, B.J., 1993: Using GPS to track forest machines. In: International Winter Meeting sponsored by the American Society of Agricultural Engineers. ASAE, Chicago, Illinois, December 14–17. Stenzel, G., Kenneth, J.P., Thomas, A.W., 1985: Logging and pulpwood production, John Wiley & Sons, New York. Veal, M.W., Taylor, S.E., McDonald, T.P., McLemore, D.K., Dunn, M.R., 2000: Accuracy of tracking forest machines with GPS. In: ASAE Annual International Meeting. American Society of Agricultural Engineers, Wisconsin, USA, 9–12 July. Veal, M.W., Taylor, S.E., McDonald, T.P., Tackett, D.K., Dunn, M.R., 2001: Accuracy of tracking forest machines with GPS. Transaction of ASAE: Chicago, Illinois.

Authors’ address: Pierre Ackerman, PhD.* e-mail: packer@sun.ac.za Benedict Odhiambo, PhD. e-mail: beneodhis@yahoo.co.uk Department of Forest and Wood Science Stellenbosch University Private Bag X1 7602 Matieland, Stellenbosch SOUTH AFRICA

Received: June 10, 2015 Accepted: October 27, 2015

268

Reino Pulkki, PhD. e-mail: rpulkki@lakeheadu.ca Faculty of Natural Resources Management Lakehead University 955 Oliver Road, Thunder Bay, ON, P7B 5E1 CANADA * Corresponding author Croat. j. for. eng. 37(2016)2

Original scientific paper

Designing Mobile Anchors to Yield: A Tension Relief System for Tail Anchoring Ben Leshchinsky, John Sessions, Jeffrey Wimer, Milo Clauson Abstract Cable yarding systems are a common method for transporting materials in mountainous terrain where environmental, safety or economic considerations do not permit ground-based methods such as skidding, forwarding or shovel-logging. Although effective in steep terrain, cable yarding requires anchored support for skyline and operating line operation. Furthermore, mechanized steep slope harvesting, which often uses cable assistance, also necessitates sufficient anchoring for large cable loads. These systems rely on a fixed anchor, or tailhold, to provide adequate line restraint for operation. Anchored guylines and skylines usually depend on available stumps or trees (Peters and Biller 1985). Where these are not available, mobile anchors such as bulldozers or excavators can be used. Mobile equipment anchors have an advantage of predictable resisting capacity, as opposed to trees and stumps, and may present a means of relieving excessive cable tensions in skyline systems without a catastrophic failure. We present a design approach for skyline tension relief with a comparison to actual field data. The analysis demonstrates that exceedance of a mobile anchor sliding resistance results in movement, limiting maximum skyline tensions and subsequently reducing them after the anchor shifts forward. Other anchor types, like stumps, anchored deadmen and engineered anchors, do not have the tendency for gradual movement to relieve skyline tensions as failure is often catastrophic, resulting in higher and potentially dangerous cable tensions, and complete loss of cable tension at failure. For mobile anchors, a relationship between cable tension and length presents an efficient means of predicting the anchor movement to facilitate design of appropriate equipment setback. A comparison of an analytical approach based on cable tensions, cable lengths, and anchor capacity with instrumented field tests demonstrates that the use of equipment, as a primary or auxiliary anchoring system, can be effective and potentially safer when adhering to design constraints based on equipment, equipment placement and in-situ soil properties. Keywords: cable logging, artificial anchors, failure capacity, movement, design, skylines, mobile anchors, dynamic loading

1. Introduction Cable logging systems require anchored support(s) to maintain safe and effective yarding of materials in steep terrain. This support can come from a variety of techniques including buried deadman anchors, stumps or trees, rock anchors or heavy equipment (also known as a mobile anchor, machine anchor or equipment anchor). These anchoring types are often used for guylines to support yarding towers, end support (tailhold) for skylines under loading and high tensions or cableassisted mechanized harvesting equipment. The most Croat. j. for. eng. 37(2016)2

common method for tail support is attachment to nearby, adequately sized trees or stumps (Pyles et al. 1991, Smith 1995) â&#x20AC;&#x201C; a resource that can be a challenge with increasingly shorter stand rotations and the smaller available trees, younger or smaller adjacent stands, and property boundaries. Designing with stump anchors typically involves consideration of tree diameter and species (Pyles et al. 1991, Smith 1995, Peltola et al. 2000), but tends to have a highly variable load capacity that is difficult to define, and their application is often dependent on the subjective judgment of workers installing a logging system. Alternatively, engineered anchors,

269

B. Leshchinsky et al.

Designing Mobile Anchors to Yield: A Tension Relief System for Tail Anchoring (269â&#x20AC;&#x201C;278)

such as buried deadmen, augered anchors, plate anchors and mobile anchors may present less variability in capacity and can account for specific soil conditions (Copstead and Studier 1990, Hartsough et al. 1997). However, fixed anchors like buried deadmen, augered anchors, or plate anchors are subject to catastrophic failure (pullout, rupture, breakage) when their capacity is exceeded, and can result in significant, potentially unsafe cable tensions if of sufficient capacity. When capacity is exceeded, anchor failures can have catastrophic results. Oregon has had at least four fatalities in the past decade from insufficient anchoring, including one fatality from a failed mobile anchor in 2006 (OR-FACE 2016). One method of preventing overloaded skylines is to have tension limiting slipping brakes on the skyline, but this alone will not protect against skyline anchors of unknown capacity. In this study, the focus is on mobile anchors, i.e. equipment anchors, and their capability to avoid catastrophic failure and excessive cable tensions under optimal design conditions. Mobile anchors often use heavy construction or logging equipment as a dead weight that can resist cable tensions during yarding. Typically, this equipment consists of excavators, bulldozers, skidders, and other heavy machinery (see Fig. 1). Soil berms, hillslopes, or embedded equipment shovels or blades can be used for extra resistance when necessary. The soil in front of the equipment can provide added anchor capacity by means of soil self-weight and passive shear resistance based on soil cohesion and angle of internal friction, in turn providing more resistance (Oregon OSHA 2008, Leshchinsky et al. 2015). Embedding the mobile anchor, although increasing the potential for equipment overturning, does allow added passive soil resistance. Mobile anchors present advantages in a variety of scenarios where cable resistance is required. Where

Fig. 1 Bulldozer using a mobile anchor

270

alternative anchoring methods are inadequate, mobile anchors present a means of providing resistance to cable loads, especially for skylines or cable-assisted harvesting equipment, which incur high tensions (Visser and Stampfer 2015, Visser and Berkett 2015, Leshchinsky et al. 2015, Olund 2001). For example, when stump anchors are used as tail support, anchor failure will often involve complete pullout of the stump, and potentially, a rapid subsequent succession of failures of other stump anchors in the system. This can result in a swift loss of tension in the skyline, potentially endangering workers near the skyline or carriage. Furthermore, anchoring plays a critical role in mechanized harvest on steep slopes, particularly in Europe and New Zealand, where cable-assisted fellerbunchers or harvesters rely on mobile anchoring to ensure that cable assistance is stable and sufficient (Visser and Stampfer 2015, Visser and Berkett 2015, Stampfer 1999). If anchoring is sufficiently strong, excessive cable loads can develop during yarding, resulting in cable tensions that may seriously overtension the skyline or guylines that could destabilize the yarding tower or bring a cable to rupture. Cable tension is a function of cable geometry and the applied forces to the cable system, including selfweight of the cable. For a given external loading and horizontal distance between supports, the greater the sag in the skyline between supports, the more efficient the skyline is at providing vertical carrying capacity per unit of horizontal force. If a mobile anchor moves forward, the sag in skyline increases the vertical/horizontal force efficiency reducing the skyline tension. Mathematical models for skyline yarding have been embedded in a number of software including the publically available Skyline XL software (USFS 2015). The reduced tensions resulting from increasing the line length between supports coupled with a mobile anchor tendency to move forward under increasing tension presents a unique opportunity to prevent catastrophic anchor or skyline failures, as well as tipping of yarding towers. For a given payload, cable sizing and length, it is well known that an increase in line length between fixed ends results in significantly reduced cable tensions (Kendrick and Sessions 1991, Brown and Sessions 1996). Concurrently, an increase in cable tensions results in increased mobilization of mobile anchor resistance, ultimately resulting in yielding soil located at both the equipment suspension (treads, tires, grousers, etc.) and embedment zone (berms, embedded blades, etc.). When designed appropriately, these regions of frictional resistance yield, enabling sliding of the vehicle and a rapid decrease in skyline tensions while maintaining cable suspension and yarder stability. After soil yield and movement, Croat. j. for. eng. 37(2016)2

Designing Mobile Anchors to Yield: A Tension Relief System for Tail Anchoring (269–278)

B. Leshchinsky et al.

the equipment can simply be moved back into place under its own operating power. Thus, such a system may present a safety relief system to prevent catastrophic failures of anchors, cable rupture and yarder tip-overs in an economically feasible way. For safety, the design failure mode must be sliding rather than equipment rollover. That is, the skyline must be attached to a point on the equipment that is relatively low to the ground (frequently to the drawbar or a winch), the equipment center of gravity must be set back a sufficient distance from point of rotation, and/or the passive resistance from embedment must be sufficiently small. Such a requirement can be challenging for bulldozers, which have a center of gravity that is not located far from anchor attachment point. A low anchor attachment point, which is often a drawbar or an axle (Oregon OSHA 2008) may counteract large overturning moments, enabling equipment sliding to the governing mode of yielding. The center of gravity is a factor specific to the equipment chosen to serve as a mobile anchor. Furthermore, if sliding is to govern, it is critical that a sufficient distance is chosen between the equipment location and any potential precipices, like steep downslopes or cliffs. In this study, a formulation is presented coupling skyline tension behavior with yield of mobile anchors based on an analytical solution. The theorized relationship between equipment movement from soil yield and reduced skyline tensions are validated with data from an instrumented field test involving an equipment anchor serving as a tailhold.

2. Analytical design The design involves a calculation of mobile anchor capacity based on prior research (Leshchinsky et al. 2015), specifically for sliding. A comparison of anchor capacity and skyline tensions from a field test are compared. A curve of cable tensions was generated from the publically available payload analysis program, Skyline XL (USFS 2015) for a variety of yarding distances to represent the reduction of cable tension with equipment movement from sliding. Based on the coupled relationship of the loading curve and anchor sliding capacity, a design approach is presented to reduce risk of catastrophic yarder, skyline, or anchoring failure by means of mobile anchor yield.

2.1 Anchor capacity A static force equilibrium analysis is used to determine anchor capacity for an array of scenarios described by slope gradient, cable angle, equipment weight, soil strength, equipment embedment depth, blade width and track interaction parameters (Leshchinsky et al. 2015).   1 cos b − tan d t sin b  cos b − tan d t sin b  + cAt  sin b + cos b + Wt ( g Db2 wb Kp + 2cwb Kp )  sin b + cos b    2 sin b + tan d t cos b  sin b + tan d t cos b  F=  cos b − tan d t sin b   sin q + cos q sin b + tan d cos b  t Where:

(1)

F ultimate anchor capacity at failure Kp rankine passive earth pressure coefficient = tan (45˚+ f'/2) f soil internal angle of friction c soil cohesion dt interaction between soil and vehicle support (tracks, tires, etc.) β angle of hillslope supporting equipment q angle of skyline pull Db depth of blade below ground surface wb width of blade At tracked footprint g unit weight of soil Wt weight of equipment With wheeled equipment, the track area can be defined as zero. The weight of the equipment, Wt, is assumed to maintain full interaction (f=d) with the ground surface due to the aggressive treads (grousers) common to tracked equipment. The soil within and beneath the treads is assumed to shear together. Uneven loading and contact of Croat. j. for. eng. 37(2016)2

271

B. Leshchinsky et al.

Designing Mobile Anchors to Yield: A Tension Relief System for Tail Anchoring (269–278)

yarding lengths, including those if a vehicle were to slide forward. These values could be generated using static equilibrium equations or publicly available software (e.g. Skyline XL, USFS 2015) and should be done for any specific situation (e.g. multi-span cable lines). The curve representing the change in skyline tension follows a general exponential decay function of the form: Y = AXB

Fig. 2 Free-body diagram of mobile anchor system the tracks with ground, called eccentricity, occurs due to large moments that can lift part of the equipment off of the ground. Loss of full ground contact can occur when the moment placed on the equipment due to the load becomes large. To mobilize any resistance due to equipment self-weight or traction, the brakes must be engaged.

2.2 Movement and clearance distance design The presented anchor capacity equation presents a means of establishing a cable resistance that represents the point at which the soil supporting anchoring equipment fails in shear, allowing sliding and movement in the direction of pull and a relief in skyline tension. This point of yield and associated ability to slide enables cable tensions to subsequently drop after a peak loading exceeds the soil resisting capacity, a phenomenon that occurs due to the direct, exponential relationship between skyline tension and cable yarding distance. That is, the decrease of the distance between the two points where a cable is fixed (e.g. a yarding tower and a tailhold) increases the cable sag, consequently reducing cable skyline tension with movement of the mobile anchor in the direction of pull. For a given configuration or corridor for yarding, a relationship cable tension and stretched length (anchor movement) can be calculated with associated cable tensions under loaded conditions (Kendrick and Sessions 1991). The maximum tension for any configuration can be found and adjusted to accommodate incremental movement of a tailhold towards the direction of pull, creating a load-cable characteristic curve (Fig. 3). The load-cable characteristic curve (LCCC) represents the reduction in maximum cable tension for a given yarding configuration for a range of potential

272

(2)

This exponential function can be represented by a given load factor (A), representative of a load that is diminished with movement, multiplied by a site constant (S, dependent on site geometry, like angle at which a tailhold moves if shifted towards the direction of pull), to an exponent that is the horizontal movement of the tail point towards the direction of pull (DL). The simple exponential function representative of the LCCC is defined in equation (3) as:

F = Load factor × Site constantMovement = ASDL

(3)

Which can be rearranged as:

F = S DL A

(4)

The exponents can be simplified and separated to establish the relationship of movement with tension: and,

 F log   = log( S DL )  A

(5)

 F log   = DL × log( S )  A

(6)

Finally resulting in the relationship for movement with cable loading, defined as:

 F log    A DL = log( S )

(7)

The relationship presented in equation (7) defines the relationship presented between the reduction in yarding distance and the tension in a skyline, with factors L and S available given a yarding profile and a cable tension analysis. Knowing the anchor capacity against sliding, presented in equation (1), and the relationship between cable tension and movement, these relationships can be related to predict when a mobile anchor will slide forward and relieve cable tensions. Knowing a given anchor capacity for a yarding configuration, it is possible to predict movement based on the LCCC when anchor capacity is exceeded. That is, when the cable load is less than that of the preCroat. j. for. eng. 37(2016)2

Designing Mobile Anchors to Yield: A Tension Relief System for Tail Anchoring (269–278)

B. Leshchinsky et al.

Fig. 3 (a) Left, Schematic representation of load-cable characteristic curve (LCCC) and (b) Right, the associated anchor response curve (ARC) dicted anchor capacity, mobile anchor movement is predicted to be negligible. However, when cable tensions exceed the anchor capacity, the soil supporting the equipment will yield, enabling movement of the vehicle until it once again reaches equilibrium (Fig. 3a), represented by an Anchor Response Curve (ARC, Fig. 3b). Equilibrium occurs due to reduced cable tensions from increased cable sag, as well as the anchor capacity of the vehicle from soil shear. Often, this peak loading occurs due to dynamic loading during yarding, and after mobile anchor movement and cable tension relief, the system returns to equilibrium, and the anchor becomes immobilized once more (Fig. 3b). Upon repeated cycles of yarding payloads, yielding of soil beneath the mobile anchor tailhold, and subsequent movements, the vehicle moves forward notably (Fig. 3b) and may need to be returned to its initial position. After each movement, assuming a similar gradient for the mobile anchor, the LCCC is shifted accordingly. When the anchor capacity is exceeded again, then the same process occurs, predicting the relationship between movement, and anchor capacity based on the LCCC. When the equipment has shifted forward beyond the point of safety or function, it can be moved back to its initial position since the anchor is motorized and mobile. From an analytical perspective, the ARC is determined based on calculated anchor capacity and a given LCCC, which present a piecewise function for mobile anchor movement when combined. The ARC Croat. j. for. eng. 37(2016)2

follows a typical LCCC until the anchor capacity is exceeded, after which, it will move (DL) according to the amount of cable tension load that has exceeded the anchor capacity (DFcrit), as shown in Fig. 3a. Based on the LCCC and anchor capacity, the following relationship is demonstrated for movement:

 F − Fcrit  log  max   A DL = ( Fmax > Fcrit ) log( S )

(8)

The gray, shaded portion in Fig. 3a represents when anchor capacity is exceeded and movement occurs, while the unshaded portion underneath the curve represents no movement. When the cable tension does not exceed the anchor capacity, no movement occurs. This is represented by:

DL = 0( Fmax < Fcrit )

(9)

Where: Fcrit anchor capacity Thus, a piecewise function is presented to establish the ARC for a given LCCC and anchor capacity (Fcrit). The establishment of the anchor capacity to sliding and the load-cable characteristic curve enable a means of predicting the movement response, known as the anchor response curve (ARC). Knowledge of an ARC provides a means to design mobile anchor systems used for skylines to yield and move providing that

273

B. Leshchinsky et al.

Designing Mobile Anchors to Yield: A Tension Relief System for Tail Anchoring (269â&#x20AC;&#x201C;278)

3. Field tests

Fig. 4 Cable yarding system profile anchor movement will have a safe run out distance. That is, with a known yarding profile, a known payload, and a known anchor capacity, sufficient equipment setback can be employed to prevent a loss of a mobile anchor. Such a design could ensure that cable tensions do not exceed the skyline design capacity and the yarding tower is not toppled due to guyline anchor breakage. This analytical method is verified with field testing, presented in this study.

Fig. 5 Load-cable characteristic curve (LCCC) for given cable logging configuration

274

A series of field tests were performed in the McDonald-Dunn forest in Oregon State University, intended to both verify prior analytical solutions for mobile anchor capacity and serve as the yielding point for anchor movement and skyline tension relief. The field tests were performed by instrumenting a swaged skyline (diameter = 1.9 cm, rupture capacity of 375 kN) anchored by a John Deere skidder, weighing 137.9 kN and supported by a wheeled undercarriage that was initially not embedded on the gravel surfacing. As it was dragged forward, slight embedment occurred, which was accounted for in anchor capacity calculations. The yarder was a Koller 501 Trailer, yarding an Acme 15 motorized carriage. The yarding profile is presented in Fig. 4. A cable load cell was placed directly on the skyline and recorded loading at a frequency of 20 Hz. After each successive yarding cycle, movement was measured between a fixed, stationary survey point on the ground to a fixed point on the vehicle. A LCCC was generated by determining the maximum load for a given yarding profile for the initial position and 0.25 meter increments of forward, horizontal movement towards the direction of pull. These points were then fitted using an exponential function, where F was 164 kN and S was 0.888 (Fig. 5). The anchor capacity for two similar scenarios is presented, and used with this LCCC to use the ARC to predict mobile anchor movement. Test 1 involved mobile anchor conditions with a hillslope of 3.4 degrees, skyline angle of pull of 17 degrees, a subsurface of compacted road aggregate (f=45Ë&#x161;, g=18.9 kN/m3) with no initial blade embedment, resulting in a predicted anchor capacity of 122Â kN. The presence of a rubber tired undercarriage led to no assumed, added cohesive shear forces along the base (no tracked suspension). Use of the LCCC and the ARC (i.e. equations 8 and 9) allowed a comparison of recorded movement to predicted movement based on recorded load and calculated anchor capacity for the given configuration (Fig. 6). The first loading cycle resulted in a dynamic load that approached the calculated anchor capacity, resulting in recorded movement. However, since the load barely exceeded the predicted anchor capacity, very little movement was observed from the ARC. However, subsequent cycles that exceeded anchoring capacity resulted in predicted movement that agreed relatively well with recorded movement for the equipment. The final movement for the five loading cycles was 1.60 meters and 1.65 meters for the recorded and predicted movements, respectively (Fig. 6). Note that the anchor capacity increased slightly with deepening blade embedment, approaching approximately 15 cm as the anchor shifted forward. Croat. j. for. eng. 37(2016)2

Designing Mobile Anchors to Yield: A Tension Relief System for Tail Anchoring (269â&#x20AC;&#x201C;278)

B. Leshchinsky et al.

Fig. 6 Cable loading, recorded equipment movement, anchor capacity and predicted equipment movement for test 1 over a condensed timeframe (for clarity) Test 2 involved mobile anchor conditions with a hillslope of 4 degrees, skyline angle of pull of 18 degrees, a subsurface of compacted road aggregate (f=45Ë&#x161;, g=18.9 kN/m3) with no initial blade embedment, resulting in a predicted anchor capacity of 122.2Â kN. Use of the LCCC and the ARC allowed for a comparison of recorded movement to predicted movement based on recorded load and calculated an-

chor capacity for the given configuration (Fig. 7). The first two loading cycles resulted in a dynamic load that exceeded the predicted anchor capacity, resulting in movements similar to that predicted from the ARC. Later cycles that exceeded anchoring capacity resulted in predicted movement that exceeded recorded movement for the equipment, but fell within a reasonable range. The final movement for the loading cycles was

Fig. 7 Cable loading, recorded equipment movement, anchor capacity and predicted equipment movement for test 2 over a condensed timeframe (for clarity) Croat. j. for. eng. 37(2016)2

275

B. Leshchinsky et al.

Designing Mobile Anchors to Yield: A Tension Relief System for Tail Anchoring (269â&#x20AC;&#x201C;278)

pacity was exceeded. On average, the dynamic loading accounted for approximately 27% of the skyline tension, agreeing with prior studies (Jorgensen et al. 1977).

4. Conclusions

Fig. 8 Detail of cable tension relief and return to equilibrium with anchor movement 1.70 meters and 2.40 meters for the recorded and predicted movements, respectively (Fig. 7). Again, anchor capacity increased slightly with deepening blade embedment, approaching approximately 22 cm as the anchor shifted forward. Recorded cable tensions showed a unique phenomenon when exceeding anchor capacity; specifically, a sharp drop in cable tension as the vehicle shifted forward to relieve loading. This drop in dynamic loading presents a relief for dangerous cable loads that may exceed safe working conditions for the skyline or yarder. As a large dynamic load occurs in the skyline, the tail anchor or yarder must give to the force â&#x20AC;&#x201C; when the tailhold is free to slide, the tensions rapidly drop due to the increase in sag in the skyline (Fig. 8). As the equipment slides forward and cable tensions drop, the anchor capacity eventually becomes greater than the skyline tension resulting in a state of equilibrium again. This movement can be predicted using the LCCC and ARC, and subsequently can be developed into a design methodology preventing yarder overturning and skyline breakage if adequate clearance is given to allow mobile anchors to move. Dynamic loading of the skyline played a significant role in anchor movement and accounted for a significant portion of loading. As yarding of a payload occurred, dynamic loads could be significant, accounting for up to a 41% increase in skyline tension compared with the subsequent return to equilibrium after anchor movement. Dynamic loading accounted for 7% to 41% of the total cable loading in scenarios where anchor ca-

276

Tailhold mobile anchor capacity combined with the cable tension behavior of a loaded skyline enables a prediction of movement upon exceedance of soil resistance, presenting a means of relieving excessive cable tensions for safer yarding operations. The analytical solution based on a static-based anchor capacity equation (Leshchinsky et al. 2015) and a load-cable characteristic curve, easily determined from cable mechanics analyses or publicly available software (Skyline XL, USFS 2015), present a means of predicting movement for a mobile anchor. This prediction can be used for designing adequate clearance distance for a mobile anchor tailhold with a known skyline load. The following conclusions resulted from this study: The predicted anchor capacity solution presented good agreement with field testing of mobile anchors employed as skyline tailholds. In the field tests, the predicted anchor capacity was within 15% of the actual loading when sliding failure occurred. This method of determining anchor capacity presents a means for designing mobile anchors for stability, or to yield under controlled conditions when skyline tensions may be relieved. Combined use of the load-cable characteristics curve (LCCC) and mobile anchor capacity equation present a piecewise function (the anchor reaction curve, ARC) that presents a means of predicting the point at which an anchor will move, and quantifying approximate movement when it does occur. Inversely, the use of the ARC function with known cable loading, site conditions and yarding configurations will allow for appropriate clearance for mobile anchor application, ensuring safe yielding conditions without catastrophic failure or loss of heavy equipment in steep terrain. When a mobile anchor capacity is exceeded, it may slide forward (when sliding conditions are critical and overturning is not a concern), relieving cable tensions by increasing the cable sag, eventually returning to equilibrium. Equilibrium is reached due to a relief in skyline tensions and an eventual recovery of soil resistance, sometimes occurring due to increased equipment embedment. Dynamic loads within a skyline may contribute over 40% to the skyline tension in yarding. On average, when the mobile anchor failed, the dynamic loads accounted for 26% of the cable tension, subsequently decreasing back to equilibrium after yield. Yield was defined as sliding. Croat. j. for. eng. 37(2016)2

Designing Mobile Anchors to Yield: A Tension Relief System for Tail Anchoring (269–278)

This study outlines an approach which may enable safer use of yarding systems due to mobile anchors serving as a tension relief system for a skyline. This function prevents excessive skyline loads and yarder overturning due to the sliding mechanism of the mobile anchor and subsequent reduction of tension. Like any design consideration, one must consider potential drawbacks in application. The most important consideration is that sliding failure must govern the mobile anchor application. Leshchinsky et al. (2015) presented a design approach that accounted for not only sliding failure, but overturning of mobile anchors, where the critical failure mechanism had to be accounted for. To employ yielding mobile anchors, it is critical to ensure that the maximum anchor resistance attained from the sliding mechanism is less than that of overturning. If sliding is not the failure mechanism, the equipment may overturn, resulting in a loss of function. Correct calculation of mobile anchor capacity involves an understanding of soil properties. It is important that the in-situ shear strength of the soil is estimated using reasonable methods. Mobile anchors require adequate clearance from dangerous drop-offs or slopes. Although movement can be predicted, the ARC is sensitive to extremely large loads and a sufficient buffer must be implemented to ensure that catastrophic movements do not occur. Like any design, conservatism in anchor setback ensures best results. Cable attachments must also be adequately protected from potential damage upon sharp surfaces on the equipment. Yielding and tension relief typically requires the use of lighter equipment with additional resistance provided by blade emplacement and possibly supplemental weights to match the design skyline tension. The use of very heavy equipment could lead to no anchor movement under significant skyline loads, placing the cable under potentially unsafe tension loads. Movement of equipment could change anchoring configuration and the interface properties between the machine and surface. Considerations need to be made based on available equipment, skyline selection and a given yarding profile. Despite these constraints, the use of mobile anchors presents a means of anchoring skylines that may improve safety as it does not catastrophically fail like stumps, buried deadmen, or plate anchors, and does not exceed safe working conditions for the skyline or yarding tower. The tension relief that occurs in the skyline due to enabling anchor movement may present a new alternative for increasing safety and efficiency of cable logging operations. Croat. j. for. eng. 37(2016)2

B. Leshchinsky et al.

Acknowledgement The authors would like to acknowledge the United State Forest Service (USFS) San Dimas Technology and Development Center and Oregon OSHA for support of this work. Input and advice from Mike Barger (USFS) and Mark Russell (USFS) were greatly appreciated.

5. References Brown, C., Sessions, J., 1996: A maximum load path solution for the standing skyline. Forest Science 42(2): 220–227. Chen, W.F., 2008: Limit analysis and soil plasticity. Ft. Lauderdale, FL: J.Ross Publishing, 638 p. Cole, R., Rollins, K., 2006: Passive earth pressure mobilization during cyclic loading. Journal of Geotechnical and Geoenvironmental Engineering 132(9): 1154–1164. Copstead, R.L., Studier, D.D., 1990: An earth anchor system: installation and design guide. U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station, 35 p. Fakharian, K., Evgin, E., 1997: Cyclic simple-shear behavior of sand-steel interfaces under constant normal stiffness condition. Journal of Geotechnical and Geoenvironmental Engineering 123(12): 1096–1105. Hartsough, B.R., Visser, R.J.M., Miles, J.A., Drews, E.S., 1997: Improved cable logging anchors for rocky soils. Transactions of the ASAE 40(1): 261–266. Hu, L., Pu, J., 2004: Testing and modeling of soil-structure interface. Journal of Geotechnical and Geoenvironmental Engineering 130(8): 851–860. Jorgensen, J., Carson, W., Chalupnik, J., Garbini, J., 1977: Skyline anchor dynamics test. Technical report FE-UW-7702. Seattle, WA: University of Washington, Mechanical Engineering, 74 p. Kendrick, D., Sessions, J., 1991: A solution procedure for calculating the standing skyline load path for partial and full suspensions. Forest Products Journal 41(9): 57–60. Leshchinsky, B., Sessions, J., Wimer, J., 2015: Analytical design charts for mobile anchor systems. International Journal of Forest Engineering 26(1): 10–23. Olund, D., 2001: The future of cable logging. International Mountain Logging and 11th Pacific Northwest Skyline Symposium, Seattle, Washington, USA, 263-267. Oregon Occupational Health and Safety Administration. (OR-OSHA), 2008: Division 7 Forest Activities. Oregon Occupational Safety and Health Standards 2008. Salem, OR. OR-FACE, 2016: Logging and Forestry Accidents. Retrieved April 01, 2016, from http://www.ohsu.edu/xd/research/centers-institutes/oregon-institute-occupational-health-sciences/ outreach/or-face/reports/logging-and-forestry.cfm Peters, P.A., Biller, C.J., 1985: Preliminary evaluation of the effect of vertical angle of pull on stump uprooting failure. Proceedings of 9th Annual Council on Forest Engineering Meeting. Mobile, AL. 90–93.

277

B. Leshchinsky et al.

Designing Mobile Anchors to Yield: A Tension Relief System for Tail Anchoring (269–278)

Peltola, H., Kellomäki, S., Hassinen, A., Granander, M., 2000: Mechanical stability of Scots pine, Norway spruce and birch: an analysis of tree-pulling experiments in Finland. Forest Ecology and Management 135(1): 143–153. Pyles, M.R., Anderson, J.W., Stafford, S.G., 1991: Capacity of second-growth Douglas-fir and Western Hemlock stump anchors for cable logging. International Journal of Forest Engineering 3(1): 29–37. Pyles, M.R. 1984: Vane shear data on undrained residual strength. Journal of Geotechnical Engineering, 110(4): 543– 547. Smith, C.C., Gilbert, M., 2007: Application of discontinuity layout optimization to plane plasticity problems. Proceedings of the Royal Society of London A: Mathematical, Physical and Engineering Sciences 463(2086): 2461–2484. Smith, J.F., 1995: The anchorage capacity of Pinus radiata guyline stump anchors used in cable logging operations. Doctoral dissertation, Lincoln University. Stampfer, K., 1999: Influence of terrain conditions and thinning regimes on productivity of a track-based steep slope harvester. Proceedings of the International Mountain Logging and 10th Pacific Northwest Skyline Symposium. Corvallis, OR. 78–87. Visser, R., Stampfer, K., 2015: Expanding ground-based harvesting onto steep terrain. Croation Journal of Forest Engineering 36(2):133–143. Visser, R. Berkett, H., 2015: Effect of terrain steepness on machine slope when harvesting. International Journal of Forest Engineering 26(1): 1–9. Tabucanon, J.T., Airey, D.W., Poulos, H.G., 1995: Pile skin friction in sands from constant normal stiffness tests. ASTM geotechnical testing journal 18(3): 350–364. USFS, 2015: SKYLINE XL. Last accessed on January 25, 2015. http://www.fs.usda.gov/detail/r6/landmanagement/resourc emanagement/?cid=fsbdev2_027048.

Appendix Table 1 The symbols and abbreviations used in this study Symbol

Description

A At c’ Db Fmax Fcrit Kp Pp S Tg wb Wt wt b DL g

Load factor Footprint area of vehicle Soil cohesion Embedment depth of blade Maximum skyline load Calculated anchor capacity Passive earth pressure coefficient Soil passive earth pressure in front of blade Site constant Anchor capacity Width of embedded blade Vehicle weight Width of tracks/tires (if applicable) Slope angle Movement Unit weight of soil Interface friction angle between vehicle base and soil Cable angle of pull Soil angle of internal friction

dt Q f’

Units kN

m2 kPa m kN kN – kPa

m-1 kN m kN m ° m

kN/m3 ° ° °

Authors’ address:

Received: February 9, 2015 Accepted: April 4, 2016

278

Assist. prof. Ben Leshchinsky, PhD. * e-mail: ben.leshchinsky@oregonstate.edu Prof. John Sessions, PhD. e-mail: john.sessions@oregonstate.edu Jeffrey Wimer e-mail: jeffrey.wimer@oregonstate.edu Oregon State University Department of Forest Engineering Resources and Management Milo Clauson e-mail: milo.clauson@oregonstate.edu Oregon State University Department of Wood Science and Engineering OR-97330 Corvallis USA * Corresponding author Croat. j. for. eng. 37(2016)2

Original scientific paper

Frequency and Advantages of Animal-Powered Logging for Timber Harvesting in Hungarian Nature Conservation Areas Ă kos Malatinszky, Csilla Ficsor Abstract Despite the fact that the management of forests in nature conservation areas represents a key interest in Europe, animal-powered logging is only rarely covered by scientific papers. The main aim of this study was to explore the occurrences of this practice in Hungarian nature conservation areas (i.e. that belong to IUCN Categories Ia, II or V as well as Natura 2000 SPA or SAC sites) that are owned by the state (i.e. about 75% of all the forested nature conservation areas), and compile the characteristics, advantages and disadvantages of animal-powered logging according to interviews with loggers. All the Forest Districts in Hungary (116) were contacted by phone. 26% of the Forest Districts apply animal logging regularly (draft horse in every case), but only 2.6% (a mere 3 Forest Districts) own a horse stock for this purpose, while the others (27) employ contractors. 7.76% (9 Forest Districts) use animal power for skidding only occasionally and none of them own horses. All operate at least partially on nature conservation areas. Although 2.6% of the Forest Districts own a horse stock, they do not use animal power for skidding. The average animal logging operation consists of two animals and three people (one of them leads the horses) that do thinning in young stands and selection work. Data on advantages and disadvantages was gathered via semi-structured onsite interviews. The daily logging capacity of a single draft horse ranges from 4.5 m3 to 30 m3, depending on topography and weather conditions. Resulting advantages include: less harm to topsoil, wood stands, and saplings, the support of natural regeneration of forests, horses are more economical than machines in thick snow, there is no use of fossil fuels, more jobs are created, and preservation of native horse breeds is supported. Conversely, output capacity is less compared to the use of machines, and only few people want or are able to work with draft horses. It is mostly the terrain conditions (especially efficient in steep terrain) and the environmental constraints that determine the use of horses instead of skidders. Keywords: animal logging, draft horse, horse breed, nature conservation area, log skidding, forest district

1. Introduction Skidding of logs requires high attention to the forest topsoil, the wood stand, and saplings. Animalpowered logging is considered to be gentle to the soil (Zimmermann 1994) and reduces damage to residual trees and seedlings (Rodriguez and Fellow 1986, Wang 1997). This is why skidding with animal power still carries a high value, especially in nature conservation Croat. j. for. eng. 37(2016)2

areas, despite the fact that fully mechanized harvesters and forwarders have almost completely ousted the draft horse from the woods (Engel et al. 2012). Before the mechanization of the timber harvesting industry (i.e. the 1950s), animal power had been extensively used for skidding of logs throughout Hungary. Since the 1990s, however, it has almost disappeared from practice (FirbĂĄs 1996). This is similar to the

279

Ă . Malatinszky and C. Ficsor

Frequency and Advantages of Animal-Powered Logging for Timber Harvesting ... (279â&#x20AC;&#x201C;286)

worldwide trend of the fast mechanization of logging, leaving behind the traditional systems such as horse and mule logging (Rodriguez and Fellow 1986). The timber harvesting industry went through a rather quick mechanization process in Hungary after World War II in that 50% of skidding had been mechanized by 1954 (Keresztesi 1982). Strengthening environmental consciousness and the commitment to sustainable development have brought animal-powered skidding to light, especially after the adoption of the new Forest Act in 2009, which ordered a step-by-step conversion from intensive forest management methods towards continuous cover forestry (CCF) in the case of the state Forest Districts. Based on this legislation, small-scale timber harvesting methods, as well as close-to-natural skidding, are expected to prevail. National park directorates, which coordinate conservation management of nature conservation areas, have also urged for the application of less harmful methods e.g. by logging invasive tree trunks from valuable areas in ways that cause minimal harm to the forest floor. The main constraint in terms of forestry utilization process in national parks and landscape protection areas (i.e. IUCN Category II and V) is that the condition (state of naturalness) of forests must not decay, and it should be subordinated to nature conservation goals. The same applies for Natura 2000 SAC and SPA sites. In strictly protected areas (IUCN Cat. Ia), only conservation management is permitted, which is without any economic aim. Despite the fact that the management of forests in nature conservation areas represents a key interest in Europe (especially after the introduction of the Natura 2000 regulations), animal-powered logging is only rarely covered by scientific papers. Thus, our objective was to provide new information on the utilization of this forestry management process. The main aim was to explore the occurrences of animal-powered logging in Hungarian nature conservation areas (i.e., that belong to IUCN Categories Ia, II or V as well as Natura 2000 SPA or SAC sites) that are owned by the state (this means about 75% of all the forested nature conservation areas), and compile the characteristics of the practice, the aspects that determine why horses are used for logging instead of machines, and the advantages and disadvantages according to interviews with loggers.

2. Materials and methods This study covers the whole area of Hungary. Altogether, 116 Forest Districts manage the National Forest Estate. All of them were contacted by telephone between February and October 2014. Besides basic data

280

(operational area, and nature conservation areas within that, etc.), they were asked about using animal power for skidding, and whether they own a horse stock for this purpose, or employ contractors (enterprises). Contractors and forestry workers (if the horse stock is owned by the Forest District), who use animal power (i.e. loggers) for skidding in nature conservation areas (either national parks (IUCN Cat. II) or landscape protection areas (IUCN Cat. V), and strictly protected sites within them (IUCN Cat. Ia) that are also designated Natura 2000 SPA or SAC sites) were contacted in person between June and September 2014. Their names and addresses were provided by the Forest Districts during their phone call. We prepared semi-structured interviews based on open ended questions following the method of Babbie (2012). The interviews were done on-site with individual responses given in person with a dialogue structure that gave opportunity for discussions. Participants were asked about why they log with animals, where they work with them, what equipment they use, what is the average quantity logged by a horse for 1 turn and the daily capacity of a single horse, the size of the area that is logged by horses, the number of horses and workers, and the nature conservation values of the area that justify the use of animal power. After this, we asked them to present their opinion on the advantages and disadvantages of using animal power for skidding, horse breeds and types used for skidding, and the characteristics that make them beneficial for this purpose. Loggers were interviewed in 17 different areas while logging (Fig. 1). These study areas are all covered by mixed broadleaved stands: 9 of them in the Northern Hungarian Mountain Range, 6 of them in the Trans-danubian Mountains, and 2 in the Southern Transdanubian Region. They belong to 11 different state Forest District companies. Photo and video documentation was recorded during field observations.

