Journal of Mechanical Engineering 2013 1 by Darko Svetak

http://www.sv-jme.eu

59 (2013) 1

Strojniški vestnik Journal of Mechanical Engineering

Since 1955

Papers

Blaž Florjanič, Edvard Govekar, Karl Kuzman: Neural Network-Based Model for Supporting the Expert Driven Project Estimation Process in Mold Manufacturing

Yikai Chen, Jie He, Mark King, Wuwei Chen, Changjun Wang, Weihua Zhang: Model Development and Dynamic Load-Sharing Analysis of Longitudinal-Connected Air Suspensions

Liane Roldo, Ivan Komar, Nenad Vulić: Design and Materials Selection for Environmentally Friendly Ship Propulsion System

Lovro Kuščer, Janez Diaci: Measurement Uncertainty Assessment in Remote Object Geolocation

Hirpa G. Lemu, Tomasz Trzepieciński: Numerical and Experimental Study of Frictional Behavior in Bending Under Tension Test

Andrew J. Dick: Characterizing Effective d31 Values for PZT from the Nonlinear Oscillations of Clamped-Clamped Micro- Resonators

Andrej Debenjak, Matej Gašperin, Boštjan Pregelj, Maja Atanasijević-Kunc, Janko Petrovčič, Vladimir Jovan: Detection of Flooding and Drying inside a PEM Fuel Cell Stack

Journal of Mechanical Engineering - Strojniški vestnik

Contents

1 year 2013 volume 59 no.

Technical Editor Pika Škraba University of Ljubljana Faculty of Mechanical Engineering, Slovenia Editorial Office University of Ljubljana (UL) Faculty of Mechanical Engineering SV-JME Aškerčeva 6, SI-1000 Ljubljana, Slovenia Phone: 386-(0)1-4771 137 Fax: 386-(0)1-2518 567 E-mail: info@sv-jme.eu, http://www.sv-jme.eu

Print DZS, printed in 450 copies

Founders and Publishers University of Ljubljana (UL) Faculty of Mechanical Engineering, Slovenia University of Maribor (UM) Faculty of Mechanical Engineering, Slovenia Association of Mechanical Engineers of Slovenia

http://www.sv-jme.eu

59 (2013) 1

Chamber of Commerce and Industry of Slovenia Metal Processing Industry Association

Since 1955

Strojniški vestnik Journal of Mechanical Engineering

rd Govekar, Karl Kuzman: sed Model for Supporting the Expert Driven Project in Mold Manufacturing

Mark King, Wuwei Chen, Changjun Wang, Weihua Zhang: t and Dynamic Load-Sharing Analysis of ected Air Suspensions

omar, Nenad Vulić: ls Selection for Environmentally Friendly Ship

Cover: Cover photo shows 3D CAD model of a mold for injection molding of thermoplastic polymers consisting of injection mold half and ejection mold half. Shown layout is typically used in automotive industry projects where the injection mold holds mirrored part geometry. These are usually referred as 1+1 cavity molds.

ctive d31 Values for PZT from the Nonlinear Oscillations ed Micro- Resonators

atej Gašperin, Boštjan Pregelj, Maja Atanasijević-Kunc, dimir Jovan: ng and Drying inside a PEM Fuel Cell Stack

Journal of Mechanical Engineering - Strojniški vestnik

Diaci: rtainty Assessment in Remote Object Geolocation

asz Trzepieciński: erimental Study of Frictional Behavior in Bending

year

no. 1 2013 59

Image Courtesy: iMold d.o.o., Slovenia and University of Ljubljana, Faculty of Mechanical Engineering, Slovenia

volume

International Editorial Board Koshi Adachi, Graduate School of Engineering,Tohoku University, Japan Bikramjit Basu, Indian Institute of Technology, Kanpur, India Anton Bergant, Litostroj Power, Slovenia Franci Čuš, UM, Faculty of Mech. Engineering, Slovenia Narendra B. Dahotre, University of Tennessee, Knoxville, USA Matija Fajdiga, UL, Faculty of Mech. Engineering, Slovenia Imre Felde, Bay Zoltan Inst. for Mater. Sci. and Techn., Hungary Jože Flašker, UM, Faculty of Mech. Engineering, Slovenia Bernard Franković, Faculty of Engineering Rijeka, Croatia Janez Grum, UL, Faculty of Mech. Engineering, Slovenia Imre Horvath, Delft University of Technology, Netherlands Julius Kaplunov, Brunel University, West London, UK Milan Kljajin, J.J. Strossmayer University of Osijek, Croatia Janez Kopač, UL, Faculty of Mech. Engineering, Slovenia Franc Kosel, UL, Faculty of Mech. Engineering, Slovenia Thomas Lübben, University of Bremen, Germany Janez Možina, UL, Faculty of Mech. Engineering, Slovenia Miroslav Plančak, University of Novi Sad, Serbia Brian Prasad, California Institute of Technology, Pasadena, USA Bernd Sauer, University of Kaiserlautern, Germany Brane Širok, UL, Faculty of Mech. Engineering, Slovenia Leopold Škerget, UM, Faculty of Mech. Engineering, Slovenia George E. Totten, Portland State University, USA Nikos C. Tsourveloudis, Technical University of Crete, Greece Toma Udiljak, University of Zagreb, Croatia Arkady Voloshin, Lehigh University, Bethlehem, USA President of Publishing Council Jože Duhovnik UL, Faculty of Mechanical Engineering, Slovenia General information Strojniški vestnik – Journal of Mechanical Engineering is published in 11 issues per year (July and August is a double issue). Institutional prices include print & online access: institutional subscription price and foreign subscription €100,00 (the price of a single issue is €10,00); general public subscription and student subscription €50,00 (the price of a single issue is €5,00). Prices are exclusive of tax. Delivery is included in the price. The recipient is responsible for paying any import duties or taxes. Legal title passes to the customer on dispatch by our distributor. Single issues from current and recent volumes are available at the current single-issue price. To order the journal, please complete the form on our website. For submissions, subscriptions and all other information please visit: http://en.sv-jme.eu/. You can advertise on the inner and outer side of the back cover of the magazine. The authors of the published papers are invited to send photos or pictures with short explanation for cover content. We would like to thank the reviewers who have taken part in the peerreview process.

ISSN 0039-2480 © 2013 Strojniški vestnik - Journal of Mechanical Engineering. All rights reserved. SV-JME is indexed / abstracted in: SCI-Expanded, Compendex, Inspec, ProQuest-CSA, SCOPUS, TEMA. The list of the remaining bases, in which SV-JME is indexed, is available on the website.

The journal is subsidized by Slovenian Book Agency. Strojniški vestnik - Journal of Mechanical Engineering is also available on http://www.sv-jme.eu, where you access also to papers’ supplements, such as simulations, etc.

Instructions for Authors All manuscripts must be in English. Pages should be numbered sequentially. The maximum length of contributions is 10 pages. Longer contributions will only be accepted if authors provide justification in a cover letter. Short manuscripts should be less than 4 pages. For full instructions see the Authors Guideline section on the journal’s website: http://en.sv-jme.eu/. Please note that file size limit at the journal’s website is 8Mb. Announcement: The authors are kindly invited to submitt the paper through our web site: http://ojs.sv-jme.eu. Please note that file size limit at the journal’s website is 8Mb. The Author is also able to accompany the paper with Supplementary Files in the form of Cover Letter, data sets, research instruments, source texts, etc. The Author is able to track the submission through the editorial process - as well as participate in the copyediting and proofreading of submissions accepted for publication - by logging in, and using the username and password provided. Please provide a cover letter stating the following information about the submitted paper: 1. Paper title, list of authors and affiliations. 2. The type of your paper: original scientific paper (1.01), review scientific paper (1.02) or short scientific paper (1.03). 3. A declaration that your paper is unpublished work, not considered elsewhere for publication. 4. State the value of the paper or its practical, theoretical and scientific implications. What is new in the paper with respect to the state-of-the-art in the published papers? 5. We kindly ask you to suggest at least two reviewers for your paper and give us their names and contact information (email). Every manuscript submitted to the SV-JME undergoes the course of the peer-review process. THE FORMAT OF THE MANUSCRIPT The manuscript should be written in the following format: - A Title, which adequately describes the content of the manuscript. - An Abstract should not exceed 250 words. The Abstract should state the principal objectives and the scope of the investigation, as well as the methodology employed. It should summarize the results and state the principal conclusions. - 6 significant key words should follow the abstract to aid indexing. - An Introduction, which should provide a review of recent literature and sufficient background information to allow the results of the article to be understood and evaluated. - A Theory or experimental methods used. - An Experimental section, which should provide details of the experimental set-up and the methods used for obtaining the results. - A Results section, which should clearly and concisely present the data using figures and tables where appropriate. - A Discussion section, which should describe the relationships and generalizations shown by the results and discuss the significance of the results making comparisons with previously published work. (It may be appropriate to combine the Results and Discussion sections into a single section to improve the clarity). - Conclusions, which should present one or more conclusions that have been drawn from the results and subsequent discussion and do not duplicate the Abstract. - References, which must be cited consecutively in the text using square brackets [1] and collected together in a reference list at the end of the manuscript. Units - standard SI symbols and abbreviations should be used. Symbols for physical quantities in the text should be written in italics (e.g. v, T, n, etc.). Symbols for units that consist of letters should be in plain text (e.g. ms-1, K, min, mm, etc.) Abbreviations should be spelt out in full on first appearance, e.g., variable time geometry (VTG). Meaning of symbols and units belonging to symbols should be explained in each case or quoted in a special table at the end of the manuscript before References. Figures must be cited in a consecutive numerical order in the text and referred to in both the text and the caption as Fig. 1, Fig. 2, etc. Figures should be prepared without borders and on white grounding and should be sent separately in their original formats. Pictures may be saved in resolution good enough for printing in any common format, e.g. BMP, GIF or JPG. However, graphs and line drawings should be prepared as vector images, e.g. CDR, AI. When labeling axes, physical quantities, e.g. t, v, m, etc. should be used whenever possible to minimize the need to label the axes in two languages. Multi-curve graphs should have individual curves marked with a symbol. The meaning of the symbol should be explained in the figure caption. Tables should carry separate titles and must be numbered in consecutive numerical order in the text and referred to in both the text and the caption as

Table 1, Table 2, etc. In addition to the physical quantity, e.g. t (in italics), units (normal text), should be added in square brackets. The tables should each have a heading. Tables should not duplicate data found elsewhere in the manuscript. Acknowledgement of collaboration or preparation assistance may be included before References. Please note the source of funding for the research. REFERENCES A reference list must be included using the following information as a guide. Only cited text references are included. Each reference is referred to in the text by a number enclosed in a square bracket (i.e., [3] or [2] to [6] for more references). No reference to the author is necessary. References must be numbered and ordered according to where they are first mentioned in the paper, not alphabetically. All references must be complete and accurate. All non-English or. non-German titles must be translated into English with the added note (in language) at the end of reference. Examples follow. Journal Papers: Surname 1, Initials, Surname 2, Initials (year). Title. Journal, volume, number, pages, DOI code. [1] Hackenschmidt, R., Alber-Laukant, B., Rieg, F. (2010). Simulating nonlinear materials under centrifugal forces by using intelligent crosslinked simulations. Strojniški vestnik - Journal of Mechanical Engineering, vol. 57, no. 7-8, p. 531-538, DOI:10.5545/sv-jme.2011.013. Journal titles should not be abbreviated. Note that journal title is set in italics. Please add DOI code when available and link it to the web site. Books: Surname 1, Initials, Surname 2, Initials (year). Title. Publisher, place of publication. [2] Groover, M.P. (2007). Fundamentals of Modern Manufacturing. John Wiley & Sons, Hoboken. Note that the title of the book is italicized. Chapters in Books: Surname 1, Initials, Surname 2, Initials (year). Chapter title. Editor(s) of book, book title. Publisher, place of publication, pages. [3] Carbone, G., Ceccarelli, M. (2005). Legged robotic systems. Kordić, V., Lazinica, A., Merdan, M. (Eds.), Cutting Edge Robotics. Pro literatur Verlag, Mammendorf, p. 553-576. Proceedings Papers: Surname 1, Initials, Surname 2, Initials (year). Paper title. Proceedings title, pages. [4] Štefanić, N., Martinčević-Mikić, S., Tošanović, N. (2009). Applied Lean System in Process Industry. MOTSP 2009 Conference Proceedings, p. 422-427. Standards: Standard-Code (year). Title. Organisation. Place. [5] ISO/DIS 16000-6.2:2002. Indoor Air – Part 6: Determination of Volatile Organic Compounds in Indoor and Chamber Air by Active Sampling on TENAX TA Sorbent, Thermal Desorption and Gas Chromatography using MSD/FID. International Organization for Standardization. Geneva. www pages: Surname, Initials or Company name. Title, from http://address, date of access. [6] Rockwell Automation. Arena, from http://www.arenasimulation.com, accessed on 2009-09-07. EXTENDED ABSTRACT By the time the paper is accepted for publishing, the authors are requested to send the extended abstract (approx. one A4 page or 3.500 to 4.000 characters). The instructions for writing the extended abstract are published on the web page http://www.sv-jme.eu/ information-for-authors/. COPYRIGHT Authors submitting a manuscript do so on the understanding that the work has not been published before, is not being considered for publication elsewhere and has been read and approved by all authors. The submission of the manuscript by the authors means that the authors automatically agree to transfer copyright to SV-JME and when the manuscript is accepted for publication. All accepted manuscripts must be accompanied by a Copyright Transfer Agreement, which should be sent to the editor. The work should be original by the authors and not be published elsewhere in any language without the written consent of the publisher. The proof will be sent to the author showing the final layout of the article. Proof correction must be minimal and fast. Thus it is essential that manuscripts are accurate when submitted. Authors can track the status of their accepted articles on http://en.svjme.eu/. PUBLICATION FEE For all articles authors will be asked to pay a publication fee prior to the article appearing in the journal. However, this fee only needs to be paid after the article has been accepted for publishing. The fee is 300.00 EUR (for articles with maximum of 10 pages), 20.00 EUR for each addition page. Additional costs for a color page is 90.00 EUR.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1 Contents

Contents Strojniški vestnik - Journal of Mechanical Engineering volume 59, (2013), number 1 Ljubljana, January 2013 ISSN 0039-2480 Published monthly

Papers Blaž Florjanič, Edvard Govekar, Karl Kuzman: Neural Network-Based Model for Supporting the Expert Driven Project Estimation Process in Mold Manufacturing Yikai Chen, Jie He, Mark King, Wuwei Chen, Changjun Wang, Weihua Zhang: Model Development and Dynamic Load-Sharing Analysis of Longitudinal-Connected Air Suspensions Liane Roldo, Ivan Komar, Nenad Vulić: Design and Materials Selection for Environmentally Friendly Ship Propulsion System Lovro Kuščer, Janez Diaci: Measurement Uncertainty Assessment in Remote Object Geolocation Hirpa G. Lemu, Tomasz Trzepieciński: Numerical and Experimental Study of Frictional Behavior in Bending Under Tension Test Andrew J. Dick: Characterizing Effective d31 Values for PZT from the Nonlinear Oscillations of Clamped-Clamped Micro- Resonators Andrej Debenjak, Matej Gašperin, Boštjan Pregelj, Maja Atanasijević-Kunc, Janko Petrovčič, Vladimir Jovan: Detection of Flooding and Drying inside a PEM Fuel Cell Stack List of reviewers in 2012

3 14 25 32 41 50 56 65

Received for review: 2011-07-12 Received revised form: 2012-08-22 Accepted for publication: 2012-10-25

Neural Network-Based Model for Supporting the Expert Driven Project Estimation Process in Mold Manufacturing Florjanič, B. – Govekar, E. – Kuzman, K. Blaž Florjanič1,* – Edvard Govekar2 – Karl Kuzman2 1 iMold d.o.o., Slovenia

2 University

of Ljubljana, Faculty of Mechanical Engineering, Slovenia

One of the crucial activities for running a successful mold manufacturing business is project estimation. The estimation process is an early project activity which is usually handled by highly skilled, in-house experts. One of the most important parameters affecting the estimation process is the volume of manufacturing hours (VMH) to produce the mold. This article suggests how to address the problem of estimating the volume of manufacturing hours by using the support of an artificial neural network (ANN) model, and its inclusion into the expert driven project estimation process. Based on the histogram of ANN estimations the percentage of unwanted underestimations of the VMH can be estimated as well and decreased by an introduced safety factor. The developed model-based estimation enables an expert to improve project estimation by using easily obtainable input data. Keywords: mold making, manufacturing, artificial neural networks, estimation process

0 INTRODUCTION The mold making industry is project driven, and as such it has to cope with the characteristics of an individual production process. One of the major sources of risk in project management is the inaccurate forecast of project costs, demand, and other impacts [1]. In the mold production process it is crucial to minimize uncertainty in the early project estimation phase. The estimation phase is commonly a human expert driven activity which is sensitive to the expert’s bias. This bias can lead to an underestimation of project resources when the estimator is overconfident, or to over-estimation of project resources when the estimator does not have sufficient confidence that all aspects of the project can be properly covered. Both scenarios have a negative impact on future business. In the case of underestimation, the project will bring economic loss, and in the case of overestimation, it will most likely be assigned to a competitive supplier. The estimator’s key competence is to properly collect and evaluate all significant information for making the project estimation successful. The contradiction lies in the fact that the estimator should spend minimal time necessary on estimation activity since usually less than 10% of all offers turn into orders in the mold making industry, as stated in [2] to [4]. Estimations in the mold manufacturing business still rely heavily on intuitive methods, which are subjective and prone to reliability and repeatability problems. A solution for these problems is addressed in this article, with the development of a supported expert driven project estimation process. In the project estimation process the volume of manufacturing hours represents one of the most

important pieces of information. It reflects the majority of costs in the final project price, and it most significantly shapes the project schedule. The research objective is to develop an ANN-supported, expert driven project estimation process to improve the estimation of the volume of manufacturing hours in the mold production. In addition to the development of a reliable estimation model, it is also very important to properly position the supporting model in the expert driven estimation process. Therefore, in addition to model building, the problem of proper position of the supporting model will be addressed in the paper. Following these aims, first an overview of estimation process is given. Then, the solution for the problem of proper placement of an estimation support model is addressed. Furthermore, the proposed ANN-based model for estimation of the volume of manufacturing hours is presented. Finally, the results of ANN modelling are presented and discussed. 1 THE PROJECT ESTIMATION PROCESS A major challenge of the project estimation process in general is to achieve sufficient project estimation reliability within minimal time consumption for this operation. Estimation reliability is directly related to the amount and quality of the data available at the moment the estimation process takes place. As shown in Fig. 1, the availability of data differs during different project stages. As we move along the timeline of the project the availability of data increases. Consequently, estimation uncertainty and risks decrease, so more accurate and reliable results can be expected. Estimation methods differ in

*Corr. Author’s Address: iMold d.o.o., Cesta v Pečale 33, 1231 Ljubljana-Črnuče, Slovenia, blaz.florjanic@imold.si

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 3-13

accordance with the project stage in which they are used [3], [5] and [6] and are divided into: • intuitive, • analogical, • parametric, • analytical. Intuitive estimation methods are based on the human expert’s prior knowledge and experience. A major downside of these methods is that results are very susceptible to many different subjective factors. So, the results obtained face problems regarding reliability and repeatability. These problems can be reduced to a certain extent by applying methods that use more than one estimator or estimation method [7]. A major benefit of these methods is moderate time consumption. They are usually applied in early project stages. Analogical estimation methods are based on finding successful projects with similar characteristics like the one being estimated. On the basis of detected similarities corresponding values are assigned to the estimated project. These methods become applicable when the basic product shape is defined. They are also considered as conditionally reliable methods since the relations between similarities are usually estimated by an expert [7]. Their main strengths are transparency of gained results and the ability to achieve the solution rapidly. These methods strongly rely on the database of previous projects, and become unreliable if proper mapping of similar characteristics cannot be obtained. IDEA/PRODUCT DEFINITION

PRODUCT DESIGN

CONCEPT

ENGINEERING DESIGN

Parametric estimation methods are used to make estimations on the basis of parameters that are able to directly translate the properties of the product or project into an estimated value. These methods are built on the databases of past projects. Estimations are obtained by collecting input parameters and processing those to formulate a proper estimation impact. These methods are usually seen as ‘black box’ solutions. A major challenge is in defining a proper set of input parameters. These methods offer both speed and sufficient reliability if used properly. By keeping the database of a past project open and adding the data of new projects, this model gains the ability of adaptation and learning, which comes forward significantly when used properly with ANN platforms. Parametric methods are prone to use both parametric (e.g. multiple regression-model) and nonparametric models (e.g. ANN model), which were all found to give acceptable estimates. Analytical estimation methods are usually applicable in the later stages of the product life cycle, when both product data and manufacturing technology are defined in details. The estimation is made on a detailed breakdown of the complete process into elementary tasks [8]. For every task the relations between inputs and corresponding outputs are analytically determined. These methods are usually rigid and relations between parameters are not easily modified. They do not have adaptation ability [8]. Gained results give the most accurate estimations. TESTING and PROTOTYPING

INDUSTRIALIZATION*

SERIAL PRODUCTION

EXPLOITATION

Intuitive methods Analogical methods Parametric methods Analytical methods

Data definition/availability Estimation reliability, Time consumption for calculation

Risk Uncertainty * Manufacturing technology definition

Fig. 1. Estimation methods applicable in different stages of the project

Florjanič, B. – Govekar, E. – Kuzman, K.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 3-13

Their major downsides are time consumption and limited applicability in the early project stages. A review of significant, recently published literature and articles dealing with estimation methods is presented in Table 1. They are sorted with regard to the used estimation method. The table also defines the industry for which research was done and what problem they were trying to solve. From the literature it is evident that the majority of research activities, related to the problems of project estimation, are focused on defining estimation models that should be able to define the link between geometric characteristics of the product and price/cost of the product/project. By focusing on these economic values, the estimating process is contaminated

with influences that do not possess technical and technological characteristics of the manufacturing process. These are actually influences of the market, reflecting request and demand, and have very little to do with technological issues. Articles which are the most significant for this research are related to product complexity [9] to [11], and the implementation of ANN in the mold production estimation process [9] and [12]. All these approaches give quite accurate estimates only when used for very specific types of products. The above mentioned articles offer the solution of the complex estimation process by using a single estimation model, taking into account all its limitations and benefits. The idea presented in this article is to

Table 1. Literature overview

[3] Duverlie and Castelain [13] Wang et al.

METHOD SUB-TYPE (ANN, Regression, Case-Based Reasoning, etc.) Retrieval of similar data from database Case-based Reasoning Case-based Reasoning

[5] Ficko et al.

Case-based Reasoning

PARAMETRIC

ANALOGICAL

METHOD

SOURCE [2] Fonseca et al.

Mold Making / Tools for injection molding Product Design Mold Making Mold Making / Tools for Sheet Metal Forming

PROBLEM SOLVING Assisting mold quotation Cost estimation Mold cost estimation Manufacturing costs estimation for stamping tools

[6] Farineau et al.