3. Results 3.1 Occurrences of animal-powered logging in Hungarian nature conservation areas owned by the state Interviews by phone showed that almost 26% of Hungarian Forest Districts (i.e. 30 Forest Districts) regularly use horses for skidding (and on-site processing). However, only 2.59% of them (a mere 3 Forest Districts) own a horse stock for this purpose (number of horses is 2 to 8), while the other 27 Forest Districts employ contractors (enterprises) for skidding. 7.76% Croat. j. for. eng. 37(2016)2

Frequency and Advantages of Animal-Powered Logging for Timber Harvesting ... (279â&#x20AC;&#x201C;286)

Ă . Malatinszky and C. Ficsor

Fig. 1 Study sites, venues of field interviews (9 Forest Districts) have reported that skidding with animal power is used only occasionally (2 or 3 times per year, skidding about 100â&#x20AC;&#x201C;200 m3 timber); and none of them own horses for these works. Although 2.59% of the Forest Districts own a horse stock, they use animal power for transporting forage for games (winter feeding) or for hunting rather than skidding (Fig. 2). If animal power is applied for logging, it is draft horses that are used for skidding (100%) and no mules are used in Hungarian forests. A probable reason for this might be that timber may be harvested only beyond the vegetation period in nature conservation areas and horses tolerate cold weather much better than mules, as argued by Shrestha and Lanford (2005). All of the Forest Districts that use horses for skidding (either occasionally or regularly) operate (at least partially) on nature conservation areas as well. 66.38% of Hungarian Forest Districts do not apply animal power in any way. 6 of them claim to use animal power, but due to various difficulties they have so far Croat. j. for. eng. 37(2016)2

not been able to implement horses for skidding. These difficulties include the lack of people with deep knowledge of the skidding methods and horse care. Another reason is the lack of a state-owned horse stud that might be selected for forestry works, and in some cases, the amount of work is too low to maintain a contractor. 22 of those Forest Districts that do not apply animal power were asked about the last year they did it; they reported that it was 2 to 33 years ago. Some of them owned as many as 60 horses at that time (and even more before the mechanization of the timber harvesting industry). This suggests that this historical way of logging has only just recently started to disappear from the Hungarian forests.

3.2 Characteristic factors of animal-powered logging in Hungarian forests The size of each area that was logged by horse(s) at the time of our on-site interviews ranged between 0.5 ha and 23 ha, with the average being 6.05 ha. This

281

Á. Malatinszky and C. Ficsor

Frequency and Advantages of Animal-Powered Logging for Timber Harvesting ... (279–286)

Most of the loggers (37%) apply the cross-breeding of warm-blooded and cold-blooded horse due to their pace and smaller size. Others reported the use of coldblooded types due to their capacity, such as Muraközi, Percheron, Nonius and the Hungarian cold-blooded horse. The cold-blooded ty pes are slower and calmer, and therefore they rarely suffer foot injuries. Horses are usually fed three times a day; forages are oat, maize, barley, alfalfa and hay. The logging capacity of a single draft horse in an average day is 15.5 m3. However, this amount ranges on a wide scale from 4.5 m3 (this was measured on a steep area, covered by snow) till 30 m3 (on a plain area, with dry weather conditions and 5 workers). Based on these data, it can be stated that the daily average capacity of a single draft horse mainly depends on the surface relief of the land and the weather conditions. Fig. 2 Frequency of horse logging in Hungarian state-owned forests also means that all of the studied animal loggers work on tracts with low timber volumes. The ratio of animal logging, compared to the use of machines, varies highly among the Forest Districts. In some mountain areas, the use of animals is almost exclusive. Still, horse logging is not negligible in several flat areas. Among forestry works, horses are mostly used for thinning and selection works, but clearcutting also occurs. The types of forest stands logged by horses are in most cases beech or beech-hornbeam mixed forests (40%), Turkey oak – sessile oak stands (18%), Sessile oak – hornbeam stands (12%), European black pine (in 1 case mixed with hornbeam) (18%), Norway spruce (6%), and black locust (6%). Trees are felled at the stump and then delimbed in order to avoid harm to the topsoil, wood stand, and saplings during skidding. The average animal logging operation consists of two horses (varying between 1 and 3; being 2 in 14 cases out of 17) and three people (varying between 2 and 5; being 3 in 9 cases out of 17) (Table 1). Usually one worker performs tree felling, another performs tree processing, and a third one is engaged with leading the (usually two) horses. This is similar to the data found by Dubois et al. (2001), which was collected in a comparable US area, and also strengthens the statement of Magagnotti and Spinelli (2011b), who said that if there are two horses per driver, it increases the costefficiency of horse skidding with a dramatic increase of the average payload.

282

The mean quantity of timber a draft horse can pull in one turn is 0.81 m3, with 1 m3 of timber per turn in most cases. The timber is usually dragged to the horse with a chain or a rope. Most of the people who care for the horses gained their knowledge from their family. Using horses for forestry work is a tradition in several families.

3.3 Advantages and disadvantages of animal-powered logging from the aspects of forestry and nature conservation Skidding with animal power is applied in hilly areas (covered by mixed broadleaved stands) rather than on flat areas. At the time of field observation, 15 of the 17 loggers were working in nature conservation areas, 3 of them being under strict protection. This fact verifies the demand for animal-powered skidding in those forests where the priority is the protection of natural values. It was also reported, however, that the use of horses might be hard in case of selective logging (which is a general practice in continuous cover forestry) due to the high weight of the trunks. According to the most common opinion, loggers use horses for thinning in young stands (10 to 50 years old), because the stand is too dense for machines without causing a lot of extra damage. Horses, however, can maneuver easily in these areas. Borz and Ciobanu (2013) also reported from Romanian areas that horses are mostly applied in very young and dense stands where thinning operations are done. It was reported in several cases that using animals instead of machines reduces damage to residual trees and seedlings. The Forest Districts favor the natural regeneration of forests, which is better supported by animal logging. Contractors, who work in a national Croat. j. for. eng. 37(2016)2

Frequency and Advantages of Animal-Powered Logging for Timber Harvesting ... (279–286)

Á. Malatinszky and C. Ficsor

Table 1 Characteristic factors of animal-powered logging in Hungarian forests Name of Forest District

Quantity Quantity Studied logged by 1 logged by 1 Used horse Number of Number of IUCN area horse during horse in 1 breeds horses workers Category 1 day ha turn m3 m3

Forest stand type

Harvesting method

Reach in the area

Eger

23.0

12.5

0.28

Percheron

2+1

Sessile oak Hornbeam

Group selection

On foot

Eger

5.5

17.5

1.50

Hungarian c.-blooded, cr.-breed

Beech

Group selection

Local accommodation

Nagymaros

7.4

20.0

1.00

Cr.-breed

Beech

Thinning

On foot

Telkibánya

0.5

15.0

1.00

Muraközi

Beech

Other

On foot

Telkibánya

2.5

10.0

1.00

Nonius, Hungarian c.-blooded

Beech

Single-tree selection

On foot

Pilisszent-kereszt

4.5

15.0

1.00

Belgian and Hungarian c.-blooded

Turkey oak Sessile oak

Single-tree selection

On foot

Királyrét

5.5

25.0

0.50

Belgian c.-blooded

Beech

Single-tree selection

On foot

Sásd

2.5

12.0

0.50

Cr.-breed

Turkey oak Sessile oak

Thinning

Motor-horsebox

Kemence

7.5

8.5

1.00

Cr.-breed

Beech Hornbeam

Thinning

Local accommodation

Bajna

15.9

4.5

0.50

Cr.-breed, Hungarian c.-blooded

Sessile oak Hornbeam

thinning

On foot

Vasvári

5.0

30.0

1.50

Cr.-breed, Muraközi

Hornbeam Black pine

Thinning

On foot

Bakonybél

2.0

10.0

1.00

Percheron

Norway spruce

Other

On foot

Bakonybél

3.0

8.5

1.00

Muraközi

Beech – Hornbeam

Thinning

On foot

Szombathely

6.5

20.0

0.50

Cr.-breed, Muraközi

Black pine

Thinning

On foot

Mátraszőlős

5.0

25.0

0.50

Cr.-breed

Turkey oak Sessile oak

Group selection

On foot

Cserépváralja

0.5

25.0

0.50

Cr.-breed

Black locust

Clearcut

On foot

Pécsi Parkerdő

80.0

14.0

0.25

Cr.-breed

Black pine

Other

Motor-horsebox

Average

6.05

15.5

0.81

Cr.-breed

II or V

Beech

Thinning

On foot

park (IUCN Cat. II) and in a landscape protection area (IUCN Cat. V), reported that logging in European black pine (Pinus nigra) stands is always made by using draft horses (not machines) as this is harmless for the saplings of indigenous species (especially manna ash (Fraxinus ornus)) and topsoil. Therefore, horse Croat. j. for. eng. 37(2016)2

skidding might be beneficial for the regeneration of native forests. Many interviewees highlighted that horse logging is gentle to the soil. In areas with shallow topsoil, skidding with animal power is favorable as it causes less harm to the topsoil than machines. Therefore, native

283

Á. Malatinszky and C. Ficsor

Frequency and Advantages of Animal-Powered Logging for Timber Harvesting ... (279–286)

species continue to be dominant, while invasive species that flourish in damaged and disturbed areas, such as the tree of heaven (Ailanthus altissima), are kept out. Since most of the Forest Districts do not own a horse stock any more, the former stables (maintained by them) also ceased to exist. Thus, it is hard to find accommodation for the contractors and their horses. As a result, contractors then have to reach the area to be logged on foot, which sometimes means a 30 km ride back and forth, and adding up to 3 hours of activity per day. It is beneficial that the horses are mostly privately owned (not by the Forest District). It is necessary, from a horse’s perspective, as the same person always works with it. This makes the teamwork between the two go smoother when completing logging. Using horses for logging is possible in any kind of weather conditions, except for deep mud, heavy rainstorm, and thick snow cover. In the case of snow, however, interviewed loggers reported deep snow to be problematic from about 20 to 50 cm or more. If the snow is not too deep for horses, then they should be used, as it is more economical than machines. In deepsnow conditions, machines need to be chained, thus increasing fuel consumption. The revenue for horse logging per cubic meter is usually the same price as for using machines (in some exceptional cases, however, horse loggers earn more). This is disadvantageous for those who use horses. Therefore, the benefits for nature conservation and forest regeneration aspects that result from using animalpower over machines should be made more known. Nature conservation areas can be logged only during winter. Therefore, the horses and the people working with them have to find another job for the vegetation period. This usually means operating a carriage or a chariot for tourism.

4. Discussion and conclusions It has been proven that the main advantage of horse logging is that it causes less logging damage to the wood stand and saplings, due to its low pressure impact to the ground. Furthermore, draft horses are able to maneuver easier than machines in dense young forest stands and thus cause less harm to the wood. The environmental significance of animal-powered logging was also highlighted by several interviewees as being an almost »zero emission« method, as there is no direct use of any fossil fuels (see also Rydberg and Jansén 2002) and, therefore, has the lowest green-

284

house gas emissions (Engel et al. 2012). While animal traction is inherently environmentally friendly, properly managed mechanized operations can also offer good environmental performances (Spinelli et al. 2013). As it is a labor-intensive operation (Rodriguez and Fellow 1986) and through ensuring forage supply for the horses and caring for them, animal-powered logging may create jobs in distant rural areas, which is beneficial for achieving the aims of rural development policies and the principles of sustainability. It has already been mentioned that animal logging is especially recommended for deployment in conservation areas with strict environmental constraints (McCabe and Tiner 1992). The new Hungarian Forest Act (adopted in 2009) prescribes conversion of forest management towards continuous cover forestry (CCF) in state forest areas, which means harvests that use selective thinnings. These low-intensity cuts reduce the performance and the profit margins expected from highly mechanized timber harvesting (Mason et al. 1999), therefore, they claim for a switch towards slighter methods (not mentioning reduced impact logging (RIL) in the Hungarian legislation), creating space for traditional small-scale harvesting alternatives (such as animal-powered skidding) leading to a new flourish of draft horse logging. Assessment of ecological and economic impacts of different forest management methods are among top priority research questions for the conservation of biodiversity in Hungary, as compiled during a recent participatory process (Mihók et al. 2015). Another aspect is the high cost of mechanized harvesting on small woodlots. This makes animal-powered logging more cost effective, especially when preexisting skidding trails are not available (Magagnotti and Spinelli 2011b), even if it contains more non-productive time elements (Shrestha et al. 2005). Several loggers reported that the use of horses is an especially efficient tool for log extraction on steep terrain, which was also stated by Magagnotti and Spinelli (2011b). Although several authors report that the main reason for using horses for timber logging is the lack of capital for modern technology (Toms et al. 2001, Jourgholami et al. 2010, Magagnotti and Spinelli 2011a), our interviews showed that it is much more the terrain conditions of the area and the constraints by nature conservation authorities that determines the use of horses instead of skidders (Wang (1997) strengthens this argument). Only a few people intend and are actually able to work with draft horses, or have knowledge and experience in this field, as it is a strict and hard way of life. They must find another job for the vegetation period, Croat. j. for. eng. 37(2016)2

Frequency and Advantages of Animal-Powered Logging for Timber Harvesting ... (279–286)

a time when logging is prohibited by the Forest Act. There is also a strong need to find a role with a reasonable income for the horses during the vegetation period. 6 Forest Districts have so far not been able to implement horse skidding due to the lack of people with in-depth knowledge on how to work with the animals and horse care. Meanwhile, we strongly recommend further development of the equipment of animal-powered skidding in Hungary, as most loggers use almost century-old tools and devices. This process would increase the efficiency of skidding and reduce the gap of output values between machines and horses. The fact that even distant sites can be reached and logged by draft horses without excessive damage might be considered either an advantage or a disadvantage, as it might be undesirable for strictly protected natural areas (IUCN Category Ia) to harvest timber at all. Horse skidding can, however, still be advisable for both national park directorates and for small-scale timber producers. It is a viable and sustainable method with relatively small investments and operating costs, and it also carries benefits for rural development. In conclusion, it can be stated that this historical way of logging started to disappear from the Hungarian forests 1 to 4 decades ago. Currently 26% of Hungarian Forest Districts apply animal logging regularly and 7.76% occasionally (but only 2.59% among them own their own horses); all operate at least partially on nature conservation areas and in every case draft horses are used. All animal loggers work on tracts with low timber volumes, do thinning and selections in young stands (10 to 50 years old), and the use of animals is almost exclusive in steep areas with strict environmental constraints. Using animals instead of machines reduces damage to the topsoil, the residual trees and seedlings, and supports the natural regeneration of forests. Furthermore, it is more economical than machines in thick snow. Usually two horses (14 cases out of 17) and three people (9 cases out of 17) work together: one worker performs tree felling, another performs tree processing, and a third one is engaged in leading the horses. The daily logging capacity of a single draft horse ranges from 4.5 m3 to 30 m3, depending on the surface relief of the land and the weather conditions. Loggers and horses sometimes spend up to 3 hours per day to reach the area to be logged. Horse logging should play a crucial role in the preservation of native horse breeds such as the Hungarian coldblooded and Muraközi, which are especially suitable for work as they are characterized as being resistant and calm. Croat. j. for. eng. 37(2016)2

Á. Malatinszky and C. Ficsor

Acknowledgements Authors are grateful for the support of the Kutató Kari Kiválósági Támogatás – Research Centre of Excellence (11476-3/2016/FEKUT).

5. References Babbie, E., 2012: The Practice of Social Research, 13th edition. Cengage Learning 2012, 608 p. Borz, S.A., Ciobanu, V., 2013: Efficiency of motor-manual felling and horse logging in small-scale firewood production. African Journal of Agricultural Research 8(24): 3126– 3135. Dubois, M.R., Toms, C.W., Bliss, J.C., Wilhoit, J.H., Rummer, R.B., 2001: A Survey of Animal-Powered Logging in Alabama. Southern Journal of Applied Forestry 25(1): 17–24. Engel, A.M., Wegener, J., Lange, M., 2012: Greenhouse gas emissions of two mechanised wood harvesting methods in comparison with the use of draft horses for logging. European Journal of Forest Research 131(4): 1139–1149. DOI 10.1007/s10342-011-0585-2. Firbás, O., 1996: Erdőhasználattan I., Mezőgazdasági Szaktudás Kiadó, Budapest, 260 p. Jourgholami, M., Majnounian, B., Feghhi, J., Visser, R.J.M., 2010: Timber extraction with mules: A case study in the Hyrcanian Forest. African Journal of Agricultural Research 5(22): 3108–3115. Keresztesi, B., 1982: Magyar erdészet 1954-1979. Akadémiai Kiadó, Budapest, 390 p. Magagnotti, N., Spinelli, R., 2011a: Integrating animal and mechanical operations in protected areas. Croatian Journal of Forest Engineering 32(2): 489–499. Magagnotti, N., Spinelli, R., 2011b: Financial and energy cost of low-impact wood extraction in environmentally sensitive areas. Ecological Engineering 37(4): 601–606. Mason, B., Kerr, G., Simpson, J., 1999: What is continuous cover forestry? Forestry Commission Information Note 29. Forestry Commission, Edinburgh, 8 p. McCabe, P., Tiner, E., 1992: Mule logging: a dying art? Treasures Forests (Spring Issue): 14–15. Mihók, B., Kovács, E., Balázs, B., Pataki, G., Ambrus, A., Bartha, D., Czirák, Z., Csányi, S., Csépányi, P., Csőszi, M., Dudás, G., Egri, C., Erős, T., Gőri, S., Halmos, G., Kopek, A., Margóczi, K., Miklay, G., Milon, L., Podmaniczky, L., Sárvári, J., Schmidt, A., Sipos, K., Siposs, V., Standovár, T., Szigetvári, C., Szemethy, L., Tóth, B., Tóth, L., Tóth, P., Török, K., Török, P., Vadász, C., Varga, I., Sutherland, W.J., Báldi, A., 2015: Bridging the research-practice gap: Conservation research priorities in a Central and Eastern European country. Journal for Nature Conservation 28: 133–148. Rodriguez, E.O., Fellow, A.M., 1986: Wood extraction with oxen and agricultural tractors. FAO forestry paper no. 49, Rome, 92 p.

285

Á. Malatinszky and C. Ficsor

Frequency and Advantages of Animal-Powered Logging for Timber Harvesting ... (279–286)

Rydberg, T., Jansén, J., 2002. Comparison of horse and tractor traction using emergy analysis. Ecological Engineering 19(1): 13–28.

Toms, C.W., Dubois, M.R., Bliss, J.C., Wilhoit, J.H., Rummer, R.B., 2001: A Survey of animal-powered logging in Alabama. Southern Journal of Applied Forestry 25(1): 17–24.

Shrestha, S.P., Lanford, B.L., Rummer, R.B., Dubois, M., 2005: Utilization and Cost of Log Production from Animal Logging Operations. International Journal of Forest Engineering 16(2): 167–180.

Wang, L., 1997: Assessment of animal skidding and ground machine skidding under mountain condition. Journal of Forest Engineering 8(2): 57–64.

Spinelli, R., Lombardini, C., Magagnotti, N., 2013: Salvaging windthrown trees with animal and machine systems in protected areas. Ecological Engineering 53: 61–67.

Zimmermann, M., 1994: Energieaspekte des Pferdeeinsatzes. Das Zugpferd 2–3: 22–25.

Authors‘ address:

Received: July 16, 2015 Accepted: January 27, 2016

286

Ákos Malatinszky, PhD.* e-mail: malatinszky.akos@mkk.szie.hu Csilla Ficsor e-mail: csilla.ficsor@gmail.com Szent István University Faculty of Agricultural and Environmental Sciences Páter Károly utca 1 H-2100 Gödöllő HUNGARY * Corresponding author Croat. j. for. eng. 37(2016)2

Original scientific paper

Modeling Harvest Forest Residue Collection for Bioenergy Production Rene Zamora-Cristales, John Sessions Abstract Forest harvest residues are often available at roadside landings as a byproduct of the log manufacturing process. This residue is usually available for renewable energy production if desired, however there is a significant amount of residues that do not reach the landing during the harvesting process and could potentially increase the supply of forest biomass from each harvest unit. The proportion of recoverable residues depends on their collection costs, which are a function of the distance from roadside landing, terrain conditions, and collection method. In this study, a forest residue collection model using forwarders and excavator-base loaders was developed to estimate the potential cost of biomass extraction from the forest to roadside landings. At the operational level, the model calculates the potential forwarder paths to estimate the cost depending on slope, machine arrangement and distance. For the analyzed harvest unit, the use of the excavator-base loader working alone is the most cost effective system for distances of less than 50 m and two forwarders and one excavator-base loader is the most cost effective system for distances beyond 50 m. The optimal solution collection costs ranged from USD 7.2 to 27.5 per oven-dry tonne for a range of distance between 15 and 350 m. The use of one operator to trade positions as forwarder operator and excavator-base loader operator resulted in lower productivity and higher cost compared to the use of a separate operator for each machine. Keywords: biomass, forwarder, simulation, spatial analysis

1. Introduction Forest harvest residues are a potential source of renewable energy to generate electricity and produce liquid biofuels (NARA 2011, SENECA 2015). In whole tree logging, forest harvest residues are often available at roadside landings as a byproduct of the log manufacturing process. However, there are a significant amount of residues that do not reach the landing during logging (breakage during dragging) and could potentially increase the supply of residues from each harvest unit. In cut-to-length operations, forest residues such as tops and limbs are usually left dispersed on the ground during the delimbing and log bucking process. Forest residues could be chipped in place by mobile chippers or collected and moved to roadside for chipping or grinding at roadside points. Or, it could be bundled and bundles forwarded to roadside. In western North America neither the bundler nor the mobile chipper have been economical (Zamora-CrisCroat. j. for. eng. 37(2016)2

tales et al. 2015). Instead, the gathering of residues involves excavator-base loaders or forwarders to collect and transport the residue to roadside locations for processing. Collection costs are a function of the distance from the collection point to the roadside landing, terrain conditions, and system productivity. The farther the collection point is, the higher the biomass cost will be. Equipment balancing is important in some system configurations, where loaders and forwarders interact between different tasks that can affect the productivity of the whole collection system. Terrain conditions affect maneuverability and may prevent the forwarder from using the shortest route to reach the landing due to ridges and severe slope changes. The objective of this study was to develop a spatial simulation model to estimate the collection cost of harvest residues for different forwarder-loader configurations at the operational level. Identifying the collection cost of forest residues will help to improve biomass supply cost estimation. The scope of this paper considers har-

287

R. Zamora-Cristales and J. Sessions

Modeling Harvest Forest Residue Collection for Bioenergy Production (287â&#x20AC;&#x201C;296)

vest units with slope gradients less than 30%. The model calculates the cost of collection from different locations in the forest to the most cost effective landing. The problem to be solved is to accurately estimate the cost of collection given the distance, terrain conditions and machine productivity.

1.1 Relevant literature Previous studies concentrate their analyses in the processing (grinding or chipping) and transportation, and very few involve the collection from the forest to the landing. Anderson et al. (2013) discussed the use of end-dump trucks to transport the material to a centralized yard; however, collection from the forest site to roadside was not discussed. In Canada, Yemshanov et al. (2014) found that forwarding biomass from the forest to the landing is inefficient given the low bulk density of the harvest residues, but the effect of cost at different distances from the landing was not discussed. Grushecky et al. (2007) evaluated extraction costs in southern West Virginia, using grapple skidders. The authors identified the extraction cost versus average extraction distance; however, the study only considered straight line average skidding distance thus not considering the effect of terrain conditions. Others have used digital terrain models to plan skid trails (Tucek 1999, Bohle 2005) and evaluate optimal landing location (Contreras and Chung 2007). RĂ¸rstad et al. (2010) developed an engineering model for estimating forest harvest residue cost using a forwarder with selfloader. Lacking actual data on harvest residues, they adjusted data from Laitila et al. (2007). Their distance from stand to landing was estimated in SGIS, but was done at a regional level. Spinelli et al. (2014) develop a simulation model to compare productivity and cost of chipping at the yarding site (not accessible for large trucks) and chipping at a roadside landing using a forwarder to transport the unprocessed residue from the yarding site to the roadside landing. Forwarding residues to the landing resulted in a more expensive operation having the forwarding distance as the most important factor affecting the cost. The model proposed here, on a harvest unit basis, is based on field collected data, and considers system configurations not previously documented in the literature. A GIS-based raster system is used to limit the travel of the forwarder to gentle terrain when possible or at least to minimize the travel on steep slope zones although this may require traveling through a longer trail. It assumes that rubber-tired vehicles are permitted on the forest harvest site. Beginning in the 1960s, some landowners in western Oregon and western Washington stopped using rubber-tired skidders on

288

compactable, high site forest soils, preferring cable logging to protect soil productivity (Fisher 1999). In the early 1970s, excavator-base loaders were introduced for yarding logs and trees to roadside. The excavatorbase loader (shovel) equipped with wide tracks (low ground pressure) and high clearance makes one pass across the harvest site limiting soil disturbance. The high productivity of this one man system for yarding and loading led to its quick adoption throughout the region. Concern over using rubber-tired vehicles lingers; some forest managers remain concerned about potential post-harvest site damage from high tonnage rubber-tired forwarders collecting low value harvest residues after the forest site was protected using the one pass shovel logging method.

1.2 Collection systems The collection of forest biomass requires concentrating the scattered residues at collection points. In the Pacific Northwest, USA this is usually performed by an excavator-base loader. If the residues are close to the landing (usually less than 50 m), they can be collected using an excavator-base loader that swings the residues directly to the landing. At longer distances, the use of alternative and more productive equipment, such as forwarders, are used to access the material and transport it to the landing. Forwarders are equipped with a self-loading grapple crane that allows the forwarder to operate independent of a dedicated loading machine. The conventional forwarder was designed for loading logs, not forest residues. Using the selfloading system for forest residues can be challenging due to the limited visibility of the operator while putting the material in the bunk and the limited reach and capacity of the loading boom. In biomass recovery operations in the Pacific Northwest, USA, forwarders are sometimes loaded using excavator-base loaders equipped with fully rotating grapples that facilitate the handling of residues. Once the forwarder is fully loaded, it returns to the landing and unloads. Equipment balancing is important to keep all equipment elements producing to optimal capacity. The farther the collection point is from the landing the more expensive it is to collect the residue because the forwarder has to spend more time traveling, thus decreasing the forwarder productivity (Fig. 1). The use of two forwarders per loader help to minimize the impact of the distance on forwarding productivity, however traffic along the trails can cause machine interference. Once the material is at the landing, it is commonly processed using grinding to increase the bulk density of the material and facilitate transport and further handling. Other equipment such as off-highway dump trucks with skidder tires could be used to move the residues; Croat. j. for. eng. 37(2016)2

Modeling Harvest Forest Residue Collection for Bioenergy Production (287–296)

R. Zamora-Cristales and J. Sessions

Fig. 1 Description of the forest residue collection problem however, the use of this equipment on forest soils could cause more soil compaction compared to the multi-axle forwarders using wheel tracks. Thus, at least five systems can be used alone or in combination: Þ System 1 – Excavator-base loader, working alone Þ System 2 – Forwarder self-loading Þ System 3 – Forwarder loaded by excavator-base loader Þ System 4 – Two forwarders loaded by one excavator-base loader Þ System 5 – As above, but the loader is manned by the forwarder operators, in turn The time and productivity of each system, s, can be defined by: (1) Ts = as + bsx

Ps =

60 L Ts s

(2)

Cs =

Costs Ps

(3)

Where: Ts time per trip in minutes, as the fixed component of the trip not related to distance bs time per ton-km x travel distance in km Ps productivity in tonnes per hour Ls load per trip in tonnes Costs cost per unit time (hours) Cs cost per unit volume in dollars per tonne. Croat. j. for. eng. 37(2016)2

The objective is to find the system or combination of systems that minimizes total collection cost including mobilization costs.

2. Material and methods The analysis for modeling forest residue collection starts at the forest unit by identifying the boundaries, potential spatial location of residues and candidate landings. In this model, a grid-type approach is used to cover the entire unit and estimate the cost of each potential residue location to the roadside landings. A point every 30 m is generated and stored to represent the location of the forest residue (Fig. 2).

Fig. 2 Spatial description of the residue collection problem from different locations within the harvest unit to potential roadside landings

289

R. Zamora-Cristales and J. Sessions

Modeling Harvest Forest Residue Collection for Bioenergy Production (287â&#x20AC;&#x201C;296)

Landing locations are typically selected by their accessibility for trucks and available turnarounds. In this model, we selected all the logging landings and loading points as potential candidates for roadside residue concentration. The main criteria for establishing a residue landing is that it has to provide good access for chip vans and enough space to place residue and processing equipment. Chip vans compared to log trucks have several limitations depending on road characteristics such as trailer low ground clearance and less traction in the rear axles when traveling empty among others (Sessions et al. 2010, Zamora-Cristales 2013). Once the unit boundaries, potential landings and concentration points were defined, we developed a computerized GIS model to design the forwarder trails given the terrain conditions.

2.1 Computerized identification of forwarder trails Forwarder trails need to be identified to accurately estimate the forwarding cost. Assuming an average forwarding distance for the entire harvest unit could lead to underestimating or overestimating the cost depending on the assumed distribution of residues among candidate roadside landings. Assuming a straight line distance from the collection point to the landing could also lead to misleading results, given that in actual conditions operators tend to avoid difficult terrain or abrupt edges when traveling in the forest, thus traveling longer paths. To create the computerized forwarder trails, a 10 m digital elevation model (DEM) was used to derive a slope raster image to create the feasible paths. All the spatial data processing was made using ArcMap 10.0 (ESRI 2012). The slope raster image allowed us to analyze potential areas that will be difficult for the forwarder to travel on. The slope raster image was then reclassified to clearly separate areas with slopes greater than 30%. Once the slope raster images were reclassified, we created a cost distance raster image to estimate the cost of each pixel to each of the potential landings. Then, a cost path raster image was created to calculate the least cost path from each harvest residue location to the most costeffective landing. Once the least cost paths were created, we converted them into a vector polyline for further processing, using the network analyst extension to create the optimal forwarder paths. Finally, a kriging technique (Oliver 1990) was used to create a continuous cost map that clearly shows the cost of collecting the residues at different distances.

simple time study determined the production coefficients as there was no significant effect of equipment interaction. However, systems 3â&#x20AC;&#x201C;5 depend upon equipment interactions (Fig. 3). A simulation model was created in a Rockwell Arena software environment (ROCKWELL 2015). System 3 is represented by one forwarder loaded by the excavator-base loader. The simulation model in this system starts when one of the forwarders is moving unloaded to the forest residue collection point. At the collection location, the excavator-base loader is simultaneously concentrating material for the forwarders. As the forwarder arrives at the collection point, the excavator-base loader proceeds to load it as long as there is enough piled material. If not enough material is available for the forwarder to be loaded, the forwarder has to wait. After the forwarder is loaded, it travels back to the landing and unloads the residue. System 4, two forwarders loaded by one excavator-base loader is similar to system 3, except that only one forwarder is allowed to travel along the trail at a time, thus minimizing interference along the trail. System 5 includes the use of two forwarders loaded by one excavator-base loader. This

2.2 Simulation model For system 1, the excavator-base loader worked alone; in system 2, one forwarder worked alone. A

290

Fig. 3 General description of the simulation model Croat. j. for. eng. 37(2016)2

Modeling Harvest Forest Residue Collection for Bioenergy Production (287–296)

system is different from the other systems in the sense that the same operator operates the forwarder and the loader. This is similar to sharing a log loader among truck drivers. This system is only feasible if the material is already concentrated so the excavator-base loader is only used for loading the forwarder. As the forwarder reaches the collection point, the operator moves to the excavator-base loader and proceeds with loading the forwarder. For the purpose of determining production rates, a time study was undertaken to calculate average time per swing of the excavator grapple, time per swing of the forwarder grapple, volume in the excavator grapple, volume in the forwarder grapple, load on the forwarder, and speed of the forwarder. During the time study the number of grapple loads per forwarder load of each type of grapple was recorded and each individual forwarder load was put into an end-dump truck and weighed at the mill yard.

2.3 Study site Source data for the simulation model was collected from a residue collection operation. We performed a time and motion study on a harvest unit located 24.5 km southwest of Springfield, Oregon, USA (43°53’59”N, 122°47’9”W). Douglas-fir (Pseudotsuga menziesii) forest residues were dispersed over a 16.7 ha unit following whole tree harvest by shovel logging. Residue consisted of branches and tops with an average diameter ranging from 5 to 15 cm (m=5.96 cm and s=2.80 cm). Average piece length was 1.2 m. A Caterpillar 564 forwarder with a maximum load capacity of 13,608 kg was used for the test. A Kobelco SK290 LC hydraulic excavatorbase loader was used to concentrate the residue at the loading points and load the forwarder except for the system where the forwarder self-loaded. A GPS Visiontac VGPS-900 was placed in the forwarder to track the movements of the machine when collecting the residues. Each forwarder load was then placed in a 90 m3 end-dump truck and transported to a local mill, where the material was weighed. A total of 180 wet tonnes were collected and transported as part of the study. Thirty forwarder cycles were recorded and data processed. Samples for moisture content were taken from each load and transported to the laboratory for moisture content estimation using standard ASTM D4442 for direct moisture content measurement of wood and wood-based materials.

2.4 Cost estimation The forwarder and the excavator-base loader costs were estimated by adapting Brinker’s (2002) machine rate method and validated with the actual contractor Croat. j. for. eng. 37(2016)2

R. Zamora-Cristales and J. Sessions

costs. All the costs were expressed in USD 2015 dollars. Hourly costs include depreciation, insurance/taxes, and interest, labor, repair and maintenance, fuel and lubricants and profit and risk (10% of total hourly costs). Fuel cost was estimated to be $0.8 l-1. If the machine is operating (forwarding/loading/piling) then the cost included all previously listed items. If the machine is idling, (e.g. forwarder waiting for the loader) then only interest, insurance/taxes, labor cost and profit and risk are included. Profit and risk is included in the idling time to recognize the opportunity cost of being not productive in addition to interest on average investment. In this study, depreciation due to use is considered negligible when the machine is not operating since the parts are not wearing out. Depreciation due to obsolescence is considered low for relatively new forest machinery in the Pacific Northwest region and depends more on the hours of use rather than the year of manufacture (Personal Communication, Larry Cumming, PetersonPacific Industries, December 9, 2016). Our accounting approach offers advantages over the scheduled/productive hour approach when dynamic equipment balancing decisions are being made. Mobilization costs were based on a fixed rent rate of $100 per hour for a lowboy truck. It was assumed that one machine is transported per truck and it takes 8 hours to complete delivery of the machine (rates in the region are calculated from the time the truck leaves the yard until it returns).

2.5 Supply economics We estimated the impact of collection cost on the amount of residue that could be supplied. This was performed by integrating the collection costs with the processing and truck transportation cost. Transportation costs were calculated for a truck equipped with a drop-center (possum-belly) trailer with a capacity of 100 m3. The truck-trailer combination has a maximum allowable legal weight of 40,823 kg. It was assumed residue is evenly distributed at each collection point defined in the GIS grid (30 meter) with a biomass volume of 42.43 dry tonnes per ha in 16.7 ha, giving a total of 707.6 dry tonnes of residues. With the transportation cost, we ran a sensitivity analysis to evaluate the amount of residue that could be economically feasible depending on the distance from the harvest unit to the bioenergy conversion facility. As the distance from the forest to the bioenergy facility increases the transportation cost increases, thus limiting the amount of harvest residue that could economically be recovered. We set four potential prices, $50, $60, $70 and $80 dollars per oven-dry tonne in order to estimate the maximum collection cost to break even. Grinding cost and productivity were extracted from Zamora-Cristales (2013) for a Peterson 4710B horizontal grinder.

291

R. Zamora-Cristales and J. Sessions

Modeling Harvest Forest Residue Collection for Bioenergy Production (287–296)

Table 1 Forwarder and excavator-base loader hourly costs, USD Operating costs Item Purchase price, $

Waiting cost

Forwarder CAT 564

Loader Kobelco SK290 LC

Forwarder CAT 564

Excavator Loader Kobelco SK290 LC

361,160

280,000

–

Ownership costs Depreciation cost, $ h-1 -1

Annual interest, $ h

-1

Annual insurance and taxes, $ h

38.52

29.87

–

16.37

12.69

16.37

12.69

12.04

9.33

12.04

9.33

Annual productive machine hours, h

1500

Hourly ownership cost, $ h-1

66.93

51.89

28.41

22.03

Variable costs -1

Labor, $ h

33.75

Repair and maintenance, $ h-1

23.11

17.92

–

16.41

16.32

–

73.28

67.99

33.75

14.02

11.99

14.02

11.99

154.23

131.87

76.18

67.77

-1

Fuel and lubricants cost, $ h -1

Hourly variable costs, $ h -1

Profit and risk, $ h (10% of hourly variable and ownership cost) Total cost, $ h-1

3. Results and discussion 3.1 System costs Moisture content of the samples was estimated in 44% (wet basis). Hourly operating and waiting costs for the forwarder and excavator-base loader are shown in Table 1. Labor cost was included in the excavatorbase loader cost although in the case of the one operator system simulation only the cost of one operator was counted. Results from the time and motion study are shown in Table 2. If the forwarder is self-loading, then it is difficult to completely fill the bunk. Additionally, it took more time to load the forwarder due to the reduction in visibility and maneuverability. Loading the forwarder with the excavator-base loader resulted in significant decreases in time and increased load volume (Fig. 4), however this affected the time for the loader to concentrate residue at the forwarder collection points. The unloading time was consistent with the load size and was considerably faster than self-loading by the forwarder because the material is partially pushed out of the bunks instead of grabbed and unloaded. The excavator-base loader spent 12.6 (d=0.4) minutes in average to pile 7.6 t of wet residue at the concentration points. During this time, the excavator-base loader spent 0.6 minutes per swing, with an average grapple load size of 0.36 t of wet residue.

292

For the harvest unit analyzed, simulation results suggest that the use of two forwarders and one loader could be the most productive system (Fig. 5) at longer distances. The productivity of this system is maintained until it reaches a distance from the landing of 255 m after which the excavator-base loader wait time is increasing. Using the same operator for both the forwarder and the loader will maintain productivity but it requires the operator to move between machines increasing the forwarder waiting time. The self-loading system appears to be the least productive of the forwarder systems due to the longer loading time

Table 2 Time and motion study results for forwarder productivity in wet tonnes (t) from 30 recorded cycles Item

Mean

Forwarder self-loading, min load-1

8.9

2.5

Forwarder self-unloading, min load-1

5.1

2.8

Excavator loading forwarder, min

5.2

1.3

Forwarder self-unloading excavator loaded, min

6.9

1.3

Travel loaded speed, km h-1

3.0

1.0

Travel unloaded speed, km h

4.2

0.8

Forwarder load, excavator loaded, t

7.6

1.2

Forwarder load size self-loaded, t

4.8

0.2

-1

Croat. j. for. eng. 37(2016)2

Modeling Harvest Forest Residue Collection for Bioenergy Production (287â&#x20AC;&#x201C;296)

R. Zamora-Cristales and J. Sessions

Fig. 4 a) Forwarder being loaded by the excavator-base loader; b) Forwarder traveling to the landing compared to the excavator-base loading system and the reduced payload due to difficult visibility and grapple maneuverability when loading the forwarder. The most cost effective option for distances less than 50 m from the roadside landing to the collection point is the use of the excavator-base loader working alone. Between 50 and 100 m, the use of one forwarder loaded by the excavator-base loader is the most cost effective system. Beyond 100 m, the two forwarders loaded by a single excavator-base loader is the most cost efficient and its comparative advantage grows with distance (Fig. 6). Although the system that uses the same operator for both machines is highly productive, it has high-

Fig. 6 Collection cost in USD per oven dry tonne as a function of the distance from the roadside landing (mobilization costs are not considered)

Fig. 5 Productivity in oven dry tonnes per hour for each of the analyzed options Croat. j. for. eng. 37(2016)2

er cost because the residues would need to be pre-piled before forwarding operations can begin. The self-loading forwarder has the highest per unit cost due to the longer collection time and smaller load size (Table 2). Fig. 6 shows cost as a function of distance. If the average collection distance for the harvest unit was greater than 50 m and less than 70 m, then mobilization costs would probably determine if the excavator-base loader would be used alone or in combination with a single forwarder. This decision will depend upon the mobilization cost per unit volume that is a function of the amount of residual material available. In this example, we assumed a mobilization cost of $800 per machine ($100 h-1 of low-

293

R. Zamora-Cristales and J. Sessions

Modeling Harvest Forest Residue Collection for Bioenergy Production (287â&#x20AC;&#x201C;296)

Table 3 Collection cost for each system in the 16.4 ha harvest unit of study (707.6 dry tonnes) and the optimal solution considering a combination of systems 1, 3, and 4 (this includes mobilization cost)

marginal costs exceeded the marginal cost of the alternatives. Cost for harvesting the unit under study are shown in Table 3.