Regression model

Product Design

Cost estimation

[9] Raviwongse and Allada

ANN

Mold Making

[12] Che

ANN

Mold Making and Injection molding

[14] Cavalier et al. [15] Farineau et al. [16] Elhag and Boussabine

Regression model, ANN Regression model Regression model, ANN

Automotive Product Design Construction/Buildings

[17] Verlinden et al.

Regression model, ANN

Mold complexity computation Product and mold cost estimation Production cost estimation Cost estimation Tender price estimation Sheet metal parts cost estimation

[19][18] Kim et al. [4] Denkena et al. [10] Fagade and Kazmer [11] Fagade and Kazmer

Regression model, ANN, Casebased Reasoning Rule-based

[19] Chan et al.

ANALYTICAL

INDUSTRY (Mold-making, Construction, etc.)

[20] Denkena et al.

Accessibility Analysis

[21] Chin and Wong

Decision Tables Boothroyd-Dewurst Dixon-Poli

[22] Fagade and Kazmer [23] Fagade and Kazmer [24] Nagahanumaiah et al. [25] Navodnik and Kopčič [26] Menges et al. [27] Kazmer

Construction/Buildings

Construction costs

Mold Making / Tools for die casting Mold Making and Injection molding Mold Making and Injection molding Mold Making / Tools for injection molding / Toy industry Mold Making/ Tools for injection molding and die casting Mold Making

Die cost calculation Lead time estimation Lead time estimation

Mold Making Mold Making and Injection molding Tools for injection molding and die casting Mold Making Mold Making Tools for injection molding and die casting Mold Making Mold Making

Mold cost estimation Manufacturing cost calculation Mold cost estimation Product and mold cost estimation Product and mold cost estimation Die or mold cost estimation Mold cost estimation Mold cost estimation Mold cost estimation

Neural Network-Based Model for Supporting the Expert Driven Project Estimation Process in Mold Manufacturing

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 3-13

develop an ANN supported project estimation process, which combines benefits from both intuitive methods and ANN estimation models. 2 ANN-SUPPORTED ESTIMATION PROCESS In general, expert estimations can represent a very broad solution space (see Fig. 2). This is mainly due to limited information availability; expert’s limited capability of simultaneously processing multiple information; and the expert’s bias. By using a supported expert estimation process the solution space gets narrower and the risk of underestimating or overestimating minimizes.

Fig. 2. Expert estimation solution space

The estimation process in the mold making business most commonly relies on human intuitive methods [5], or a combination of intuitive and analogical estimation methods. Mold makers put major emphasis on retrieving accurate project estimation with minimal time consumption, because a large number of quotations have to be processed in order to achieve sufficient order load. The reason for that lies in a very moderate success rate of all submitted offers. In order to achieve a sufficient level of result credibility, the estimation process has to be systematically approached. With this aim, a detailed step-by-step, expert driven, and ANN-supported estimation approach has been developed as shown schematically in Fig. 3. The complete estimation process consists of several phases: • Input data retrieval (IDR); • Conceptual design and Product manufacturability verification (CDPMV); • Resource estimation phase (REP); • Economic calculation phase (ECP). In IDR phase all input data necessary for completing the estimation is collected and evaluated. Having all the prescribed input data (a 3D CAD model 6

of product, a part drawing, and technical requirements for mold design) at disposal is a necessary condition for moving to the next phase. In the CDPMV phase an expert defines the basic mold concept, starting with: proper part orientation; undercut area definition; basic mold dimension definition; and mold subsystems definition. To support his/her decisions in this phase the expert usually uses set of design rules, decision trees, and a past mold design database. In the CDPMV phase the expert also verifies product manufacturability for the prescribed manufacturing technology, in this case injection molding. For this step commercial CEA software is available. In the REP phase an expert is faced with estimation of proper resources for a complete project. This is a crucial phase of the estimation process. To formulate estimation in this phase the expert usually relies on information from a mold material database, a post-calculation database, and a manufacturing technology database. The REP phase is followed by the ECP phase in which the estimation is translated into corresponding financial values. In the REP phase experts usually use intuitive estimation methods, which have the aforementioned reliability disadvantages. To minimize the problem regarding the reliability of the estimation results it is proposed to place estimation supporting model. The position of the supporting model in the estimation process shown in Fig. 2 is denoted in red colour. The estimation support can be achieved by different modelling methods like regression, ANN, support vector machines, etc. By applying the estimation supporting model the unsupported estimation process is upgraded to a supported estimation process (see Fig. 2). In this article the focus is on the most influential factor in the project estimation process – the volume of manufacturing hours (VMH). VMH is defined by:

VMH = ∑∑ ( tl + tm + tu ) , (1) P OP

and represents the total of all machining hours spent to complete all parts (P) of the mold. Only the hours when machines are actually occupied are taken into account. This means that at each operation (OP) machining time (tm), the loading time (tl) and unloading time (tu) are taken into account. To support the estimation of the VMH, the ANN-based model is used, which is described in the following section.

Florjanič, B. – Govekar, E. – Kuzman, K.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 3-13

Fig. 3. The systematic, expert driven project estimation process supported by ANN

Neural Network-Based Model for Supporting the Expert Driven Project Estimation Process in Mold Manufacturing

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 3-13

3 ARTIFICIAL NEURAL NETWORK MODEL ANNs are recognized as universal function approximators and can be efficiently used to model high dimensional and nonlinear relations [14]. They represent a valid alternative, especially when relationships are not known in either parametric or in an analytical form. This is an empirical model that learns from past examples and generalizes the solution for new cases. In our case, the purpose of ANN is to generate mapping from selected input data into a corresponding estimation of the VMH, based on learning by using empirical data without any prior knowledge of the mapping function. The ANN output retrieved from the model is categorized as an evaluation indicator for the expert to confirm their estimation or to re-evaluate and correct it accordingly. The methodology for the implementation of an ANN-based estimation of the VMH consists of three major phases: input variable definition; ANN architecture definition and training; and model validation, as shown in Fig. 4. After ANN architecture and input variables are optimized, and the ANN model performance is approved by an expert, it is ready for implementation as a support in the estimation process as presented in Fig. 3.

Fig. 4. General ANN-based estimation model creation

CAD model, part drawing, and special technical requirements [28], • technical requirements for the injection mold (TRFIM) that define the environment in which the mold will operate in serial production (molding facility), • mold design principles/rules (MDP/R). Production environment characteristics in which mold manufacturing takes place (mold shop equipment, organization, technology utilization, corporate culture, etc.) can also be used as ANN input variables. However, these characteristics are more applicable for estimations used in later project stages, when mold design is already completed. In the case when a cumulative variable like the VMH is observed, it can be presumed, that the production environment influence is already captured within the expected ANN output. These are the outputs that are collected through the samples described in Section 2.3. When the selection of ANN input variables was considered an expert opinion was taken into account. Based on this, 22 input variables were used of which 11 describe the MMPGOR, five describe the TRFIM, and six describe the MDP/R characteristics. Names and the corresponding variable value type are shown in Table 2.

Fig. 5. Dominant factors defining ANN inputs

2.1 Input Variable Definition When implementing an ANN model for the VMH estimation one of the most vital steps is to define an appropriate set of input variables that are presumably related to the VMH. In our case the VMH is mostly influenced by (see Fig. 5): • micro and macro part geometry and quality requirements (MMPGQR), prescribed with a 3D 8

2.2 ANN Architecture Definition To model a multivariable relation between the 22 selected input variables and the corresponding VMH value a multi-layer feed-forward network is used. For ANN training a Levenberg-Marquard learning rule is applied. It is a method which is fast and most appropriate for training moderate-sized, feed-forward neural networks [29].

Florjanič, B. – Govekar, E. – Kuzman, K.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 3-13

Table 2. ANN Input and output variables with corresponding value type, and encoding Real Value

Ejection

Real Value

Injection system

Part envelope height [mm] Part surface area [mm2] Part volume [mm3]

Real Value

MC,IS

Real Value

Cavity material, Injection side Cavity material, Ejection side Overall dimensional tolerance requirements of the part

Nominal part thickness [mm] Part material

Real Value

Mold length [mm]

Encoding 0=Simple/ Single stroke 1=Multiple strokes -1=Cold runner system 0=Combined system 1=Hot runner system 0=Non Hardened or Pre-Hardened 1=Hardened steel 0=Non Hardened or Pre- Hardened 1=Hardened steel 0=Class 4 (<0.5), Class 5 (<1), Class 6 (>1) 1=Class 3 (<0.1), Class 2 (<0.05), Class 1 (<0.01) Real Value

Mold width [mm]

Real Value

Envelope volume [mm3] Part complexity /Cavity detail

VE CXP

-1= Semi-crystalline 1=Amorphous Real Value -1=Simple/ Low detail 0=Moderately complex 1=Complex/ High detail

Mold height [mm] Parting line/surface complexity

HM CXPL

Surface finish, Injection side

SFIS

Number of sliders per cavity, Ejection side

NS,ES

Surface finish, Ejection side

SFES

0=Polished with sandpaper, Fine EDM, Fine milled/ Machined, etc. 1/2=High polished 1=High polished-Class A surfaces 0=Polished with sandpaper, Fine EDM, Fine milled/ Machined, etc. 1/2=High polished 1=High polished-Class A surfaces

Real Value -1=Simple / Flat 0=Moderately complex (Smoothly shaped, Small steps) 1=Free-form (Complex, nontangential surfaces, big steps) Real Value

INPUTS

Fig. 6. ANN initial architecture

DTP

Number of lifter cores NLC,ES per cavity, Ejection side

OUTPUTS Volume of manufacturing hours

As a performance function for feed-forward networks a mean square error (MSE), has been used, which defines the average squared difference between the network outputs and the target VMH Outputs. The initial ANN architecture is shown in Fig. 6. In addition to the 22 units in the input layer it consists of 10 neurons with a sigmoid activation function in the hidden layer, and a single output neuron with a linear activation function. The ANN structure is implemented in a MATLAB environment.

MC,ES

VMH

MDP/R

Encoding

TRFIM

MMPGQR

INPUTS Part envelope length [mm] Part envelope width [mm]

Real Value

2.3 Validation of the ANN Model Training and validation of ANN model relies on the large amount of samples comprised of ANN input and the corresponding target output data. Obtaining a large number of samples in an individual production, such as mold manufacturing, represents a certain obstacle, because companies hold this information as internal know-how. In our case 105 samples were obtained from a mid-sized mold shop. The samples were taken from automotive industry projects where the injection mold typically holds mirrored part geometry. These are usually referred as 1+1 cavity molds (see Fig. 7). By narrowing the research to a certain type of molds,

Neural Network-Based Model for Supporting the Expert Driven Project Estimation Process in Mold Manufacturing

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 3-13

improved results are expected, and a narrower and denser decision space is achieved. In order to overcome the obstacle of a restricted number of samples in ANN performance validation a multifold cross-validation procedure was used. For this purpose, a set of 105 input samples was randomized and divided into five subsets, each containing 21 samples. For each training the assigned subset was selected as a testing subset. The remaining four subsets were used for training. For statistical relevance of the ANN performance the ANN training and testing with the defined subsets was repeated five times. An average value of the output error was used as a measure of ANN performance for assigned testing and training subsets.