Cost, $ Cost, $ t-1

3.2 Application to the trail network

System 1: Excavator-base loader

42,994

60.8

System 2: Forwarder self-loading

For each potential residue spatial location a least cost path to landing was determined. The processing of the digital elevation model, residue and landing locations resulted in the optimal location of the forwarder trails (Fig. 7) according to a slope-weighted shortest path to the closest landing. The forwarder trails were designed to avoid traveling over abrupt changes in slope and steep areas (<30% in slope) by penalizing cost rasters on steeper slopes. The total length of forwarder trails was 8660 m, occupying about 15% of the harvested area. In Fig. 7a, costs were assigned using the results of Fig. 6 resulting in what we define as the optimal system cost. At shorter distances (less than 50 m), the excavator-base loader was used, at distances between 50 and 70 m, one forwarder and one excavator-base loader was used and for longer distances greater than 70 m, the two forwarder and one excavator-based loader system was used.

System

26,399

37.3

System 3: Forwarder loaded by excavator-base loader 17,613

24.9

System 4: Two forwarders loaded by one excavator base- loader

16,447

23.2

System 5: Two forwarders loaded by one excavator base-loader sharing operator

22,630

32.0

Optimal Solution, System 1<50 m; 50 m<System 3<70 m; System 4>70 m

16,180

22.9

boy cost, contracted for 8 hours). This gave a cost of 1.1, 2.3 and 3.4 dollars per oven dry tonne for one, two and three machines respectively. This cost assumes that 707.6 oven-dry tonnes are available and recoverable. In all cases, the excavator-base loader would be used to directly collect residues until at least the point where its

Fig. 7 Cost raster map for: a) optimal costs; b) one forwarder self-loading; c) one forwarder, one loader; d) two forwarders one loader (optimal costs combine the excavator-base loader working alone at short distances with the two forwarders and the excavator-base loader working together at longer distances)

294

Croat. j. for. eng. 37(2016)2

Modeling Harvest Forest Residue Collection for Bioenergy Production (287â&#x20AC;&#x201C;296)

Figs. 7b-d show the cost mapping if the self-loading forwarder (system 2), one forwarder and one excavator-base loader (system 3), and two forwarders and one excavator-base loader were used (system 4) without using the excavator-base loader working alone at the shorter distances. Average forwarding distance using this harvest unit was 156.4 m. On the other hand, using the straight line method, the average forwarding distance for the same unit would be 124.5 m. The straight line average forwarding distance is 20% less than the actual distance calculated using the raster method, thus underestimating the forwarding cost. As the collection cost varies over the harvest unit, it is possible that, depending on price and the transportation cost to the bioenergy facility, not all of the residues will be delivered to the landing, but may either be left piled or burned in place. Assuming no other forest management benefit to the landowner (for example, reduced disposal costs, added available planting space, reduced fire risk), the percentage of biomass that could be available as a function of the distance from the forest to the bioenergy facility is shown in Fig. 8. At distances longer than 60 km no residue could be economically recoverable at a gate price of $50 t-1. Similarly at $60 t-1, the maximum transportation distance is 100 km. This procedure can be adapted for different processing and other transportation configurations to evaluate potential biomass availability from an economical point of view and can include other forest management benefits to the owner such as avoided disposal costs, increased planting space, or reduced fire risk.

R. Zamora-Cristales and J. Sessions

4. Conclusions The utilization of forest residues offers an additional, but low value product from the forest. In order to provide economic value, the collection model must be well rationalized. A number of methods can be used to collect forest residues. Currently in the Pacific NW, USA, only residues close to the landing are utilized and those are primarily collected by an excavator-base loader working alone. We have demonstrated that a number of methods can be used to collect residues. For the conditions in our simulation, the excavator-base loader is the least expensive option within 50 m, between 50 and 70 m a combination of one forwarder and one excavator-base loader is the most cost effective option and beyond 100 m, a combination of two forwarders loaded by an excavator-base loader is the least expensive option with collection costs increasing modestly up to 240 m. However, if the total forwarding distance is less than 100 m, it is possible that excavator-base loader working alone may still be the lowest total cost option due to mobilization costs to bring in a forwarder. The mobilization cost to move the machinery (forwarders and loader) to the site is a fixed cost, thus it is important to have a significant amount of biomass available at the unit to justify the transport and placement of the machinery, especially for systems that require the use of two forwarders. The excavator-base loader would always be used to forward the closest material regardless of the system used at longer distances. The model developed in this research could be adapted and used in other conditions. The only required input for the GIS trail identification is the use of the digital elevation model. Additionally, the model can be extended by adding other land features such as streams. In this simulation the only physical barrier for the forwarder was ground slope. It was assumed that the use of forwarders would be permitted. In this example, forwarder trails covered 15% of the area. Depending on soil considerations, forwarder trails could be reduced by increased piling by the excavator-base loaders. This could be represented by larger pixels. An alternative analytical modeling approach could be mathematical programming that includes soil compaction and mitigation methods and permits direct control of the area in forwarder trails.

Fig. 8 Non-roadside biomass available per oven-dry tonne at different potential prices at the bioenergy facility gate Croat. j. for. eng. 37(2016)2

Regardless of the collection system, there is a price point at which some residues in a harvest unit will not be recovered suggesting that there is a tradeoff between off-road collection distance and on-road transportation cost. Including other forest management benefits such as avoided disposal costs will increase economic collection distances.

295

R. Zamora-Cristales and J. Sessions

Modeling Harvest Forest Residue Collection for Bioenergy Production (287–296)

Acknowledgements This research was funded by the Northwest Renewables Alliance (NARA). NARA is supported by the Agriculture and Food Research Initiative Competitive Grant 2011-68005-30416 from the US Department of Agriculture, National Institute of Food and Agriculture. We also want to thank Lane Forest Products, and TOPS Weyerhaeuser for giving us access to their operations and operational assistance.

5. References Anderson, N., Chung, W., Loeffler, D., Jones, J.G., 2012: A productivity and cost comparison of two systems for producing biomass fuel from roadside forest treatment residues. Forest Products Journal 62(3): 222–233. ASTM D4442-15, 2015: Standard Test Methods for Direct Moisture Content Measurement of Wood and Wood-Based Materials, ASTM International, West Conshohocken, PA. Bohle, K., 2005: Forcasting harvest production rates using GIS: a case study. Master of Forestry paper, Dept. of Forest Engineering, Oregon St. Univ., Corvallis, 72 p. Brinker, R., Kinard, J., Rummer, B., Lanford, B., 2002: Machine rates for selected forest harvesting machines. Circular 296 (Revised). Alabama Agricultural Experimental Station, Auburn, 31 p. Contreras, M., Chung., W., 2007: A computer approach to finding an optimal log landing location and analyzing influencing factors for ground-based timber harvesting. Canadian Journal of Forest Research 37(2): 276–292. ESRI, 2012: ArcGIS 10. Available from www.esri.com/software/arcgis/arcgis10 [accessed on October 3, 2012]. Fisher, J., 1999: Shovel logging: cost effective systems gains ground. Proceedings of International Mountain Logging and 10th Pacific Northwest Skyline Symposium; March 28 – April 1; Corvallis, Oregon: 61–67. Grushecky, S., Wang, J., McGill, D., 2006: Influence of site characteristics and costs of extraction and trucking on logging residue utilization in southern West Virginia. Forest Products Journal 57(7–8): 63–67. Hakkila, P., 1989: Utilization of Residual Forest Biomass. Springer-Verlag, Berlin, 568 p.

Laitila, J., Asikainen, A., Hotari, S., 2005: Residue recovery and site preparation in a single operation in regeneration areas. Biomass and Bioenergy 28(2): 161–169. NARA 2011: Nortwest Advanced Renewables Alliance. NARA. https://www.nararenewables.org/ Oliver, M.A., 1990: Kriging: A Method of Interpolation for Geographical Information Systems. International Journal of Geographic Information Systems 4(3): 313–332. ROCKWELL, 2015: Rockwell Arena simulation software available from https://www.arenasimulation.com/ [last accessed on May 21, 2015]. Rørstad, P.K., Trømborg, E., Bergseng, E., Solberg, B., 2010: Combining GIS and Forest Modelling in Estimating Regional Supply of Harvest Residues in Norway. Silva Fennica 44(3): 435–451. SENECA, 2010: Biomass cogeneration facility. http://senecasawmill.com/seneca-sustainable-energy/biomass/ [last accessed May 21, 2015]. Sessions, J., Wimer, J., Costales, F., Wing, M., 2010: Engineering considerations in road assessment for biomass operations in steep terrain. Western Journal of Applied Forestry 25(5): 144–154. Spinelli, R., Di Gironimo, G., Esposito, G., Magagnotti, N., 2014: Alternative supply chains for logging residues under ac cess constraints. Scadinavian Journal of Forest Research 29(3): 266–274. http://dx.doi.org/10.1080/02827581.2014.896939 Tucek, J., 1999: Algorithms for skidding distance modeling on a raster digital terrain model. International. Journal of Forest Engineering 10(1): 67–79. Yemshanov, D., McKenney, D.W., Fraleigh, S., McConkey, B., Huffman, T., Smith, S., 2014: Cost estimates of post harvest forest biomass supply for Canada. Biomass and Bioenergy 69: 80–94. Zamora-Cristales, R., Sessions, J., Boston, K., Murphy, G., 2015: Economic Optimization of Forest Biomass Processing and Transport in the Pacific Northwest. Forest Science 61(2): 220–234. http://dx.doi.org/10.5849/forsci.13-158 Zamora-Cristales, R., Sessions, J., Murphy, G., Boston, K., 2013: Economic impact of truck-machine interference in forest biomass recovery operations on steep terrain. Forest Products Journal 63(5–6):162–173. http://dx.doi.org/10.13073/FPJD-13-00031

Authors’ address:

Received: May 24, 2015 Accepted: December 18, 2015

296

Rene Zamora-Cristales, PhD. * e-mail: rene.zamora@oregonstate.edu Prof. John Sessions, PhD. e-mail: john.sessions@oregonstate.edu Oregon State University Department of Forest Engineering, Resources and Management Corvallis, Oregon 97331 USA * Corresponding author Croat. j. for. eng. 37(2016)2

Original scientific paper

The Impact of Log Moisture Content on Chip Size Distribution When Processing Eucalyptus Pulpwood Jaco-Pierre van der Merwe, Pierre Ackerman, Reino Pulkki, Dirk Längin Chip moisture content and especially its uniformity impact kraft pulping. However, the effect of pulp log moisture content on chip quality during chip production is not well known. Chip size distribution is important in kraft pulping as it impacts chemical use, pulp quality and recovery. This study investigated the influence of two pulp log drying periods (1 and 2 weeks) on chip moisture content and chip size distribution when chipping eucalypt pulp logs. In addition, the effect of three log classes (base, middle and top logs) on chip moisture content and chip size distribution were also analysed. Within the respective log classes, moisture content of chips produced from logs dried for 2 weeks was 5.5% to 13.2% lower than moisture content of chips produced from logs dried for 1 week. Chip moisture content also decreased with decreasing log size for both log drying periods. One week dried logs produced chips with 1.0% less over-thick chips than 2 week dried logs (1.5% versus 2.5%). One week dried logs also produced chips with 4.2% to 7.2% less accepts than chips produced from 2 week dried logs within respective log classes. Across both drying periods, over-thick chip production increased with decreasing log size, while the amount of accepts produced decreased with decreasing log size. Logs dried for 2 weeks produced chips with significantly less under-sized chips than logs dried for 1 week. Two week dried logs produced chips with 4.4% to 7.7% less pins and 0.7% to 1.0% less fines than 1 week dried logs within respective log classes. For both log drying periods, the amount of under-sized chips produced increased with decreasing log size. Keywords: Pulp logs, eucalyptus, moisture content, chip size

1. Introduction Commercial forestry is practiced on 1.273 million ha or 1.1% of South Africa’s total surface area. South African commercial forests serve various wood based industries, of which the pulp and paper industry is the largest (FES 2011). During 2011, the industry produced a total 18.5 million m3 of roundwood, of which 12.6 million m3 was harvested for pulp and paper production. Revenue for these pulp log sales amounted to EUR 279 million and pulp product sales from primary processing plants was EUR 799 million (FSA 2013). Fast growing eucalypt hardwood species supply 83% of wood resources used for pulp and paper manufacturing (FES 2011). Pulp logs are purchased and harvested on a per green tonne basis. Therefore, the moisture content of pulp logs at the time of purchase has a significant impact on the pulp log price as it influences wood mass. Croat. j. for. eng. 37(2016)2

Previous studies have investigated the impact of log moisture on chip size distribution, when producing chips for the bioenergy market (Spinelli et al. 2011, Mihelič et al. 2015). However, little is known on how log moisture content or log drying period length influence wood chip quality (i.e. thickness, size distribution and chip fracturing) during chip production, when producing chips from eucalypt pulp logs in a plantation setting. Chip quality is important as it influences pulp recovery and quality in kraft pulping (True 2006, Macleod 2007, Gulsöy 2012). Pulp log (from here on referred to as log) moisture content (MC) influences mechanical wood properties such as wood hardness, strength and processing ability (Niedźwiecki 2011). Physical log properties such as log size (length and diameter), degree of debarking, log surface damage and wood density influence the rate of moisture loss. Logs with bark experience slower moisture loss as opposed to debarked logs or tree sections

297

J.-P. van der Merwe et al.

The Impact of Log Moisture Content on Chip Size Distribution When Processing ... (297–307)

(Connel 2003, Röser et al. 2011). Sapwood is more exposed to climatic elements after debarking and, therefore, has higher moisture loss rates when compared to heartwood (Defo and Brunette 2007, Färlin 2008). Freshly harvested logs lose moisture while in storage either in the plantation, at roadside or at the mill (Röser et al. 2011). Various storage practices such as stack geometry, orientation to sun and wind, locality and individual log exposure will either accelerate or inhibit the rate of log drying (Persson et al. 2002, Defo and Brunette 2007, Färlin 2008, Gjerdrum and Salin 2009, Phanphanich and Mani 2009, Röser et al. 2011, Eisenbies et al. 2014, Erber et al. 2014, Routa et al. 2015). Smaller log stacks dry quicker due to higher log surface exposure and log stacks sheltered from the wind and sun will have slower drying rates (Persson et al. 2002, Defo and Brunette 2007, Färlin 2008). Seasonal variations in temperature, precipitation, relative humidity, wind speed and wind direction will influence log drying rates (Gjerdrum and Salin 2009, Defo and Brunette 2007, Röser et al. 2011). Log moisture loss is greater at higher ambient temperatures, low atmospheric humidity and/or when logs are exposed to a prevailing wind (Persson et al. 2002, Connel 2003, Defo and Brunette 2007, Gjerdrum and Salin 2009, Röser et al. 2011). Precipitation replenishes log moisture and will reduce moisture loss (Defo and Brunette 2007, Gjerdrum and Salin 2009, Röser et al. 2011). Processing efficiencies and pulp yield can be directly related to chip quality, as chip quality plays an important role in pulp recovery (MacLeod 2007). Chip moisture content, and especially uniformity, have a major impact on the kraft pulping process (Pulkki 1991). The quality of chips derived from chippers is expressed in terms of chips size distribution: i.e., the percentage of accepted chips (prime and small-size chips), over-size chips, over-thick chips, pins and fines, and whether the chips contain any impurities in the form of bark, knots and rot. In kraft pulping, chemical penetration times vary in relation to chip size, thickness and uniformity. Uniform chips lead to more uniform pulping conditions and higher pulp recovery (Pulkki 1991, Twaddle and Watson 1992a, Twaddle and Watson 1992b, Twaddle and Watson 1992c, MacLeod et al. 1995, Uhmeier 1995, Hartler 1996, Uhmeier and Persson 1997, Broderick et al. 1998, Tessier et al. 1999, Bjurulf 2005, Ding et al. 2005, Bjurulf 2006, True 2006, MacLeod 2007, Balakrishnan 2008, Färlin 2008, Santos et al. 2008, Hellström 2010, Walton et al. 2010, Mafia et al. 2012, Patt et al. 2012). Timber freshness is expressed in terms of the MC in the wood itself and will influence chip size distributions during chip production (Qian et al. 1994, Hellström

298

2008, Hellström 2010, Isokangas 2010, Niedźwiecki 2011). An increase in chip thickness during chip production has been observed, with decreasing log MC. When log MC is very high or low, chip size and uniformity will be negatively affected. Logs with very low MC produce greater quantities of undesirable small chips (fines and pins) and large chips (over-sized and overthick chips) due to decreasing wood plasticity. At higher MC, the wood is softer and greater quantities of pins and fines are produced (Pulkki 1991, Uhmeier and Persson 1997, Persson et al. 2002, Bjurulf 2006, Watson and Stevenson 2007, Färlin 2008, Hellström 2008, Niedźwiecki 2011, Mihelič et al. 2015). No studies could be found to indicate optimal MC for chipping as wood processing is a function of the interactions between wood density and MC (Niedźwiecki 2011). Watson and Stevenson (2007) investigated the influence of seasonal variations in log MC of softwood and hardwood species on chip size and uniformity and their effect on kraft pulping. The authors found that over-sized chip production increased with decreasing seasonal log MC. While as MC increased, under-sized chips production increased. No literature was found as to how log MC influence size and uniformity of eucalypt chips, nor have critical log moisture values been associated with eucalypt chip quality. The objective of the study was to determine the influence of two log drying periods and three log size classes on the quality of chips produced in relation to chip size distribution, including any fracturing, chip MC and chip uniformity.

2. Materials and methods 2.1 Site selection, treatments and harvesting The study was done near Kwambonambi in the Northern KwaZulu-Natal forestry region of South Africa. The coastal region is subject to sub-tropical climates, with mean annual temperature and precipitaTable 1 Study site and tree details Species

E. grandis x urophylla

Age, years

Establishment spacing, m

3x2.5

26.20 3

-1

MAI (6 years), m ha yr

31.4

DBH, cm

15–20

Slope, %

Croat. j. for. eng. 37(2016)2

The Impact of Log Moisture Content on Chip Size Distribution When Processing ... (297–307)

tion of 22˚C and 1196 mm, respectively (Dovey 2012). An even-aged Eucalyptus grandis×urophylla cloned compartment of relatively uniform tree size was selected for the study (Table 1). The trees were harvested during spring, September 2012, when eucalyptus sapflow has been observed to be high in the Kwambonambi area (Dye et al. 2004). Size distribution and MC of chips produced from logs dried over 2 drying periods were compared. This includes drying period 1 – as fresh as possible (in this case 1 week drying); and period 2 – two weeks of drying before chipping was initiated. In addition, a distinction was made between 3 log size classes and it was taken into account how class related to chip size distribution and uniformity with log drying period. Three logs were removed from every tree (base, middle and top logs). The top log was the third log up the tree, but was not necessarily the last possible log available from any specific tree. For this study, 120 trees were felled in harvester settings with a single grip harvester with up to 5 feed roller passes along the log surface. A harvester setting comprised of 10 trees (5 rows wide and 2 rows deep), and 60 trees were allocated to each of the log drying treatment. Log drying periods were randomly assigned within the experimental design. Trees within each setting were colour coded according to log drying period and tree position. Each tree was also sequentially numbered. From each of the 120 tree, three 5.5 m logs were processed: one from the base, one from the middle and one from the top of the tree. The study produced 360 individual logs. There were equal numbers of logs for each of the log classes (base, middle and top logs). Trees with growth deformities, such as double leaders and butt sweep, within the experimental layout were excluded and formed part of the buffer zones to maintain design continuity. A SP Maskiner 591LX harvesting head mounted on a tracked Hitachi IS200 excavator base was used for the study. The feed roller pressures were pre-calibrated for tree size (DBH) and the particular bark characteristics (thickness and adherence to the stem) to minimise potential log surface damage induced by the feed roller before the study commenced. A Timberpro TF840-B forwarder extracted and loaded the log assortments directly onto a timber truck. The load was securely covered with a tarpaulin to limit moisture loss during transport to the chipping facility. Logs were chipped in a Bandit 250 XP mobile disc chipper. Chipper maintenance was done by a chipper technician prior to chipping. Chipper maintenance Croat. j. for. eng. 37(2016)2

J.-P. van der Merwe et al.

included knife change and anvil clearance adjustments. The chipper knife angles were fixed at 45˚. Logs were separated and stacked according to the drying period which they would be subjected to. When the logs reached the predetermined drying period, chipping was initiated. The logs were manually fed into the chipper to avoid potential grapple induced log surface damage from the mechanical loader. The chips were ejected from the chipper spout by means of the standard blower into an industrial tumbler. The tumbler container opening was covered by plastic bags, to prevent any chips from escaping the container. Chips produced from individual logs were mixed thoroughly in the tumbler for one minute before a 12l sample was extracted. Samples were immediately placed in plastic bags and sealed to avoid further moisture loss. Each log tag information was copied onto the bag containing the chips produced from it. The remaining chips were discarded. The green mass of each sample was recorded before being repacked into brown paper bags to facilitate moisture loss while in storage. Individual log information was replicated onto the paper bag. Samples were stored off the ground for 1 month to allow for air drying. Subsequently, the chip samples were screened for 5 min according to SCAN-CM 40:94 standards into 5 chip size classes (over-sized, over-thick, accepted, pins and fines) using a mechanical chip size screener (SCAN-Test 1994). Each of the 1800 individual fractioned chip class sub-samples (5 chip class sub-samples per chip sample), were marked for identification and bagged separately. Chip class sub-samples were dried at a temperature of 105˚C for 24 h according to SCAN-CM 39:94 standards to determine dry matter content. Individual chip class sub-samples were expressed as a mass percentage of the total sample bone dry mass (SCAN-Test 1994). Chip MC was calculated for individual samples according to D4442-07 standards for the direct moisture content measurement of wood and wood-base materials (ASTM International 2007).

2.2 Statistics Two way multi factorial analysis of variance (ANOVA) was used to analyse the data using the STATISTICA 10 software package (StatSoft 2012). The null hypothesis tested was for no treatment interaction effect. If the null hypothesis was rejected, individual treatment effects were compared. However if the null hypothesis was not rejected, treatment interactions are significant and only the interactions between treatments were analysed, as treatment effects were depen-

299

J.-P. van der Merwe et al.

The Impact of Log Moisture Content on Chip Size Distribution When Processing ... (297–307)

dent of each other (Milton and Arnold 1999). When significant differences were found between treatments or treatment interaction effects (α=0.05), significant differences between individual means were determined using a post hoc Bonferroni t-test, as ANOVA residuals were normally distributed. The least square means (LSM) method was used for the representation of significant treatment interactions.

3. Results 3.1 Chip moisture content The interaction between log drying period and log class had a significant effect on wood chip MC (p<0.001), Table 2. Two week dried top logs with a MC of 31.8% produced chips 12.1% lower in MC than chips produced from two week dried base logs (31.8% versus 43.9%) Table 2 Effect of log drying period and log class on chip MC Log drying period

1 week

2 weeks

Log class

MC, %

Base

49.4±0.31e

Middle

47.9±0.32d

Top

45.0±0.30c

Base

43.9±0.24c

3.2 Chip size distribution Chip size and uniformity differed significantly in relation to log drying period (Table 3), log class (Table 4) and the interaction between the treatments (Table 5).

Middle

38.1±0.27

Top

31.8±0.38a

and 6.2% lower in MC than chips produced from two week dried middle logs (31.8% versus 38.1%). Two week dried top logs also produced chips 17.6% lower in MC than chips produced from one week dried base logs (31.8% versus 49.4%), 16.1% lower in MC than chips produced from one week dried middle logs (31.8% versus 47.9%) and 13.1% lower in MC than one week dried top logs (31.8% versus 45.0%). Two week dried middle logs with a MC of 38.1% produced chips 5.9% lower in MC than two week dried base logs (38.1% versus 43.9%). Two week dried middle logs also produced chips 11.3% lower in MC than one week dried base logs (38.1% versus 49.4%), 9.9% lower in MC than one week dried middle logs (38.1% versus 47.9%) and 6.9% lower in MC than one week dried top logs (38.1% versus 45.0%). Two week dried base logs with a MC of 43.9% produced chips 5.5% lower in MC than one week dried base logs (43.9% versus 49.4%) and 4.0% lower in MC than one week dried middle logs (43.9% versus 47.9%). One week dried top logs with a MC of 45.0% produced chips 4.4% lower in MC than one week dried base logs (45.0% versus 49.4%) and 2.9% lower in MC than one week dried top logs (45.0% versus 47.9%). One week dried middle logs with a MC of 47.9% produced chips 1.5% lower in MC than one week dried base logs (47.9% versus 49.4%).

Table 3 Means table for over-size chips, over-thick chips, accept chips, pin chips and fines produced from logs subject to respective drying periods, expressed as a percentage Log drying period 1 week 2 weeks

Chip size distribution, % Over-size 0.08±0.02 0.15±0.04

Over-thick

Accepts

Pins

Fines

74.2±0.36

20.8±0.30

3.5±0.044

79.9±0.27

14.8±0.20

2.6±0.045

1.5±0.08 2.5±0.10

Table 4 Means table for over-size chips, over-thick chips, accept chips, pin chips and fines produced from respective log classes, expressed as a percentage Log class Base

300

Chip size distribution, % Over-size 0.09±0.03

Over-thick

Accepts

Pins

Fines

81.1±0.28

14.6±0.26

2.6±0.057

1.6±0.09

Middle

0.12±0.04

1.9±0.10

77.1±0.34

17.9±0.33

3.0±0.053

Top

0.13±0.04

2.6±0.13b

72.9±0.43

20.9±0.42

3.5±0.061

Croat. j. for. eng. 37(2016)2

The Impact of Log Moisture Content on Chip Size Distribution When Processing ... (297–307)

J.-P. van der Merwe et al.

Table 5 Means table for over-size chips, over-thick chips, accept chips, pin chips and fines produced for the interaction between log drying periods and log classes, expressed as a percentage Log drying period

1 week

2 weeks

Log class

Chip size distribution, % Over-size

Over-thick

Accepts

Pins d

Fines c

Base

0.05±0.02

1.0±0.10

79.0±0.31

16.8±0.28

3.1±0.062c

Middle

0.13±0.06

1.3±0.10

74.3±0.37b

20.9±0.31d

3.4±0.050d

Top

0.06±0.03

2.1±0.16

69.3±0.39

24.7±0.32

3.9±0.077e

Base

0.13±0.06

2.1±0.12

83.2±0.25e

12.4±0.19a

2.1±0.038a

Middle

0.12±0.05

2.4±0.15

80.0±0.23d

14.9±0.21b

2.5±0.042b

3.1±0.20

3.2±0.072c

Top

0.20±0.08

No significant interactions were observed between drying period and log class (p=0.370) and the amount of over-sized chips produced. The individual main effects of drying period and log class also had no significant effect on the amount of over-sized chips produced (p=0.085 and p=0.711), (Table 3, Table 4 and Table 5). Main effect log drying period and log class had a significant effect on the amount of over-thick chips produced (p<0.001), (Table 3 and Table 4). One week dried logs produced 1.0% less over-thick chips than logs dried for two weeks (1.5% versus 2.5%). Base logs produced 1.0% less over-thick chips than top logs (1.6% versus 2.6%). Middle logs produced 0.7% less over-thick chips than top logs (1.9% versus 2.6%). The interaction between log drying period and log class had a significant effect on amount of accepts produced (p<0.001), (Table 5). One week dried top logs produced 9.7% less accepts than one week dried base logs (69.3% versus 79.0%) and 5.0% less accepts than one week dried middle logs (69.3% versus 74.3%). One week dried top logs also produced 13.9% less accepts than two week dried base logs (69.3% versus 83.2%), 10.7% less accepts than two week dried middle logs (69.3% versus 80.0%) and 7.2% less accepts than two week dried top logs (69.3% versus 76.5%). One week dried middle logs produced 4.7% less accepts than one week dried base logs (74.3% versus 79.0%). One week dried middle logs also produced 9.0% less accepts than two week dried base logs (74.3% versus 83.2%), 5.7% less accepts than two week dried middle logs (74.3% versus 80.0%) and 2.2% less accepts than two week dried top logs (74.3% versus 76.5%). One week dried base logs produced 4.2% less accepts than two week dried base logs (79.0% versus 83.2%). Two week dried top logs produced 6.7% less accepts than two week dried base logs (76.5% versus 83.2%) and 3.5% less accepts than two week dried Croat. j. for. eng. 37(2016)2

76.5±0.40

17.0±0.31

middle logs (76.5% versus 80.0%). Two week dried top logs also produced 2.5% less accepts than one week dried base logs (76.5% versus 79.0%). Two week dried middle logs produced 3.2% less accepts than two week dried base logs (80.0% versus 83.2%). The interaction between log drying period and log class had a significant effect on amount of pins produced (p<0.001), (Table 5). Two week dried base logs produced 2.5% less pins than two week dried middle logs (12.4% versus 14.9%) and 4.6% less pins than two week dried top logs (12.4% versus 17.0%). Two week dried base logs also produced 4.4% less pins than one week dried base logs (12.4% versus 16.8%), 8.4% less pins than one week dried middle logs (12.4% versus 20.9%) and 12.3% less pins than one week dried top logs (12.4% versus 24.7%). Two week dried middle logs produce 2.1% less pins than two week dried top logs (14.9% versus 17.0%). Two week dried middle logs also produce 1.9% less pins than one week dried base logs (14.9% versus 16.8%), 5.9% less pins than one week dried middle logs (14.9% versus 20.9%) and 9.8% less pins than one week dried top logs (14.9% versus 24.7%). Two week dried top logs produced 3.8% less pins than one week dried middle logs (17.0% versus 20.9%) and 7.7% less pins than one week dried top logs (17.0% versus 24.7%). One week dried base logs produced 4.0% less pins than one week dried middle logs (16.8% versus 20.9%) and 7.9% less pins than one week dried top logs (16.8% versus 24.7%). One week dried middle logs produced 3.9% less pins than one week dried top logs (20.9% versus 24.7%). The interaction between log drying period and log class had a significant effect on amount of fines produced (p=0.016), (Table 5). Two week dried base logs produced 0.4% less fines than two week dried middle logs (2.1% versus 2.5%) and 1.1% less fines than two week dried top logs (2.1% versus 3.2%). Two week

301

J.-P. van der Merwe et al.

The Impact of Log Moisture Content on Chip Size Distribution When Processing ... (297â&#x20AC;&#x201C;307)

dried base logs also produced 1.0% less fines than one week dried base logs (2.1% versus 3.1%), 1.3% less fines than one week dried middle logs (2.1% versus 3.4%) and 1.8% less fines than one week dried top logs (2.1% versus 3.9%). Two week dried middle logs produced 0.7% less fines than two week dried top logs (2.5% versus 3.2%). Two week dried middle logs also produced 0.6% less fines than one week dried base logs (2.5% versus 3.1%), 0.9% less fines than one week dried middle logs (2.5% versus 3.4%) and 1.3% less fines than one week dried top logs (2.5% versus 3.9%). Two week dried top logs produced 0.2% less fines than one week dried middle logs (3.2% versus 3.4%) and 0.7% less fines than one week dried top logs (3.2% versus 3.9%). One week dried base logs produced 0.3% less fines than one week dried middle logs (3.1% versus 3.4%) and 0.8% less fines than one week dried top logs (3.1% versus 3.9%). One week dried middle logs produced 0.5% less fines than one week dried top logs (3.4% versus 3.9%).

4. Discussion 4.1 Moisture content Previous studies have found that drying rates increased with decreasing logs size, hence the lowest chip MC was expected to be recorded for chips produced from the smaller top logs (Hartsough et al. 2000, Connel 2003, Defo and Brunette 2007). Chip MC varied according to log drying period and log classes used for chip production. Differences in chip MC produced from the log classes were greater during the second week of drying (Table 2). For 2 week dried logs, the MC of chips produced from top logs was 6.3% lower than that of middle logs (31.8% vs. 38.1%) and chips produced from middle logs were 5.8% lower than base logs (38.1% vs. 43.9%). However, chips produced from 1 week dried top logs were 2.9% lower in MC than chips produced from middle logs (45.0% vs. 47.9%) and chips produced from middle logs were 1.5% lower in MC than chips produced from base logs (47.9% vs. 49.4%). The relatively low rate of moisture loss during the first week of drying was most likely due to the logs being protected by a tarpaulin for 6 days during the transport from the harvesting site to the chipping facility 1800 km to the south-west of the country (Persson et al. 2002, Gjerdrum and Salin 2009). Differences in chip MC also gradually increased with decreasing log size when compared to chips produced from the different log section classes subjected to 1 and 2 week drying periods. Two week dried base logs produced chips 5.5% lower in MC than 1 week

302

dried base logs (43.9% vs. 49.4%), 2 week dried middle logs produced chips 9.8% lower in MC than 1 week dried middle logs (38.1% vs. 47.9%) and 2 week dried top logs produced chips 13.2% lower in MC than 1 week dried top logs (31.8% vs. 45.0%).

4.2 Chip size distribution The methodology developed for this study to investigate the influence of log drying period and log size on chip quality is unique. The method applied to separate chip fractions produced during chipping was sound in relation to the study objectives. However, an additional screen separating small sized accepts from prime sized accepts during screening would have been beneficial to better understand the trends observed regarding chip size distributions and the factors affecting them. Unfortunately, the necessary equipment was not available. The main effects, as well as the interactions between the main effects, had no significant influence on oversize chip produced. Previous studies have shown that chips produced from horizontal feed disc chippers have significantly less over-size chips than chips produced from drop feed disc chippers (Twaddle and Watson 1992a, Twaddle and Watson 1992d, Nati et al. 2014). It has been found that logs fed into drop feed chippers have highly variable log orientations during chipping (Isokangas 2010). Logs from the thin ends of trees are often forced into chipping knives causing fracturing and even breakages due to uncontrolled log feeding speeds (Isokangas 2010). These factors could often lead to greater over-size chip production (Isokangas 2010). Watson and Stevenson (2007) found that the amount of over-thick chips produced during chipping increased with decreasing log MC. One week dried logs produced chips with significantly less over-thick chips than 2 week dried logs (1.5% versus 2.5%). Over-thick chip production also increased with a decreasing log size (Table 4). Base logs produced 1.0% less over-thick chips than top logs (1.6% versus 2.6%) and middle logs produced 0.7% less over-thick chips than top logs (1.9% versus 2.6%). Previous studies have suggested that the production of over-thick chips are related to wood quality defects such as irregular grain in wood and knots (Bjurulf 2006 and CĂĄceres et al. 2016). Knot content is a function of branch frequency and size (Malan 2003). For eucalypts, branch sizes increase with tree height (Dye et al. 2004, Kearney et al. 2007). Knot content will therefore proportionally increase with tree height and decrease with age, which would explain the greater production of over-thick chips for top logs (Dye et al. 2004, Kearney et al. 2007). Croat. j. for. eng. 37(2016)2

The Impact of Log Moisture Content on Chip Size Distribution When Processing ... (297–307)

In addition, the interactions between drying period and log class had a significant effect on the percentage of accept chips produced (Table 5). One week dried logs produced chips with significantly less accept chips than 2 week dried logs. This trend was also observed within each log section class. One week dried top logs produced chips with 7.2% less accept chips than 2 week dried top logs (69.3% vs. 76.5%). One week dried middle logs produced chips with 5.7% less accept chips than 2 week dried middle logs (74.3% vs. 80.0%) and 1 week dried base logs produced chips with 4.2% less accept chips than 2 week dried base logs (79.0% vs. 83.2%). The percentage of accept chips produced during chipping is a function of the percentage of undesirable chip size fractions produced. As the percentage of under-sized and over-sized chip fractions increases, the proportion of accept chips correspondingly decreases. Individual log sections also had a significant effect on the amount of accept chips produced during chip production. With decreasing log size, the percentage of accept chips produced decreased linearly. The trend was also observed for chips produced from logs dried for both 1 week and 2 week drying periods (Table 5). Comparing log sections dried for 1 week, it was found that top logs produced chips with 5.0% less accept chips than middle logs (69.3% vs. 74.3%) and middle logs produced chips with 4.7% less accept chips than base logs (74.3% vs. 79.0%). Comparing log section classes dried for 2 weeks it was found that top logs produced chips with 3.5% less accept chips than middle logs (76.5% vs. 80.0%) and middle logs produced chips with 3.2% less accept chips than base logs (80.0% vs. 83.2%). The effect of wood MC on chip size and uniformity has been investigated internationally and it was shown that chips produced from logs with low or high MC produced greater amounts of non-optimum chips during chipping (Pulkki 1991, Uhmeier and Persson 1997, Watson and Stevenson 2007, Färlin 2008, Hellström 2010, Niedźwiecki 2011, Spinelli et al. 2011 and Mihelič et al. 2015). Surface wood dried quicker than sub-surface wood (Defo and Brunette 2007). With log surface to volume ratios increasing with decreasing log size, smaller logs have a larger portion of surface wood with greater drying rates leading to larger portions of excessively dry wood and lower proportions of accept chips produced (Bassler 1987, Pulkki 1991, Uhmeier and Persson 1997, Defo and Brunette 2007, Färlin 2008, Hellström 2008). After the 1 week drying period, the surface wood is dryer than the sub-surface wood, which then negatively affects accept chip production (Araki 2002, Defo and Brunette 2007, Watson and Stevenson 2007, Niedźwiecki 2011). Croat. j. for. eng. 37(2016)2

J.-P. van der Merwe et al.

Interactions between the drying period and log class also had a significant effect on the amount of pins produced during chip production (Table 5). Two week dried logs produced chips with significantly less pins than 1 week dried logs. This trend was also observed within each log class. Two week dried base logs produced chips with 4.4% less pins chips than 1 week dried base logs (12.4% vs. 16.8%). Two week dried middle logs produced chips with 6.0% less pins than 1 week dried middle logs (14.9% vs. 20.9%) and 2 week dried top logs produced chips with 7.7% less pins than 1 week dried top logs (17.0% vs. 24.7%). Moreover, log class had a significant effect on the amount of accept chips produced during chip production. With decreasing log size, the percentage of pins produced increased linearly. The trend was also observed for chips produced from logs dried for both 2 week and 1 week drying periods. Comparing log sections dried for 2 weeks, it was found that base logs produced chips with 2.5% less pins than middle logs (12.4% vs. 14.9%) and middle logs produced chips with 2.1% less pins than top logs (14.9% vs. 17.0%). Comparing log sections dried for 1 week, it was found that base logs produced chips with 4.1% less pins than middle logs (16.8% vs. 20.9%) and middle logs produced chips with 3.8% less pins than top logs (20.9% vs. 24.7%). Wood MC has been found to have a significant impact on chip size and uniformity. Especially when wood moisture content was excessively high or low, the amount of non-optimum chips produced increased (Pulkki 1991, Uhmeier and Persson 1997, Watson and Stevenson 2007, Färlin 2008, Hellström 2010, Niedźwiecki 2011, Spinelli et al. 2011 and Mihelič et al. 2015). As previously mentioned, surface wood dried faster than sub-surface wood (Defo and Brunette 2007). As log surface to volume ratios increased with decreasing log size, smaller logs have larger portions of surface wood with greater drying rates that lead to larger portions of excessively dry wood and greater amounts of pins produced (Bassler 1987, Pulkki 1991, Uhmeier and Persson 1997, Defo and Brunette 2007, Färlin 2008, Hellström 2008). After the 1 week drying period, the surface wood is dryer than the sub-surface wood, which results to the production of greater quantities of pins (Araki 2002, Defo and Brunette 2007, Watson and Stevenson 2007, Niedźwiecki 2011). Interactions between respective log drying periods and log classes had a significant effect on chip fines produced during chipping (Table 5). In fact, across all log classes, 2 week dried logs produced chips with significantly less fines than 1 week dried logs. Comparing fines content across the drying periods for in-

303

J.-P. van der Merwe et al.