0.92545 is evident. Although this result is very encouraging, it is also very deceptive. To further analyze the ANN model performance, additional indicators are used. ANN model performance was characterized by relative percentage error (RPE) and mean absolute percentage error (MAPE) of the estimated VMH defined by Eqs. (2) and (3):

RPE =

MAPE =

yi − ti ⋅100, (2) ti 1 N

∑ i =1

yi − ti . (3) ti

Fig. 8. Scatter plot of network outputs vs. target outputs Fig. 7. Example of typical injection mold for automotive industry holding geometry for mirrored parts (left and right side of the vehicle)

Through an iteration process the number of neurons in the hidden layer was optimized, keeping in mind the fundamental ANN rules of minimizing the output error and keeping the network small. The final ANN architecture consists of four neurons in the hidden layer with a sigmoid activation function and one neuron with a linear activation function in the output layer. As ANN inputs in our case all 22 variables presented in Table 2 were used. 3 ANN MODEL ESTIMATION RESULTS An example of the comparison between network outputs and target outputs is shown in Fig. 8. From the figure a low scatter and an acceptable correlation between the target value and corresponding estimation of the VMH with a correlation coefficient 10

In the Eqs. (2) and (3) ti and yi denote target and by ANN estimated value of the VMH and N denotes the number of input samples. From the above defined errors (Eqs. (2) and (3)) the MAPE is used for statistical characterisation of ANN performance, whereas the RPE has an additional practical interpretation as negative and positive RPE correspond to underestimation and overestimation of VMH, respectively. While overestimation represents either profit or in the worst case, a non-competitive offer, underestimation means very dangerous nonprofitability of the project. The RPE for each sample i is shown in Fig. 9 and the corresponding histogram is presented in Fig. 10. From Fig. 10 it is evident that the majority (89.5%) of the results predicted the VMH values have an RPE in a range between –25 and +25%. However, the fact that in 4.8% of the predicted VMH values the corresponding RPE is below –25% should not be overlooked. In the most extreme case underestimation

Florjanič, B. – Govekar, E. – Kuzman, K.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 3-13

Fig. 9. RPE for each sample Table 3. Network output indicators min max

Output range [VMH] MAPE Mean absolute percentage error RPE Relative percentage error [%]

max min

Subset 1

Subset 2

Subset 3

Subset 4

Subset 5

NETWORK

384 1407 0.085 21.9 –22.9

289 1209 0.124 20.9 –23.5

229 1283 0.192 34.1 –38.1

453 1604 0.123 26.2 –24.6

303 2006 0.140 29.2 –24.4

229 2006 0.133 34.1 –38.1

shows RPE –38.1%. An estimator should keep in mind the level of underestimation that can be expected from using ANN model. The results of the ANN model performance for a particular validation subset are shown in Table 3. In addition to RPE and MAPE, the minimum and maximum ANN output of the VMH are also given, indicating the ANN output range. The overall network output based on performed cross-validation using five subsets yields a MAPE 0.133. These results show that additional instruction should be implemented in order to apply the results gained from the ANN model in the estimation process.

For this purpose, the RPE shown in Fig. 9 was reshaped in histogram form and the corresponding cumulative function as shown in Fig. 10.

Fig. 11. RPE sample histogram and cumulative distribution using 15% safety factor

Fig. 10. RPE sample histogram and cumulative distribution

For an expert it is important to have sufficient confidence in estimations given by the ANN model.

The RPE sample histogram gives better information regarding ANN model behaviour. The information gained from this diagram is the basis for proposing a practical safety-factor approach. The goal of this approach is to give an expert the guidance on interpretation and how to use the ANN network estimations in order to shape the conservative decision in real life application. For the purpose of practical safety-factor approach, the 80/20 Pareto principle was applied. From the cumulative distribution it can be seen that 20% of all outputs have an RPE of –15% or less. This gives a basis for defining safety-

Neural Network-Based Model for Supporting the Expert Driven Project Estimation Process in Mold Manufacturing

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 3-13

factor approach. To achieve the 15% safety factor on the gained ANN model output, we artificially shift the obtained histogram into overestimating interval. It can be can expected that by applying this factor approximately 20% of all cases will fall in a safer underestimation interval, as shown in Fig 11. To achieve even more conservative decisions a higher, 25% safety factor is advised. In this case it can be expected that only around 4.8% of all cases will fall in an underestimation interval. 4 CONCLUSIONS This paper proposes an implementation of the ANN based model that can be used as expert support in the project estimation process. The proposed supported project estimation process defines a bridge between expert-driven intuitive models and datadriven models. As an example, an ANN model to estimate VMH is considered. The results show that the presented ANN model fulfils the requirements of relevancy, simplicity, and reliability. A major benefit of ANN is the ability to model multivariable relations, but on the other hand the model showed in some cases output deviations that should not be neglected in reallife application of the model. By implementing a safety-factor approach, guidance is given to the expert on how to handle network output in order to decrease the probability of unwanted project underestimation, and to achieve acceptable confidence of estimation, respectively. The following benefits can be expected by applying proposed supported estimation approach: • lowered risk of underestimating the complexity of the project, • embedded repeatability and stability in the decision making process, • improvement in expert estimation reliability, • a significantly shorter estimation process, • allowing an enterprise to foresee sufficient manufacturing resources in the early project stage, • by adapting input data specific to the estimator’s environment this model can be applied in any mold shop, • it can be used as a learning assistant for novice estimators. The major limitation of the proposed modelbased, supported project estimation process is a limited number of samples. In addition, the assumption that by implementing a limited number of parameters the information is incomplete from a wider perspective cannot be neglected. As a result, in decision making 12

processes, experts frequently rely on information that is incomplete. To overcome this obstacle, future research activities will consider implementation and development of a specially tailored expert elicitation model. 5 REFERENCES [1] Flyvbjerg, B. (2006). From Nobel prize to project management: getting risks right. Project Management Journal, vol. 37, no. 3, p. 5-15. [2] Fonseca, M., Henriques, E., Ferreira, A., Jorge, J. (2007). Assisting mould quotation though retrieval of similar data. Digital Enterprise Technology, Session 5, p. 527-534, DOI:10.1007/978-0-387-49864-5_62. [3] Duverlie, P., Castelain, J.M. (1999). Cost estimation during design step: Parametric method versus case based reasoning method. The International Journal of Advanced Manufacturing Technology, vol. 15, no. 12, p. 895-906, DOI:10.1007/s001700050147. [4] Denkena, B., Lorenzen, L.E., Schürmeyer, J. (2009). Rule-based quotation costing of pressure die casting moulds. Production Engineering, vol. 3, no. 1, p. 8794, DOI:10.1007/s11740-008-0139-8. [5] Ficko, M., Drstvenšek, I., Brezočnik, M., Balič, J., Vaupotič, B. (2005). Prediction of total manufacturing costs for stamping tool on the basis of CAD-model of finished product. Journal of Materials Processing technology, vol. 164-165, p. 1327-1335, DOI:10.1016/j. jmatprotec.2005.02.013. [6] Farineau, T., Rabenasolo, B., Castelain, J.M., Meyer, Y., Duverlie, P. (2001). Use of parametric models in an economic evaluation step during the design phase. The International Journal of Advanced Manufacturing Technology, vol. 17, no. 2, p. 79-86, DOI:10.1007/ s001700170195. [7] Rihar, L., Kušar, J., Duhovnik, J., Starbek, M. (2010). Teamwork as a pre-condition for simultaneous product realization. Concurrent Engineering Research and Application, vol. 18, no. 4, p. 261-273, DOI:10.1177/1063293X10389789. [8] Kušar, J., Rihar, L., Duhovnik, J., Starbek, M. (2008). Product management of product development, Strojniški vestnik - Journal of Mechanical Engineering, vol. 54, no. 9, p. 588-606. [9] Raviwongse, R., Allada, V. (1997). Artificial neural network based model for computation of injection mold complexity. The International Journal of Advanced Manufacturing Technology, vol. 13, no. 8, p. 577-586, DOI:10.1007/BF01176302. [10] Fagade, A.A., Kazmer, D.O. (1999). Modelling the effects of complexity on manufacturing costs and time-to-market of plastic injection molded products. Proceedings of the Tenth Annual Conference of Production and Operations Management Society, Charleston. [11] Fagade, A.A., Kazmer, D.O. (1999). Effects of complexity on tooling cost and time-to-market of

Florjanič, B. – Govekar, E. – Kuzman, K.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 3-13

plastic injection molded parts. Proceedings ANTEC, p. 3304-3308. [12] Che, Z.H. (2010). PSO-based back-propagation artificial neural network for production and mold cost estimation of plastic injection molding. Computer & Industrial Engineering, vol. 58, no. 4, p. 625-637, DOI:10.1016/j.cie.2010.01.004. [13] Wang, H., Zhou, X.H., Rouan, X.Y. (2003). Research on injection mould intelligent cost estimation system and key technologies. The International Journal of Advanced Manufacturing Technology, vol. 21, no. 3, p. 215-222, DOI:10.1007/s001700300024. [14] Cavalier, S., Maccarone. P., Pinto, R. (2004). Parametrical vs. neural network models for estimation of product costs: a case study in the automotive industry. International Journal of Production Economics, no. 91, p. 165-177, DOI:10.1016/j.ijpe.2003.08.005. [15] Farineau, T., Rabenasolo, B., Castelain, J.M. (2002). Choice of cost-estimation functions based on statistical quality criteria and technical coherence. The International Journal of Advanced Manufacturing Technology, vol. 19, no. 7, p. 544-550, DOI:10.1007/ s001700200058. [16] Elhag, T.M.S., Boussabine, A.H. (1999). Tender price estimation: Neural networks vs. regression analysis. Proceedings of COBRA RICS Research Foundation, Salford. [17] Verlinden, B., Duflou, J.R., Collin, P., Cattrysse, D. (2008). Cost estimation for sheet metal parts using multiple regression and artificial neural networks: A case study. International Journal of Production Economics, vol. 111, no. 2, p. 484-492, DOI:10.1016/j. ijpe.2007.02.004. [18] Kim, G.H., An, S.H., Kang, K.I., (2004). Comparison of construction cost estimating models based on regression analysis, neural networks, and case-based reasoning. Building and Environment, vol. 39, no. 10, p. 1235-1242, DOI:10.1016/j.buildenv.2004.02.013. [19] Chan, S.F., Law, C.K., Chan, K.K. (2003). Computerised price quoting system for injection mould manufacture. Journal of Materials Processing Technology, vol. 139, no. 1-3, p. 212-218, DOI:10.1016/ S0924-0136(03)00222-X.

[20] Denkena, B., Schürmeyer, J., Böß, V., Kaddour, R. (2011). CAD-based cost calculation of mould cavities. Production Engineering, vol. 5, no. 1, p. 73-79, DOI:10.1007/s11740-010-0277-7. [21] Chin, K.S., Wong, T.N. (1996). Developing a knowledge-based injection mold cost estimation system by decision tables. The International Journal of Advanced Manufacturing Technology, vol. 11, no. 5, p. 353-365, DOI:10.1007/BF01845694. [22] Fagade, A.A., Kazmer, D.O. (1998). Economic design of injection molded parts using DFM guidelines - a review of two methods for tooling cost estimation. Proceedings ANTEC, Atlanta, p. 869-873. [23] Fagade, A.A., Kazmer, D.O. (2000). Early cost estimation for injection molded components. Journal of Injection Molding Technology, vol. 4, no. 3, p. 97-106. [24] Nagahanumaiah, Ravi, B., Mukherjee, N.P., (2005). An integrated framework for die and mold cost estimation using design features and tooling parameters. The International Journal of Advanced Manufacturing Technology, vol. 26, no. 9-10, p. 1138-1149, DOI:10.1007/s00170-004-2084-9. [25] Navodnik, J., Kopčič, M. (1998). Plastic – moldmaking handbook, 3rd. revised ed. Navodnik-Chemical Engineering, Velenje. (in Slovene). [26] Menges, G., Michaeli, W., Mohren, P. (2000). How To Make Injection Molds, 3rd ed., Hanser Gardner Publications, Cincinatti. [27] Kazmer, D.O. (2007). Injection Mold Design Engineering, Hanser Gardner Publications, Cincinnati, DOI:10.3139/9783446434196. [28] Kušar, J., Duhovnik, J., Tomaževič, R., Starbek, M. (2007). Finding and evaluating customers needs in the product-development process. Strojniški vestnik – Journal of Mechanical Engineering, vol. 53, no. 2, p. 78-104. [29] Beale, M.H, Hagan, M.T., Demuth, H.W. (2010). MATLAB, Neural Network Toolbox™ 7, User’s Guide, online edition, The MathWorks, Inc., Natic, from: http://www.mathworks.com/help/pdf_doc/nnet/ nnet_ug.pdf, accessed at 11-10-25.

Neural Network-Based Model for Supporting the Expert Driven Project Estimation Process in Mold Manufacturing

Received for publication: 2012-08-23 Received revised form: 2012-10-06 Accepted for publication: 2012-11-16

Model Development and Dynamic Load-Sharing Analysis of Longitudinal-Connected Air Suspensions Chen, Y. – He, J. – King, M. – Chen, W. – Wang, C. – Zhang, W. Yikai Chen1,5,* – Jie He2 – Mark King3 – Wuwei Chen4 – Changjun Wang5 – Weihua Zhang1 1 Hefei

University of Technology, School of Transportation Engineering, China 2 Southeast University, School of Transportation, China 3 Queensland University of Technology, Centre for Accident Research and Road Safety- Queensland, Australia 4 Hefei University of Technology, School of Mechanical and Automotive Engineering, China 5 Traffic Management Research Institute of the Ministry of Public Security, China The objective of this research was to investigate the effect of suspension parameters on dynamic load-sharing of longitudinal-connected air suspensions of a tri-axle semi-trailer. A novel nonlinear model of a multi-axle semi-trailer with longitudinal-connected air suspension was formulated based on fluid mechanics and thermodynamics and was validated through test results. The effects of suspension parameters on dynamic load-sharing and road-friendliness of the semi-trailer were analyzed. Simulation results indicate that the road-friendliness metric DLC (Dynamic Load Coefficient), is generally in accordance with the load-sharing metric - DLSC (Dynamic Load Sharing Coefficient). When the static height or static pressure increases, the DLSC optimization ratio declines monotonically. The effect of employing larger air lines and connectors on the DLSC optimization ratio gives varying results as road roughness increases and as driving speed increases. The results also indicate that if the air line diameter is always assumed to be larger than the connector diameter, the influence of air line diameter on loadsharing is more significant than that of the connector. Keywords: modeling, driving condition, dynamic load-sharing, longitudinal-connected, air suspension, heavy truck

0 INTRODUCTION Extensive studies of “road-friendly” heavy vehicles have been performed during the last few decades to reduce road damage and increase the rated load of vehicles. However, the load-sharing ability of multiaxle heavy vehicles, which has a strong correlation with road-friendliness, has been far from adequately investigated. Load-sharing is defined as the equalization of the axle group load across all wheels/ axles [1]. When a multi-axle heavy vehicle with leaf suspensions travels on a rough road or hits a bump/ pothole, such as a bridge-head, or speed control humps, unequally distributed loads among the axles of an axle group tend to appear due to the ineffectiveness of the load-sharing mechanism (centrally pivoted walking beam, trunnion shaft, etc.) and the high stiffness of leaf springs [2]. This phenomenon causes overloading of a single axle of the axle group, which has at least two disadvantages: (a) it increases the possibility of a tire bursting as well as reducing the maneuverability and stability of the vehicle; (b) it accelerates the rutting and fatigue that contributes to pavement damage [3]. As a consequence, the improvement of load-sharing within axle groups has attracted much attention among both vehicle manufacturers and road management departments. Load-sharing performances of axle groups are specified in regulations for road-friendly vehicles in many countries. The DIVINE (Dynamic Interaction between Vehicle and Infrastructure Experiment) 14

project undertaken by OECD (Organization for Economic Cooperation and Development) suggests that in order to qualify as a road-friendly tandem suspension, the average load variation per unit of relative vertical suspension displacement must be less than 0.3 kN/mm [4]. The Australian specification for road-friendly suspensions nominates that roadfriendly suspensions must have static load-sharing, i.e., load-sharing when the vehicle is stationary, to a defined value, between axles in an axle group or tires in an axle group. However, the formal methodology to determine the static load-sharing value on a heavy vehicle is not defined [5]. In Europe, an air suspension needs to have fully-functioning hydraulic shock absorbers to pass a static road-friendliness test [6], and heavy vehicles with road-friendly suspensions are allowed higher static axle loads. A common problem with these regulations is that only the static loadsharing of vehicles is specified and that there is no requirement for suspensions to retain their dynamic load-sharing performance, i.e., load-sharing when the vehicle is driving. Many other load-sharing metrics have also been proposed by researchers. LSC (Load-Sharing Coefficient) [7] and DLSC (Dynamic Load-Sharing Coefficient) [8] have been used to evaluate static and dynamic load-sharing, respectively. Noting that perfect load equalization would give a LSC of 1.0 [9], LSC values for steel suspensions were documented in the range 0.791 to 0.957 [7]. Air suspensions with conventional-size longitudinal air lines were placed

*Corr. Author’s Address: Hefei University of Technology, School of Transportation Engineering, 193 # Tunxi Road, Hefei, China, leochen079307@hotmail.com

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 14-24

in the middle of this range with LSCs of 0.904 to 0.925 [4]. More recent studies commissioned by the National Road Transport Commission of Australia found that installation of larger air lines on multiaxle air suspensions increased longitudinal air flow between air springs on adjacent axles [10] and [11] Follow-up tests funded by the Queensland Department of Main Roads discovered that an improvement in DLSC of 4 to 30% for a tri-axle coach and 37 to 77% for a tri-axle semi-trailer were obtained by alternating the conventional-size longitudinal air connection (three 6.5 mm inside diameter connectors connecting 6.5 mm inside diameter air lines) with a larger air connection (three 20 mm inside diameter connectors connecting 50 mm inside diameter air lines) [12] and [13]. However, due to the limitations of laboratory equipment, only vehicle speed and a limited number of air connections were considered in most tests, so that the effect of some other factors like the static absolute air pressure of the air spring and static height of the air spring on load-sharing have not yet been reported before. Limitations of laboratory and on-road tests can be addressed by developing realistic models of longitudinal-connected air suspensions. Potter et al developed a simplified tandem bogie model, and by changing the damping coefficient and torsional stiffness of the leveling beam of the model [2], it can represent load-leveling steel suspensions, independent steel suspensions, longitudinal-connected air suspensions, and independent air suspensions, respectively. Davis proposed a model of a tri-axle semitrailer with longitudinal-connected air suspensions [12], and used a variable “load-sharing fraction” to represent the load-sharing ability of the suspension. However, the physical meaning of the variable was unclear. A more realistic model of a similar tri-axle semi-trailer was developed by Roebuck et al based on aerodynamics and thermodynamics [14]. In the model, the volumetric flow rate [m3/s] between two air springs was assumed to be simply proportional to the difference in air pressure with a constant coefficient Cflow [m3/(kPa·s)]; in addition, the volumes and effective areas of the air springs were simplified as constants while the vehicle was travelling. Unfortunately, these simplifications of nonlinearities reduced the precision of the proposed models. A more realistic model of longitudinalconnected multi-axle air suspensions is urgently needed for precise analysis and optimization of loadsharing in multi-axle semi-trailers. The rest of this paper is organized in the following order. In Section 1, a novel nonlinear

model of longitudinal-connected tri-axle air suspensions is derived based on fluid mechanics and thermodynamics. The accuracy of the model is validated and load-sharing criteria are chosen in Section 2. Based on the model, the effects of air suspension parameters (static height and static absolute air pressure of air spring, inside diameters of air line and connector) on dynamic load-sharing are analyzed in Section 3. Finally, Section 4 presents a summary of the results and draws conclusions. 1 INTEGRATED MODEL OF VEHICLE AND ROAD EXCITATION 1.1 Mathematic Model of the Tri-Axle Semi-Trailer A basic half model representing a typical tri-axle semitrailer with longitudinal-connected air suspensions in most western countries was employed, as shown in Fig. 1. This model includes five degrees of freedom (DOF), which are vertical displacements of sprung mass and three unsprung masses, z, x1, x2, x3, as well as the pitch angle of the sprung mass, ϕ. Front

Rear

x1 q1

x2 q2

x3 q3

Fig. 1. Schematic of the tri-axle semi-trailer with longitudinalconnected air suspensions

The equations of motion of the semi-trailer are given by: mt1  x1 = (q1 − x1 )kt1 + c1 ( z − x1 − φl ) − (1) 1 − ( Ps1 − P0 ) As1 + mg , 3

mt 2  x2 = (q2 − x2 )kt 2 + c2 ( z − x2 ) − 1 − ( Ps 2 − P0 ) As 2 + mg , 3

(2)

mt 3  x3 = (q3 − x3 )kt 3 + c3 ( z − x3 + φl ) − 1 − ( Ps 3 − P0 ) As 3 + mg , 3

Model Development and Dynamic Load-Sharing Analysis of Longitudinal-Connected Air Suspensions

(3)

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 14-24

1   J φ = ( Ps 3 − P0 ) As 3 − c3 ( z − x3 + φl ) − mg )  l − 3   (4) 1   − ( Ps1 − P0 ) As1 − c1 ( z − x1 − φl ) − mg )  l , 3  

mz = ( Ps1 − P0 ) As1 + ( Ps 2 − P0 ) As 2 + + ( Ps 3 − P0 ) As 3 − c1 ( z − x1 − φl ) − (5) − c ( z − x ) − c ( z − x − φl ) − mg , 2

where mt1, mt2, mt3 and q1, q2, q3 are the unsprung masses and road excitation of the three axles, respectively. m is the sprung mass of the semi-trailer. J is the moment of inertia of the gross sprung mass around the lateral axis. Ps1, Ps2, Ps3 and As1, As2, As3 are the dynamic absolute air pressure and dynamic effective area of the three air springs, respectively. P0 is atmospheric pressure. l is the wheelbase, c1, c2, c3 are the damping coefficients of three dampers, and kt1, kt2, kt3 are the stiffness of the three tires. 1.2 Road Roughness Excitation Many methodologies have been proposed to model road surface profile [15] to [17]. One method is to describe the profile as a realization of a random process that is represented by its PSD (Power Spectral Density). A concise spectral model is used in this study as [18]:

Gq (n) = Gq (n0 )(

n −2 ) (n1 < n < n2 ), (6) n0

where Gq(n) is the PSD function [m3/cycle] for the road surface elevation; n is the spatial frequency [cycle/m]; n0 is the reference spatial frequency, n0 = 0.1 cycle/m; Gq(n0) is roughness coefficient [m3/cycle], whose value is chosen depending on the road condition. Classification of road roughness is based on the index of the ISO standard [19]. The ISO has proposed a road roughness classification from Class “A” (very good) to Class “H” (very poor) according to different values of Gq(n0). n1 and n2 are lower and upper spatial cutoff frequencies when Gq(n) reaches 1 and 10-5 m3/cycle, respectively [19]. When travelling along the road surface at a constant vehicle speed u, the temporal frequency, f, and n are related as f = un. Therefore, the relationship between spatial PSD and temporal PSD becomes:  n 1 1 Gq ( f ) = Gq (n) = Gq (n0 )   u u  n0  16