The Impact of Log Moisture Content on Chip Size Distribution When Processing ... (297–307)

dividual log classes, the fines content difference increased with increasing log size. Two week dried base logs produced chips with 1.0% less fines than 1 week dried base logs (2.1% vs. 3.1%), 2 week dried middle logs produced chips with 0.9% less fines than 1 week dried middle logs (2.5% vs. 3.4%) and 2 week dried top logs produced chips with 0.7% less fines than 1 week dried top log classes (3.2% vs. 3.9%). A higher rate of moisture loss for smaller sized logs may explain why smaller sized logs have a smaller difference in the amount of fines produced, as smaller logs may be closer to optimum log MC for limiting fines production during chipping. It can be concluded that log MC has a significant effect on fines production (Bassler 1987, Araki 2002, Watson and Stevenson 2007, Niedźwiecki 2011). Chip fines content also increased with decreasing log size for chips produced from logs dried for respective drying periods. One week dried base logs produced chips with 0.3% less fines than 1 week dried middle logs (3.1% vs. 3.4%) and 1 week dried middle logs produced chips with 0.5% less fines than 1 week dried top logs (3.4% vs. 3.9%). Two week dried base logs produced chips with 0.4% less fines than 2 week dried middle logs (2.1% vs. 2.5%) and 2 week dried middle logs produced chips with 0.7% less fines than 2 week dried top logs (2.5% vs. 3.2%). Log surface to volume ratios increased exponentially as log size decreased; therefore, smaller log classes have greater proportions of exposed surface wood with low MC. Larger proportions of drier surface wood potentially led to greater quantities of chip fines during chip production (Araki 2002, Watson and Stevenson 2007, Niedźwiecki 2011).

5. Conclusion A study to determine the impact of log drying period on wood chip size distributions was conducted. The study included 2 log drying periods with logs dried for 1 week and 2 weeks, respectively. In addition, the effect of log size on the production of chip size distributions was analysed. Trees included in the study were harvested during the relatively wet spring months in the Kwambonambi area in Northern KwaZulu-Natal of South Africa. The logs samples were chipped at a chipping facility located in the WesternCape province of South Africa. The chip sample analysis was done at Stellenbosch University. Results show that drying period and log size class had a significant impact on chip size fractions produced during chipping (over-thick chips, accepts, pins and fines). Drying period and log class had a signifi-

304

cant effect on the amount of over-thick chips produced. One week dried log classes produced 1.0% less over-thick chips than 2 week dried log classes (1.5% vs. 2.5%). Over-thick chip production also increased with decreasing log size. Base logs produced 1.0% less over-thick chips than top logs (1.6% vs. 2.6%) and middle logs produced 0.7% less over-thick chips than top logs (1.9% vs. 2.6%). In addition, the interaction between log drying period and log class had a significant effect on accept chip production. One week dried logs produced significantly less accept chips than 2 week dried logs across all log classes. One week dried base logs produced 4.2% less accepts than 2 week dried base logs (79.0% vs. 83.2%). Likewise, 1 week dried middle logs produced 5.7% less accepts than 2 week dried middle logs (74.3% vs. 80.0%) and 1 week dried top logs produced 7.2% less accepts than 1 week dried top logs (69.3% vs. 76.5%). Accept chip production also decreased with decreasing log size for logs subject to both log drying periods. Moreover, results show that the interaction between log drying period and log class had a significant effect on the amount of pins produced during chipping. Two week dried logs produced significantly less pins than 1 week dried logs across all log classes. Two week dried base logs produced 4.4% less pins than 1 week dried base logs (12.4% vs. 16.8%). Meanwhile, 2 week dried middle logs produced 6.0% less pins than 1 week dried middle logs (14.9% vs. 20.9%) and 2 week dried top logs produced 7.7% less pins than 1 week dried top logs (17.0% vs. 24.7%). The production of pins also increased with decreasing log size for logs subject to both drying periods. Finally, findings demonstrate that the interaction between log drying period and log class had a significant effect on chip fines production. Two week dried logs produced less fines when compared to 1 week dried logs. Further, 2 week dried base logs produced 1.0% less fines than 1 week dried base logs (2.1% vs. 3.1%). Meanwhile, 2 week dried middle logs produced 0.9% less fines than 1 week dried middle logs (2.5% versus 3.4%) and 2 week dried top logs produced 0.7% less fines than 1 week dried top logs (3.2% versus 3.9%). Fines production also increased with decreasing log size for both log drying periods.

6. Recommendations Based on the results of this study, the following are the recommendations that forestry companies could follow in order to improve the quality of pulp logs used for pulp and paper manufacturing in relation to: Croat. j. for. eng. 37(2016)2

The Impact of Log Moisture Content on Chip Size Distribution When Processing ... (297–307)

Þ Debarking practices Þ Log MC Þ Tree size. Log MC greatly influences chip size and uniformity during chip production. Infield log drying periods need to be adjusted according to climatic conditions, tree species and tree size. Log assortments extracted from individual trees during harvesting vary in size, and therefore will have a wide range of drying rates. Thus, log drying periods need to be suited to a variety of log assortments to ensure that log MC is as close as possible to the optimal MC for chip production. This is also essential to ensure uniformity in chip moisture content. Additionally, it was found that tree size has a significant effect on chip quality. With decreasing log size, the amount of undesired chip fractions produced during chipping increased. As such, plantation compartments scheduled for annual harvesting operations should be revised to avoid the harvesting of undersized trees. Forestry companies should consider adjusting plantation felling ages to ensure larger tree size at the time of felling. Closer investigation is needed to determine the optimum tree size to facilitate debarking, and to maximize chip quality and pulp value recovery. Further research is also needed to determine the best MC range for chipping from a chip quality point of view.

Acknowledgments Thanks and appreciation to Mondi Forests for the allocation of a study area and for financing the project. Thanks to Bandit South Africa for assistance in chipping operations.

7. References

J.-P. van der Merwe et al.

Biermann, C.J., 1996b: Pulping fundamentals. In: Biermann, C. J. (ed.) Handbook of pulping and papermaking, 2nd edition. Oxford, Academic Press 3: 55–97. Bjurulf, A., 2006: Chip Geometry: Methods to impact the geometry of market chips. Doctoral Thesis, Swedish University of Agricultural Sciences, 43 p. Bjurulf, A., 2005: Effects on chip size distribution when an extra screen is added in SCAN-CM 40:01. TAPPI Journal 4(4): 28–31. Broderick, G., Cacchione, E. Héroux, Y., 1998: The importance of distribution statistics in the characterization of chip quality. TAPPI Journal 81(2): 131–142. Cáceres, C.B., Hernández, R.E., Koubaa, A., 2016: Effects of log position in the stem and cutting width on size distribution of black spruce chips produced by a chipper-canter. Wood and Fiber Science 48(1): 1–18. Clarke, C., Palmer, B., Gounden, D., 2008: Understanding and adding value to Eucalyptus fibre. Southern Forests: A Journal of Forest Science 70(2): 169–174. Connel, M.J., 2003: Log Presentation : Log damage arising from mechanical harvesting or processing. Forestry and Forest Products Report No. PN02.1309. Commonwealth Scientific and Industrial Research Organisation, Yarralumla, 64 p. Dang, V.Q., Nguyen, K.L., 2008: A universal kinetic model for characterisation of the effect of chip thickness on Kraft pulping. Bioresource technology 99(5): 1486–1490. Defo, M., Brunette, G., 2007: Application of a mathematical model to the analysis of the influence of length and diameter on log drying rate. Wood and Fiber Science 39(1): 16–27. Ding, F., Benaoudia, M., Bédard, P., Lanouette, R., Lejeune, C., Gagné, P., 2005: Wood chip physical quality definition and measurement. Pulp and Paper Canada 106(2): 27–32. Dovey, S.B., 2012: Effects of clear felling and residue management on nutrient pools, productivity and sustainability in a clonal eucalypt stand in South Africa. Doctoral thesis, Stellenbosch University, 246 p. Dye, P., Versfeld, D., 2007: Managing the hydrological impacts of South African plantation forests: an overview. Forest Ecology and Management 251(1–2): 121–128.

Araki, D.S., 2002: Fibre recovery and chip quality from debarking and chipping fire-damaged stems. Report, Forest Engineering Research Institute of Canada, 32 p.

Dye, P., Jacobs, S., Drew, D., 2004: Verification of 3-PG growth and water-used predictions in twelve Eucalyptus plantation stands in Zululand, South Africa. Forest Ecology and Management 193(1–2): 197–218.

ASTM international, 2007: Standard Test methods for direct moisture content measurement of wood and wood-base materials, 6 p.

Eisenbies, M.H., Volk, T.A., Pate, A., 2014: Changes in feedstock quality in willow chip piles created in winter 1 from a commercial scale harvest. Biomass and Bioenergy 86: 180–190.

Balakrishnan, J., 2008: Effect of wood storage and seasonal variations of hardwoods on pulp and paper properties. MEng Thesis, University of New Brunswick, 7–25.

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

Bassler, G.A., 1987: Producing pulp quality chips in the woods: A short term study on three portable delimber-debarker systems. Master thesis, Oregon State University, 135 p. Biermann, C.J., 1996a: Wood and fiber fundamentals. In: Biermann, C.J. (ed.). Handbook of pulping and papermaking, 2nd edition. Oxford, Academic Press 2: 13–50. Croat. j. for. eng. 37(2016)2

Färlin, A., 2008: Analysis of chip quality and value at Norske Skog Pisa Mill. Master thesis, Swedish University of Agricultural Science, 16–48. Forestry Economics Services CC, 2011: Report on commercial timber resources and primary roundwood processing in

305

J.-P. van der Merwe et al.

The Impact of Log Moisture Content on Chip Size Distribution When Processing ... (297–307)

South Africa. Department of Agriculture, Forestry and Fisheries, Pretoria, 138 p.

computing sciences. ISBN-10: 007246836X, 2nd edition, New York: McGraw-Hill, 819 p.

Forestry South Africa, 2013: Abstract of South African forestry facts for the year 2010/2011. Department of Agriculture, Forestry and Fisheries, Johannesburg, 1 p.

Nati, C., Eliasson, L., Spinelli, R., 2014: Effect of chipper type, biomass type and blade wear on productivity, fuel consumption and product quality. Croatian Journal of Forest Engineering 35(1): 1–7.

Gjerdrum, P., Salin, J., 2009: Open-air drying of scots pine transmission poles prior to creosote treatment. The 7th International Conference: Wood Science and Engineering in the Third Millennium, International Union of Forest Research Organizations, Brasov, Romania, 95–102. Gulsöy, S.K., 2012: Factors affecting Kraft pulp yield. Kastamonu University, Journal of Forestry Faculty 12(1): 144–156. Hartler, N., 1996: Achievement and significance of optimal chip quality. TAPPI Journal 79(2): 259–264. Hartsough, B., Spinelli, R., Pottle, S., Klepac, J., 2000: Fiber recovery with chain flail delimbing/debarking and chipping of hybrid poplar. Journal of Forest Engineering 11(2): 37–41. Hellström, L.M., 2010: On the wood chipping process – a study on basic mechanisms in order to optimize chip properties for pulping. Doctoral thesis, Mid Sweden University, 46 p. Hellström, L.M., 2008: Fracture processes in wood chipping. Licentiate thesis, Mid Sweden University, 32 p. 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. Isokangas, A., 2010: Analysis and management of wood room. Doctoral thesis, University of Oulu, 118 p. Kearney, D., James, R., Montagu, K., Smith, R.G.B., 2007: The effect of initial planting density on branching characteristics of Eucalyptus pilularis and E. grandis. Australian Forestry 70(4): 262–268. Laitinen, J., 2001: Determination of residual bark in debarked pulpwood using image analysis. TAPPI Journal 84(1): 1–10. MacLeod, M., 2007: The top ten factors in Kraft pulp yield. Paper and Timber 89(4): 1–7. MacLeod, J.M., McPhee, F.J., Kingsland, K.A., Tristram, R.W., O’Hagan, T.J., Kowalska, R.E., Thomas, B.C., 1995: Pulp strength delivery along complete Kraft mill fiber lines. TAPPI Journal 78(8): 153–160. Mafia, R.G., Santos, P.C, Demuner, B.J., Massoquete, A., Sartório, R.C., 2012: Eucalyptus wood decay: effects on productivity and quality of cellulose. Forest Pathology 42(4): 321–329. Malan, F.S., 2003: The wood quality of the South African timber resource for high-value solid wood products and its role in sustainable forestry. The Southern African Forest Journal 198(1): 53–62. Mihelič, M., Spinelli, R., Magagnotti, N., Poje, A., 2015: Performance of a new industrial chipper for rural contractors. Biomass and Bioenergy 83: 152–158. Milton, J.S., Arnold J.C., 1999: Introduction to probability and statistics: principles and applications for engineering and the

306

Niedźwiecki, Ł., 2011: Energy requirements for comminution of fibrous materials – qualitative chipping model. Master thesis, Linnaeus University, 74 p. Patt, R., Kordsachia, O., Süttinger, R., 2012: Ullmann’s encyclopedia of industrial chemistry. 8th edition. Wiley-VCH, ed. Wiley-VCH Verlag GmbH, 477–534. Persson, E., Sjöström, M., Sundblad, L., Wiklund, S., Wilhelmsson, L., 2002: Färskt virke–en utmaning för skogsbruk och virkesmätning [Fresh wood-a challenge for forestry and timber measurement]. Resultat: Skogforsk Report No.8. Skogforsk, 1–4. Phanphanich, M., Mani, S., 2009: Drying characteristics of pine forest residues. BioResources 5(1): 108–121. Pulkki, R., 1991: A literature synthesis on the effects of wood quality in the manufacture of pulp and paper. Technical Note TN-171. Forest Engineering Research Institute of Canada 10 p. Qian, Y., Tessier, P.J.C., Dumont, G.A., 1994: Fuzzy logic modelling and optimization of a wood chip refiner. TAPPI Journal 77(2): 181–186. Röser, D., Mola-Yudego, B., Sikanen, L., Prinz, R., Gritten, D., Emer, B., Väätäinen, K., Erkkilä, A., 2011: Natural drying treatments during seasonal storage of wood for bioenergy in different European locations. Biomass and Bioenergy 35(10): 4238–4247. 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): 111–119. Santos, A.J., Anjos, A., Simões, R.M.S., 2008: Influence of kraft cooking conditions on the pulp quality of Eucalyptus globulus. Appita Journal 61(2): 148–155. Scandinavian pulp and paper board testing committee, 1994: Wood chips for pulp production, size distribution, SCAN-CM 39:94: 1–2. Scandinavian pulp and paper board testing committee, 1994: Wood chips for pulp production, size distribution, SCAN-CM 40:94: 1–4. 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. Spinelli, R., Hartsough, B., 2001: A survey of Italian chipping operations. Biomass and Bioenergy 21(6): 433–444. Svedman, M., Tikka, P., Luhtanen, M., 1998: Effects of softwood morphology and chip thickness on pulping with a displacement Kraft batch process. TAPPI Journal 81(7): 157–168. Tessier, P., Broderick, G., Desrochers, C., 1999: Chip size distribution for an ultra-high-yield sulphite process. TAPPI Journal 82(12): 1–8. Croat. j. for. eng. 37(2016)2

The Impact of Log Moisture Content on Chip Size Distribution When Processing ... (297–307)

J.-P. van der Merwe et al.

Tikka, P., Tähkänen, H., Kovasin, K., 1993: Chip thickness vs. Kraft pulping performance: Experiments by multiple hanging baskets in batch digesters. TAPPI Journal 76(3): 131–136.

Twaddle, A.A., Watson W.F., 1992d: Survey of disc chippers in roundwood chipping yards of south eastern United States. TAPPI Journal 75(12): 77–81.

True, G.R., 2006: Pins and fines: a small cause of big losses. TAPPI Frontline Focus 2(12): 2–3.

Uhmeier, A., Persson, K., 1997: Numerical Analysis of Wood Chipping. Holzforschung 51(1): 83–90.

Twaddle, A., 1997: The influence of species, chip length, and ring orientation on chip thickness. TAPPI Journal 80(6): 123– 131.

Uhmeier, A., 1995: Some fundamental aspects of wood chipping. TAPPI Journal 78(10): 79–86.

Twaddle, A.A., Watson W.F., 1992a: Shortwood vs. longwood chipping what impact on chip distribution? TAPPI Journal 75(6): 107–111. Twaddle, A.A., Watson W.F., 1992b: Quantification of factors affecting chip quality from disc chippers. TAPPI Journal 75(10): 135–140. Twaddle, A.A., Watson W.F., 1992c: Survey of drum debarkers in roundwood chipping yards of south eastern United States. TAPPI Journal 75(11): 105–107.

Walton, S.L., Hutto, D., Genco, J.M., van Walsum, G.P., van Heiningen, A.R.P., 2010: Pre-extraction of hemicelluloses from hardwood chips using an alkaline wood pulping solution followed by Kraft pulping of the extracted wood chips. Industrial and Engineering Chemistry Research 49(24): 12638– 12645. Watson, W.F., Stevenson, R., 2007: The effect of seasonal variation in wood moisture content on chip size and Kraft pulping. In Engineering, Pulping and Environmental Conference: Innovations in Engineering, Pulping and Environmental. Jacksonville, Florida, USA: Technical Association of the Pulp and Paper Industry, 1–8.

Authors’ address: Jaco-Pierre van der Merwe, MSc. e-mail: JPvdMerwe@york.co.za Pierre Ackerman, PhD.* e-mail: packer@sun.ac.za Stellenbosch University Department of Forest and Wood Science Private Bag X1, Matieland SOUTH AFRICA Reino Pulkki, PhD. e-mail: rpulkki@lakeheadu.ca Stellenbosch University Department of Forest and Wood Science Private Bag X1, Matieland SOUTH AFRICA Lakehead University Faculty of Forestry and Forest Environment 955 Oliver Road, Thunder Bay, Ontario CANADA

Received: February 05, 2016. Accepted: March 18, 2016. Croat. j. for. eng. 37(2016)2

Dirk Längin, PhD. e-mail: Dirk.Laengin@mondigroup.co.za Mondi 380 Old Howick Road Hilton 3245 SOUTH AFRICA * Corresponding author

307

Original scientific paper

Evaluation of Chipping Productivity with Five Different Mobile Chippers at Different Forest Sites by a Stochastic Model Mika Yoshida, Simon Berg, Rin Sakurai, Hideo Sakai Abstract It is important to evaluate chipping productivity that often differed according to the timing of observations and varied unexpectedly. A variation in production was the major concern of stakeholders for sustainable forest operation to establish regularly attainable production schedules on many operational levels. The aim of this study was to estimate the variance of chipping productivity by using a stochastic simulation model to achieve the objective evaluation of chipper performances. Chipping operations of five different kinds of mobile chippers, i.e. three smaller and two middle and larger ones in horse powers, were investigated. Probability distributions of material size and feeding time for chipping in a log-normal distribution were estimated. The estimates were made based on chipping operations performed 2000 or 4000 times by mechanical repetitions. Except for the largest chipper, whose observed productivity was 338 loose m3/hr, all of the observed productivities, varying from 18 to 68 loose m3/hr, were located within a two-sided confidential interval whose difference between both ends was 4 to 10 loose m3/hr. The estimates were, generally, reliable with small variances around the median productivity values in the model. By this stochastic model, chipper productivity could be shown objectively, while the accuracy would be improved more by increasing sample size and accurate material size measurement. It was elucidated that the operations followed by chipping should encompass enough volume capacity to provide stable chipping productivity. Keywords: biomass energy, feeding operation, forestry operation, stochastic simulation, supply chain management

1. Introduction The wood use as a renewable energy resource is aÂ global trend. In EU, for instance, the target rate of renewable energy share was clearly stated as 20% in 2020 (European Commission 2013) and the share of woody biomass might account for two-thirds of the target (European Commission 2015). In Canada and the US, the regular share of wood as a renewable energy was also reported (Government of Canada 2014, US Energy Information Administration 2015). Wood demand as an energy resource enabled the use of lower quality timbers and logging residues. It was, therefore, expected to provide additional resources and profit for forest companies or forestland owners while improving forest health and quality. Croat. j. for. eng. 37(2016)2

In Japan, the planted forests occupied a quarter of land, and about half of them were under 50 years old (Japanese Forestry Agency 2015). It was a major concern to manage and utilize the lower quality of timbers produced from such planted forests. While there were conventional chip supply chains for pulp industries using fixed chippers at factories or storage sites, some mobile chippers had begun to be introduced to produce wood chips at forest roadside with the expectation of stable chip supply to power plants as a resource. One of the emerging issues on mobile chippers was estimating the productivity of chipping operations because their working conditions and wood materials always changed compared to fixed chippers. At the same time, the supply chain with chipping at forest roadside for the timely production of chips

309

M. Yoshida et al. Evaluation of Chipping Productivity with Five Different Mobile Chippers at Different Forest ... (309–318)

required a well-balanced transportation and logistics to factory or storage sites to realize the maximum utilization of chipper (Talbot and Suadicani 2005), because the price of chippers was generally expensive (Yoshida and Sakai 2014). Indeed, it was possible to discharge chips on the ground and to reload them onto the truck, but additional cost would occur in the reloading process (Stampfer and Kanzian 2006) and chips would lose quality by getting contaminated with soil (Spinelli and Hartsough 2001). Therefore, the importance of chipping productivity estimation of mobile chippers has been gradually recognized for optimized transportation scheduling. It was well known that the productivity of wood chipping was mainly affected by the total solid volume of a piece of material and the relationship could be expressed in various ways: such as chipping time per ton/volume for average piece size (Spinelli and Hartsough 2001, Assirelli et al. 2013); for load size of a loader (Röser et al. 2012); for piece size under a multilinear regression analysis (Ghaffariyan et al. 2013); and for butt-end diameter of logs (Yoshida and Sakai 2014). Previous models of productivity estimation were generally analyzed by using linear regression for the average of material sizes in a forest area. Furthermore, productivity usually differed according to the timing of observations. It was also proved that there were differences between estimated and actual productivities across chipper machines (e.g. Spinelli and Magagnotti 2010, Ghaffariyan et al. 2013). As the material supply and the product delivery were scheduled under the typical productivity, unexpected variances of chipping productivity sometimes diminished or eroded profit in its supply chain. Therefore, the realistic and typical value of chipping productivity and its possible variance should be grasped in advance for profitable and sustainable management of chip supply chain. The variance of chipping productivity seemed to be decided mainly by fluctuations in feeding time of a piece of material to a chipper. Compared to other forestry operations, such as timber harvesting, chipping operation had less complex operations (Röser et al. 2012), and operator’s effect could be regarded as secon dary and minute to overall production (Spinelli and Magagnotti 2010). Thus, the variance of chipping productivity might be mainly derived from the accumulation of uncertainty in material volume and feeding operations, and the machinery condition such as blade wear (Spinelli et al. 2014). The aim of this study is to estimate the variance of chipping productivity by using a stochastic simulation model for the objective evaluation of chipper performance. Stochastic simulation method was selected to analyze the productivity variance since the method is suitable to describe complex problems including in-

310

teractions and uncertainty. This approach had been used in previous studies such as Gallis (1996), Talbot and Suadicani (2005), Asikainen (2010) and ZamoraCristales et al. (2013) to analyze interactions among production and transportation processes. Since feeding and chipping operations were in an interactive relationship (Röser et al. 2012), this study analyzed productivity of five different types of currently used mobile chippers considering the interaction within a process and verified the robustness of the method.

2. Material and methods 2.1 Time observation and machine selection To obtain real data at forest chipping site, time observations recorded by stop watch were conducted nearby the five mobile chippers during operations. The observed productive chipping system factors in a cycle were divided into two main elements: feeding to chipper and operational delay. A cycle of feeding to chipper operation, which essentially determined the productivity, was defined as the time between the start of grappling by the loader movement onto the material pile after the previous releasing of materials and the end of releasing materials on the feeder. The operational delay was time of extra operations that included cleaning the landing site; preparing logs from piles; and removal of stuck logs in the feeder, which randomly occurred. All operational delays less than 15 minutes were taken into consideration in the productivity calculation for securing comparability; they were used as one of the criteria to classify delays (Samset 1990). Other delays over 15 minutes could be regarded as avoidable delays that could be excluded from the analysis by giving appropriate considerations. Mechanical and observation delays were also excluded. To derive productivity, the common formula for chipping productivity P (loose m3/hr) was used as shown below:

P = d∑ Vi / h

(1)

i =1

Where: d density coefficient for converting solid volume to chip volume (loose m3/solid m3) n total number of feeding operations in a production period Vi material solid volume in each i time of feeding operation (solid m3) h gross working time of a chipping operation (hr) consisting of the time for feeding operations and operational delays less than 15 minutes. Croat. j. for. eng. 37(2016)2

Evaluation of Chipping Productivity with Five Different Mobile Chippers at Different Forest ... (309–318) M. Yoshida et al.

Table 1 Chipper machine details Chipper

TP250 mobile turnable

Farmi 380

YM-400C

MUS-MAX WT8-XL

CBI chip max 484VR

Country of origin

Denmark

Finland

South Korea

Austria

USA

Chipping type

Disc (2 knives)

Disc (4 knives)

Drum (8 knives)

Drum (4 knives)

Mobility

Traction

Tractor attachment

Self-propelled

Truck mounted

Traction

Engine

Internal

External

Internal

Power, kW

53.7

140.0

150.0

353.0

570.7

Feeder dimension, mm

H 260 x W 350

H 380 x W 420

H 500 x W 400

H 600 x W 640

H 762 x W 1,219

Feeding assistant

Chain conveyor

Belt conveyor

Discharger type

Blower

Material type used in the investigation

Sugi and hinoki logs of 4m in length

Sugi short logs of 2m in length

Sugi logs of 4m in length

Sugi and hinoki logs of 4m in length

Tree tops and short logs

Observation place

Paved landing

Unpaved landing

Paved landing

In this study, analysis was made of five different types and sizes of mobile chippers: TP250 mobile turnable disc chipper (expressed as TP250); Farmi 380 disc chipper; YM-400C disc chipper; MUS-MAX WT8-XL drum chipper (expressed as MUS-MAX); and CBI chip max 484VR drum chipper (expressed as CBI). They had been already in active use in their countries of origin and could be considered as representative. The chipper machine details are summarized in Table 1. TP250 was the smallest chipper in size with engine power of 53.7 kW, and it was chosen for its small size. This chipper was designed for manual loading and vehicle traction, and an independent wheel loader was used to assist its loading operation. Farmi 380 and YM-400C had similar engine power of 140 kW and 150 kW, respectively. Farmi 380 was one of the attachments connected to the tractor power take off (PTO). It had a feeding assistant chain conveyor and an integrated grapple loader. YM-400C was an independent machine mounted on a four-wheeled vehicle equipped with a feeding assistant belt conveyor and it had to be supported by a separate grapple loader. Because of their relatively smaller size and mobility, these three chippers seemed to be fitted to small-scale forestry and operations along forest roads. MUS-MAX was a truck mounted chipper with a feeding assistant belt conveyor and an integrated grapple loader. It was chosen for its all-in-one system. It had an engine with 353 kW power, and could be regarded as a middle class chipper. CBI was the representative of large chippers both in size and engine power. It was equip ped with an engine of 570.7 kW power and a belt conveyor in front of the feeder. Croat. j. for. eng. 37(2016)2

At a feeding operation, plural pieces of material were fed at every feeding cycle in the investigations except for YM-400C. Although YM-400C had a feeder big enough for plural pieces, only a piece of material was fed per cycle because plural pieces were stuck when fed. The investigation sites were located in different areas of Japan: TP250, YM-400C, and CBI were observed in Shimane Prefecture; Farmi 380 was observed in Yamagata Prefecture; and MUS-MAX in Akita Prefecture. The operational sites were at paved landings except for Farmi 380 at a storage landing.

2.2 Material size measurement The number of logs of each feeding operation was observed and recorded by a video camera to verify the number or quantity of fed materials. The materials were prepared in accordance with the usual practice and the size was measured during actual investigations by different methods. In the investigations of TP 250 and YM-400C, logs of 4 m in length were used, and their top and butt-end diameter and length were measured for the volume calculation by Smalian’s formula. In the investigation of Farmi 380, short logs of 2 m in length were used; the top-end diameter of 124 logs was randomly measured from the pile to estimate the material size distribution. It was possible to regard such short logs as column-shaped timber, and the volume was calculated by squared diameter method from topend diameters. In the investigation of CBI, the total weight of logs was measured as 7.3 wet-tonnes before chipping. The average volume of a piece of material was calculated by the number of logs and the weight density coefficient of 0.6 wet-tonne/solid m3 assuming

311

M. Yoshida et al. Evaluation of Chipping Productivity with Five Different Mobile Chippers at Different Forest ... (309–318)

that the wet-based moisture content was 50% moisture base whose oven-dry weight was 0.3 oven-dry tonnes (Greenhouse Gas Inventory Office of Japan et al. 2014). The number of fed logs at a feeding operation was regarded as the material size distribution. In the investigation of MUS-MAX, logs of 4 m in length were used; thier top-end diameters were previously classified in 2 cm increment by the operation contractor for the calculation of volume by using a volume table used in practice. The unit »loose m3« meant a cubic meter of chip volume.

2.3 Simulation method The probability distributions of feeding time and material size were assumed to follow a log-normal distribution because all of the figures were positive numbers and theoretically could be infinite. To obtain appropriate location m and scale s parameters, feeding time, and material size data, these parameters were re-sampled by bootstrapping 2000 repetitions on TP250, YM-400C, Farmi 380 and MUS-MAX, and by 4000 repetitions on CBI because of its small original sample size. The normality of generated parameters was tested by Shapiro-Wilk normality test, and the adaptability of mean and median values was also examined. By using these obtained parameters, chipping productivity at j time of repetitions Pj (loose m3/hr) was expressed by using two independent probability distributions, as the formula (2):

Pj = (1 − a)d

∑ ∫ ( f ( M ) / g( F ))

(2)

M ,F ∈D

Where: a operational delay time ratio observed in actual operations data set of trials indicating the number of operation cycles F and materials M f(M) probability distribution of material size at a feeding operation g(F) probability distribution of feeding time. These denominators and numerators were independent. The density coefficient d was set at 2.8 loose m3/solid m3 here (Serup et al. 2002). R-language ver. 3.1.2 (R Core Team 2014) was used for all analyses and simulations.

3. Results 3.1 Result of chipping observation Investigation details are summarized in Table 2. The productivity of MUS-MAX was higher than that of other three smaller chippers with the relatively lower operational delay time ratio. CBI showed the highest productivity. In all of the observations, there were no operational delays exceeding 15 minutes. The delay ratio of Farmi 380 was greater than that of others and the delay time of CBI was not observed.

3.2 Result of simulation Fig. 1 shows the histograms and fitted probability distributions for the time of feeding operation and material size of each chipper. The fitted distribution of

Table 2 Investigation details and simulation constraints TP250 mobile turnable

Farmi 380

YM-400C

MUS-MAX WT8-XL

CBI chip max 484VR

64.2

48.6

90.0

60.6

5.16

Observed cycles of feeding operation, cycles

115

Number of logs, pieces

464

115

569

111

Number of simulation cycles for feeding time function F Î

115

Number of simulation cycles for material size M Î

464

115

569

Log volume

Diameter of a short log

Log volume

The number of logs at a feeding operation

2000

4000

0.08

0.02

0.08

0.05

0.09

0.15

0.43

0.24

0.08

0.00

18.7

22.6

18.3

68.2

338.0

Chipper Total time of investigations, min

Representation of material size probability distribution f(M) The number of repetitions i 3

Average material size, solid m Operational delay time ratio

Productivity from the actual observation, m /hr

312

Croat. j. for. eng. 37(2016)2

Evaluation of Chipping Productivity with Five Different Mobile Chippers at Different Forest ... (309â&#x20AC;&#x201C;318) M. Yoshida et al.

Fig. 1 Histograms and fitted probability distributions for the time of feeding operation and material size of each chipper Croat. j. for. eng. 37(2016)2

313

M. Yoshida et al. Evaluation of Chipping Productivity with Five Different Mobile Chippers at Different Forest ... (309–318)

Table 3 Parameters of log-normal distribution for material size and feeding time, and p-value of Shapiro-wilk normality test Chipper

TP250

Farmi 380

YM-400C

MUS-MAX

CBI

Location parameter m

Scale parameter s

Probability distribution

Median

Mean

p-value

Median

Mean

p-value

f(M)

–2.59

–2.59±0.02

0.25

0.15

0.15±0.01

0.10

g(F)

4.70

4.70±0.06

0.065

0.27

0.27±0.04

0.32

f(M)

2.26

2.26±0.02

0.77

0.23

0.23±0.01

0.47

g(F)

3.47

3.47±0.04

0.014*

0.30

0.30±0.03

0.25

f(M)

–2.63

–2.63±0.05

0.74

0.48

0.48±0.04

0.35

g(F)

3.40

3.39±0.03

0.75

0.32

0.32±0.02

0.53

f(M)

–3.06

–3.06±0.007

0.47

0.18

0.18±0.005

0.55

g(F)

3.12

3.61±0.04

0.33

0.38

0.38±0.04

0.00067***

f(M)

2.08

2.07±0.11

5.95e–05***

0.38

0.37±0.06

2.2e–16***

g(F)

3.12

3.12±0.09

0.073

0.30

0.30±0.06

3.667e–08***

* Significant (p<0.05). Distribution could not clarify its normality by normality test, but median values could be used as representative *** Significant (p<0.001). Distribution could not clarify its normality by normality test, but median values could be used as representative

material size of CBI had a peaked distribution higher than its histogram, and that of MUS-MAX had a gentler peak compared to its histogram, while the others were generally well-fitted to log-normal distribution. The location and scale parameters and p-values by Shapiro-Wilk normality test of these parameters are shown in Table 3. Although the p-values were partly significant for Farmi 380, MUS-MAX and CBI, all of the location and scale parameters were concentrated in a very narrow range around the mean/median values as normal distribution. The median values were, therefore, represented as both location and scale parameters for probability distribution. The estimated productivity distributions are summarized in Table 4, and illustrated visually in Fig. 2. Productivities could be estimated with variance, and objectively evaluated. The productivities from actual observations were within each two-sided confidential interval (p =0.05) except for CBI, and the obtained productivity distributions generally seemed to be reliable. All of the productivity distributions were positively skewed, and the median values were available as the representative productivity. The productivity distribution of TP250 and YM-400C was similar to each other despite their differences in machine size and engine power. Comparing the estimated productivities of YM-400C and Farmi 380, that of Farmi 380 was higher while the machine size and engine power were similar to each other.

314

The productivity of chippers TP250, YM-400C, and Farmi 380 concentrated in a narrow variance with about 4 loose m3/hr difference in two-sided confidential interval. The estimated productivity of middle class machine MUS-MAX, which had bigger mechanical power potential, also varied only in 10 loose m3/hr differences in the two-sided confidential interval, and the productivity was also concentrated around the median value as well as those of other three smaller chippers. On the contrary, the productivity of CBI, whose engine power was about double to 10 times larger than those of others, varied in about 60 loose m3/hr difference Table 4 Summary of chipping productivity estimation using a stochastic modeling method Chipper Min, m3/hr

TP250

Farmi 380 YM-400C MUS-MAX

CBI

13.55

20.12

15.44

58.15

226.77

17.01

23.72

18.40

66.33

278.44

17.04

23.72

18.41

66.37

270.09

Max, m /hr

20.63

27.69

23.04

75.83

343.62

Two sided Lower confidential interval, m3/hr Upper

15.25

21.84

16.42

60.94

250.98

19.12

25.79

20.51

71.90

309.65

Skewness

0.14

0.15

0.16

0.06

0.20

Kurtosis

0.002

0.11

–0.018

0.10

Median, m /hr 3

Mean, m /hr 3

The term »m3« means loose m3 here

Croat. j. for. eng. 37(2016)2

Evaluation of Chipping Productivity with Five Different Mobile Chippers at Different Forest ... (309â&#x20AC;&#x201C;318) M. Yoshida et al.

Fig. 2 The productivity distribution of each chipper in the two-sided confidential interval. Indeed, the CBI had high potential of productivity, but this large difference could be a problem in production scheduling and investment. Business planning would need improvement.

4. Discussion 4.1 Suggestions for productivity improvement The operational delay time ratios of chippers varied and it implied the possibility of productivity improvement because delay had a big influence on chipping operation (Spinelli and Visser 2009). The productivity of YM-400C was similar to that of smaller chipper, TP250. Increasing the average size of material and consecutive feeding were an effective way to improve productivity. As only a log could be fed at one feeding operation for YM-400C due to its weak power of feeding roller, reducing operational delays at least to the delay time ratio that TP250 showed was also another way to improve productivity. Farmi 380 also had a larger operational delay time ratio, which might have a negative effect on the productivity at the investigation site, where material piles were located behind the feeder entrance. It would be possible to decrease such operational delay by locating the material pile in front of the feeder. The other three chippers, TP250, CBI and MUS-MAX, showed smaller operational delay time ratios. The productivity of bigger class chippers could be improved and made sustainable by preparing sufficiently large amounts of material. Croat. j. for. eng. 37(2016)2

The size of TP250 was, however, small and manual feeding system restricted log sizes even if the operator used a wheel loader for feeding assistance. Hence, it had few possibilities of productivity improvement. Conversely, MUS-MAX had a larger entrance and a feeding-assistant belt conveyor. The capacity seemed to be used at maximum because the chipper could have multiple logs fed at the same time at the feeding operation during the observation. As for MUS-MAX, its chipping productivity improvement could only be possible by extending the length of timber logs. The largest machine, CBI, had a greater feeding space for putting more material onto the conveyor during the observations. Furthermore, utilization of a grapple loader with larger grappling capacity would increase the volume of a feeding operation. In general, the appropriate use and setup of operation on site are important to achieve the full performance of chippers. It is also necessary to determine how the system could be improved to increase its performance and efficiency.

4.2 Discussion of the model There were almost no differences between the mean and median values. Log-normal distribution was generally well-correlated to express feeding time and material size distributions because chipping operation was simple to form a supply chain and stable in its production. The location and scale parameters estimated from one dataset of an actual operation could represent feeding time and material size distributions at an acceptable level by applying re-sampling method used in this study.

315

M. Yoshida et al. Evaluation of Chipping Productivity with Five Different Mobile Chippers at Different Forest ... (309–318)

The observed productivities were in the range of two-sided confidential intervals except for CBI. One possible reason for this result on the CBI was that the actual number of material during investigation might be more than that recorded. The true material size distribution of CBI in the investigation might be wider and the average would actually be smaller. The productivity variance expressed by two-sided confidential intervals could be derived from the number of trials in a simulation . Based on a common law known as the law of large numbers, the average of the results obtained from a larger number of trials should be close to the real mean. The trial numbers of CBI was smaller than others in the material size distribution, which caused wider productivity variance of more about 60 loose m3/hr difference in the two-sided confidential interval. A continuous operation with larger operation cycles and materials was, therefore, preferable to achieve productivities in a narrow variance and to establish a stable chip supply chain based on a reliable chipping productivity by neutralizing the variance in chipping productivity. For MUS-MAX, the material size was measured by the conventional method using volume table with a diameter increment of 2 cm. For example, the volume of a log with the diameter of 17.9 cm was actually calculated as that of a log with the diameter of 16.0 cm. Therefore, the histogram was discrete and concentrated on some specific volumes. The volume of materials used in this study seemed to be underestimated, causing differences between the histogram from actual data and the fitted distribution. However, this underestimation was minor in this study because the size distribution would not be varied enough to affect the result of productivity estimation, and because the number of feeding cycles in a continuous operation was enough to neutralize the variation of each feeding operation. Nevertheless, the measurement of material size was important and should be precise as much as possible to provide the estimated chipping productivity. For further development of this simulation method, it is necessary to have a large and accurate data pool of feeding time, material size distribution, and operational delay time obtained from practice on different kinds of chippers. Recently, some grapple loaders have been equipped with an automatic data collection system to record their actual productive time ratio and fuel consumption. Such technology is useful to get feeding and operational delay time automatically. Data sharing among researchers and stakeholders of chip supply chain are also useful to increase the amount of verified data for objective comparison of chippers with a comprehensive database.

316

4.3 Evaluation of five chippers and suggestion Considering the hot system in which chips were transported directly after chip production (Asikainen 1998), MUS-MAX would have an advantage in terms of interaction with transportation because the capacity of typical semi-trailers/large trucks was around 40 m3 in Japan, so that chipping time would be less than one hour by MUS-MAX and chipping and transport could be harmonized. The other three smaller chippers would take more time for chipping and they would be suitable for the combination with smaller trucks, whereas all of truck capacities commonly used in Japan seemed to be too small for CBI to show its potential. Therefore, for CBI, it was preferable to arrange a system with no interaction between chipping operation and chip transportation systems. Table 4 showed that the productivity began to vary increasingly as the chipper size became larger; therefore, the conditions of operation at chipping site, and quantity and quality of material should be made as good as possible especially for large class chippers.

5. Conclusions The variance of productivities for each chipper could be estimated from recorded observations, and it could be neutralized by increasing the number of feeding cycles. This stochastic model applying log-normal distributions both to feeding operations and material size was useful to estimate the objective productivity and its variance. Accurate material size measurement was indispensable to make this estimation more precise, and the optimal operational setup for each machine characteristics should be applied in order to present its productivity. Typical productivities could be represented by the median values of the productivity distribution, and these corresponded to their engine powers. At the same time, the variance of productivity should be taken into account when choosing chippers and planning chip supply chain.

Acknowledgements We are grateful to Mr. Ishiyama from Forest Realize Co., Ltd., Kaneyama-machi forest owners’ association, Kazuno forest owners’ association, Sekiou forest owners’ association, Okidouzen forest owners’ association and Shinwa-Sangyo Co., Ltd. for offering opportunities of chipping observation. We would like to thank Dr. Dillon Chrimes for his English revision. This work was partly supported by Grant-in-Aid for JSPS Fellows. Croat. j. for. eng. 37(2016)2

Evaluation of Chipping Productivity with Five Different Mobile Chippers at Different Forest ... (309–318) M. Yoshida et al.

6. References Asikainen, A., 1998: Chipping terminal logistics. Scandinavian Journal of Forest Research 13(1–4): 386–392. Asikainen, A., 2010: Simulation of stump crushing and truck transport of chips. Scandinavian journal of forest research 25(3): 245–250. Assirelli, A., Civitarese, V., Fanigliuo, 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. European Commission, 2013: Report from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions, Renewable Energy Progress Report. European Commission. Brussels, Belgium, 15 p. European Commission, 2015: Biomass potential. Webpage on the Internet; last access 2016 March 7. Available from http://ec.europa.eu/agriculture/bioenergy/potential/index_ en.htm.

Röser, D., Mola-Yuego, B., Prinz, R., Emer, B., Sikanen, L., 2012: Chipping operations and efficiency in different operational environments. Silva Fennica 46(2): 275–286. Samset, I., 1990: Some observations on time and performance studies in forestry. Communication of the Norwegian Forest Research Institute 43(5): 1–80. Serup, H., Falster, H., Gamborg, C., Gundersen, P., Hansen, L., Heding, N., Jakobsen, H.H., Kofman, P., Nikolaisen, L, Thomsen, I.M., 2002: Wood for Energy Production, Technology-Environment-Economy. Second revised edition. The Centre for Biomass Technology. 69 p. Spinelli, R., Hartsough, B., 2001: A survey of Italian chipping operations. Biomass and Bioenergy 21(6): 433–444. Spinelli, R., Magagnotti, N., 2010: A tool for productivity and cost forecasting of decentralized wood chipping. Forest Policy and Economics 12(3): 194–198. Spinelli, R., Visser, R., 2009: Analyzing and estimating delays in wood chipping operations. Biomass and Bioenergy 33(3): 429–433.

Gallis, C.T., 1996: Activity oriented stochastic computer simulation of forest biomass logistics in Greece. Biomass and Bioenergy 10(5–6): 377–382.

Spinelli, R., Glushkov, S., Markov, I., 2014: Managing chipper knife wear to increase chip quality and reduce chipping cost. Biomass and Bioenergy 62: 117–122.

Ghaffariyan, M.R., Spinelli, R., Brown, M., 2013: A model to predict productivity of different chipping operations. Southern Forests 75(3): 129–136.

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.

Greenhouse Gas Inventory Office of Japan, Center for Global Environmental Research, National Institute for Environmental Studies, Japan, 2014: National Greenhouse Gas Inventory Report of Japan. National Institute for Environmental Studies, Japan, 712 p.

Talbot, B., Suadicani, K., 2005: Analysis of two simulated in-field chipping and extraction system in spruce thinnings. Biosystems Engineering 91(3): 283–292.

Government of Canada, 2014: About electricity, Natural resource Canada. Webpage on the Internet; last access 2015 November 24. Available from http://www.nrcan.gc.ca/energy/electricity-infrastructure/about-electricity/7359. Japanese Forestry Agency, 2015: Annual Report on Forest and Forestry in Japan, Fiscal Year 2014. Japanese Forestry Agency, Tokyo, 206 p. (In Japanese) R Core Team., 2014: R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-project.org/.

Croat. j. for. eng. 37(2016)2

US Energy Information Administration, 2015: Annual Energy Review 10.1 Renewable energy production and consumption by primary energy source, 1949–2012. Web page on the Internet; Last access 2015 October 20. Available at http://www.eia.gov/totalenergy/data/monthly/index.cfm. Yoshida, M., Sakai, H., 2014: Importance of capital cost reduction of chippers and their required productivity. Journal of Forest Research 19(4): 361–368. Zamora-Cristales, R., Boston, K., Sessions, J., Murphy, G., 2013: Stochastic simulation and optimization of mobile chipping and transport of forest biomass from harvest residues. Silva Fennica 47(5): 1–21.

317

M. Yoshida et al. Evaluation of Chipping Productivity with Five Different Mobile Chippers at Different Forest ... (309â&#x20AC;&#x201C;318)

Authorsâ&#x20AC;&#x2122; address: Mika Yoshida, PhD.* e-mail: yoshida.mika.kf@u.tsukuba.ac.jp University of Tsukuba Faculty of life and Environmental Sciences 1-1-1 Tennoudai, Tsukuba-shi Ibaraki, 305-8577 JAPAN Prof. Hideo Sakai, PhD. e-mail: sakaih@fr.a.u-tokyo.ac.jp Simon Berg, PhD. e-mail: phd.simon.berg@gmail.com Department of Forest Science The University of Tokyo Graduate School of Agricultural and Life Sciences 1-1-1, Yayoi, Bunkyo-ku Tokyo, 113-8657 JAPAN

Received: November 2, 2015 Accepted: March 9, 2016

318

Rin Sakurai, PhD. e-mail: sakurai@cc.miyazaki-u.ac.jp University of Miyazaki Faculty of Agriculture 1-1, Gakuen-Kibanadai-Nishi, Miyazaki City Miyazaki, 889-2192 JAPAN * Corresponding author Croat. j. for. eng. 37(2016)2

Original scientific paper

A Methodological Approach Exploiting Modern Techniques for Forest Road Network Planning Andrea Laschi, Francesco Neri, Niccolò Brachetti Montorselli, Enrico Marchi Abstract A well-developed road network allows all forest activities, including wood harvesting, firefighting and recreational activities. However, forest road construction and maintenance involve economic and environmental costs. For these reasons, forest road network planning is a fundamental phase of forest management, maximising the benefits and reducing costs and impacts. Thanks to modern technologies in data collection both for terrestrial and forest characteristics, new methods and tools have been developed to improve and facilitate road planning. The aim of this study was the development of a Decision Support System for helping managers during forest road network planning, exploiting Multi-Criteria Analysis, an Analytic Hierarchy Process and Geographic Information Systems. Three steps characterised the study: Þ an in-depth survey of the existing forest road network Þ an accessibility evaluation, based on a commonly applied Italian definition, taking into account the morphological characteristics of the land Þ an estimation of the accessibility requirements through the analysis of experts’ opinions, defined as Road Needs Index, based on different factors These phases were applied to a forest property located in northern Italy, and some improvements were proposed simulating a manager’s approach during planning. The results showed interesting features in accessibility evaluation, which identified three different classes of accessibility, represented in a map. The estimation of Road Needs Index assigned a class regarding road requirements to each forest management unit: »low«, »medium«, »high« and »very high«. This information was merged, becoming a useful tool to identify the forest areas with the highest problems in relation to the forest road network. Keywords: forest management, GIS, Analytic Hierarchy Process, accessibility, road needs, Decision Support System

1. Introduction A sustainable perspective of forest management cannot ignore careful and accurate forest road network planning (Çalişkan 2013, Hippoliti 1997, Hippoliti 1976). A forest road network traditionally ensures access to forests and grazing lands, allowing forest operations and other productive activities. In the last decades, the increasing importance of the multi-functionality of forests has highlighted the key role of forest road management for tourism and recreational tasks (Gumus et al. 2008, Chirici et al. 2003). Moreover, forest Croat. j. for. eng. 37(2016)2

roads allow access to remote areas in case of natural hazards (Enache et al. 2013) and are a fundamental infrastructure in extinguishing forest fires (Hayati et al. 2012). Construction characteristics should be different, depending on the kind of machines that are expected to run on different road branches. In particular, width, slope and radius of curvature are the most important elements for forest roads that can limit vehicles trafficability (i.e. dimensions and payload of vehicles). The quality of roads is related to building and maintenance quality, in terms of both techniques and materials, and it can vary during road lifespan (Kiss et al. 2015). Con-

319

A. Laschi et al.

A Methodological Approach Exploiting Modern Techniques for Forest Road Network Planning (319–331)

sidering these aspects, an accurate road network plan is mandatory to allow the best efficiency and cost effectiveness of forest roads for all forest activities. Despite the essential role of forest roads in forest management, there are several potential negative effects in relation to these infrastructures. Despite the reduced width and excavation volume of a forest road in comparison with public roads, several environmental impacts relating to its construction, maintenance and use should affect this infrastructure (Trombulak and Frissell 2012, Avon et al. 2010, Demir 2007, Delgado et al. 2007), especially taking into account the natural context where it is located. However, these impacts can be reduced thanks to accurate planning and management (Akbarimehr and Naghdi 2012). Furthermore, forest roads should also be considered as ecosystems with an active role in the forest environment, which is not necessarily negative (Lugo and Gucinski 2000). For these reasons, a well-designed and well-developed road network plan must support the forest management plan to permit the best maintenance and enhancement of the road network, focused on the real management needs, through an integrated information exchange. Taking into account different interests related to forests, many analyses under various viewpoints have to be considered during planning. The main aspects under consideration should regard technical, economic and environmental issues. A technical approach identifies the strengths and weaknesses in the current road network to permit the best application of the management plan. Economic implications always have fundamental impacts; in the forest sector, the construction of a road is a significant cost (Ghajar et al. 2012, Samani and Hosseiny 2010) and accurate planning permits optimization of the cost effectiveness of road network management and enhancement. There are some fundamental steps that characterise a well-developed forest road network plan. Þ a complete knowledge of the actual conditions of each road segment in the entire network examined, both in terms of construction characteristics and maintenance conditions Þ a cautious evaluation of the actual accessibility state of different forest areas Þ an evaluation concerning the real needs of forest roads i.e. at management unit level in different sub-areas of the managed area, considering all of the functions provided by the analysed forest, such as wood production, hydrogeological protection, nature conservation, tourist interests and landscape tasks The best system in forest management planning, where different needs have to be considered, is the

320

development of a Decision Support System (DSS). In the last few years, this system has made possible an organised and integrated overview of relevant parameters related to forest functions, helping forest managers in decision processes (Vacik and Lexer 2014). In this context, considering the huge number of variables to be considered to represent the main interests related to forest multi-functionality, the approach of a MultiCriteria Analysis (MCA) is recommended (Çalişkan 2013, Sacchelli et al. 2013). Moreover, the Analytic Hierarchy Process (AHP) (Saaty 1980) has been, and continues to be, one of the most common DSSs in defining the priority of different parameters considered, organising them in a hierarchy (Çalişkan 2013). Furthermore, Geographical Information Systems (GISs) play a key role in managing and displaying terrestrial data for spatial forest planning (Zeki Baskent and Keles 2005) and specifically also in forest road planning (Najafi and Richards 2013, Hayati et al. 2012, Mohtashami and Bergkvist 2012, Dean 1997). These approaches and technologies are widely used in forest planning under a sustainable perspective (Ducey and Larson 1999, Miettinen and Hämäläinen 1997) with a huge number of studies based on multi-criteria and hierarchical approaches, also on the specific topic of forest road planning (Pellegrini et al. 2013, Ghajar et al. 2012, Samani and Hosseiny 2010). Because of the integration between terrestrial data and information included in the forest management plan, a valuable road network plan, which takes into account the most important aspects of road management, can be developed. With such information, it is possible to make informed choices in road maintenance, enhancement and building. To summarise, short-term needs should not be the main influence in the construction of a forest road, but the intervention should be integrated in mid/longterm planning, to maximise the related functionality and economic benefits, minimising the environmental impacts. The aim of this work was the development of an MCA integrating GIS and AHP to obtain a DSS with reliable information regarding accessibility and road needs, in a case study area in northern Italy. Moreover, an evaluation of the obtained results was performed; the results were used in simulating a forest manager’s interpretation and identifying weaknesses in the road network. Hypotheses for enhancement and improvements in the analysed forest road network were defined and suggested to the managers.

2. Materials and methods The study area was Paneveggio forest, an alpine public property located in the Autonomous Province of Trento, in northern Italy. The property has a total Croat. j. for. eng. 37(2016)2

A Methodological Approach Exploiting Modern Techniques for Forest Road Network Planning (319–331)

Table 1 Road classification following Italian standards Road for trucks Main

Secondary

Road for tractors

Road Minimum width Prevalent

3.5 m

3.0 m

2.5 m

5–6 m

4–5 m

3–4 m

Optimum

3–8%

Max. average

10%

12%

14%

Max. for short stretches

14%

18%

20 (25)%

Max. counter slope*

10%

12%

14%

Minimum curvature radius in hairpin turns

10 m

Slope

* counter slope is defined as a grade that is opposite to the general running grade of a road, i.e. the slope of the road section/s that tractors or trucks have to cover loaded

area of 4300 ha, of which 2803 ha are forests. Paneveggio is entirely included in a mountainous area above 1300 m above sea level. Characteristic tree species are spruce (Picea abies (L.) H. Karst), larch (Larix decidua Mill.) and Swiss stone pine (Pinus cembra L.). A closeto-nature silviculture is applied in this area and the average annual yield is 6000 m3 y–1 of roundwood. Paneveggio is included in the »Provincial Natural Park of Paneveggio and Pale di San Martino«. The study was divided into three main parts to organise data collection and processing in sequential steps: Þ forest road network analysis Þ accessibility analysis Þ evaluation of forest road needs These steps are described and illustrated below.

2.1 Forest road network analysis At this stage, the objective was to collect all useful information regarding characteristics of the network, to implement an up-to-date database with geographical information, construction characteristics and maintenance level of each forest road. Each forest road was surveyed by car and several types of information were collected. With a set of basic instruments and a portable GPS, a survey form was filled in with the following main information: co-ordinates and altitude of starting and ending points, total length, maximum and average slope, prevalent and minimum width, minimum curvature radius of hairpin turns. Moreover, accessory elements such as water-bars, cross-drainage culverts, etc. were counted and described. For each road, a general description, an evaluation of maintenance condiCroat. j. for. eng. 37(2016)2

A. Laschi et al.

tions and the identification of criticisms were added to the form and then to the dataset. All of the collected information was implemented on a GIS platform (both ArcGIS 9.3 and QGIS 2.10 software were used) and a shape file with up-to-date information for each road was obtained. Roads were represented as lines, merged by nodes in each road intersection. A Digital Terrain Model (DTM) and orthophotographs were helpful during graphic representation of forest roads to correct the errors in accuracy made by the portable GPS during the field survey. Because of the dimensional characteristics collected, it was possible to classify each road following the Italian classification suggested by Hippoliti (1976), which is related to the kind of machine that could drive on the road. Characteristics of the roads are reported in Table 1. All of the collected information was organised in a specific geo-referred database. The total length and road density (m ha–1) were calculated.

2.2 Accessibility analysis In this phase, an in-depth analysis of forest accessibility was developed by means of GIS, thanks also to the information collected and organised in the first step of this work. First of all, it was necessary to define the criteria of forest accessibility in relation to forest roads. The method based on the access time, suggested by Hippoliti (1976), was applied in detail. This method has been applied in other studies (Grigolato et al. 2013, Cavalli and Grigolato 2009, Chirici et al. 2003), because it is functional in mountainous conditions, and it takes into consideration terrain slope variations. Hippoliti (1976) defined three different accessibility classes based on the time required for a forest worker to make a round trip on foot from the nearest road to a given point in the forest. This time was called the »access time«. Hippoliti assumed an average walking speed for a worker in forest of 4 km h–1 on flat terrain, and a walking speed of 400 ma h–1 on steep terrain, where ma is the differential levelling in metres. When the access time is up to 30 minutes (for going in the morning and coming back at the end of workday), the area is classified as »served«, while when it is between 30 minutes and two hours, it is »barely served«. When a worker spends more than two hours reaching the work site, the area is considered as »not served« by the forest road network. This concept was introduced in the ‘70s, when the manpower price in Italy was »very low«. Nowadays, considering the cost of manpower, it is not economically viable to spend two hours to reach the work site. For that reason, in this work Hippoliti’s approach was modified in time ranges, reducing the maximum access time to one hour for

321

A. Laschi et al.

A Methodological Approach Exploiting Modern Techniques for Forest Road Network Planning (319–331)

Table 2 Accessibility categories based on a revised definition of Hippoliti Accessibility

Access time (return trip)

Distance in flat terrain (when slope<10%)

Differential levelling (when slope³10%)

»Served«

Up to 30'

Max. 1000 m

Max. 100 m

»Barely served«

Between 30' and 60'

Max. 2000 m

Max. 200 m

»Not served« More than 60' More than 2000 m More than 200 m

»barely served« areas. The revised accessibility characteristics are summarised in Table 2. Three different methods were implemented in the Paneveggio forest to calculate and graphically represent the degree of accessibility. The classical method, based on a field survey and manual representation on maps, was implemented to identify the second best application in terms of accuracy and reliability. The descriptions of each method are reported below. 2.2.1 »Method H« manual approach »Served« and »barely served« areas were manually drawn starting from each road branch and considering altitude gaps (100 m for »served« and 200 m for »barely served« areas). Once the draft accessibility map was made, a field survey permitted correction of the map identifying natural barriers that impede access to some areas, e.g. cliffs and deep gullies. Finally, the results were manually reported on a GIS platform to obtain a digital map, comparable with the results of the other methods. 2.2.2 »Method A« travel time This was the first GIS approach applied in this study for accessibility estimation. Hippoliti’s definition for accessibility could be described taking into account the travel time or travel distance. In this method, the travel time was considered to determine the accessibility class for each point in the forest property (Chirici et al. 2003). This operation required several steps and information. A DTM with a resolution of 5´5 m created from LIDAR data, and the road network map previously developed (see Section 2.1) were used as starting information. A »Cost Distance« tool was applied to calculate the accessibility degree of the forest surface. In particular, this tool is a procedure for determining least cost paths across continuous surfaces, typically using grid representations (de Smith et al. 2015). In this study, Cost Distance identified the surface, around each forest road, with a maximum cost in terms of time of walking transfer for a forest

322

worker. Therefore, several steps were made to obtain a »cost-map« including the information concerning the crossing time for each pixel, which strictly depends on the slope gradient: Þ slope map creation, starting from the DTM, in raster format containing the slope gradient in % for each pixel, p% Þ differential levelling (d), covered for each pixel, was calculated considering a horizontal distance (d0) of 6 m (the average of the side of the pixel, 5 m, and the diagonal, 7 m), using the formula: differential levelling = d =

p% ⋅ d0 100

(1)

Þ »d« was reclassified with high value (900 m) to exclude extreme slopes (>100%), that could not be crossed by walking. For slopes lower than 10%, a fixed value of 0.6 m was assigned to obtain efficient results for flat terrains, too Þ considering the climbing speed of 400 ma h–1, corresponding to 0.1 ma s–1 (tu), a raster containing information regarding the time of pixel crossing was made by the following equation: pixel crossing time = tpc =

d d = – tu 0.1

(2)

Þ starting from the vector file of the forest road network, a raster file representing presence/absence of roads was created Þ the Cost Distance tool was applied to define the »served« and »barely served« areas, starting from the raster of road presence and the costmap previously created, applying the time limits as maximum cost (15 minutes and 30 minutes for »served« and »barely served« areas, respectively) 2.2.3 »Method B« fraction of the maximum distance In the previous method, »Cost Distance« was applied considering the travel time. In this case, a method that interprets the accessibility definition considering the travel distance was developed. As explained in Section 2.2, on flat terrain we considered the linear distance covered, while on a slope we considered the change of altitude between the starting point at the roadside and the work point in the forest. A DTM of 5´5 m in resolution guaranteed the morphological information. The Cost Distance procedure was also applied in this case, and several steps were necessary: Þ a slope map (p%) based on DTM was created Þ the slope map was reclassified, as in »method A«, to obtain a unique »cost-map« efficient on both slope and flat terrain. Therefore, the values Croat. j. for. eng. 37(2016)2

A Methodological Approach Exploiting Modern Techniques for Forest Road Network Planning (319–331)

»10« were assigned to all the pixels characterised by a slope up to 10%, and the value »99999«, to all the pixels with a slope higher than 100% Þ a »weight« was given to each pixel, which was determined as the ratio between the maximum distance (dmax) on flat terrain (1000 m) and the horizontal distance (Df(p%)) corresponding to the maximum differential levelling (dmax) (»served« = 100 m, »barely served« = 200 m) on a terrain with the slope of the pixel taken into account. Finally, the formulas for weight calculation were the following: weight = w =

Dmax 100 ⋅ dmax , Df (p ) = % Df (p ) p%

(3)

Þ the final cost-map was obtained by multiplying the weight value by the pixel size Þ starting from the vector file of the forest road network, a raster file representing presence/absence of roads was created Þ starting from the road raster (presence/absence) and the cost-map previously created, applying the distance limits as maximum cost (1000 m and 2000 m for »served« and »barely served« areas, respectively), the Cost Distance tool was applied defining the »served« and »barely served« areas 2.2.4 »Method C« fixed distance buffer This is an expeditious approach, which did not consider Hippoliti’s definition of accessibility. It was implemented to compare a quick method with the other models described. It simply drew a buffer at a fixed distance from each road branch. Several examples of this accessibility evaluation have been developed in other studies, considering different distances in relation to the type of machines and techniques applied. Some studies set a unique limit value, e.g. 300 m without slope considerations (Bååth et al. 2002), and 3 km but limited to the first slope class (max. 20%) (López-Rodríguez et al. 2009). Moreover, in other studies, different classes with a varying distance were considered, comparing buffers within 150, 200 and 250 metres from the road (Pentek et al. 2008). In this study, two buffers were applied, the first in a range of 100 m from the road and the second in a range of 300 m. The first corresponds to the »served« area, which was identified with the operative distance generally considered for extraction with tractor and winch. The second one was assumed as the »barely served« area, which corresponds to the operative distances of a lightweight cable-yarder. Croat. j. for. eng. 37(2016)2

A. Laschi et al.

Finally, a comparison between the different methods was made. »Method H« was considered as »truth« to have a reference for comparison.

2.3 Evaluation of forest road needs Thanks to the previous analysis, the accessibility degree for all forest management units was available. This was important information, but it needed to be integrated with another analysis, in a typical MCA perspective, to obtain a complete set of information useful for roads management. In particular, it was fundamental to support accessibility information with the real needs of roads for each forest management unit. The »needs of roads« (or »needs of accessibility«) aimed to identify the accessibility requirements in different areas, concerning different characteristics of forest management units and the related silvicultural prescriptions applied. Effectively, forest located at an altitude close to the upper tree line, with low tree growth and managed under a protective perspective, generally requires a lower accessibility level than others located in a more fertile area, with high yield, and managed under a productive perspective. In this context, a GIS-based method for evaluating the forest needs in terms of accessibility was developed. It was assumed that forest operations would be the forestrelated activities characterised by the highest accessibility requirements, compared with other services provided by the forest, and activities carried out inside it. Therefore, the methodology was developed considering factors related to forest productivity. On the basis of Paneveggio’s forest management plan, a set of data was chosen as factors useful for determining the forest-road needs. Three categories were identified as more suitable than the others considering their reliability and the impact on this study: growing stock (GS), site fertility class (FC) and productive potential index (PPI) for high forests. Other factors were discarded for specific reasons, e.g. the annual volume increment, which should be a good indicator characterising a forest management unit in terms of productive capacity. In our case, the increment information presented some out-of-scale values, presumably attributable to mistakes in reporting, which did not guarantee a good reliability level. In particular, the three cited factors were chosen because: Þ GS represented a reference point in a short-term productivity perspective Þ FC, which was divided into nine classes, from the most fertile (class »1«) to the least fertile (class »9«), assigned to each forest management

323

A. Laschi et al.

A Methodological Approach Exploiting Modern Techniques for Forest Road Network Planning (319–331)

unit. It was chosen for representing a fixed condition along the time regarding the production capacity of each management unit Þ PPI, this was a classification made by a pool of experts regarding the whole Province of Trento. Following different parameters, including the main tree species, hydrogeological risks and environmental significance, they assigned to forest sub-areas a synthetic value representative of their productive potential. This index was divided into nine classes, from the lowest productive potential (class »1«) to the highest (class »9«). This factor was chosen because it was relevant from a long-term management perspective. To calculate a »road needs index« (RNI), after normalization, the factors were merged by means of a weighted sum. To assign a »weight« to each factor, an Analytic Hierarchy Process (AHP) was applied to the results of local forest expert interviews. Each factor was normalised (in a range between 0 and 1) following the processes explained below: »Growing Stock« (GS): considering the distribution of GS values, we applied an exponential function for GS values up to 700 m3 ha–1 y–1, with increasing values from 0 to 1; for GS values higher than 700 m3 ha–1 y–1, the normalised value assigned was always 1: GSnorm

 −0.00000198835 GS 2 + 0.00280713 GS + 0.0093 , for 0 ≤ GS ≤ 700  =  1 , for GS > 700 

(4)

»Fertility Class« (FC): forest management unit distribution in fertility classes showed a reduced frequency in the most fertile classes. For this reason, this factor was normalised assigning value 1 to the first three classes, and a linear regression was applied for the classes from the fourth to the ninth, with decreasing values from 1 to 0:

FC norm

 1 3  − 6 FC + 2 , for FC > 3  =  1 , for 1 ≤ FC ≤ 3  

(5)

»Productive Potential Index« (PPI): a linear regression was applied assigning increasing values from the lowest potential to the highest, in the range 0 to 1: PPI norm =

1 1 PPI − 8 8

(6)

Sixteen forest experts, including forest managers and Provincial Forest Service members, were interviewed to collect their personal opinion on the importance of the three factors previously described regarding forest management, with special attention on the topic of forest roads. In particular, a compilation of questions, based on a comparison of the three factors, was submitted to the experts by a personal interview. Following AHP concepts, personal opinions were implemented by Expert Choice® software to obtain an impartial result on the importance of analysed factors for forest road management. The software compared the answers given by people interviewed and provided the »weight« value in% for each factor. Finally, one raster file was calculated for each factor involved in the analysis. The RNI map was calculated by the weighted sum between GSnorm, FCnorm and PPInorm: RNI = wGS ´ GSnorm + wFC ´ FC norm + wPPI ´ PPI norm

(7)

where wGS, wFC and wPPI were the weights obtained for GS, FC and PPI, respectively. The result was a map of the Paneveggio forest in which each pixel was classified in relation to accessibility requirements. The Paneveggio forest was divided into four road needs classes: »Low«: with RNI < 0.25, »Medium«: with 0.25 ≤ RNI < 0.50, »High«: with 0.50 ≤ RNI < 0.75, »Very high«: with RNI ≥ 0.75.

324

Croat. j. for. eng. 37(2016)2

A Methodological Approach Exploiting Modern Techniques for Forest Road Network Planning (319–331)

A. Laschi et al.

Fig. 1 Paneveggio forest road network

3. Results 3.1 Forest road network analysis The first step of this study collected information regarding the state of the forest road network. A shape file was created and the road network is represented in Fig. 1. In the related database, all of the information collected and cited in Section 2.1 was recorded. The total length of road network within the forest area (2803 ha) was 37.1 km, corresponding to a density of 13.2 m ha–1, while in the overall property (4366 ha) there were 52.4 km of roads, with a density of 12 m ha–1. According to the Italian classification, the majority of forest roads (69% of total, 36 km) were classified as »Road for tractor«, while 25% were classified as »Secondary road for truck« and only 6% were included in the category »Main road for truck«.

developed to evaluate the correspondence rate. In particular, the maps were overlapped and compared, identifying the percentage of pixels with the same accessibility class assigned (correspondence) and the others, where the model assigned different classes. The results showed that correspondences of 83.1%, 87.2% and 51.5% were calculated for methods A, B and C, respectively, using »Method H« as the reference.

3.3 Evaluation of forest road needs The AHP results defined the values for the »weights« to be used in the RNI calculation. In parTable 3 Surface distribution in accessibility classes identified by means of four analysed methods »Served«

3.2 Accessibility analysis

»Surface«

The four methods described in Section 2.2 were implemented and compared. Results regarding the surface distribution between the three accessibility classes are reported in Table 3. A comparison between »Method A« (Fig. 2), »Method B« (Fig. 3) and »Method C« (Fig. 4), considering »Method H« (Fig. 5) as reference, was Croat. j. for. eng. 37(2016)2

»Barely served«

»Not served«

»Method H«

56.2

1574

20.0

562

23.8

667

»Method A«

54.7

1534

21.2

594

24.1

675

»Method B«

60.2

1689

21.4

599

18.4

515

»Method C«

26.8

750

34.2

960

39.0

1093

325

A. Laschi et al.

A Methodological Approach Exploiting Modern Techniques for Forest Road Network Planning (319–331)

Fig. 2 Results of »Method A« applied to Paneveggio forest

Fig. 3 Results of »Method B« applied to Paneveggio forest

326

Croat. j. for. eng. 37(2016)2

A Methodological Approach Exploiting Modern Techniques for Forest Road Network Planning (319–331)

Table 4 Forest area distribution based on RNI. For each RNI class, the distribution of the area in each of the current accessibility classes is shown At-present accessibility class Total

Road Needs Index

»Low«

»Medium«

»High«

»Very High«

»Served«

»Barely served«

»Not served«

75.7

0.0

22.4

53.3

2.7

0.0

0.8

1.9

765.2

185.0

238.3

341.9

27.3

6.6

8.5

12.2

1202.5

857.7

274.7

70.1

42.9

30.6

9.8

2.5

759.6

644.7

64.5

50.4

27.1

23.0

2.3

1.8

ticular, the elaboration of experts’ answers showed that the most important factor was the GS, with a weight of 45%, while weights of 30% and 25% were determined for FC and PPI, respectively. The final for-

A. Laschi et al.

mula applied to each forest management unit in calculating the RNI was: RNI = 0,45 ´ GSnorm + 0,30 ´ FC norm + 0,25 ´ PPI norm

(8)

Summarised information on RNI distribution of the total surface, also in relation with the current accessibility results, is reported in Table 4. Fig. 6 gives a graphical representation of the RNI results, distributed following four classes established in Section 2.3. Overall results were interpreted to identify the areas in terms of road accessibility. Taking into account all of the results, two new road branches were proposed. The two hypotheses are reported in Fig. 7.

4. Discussion Due to different analyses performed, different topics were investigated in this study and useful information was obtained to facilitate and improve the decision processes related to forest road network planning. Analysis of the existing road network highlighted the predominance (69%) of forest roads only suitable for the transit of tractor and trailer, while 25% were roads with better characteristics in terms of width, radius of

Fig. 4 Results of »Method C« applied to Paneveggio forest Croat. j. for. eng. 37(2016)2

327

A. Laschi et al.

A Methodological Approach Exploiting Modern Techniques for Forest Road Network Planning (319–331)

Fig. 5 Results of »Method H« applied to Paneveggio forest

Fig. 6. Distribution of forest management units in classes depending on RNI

328

Croat. j. for. eng. 37(2016)2

A Methodological Approach Exploiting Modern Techniques for Forest Road Network Planning (319–331)

A. Laschi et al.

Fig. 7 Proposals for two new forest roads curvatures and slope; only 6% of roads could allow the transit of heavy trucks. This information should suggest the enhancement of the main forest roads to improve and rationalise wood transport by means of heavy trucks instead of tractors and trailers. The slope is a crucial factor in accessibility estimation. In mountainous areas, the slope heavily influences the accessibility (Cavalli and Grigolato 2009) and strongly affects the logging system and costs of forest operations (Hippoliti 1997). In this study, the accessibility analysis added essential information regarding the rationality in forest roads distribution around the forest property. Comparing the applied methods, »Method C« showed the worst performance, in terms of correspondence, in comparison with results obtained by »Method H«, which was taken as the reference. »Methods A« and »B« had similar responses; »Method A« obtained better results than »Method B« considering the general surface percentage allocated to the three different categories, which were very similar in value to »Method H«. On the other hand, »Method B« showed better accuracy than »Method A«, considering the correspondence in pixel allocation. This similarity was expected, considering the similar approach used in the two methods. However, accuracy was considered as the main objective in identifying the best method for Croat. j. for. eng. 37(2016)2

accessibility estimation, and »Method B« had the better performance in this sense. It showed a »high« rate of »served« management units in Paneveggio forest (about 60% of the total) and a limited »not served« area (18%). In this context, introducing the evaluation of the real needs of forest roads in different areas, especially regarding »barely« and »not served« areas, could give an added value to the roads planning process according to the multi-criteria approach. As reported in Table 4, »not served« areas mainly referred to forest management units with a »low« or »medium« RNI, while only a limited area with »high« and »very high« needs was »not served«. Regarding »barely served« management units, there was a reasonable rate of forest with »medium« and »high« RNI. In this case, an enhancement of the road network should improve accessibility conditions, optimising productivity in forest operations. Regarding the three factors analysed in the AHP, technicians focused their main interest on growing stock (weight: 45%), which is a factor influencing short-term decisions. Fertility class had an intermediate weight (30%), thus highlighting the attention of forest technicians also on long-term aspects, while the productive potential index (25%) received a quite low weight. In practice, RNI analysis took into account a balanced mix of information aimed both to short and long-term

329

A. Laschi et al.

A Methodological Approach Exploiting Modern Techniques for Forest Road Network Planning (319–331)

approaches for forest road network planning. However, the method could be improved through the implementation of more, or different, factors; in particular, the increments of volume should be considered when available to obtain a representative parameter related to stand productivity. Moreover, in RNI analysis, different solutions or alternatives could be implemented to optimise the methods depending on management priorities. In the case of special needs, specific factors could be introduced into the RNI calculation formula, enlarging the analysis to specific topics. An example of that would be the use of indicators regarding wildfire risk, or tourism points of interest. In these cases, the RNI analysis could give useful information in addition to the ordinary needs, normally satisfied when the road network is functional for forest operations.

5. Conclusion In this study, a methodological approach aimed to improve forest road network planning was developed and applied to a forest property. MCA and AHP criteria were applied by means of GIS, exploiting terrestrial information and forest characteristics. The state and accessibility requirements in a forest area have key roles in optimising road network management. The analysis developed in this study could be a real added value for managers during planning. In particular, the examined methods facilitate objective forest road planning, which permits optimisation of resources and avoids the building of a useless road and/ or oversized road network. Enhancement of the forest road network, through well-defined improvements, guarantees the correct management of the forest, allowing all of the forest services, such as wood production, hydrogeological protection, tourism use and habitat conservation. In practice, a well-planned and well-realised road network maximises the efficiency of all forest activities, minimising the costs, both economically and environmentally. The approach described and achieved in this study perfectly fit in this perspective, reporting fundamental information useful for sustainable maintenance and development of the forest road network.

6. References Akbarimehr, M., Naghdi, R., 2012: Reducing erosion from forest roads and skid trails by management practices. Journal of Forest Science 58(4): 165–169. Avon, C., Bergès, L., Dumas, Y., Dupouey, J.L., 2010: Does the effect of forest roads extend a few meters or more into

330

the adjacent forest? A study on understory plant diversity in managed oak stands. Forest Ecology and Management 259(8):1546–1555. Bååth, H., Gällerspång, A., Hallsby, G., Lundström, A., Löfgren, P., Nilsson, M., Ståhl, G., 2002: Remote sensing, field survey, and long-term forecasting: An efficient combination for local assessments of forest fuels. Biomass and Bioenergy 22(3): 145–157. Çalişkan, E., 2013. Planning of forest road network and analysis in mountainous area. Life Science Journal 10(2): 2456–2465. Cavalli, R., Grigolato, S., 2009: Influence of characteristics and extension of a forest road network on the supply cost of forest woodchips. Journal of Forest Research 15(3): 202–209. Chirici, G., Marchi, E., Rossi, V., Scotti, R., 2003: Analisi e valorizzazione della viabilità forestale tramite GIS: la foresta di Badia Prataglia (AR). L’Italia Forestale e Montana 6: 460–481. Dean, D.J., 1997: Finding optimal routes for networks of harvest site access roads using GIS-based techniques. Canadian Journal of Forest Research 27(1): 11–22. 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 and Urban Planning 81(4): 328–340. Demir, M., 2007: Impacts, management and functional planning criterion of forest road network system in Turkey. Transportation Research Part A: Policy and Practice 41(1): 56–68. Ducey, M., Larson, B., 1999: A fuzzy set approach to the problem of sustainability. Forest Ecology and Management 115(1): 29–40. Enache, A., Ciobanu, V.D., 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. Ghajar, I., Najafi, A., Ali Torabi, S., Khamehchiyan, M., Boston, K., 2012: An Adaptive Network-based Fuzzy Inference System for Rock Share Estimation in Forest Road Construction. Croatian Journal of Forest Engineering 33(2): 313–328. Grigolato, S., Pellegrini, M., Cavalli, R., 2013: Temporal analysis of the traffic loads on forest road networks. iForestBiogeosciences and Forestry 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. Environmental Monitoring and Assessment 142(1–3): 109–116. Hayati, E., Majnounian, B., Abdi, E., 2012: Qualitative evaluation and optimization of forest road network to minimize total costs and environmental impacts. iForest – Biogeosciences and Forestry 5(3): 121–125. Croat. j. for. eng. 37(2016)2

A Methodological Approach Exploiting Modern Techniques for Forest Road Network Planning (319–331) Hippoliti, G., 1997: Appunti di meccanizzazione forestale, Firenze: Società Editrice Fiorentina. Hippoliti, G., 1976: Sulla determinazione delle caratteristiche della rete viable forestale. Italia Forestale e Montana – Italian Journal of Forest and Mountain Environments 6: 242–255. Kiss, K., Malinen, J., Tokola, T., 2015: Forest road quality control using ALS data. Canadian Journal of Forest Research 45(11): 1636–1642. López-Rodríguez, F., Atanet, C.P., Blázquez, F.C., Celma, A.R., 2009: Spatial assessment of the bioenergy potential of forest residues in the western province of Spain, Caceres. Biomass and Bioenergy 33(10): 1358–1366.