−2

= Gq (n0 )n02

u . (7) f2

As the angular frequency, ω, is related to f as ω = 2 π f, Eq. (7) is rewritten as:

Gq (ω ) = 4π 2 Gq (n0 )n02

u . (8) ω2

Eq. (8) is transformed to the following equation when inserting a lower cutoff angular frequency, ω1 [20]:

Gq (ω ) = 4π 2 Gq (n0 )n02

u , (9) ω + ω12 2

where ω1 = 2 π n1 u. Standard road roughness is a response of a first order linear to a white noise, w(t) [21], therefore: 2

Gq (ω ) = H (ω ) Sω , (10)

where Sω is the PSD of the white noise, Sω =1. Substituting Eq. (10) into Eq. (9) yields: 2π n0 Gq (n0 )u H (ω ) = . (11) ω1 + jω Then the road roughness q(t) is given by:

q (t ) = −2π n1uq (t ) + 2π n0 Gq (n0 )uw(t ). (12)

The upper cut-off frequency n2 was modeled by setting the sampling frequency of w(t) based on Nyquist sampling theory, i.e., the sampling frequency should be at least 2n2u Hz. A time delay of l / v for road excitation is applied between adjacent axles. 1.3 Detailed Model of Longitudinal-Connected Tri-Axle Air Suspensions To solve the equations in section 1.1, a detailed model of longitudinal-connected tri-axle air suspensions is needed to express Ps1, Ps2 and Ps3 as functions of the 5 variables (5 DOF). It is assumed that all the air springs are stroked fast enough so that all the heat of the operation is conserved when the vehicle is travelling, i.e., an adiabatic process occurs. Thus, the formula for calculating the dynamic absolute air pressure inside the front air spring, Ps1, is [22]:

V Vs10 k = Ps1 ( s1 ) k P= ) constant. (13) s10 ( ms1 ms10

Vs1, Vs10 are the dynamic volume, and static volume of the front air spring; ms1, ms10 are the

Chen, Y. – He, J. – King, M. – Chen, W. – Wang, C. – Zhang, W.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 14-24

dynamic air mass, and static air mass inside the front air spring; and Ps10 is the static absolute air pressure inside the front air spring. The value of the above exponent, k, varies with the gas used and is a function of the specific heat of the gas. Air suspension operation is characterized by neither an isothermal nor an adiabatic process, but is instead polytropic. In normal use, however, the process is much closer to adiabatic than isothermal. Accordingly, the value of k is set to 1.4. Ps1 is obtained from Eq. (13) as follows: Ps1 = (

Vs10 ms1 1.4 ) Ps10 , (14) Vs1ms10

where Vs1 is a function of the dynamic height of the front air spring, and is given by:

Vs1 = ( z − x1 − φ l ) As1 + Vs10 , (15)

where As1 can also be approximated as a function of dynamic height based on experimental data. ms1 depends on the air flow rate inside the front connector, G1 [kg/s], is given by: t

ms1 = m10 + ∫ G1dt. (16)

Since only small variations of temperature, air pressure and air spring volume exist when the semitrailer is travelling, the air flow inside the front connector is considered to be an incompressible steady flow, which satisfies the following formula, according to Bernoulli’s equation [23]:

Ps1 1 2 Pf 1 1 2 + vs1 = + v f 1 , (17) ρ 2 ρ 2

where Pf1 is the dynamic absolute air pressure inside the front connector; vs1, vf1 are the air flow speed [m/s] inside the front air spring and the front connector, respectively; and the air inside all the air springs, connectors and the air line is assumed to have a same constant density, ρ, when the semi-trailer is travelling; Af1 is the effective area of the front connector, which is equal to the actual area multiplied by a contraction coefficient, 0.7 [24]. Noting that As1 is related to Af1 because As1·vs1 = Af1·vf1, inserting vs1 = Af1·vf1 / As1 into Eq. (17) yields:

vf1 =

2 Ps1 − Pf 1 ρ

where vf1 is modified with a coefficient, cd (0.8), to reflect the friction of the connector. Therefore, the actual air flow speed inside the front connector, v'f1, and G1 are given by:

2 Ps1 − Pf 1 ρ

v′f 1 = cd

 A2f 1  1 − ( )  , (19)  As21  

G1 = sgn( Pf 1 − Ps1 )cd Af 1 ×

× 2 ρ Ps1 − Pf 1

 A2f 1  1 ( )  . (20) −  As21  

Substituting Eqs. (15), (16) and (20) into Eq. (14) yields: Ps1 = Ps10Vs110.4 [ ( z − x1 − φ l ) As1 + Vs10 ]

−1.4

m s−101.4 ×

× [m10 + ∫ sgn( Pf 1 − Ps1 )cd Af 1 ×

(21)

× 2 ρ Ps1 − Pf 1

(1 − (

2 f1 2 s1

))dt ]1.4 .

The dynamic absolute air pressures inside the three connectors and the longitudinal air line are assumed to have the same value, Pf1, during travel. Thus, similar expressions for the air flow rate inside the middle and rear connectors (G2 and G3), as well as the dynamic absolute air pressure inside the middle and rear air springs (Ps2 and Ps3) are derived as follows: G2 = sgn( Pf 1 − Ps 2 )cd Af 2 ×

× 2 ρ Ps 2 − Pf 1

 A2f 2  1 ( )  , (22) −  As22  

G3 = sgn( Pf 1 − Ps 3 )cd Af 3 ×

× 2 ρ Ps 3 − Pf 1

 A2f 3  1 ( )  , (23) −  As23  

Ps 2 = Ps 20Vs120.4 [ ( z − x2 ) As 2 + Vs 20 ]

−1.4

m s−201.4 ×

× [m20 + ∫ sgn( Pf 1 − Ps 2 )cd Af 2 × 0

× 2 ρ Ps 2 − Pf 1

(1 − (

A2f 2 As22

))dt ]1.4 ,

(24)

 A2f 1  1 − ( 2 )  , (18) As1  

Model Development and Dynamic Load-Sharing Analysis of Longitudinal-Connected Air Suspensions

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 14-24

Ps 3 = Ps 30Vs130.4 [ ( z − x3 + φ l ) As 3 + Vs 30 ]

−1.4

Table 1. Parameters of the Tri-axle Semi-trailer Model

m s−301.4 ×

Parameter

× [m30 + ∫ sgn( Pf 1 − Ps 3 )cd Af 3 × 0

× 2 ρ Ps 3 − Pf 1

(1 − (

A2f 3 As23

1.4

))dt ] ,

(25)

where Ps20, Vs20, ms20 are the static absolute air pressure, static volume and static air mass of the middle air spring, Ps30, Vs30, ms30 are the corresponding parameters for the rear air spring; and Af2, Af3 are the areas of the middle and rear connectors, respectively. The three air springs have the same static absolute air pressure, static volume and static air mass, and the three connectors have the same inside diameter. The air line is made of steel, so the volume of air line is constant. Based on Eq. (14), the dynamic absolute air pressure inside the air line, Pf1, is expressed as a function of the static air mass of the air line, ml0, the static absolute air pressure inside the air line, Ps10, and the gross air flow rate inside the air line, Gl, shown as follows: t ml 0 + ∫ Gl dt . 0 Pf 1 = ( )1 4 Ps10 , (26) ml 0

Gl = −G1 − G2 − G3 . (27)

Substituting Eqs. (20), (22), (23) and (27) into Eq. (26) yields: Pf 1 = ml 0 −1.4 Psl 0 (ml 0 + ∫ {− sgn( Pf 1 − Ps1 ) ×

As10 Vs10 hs0 df ds Ps10 P0 m mt1 J kt1 ρ crebound

cbump

Value

Dimension

Description

Static effective area of each air spring 0.0125 m3 Static volume of each air spring 0.16 m Static height of each air spring 0.0065 m Inside diameter of each connector Inside diameter of the longitudinal 0.0065 m air line Static absolute air pressure inside 464288 Pa each air spring, each connector and the air line 101325 Pa Atmosphere pressure Gross sprung mass of the semi8700 kg trailer 336 kg Unsprung mass of each air spring Moment of inertia of the gross 5684 kg·m2 sprung mass around the lateral axis 1960000 N/m Stiffness of dual tires on each hub Density of air inside air springs, air 6.5417 kg/m3 connectors and the air line Damping coefficient of each damper when dynamic height 288600 N·s/m of respective suspension is increasing Damping coefficient of each damper when dynamic height 184500 N·s/m of respective suspension is decreasing 0.0783

2 LOAD-SHARING CRITERIA AND MODEL VALIDATION

× cd Af 1 2 ρ Ps1 − Pf 1

 A2f 1  1 − ( 2 )  − As1  

2.1 Load-Sharing Criteria

− sgn( Pf 1 − Ps 2 )cd Af 2 × 2 ρ Ps 2 − Pf 1

 A2f 2  1 − ( 2 )  − As 2  

− sgn( Pf 1 − Ps 3 )cd Af 3 × 2 ρ Ps 3 − Pf 1

 A2f 3    1.4 1 − ( 2 )   dt) . (28) As 3    

Based on the equations in Section 1.1, Section 1.2 and Eqs. (21), (24), (25) and (28), an integrated model of road excitation and a fully-loaded tri-axle semitrailer with longitudinal-connected air springs was developed with Matlab/ Simulink. Parts of the key parameters are tabulated in Table 1. The expression of the effective area of each air spring as a function of the dynamic height of corresponding air spring will be obtained based on test results in Section 2.2.

Criteria need to be chosen to evaluate the load-sharing of the semi-trailer. A metric often used to characterize the magnitude of dynamic forces of a wheel in an axle group is the DLC (Dynamic Load Coefficient) [7], defined as: σi DLC (i ) = , (29) Fmean (i ) where σi denotes the standard deviation of wheelforce i, and Fmean(i) denotes the mean wheel-force of wheel i. Although DLC is usually referred to as a road-friendliness criterion and has been criticized for its mutually exclusivity with another load-sharing criterion, LSC [1], it still has been widely used as one measure to differentiate suspension types from each other (e.g., steel vs. air) [12], [13] and [25].

Chen, Y. – He, J. – King, M. – Chen, W. – Wang, C. – Zhang, W.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 14-24

De Pont points out that LSC does not address dynamic load-sharing [8]. The DLSC was proposed as an alternative to LSC, to account for the dynamic nature of wheel-forces and instantaneous load-sharing during travel, and is defined as [8]. 1

DLSCi =

∑ ( DLS ( j ) − k ∑ DLS ( j )) j =1

j =1

. (30)

The dynamic load-sharing of wheel i, DLSi(j), is:

DLSi ( j ) =

nFi ( j ) n

∑ Fi ( j )

, (31)

i =1

where n is the number of wheels on one side of an axle group; k is the number of terms in the dataset; and Fi(j) is the instantaneous force at wheel i. In this study, the average DLSC of tires on the same side of the semi-trailer axle group was employed as a metric of load-sharing. The average DLC of tires on the same side of the semi-trailer axle group was used to evaluate road-friendliness as well as to analyze the relationship between load-sharing and road-friendliness. 2.2 Model Validation The prototype of the tri-axle semi-trailer was tested on various road sections for verification of the integrated model of vehicle and road excitation, as shown in Fig. 2. The tests were part of a joint project titled “Heavy vehicle suspensions – testing and analysis” between the Queensland University of Technology (QUT) and the Department of Transport and Main Roads, Queensland (TMR) [26]. The setups of the tests are shown in Fig. 2. Two types of longitudinal connections were used to connect the passive air suspensions on the same side: conventional (three 6.5 mm inside diameter connectors connecting a 6.5 mm inside diameter air line) and large (three 20 mm inside diameter connectors connecting a 50 mm inside diameter air line). Strain gauges (one per hub) were mounted on the neutral axis of each axle between the spring and the hub to record the shear force on the hubs, i.e., air spring force, and accelerometers were mounted as closely as possible to each hub and to the corresponding upper positions at the chassis to derive the height of each air spring. In addition, six air pressure transducers were employed to obtain the pressures inside the air springs, and a

TRAMANCO P/L on-board CHEK-WAY telemetry system was used to record all the data. The dynamic force of each tire was derived based on the shear force on the respective hub and the acceleration on the respective axle. The effective area of each air spring was obtained by dividing the respective shear force by the respective pressure inside the air spring, and the volume of each air spring was derived by multiplying the respective effective area by the respective spring height. The tests comprised of driving the semi-trailer over three typical urban road sections at speeds ranging from 60 to 80 km/h; the sections of road varied from smooth with long undulations to rough with short undulations. The IRI (International Roughness Index) values of each road section were provided by TMR, and the IRI is related to Gq(n0) in Eq. (6) as IRI = 0.78 × 103 Gq (n0 ) [27]. Ten seconds of dynamic signal data were recorded per road section, and this was done for both experimental cases (i.e., conventional longitudinal connection vs. large longitudinal connection) for the fully loaded condition. Thus, the dynamic effective area of each air spring is approximated as a function of the dynamic height of corresponding air spring, y, based on the experimental results, shown as: 3 2 As1 = −7.670500y + 2.866880y −

−0.354226y + 0.093002.