A. Laschi et al.

Pentek, T., Pičman, D., Nevečerel, H., Lepoglavec, K., Poršinsky, T., 2008: Road network quality of the management unit Piščetak – GIS analysis. In Proceedings of the 3rd international scientific conference FORTECHENVI 2008, 45–53. Saaty, T.L., 1980: The analytic hierarchy process: planning, priority setting, resources allocation. New York: McGraw. Sacchelli, S., De Meo, I., Paletto, A., 2013: Bioenergy production and forest multifunctionality: A trade-off analysis using multiscale GIS model in a case study in Italy. Applied Energy 104: 10–20.

Lugo, A.E., Gucinski, H., 2000: Function, effects, and management of forest roads. Forest Ecology and Management 133(3): 249–262.

Samani, K., Hosseiny, S., 2010: Planning road network in mountain forests using GIS and Analytic Hierarchical Process (AHP). Caspian Journal of environmental sciences 8(2): 151–162.

Miettinen, P., Hämäläinen, R.P., 1997: How to benefit from decision analysis in environmental life cycle assessment (LCA). European Journal of Operational Research 102(2): 279–294.

de Smith, M.J., Goodchild, M.F., Longley, P.A., 2007: Geospatial Analysis. A comprehensive guide to principles. techniques and software tools. Leicester, UK: Matador, Troubador Publishing Ltd.

Mohtashami, S., Bergkvist, I., 2012: A GIS Approach to Analyzing Off-Road Transportation: a Case Study in Sweden. Croatian Journal of Forest Engineering 33(2): 275–284.

Trombulak, S.C., Frissell, C.A., 2012: Review of Ecological Effects of Roads on Terrestrial and Aquatic Communities and of Aquatic Ecological Effects of Roads on Terrestrial Commnunities 14(1): 18–30.

Najafi, A., Richards, E.W., 2013: Designing a Forest Road Network Using Mixed Integer Programming. Croatian Journal of Forest Engineering 34(1): 17–30. 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.

Vacik, H., Lexer, M.J., 2014: Past, current and future drivers for the development of decision support systems in forest management. Scandinavian Journal of Forest Research 29(1): 2–19. Zeki Baskent, E., Keles, S., 2005: Spatial forest planning: A review. Ecological Modelling 188(2–4): 145–173.

Authors’ address:

Received: February 1, 2016 Accepted: April 4, 2016 Croat. j. for. eng. 37(2016)2

Andrea Laschi, PhD. * e-mail: andrea.laschi@unifi.it Francesco Neri, PhD. e-mail: francesco.neri@unifi.it Niccolò Brachetti Montorselli, PhD. e-mail: n.montorselli@gmail.com Prof. Enrico Marchi, PhD. e-mail: enrico.marchi@unifi.it University of Florence GESAAF – Department of Agricultural, Food and Forestry Systems Via San Bonaventura 13 50145 Florence ITALY * Corresponding author

331

Original scientific paper

Optimum Utilization of Rice Husk Ash for Stabilization of Sub-base Materials in Construction and Repair Projects of Forest Roads Mehran Nasiri, Majid Lotfalian, Amir Modarres, Wei Wu Abstract Forest roads play an important role in forest management, timber transportation and forest protection. However, minimum standards are considered for pavement materials due to the traffic volume and economic situation of forestry projects. Therefore, this paper aims to (a) evaluate the role of Rice Husk Ash (RHA) as a soil stabilizer of sub-base layer to improve the quality of materials and (b) determine the optimum utilization among 10 combinations of soil, lime and RHA regarding the environmental factors. The results of laboratory studies on soil A–6 (AASHTOO classification) indicates a general decrease in the maximum dry density (MDD) and an increase (21.9%) in optimum moisture content (OMC) with increase in RHA content. Adding RHA (9%) causes a decrease (13.3%) in liquid limit and plasticity index (PI) of soil. However, this improving effect is not as much as the influence of lime. The California bearing ratio (CBR) of stabilized soil in both saturated condition and optimum water content was 28% and 37.5% more than the natural soil, respectively. The maximum unconfined compressive strength (UCS) values were recorded for 9% RHA, 237 KN m-2 after 28 days curing time, which was 23 KN m-2 more than the natural soil. According to the results, the combination of Soil+4% Lime+9% RHA could be used as the optimum consumption of materials for stabilization of sub-base layer in construction of forest roads. Keywords: forest road, stabilization, rice husk ash (RHA), CBR, UCS, OMC, MDD

1. Introduction Forest roads are designed under substantially limited conditions. In most cases, minimum standards are considered for layers due to the traffic volume and economic situation of forestry projects. The most important pavement layer of forest roads (access roads) is sub-base layer, since the construction of base layer on forest roads is not justified according to its standards. Several types of sub-base materials can be used considering the type of materials in borrow area, weather conditions, number of traffic and economic situation. Usually, borrow materials are used for the construction of forest roads. Experiences have revealed that according to AASHTOO classification (AASHTO 1986), the materials used for the access roads are usually placed in classes of A–4, A–5, A–6 Croat. j. for. eng. 37(2016)2

and A–7. These soils generally contain silt and clay and are classified in bad to average classes in terms of road construction (AASHTO 1986). To have the required strength against tensile stresses and strains spectrum, the materials used for constructing sub-base layer should have suitable specification. The use of soil stabilizers is one of the ways to improve the mechanical specification of these soils (Nair et al. 2008, Jamil et al. 2013). Lime has been used several times in forest road projects as a stabilizer (Péterfalvi et al. 2015). The reaction between soil and lime is performed very slowly. In addition, in the presence of organic matters in forest road materials, pure lime cannot be a suitable stabilizer. However, the extensive use of lime is harmful to the environment and roadside vegetation in the area with non-calcareous

333

M. Nasiri et al. Optimum Utilization of Rice Husk Ash for Stabilization of Sub-base Materials in Construction ... (333â&#x20AC;&#x201C;343)

bedrock. Due to limitations in the use of stabilizers such as cement in forest roads (Khan et al. 2012), rice husk ash (RHA) with a huge amount of silica and high specific surface could play a major role in soil stabilization method with lime (Weiting et al. 2012). Given that RHA is not suitable for cattle feeding and it is also non-biodegradable, it could be considered as an inexpensive stabilizer. Low cost and availability as well as the stabilization specification of RHA have led many researchers to investigate RHA as an alternative for soil stabilization (Basha et al. 2005, Jha and Gill 2006, Onyango et al. 2007, Nair et al. 2008, Ramezanianpour et al. 2009, Chobbasti et al. 2010, Sabat and Nanda

2011, Harichane et al. 2011, Milani et al. 2012, Weiting et al. 2012, Trivedi et al. 2013, Jamil et al. 2013). Onyango et al. (2007) in Tanzania used the natural pozzolan as the road pavement stabilizer. He showed that pozzolan was one of the most appropriate stabilizers in low-volume road and that it was economically effective. Chobbasti et al. (2010) showed that RHA was effective in reducing the liquid and plastic limits of soil. Alhassan (2008) reported that the Unconfined Compressive Strength (UCS) would improve with the use of RHA and the highest UCS was measured by 6â&#x20AC;&#x201C;8% in the combination of soil and RHA. The values of optimum moisture content (OMC) and

Fig. 1 Map and geographical location of the study area

334

Croat. j. for. eng. 37(2016)2

Optimum Utilization of Rice Husk Ash for Stabilization of Sub-base Materials in Construction ... (333–343) M. Nasiri et al.

maximum dry density (MDD) were also reported in studies of Basha et al. 2005 and Trivedi et al. 2013. They concluded that the OMC increased and MDD reduced by increasing in RHA percent. Several studies have examined the California Bearing Ratio (CBR) of soil (Jha and Gill 2006, Chobbasti et al. 2010, Purbi Sen et al. 2011, Sabat and Nanda 2011). They stated that the addition of lime to the soil had quickly affected the soil CBR, which had increased over time. By conducting a CBR test, Sabat and Nanda (2011) showed that the addition of mentioned materials to the soil had increased its strength by 20% compared to the natural soil samples. By further investigations, it was found that most of the conducted studies were performed for urban and rural roads and important environmental factors have not been examined. The amount of lime used in some of these studies was very high and reached an amount of 10%, while in this study, the environmental issues i.e. the use of less lime and finding a combination that improves the mechanical properties of soil were considered. Therefore, this paper aims to (a) evaluate the role of Rice Husk Ash (RHA) as a soil stabilizer of the sub-base layer and (b) determine the optimum utilization among 10 combinations of soil, lime and RHA regarding the environmental factors.

2. Materials and Method 2.1 Study Area This study was conducted in a road of Caspian forest (Azarrood basin), north of Iran (Fig. 1). The latitude and longitude of Azarrood basin are 36°08′5″ N and 52°45′58″ E, respectively. In this region, there is a moderate mountainous climate with cold winters and humid summers. The forest altitude ranges from 360 to 1490 meters above sea level and average annual precipitation is 800 mm. Alborz earth dam with a height of 72 m was constructed in this area (Latifi et al. 2012). Therefore, due to the earth dam and soil erosion, one of the most important solutions to prevent sediment delivery from forest roads and to strengthen the road materials is the use of stabilizers.

2.2 Identification of Soil About 350 kg of soil were collected from the borrow area (excavation slopes) near the Alborz Dam. Soils were put in 50 kg bags and transferred to the soil mechanics laboratory. In the laboratory, soils were remixed well together and used with respect to the objectives of this study. In order to identify the soil, tests such as sieve analysis and specific gravity (Gs) were Croat. j. for. eng. 37(2016)2

Table 1 Properties of the natural soil Laboratory tests

Rate

Liquid limit, %

39.4

Plastic limit, %

28.2

Plasticity index, %

11.2

Specific gravity, Gs

2.74

OMC, %

16.2

MDD, Mg m–3

1.54

–2

UCS, KN m

214

CBROWC, %

9.5

CBR saturated, %

7.5

AASHTO classification

A–6

USCS classification

done. For this purpose, the dried soil samples were placed in the topmost sieve and dry sieving was performed using a mechanical sieve shaker for a minimum period of 10 min. The specific gravity (Gs) of soil samples were determined using a pycnometer (ASTMD854) and the average of soil Gs was recorded 2.74. Also, the liquid and plastic limits of the soil were determined in order to study the Aterrberg limits and classify the soil. According to the AASHTO soil classification system, the studied soil is classified as A–6 soil and according to the Unified Soil Classification System (USCS), it is named SM (Silty Sand with Gravel). Table 1 shows more information about engineering properties of the studied soil. In order to stabilize the sub-base soil, at first rice husk and hydrated lime were prepared from farm fields of northern Iran and Alborz Industrial factory, respectively. The type of used lime is slaked lime and some of its characteristics are shown in Table 2. To make rice husk, rice husk must be burned at high temperatures (about 500°C) (Chobbasti et al. 2010). After burning, the obtained RHA were placed in the open air to complete the ash production process (Fig. 2b). Some characteristics of Tarom husk ash are shown in Table 3 (Chobbasti et al. 2010). Table 2 Characteristics of the used lime Non Hydrated CaO Non Hydrated MgO % % 1.62

0.8

CO2 %

Ca(OH)2 %

11.2

92.8

335

M. Nasiri et al. Optimum Utilization of Rice Husk Ash for Stabilization of Sub-base Materials in Construction ... (333–343)

Fig. 2 Laboratory tests to identify specification of different combinations (a) and produced RHA and rice husk (b) Table 3 Chemical characteristics of the Tarom husk ash (* 10-6) P, %

Mg2+, %

Ca2+, %

K, %

Na, %

SiO2, %

24.7

66.1

0.016

0.124

0.114

83.7

2.3 Mix Designs By examining different studies, it can be concluded that soil properties are one of the most important factors of mix design. Some studies have considered the value of 6% of lime consumption (Muntohar 2002); some other proposed the value of 10%. According to the objects i.e. the minimum use of lime, a combination of 4% and 6% was considered in this study. However, the addition of RHA varied in different studies. For example, Alhassan (2008) and Chobbasti et al. (2010) considered a maximum of 12% and 7% RHA for A–7–6 and A–4 soils, respectively. In this study, the RHA values of 5, 7, 9 and 12% were considered regarding the Liquid limit (LL) = 39.4, Plastic Limit (PL) = 28.2 and A–6 soil, respectively. Table 4 shows the mix design of materials.

336

2.4 Methods of Testing Soil mechanics tests were designed in order to evaluate the effect of stabilizers. So, materials were prepared according to the mix design and studied using tests of CBR, UCS (7, 14 and 28 days curing times), ATERBERG and Compaction. For a more accuracy, the number of replications for CBR, UCS, ATERBERG and Compaction test were considered 3 times (for each sample). The California Bearing Ratio test, or CBR-test, is an empirical test, which is used as an important criterion in pavement design. With this test, the bearing value of road sub-base and sub-grade can be estimated. In addition, it is one of the common tests for assessing the strength of stabilized soils. This test was conducted in accordance with the standard ASTM D1883-07(2002) in two conditions of saturation and optimum water content. According to the purpose of the study, this test can examine the strength of stabilized samples compared to the natural soil samples. Also, the proctor compaction tests were performed in accordance with the standard ASTM D069807E01(2002) in CBR mould; by conducting this test, it can be seen Croat. j. for. eng. 37(2016)2

Optimum Utilization of Rice Husk Ash for Stabilization of Sub-base Materials in Construction ... (333–343) M. Nasiri et al.

Table 4 Investigated mixtures and mentioned indices Mix design

Index

Soil

Soil+6% Lime

S+6L

Soil+4% Lime+5% RHA

S+4L+5R

Soil+4% Lime+7% RHA

S+4L+7R

Soil+4% Lime+9% RHA

S+4L+9R

Soil+4% Lime+12% RHA

S+4L+12R

Soil+6% Lime+5% RHA

S+6L+5R

Soil+6% Lime+7% RHA

S+6L+7R

Soil+6% Lime+9% RHA

S+6L+9R

Soil+6% Lime+12% RHA

S+6L+12R

how much free space of aggregates is filled by stabilizers and how the optimum water content is affected. The CBR test is carried out on material passing the 20 mm test sieve. Regarding weather conditions of northern Iran, CBR samples are saturated for about 4d (Chobbasti et al. 2010). Soil samples were moved into cylindrical plunger (proctor cylinder) and compacted in 5 layers with 56 blows per layers. The CBR tests were performed using the load-measuring device, which is connected to the compression machine with the loading rate of 1 mm/min. The mould with the sample and the surcharge weights were placed in the machine and loading rates were measured at every 2.5 mm displacement and CBR rates were calculated for each sample. One of the most useful methods of evaluating the strength of stabilized soil is UCS test. The required amount of additive to be used in stabilization of soil could be determined by this test. In this study, the test was performed for samples with 7, 14 and 28 days curing time to consider the time factor (Alhassan 2008). This test was performed in accordance with the standard ASTM D 2166. The compression device (a hydraulic-actuated loading piston with the capability of infinite rates of strain and stress loads) was used to measure the UCS rates (Fig. 2a). Also, special trimming was not needed because the California sampler was used in this test. After extruding the soil sample, specimens were placed on the bottom plate of the compression machine and compression load was applied. Then the rates of UCS were calculated. The ATERBERG limits test was performed in accordance with the standard ASTM D4318-05 (2002). According to the purpose of this study, this test can show how much the added pozzolanic materials could improve soil propCroat. j. for. eng. 37(2016)2

erties in terms of ATERBERG limits. To measure the liquid limit, about 250 gr of the studied soils (the material passing the No. 40 sieve) were weighted and thoroughly mixed with 15 to 20 ml of distilled water. Mixing was done on a glass plate to keep the whole sample at the same moisture content. The liquid limit was measured using a Casagraunde cup. Plastic limit of fine-aggregates in sandy materials have a major influence on the strength of used materials. Materials with a PI (plasticity index) exceeding the allowed values should not be used on base and sub-base layers, because materials with higher PI have lower shear strength. The plastic limit is often used together with the liquid limit to determine the plasticity index. To measure the plastic limit, about 20 g of the studied soils (the material passing the No. 40 sieve) is needed. Samples were mixed with the distilled water for 10 min to form a plastic ball and then rolled out under the fingers on a glass plate. The rate of rolling should be between 80 to 90 strokes per minute to form a 3 mm diameter. After crumbling, the pieces of crumbled soil were collected to determine the moisture content.

2.5 Data Analysis Statistical comparisons were done using the SPSS 16 software to compare means among treatments. Data (different combination of materials) were analyzed using ANOVA and Tukey HSD test in terms of soil mechanical properties. Also, the graphical displays were made using the SigmaPlot 12.0 software.

3. Results and Discussion 3.1 Compaction Characteristics There was an increase in OMC with the increase of RHA contents (Fig. 3). The trend is in line with Alhassan 2008 and Roy et al. 2010. According to Fig. 3, it can be found that the soil OMC for combinations of S+4L+7R and S+6L+9R had increased 19% and 21.9%, respectively, compared to the natural soil samples. The increase was a result of the addition of RHA, which reduced the quantity of free silt and clay fraction and coarser materials with larger surface areas were formed (these processes require water). Furthermore, this indicates that more water was needed to compact the soil-RHA mixtures (Ramezanianpour et al. 2009). According to Fig. 4, adding RHA to the soil caused a decrease in MDD. The trend is in line with Alhassan 2008, Basha et al. 2005, Trivedi et al. 2013. Given that the MDD for the natural soil and S+6L was 1.54 Mg m-3 and 1.42 Mg m-3, respectively, by adding 9% RHA it was possible to reduce the soil MDD to 1.21 Mg m-3.

337

M. Nasiri et al. Optimum Utilization of Rice Husk Ash for Stabilization of Sub-base Materials in Construction ... (333â&#x20AC;&#x201C;343)

Fig. 3 Influence of RHA and lime on OMC parameter

Fig. 4 Influence of RHA and lime on MDD parameter

The reduction in MDD can be related to the replacement of soil by RHA in the mixture, which has a relatively lower specific gravity compared to the soil (Osinubi and Katte 1997). It may also be related to the coating of soil by RHA, which results to large particles with larger voids and less density (Chobbasti et al. 2010). The reduction in the MDD may also be explained by considering the RHA as filler (with lower specific gravity) in the soil voids.

3.2 ATERBERG Limits The results showed that the soil liquid limit decreased with an increase in the RHA (Fig. 5). The trend is in line with Jha and Gill 2006, Onyango et al. 2007, Purbi Sen et al. 2011. The liquid limit of the natural soil and S+6L was 39.4 and 29.7, respectively. According to Fig. 5, it can be found that 9% RHA can reduce the soil liquid limit to 26.1%. The comparison of re-

Fig. 5 Influence of RHA and lime on plastic and liquid limits

338

Croat. j. for. eng. 37(2016)2

Optimum Utilization of Rice Husk Ash for Stabilization of Sub-base Materials in Construction ... (333–343) M. Nasiri et al.

Fig. 6 Influence of RHA and lime on CBR parameters; (a) Saturated condition; (b) Optimum water content duction ratio between treatments indicated that the lime can have a much greater effect on the reduction of the soil liquid limit (a reduction of 10%). The reason for this reduction could be the chemical reaction between the limes and soil that require more moisture and provides this moisture from the soil. The results showed the plastic limit decreased with an increase in the RHA (Fig. 5). The results are consistent with Muntohar 2002, Trivedi et al. 2013. The plastic limit of the natural soil and S+6L was 28.2% and 25.8%, respectively. Similar to the soil liquid limit, the comparison of the reduction ratio between treatments revealed that lime can have a much greater effect on the soil plastic limit.

3.3 California Bearing Ratio The results indicated in both saturated condition and optimum water content that the addition of RHA could increase the soil CBR; however, the increasing trend varied for different percentages (Fig. 6). Trends shown in Fig. 6 are in agreement with Chobbasti et al. 2010, Purbi Sen et al. 2011, Sabat and Nanda 2011. In the saturated condition, the treatment of S+4L+7R reached the highest effectiveness on CBR (an increase of 27%) and then decreased. Alhassan (2008) reported that the highest effectiveness on CBR for saturated condition was 18.5%. The comparison results showed that the CBR of saturated condition was 28% and 19% higher than the natural soil and S+6L, respectively. The results of optimum water content also indicated that the CBR reached its maximum values in the treatment Croat. j. for. eng. 37(2016)2

of S+4L+7R (46%) and then dropped. The maximum differences recorded for CBR were 37.5% and 28% higher than the natural soil and S+6L.

3.4 Unconfined Compressive Strength According to Fig. 7, it can be found that the UCS did not have a lot of changes with increasing in lime (4% lime or 6% lime). The UCS values increased with the addition of RHA to its maximum at between 7–9% RHA and then dropped from 12% RHA. According to Fig. 7, the highest rate of increase in UCS occurs in the first 28 days. The trend is in line with Alhassan 2008 and Jha and Gill 2006. Alhassan (2008), stated that the highest UCS was related to 8% RHA and 28 days samples. The comparison results showed that the maximum UCS of stabilized soil (237 KN m-2) was 23% and 16% higher than the natural soil (214 KN m-2) and S+6L (221 KN m-2). The subsequent increase in the UCS is attributed to the formation of cementitious compounds between the CaOH present in the soil and RHA as well as the pozzolans present in the RHA (Jami et al. 2013). The reason of this reduction in the UCS values after the addition of 9% RHA could be excess RHA introduced to the soil and hence forming weak bonds between the soil and cementitious compounds formed.

4. Conclusion We concluded that pozzolanic materials, such as rice husk ash (RHA), increase the materials quality in soil stabilization method with lime. Since the rice husk

339

M. Nasiri et al. Optimum Utilization of Rice Husk Ash for Stabilization of Sub-base Materials in Construction ... (333â&#x20AC;&#x201C;343)

Fig. 7 Influence of RHA and lime on UCS parameters; (a) 7 days curing time; (b) 14 days curing time; (c) 28 days curing time is an agricultural waste material and it is available in many countries at little or no cost (1$ for each gunnybags in Iran), the produced ash from rice husk can be used as an appropriate and inexpensive stabilizer in sub-base layer of forest roads. So, this study introduces the best combination of sub-base soil, rice husk ash and lime by examining ten mix designs: According to CBR results, the combinations of S+4L+7R and S+6L+12R had best performance. By examining the values of UCS, the combinations of

340

S+4L+9R and S+6L+9R were selected as the most effective ones. According to the OMC, the combinations of S+4L+9R, S+6L+9R as well as S+6L+12R were selected as the best combinations and according to MDD, the combinations of S+4L+9R, S+4L+12R and S+6L+9R had almost the same performance. The results of soil plastic and liquid limits indicated that the combinations of S+4L+9R and S+6L+9R had the best performance in reduction of both plastic and liquid limits. So, accurate examinations between the menCroat. j. for. eng. 37(2016)2

Optimum Utilization of Rice Husk Ash for Stabilization of Sub-base Materials in Construction ... (333â&#x20AC;&#x201C;343) M. Nasiri et al.

Fig. 8 Comparison between the natural soil and combinations of S+4L+9R and S+4L+9R in improvement of mechanical properties of materials; CBR-OPT: CBR owc; CBR-S: CBR saturated; UCS-28: 28 days curing time; Where letters in superscript differ, data are significantly different between combinations for each soil mechanical test (p<0.05) tioned treatments showed that the combination of S+4L+9R and S+6L+9R could be considered as an optimum utilization to stabilize sub-base materials. The positive effects of these two combinations in improvement of mechanical properties of Aâ&#x20AC;&#x201C;6 soil are shown in Fig. 8. However, the important aspect is the use of less lime due to environmental issues surrounding forest roads and economic situation of forest roads projects. Thus, the combination of S+4L+9R was selected as the best combination for stabilizing Croat. j. for. eng. 37(2016)2

the sub-base materials of forest roads due to less consumption of lime.

Acknowledgements The authors would like to thank the University of Sari Agricultural Sciences and Natural Resources for funding this research and to acknowledge the support of Mr. Bozorgpour, Mr. Valizadeh and Mr. Savadkoohi for their invaluable assistance in soil sampling and laboratory tests.

341

M. Nasiri et al. Optimum Utilization of Rice Husk Ash for Stabilization of Sub-base Materials in Construction ... (333–343)

5. References Alhassan, M., 2008: Permeability of lateritic soil treated with lime and rice husk ash. Assumption University Journal of Thailand 12(2): 115–120. American Association of State Highway and Transportation Officials (AASHTO), 1986: Standard Specifications for Classification of Soils and Soil-Aggregate Mixtures for Highway Construction Purposes (M145). Washington D C, USA. American Society for Testing and Material (ASTM), 2002: Test Methods for Laboratory Compaction Characteristics of Soil Using Standard Effort (D0698-07E01). Annual Books of ASTM Standard. Section 4 Vol. 0408 USA. American Society for Testing and Material (ASTM), 2002: Test Methods for Liquid Limit, Plastic Limit, and Plasticity Index of Soils (D4318-05). Annual Books of ASTM Standard, Section 4 Vol. 0408 USA. American Society for Testing and Material (ASTM), 2002: Test Methods for CBR (California Bearing Ratio) of Laboratory Compacted Soils (D1883-07). Annual Books of ASTM Standard, Section 4 Vol. 0408 USA. Basha, E.A., Hashim, R., Mahmud, H.B., Muntohar, A.S., 2005: Stabilization of residual soil with rice husk ash and cement. Construction and Building Materials 19(6): 448–453. Chobbasti, A.J., Ghodrati, H., Vahdatirad, M.J., Firouzian, S., Barari, A., Torabi, M., Bagherian, A., 2010: Influence of using rice husk ash in soil stabilization method with lime. Frontiers of Earth Science in China 4(4): 471–480. Jamil, M., Kaish, A.B.M.A., Raman, S.N., Zain, M.F.M., 2013: Pozzolanic contribution of rice husk ash in cementitious system. Construction and Building Materials 47: 588–593. Jha, J.N., Gill, K.G., 2006: Effect of rice husk ash on lime stabilization. Journal of the Institution of Engineers 87: 33– 39. Harichane, K., Ghrici, M., Kenai, S., 2011: Effect of curing time on shear strength of cohesive soils stabilized with combination of lime and natural pozzolana. International Journal of Civil Engineering 9(2): 90–96.

Milani, S., Silva, A., Paula, D., 2012: Physical, Mechanical and Thermal Performance of cement-stabilized rammed earthrice husk ash walls. Journal of Materials in Civil Engineering 24(6): 775–782. Muntohar, A.S., 2002: Utilization of Uncontrolled Burnt Rice Husk Ash In Soil Improvement. Dimensi Teknik Sipil Journal 4(2): 100–105. Nair, D.G., Fraaij, A., Klaassen, A.K., Kentgens, P.M., 2008: A structural investigation relating to the pozzolanic activity of rice husk ashes. Journal of Cement and Concrete Research 38(6): 861–869. Osinubi, K.J., Katte, V.Y., 1997: Effect of elapsed time after mixing on grain size and plasticity characteristics, soil lime mixes. Nigeria Soc. Engin. Tech Trans 32(4): 65–76. Onyango, M., Macha, I., Busch, C., 2007: Use of Naturally Occurring Pozzolans for Road Construction in Tanzania. 9th International Conference on Low-Volume Roads. June 24– 27, Austin, Texas. Péterfalvi, J., Primusz, P., Markó, G., Kisfaludi, B., Kosztka, M., 2015: Evaluation of the Effect of Lime-Stabilized Subgrade on the Performance of an Experimental Road Pavement. Croat. j. for. eng 36(2): 269–282. Purbi, S., Mukesh, N., Dixit, M., 2011: Evaluation of Strength Characteristics of Clayey Soil by Adding Soil Stabilizing Additives. International Journal of Earth Sciences and Engineering 4: 1060–1064. Ramezanianpour, A.A., Khani, M.M., Ahmadibeni, G.H., 2009: The Effect of Rice Husk Ash on Mechanical Properties and Durability of Sustainable Concretes. International Journal of Civil Engineering 7(2): 83–91. Roy, T.K., Chattopadhyay, B.C., Roy, S.K., 2010: Effect of Alternative Materials on Engineering Properties of Alluvial Soil for Construction of Road Subgrade. Proceeding of Traffic and Transportation Studies 1331–1340. Sabat, A.K., Nanda, R.P., 2011: Effect of marble dust on strength and durability of Rice husk ash stabilised expansive soil. International journal of civil and structural engineering 1(4): 939–948.

Khan, R., Jabbar, A., Irshad, A., Khan, W., Naeem Khan, A., Mirza, J., 2012: Reduction in environmental problems using rice-husk ash in concrete. Construction and Building Materials 30: 360–365.

Trivedi, J.S., Sandeep, N., Chakradhar, I., 2013: Optimum Utilization of Fly Ash for Stabilization of SubGrade Soil using Genetic Algorithm. Procedia Engineering 51: 250–258.

Latifi, N., Matro, A., Khari, M., 2012: Monitoring results of Alborz earth dam during construction. Electronic Journal of Geotechnical Engineering 17: 2474–2484.

Weiting, X., Tommy, Y.L., Shazim, A.M., 2012: Microstructure and reactivity of rich husk ash. Construction and Building Materials 29: 541–547.

342

Croat. j. for. eng. 37(2016)2

Optimum Utilization of Rice Husk Ash for Stabilization of Sub-base Materials in Construction ... (333â&#x20AC;&#x201C;343) M. Nasiri et al.

Authorsâ&#x20AC;&#x2122; address: Mehran Nasiri, PhD. student * e-mail: me.nasiri@sanru.ac.ir; me.nasiri66@gmail.com Assoc. prof. Majid Lotfalian, PhD. e-mail: mlotfalian@sanru.ac.ir Sari university of agricultural sciences and natural resources Department of Forest engineering P.O Box 576, Badeleh, Sari, Mazandaran IRAN

Received: October 27, 2015 Accepted: February 02, 2016 Croat. j. for. eng. 37(2016)2

Assoc. prof. Amir Modarres, PhD. e-mail: a.modarres@nit.ac.ir Babol University of Technology Department of civil engineering P.O. box 484, Shariati Av., Babol, Mazandaran IRAN Prof. Wei Wu PhD. e-mail: wei.wu@boku.ac.at University of Natural Resources and Life sciences Institute of Geotechnical Engineering P.O. Box 85084, Vienna AUSTRIA * Corresponding author

343

Original scientific paper

Improvement of Forest Road Gravel Surfacing Quality by Nano-polymer CBR PLUS Majid Lotfalian, Aidin Parsakhoo, Amir Savadkoohi Abstract In this study, CBR PLUS was used on gravel roads to reduce erosion and consequently maintenance costs of forest road network. Nano-polymer CBR PLUS with weight percentages of 0.25, 0.50, 0.75 and 1.00% in water were added to the surfacing materials of a forest road in Nekachoob Company of Iran. A comparative soil chemical and mechanical test for untreated and treated soil was carried out to observe variations in results. The results showed that by increasing the percentage of CBR PLUS, the soil fine particles content decreased. In addition, with increasing the CBR PLUS content, a decrease of the liquid limits was observed in soil. Soil treated with CBR PLUS, resulted in improvement of soil dry density and California bearing capacity (CBR). Moreover, the amount of Calcium, Magnesium, Potassium ions increased with the increasing CBR PLUS. Finally, it was concluded that 1.00% of nano-polymer CBR PLUS acted better in comparison with lower concentration. However, it is suggested that higher levels of this substance should be evaluated in future studies. Keywords: roadway, gravel surface, stabilizer, soil mechanical tests, nekachoob

1. Introduction Road network plays a major role in the movement of the countryâ&#x20AC;&#x2122;s people and goods. Most of the forest roads in the Hyrcanian zone have a gravel surface and the un-sieved materials of roadways consist of sandy to sandy clay loam soil material. Gravel is brought to the site from a quarry or stream bed. Sometimes, oversize stone or clay particles can be detected in stream materials. These poor materials of road surfaces are subject to erosion and degradation, which lead to sedimentation (Miller et al. 1998, Liu et al. 2009). The performance of a forest road depends on the quality of two layers including natural earth and improved surface layer (Inyang et al. 2007). Coarse-grained soils composed of a mixture of coarse and fine gravels with little or no fines are suitable for natural earth. Above the compacted natural earth, there is an improved layer with the thickness of 15â&#x20AC;&#x201C;20 cm of well graded materials (Sessions 2007). The capacity of these layers to support the loads applied to the ground is defined as bearing capacity. A useful parameter for determinCroat. j. for. eng. 37(2016)2

ing the strength of subgrade material is the California Bearing Ratio (CBR). CBR measurements in forest road/gravel road conditions are often carried out using the light falling weight deflectometer (LFWD), the dynamic cone penetrometer (DCP) and the conventional falling weight deflectometer (FWD) (Kaakkurivaara et al. 2015). It is crucial for forest engineers to develop gravel forest roads with a CBR value. The surface should be kept smooth and free from holes and ruts (Mishra 2012, Krishna et al. 2013). Clay in gravel road feels slippery and sticky when moist. Soil engineering properties in forest road construction depend on clay content. Plastic clays exhibit shrink/swell behavior with a change in moisture content (Sharma et al. 2008). Swelling of clays causes serious damages in the gravel surface of forest road especially during the harvesting operation (Azzam 2014). Al-Rawas et al. (2005) used various stabilizers including lime, Portland cement and lime/cement mixtures to consolidate subgrade or base materials and concluded that lime shows the greatest improvement

345

M. Lotfalian et al.

Improvement of Forest Road Gravel Surfacing Quality by Nano-polymer CBR PLUS (345–352)

to compressibility CBR values and swelling. Nowadays, numerous types of nano-polymer particles/nano composites are used for soil stabilization in engineering operations. CBR PLUS is a unique cation reactive organic compound, which forms protective, oily layers on the surface of soil and clay particles. It reduces ion mobility and ion exchange and simultaneously makes the clay material hydrophobic by eliminating the adsorption of water. CBR PLUS has been successfully used on gravel roads to reduce maintenance costs (Petry and Little 2002). CBR PLUS is a concentrated liquid, of which 100 liters consolidate between 10,000 and 20,000 square meters of soil to a depth of 15 cm when diluted with water. Taherkhani et al. (2012) used CBR PLUS for improving physical and mechanical properties of different combinations of clay, sand and gravel. They concluded that the CBR PLUS decreases the plasticity index and permeability and increases the compressive strength and CBR of the combinations. The combination of 50% gravel, 30% sand and 20% of clay is found to attain the highest strength with the stabilizer (Piratheepan et al. 2010). Unfortunately, despite of wide distribution of clay soils in Hyrcanian forest and their high destructive influence on road surface materials, no research has been conducted regarding improvement of such clay by nano-polymers. Therefore, finding of low cost nano-polymers and efficient concentration to stabilize these materials is important for the prosperity in forestry operations. CBR PLUS has the potential to increase the strength and durability of road construction materials. The aim of the present study was: Þ t o investigate the effects of nano-polymer CBR PLUS on some physical and mechanical properties of road surface materials

Þ t o determine the optimum percentage of CBR PLUS for improving the quality of forest road gravel surface Þ to assess the effects of CBR PLUS on some chemical properties of runoff produced from road surface materials

2. Materials and methods 2.1 Study area This study was conducted for forest roads in district one of Nekachoob forestry plan. This forest is located in the South-East of Neka city (from 36˚28’55’’ to 36˚36’55’’ North Latitude and from 53˚20’15’’ to 53˚35’00’’ East Longitude), Iran. The total length of forest road is 24 km with maximum bearing capacity of 45 tons. These roads are classified as secondary forest roads and used for the aims of forestry and wood transportation. The forest is located at an altitude of 140–600 m above sea level. The precipitation is 618 mm throughout the year. The climate of the region is mid moist and temperate. In this study, a segment of the road with a dimension of 50 meter × 4 meter and gravel surfacing depth of 0.15 meter was selected for treatments.

2.2 Data collection and measurements In this study, the surfacing material was brought to the site from a stream bed to improve the surfacing quality of a forest road. Five macro plots with dimensions of 2×3 meters were established in longitudinal slope direction of the road. One of these macro plots was untreated and others were treated by CBR PLUS nano-polymer with weight percentages of 0.25, 0.50, 0.75 and 1.00% in relation to the weight of water, in

Fig. 1 Schematic of sampling design in study area

346

Croat. j. for. eng. 37(2016)2

Improvement of Forest Road Gravel Surfacing Quality by Nano-polymer CBR PLUS (345–352)

which CBR PLUS was solved. In this study, the technical performance of CBR PLUS nano-polymer showed lower concentration, because CBR PLUS nano-polymer is still considerably more expensive than the conventional methods and a small increase in concentration can enormously increase overall costs when considering the total length of road network. Besides, the biological effects of this nano-polymer have not yet been recognized. Each macro plot was divided into 24 micro plots with dimensions of 0.5×0.5 meters and then four of them were randomly selected for soil analysis (Fig. 1). After two weeks from the treatment time, 48 kg of the surfacing material was collected from the selected micro plots to investigate the variations in mechanical status of materials. Soil mechanic test, including Atterberg limits (liquid limit, plastic limit and index), moisture, density AASHTO (Proctor compaction) and CBR, was carried out in the laboratory. Sieve Analysis: Dry and wet sieve analysis of the soil was performed in the laboratory. The sieve sizes were 0.075, 0.15, 0.60, 2.00, 2.36, 4.00 and 8.00 mm. The particle size distribution was obtained from records of the weight of soil particles retained on each sieve and is usually shown as a graph of ‘percentage passing by weight’ as a function of particle size. The soil particles larger than 0.06 mm in diameter can be described as coarse and the smaller ones as fine particles. Water content: Soil moisture was calculated using Equation (1):

W1 − W2 × 100 W2 − W3

(1)

Where: W1 weight of the can (g) + wet soil (g) W2 weight of the can (g) + dry soil (g) W3 weight of the empty can (g) The tested soil samples were oven-dried at 105 °C for 24 hours. Atterberg Limits: Liquid Limit (LL) was determined using Equation (2) (Atterberg 1911):

 N LL = WN ×    25 

0.121

(2)

Where: N number of drops of the cup required to close the groove W soil moisture content (%) at which the groove is closed The moisture content, as determined in Equation (1), when the soil sample is cracking, is the Plastic Limit (PL). Croat. j. for. eng. 37(2016)2

M. Lotfalian et al.

The Plasticity Index (PI) of the soil is the numerical difference between its liquid limit and its plastic limit (Equation 3, Atterberg 1911):

PI = LL − PL

(3)

Standard Proctor test: To assess the amount of compaction and the water content required in the field, compaction test (Standard Proctor test) was done on the soil. The water content, at which the maximum dry density is attained, was obtained from the relationships provided by the test (Equation 4, Davidson and Gardiner 1949).

Pd =

 W 1  100 + G  s

(4)

Where: Ρd dry density of the soil g/cm3 Gs specific gravity of the soil being tested (assume 2.70 if not given) PW density of water in g/cm3 (approximately 1 g/cm3) W moisture content (%) In the Proctor test, the soil was first air dried and then separated into samples. The water content of each sample was adjusted by adding water. The soil was then placed and compacted in the 4-inch-diameter Proctor compaction mould using 25 blows by a 5.5 lb standard hammer falling 12 inches. At the end, the sample was removed from the mould and the dry density and the water content of the sample were determined for each Proctor compaction test. Then, a curve was plotted for the dry density as a function of the water content. From this curve, the optimum water content to reach the maximum dry density can be obtained. California Bearing Ratio (CBR): The CBR value of a soil is an index related to the strength of the soil (Equation 5, Yetimoglu et al. 2005).