(32)

The effective area multiplied by the dynamic spring height yields: Vs1 = −7.670500y 4 + 2.866880y 3 −

−0.354226y + 0.093002y.

(33)

The comparisons between the test and simulation results in terms of load-sharing performance are listed in Table 2. As shown in Table 2, a reasonable agreement exists between test and simulation results for both types of connection, under various road roughness conditions and vehicle speeds. The absolute error ratio of the DLC between each test and corresponding simulation is less than 10%; except for test/simulation 4, the absolute error ratios of the DLSC are less than 20%. It is also noted that all the simulation results are smaller than the corresponding test results, which mainly dues to wear of the suspensions of the test vehicle after a period of use and some simplifications of the model of longitudinal-connected air suspensions.

Model Development and Dynamic Load-Sharing Analysis of Longitudinal-Connected Air Suspensions

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 14-24

g) Fig. 2. Setups of the tests; a) prime mover, b) test semi-trailer with test load, c) large longitudinal air line, d) strain gauge, e) accelerometer mounted on the axle, f) air pressure transducer, g) CHEK-WAY telemetry system underneath the semi-trailer [26] Table 2. Comparison of Load-Sharing Performances between Tests and Simulations Test/ simulation number

Type of longitudinal connection

IRI

Velocity [km/h]

conventional

6.213

large

6.213

conventional

7.602

large

7.602

conventional

8.880

large

8.880

It can be concluded from Table 2 that the simulation results correlated well with the measurements. Therefore, the integrated model 20

Load-sharing criteria

Test results

Simulation results

DLC DLSC DLC DLSC DLC DLSC DLC DLSC DLC DLSC DLC DLSC

0.0791 0.0505 0.0699 0.0440 0.1034 0.0851 0.0983 0.0819 0.1773 0.1506 0.1775 0.1474

0.0733 0.0431 0.0637 0.0357 0.1001 0.0721 0.0926 0.0645 0.1679 0.1256 0.1626 0.1256

Error ratio (compared with the test results) [%] -7.3 -14.6 -8.9 -18.8 -3.2 -15.3 -5.8 -21.2 -5.3 -16.6 -8.4 -14.8

of vehicle and road excitation in this study can be employed for further analysis.

Chen, Y. – He, J. – King, M. – Chen, W. – Wang, C. – Zhang, W.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 14-24

3 EFFECT OF SUSPENSION PARAMETERS ON LOAD-SHARING 3.1 Effect of Static Height and Static Pressure The effect of the static height and static pressure of air springs for the fully-loaded semi-trailer are discussed in this section, for a constant vehicle speed of 20 m/s and a standard “B” class road profile. Note that the effective area of each air spring at a given height is nearly constant with various static absolute air pressures, i.e., Eq. (32) always holds. The new static pressure of each air spring under a new static height is related to the corresponding static spring height as follows: 3 Ps10 ( new ) = 2900 × 9.8 × ( −7.670500hs 0 ( new ) + 2

+ 2.866880hs 0 ( new ) − 0.354226hs 0 ( new ) + 0.093002)

−1

+ 1.01325 * 105.

(34)

0.056 Type 1 Type 2

0.054

DLSC

0.052 0.05 0.048 0.046

The influence of the static height (varying from 0.15 to 0.22 m) and corresponding static pressure (varying from 463431 to 495204 Pa) on load-sharing are shown in Fig. 3. It can be seen in Figs. 3a and b that both the DLSC and DLC reduce as the static height increases. In fact, when the static height increases, the absolute value of dAs10 / dy increases and dVs10 / dy decreases based on Eqs. (32) and (33). These result in the reduction of dynamic stiffness of each air spring and accordingly the decline of DLSC and DLC. Another finding is that compared with the DLC of the semi-trailer with connection “2”, which decreases at a relatively constant rate, the DLSC of the semitrailer with connection “2” decreases more slowly and becomes constant when the static height exceeds 0.19 m. This indicates that when a large connection is employed, the load-sharing will not change much as the static height increases and the air pressure of air springs decreases, but the dynamic tire forces will continue to reduce. The optimization ratios of both DLSC and DLSC decline as the static height increases, as shown in Fig. 3c. As the static height increases, Vs10 increases, and the volume of air line / Vs10 decreases, and the effect of employing large air line and connectors becomes less prominent.

0.044 0.042 0.15

0.16

0.17

0.18 0.19 0.2 Static height (m)

0.21

0.22

0.09 Type 1 Type 2

0.085

DLC

0.08 0.075 0.07 0.065 0.15

0.16

0.17

0.18 0.19 0.2 Static height (m)

0.21

0.22

18% DLSC DLC

Optimization ratio

16% 14% 12% 10% 8% 6% 4% 2% 0.15

0.16

0.17

0.18

0.19

0.2

0.21

0.22

Static height (m) c) Fig. 3. Effect of static height on load-sharing; a) DLSC, b) DLC, c) DLSC optimization ratio and DLC

3.2 Effect Of Inside Diameter of Air Line and Connector The effects of size of air line and connector (varying from 10 to 100 mm) on load-sharing are plotted in Fig. 4, with a constant vehicle speed of 20 m/s for the fully-loaded semi-trailer and a standard “B” class road profile. The diameters of the connectors are always less than or equal to those of the air lines in all the simulations. It can be seen in Figs. 4a and 4b that with a fixed diameter of air line, both DLSC and DLC reduce quickly as the air line diameter increases from 10 to 30 mm. For example, with a 100 mm diameter air line, reductions up to 11.4 and 8.4% are observed for DLSC and DLC, respectively. When the diameter of air line increases beyond 30 mm, both DLSC and DLC decrease more slowly and finally become constants. However, when the diameter of the connector is fixed, the change of DLSC with air line diameter is different from that of the DLC. With a 10 mm diameter connector, as the air line diameter increases from 10 to 100 mm, the DLSC only decreases 1.0%, while the DLC decreases 6.8%. Thus, although the load-sharing of the semi-trailer improves very slowly by increasing the size of the connector, the dynamic

Model Development and Dynamic Load-Sharing Analysis of Longitudinal-Connected Air Suspensions

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 14-24

tire force and accordingly the road-friendliness of the semi-trailer are effectively improved.

PSD (N2/Hz)

DLSC

0.052 0.05

0.05

0.048

0.046

100 Diameter of connector (mm)

100

a) 10 10

0.08

PSD (N2/Hz)

DLC

0.078 0.076

0.075

0.074 0.07

0.072 0.07

0.065 20

Type 1 Type 2 Type 3

20 Diameter of air line (mm)

0.08

10 10 10 10

-1

10 Frequency (Hz)

7 6 5

Type 1 Type 2 Type 3

4 3 2

20 80 40 60 80 100 100 Diameter of air line (mm) Diameter of connector (mm)

The change in dynamic tire force can be revealed more clearly with spectral analysis. Fig. 5 shows the PSD for dynamic tire force for three types of longitudinal connections among air suspensions, i.e., type “1”, type “2” (both as specified above), and type “3” (three 100 mm inside diameter connectors connecting a 100 mm inside diameter air line). It is evident that the dynamic tire forces are effectively isolated by using large air lines and connectors, especially at frequencies lower than 0.3 Hz and frequencies from 0.8 to 5.0 Hz. Compared with the semi-trailer with air connection “1”, the peak values around the body bounce frequency (about 1.9 Hz) of the front dual tires decreased by 18.6 and 60.6% through employing air connections “2” and “3”, and the corresponding optimization ratios of the middle, rear dual tires are 22.3, 50.5, 36.6 and 58.1%, respectively. 4 CONCLUSIONS In this study, the effects of suspension parameters on dynamic load-sharing of longitudinal-connected air suspensions of a tri-axle semi-trailer were investigated comprehensively. A novel nonlinear model of longitudinal-connected tri-axle air suspensions

10 10

1 0

b) 10

10 PSD (N2/Hz)

Fig. 4. Effect of size of air line and connector on load-sharing optimization ratios; a) DLSC, b) DLC

0.085

0.044

10 Frequency (Hz)

Type 1 Type 2 Type 3

-1

10 Frequency (Hz)

Fig. 5. Effect of size of air line and connector on optimization ratios of load-sharing; a) front axle, b) middle axle, c) rear axle

was formulated based on fluid mechanics and thermodynamics and validated through test results. The effects of road surface conditions, driving speeds, air line diameters and connector diameters on the dynamic load-sharing capability of the semi-trailer were analyzed in terms of the DLSC and DLC, and the following conclusions can be drawn: 1. The road-friendliness metric DLC, is generally in accordance with the load-sharing metric DLSC.

Chen, Y. – He, J. – King, M. – Chen, W. – Wang, C. – Zhang, W.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 14-24

2. When the static height or the static pressure is increased, the DLSC optimization ratio declines. The reason for this phenomenon is that the static height increases with more static pressure, Vs10 increases, and the volume of air line divided by Vs10 decreases. Thus, the effect of employing a large air line and connectors becomes less prominent. 3. Assuming that the diameter of the air line is always larger than that of the connector, the influence of the diameter of air line is more significant than that of the connector. When the semi-trailer is driving at 20 m/s on a standard “B” class road and the connector diameter is fixed at 10 mm, the DLSC only decreases 1.0% as the air line diameter increases from 10 to 100 mm. However, when the air line diameter is fixed at 100 mm, the reduction reaches 11.4% as the connector diameter increases from 10 to 30 mm; as the connector diameter continues to increase, the DLSC declines at a very slow rate and becomes constant. Based on the proposed model, investigation of the control methods of the tri-axle air suspension system with longitudinal air line and their influence on loadsharing will be undertaken in the future. 5 ACKNOWLEDGEMENT This work was supported by the National Natural Science Foundation of China (Grant Nos. 51078087, 51178158, 51075112), the Natural Science Foundation of Anhui Province (Grant No. 11040606Q39), and the Fundamental Research Funds for the Central Universities (Grant No. 2012HGQC0015 and 2011HGBZ0945). The assistance of Dr Lloyd Davis from the Queensland Department of Transport and Main Roads is also greatly acknowledged. 6 REFERENCES [1] Davis, L.E., Bunker, M.B. (2008). Load-Sharing in Heavy Vehicle Suspensions - New Metrics for Old. Queensland University of Technology, QUT Digital Repository, Brisbane. [2] Potter, T.E.C., Cebon, D., Cole, D.J., Collop, A,C. (1996). Road damage due to dynamic tyre forces, measured on a public road. International Journal of Heavy Vehicle Systems, vol. 3, no. 1-4, p. 346-62. [3] Cebon, D. (1987). Assessment of The Dynamic Wheel Forces Generated by Heavy Road Vehicles. Symposium on Heavy Vehicle Suspension Characteristics, p. 199212.