Yd =

Yt 1+ W

(5)

Where: Yd equivalent dry unit weight of the soil Yt total unit weight W moisture content (%) In the CBR test, soil samples were compacted using metal rammer to obtain unit weights both above and below the desired unit weight. After allowing the sample to take on water by soaking, or other specified treatment such as curing, each sample was subjected to penetration by a cylindrical rod. Stress versus penetration

347

M. Lotfalian et al.

Improvement of Forest Road Gravel Surfacing Quality by Nano-polymer CBR PLUS (345–352)

depth were plotted to determine the CBR for each specimen. The CBR at the specified density was determined from a graph of CBR versus dry unit weight. Rainfall Simulation on treated plots: A portable single nozzle (nozzle code: Schlick r86510) rainfall simulator simulated rain with drop size of 3 mm for 20 min at the intensity of 32 mm/h. Water was rained from the nozzle mounted 2 m above the ground onto 20 micro plots of 0.5×0.5 meters bordered by a steel structure. Runoff water samples were collected by water gauge and then transmitted to the laboratory. The road was a secondary road and during the study, all road traffic was stopped for 30 days. Measuring the runoff Calcium (Ca), Magnesium (Mg) and Potassium (K): Surface runoff suspension of each gauge was sampled by 0.3-L glass recipients. Immediately after the collection and after the settling of the sediment in the glass, a sample of the solution was removed for analysis of Ca, Mg and K contents. Ca was measured using the NaOH titration method. Mg was determined with atomic absorption spectrophotometer. The available K was determined by ammonium acetate extraction at pH 9. Statistical package for social sciences (SPSS) were used for data analysis and least significant difference (LSD) was used for comparing averages.

3. Results When making the surface layer of forest roads, it is not enough to apply individually the bearing capacity of clay materials, gravel and sand. State Standard Specifications (1996) recommended that the blend of different size aggregates should be 40–80% hard stone graded from 6 to 76 mm in diameter in the base and 19 mm aggregate for surface gravel (crushed stone) with 20–60% sand (less than 6 mm) and 8–15% fines. Greater amount of clay in the surfacing layer of forest road makes it prone to erosion and sensitive to freezing. For the purpose of this study, surfacing materials were used as follows: 42% of gravel, 37% of sand and 21% of silt and clay. Then, these combinations were stabilized with different percentages of nano-polymer CBR PLUS. As shown in Fig. 2, a certain procedure cannot be stated for the effects of CBR PLUS upon plastic limit and plastic index of soil. The liquid limit decreased 4% by increasing CBR PLUS (Table 2). The rate of soil dry density has increased by heightening CBR PLUS. Using different moisture contents from 4% to 9%, the dry unit weight measured at the end of each Proctor Test were plotted, and subsequently the maximum dry density and optimum water content for each CBR PLUS treatment were obtained. It was ob-

348

Fig. 2 Effect of different percent of CBR PLUS on Atterberg limits including Plastic Index (PI), Plastic Limit (PL) and Liquid Limit (LL) served that, by adding CBR PLUS to the road surface materials, the maximum dry density changed significantly from 2.09 (control sample) to 2.17 g/cm3 (1.00% CBR PLUS) (Table 2). The state at the peak is said to be that of 100% compaction at a particular compactive effort, the curve is usually of a hyperbolic form, when the points obtained from tests are smoothly joined. When water is added to dry soil, it helps in bringing the solid particles close by coating them with thin films

Fig. 3 Dry density of materials treated by CBR PLUS in Proctor test Croat. j. for. eng. 37(2016)2

Improvement of Forest Road Gravel Surfacing Quality by Nano-polymer CBR PLUS (345â&#x20AC;&#x201C;352)

Table 1 Weight distribution of particle size for GM (Silty gravel or silty gravel with sand based on unified soil classification system) materials treated by CBR PLUS at the depth of 0â&#x20AC;&#x201C;20 cm Treatment

Fine particles, %

Coarse particles, %

Control

0.25% CBR PLUS

0.50% CBR PLUS

22.9

77.1

0.75% CBR PLUS

20.9

79.1

1.00% CBR PLUS

19.9

80.1

of water. The peak of dry density was detected at water content of 7% (Fig. 3). According to the grading curve of the treated and untreated samples, it can be concluded that by heightening the CBR PLUS from 0.25% to 1.00%, the amount of soil fine particles are reduced from 28% (control sample) to 27% (0.25% CBR PLUS) and then decreased by 8.1% when adding 1.00% CBR PLUS. The 1 percent treatment holds the least weight percent of fine particles and maximum weight percent of coarse particles compared to other treatments (Table 1). Particle size distribution analysis revealed that the materials treated by 1.00% CBR PLUS contained more sand and gravel size particles (>0.06 mm) as compared to materials treated with lower percentages of CBR PLUS. Indeed, considering the grain size distribution curves for 0.25%, CBR PLUS and control sample are poorly graded (Fig. 4).

M. Lotfalian et al.

Fig. 5 shows the California bearing ratio increased by increasing the CBR PLUS. It was observed that by adding CBR PLUS to the road surface materials, the CBR values changed significantly from 73% (control sample) to 78% (0.25% CBR PLUS) (Table 2). It was identified that the amount of runoff Calcium, Magnesium, Potassium in treated plots was higher than in untreated plots after rainfall simulation. The amount of these ions increased with the increase of the percentage of CBR PLUS (Table 3).

4. Discussion The effect of using a new stabilization product, CBR PLUS, on gravel surfacing layer of forest road Table 2 Comparison of some mechanical properties of materials treated by different weight percentages of CBR PLUS Treatment

Dry density g/cm3

CBR, %

Plastic index, %

Liquid limit, %

Control

1.98b

73b

10.55a

24.79a

0.25% CBR PLUS

1.99b

78ab

9.72a

24.59a

0.50% CBR PLUS

2.03ab

82a

9.50a

24.49a

0.75% CBR PLUS

2.09a

85a

10.44a

24.00b

1.00% CBR PLUS

2.13a

88a

9.80a

23.79b

Note: Different superscripts show significant difference at probability level of 5% based on LSD test

Fig. 4 Curve of weight distribution of particle size for materials treated by different percentages of CBR PLUS Croat. j. for. eng. 37(2016)2

349

M. Lotfalian et al.

Improvement of Forest Road Gravel Surfacing Quality by Nano-polymer CBR PLUS (345â&#x20AC;&#x201C;352)

Fig. 5 Variations of California bearing ratio (dry CBR) by increasing the percentage of CBR PLUS was studied by conducting different experiments such as Atterbergâ&#x20AC;&#x2122;s limits, Standard Proctor and California bearing ratio tests on soil mixed with different percentages of CBR PLUS: 0.25%, 0.50%, 0.75%, and 1.00%. In a study in Turkey, polymer was used as an additive for the improvement of a base road material (Kavak et al. 2010). Nanotechnology is one of the most active research areas dealing with civil engineering and construction materials. CBR PLUS is often applied in small quantities on the road surface by using a water tanker normally used for dust control. Water is a natural destroyer of roads. CBR PLUS nanotechnology allows permanent waterproofing of soils and aggregate surfaces. In this study, the liquid limit decreased significantly (4%) by increasing CBR PLUS, which was in agreement with the findings of Taherkhani et al. (2012). They concluded that the CBR PLUS decreases the plasticity index and permeability of materials. In addition, Musavi et al. (2014) showed that the CBR PLUS decreased the liquid limit and plasticity index. The Atterberg liquid limit is the water

content at which the body begins to flow, using a specific apparatus. High liquid limits indicate soils of high clay content and low bearing capacity. Therefore, when a soil liquid limit decreases, the soil mechanical resistance improves (Jianqiao et al. 2012). The stabilization process of road surfacing materials by nano-polymer CBR PLUS requires more than two weeks, which is called supplementary period (Lahalih and Ahmed 1998). The density rate of dry soil increased by heightening CBR PLUS, which indicates that the shear resistance increased and soil infiltration decreased. Musavi et al. (2014) showed that the CBR PLUS can increase the optimum moisture and maximum dry density. We found that by heightening the CBR PLUS, the amount of soil fine particles reduced. The combination of 50% gravel, 30% sand and 20% of clay is found to attain the highest strength with the stabilizer (Piratheepan et al. 2010). Moreover, California bearing ratio increased by increasing the CBR PLUS and this is the indicator of high resistance and bearing capacity of materials. This finding was also observed by several other researchers (Taherkhani et al. 2012, Musavi et al. 2014). Increment in CBR value by nano-polymer CBR PLUS is helpful in reducing the quantity of gravel materials and consequently thickness of the surfacing layer of forest road. In the present research, after rainfall simulation, it was identified that the amount of runoff Calcium, Magnesium and Potassiumin was higher in treated plots than in untreated plots. The amount of these ions increased with the increase of the percentage of CBR PLUS. It was proved that the clay particles cause absorption of metal ions (Velde 1995, Jain and Ram 1997). The metal ions, while absorbing a high amount of water, cause the decrease of soil resistance and increase of its volume. These ions are called water absorbents. CBR PLUS is a kind of product made from organic Solphonic Asid, which destructs water absorbents and breaks the absorbents-clay link (Yilmaz and Civelekoglu 2009, Taherkhani and Javanmard 2015). Then it would be replaced by absorbents and make the Clay hydroponic. This leads to the release of ions like Calcium, Magnesium and Potassium, heightening the resistance and bearing ratio of soil.

Table 3 Amount of runoff Calcium, Magnesium, Potassium in treated and untreated plots (mg/Li)

350

Variables

Control

0.25% CBR PLUS

0.50% CBR PLUS

0.75% CBR PLUS

1.00% CBR PLUS

Calcium

120

145

172

196

224

Magnesium

Potassium

Croat. j. for. eng. 37(2016)2

Improvement of Forest Road Gravel Surfacing Quality by Nano-polymer CBR PLUS (345–352)

The conventional method of surfacing requires high maintenance costs. Road gravel surface making use of CBR PLUS may have a high initial cost but it has nearly no maintenance costs, which will have a positive effect on the road economy in a long-run. In this study, it was concluded that 1.00% of nano-polymer CBR PLUS acted better in relation to the weight of water, in which this material is solved, in comparison with other treatments. However, it is suggested that higher levels of this substance should be evaluated in future studies.

5. References Atterberg, A., 1911: On the investigation of the physical properties of soils and on the plasticity of clays. Internationale Mitteilungen für Bodenkunde 1: 10–43. In German. Al-Rawas, A.A., Hago, A.W., Al-Sarmi, H., 2005: Effect of lime, cement and Saroo (artificial pozzolan) on the swelling potential of an expansive soil from Oman. Building and Environment 40(5): 681–687. Azzam, W.R., 2014: Behavior of modified clay microstructure using polymer nanocomposites technique. Alexandria Engineering Journal 53(1): 143–150. Davidson, D.T., Gardiner, W.F., 1949: Calculation of standard proctor density and optimum moisture content from mechanical, analysis, shrinkage and factors and plasticity index. Highway Research Board 29(1): 447–481. Inyang, H.I., Bae, S., Mbamalu, G., Park, S.W., 2007: Aqueous polymer effects on volumetric swelling of Namontrollonite. Journal of Materials in Civil Engineering 19(1): 84–90. Jain, C.K., Ram, D., 1997: Adsorption of metal ions on bed sediments. Hydrological Sciences – Journal des Sciences Hydrologiques 42(5): 713–723. Jianqiao, L.I., Xiaodong, Z., Meng, Z., Hao, L.I., 2012: Soil liquid limit and plastic limit treating system based on analytic method. Procedia Earth and Planetary Science 5(1): 175–179.

M. Lotfalian et al.

Lahalih, S.M., Ahmed, N., 1998: Effect of new stabilizers on the compressive strength of dune sand. Construction and Building Materials 12: 321–328. Miller, W.P., Willis, R.L., Levy, G.J., 1998: Aggregate stabilization in kaolinitic soils by low rates of anionic polyacrylamide. Soil Use and Management 14(2): 1001–105. Mishra, E.N.K., 2012: Strength characteristics of clays subgrade soil stabilization with fly-ash and lime for road works. Indian Geotechnical Journal 4(7): 206–211. Musavi F., Abdi E., Estabragh, A.R., Majnounian, B., 2014: Assessing the capability of polymer stabilizer in forest road stabilization (Case study: Kheyrud forest). Iranian Journal of Forest 6(1): 1–10. In Persian. Petry, T., Little, D., 2002: Review of stabilization of clays and expansive soils in pavements and lightly loaded structuresHistory, practice and future. Journal of Materials in Civil Engineering 14(6): 447–460. Piratheepan, J., Gnanendran, C., Lo, S., 2010: Characterization of cementitiously stabilized granular materials for pavement design using unconfined compression and IDT testing with internal displacement measurements. Journal of Materials in Civil Engineering 22(5): 495–505. Sessions, J., 2007: Forest road operations in the Tropics. Springer-Verlag Berlin Heidelberg New York. 170 p. Sharma, R.S., PhaniKumar, B.R., Rao, B.V., 2008: Engineering behavior of a remolded expansive clay blended with lime, calcium chloride and rice husk ash. Journal of Materials in Civil Engineering 20(8): 509–515. State of Wisconsin, Department of transportation, 1996: Standard specifications for highway and structure construction. 564 p. Taherkhani, H., Hashemi, A., Sharifi, V., 2012: Evaluating the use of CBR PLUS for constructing the pavement layers from stabilized soils. Journal of Transportation Engineering 3(4): 339–344.

Kaakkurivaara, T., Vuorimies, N., Kolisoja, P., Uusitalo, J., 2015: Applicability of portable tools in assessing the bearing capacity of forest roads. Silva Fennica 49(2): 1–26.

Taherkhani, H., Javanmard, M., 2015: Comparison of the effects of cement, lime and CBR PLUS on reducing the swelling of clay soils. Geology Engineering Journal 9(4): 3131– 3150.

Kavak, A., Bilgen, G., Mutman, U., 2010: In-situ modification of a road material using polymer. Scientific Research and Essays 5(17): 2547–2555.

Velde, B., 1995: Composition and mineralogy of clay minerals. In: Velde, B. (ed.) Origin and mineralogy of clays. New York, Springer-Verlag, 8–42.

Krishna, G., Padmavathi, M., Prashanth Kumar, S., 2013: Stabilization of black cotton soil treated with fly ash and Zycosoil. International Journal of Civil Engineering and Building Materials 3(3): 133–144.

Yetimoglu, T., Inanir, M., Inanir, O., 2005: A study on bearing capacity of randomly distributed fiber-reinforced sand fills overlying soft clay. Geotextiles and Geomembranes 23(2): 174–183.

Liu, J., Shi, B., Jiang, H., Bae, S., Huang, H., 2009: Improvement of water stability of clay aggregates admixed with aqueous polymer soil stabilizers. Catena 77: 175–179.

Yilmaz, I., Civelekoglu, B., 2009: Gypsum: an additive for stabilization of swelling clay soils. Journal of Applied Clay Science 44(12):166–172.

Croat. j. for. eng. 37(2016)2

351

M. Lotfalian et al.

Improvement of Forest Road Gravel Surfacing Quality by Nano-polymer CBR PLUS (345â&#x20AC;&#x201C;352)

Authorsâ&#x20AC;&#x2122; address: Assoc. prof. Majid Lotfalian, PhD.* e-mail: mlotfalian@sanru.ac.ir Amir Savadkoohi, MSc. student e-mail: savadkoohi_st59@yahoo.com Sari University of Agricultural Sciences and Natural Resources Faculty of Natural Resources 9 km Darya Road Sari, Mazandaran IRAN

Received: December 27, 2015 Accepted: February 23, 2016

352

Assist. prof. Aidin Parsakhoo, PhD. e-mail: parsakhoo@gau.ac.ir aidinparsakhoo@yahoo.com Gorgan University of Agricultural Sciences and Natural Resources Faculty of Forest Science Basij Square Gorgan, 49189-43464 IRAN * Corresponding author Croat. j. for. eng. 37(2016)2

Original scientific paper

Integrated Oak Timber Protection from Ambrosia Bark Beetles: Economic and Ecological Importance in Harvesting Operations Milivoj Franjević, Tomislav Poršinsky, Andreja Đuka Abstract Ambrosia bark beetles belong to a group of xylomycetophagous insects from the order Coleoptera, family Curculionidae and are characterized as important pests of oak timber. Galleries they form during their life cycle and infect with »ambrosia fungi« significantly decrease the economic value of oak roundwood. A state owned Croatian company »Hrvatske šume« Ltd. manages FSC certificated forests, where pedunculate and sessile oaks account for 22.6% in the annual allowable cut. Methods of oak roundwood protection, that used to be implemented in the past, are now banned in accordance with EU legislation as well as FSC criteria of forest protection. In these forest management conditions, it becomes necessary to introduce new biotechnical methods for oak roundwood protection. Available commercial products, flight barrier traps and synthetic semiochemicals, as well as pretreated insecticide treated polymer nets, were tested as means of integrated oak roundwood protection. Implementation of these products was tested in field conditions. Good knowledge of phenology of ambrosia bark beetles, thorough understanding of timber harvesting operations and field conditions that dominate in even aged oak forests, is crucial if applied methods are to be effective and taken on time. Field experiments conducted in this research showed that early seasonal deployment of semiochemically baited flight barrier traps can reduce the number of bark beetles that infest oak timber. It was also concluded that without additional protection with polymer nets, it is not possible to protect oak timber in compliance with strict FprEN 1316-1: 2012 E standards for oak roundwood classes, which do not allow any timber infestation in the highest quality grades (A and B quality class). Semiochemicals used as repellents during the research were ineffective. In the early months of spring, oak roundwood is at high risk of infestation at the roadside landings, where it is waiting to be transported. Keywords: oak roundwood, ambrosia bark beetles, integrated timber protection, FprEN 13161: 2012 E, pest control

1. Introduction In Croatian even-aged oak stands, ambrosia bark beetles are important economic pests that degrade the value of oak roundwood exposed to these beetles during main felling. Ambrosia bark beetles belong to a group of xylomycetophagous insects from the order Coleoptera, family Curculionidae. Dispersal flight of these beetles starts in early winter months and continCroat. j. for. eng. 37(2016)2

ues in the spring, interrupted only by a few weeks of cold weather throughout the harvesting season. It is a serious threat to oak roundwood in landing areas. Phenology of ambrosia bark beetles, namely their infestation period, generally matches with the period of intensive harvesting operations in Croatian oak stands. According to Official Gazette (NN 17/2015) even-aged forests, managed by regeneration cuttings (preparatory, seeding, additional and final felling), cutting,

353

M. Franjević et al.

Integrated Oak Timber Protection from Ambrosia Bark Beetles: Economic and Ecological ... (353–364)

bucking and timber extraction (from the forest stand) is prohibited during vegetation period i.e. from April 1 to September 30, unless forest stands were excessively flooded, and harvesting operations can only be performed in two months after the vegetation period starts. In both even-aged and un-even aged forests, thinning cutting is prohibited in the first two months of the vegetation period. Galleries produced by ambrosia bark beetles during the completion of their life cycle decrease the economic value of oak roundwood, both directly and indirectly by the discoloration as a consequence of their fungal symbionts (hence ambrosia beetles). In the growing number of FSC certified stands, the application of insecticides for the protection of oak roundwood from ambrosia beetles is restricted because of FSC policy (in relation to the use of pesticides in FSC-certified forests and plantations) with the aim of minimizing the negative environmental and social impacts of pesticides whilst promoting economically viable management. This policy is implemented through compliance with the requirements of FSCSTD-01-001 FSC Principles and Criteria of Forest Stewardship and the associated national or subnational indicators and means of verification (FSC-POL-30-001 (2005) EN). According to Beuk et al. (2007), the company »Hrvatske šume« Ltd. manages 76% or 302.4 million m3 of timber in Croatian state owned forests, where pedunculate oak (Quercus robur L.) accounts for 14.9% (45.0 mil m3) and sessile oak (Quercus petraea (Matt.) Liebl.) for 9.5% (29 mil. m3) of the total growing stock. Pedunculate oak accounts for 13.6% (0.79 mil m3) and sessile oak accounts for 9.0% (0.52 mil m3) in the annual allowable cut. The same authors state that the most valuable tree specie in the Croatian forests is pedunculate oak and that it is of high quality and very demanding management conditions. In 2002, the company »Hrvatske šume« Ltd. was awarded with the FSC certificate for forest management. According to the definition, »FSC certification means that forests are managed in accordance with strict ecological, social and economic standards« (Anon. 2015). According to standard FprEN 1316-1:2012:2012: E Hardwood round timber – Qualitative classification Part 1: Oak and beech insect attack is not permitted in classes A (first class timber: generally corresponding to a butt log with clear timber or with only minor features not restricting the use) and B (average to first class timber, with no specific requirements for clear wood), while in class C (timber of average to low quality, allowing all quality features, which do not seriously reduce the natural features of wood) is only permitted in sapwood. Standard classification applies

354

to the following species: Oak (Quercus sessiliflora SALISB. or Quercus petraea LIEBL., Quercus robur L. or Quercus pedunculata EHRH.) and Beech (Fagus sylvatica L). The methods of oak protection, implemented in the past, are now banned following the FSC criteria. In the last decades, growing concern about environment protection and efficient forest protection has focused on a new approach to forest protection and formerly deployed IPM strategies – integrated pest management (IPM). Holistic in approach, it takes into account all the aspects of forest ecosystem and forest management operations chaining them together into an environmentally friendly and economically efficient protection of forests. Introduction of new biotechnical methods of oak timber protection seem as an obvious tool in the new ecological, economic and legal circumstances. The purpose of this research was to explore the possibilities of targeted ethological manipulation of adult ambrosia beetles during their swarming flight in search for oak timber suitable for infestation. Some of the commercially available products, as well as a few of those in the preproduction phase, were applied in trial testing their potential in the integrated oak timber protection. Intercept flight barrier traps with various semiochemicals and FSC approved polymer nets with incorporated insecticides were tested in field conditions during the years 2003, 2009–2011. Oak logs were used for the evaluation of efficiency in each of the tested chemicals/trial designs. With the repeated population monitoring and consecutive catches of ambrosia beetles in oak stands, where trials were set up, it turned out that Trypodendron domesticum (Linnaeus 1758) has two generations per year in Croatian environment (Franjević 2013, Petercord 2006). It was also observed that the population of Asian species Xylosandrus germanus (Blandford 1894) has been established in Croatian oak stands within the last decade. Valuable biological knowledge was collected on several species of ambrosia bark beetles that inhabit our forests and causes damage to oak roundwood including Xyleborus dispar (Fabricius 1792), Xyleborus monographus (Fabricius 1792), Xyleborus saxesenii (Ratzeburg 1837). It was observed that T. domesticum is actively flying from the beginning of January, when weather conditions are favorable (characterized by warmer weather so typical for the last decade, which is consistent with the report of Intergovernmental Panel on Climate Change (IPCC 2007) that states that in 21st century ground air temperature rises from 1.8°C to 4°C depending on greenhouse gases emissions as stated by Meehl et al. 2007 and Ramanathan and Feng 2009. Trypodendron signatum (FaCroat. j. for. eng. 37(2016)2

Integrated Oak Timber Protection from Ambrosia Bark Beetles: Economic and Ecological ... (353–364)

bricius 1792) activity follows in February. Platypus cylindrus (Fabricius 1792) has a period of activity from early June to early October. Asian bark beetle species X. germanus appeared for the first time in trials in 2009, and since then it became the second most frequent species in pheromone trap catches in 2011 (Franjević 2012). It has been concluded that attention should be paid to the spreading of this invasive species and population building. The same applies to its role in the, now expanded, Croatian ambrosia beetle group.

2. Materials and Methods In lowland even-aged oak stands, methods of integrated oak roundwood protection were tested in the period from 2003 to 2011. Pheromone baited intercept panel traps were used as an olfactory manipulation, trapping and monitoring of ambrosia beetle flight periods. Two field locations were Jastrebarsko in Zagreb

M. Franjević et al.

county and Otok near Vinkovci. They were selected as good representatives of pedunculate oak stands with Jastrebarsko as the most western part of its distribution in Croatia and ecologically inferior to oak stands in the eastern part of Croatia – Vinkovci. Ambrosia bark beetles that infest oak timber are equally represented through pedunculate oak area of distribution. IPM® Tech InterceptTM panel traps were used because of their advantages over Lindgren® and Theysohn® panel traps in Cerambycid and Scolytid trapping. IPM® Tech InterceptTM panel traps catch beetles from all four quadrants and are less susceptible to weather conditions and predatory entomofauna, which can influence the results of trapping (Czokajlo et al. 2002). Also, because of different strategies that were used in oak timber protection from ambrosia beetles, traps were not always active at the same time of the year. However, they were always active in the period when ambrosia bark beetle swarming can be expected. IPM®

Fig. 1 Pin holes visible on bark (top left) and holes on debarked one-meter logs (top right) and oak roundwood before protection (below left) and protected (below right) with Woodnet® system Croat. j. for. eng. 37(2016)2

355

M. Franjević et al.

Integrated Oak Timber Protection from Ambrosia Bark Beetles: Economic and Ecological ... (353–364)

Tech InterceptTM panel traps were completed with different attractive components in the years of experiment. Simultaneously with panel trap exposure, (pedunculate) oak timber was placed in various designs and separated in control and protected group. After the experiment was concluded, all timber was debarked and bark beetle pin holes counted on site (Fig. 1). Data from the control and treated/protected group of oak timber were statistically analyzed. Also, trap catches and weather station data were used for the correlation analysis between phenology and species occurrence. Throughout the duration of these experiments, panel trap catches were collected weekly and analyzed in laboratory. In 2003, panel traps were baited with lineatin, an attractive component of ambrosia beetles (Macconell et al. 1977). Six traps, in five repetitions, were set 10 meters from the logs in circular position and spaced 60° angularly from each other. The control group in Jastrebarsko consisted of 80 oak logs, and the protected group of 85 logs. Otok control group consisted of 61 oak logs and the protected group of 55 logs (with α-pinene as repelent). Otok trials were conducted just with the repellent component imitating the smell of conifers without panel traps. The experiment was carried out from March 4, 2003 to April 2, 2003. In Jastrebarsko control group in the year 2009, GLV (Green Leaf Volatile) and Domowit-Trypowit D® were used for trapping in panel traps ETOH. Twelve traps, in five repetitions, were set 20 meters from the logs in circular position and spaced 30° from each other. ETOH and Domowit-Trypowit D® were used in panel traps as attractive components. As a repellent on exposed oak roundwood, ampoules of Tompin® were used. This is a pheromone component used for baiting of species from genus Tomicus (Tomicus piniperda, Tomicus minor and Tomicus destruens) and it contains aggregation pheromones but also α-pinen, which is the primary attractive component found in bark and resin of conifers. Six Tompin® ampoules were attached on every oak log bundle. The experiment was carried out from March 17, 2009 to April 28, 2009. Control and protected/treated group consisted of 50 oak logs each. In Jastrebarsko in the year 2010, twenty-four traps were set 20 meters from the logs in circular position and spaced 15° from each other. ETOH and DomowitTrypowit D® were used in panel traps as attractive components. As a repellent on exposed oak roundwood, ampoules of Hostowit® and Kombiwit® were used on oak log bundles. Hostowit® has a universal attractive component for bark beetles on conifers. Kombiwit® is aggregation pheromone for Ips typographus and Pityogenes chalcographus. Control and pro-

356

tected group consisted of 50 oak logs each. The experiment was carried out from April 21, 2010 to May 26, 2010. All oak logs used in trials were 1 m long. On June 8, 2011, oak tree was cut down and logs were protected with Woodnet® system product of BASF® Chemical Company (Fig. 1). According to »old« Croatian Standards of Forest Exploitation Products (Anon. 1995) that are still in use and derived from ex-Yugoslav standards JUS (Anon. 1989), oak roundwood is classified into two types of veneer (class I and II of veneer logs) and three sawlog classes (class I, II and III of sawlogs). Roundwood was protected within half an hour after cutting. Removal and evaluation of Woodnet® system protection was done on September 1, 2011. During the period of Woodnet® system evaluation, IPM Tech® panel traps were collecting data in the same forest management unit. In the year 2011, monitoring of swarming period for ambrosia bark beetles was conducted in lowland oak stands with five randomly positioned traps that were set with ETOH and Domowit-Trypowit D® and used in panel traps as attractive components. The experiment was carried out from January 11, 2011 to June 6, 2011. ETOH is a known attractant for ambrosia beetles (Moeck 1970) and Domowit-Trypowit D® is a commercially available product for trapping of beetles from Trypodendron genus. In the year 2011, monitoring of ambrosia beetle phenology was conducted from

Fig. 2 Statistical analysis of pin holes in oak roundwood per square meter for treated (T) and control group of logs in Jastrebarsko 2003 protection with lineatin baited traps Croat. j. for. eng. 37(2016)2

Integrated Oak Timber Protection from Ambrosia Bark Beetles: Economic and Ecological ... (353–364)

M. Franjević et al.

Table 1 Descriptive statistics for Jastrebarsko 2003 field measurement Effect

Factor level

Total

Mean

Standard deviation

Standard error

–95.00%

+95.00%

165

37.46703

65.26752

5.081069

27.43428

47.49978

Logs

Treated

14.82165

17.26183

1.872309

11.09836

18.54494

Logs

Control

61.52775

85.95576

9.610147

42.39923

80.65627

Table 2 Descriptive statistics for Otok 2003 field measurements Effect

Factor level

Total

Mean

Standard deviation

Standard error

–95.00%

+95.00%

116

23.62087

50.06520

4.648437

14.41321

32.82853

Logs

Treated

22.27058

57.55435

7.760627

6.71146

37.82970

Logs

Control

24.83834

42.65842

5.461851

13.91301

35.76367

early January to early June and in that time Spectrum Technologies Inc. Watchdog® Weather Station 2000 Series were used for temperature measurements.

During the period of trapping with lineatin in the year 2003, 30 panel traps were set. In that year, trials were conducted on two locations – Jastrebarsko and Otok. After debarking, the results showed insufficient level of protection regarding both experiment sites.

However, on the location Jastrebarsko, a distinct difference between the protected (15 pin holes per square meter) and control (62 pin holes per square meter) group of oak timber (Fig. 2 and Table 1) was observed. The location Otok showed no distinct difference (Fig. 3 and Table 2) between the two groups of oak timber – protected (15 pin holes per square meter) and control (24 pin holes per square meter), meaning that α-pinene as repelent was ineffective. The period of trapping in Jastrebarsko 2009 with panel traps baited with ETOH, GLV and Domowit-

Fig. 3 Statistical analysis of pin holes in oak roundwood per square meter for treated and control group of logs in Otok 2003 protection with a-pinene as repellent

Fig. 4 Statistical analysis of pin holes in oak timber per square meter for treated and control group of logs in Jastrebarsko 2009 protection with ETOH, GLV and Domowit-Trypowit D® baited traps

3. Results

Croat. j. for. eng. 37(2016)2

357

M. Franjević et al.

Integrated Oak Timber Protection from Ambrosia Bark Beetles: Economic and Ecological ... (353–364)

Fig. 5 Statistical analysis of pin holes in oak roundwood per square meter for treated and control group of logs in Jastrebarsko 2010 protection with ETOH, GLV and Domowit-Trypowit D® baited traps

Fig. 6 Statistical analysis of pin holes in oak roundwood per square meter for treated and control group of logs in Jastrebarsko 2003, 2009 and 2010

Trypowit D® is shown in Fig. 4 and Table 3. There were 60 pheromone baited panel traps positioned in 12 traps around oak roundwood bunch in five repetitions, but the level of oak roundwood protection was also insufficient. There was a distinct difference between the control and protected group of oak roundwood but protection was low with the protected group of oak roundwood (22 pin holes per square meter) and control group (59 pin holes per square meter).

The period of trapping in Jastrebarsko 2010 with panel traps baited with ETOH, GLV and DomowitTrypowit D® is shown in Fig. 5 and Table 4. There were 24 pheromone baited panel traps positioned around a single oak roundwood bunch with control group. The level of oak roundwood protection was also low, with control group less infested with ambrosia bark beetles. Protected logs had 11 pin holes per square meter and control group 9 pin holes per square meter.

Table 3 Descriptive statistics for Jastrebarsko 2009 field measurements Effect

Factor level

Total

Mean

Standard deviation

Standard error

–95.00%

+95.00%

100

39.95691

27.10859

2.710859

34.57797

45.33584

Logs

Treated

21.20417

13.53189

1.913698

17.35845

25.04989

Logs

Control

58.70965

24.16511

3.417462

51.84200

65.57729

Table 4 Descriptive statistics for Jastrebarsko 2010 field measurements Effect

Factor level

Total

Mean

Standard deviation

Standard error

–95.00%

+95.00%

100

10.22870

8.662325

0.866233

8.509903

11.94749

Logs

Treated

11.39364

9.558832

1.351823

8.677052

14.11023

Logs

Control

9.06375

7.580386

1.072028

6.909429

11.21807

358

Croat. j. for. eng. 37(2016)2

Integrated Oak Timber Protection from Ambrosia Bark Beetles: Economic and Ecological ... (353–364)

M. Franjević et al.

Table 5 Ambrosia bark beetle trap catches in 2011 (January 18 to April 5) with weekly maximum and minimum average temperatures 2011

Jan. 18

Jan. 25

Feb. 1

Feb. 8

Feb. 15

Feb. 22

Mar. 1

Mar. 8

Mar. 15

Mar. 22

Mar. 29

Apr. 5

X. germanus

T. domesticum

171

226

317

T. signatum

352

2317

351

1609

856

X. saxesenii

189

X. dispar

X. monographus

172

183

578

2637

447

1765

1154

Max (°C)

10.8

3.8

2.1

8.9

11.8

4.5

1.4

3.4

15.6

13.3

22,6

Min (°C)

–0.2

–2.9

–3.3

–5.8

–3.9

0.1

–7.3

–3.6

–4.5

2.6

–0.2

1,3

In the year 2011, phenology of ambrosia bark beetles was monitored and correlated with average weekly temperatures (Fig. 7). If daily temperatures exceed 9°C, it is sufficient for activating early ambrosia bark beetles from Trypodendron genus (Petercord 2006). Overall catches for species of ambrosia bark beetles in periods of monitoring are presented in Tables 5 and 6. Analysis of oak logs protected with Woodnet® system on September 1, 2010 was done after 3 months of exposure. Debarking of oak timber and detailed observation showed only 24 pin holes. All observed pin-

holes were on places were bark was previously damaged. In six pinholes, females of X. monographus were found. No other xylophagous insects or their remains were found on the Woodnet® system or near it. The remains of beetles from genus Geotrupes were the only evidence of system toxicity near Woodnet® system.

4. Discussion and Conclusions During the experiments of integrated oak roundwood protection, insufficient levels of protection were achieved with pheromone baited traps. The reason can

Fig. 7 Total catches of ambrosia bark beetles during monitoring Croat. j. for. eng. 37(2016)2

359

M. Franjević et al.

Integrated Oak Timber Protection from Ambrosia Bark Beetles: Economic and Ecological ... (353–364)

Table 6 Ambrosia bark beetle trap catches in 2011 (April 12 to June 7) with weekly maximum and minimum average temperatures 2011

Apr. 12

Apr. 19

Apr. 26

May 3

May 10

May 17

May 24

May 31

June 7

X. germanus

225

450

263

162

185

1466

T. domesticum

110

1356

T. signatum

878

570

2105

872

1245

1615

1103

540

289

14787

X. saxesenii

135

723

X. dispar

116

X. monographus

121

364

1283

620

2751

919

1381

2115

1524

860

415

18812

Max (°C)

23.1

18.7

24.7

18.3

21.1

26.6

26.4

27.5

–

Min (°C)

2.8

0.9

8.9

0.8

6.2

8.5

9.3

12.9

–

be explained by the fact that ambrosia bark beetles are polifagous species and each generation has to find their suitable host for development, so in managed forests there are relatively few of them. Suitable trees are randomly positioned in forests and their number varies from year to year. Position of suitable trees is unpredictable for ambrosia bark beetles and they have developed complex mechanisms for finding suitable hosts, which is generally based on semiochemicals and aggregation pheromones (Wood 1982). Ambrosia bark beetles are aggregation insects and their populations are pulse eruptive (Thalenhorst 1958, Berryman 1987). Periods of gradation usually last from 5 to 7 years and during that period bark beetles can damage a large number of trees (Bombosh 1954, Schroeder and Lindelöw 2002, Jakuš et al. 2003). Accordingly, during the experiments, protection of oak roundwood was more efficient in regard of protected/treated vs. control groups (pin holes) in the early period of ambrosia bark beetle swarming. Panel traps were more attractive during the early period of dispersion flight and sweeping forest for suitable hosts. Panel traps were more influenced by the period of exposure than by attractive component used during our experiments. Once aggregation process started in exposed oak logs, pheromone baited traps lost their efficiency in the protection of oak timber (Fig. 6). Overall levels of protection achieved with pheromone baited traps were insufficient in accordance with strict FprEN 1316-1:2012:2012: E Hardwood round timber – Qualitative classification – Part 1: Oak and beech, which does not permit any insect attacks in A and B quality class, whilst for C class insect attacks are permitted only in sapwood. In terms of »old« Croatian

360

Standards that are still in use, insect attacks are also not permitted in veneer classes. Repellents used in experiments were inefficient although some species of ambrosia bark beetles are polyphagous (X. dispar, X. germanus). Protection of oak roundwood with Woodnet® system achieved excellent results, and showed great modularity and usability in FSC certified forests. Data collected during 2011 is valuable for understanding phenology and is crucial for the implementation of integrated oak roundwood protection. Winter experiments show risks in forest operations for oak roundwood exposed to ambrosia bark beetles especially from Trypodendron genus in swarming periods (Fig. 8). Among monitored species of ambrosia bark beetles, the following species are mentioned in European literature: T. domesticum, T. signatum, X. dispar, X. germanus (Maksymov 1987, Bruge 1995), P. cylindrus, as univoltine species, and as bivoltine species X. saxesenii, X. monographus and according to some authors X. germanus (Faccoli and Rukalski 2004). Monitoring of ambrosia bark beetles in 2011 bivoltinism for T. domesticum was carried out in weather conditions favorable for early swarming, first in January and then in May (Tables 5 and 6, Fig. 7). Existence of the second generation of T. domesticum is described by some foreign authors (Eichhorn and Graf 1974, Petercord 2006). Based on the results of these experiments, the recommendation would be to protect the exposed oak roundwood from the middle of the winter (January) when performing forest operations with oak roundwood. Weather in winter with daily maximum temperatures exceeding 9°C are favorable for the beginCroat. j. for. eng. 37(2016)2

Integrated Oak Timber Protection from Ambrosia Bark Beetles: Economic and Ecological ... (353–364)

Fig. 8 Risk of pin hole damage on oak roundwood in relation to ambrosia bark beetle species swarming period and forest operations ning of T. domesticum swarming. Protective systems like BASF Woodnet® for oak roundwood that was not possible to transport from landing areas and barrier panel traps equipped with attractive components were used as early warning for the start of ambrosia bark beetle swarming.

M. Franjević et al.

Strict European standards (FprEN 1316-1:2012:2012: E) do not allow any timber infestation for the most valuable assortments (A and B quality class) and oak timber is exposed to the highest risk in the period from mid January to mid March regarding dominant species T. domesticum and T. signatum. The results of the research give some guidance to the ongoing development of a new approach to integrated oak timber protection. They reveal the actual potential of the use of semiochemicals in the process, either for their use as a monitoring tool (very usable) or means of reduction of timber damages (generally low) and mass trapping of timber beetles (also very low). A developing method of targeted mechanical protection by use of chemically treated (FSC and WHO approved) polymer reusable net has been recognized as highly promising. It identifies flight periods of species that infest oak timber and suggests methods of protection. Even though the research was done on pedunculate oak timber and roundwood, the proposed integrated protection systems can be used on any roundwood because ambrosia bark beetles are polyphagous species that can successfully develop on a broad spectrum of both broadleaves and conifers. Integrated timber protection, although with higher costs in the beginning (BASF Woodnet® price is 125 €/100 m2, while previously used, but now banned, Deltacid in dose of 15 L/100 m2 was 26 €), eventually

Fig. 9 Prices of pedunculate oak and European beech logs Croat. j. for. eng. 37(2016)2

361

M. Franjević et al.

Integrated Oak Timber Protection from Ambrosia Bark Beetles: Economic and Ecological ... (353–364)

Fig. 10 Typical forwarder roadside landing brings both ecological (through FSC certification) and economic benefits. According to Business Annual Reports of the company »Hrvatske šume« Ltd. for the year 2009 and 2010, pedunculate oak and sessile oak Class I and II of veneer (which must also be free of insect attacks) account for 42.48% of all veneer assortments together. According to timber products price list (Fig. 9) of the company »Hrvatske šume« Ltd. (Anon. 2012), class I veneer logs of pedunculate and sessile oak are 3.15 times more costly than those of beech (Fagus sylvatica), the most common (Beuk et al. 2007) tree species in the Republic of Croatia (which accounts for 37.4% i.e. 113.2 million m3 of timber), 1.52 time more costly than those of narrow-leaved ash (with overall share of 3.9% i.e. 11.8 million m3) and other ash species (Fraxinus anfustifolia L. and other) and 4.33 time more costly than those of hornbeam (Carpinus betulus L.) with an overall share of 7.7% i.e. 23.2 million m3. Further to the above, it can be concluded that integrated timber protection can bring many advantages to forest management of broadleaved species. However, it should also be mentioned that the use of systems like BASF Woodnet® will require changing/adjusting of roadside landings (Fig. 10). The process of covering/uncovering logs presents a challenge (especially in terms of poorly organized landings or for example elongated skidder roadside landings), but nevertheless BASF Woodnet® systems should be used for the most valuable assortments, situated on the landing at the end of the harvesting season waiting for the binding process. The results of the research give some guidance to the ongoing development of a new approach to integrated oak timber protection. They reveal the actual

362

potential of the use of semiochemicals in the process, either for their use as a monitoring tool (very usable) or means of reduction of timber damages (generally low) and mass trapping of timber beetles (also very low). A newly developing method of targeted mechanical protection by the use of chemically treated (FSC and WHO approved) polymer reusable net has been recognized as highly promising. It identifies flight periods of species that infest oak timber and suggests methods of protection. The use of the proposed integrated timber protection measures will also ensure the sustainability of holding (in terms of the company »Hrvatske šume« Ltd.) or receiving (in terms of private forest owners) the prestigious FSC certificate for forest management. As stated above, the FSC certificate represents a great honor, as it is an international acknowledgement that forests are being managed in accordance with very strict standards, and hence it is also a recognition to the forestry profession. The use of FSC approved measures for timber protection is aimed at minimizing the negative environmental and social impacts of pesticides while promoting economically viable management. The objective of FSC program is to promote environmentally responsible, socially beneficial and economically sustainable forest management.

Acknowledgments We would like to thank the Croatian Forest Research Institute in Jastrebarsko for the data collected from the weather station Spectrum Technologies Inc. Watchdog® Weather Station 2000 Series. We would also like to thank the Forest Department Jastrebarsko Croat. j. for. eng. 37(2016)2

Integrated Oak Timber Protection from Ambrosia Bark Beetles: Economic and Ecological ... (353–364)

for providing materials necessary for this research. Many thanks to Boris Hrašovec, Dariusz Czokajlo and Ake Lindelow for their collaboration and advices on the experimental design of field trials.

5. References Anon., 1989: Jugoslav Timber Standards. Editor Vidinska Japundža, Newspaper and Editorial Agency of Jugoslavia, Belgrade, 682 p. (in Serbian). Anon., 1995: Croatian Standards for Products of Forest Exploatation. II. Edition. Hrvatske norme proizvoda iskorištavanja šuma. II. izdanje, State Office for Mterology, Zagreb, 1–245 (in Croatian). Anon., 2012: Prices of Main Forest Products. Available on: http://nadmetanja. hrsume.hr/javnipoziv/20141215_kupnjasortimenata/03_CjenikGlavnihSumskih Proizvoda.pdf (in Croatian). Anon., 2015: FSC in »Hrvatske šume« Ltd.. Available on: http://portal.hrsume.hr/index.php/en/h-consult-doo/42certifikati/certifikati/252-consult (in Croatian). Berryman, A.A., 1987: The Theory and Classification of Outbreaks. In: Insect Outbreaks, P. Barbosa, J.C. Schultz (Eds.). San Diego: Academic Press, 3–30. 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. Bombosh, S., 1954: Zürpidemiologie des Buchdruckers (Ips typographus L.). In: Die Grosse Borkekäferkalamität in Südwestdeutchland 1944-51, G. Wellwnstein (Ed.). Ulm: Forstschutzstelle Südwest, Ringingen, Ebner, 239–83 (in German). Bruge, H., 1995: Xylosandrus germanus (Blandford, 1894) [Belg. Sp. Nov.] (Coleoptera Scolytidae). Annales de la Société royale belged’Entomologie, 131: 249–264 (in French). Czokajlo, D., Hrasovec, B., Pernek, M., Hilszczanski, J., Kolk, A., Teale, S., Wickham, J., Kirsch, P., 2003. New Lure for the Larger Pine Shoot Beetle, Tomicus Piniperda – Attractant/Trap Design Combinations Tested in North America and Europe. In: McManus, Michael L., Liebhold, Andrew M., eds. Proceedings: Ecology, Survey and Management of Forest Insects; 2002 September 1–5; Krakow, Poland. Gen. Tech. Rep. NE-311. Newtown Square, PA: U.S. Dept. of Agriculture, Forest Service, Northeastern Research Station, 6–9. Eichhorn, O., Graf, P., 1974: Uber einige Nutzholzborkenkafer und ihre Feinde [On some timber bark beetles and their enemies]. Anzeiger fur Schadlingskunde, Pflanzen- und Umweltschutz 47: 129–135. EN 1316-1: 2012: 2012 (E): Hardwood Round Timber – Qualitative Classification Part 1: Oak and Beech. (CSN EN 1316-1: Hardwood round timber – Qualitative classification – Part 1: Oak and beech. Final draft, 9 p. Faccoli, M., Rukalski, J.P., 2004: Attractiveness of Artificially Killed Red Oaks (Quercus rubra) to Ambrosia Beetles (Coleoptera Scolytidae). In: Cerretti, P., Hardersen, S., Mason, F., Croat. j. for. eng. 37(2016)2

M. Franjević et al.

Nardi, G., Tisato, M., Zapparoli, M., (eds), Invertebrati di una foresta della Pianura Padana, Bosco della Fontana – Secondo contributo. Conservazione Habitat Invertebrati, 3. Cierre Grafica Editore, Verona, 171–179. Franjević, M., 2012: Novel Biotechnical Methods within the Integrated Protection of Oak Timber Against Ambrosia Beetles. Dissertation thesis. Faculty of Forestry University of Zagreb, 1–224 (in Croatian). Franjević, M., 2013: Bivoltinism of European Hardwood Ambrosia Beetle Trypodendron domesticum in Croatian Lowland Oak Stands of Jastrebarski Lugovi. Šumarski list 137(9–10): 495–498 (in Croatian). FSC-POL-30-001 (2005) EN: FSC POLICY, FSC PESTICIDES POLICY. 2005 Forest Stewardship Council A.C., Bonn, Germany, 1–4. FSC Guide: To Integrated Pest, Disease and Weed Management in FSC Certified Forests and Plantations. FSC Technical Series No. 2009 – 001. Written by: Ian Willoughby, Carlos Wilcken, Philip Ivey, Kevin O’Grady and Frank Katto on behalf of the Forest Stewardship Council, 1–19. Jakuš, R., Grodzski, W., Jezik, M., Jachym, M., 2003: Definition of Spatial Patterns of Bark Beetles Ips typographus L. Outbreak Spreading in Tatra Mountains (Central Europe), Using GIS. In: Proceedings: Ecology, Survey and Management of Forest Insects, 2002 September 1–5, Krakow Poland, M.L. McManus, A. M. Liebhold (Eds). USDA Fores Service General Technical Repor NE 311: 25–32. Macconell, J.G., Borden, J.A., Silverstein, R.M., Stokkink, E., 1977: Isolation and Tenative Identification of Lineatin, a Pheromone from the Frass of Trypodendron lineatum (Coleoptera:Scolytidae). J. Chem. Ecol. 3(5): 549–561. Maksymov, J.K., 1987: Erstmaliger Masasenbefall des Schwarzen Nutzholzborkenkafer Xylosandrus germanus Blandf, in der Schweiz [First mass attack of Xylosandrus germanus in Switzerland]. Schweizerische Zeitschrift fur Forestwesen 138: 215–227. Meehl, G.A., Stocker, T.F., Collins, W.D., Friedlingstein, P., Gaye, A.T., Gregory, J.M., Kitoh, A., Knutti, R., Murphy, J.M., Noda, A., Raper, S.C.B., Watterson, I.G., Weaver, A.J., Zhao, Z.C., 2007: Global Climate Projections. U: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor, H.L. Miller (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 996 p. Moeck, H.A., 1970: Ethanol as the Primary Attractant fot the Ambrosia Beetle Tr y p o d e n d ro n lineatum (Coleoptera:Scolytidae). Canadian Entomologist 102(8): 985–994. NN 17/2015: Regulations on Marking of trees and timber assortments. Official Gazette 17/2015. Petercord, R., 2006: Flight Period of the Broad-Leaved Ambrosia Beetle Trypodendron domesticum L. in Luxembourg and Rhineland-Palatinate between 2002 and 2005. IUFRO Work-

363

M. Franjević et al.

Integrated Oak Timber Protection from Ambrosia Bark Beetles: Economic and Ecological ... (353–364)

ing Party 7.03.10 Proceedings of the Workshop, Gmunden/ Austria, 213–218. Schroeder, L.M., Lindelöw, A., 2002: Attacks on Living Spruce Trees by the Bark Beetle Ips typographus (Col., Scolytidae) Following a Storm-Felling: A Comparison Between Stands with and without Removal of Wind Felled Trees. Agricultural and Forest Entomology 4(1): 47–56.

Thalenhorst, W., 1958: Grundzüge der Populaationsdynamik des großen Fichtenborkenkäfers Ips typographus L. Schriftenreihe der Forstlichen Fakultät der Universität Göttingen 21: 1–126 (in German). Wood, D.L., 1982: The Role of Pheromones, Kairomones and Allomones in the Host Selection and Colonization Behavior of Bark Beetles. Anual Review of Entomology 27(1): 411–446.

Authors’ address:

Received: January 20, 2016 Accepted: March 16, 2016

364

Milivoj Franjević, PhD. e-mail: mfranjevic@sumfak.hr Prof. Tomislav Poršinsky, PhD. e-mail: tporsinsky@sumfak.hr Andreja Đuka, PhD.* e-mail: aduka@sumfak.hr Faculty of Forestry University of Zagreb Svetošimunska 25 10002 Zagreb CROATIA * Corresponding author Croat. j. for. eng. 37(2016)2

Original scientific paper

Effect of Pruning on Diameter Growth in Pinus brutia Ten. Plantations in Turkey Nesat Erkan, Erdogan Uzun, A. Cem Aydin, M. Necati Bas Abstract Pruning is a technique used to add value to trees growing in forest stands, allowing the formation of clear, knot-free wood. Although many factors affect timber value, knots are the primary cause of reduction in timber quality of conifers. On the other hand, pruning may also cause reduction in the rate of tree growth, depending on pruning intensity.The aim of this study is to assess the effects of different pruning intensities on DBH (diameter at breast height) growth of young Pinus brutia plantations. For this purpose, three field experimental sites each with different site qualities, were established in three different locations. Four different treatments were applied at each test site: 1) control, no pruning of branches, 2) pruning up to 25% of tree height, 3) pruning up to 50% of tree height, and 4) pruning up to 75% of tree height. The effects of pruning on DBH growth were observed over a period of 14 years. At »Bük« test site, which has the poorest site quality, only those specimens pruned up to 75% of tree height showed significant reduction in DBH growth compared to the control. However, at »Nebiler« and »Kursunlu« test sites, specimens pruned both up to 50% and 75% of tree height showed statistically significant decrease in DBH growth. To recover from pruning stress in terms of DBH growth rate, it took trees 6 years at the poorest test site and 4 years at the relatively better test sites. This indicates that site quality of plantation sites accounts for not only DBH growth differences between sites, but also recovery rate of trees from any disturbances. Results showed that for those trees pruned up to 75% of their height, total DBH growth was reduced by between 6.5% and 9.0% after 14 years compared to the control at the test sites. No negative effect from pruning on DBH increment was observed in the first growing season. This may be due to earlier storage of nutrition in different parts of trees, thereby enabling them to compensate for the stress of crown reduction during the first growing season following pruning. Keywords: pruning, Pinus brutia, diameter growth, growth reduction

1. Introduction Silvicultural treatments are defined as various technical operations carried out for different purposes from plantation establishment or regeneration through to final harvest. Pruning is one such operation that removes live and/or dead branches from a certain portion of tree stem, starting from ground level. The most common purpose of pruning is to produce knot-free wood called »clear wood«. It may be an attractive investment for forest management if it is economically feasible. For example in June 1993 in New Zealand, pruned logs were selling for as much as 350 $/m3 as opposed to 150 $/m3 for good unpruned logs. Even Croat. j. for. eng. 37(2016)2

though pruning is expensive and reduces tree diameter growth, managers choose to sacrifice some quantity of wood for superior quality and therefore higher profits (Maclaren 1993). Thinning is another silvicultural treatment that improves wood quality. For example, Guller et al. (2012), studying wood density traits in P. brutia, observed that thinning increased annual ring width, latewood proportion and average ring density. As pruning may be an attractive silvicultural technique from an economic standpoint, it is suggested that it be applied by forest managers. However, since it may reduce total diameter growth due to reduction in the tree leaf area, the most profitable pruning in-

365

N. Erkan et al.

Effect of Pruning on Diameter Growth in Pinus brutia Ten. Plantations in Turkey (365–373)

tensity needs to be investigated, depending on site quality and specific tree species (Saatcioglu 1971, Kalipsiz 1982, Savill et al. 1997, Schmidt and Wardle 2002). In Turkey, the State Forest Service (FS), General Directorate of Forestry (OGM), owns almost all the forests, and regulates thinning, pruning and other silvicultural treatments (OGM 2014). Pinus brutia is the most important tree species in Turkish forestry, accounting for approximately 5.8 million ha of natural and planted forest area (OGM 2012) and 4.3 million m 3/year of wood production (OGM 2013). Indeed, the species accounts for 31% of total industrial wood production of Turkey. Due to the uncertainty concerning the economic feasibility of pruning at different intensities, the FS currently runs a restricted pruning programme with a low level of implementation. The FS needs to know the economic effects of different pruning intensities on tree (Miraboglu 1983). Moreover, there is currently a lack of research results on pruning and its effects on tree growth for different forest tree species in Turkey. Accurate scientific information about the effects of different pruning intensities on diameter growth under different site conditions and qualities is needed before any large-scale implementation programme is undertaken. The earlier results of this research, which evaluated data for the first seven years following pruning, were published as a technical bulletin in 2010 by Erkan et al. However, additional evaluation on advanced ages will certainly help to obtain more reliable results. The study under report was conducted on Pinus brutia Ten. at age of 14 years. The aim of the study is to quantitatively describe the effects of different pruning intensities on diameter growth and increment at breast height (DBH) of P. brutia plantations growing on three sites, each with different site qualities.

2. Material and Methods 2.1 Study sites The study sites (»Bük«,»Nebiler« and »Kursunlu«) were located in three different areas of P. brutia plantations in Antalya Region in Southwestern Turkey (Table 1). The region has a typical Mediterranean climate with a relatively hot, dry summers, and mild and wet winters. The mean minimum temperature in the coldest month (January) is 5°C, and the mean maximum temperature of the hottest month (July) is 34°C. Annual rainfall, which is about 1091 mm, is mainly concentrated in the winter months, rainfall contribution in the summer months (from June to September) only accounts for 22 mm of the total annual precipitation. The means of climate data represent a 35 year period, from 1975 to 2010. Some of the other properties of test sites are given in Table 1.

2.2 Pruning regimes and measurements Pruning was performed prior to the growing season in February 2000 by removing all branches (live and dead) from outside the branch collar without damaging the main stem tissue, starting from ground level up to 25, 50 and 75% of total tree height. This gave control: unpruned, intensity 1: 25%, intensity 2: 50%, intensity 3: 75%. Total height was recorded as a reference to indicate the intensity of pruning because row spacing used in P. brutia plantations was 2.0×3.0 m and green branches closer to ground level on trees were still growing at this development stage. Annual DBH increment and total diameter growth were taken as dependent variables to investigate the effects of pruning on DBH growth. This response of forest trees to pruning is mainly reflected in diameter growth (Harold and Paul 1952, Kozlowski and Pallardy 1990, Kukpa 2007) whereas height growth, which is determined largely by site quality, is affected

Table 1 Test sites and some associated properties Stand age*

Mean stand DBH over bark*, cm

Site index top height at age 30, m

Mean stand height, m*

Soil type

Altitude m

Aspect

Coordinates

»Bük«

10.4

9.7

6.9

Sandy clay loam

692

N 36°57’51.45” E 30°24’42.42”

»Nebiler«

11.9

23.0

7.9

Sandy clay loam

310

Flat

N 36°57’23.38” E 30°34’13.54”

»Kursunlu«

12.0

22.8

7.8

Clay

Flat

N 37°00’59.90” E 30°49’36.99”

Test sites

* Values are for the time of pruning in year 2000

366

Croat. j. for. eng. 37(2016)2

Effect of Pruning on Diameter Growth in Pinus brutia Ten. Plantations in Turkey (365–373)

little by pruning unless the treatment is so severe that the tree vigour is reduced to the point where it simply stops its terminal growth (Schmidt and Wardle 2002). Total and pruning heights (cm) of all the treatment trees were obtained by using a CRAIN telescopic height measurement device. Total DBH growth under bark (cm) and annual DBH increment – ADI (mm) were measured for all treatments (including control trees) over a period of 14 years (from the beginning of the 2000 growing season to the end of 2013). Two increment cores were taken from opposing sides (North and South) from each sampled tree after the completion of the 2013 growing season. PRESSLER increment borers were used to obtain core samples. The annual DBH increment for the related year (annual tree ring width as the sum of the widths of two rings from two cores taken from opposite sides of a sample tree) was measured from fine sanded core samples to the nearest 0.01 mm using a PREISSER DIGI-MET measurement machine. Annual diameter increment (ADI) was determined for each tree as follows:

ADIk = RWk1 + RWk2

(1)

Where, for the related year: ADIk annual diameter increment at breast height under bark of kth tree RWk2 ring width measured from the opposite side (1 and 2) of kth tree

2.3. Experimental design and data analyses The trees were initially planted at 2.0×3.0 m spacing on the test sites. Individual stands at each test site were at the early stages of their development, mean DBH being not greater than 12.0 cm over bark in the pruning year, 2000. Test site ages varied from 12 to 25 years in the start of the pruning experiment (Table 1). Age differences between the test sites with similar diameters were due to differences in site qualities of given test site pairs. Site index was determined using site index table prepared for P. brutia plantations by Usta (1991). The trees are located in rows (2.0×3.0 m) on each test site. First, we have chosen 10 rows within a given test site. Each row (replication) consisted of 12 observation (treatment) trees (4 pruning types × 3 observation trees), in addition to buffer trees surrounding the observation trees. Each row in a given test site serves as a replication in the experiment. Each observation tree within a given row was randomly assigned to one of the four pruning types in such a way that each observation tree was surrounded by unpruned buffer Croat. j. for. eng. 37(2016)2

N. Erkan et al.

trees. This measure was taken in order to avoid any border effect, and also to more or less imitate a selective pruning regime as proposed by FS regulations (OGM 2014). Treatment trees were selected and marked amongst co-dominant trees within each test site. The mean DBH of observation trees for different treatments was not greater than 10 cm under bark, ranging from 7.9 cm to 9.9 cm. Differences amongst mean tree DBH values (under bark) within each treatment group were less than 2.0 cm. Such a measure was taken in order to minimize growth differences that may arise purely due to differences in initial DBH values of separate treatment groups. Other site variables (such as spacing, stand density, canopy closure and crown structure) that have a primary effect on growth (Erkan 1996, Erkan 1998) were supposed to be more or less similar for all treatment trees. The ANOVA (Analysis of Variance) Model for within site comparisons was as follows:

yik = µ + a i + e ik

(2)

Where: yik DBH increment of kth tree within ith pruning intensity y mean DBH increment of trees αi effect of pruning intensity and εik random error. Prior to selecting candidate test sites for the pruning experiment, a detailed field survey of available plantations sites in the region was undertaken. One of the criteria for selecting a test site was that each should be as homogenous as possible within itself in respect of its site characteristics as reflected in uniform plantation growth and ground cover properties. Effects of pruning on DBH increment were evaluated within each test site separately by using one-way ANOVA in SPSS 22.0 (SPSS Inc. 2015). ANOVA tests were performed for each of the measurement years between 2000 and 2013. Different pruning intensities were compared by applying Duncan’s multiple range tests.

3. Results ANOVA results showed that DBH increment in P. brutia was reduced by different degrees at all the test sites as a result of different pruning intensities. Differences in DBH increment were statistically significant (p<0.05) among pruning treatments starting from 2 to 7 years following pruning at the »Bük« test site (Table 2 and 3, Fig. 1). The same was true from 2 to 5 years both at »Nebiler« and »Kursunlu« test sites.

367

N. Erkan et al.

Effect of Pruning on Diameter Growth in Pinus brutia Ten. Plantations in Turkey (365–373)

Fig. 1 Annual DBH increment after pruning for different pruning intensities on three test sites (Int1, Int2 and Int3 refer to pruning intensities of 25, 50 and 75% of tree height, respectively)

According to Duncan’s tests, there were no statistically significant (p<0.05) differences between the control and the 25% pruning treatment at all experimental sites. In »Bük«, which is a relatively poor site, only the highest intensity (75%) pruning treatment showed a significant (p<0.05) decrease in diameter increment

368

compared to the control (Fig. 1a). The inhibiting effect of 75% pruning treatment lasted for six years, after which annual DBH increments were equal to that of control trees. During this six years, trees pruned to 75% of total height showed about 9% less growth in total DBH on average compared to the control group Croat. j. for. eng. 37(2016)2

Effect of Pruning on Diameter Growth in Pinus brutia Ten. Plantations in Turkey (365–373)

N. Erkan et al.

Fig. 2 Cumulative DBH growth for 14 years following different pruning intensities on three test sites (Int1, Int2 and Int3 refer to pruning intensities of 25, 50 and 75% of tree height, respectively)

in »Bük« (Fig. 2a). In »Nebiler« 50 and 75% pruning treatments had significantly (p<0.05) different effects from the control in years two-four and two-five after pruning, respectively. Similarly, in »Kursunlu«, 50 and 75% pruning treatments were significantly (p<0.05) different from the control in the third year and for the Croat. j. for. eng. 37(2016)2

period of two-five years after pruning, respectively (Table 3, Fig. 1b, 1c). Total DBH growth loss at the end of year 14, due to the reduction of DBH increment after pruning, for intensity 3 (75% pruning) compared to the control group was 6.5 and 6.7% at »Nebiler« and »Kursunlu«, respectively.

369

N. Erkan et al.

Effect of Pruning on Diameter Growth in Pinus brutia Ten. Plantations in Turkey (365–373)

Table 2 ANOVA results for DBH increment for various years after pruning under four different pruning intensities (groups) at three test sites (»Bük«, »Nebiler« and »Kursunlu«)

»Bük«

Source of

Years after pruning

»Nebiler«

variance

Sum of squares

Between treatments

1.4

0.38

0.764

5.4

0.53

0.659

5.4

0.38

0.768

Within treatments

137.1

115

384.5

115

555.4

116

Total

138.5

118

389.9

118

560.9

119

Between treatments

19.0

17.8

0.000

Within treatments

38.9

115

Total

57.9

118

Between treatments

58.8

22.71

0.000

Within treatments

72.1

Total

130.9

Between treatments

181.5

9.61

0.000

Within treatments

223.7

115

Total

405.2

118

Between treatments

45.2

2.78

0.044

Within treatments

121.2

Total

166.4

0.42

0.739

0.88

0.455

1.05

0.371

0.516

0.672

0.253

0.859

0.268

0.848

18.75

86.9

267.6

115

354.4

118

376.1

115

421.7

118

797.8

31.26

0.000

541.0

116

791.1

119

402.2

115

684.8

116

118

1087.0

119

369.2

115

558.9

118

24.6

115

161.5

118

186.1

40.4

Within treatments

216.9

115

Total

257.2

118

Between treatments

17.4

Within treatments

141.9

Total

159.3

14.29

0.000

766.3

119

37.5

115

520.9

116

118

558.4

119

220.3

118

6.2

115

175.6

118

181.8

5.1

Within treatments

85.9

115

Total

91.0

118

Between treatments

1.2

Within treatments

24.1

Total

25.3

2.27

3 116

486.7

119

8.7

115

383.8

116

118

392.5

119

3 115

217.8

118

6.8

115

225.7

118

232.6

1.34

0.001

5.2

1.4

0.129

5.84

0.000

481.4

216.4 1.93

0.084

19.70

0.000

116

115

0.004

34.18

0.000

613.7

7.4

4.69

0.000

12.44

152.6

212.8

Between treatments

7.13

0.000

250.0

189.8

Between treatments

31.09

0.000

1.36

0.25

0.265

0.258

10.7

391.9

116

402.6

119

4.5

110

301.5

103

113

306.1

106

0.38

0.859

0.762

Between treatments

0.6

3.1

1.7

Within treatments

37.6

115

191.7

110

233.8

103

Total

38.2

118

194.9

113

235.5

106

Between treatments

0.4

2.9

1.1

Within treatments

15.7

115

247.1

110

136.7

103

Total

16.1

118

250.0

113

137.8

106

0.67

1.02

0.572

0.378

4. Discussion The diameter increment in P. brutia was reduced by pruning when the above intensities were applied. Annual diameter increments for 50% and 75% pruning intensities were statistically different from the control,

370

»Kursunlu«

0.60

0.435

0.615

0.728

unlike the 25% pruning intensity treatment (Fig. 1). Thus, it can be concluded that, in all cases, pruning of about 25% of the lower crown of trees from ground level has no effect on diameter growth. Under low levels of light intensity, needles on lower branches may respire as much or conceivably more than they can Croat. j. for. eng. 37(2016)2

Effect of Pruning on Diameter Growth in Pinus brutia Ten. Plantations in Turkey (365–373)

N. Erkan et al.

2013 (14)

Sub groups

Mean, mm

Sub groups

Mean, mm

Sub groups

Mean, mm

Sub groups

Mean, mm

Sub groups

Mean, mm

Sub groups

Total DBH inside bark, cm

Mean, mm

2011 (12)

Sub groups

2009 (10)

Mean, mm

2007 (8)

Sub groups

2006 (7)

Mean, mm

2005 (6)

Sub groups

2004 (5)

Mean, mm

2003 (4)

Sub groups

2002 (3)

Mean, mm

»Nebiler«

2001 (2)

Treatments

Control

4.34

2.49

3.21

5.86

4.26

5.21

3.99

2.72

1.07

1.46

0.91

11.56

25% pruned 4.07

2.61

3.41

5.53

4.35

5.08

3.90

2.39

0.83

1.34

0.75

11.46

50% pruned 4.29

2.62

3.40

5.58

4.33

5.21

4.31

2.78

1.06

1.49

0.82

11.61

75% pruned 4.15

1.65

1.71

2.79

2.88

3.82

3.25

2.30

1.01

1.32

0.83

10.52

9.19

5.17

8.04

7.75

5.65

5.53

4.81

4.89

3.52

2.82

3.48

15.40

25% pruned 9.17

5.11

8.04

7.88

5.57

5.65

4.72

4.94

3.01

2.77

3.36

15.29

50% pruned 8.96

4.07

6.00

6.70

5.30

5.75

4.87

5.15

3.65

3.09

3.69

15.11

75% pruned 9.55

3.08

3.70

4.69

4.49

5.08

4.28

4.88

3.52

3.16

3.75

14.39

7.71

7.07

8.52

8.32

7.51

6.27

5.00

4.60

3.14

2.41

2.29

16.25

25% pruned 7.60

7.12

8.75

8.19

7.75

6.62

5.54

5.28

3.26

2.47

16.57

50% pruned 7.14

6.01

7.20

7.30

7.26 a b 6.13

5.42

5.00

3.69

2.29

2.34

15.99

75% pruned 7.48

3.56

4.16

5.50

6.28

5.90

4.56

3.47

2.66

2.19

15.15

Control

Control »Kursunlu«

2000 (1) Sub groups

Years*

Mean, mm

»Bük«

Experimental sites

Table 3 Duncan test results for DBH increment following pruning year for different pruning intensities at three test sites (p<0.05 level)

6.10

* Observation years and number of years passed since pruning (in parenthesis)

photosynthesize. Therefore, these branches contribute little to the growth and may even be a burden to the tree resources (Savill et al. 1997, Kozlowski and Pallardy 1990, Montagu et al. 2003). Moreover, Savill et al. (1997), citing research conducted in Chryptomeria japonica by Wang et al. (1980), inform that removing 10% or slightly more of the live crown actually improved the growth. Savill et al. (1997), referring to research done in Europe and North America, also indicate that pruning more than 1/3 of canopy will reduce diameter growth. Uotila and Mustonen (1994), in a study on Scots pine (Pinus sylvestris L.), found that growth reduction was statistically significant (up to 33% decrease in diameter growth compared to the control) when 40% or more of the live crown was removed by pruning. The photosynthetic surface of a tree crown, which is directly related to growth, is reduced at least for a certain period after pruning. However, in the remaining foliage, which generally functions below its maximum photosynthetic capacity, photosynthetic activity is enhanced during the years after pruning (Lovett Doust 1989). Alcorn et al. (2008) reported that in most artificial defoliation studies photosynthetic activity remained unchanged initially, and then increased following leaf-area recovery to levels even above those Croat. j. for. eng. 37(2016)2

of plants that had not been defoliated. It, therefore, appears that a pruned tree, by increasing photosynthetic activities in its remaining foliage, allocates its energy to rapidly restore crown loss, which consequently leads to reduction of diameter growth. In short, trees recover leaf-area loss and consequently growth loss by increasing their photosynthetic activity following defoliation. Each observation tree was surrounded by unpruned buffer trees to avoid any unequal competition and border effects. In this way it was expected that observation trees would be exposed to more or less equal competition, the only difference being the differential pruning effect on the trees. It was observed that the duration of the recovery period from disturbance due to pruning varied at different experimental sites depending on site quality (Fig. 1). For example, recovery from pruning effects at »Bük« experimental site, which has the lowest site quality among the three test sites, lasted longer than the other two sites (6 years vs. 4 years). This implies that on good sites, trees perform better in producing new needles and branches in order to compensate for growth reduction caused by removed foliage and branches. It appears that any intervention on stands, including pruning, influences crown dynamics de-

371

N. Erkan et al.

Effect of Pruning on Diameter Growth in Pinus brutia Ten. Plantations in Turkey (365–373)

pending on species, site quality and age at which the intervention is applied (Forrester et al. 2010). Additionally, 50% pruning had no significant effect on DBH increment at »Bük« as opposed to »Nebiler« and »Kursunlu«. This can be explained by a higher response to silvicultural treatments on better sites compared to poor sites, which means that trees on less productive sites suffer from nutrients and water deficits more than they do from solar radiation. At »Bük« test site, annual DBH increment under 75% pruning intensity was significantly lower than those under three other intensities (control, 25% and 50% pruning) during the first six years after pruning (2001–2006). However, this difference disappeared in the 7th year (in 2007). At those test sites with higher site quality, »Nebiler« and »Kursunlu«, annual DBH increments both under 50% and 75% pruning intensities were statistically lower than those under control and 25% pruning intensities (Fig. 1, Table 2 and 3). The results show that the growth rate of Pinus brutia is higher on good sites, and that tree canopy can recover faster than that on poor sites, reaching the growth rate of the control group. Indeed, Endo and Mesa (1992) suggested heavier pruning, based on prior studies, on sites with higher site quality because faster recovery is possible on such sites. Obviously, reduction in annual DBH increment would result in reduction of cumulative DBH and volume growth collectively. The results showed that pruning up to 75% of tree height reduced overall DBH growth by 9.0%, 6.5% and 6.7% in »Bük«, »Nebiler« and »Kursunlu«, respectively, some 14 years after pruning (Fig. 2). Estimates of stem volume production over 14 years suggested that 75% pruning would reduce standing volume (m3/ha) by 18%, 19% and 21% for the three sites, respectively. Endo and Mesa (1992) conducted a study in Colombia on 3.5 year old Pinus patula plantations, pruned to an intensity of 30, 50 and 70% of total canopy. Based on analyses conducted 4.5 years after pruning, they also found that the 70% pruning treatment caused statistically significant reductions in volume increment per ha, and thus suggested a lower level of pruning for the species. Pruning is targeted for providing cleartrunks in the final crop for more than 20 cm in mean stand diameter, which provide higher grade lumber. P. brutia can reach this diameter at approximately 40 to 70 years depending on site quality (Erkan 1996). Erkan et al. (2010) made an economical evaluation of pruning for P. brutia and showed that the internal rate of revenue (IRR) can reach up to 10% for 40 years rotation period for good sites. Longer rotation periods have less IRR due to the discount rate of money spent as pruning cost.

372

It should also be noted that there was no negative effect of pruning on DBH increment in the first growing season following pruning. This result implies that a tree stores nutrition in its different parts so that it compensates for the loss resulting from crown reduction during the first growing season after pruning. Kukpa (2007) found similar results in a study conducted on cherry trees.

Acknowledgements This study was funded by the Turkish General Directorate of Forestry. The authors are grateful to Dr. Kani Isik from Akdeniz University and Steven Speed, working at Office National des Forêts (ONF, Paris, France) for their review and valuable suggestions on the earlier version of the manuscript, and to Mr. Arnold J. Grayson, former Director of Research of the British Forestry Commission for reviewing the English language of the text.

5. References Alcorn, P.J., Bauhus, J., Thomas, D.S., James, R.N., Smith, R.G.B., Nicotra, A.B., 2008: Photosynthetic response to green crown pruning in young plantation-grown Eucalyptus pilularis and E. cloeziana. Forest Ecology and Management 255(11): 3827–3838. Endo, M., Mesa, G.V., 1992: Results of a Pruning Trial with Pinus patula in Colombia. IPEF International 2: 45–59. Erkan, N., 1996: Stand simulation for Pinus brutia Ten. SoutEastern Anatolia Forest Research Institute, Technical Bulletin No. 1, Elazig, 96 p. Erkan, N., 1998: Growth analyses for the plantations of Cedrus libani and Pinus nigra in Elazig Region. South-Eastern Anatolia Forest Research Institute, Technical Bulletin 3, Elazig, 44 p. Erkan, N., Uzun, E., Aydin, A.C., Bas, M.N., 2010: The effects of pruning on growth of Pinus brutia plantations and its economy. Southwest Anatolia Forestry Research Institute, Technical Bulletin 38, Antalya, 60 p. Forrester, D.I., Medhurst, J.L., Wood, M., Beadle, C.L., Valencia, C.L., 2010: Growth and physiological responses to silviculture for producing solid-wood products from Eucalyptus plantations: An Australian perspective. Forest Ecology and Management 259(9): 1819-1835. Guller, B., Isik, K., Cetinay, S., 2012: Variations in the radial growth and wood density components in relation to cambial age in 30-year-old Pinus brutia Ten. at two test sites. Trees 26(3): 975–986. Harold, E.Y., Paul, J.K., 1952: The Effect of Pruning on the Height and Diameter Growth of Loblolly Pine. Journal of Forestry 50(6): 474–479. Croat. j. for. eng. 37(2016)2

Effect of Pruning on Diameter Growth in Pinus brutia Ten. Plantations in Turkey (365–373) Kalipsiz, A., 1982: Orman Hasilat Bilgisi Ders Kitabi. I. U. Orman Fakultesi Yayin No:328, Istanbul, 349 p. Kozlowski, T.T., Pallardy, S.G., 1990: Growth Control in Woody Plants. Academic Press, London, ISBN: 978-0-12424210-4, 641 p. Kukpa, I., 2007: Growth Reaction of Young Wild Cherry (Prunus avium L.) Trees to Pruning. Journal of Forest Science 53(12): 555–560. Lovett Doust, J., 1989: Plant reproductive strategies and resource allocation. Trends in Ecology and Evolution 4(8): 230–234. Maclaren, J.P., 1993: Radiata Pine Growers’ Manual. New Zealand Forest Research Institute, Bulletin184, 140 p. Miraboglu, M., 1983: Orman Isletme Iktisadi Ders Kitabi. I.U. Orman Fakultesi Yayin No: 340, Istanbul, 248 p. Montagu, K.D., Kearney, D.E., Smith, R.G.B., 2003: The biology and silviculture of pruning planted eucalypts for clear wood production, a review. Forest Ecology and Management 179(1–3): 1–13. OGM, 2014: Silvikulturel uygulamalarin temel esaslari. Teblig No:298, Ankara, 142 p. OGM, 2012: Forest areas in Turkey. General Directorate of Forestry, Forest Management and Planning Department, Publication No: 85, Ankara, 27 p.

N. Erkan et al.

OGM, 2013: Report for forestry statistics in Turkey. Unpublished documents, General Directorate of Forestry, Exploitation and Marketing Department, Ankara, 6 p. (available at: http://www.ogm.gov.tr /ekutuphane/Sayfalar/Istatistikler. aspx?). Saatcioglu, F., 1971: Orman Bakimi Ders Kitabi. I.U. Orman Fakültesi Yayin No:160, Istanbul, 303 p. Savill, P., Evans, J., Auclair, D., Falck, J., 1997: Plantation Silviculture in Europe. Oxford University Press, 297 p. Schmidt, L.S., Wardle, T.D., 2002: Impact of Pruning Eastern Redcedar (Juniperus virginiana). Western Journal of Applied Forestry 17(4): 189–193. SPSS Inc., 2015: SPSS 22.0 Guide to Data Analysis, Published by Prentice Hall, New Jersey, USA, 637 p. Uotila, A., Mustonen, S., 1994: The effect of different levels of green pruning on the diameter growth of Pinus sylvestris L.. Scandinavian Journal of Forest Research 9(3): 226-232. Usta, H.Z., 1991: A study on the yield of Pinus brutia Ten Plantations. Turkish Forest Research Institute, Technical Bulletin 219, Ankara, 138 p. Wang, W.C., Pinto, J.P., Yung, Y.L., 1980: Climatic effects due to halogenated compounds in the Earth’s atmosphere. J. Atmos. Sci. 37(2): 333–338.

Authors’ address:

Received: March 11, 2015. Accepted: December12, 2015. Croat. j. for. eng. 37(2016)2

Nesat Erkan, PhD.* e-mail: nesaterkan@yahoo.com Erdogan Uzun e-mail: erdoganuzun@ogm.gov.tr Ali Cem Aydin e-mail: alicemfb77@hotmail.com Mustafa Necati Bas e-mail: necati_bas@yahoo.com Southwest Anatolian Forest Research Institute P.O.Box 264 07010 Antalya TURKEY * Corresponding author

373