[4] Cantieni, R., Krebs, W., Heywood, R. (1998). Dynamic Interaction between Vehicles and Infrastructure Experiment (DIVINE), Technical Report No. DSTI/ DOT/RTR/IR6(98)1/FINAL. Organisation for Economic Co-operation and Development (OECD), Paris. [5] Department of Transport and Regional Services Australia. (2004). Vehicle Standards Bulletin VSB 11 Certification of Road-Friendly Suspension Systems. Australian Department of Infrastructure, Transport, Regional Development and Local Government, Canberra. [6] Costanzi, M., Cebon, D. (2007). An investigation of the effects of lorry suspension performance on road maintenance costs. Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science, vol. 221, no. 11, p. 1265-1277, DOI:10.1243/09544062JMES639. [7] Sweatman, P.F. (1983). A Study of Dynamic Wheel Forces in Axle Group Suspensions of Heavy Vehicles. Australian Road Research Board, Special Report. Report No. SR27. Vermont South, Victoria. [8] de Pont, J.J. (1997). Assessing Heavy Vehicle Suspensions for Road Wear. Research report No 95. Transfund New Zealand, Wellington. [9] Potter, T.E.C., Cebon, D., Collop, A.C., Cole, D.J. (1996). Road-damaging potential of measured dynamic tyre forces in mixed traffic. Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering, vol. 210, no. 3, p. 215-225, DOI:10.1243/PIME_PROC_1996_210_265_02. [10] Estill & Associates Pty Ltd. (2000). Operational Stability and Performance of Air Suspension on Various Vehicle Configurations. Australia: Department of Transport and Works, South Perth. [11] Roaduser Systems Pty Ltd. (2005). Stability and OnRoad Performance of Multi-Combination Vehicles with Air Suspension Systems. Australia: National Road Transport Commission, Canberra. [12] Davis, L.E. (2010). Heavy Vehicle Suspensions- Testing and Analysis. Ph.D. thessis, Queensland University of Technology, Brisbane. [13] Davis, L.E., Bunker, J.M. (2011). Altering heavy vehicle air suspension dynamic forces by modifying air lines. International Journal of Heavy Vehicle Systems. vol. 18, no. 1, p. 1-17, DOI:10.1504/ IJHVS.2011.037957. [14] Roebuck, R.L., Cebon, D., Dale, S.G. (2006). Optimal control of a semi-active tri-axle lorry suspension. Vehicle System Dynamics, vol. 44, no. supl., p. 892903, DOI:10.1080/00423110600907493. [15] Papagiannakis, A.T., Zelelew, H.M, Muhunthan, B. (2007). Wavelet analysis of energy content in pavement roughness and truck dynamic axle loads. Transportation Research Record: Journal of the Transportation Research Board, vol. 2005/2007, p. 153-159, DOI:10.3141/2005-16.

Model Development and Dynamic Load-Sharing Analysis of Longitudinal-Connected Air Suspensions

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 14-24

[16] Taskin, Y., Hacioglu, Y., Yagiz, N. (2007). The use of fuzzy-logic control to improve the ride comfort of vehicles. Strojniski vestnik - Journal of Mechanical Engineering, vol. 50, no. 10, p. 462-468. [17] Guclu, R. (2004). The fuzzy-logic control of active suspensions without suspension-gap degeneration. Strojniski vestnik - Journal of Mechanical Engineering, vol. 53, no. 4, p. 233-240. [18] Shi, X.M., Cai, C.S. (2009). Simulation of dynamic effects of vehicles on pavement using a 3D interaction model. Journal of Transportation Engineering - ASCE, vol. 135, no. 10, p. 736-744, DOI:10.1061/(ASCE) TE.1943-5436.0000045. [19] ISO 8068:1995 (E) (1995). Mechanical VibrationRoad Surface Profiles-Reporting of Measured Data. International Organization for Standardization, Geneva. [20] Chen, J.P., Chen, W. W., Zhu, H., Zhu, M.F. (2010). Modeling and simulation on stochastic road surface irregularity based on Matlab/Simulink. Transactions of the Chinese Society of Agricultural Machinery, vol. 41, no. 3, p. 11-15. [21] Prabakar, R.S., Sujatha, C., Narayanan, S. (2009). Optimal semi-active preview control response of a half

car vehicle model with magnetorheological damper. Journal of Sound and Vibration, vol. 326, no. 3-5, p. 400-420, DOI:10.1016/j.jsv.2009.05.032. [22] Wang, J.S., Zhu ,S.H. (2009). Linearized model for dynamic stiffness of air spring with auxiliary chamber. Journal of Vibration and Shock, vol. 28, no. 2, p. 72-76. [23] White, F.M. (2011). Fluid Mechanics. McGraw-Hill, Columbus. [24] Chen, S.M. (2011). Hydraulic and Pneumatic Transmission. China Machine Press, Beijing. [25] Davis, L.E., Bunker, J.M. (2009). Suspension Testing of 3 Heavy Vehicles - Dynamic Wheel Force Analysis - Report. Australia: Department of Main RoadsQueensland Government, Brisbane. [26] Davis, L.E., Bunker, J.M. (2009). Heavy Vehicle Suspension Testing and Analysis-Dynamic Load Sharing. Australia: Department of Main Roads Queensland Government, Brisbane. [27] Chen, H.X., He, Z.Y. (2008). A study on simulation of road roughness based on international roughness index. Highway, vol. 11, p. 155-160.

Chen, Y. – He, J. – King, M. – Chen, W. – Wang, C. – Zhang, W.

Received for review: 2012-05-21 Received revised form: 2012-10-19 Accepted for publishing: 2012-11-06

Design and Materials Selection for Environmentally Friendly Ship Propulsion System Roldo, L. – Komar, I. – Vulić, N. Liane Roldo1,* – Ivan Komar2 – Nenad Vulić3

1 Federal

University of Rio Grande do Sul, Materials Department, Brazil

2 University of Split, Faculty of Maritime Studies, Croatia 3

Croatian Register of Shipping, Croatia

The selection of the material regarding specific design of the stern tube journal bearings in vessels is critical taking into account the lubrication system, whether oil or water based, and the consequent lubricant leakage. Therefore, the present study examines the feasibility and the advantages of implementing water lubricated polymer shaft bearings instead of conventional white metal bearings lubricated by oil. The investigation method is related to the numerical model and the software application based on finite difference method and isoviscous model. Results, based upon data collected from three different types of actual vessels in service, have shown that power loss in polymer bearings is at least 6 times less than in conventional white metal ones indicating that polymer journal bearings are significantly more energy efficient and environmentally friendly. Keywords: ship propulsion system, selection of materials, polymer, babbitt metal, journal bearing, software application

0 INTRODUCTION General machine building industry conventionally uses metallic antifriction materials or white metal (WM) bearing alloys in friction units, known commercially as babbitt alloys. ASTM Standard B 23-00 specification covers eight typical white metal bearing alloys, the tin-based alloys and lead-based alloys [1]. Babbitts are among the most widely used materials for hydrodynamic lubricated bearings. They have excellent embeddability and conformability characteristics. They are unsurpassed in compatibility and thus prevent shaft scoring. Tin- and lead-base babbitts have a relatively low load-carrying capacity. This capacity is increased by metallurgically bonding these alloys to stronger backing materials such as steel, cast iron, or bronze [2]. Copper-lead alloys, copper-nickel, bronzes and aluminium alloys are also used. Studies of sintered self-lubricating bronze bearings have shown that additional oil at bearing considerably decreased friction coefficient especially at high velocities and pressures. In addition, it has been observed that friction coefficient decreases more by additional additive, as well as effects of loads, spindle speed and oil types influence on friction coefficient [3]. On the other hand, the Croatian Register of Shipping (CRS) rules for classification of ships in 2009 notified that cast copper alloys are recommended for applications such as shaft liners and bearing bushes [4]. Feyzullahoǧlu and Şakiroǧlu [5] in 2010 pointed out that aluminium alloys due to fine properties like low cost, resistance to corrosive effects, co-activation with steel shafts, high

thermal conductivity, fatigue strength, lightness and workability, are entitled materials for journal bearings. In recent years, an increasingly close attention has been given to industrial products environmental safety of the friction units of modern ships, hydraulic turbines, pumps, shipping locks, as well as oilextracting and oil-processing equipment operating in water. For this reason oil lubrication of friction units is eliminated by using such natural lubricant as water or even without lubricants [2] and [6]. Some of the water durability multiphase systems are based on polymeric materials such as thermoplastic, thermosetting, rubber and composites. Still the major problems in designing polymer bearings are to decide the optimal dimensions and material type for a long life and for obtaining lower friction and wear losses [3]. The mechanical properties generally limit the application to lightly loaded conditions and often to low speeds and conforming surfaces [2]. To Sedlaček et al. [7] a possible way to design surface texturing parameters, which would result in contact surfaces with lower friction, is by treating surface texturing as a controlled roughness. Polypropylene (PP), polyethylene (PE), polyoxymethylene (POM), polyamide (PA), polyimide (PI), polytetrafluoroethylene (PTFE), polyurethane (PU) and polyesters with thermoplastic matrices are used to fabricate journal bearings. Also, other thermosetting materials like phenol-formaldehyde and epoxy resins, as well as composites with thermoplastic or thermosetting matrices reinforced with fillers such as carbon or glass fibre, are used as well [2], [6] and [8] to [10].

*Corr. Author’s Address: Federal University of Rio Grande do Sul,. Av. Osvaldo Aranha, 99/604, Porto Alegre - RS, 90035-190 Brazil liane.roldo@ufrgs.br

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 25-31

PP is used where higher-stiffness materials are required. It has a good environmental stress cracking resistance but is less resistant to degradation. If higher stiffness is required, reinforcement such as short glass-filled, calcium carbonate and talc can be added. PE has high toughness, ductility, excellent chemical resistance and very low water absorption. POM, PA and PI have good sliding and wear properties at low frictions. POM is a material generally used in engineering applications and is highly self-lubricating [8] and [9]. PI and its composites filled with solid lubricants and reinforced with carbon fibre show better friction reducing and antiwear behaviour under water-lubrication than under dry sliding [11]. Polyurethanes are very versatile polymers. They are used as flexible and rigid foams, elastomers, and coatings. Polyurethanes are available as both thermosets and thermoplastics. In addition, their hardness spans the range from rigid material to elastomer. Also, by providing good abrasion resistance with a low coefficient of friction and water-lubricated antifriction, cast PU’s find application in roller coatings and press pads as well as gaskets, casting molds, timing belts, wear strips, liners, heels, soles, etc. [6] and [8]. Solid lubricant properties and the low friction characteristics make PTFE suitable for use in bearings, mould release devices, and anti-stick cookware [8] and [9]. For polymeric and composites materials such as epoxy resins, polyacetals, polyesters, PTFE and PI the friction coefficient frequently decreases in the presence of water while the wear rate increases. This property depends on the type of filler; at least, PTFE with different fillers may exhibit both a decrease and increase in the wear resistance [6]. Glass fibres and carbon fibres, which are short fibre reinforcements, have been successfully used to improve the strength to high pressure and wear resistance. In addition, glass fibres improve the load carrying capability and thermal conductivity. This is a positive effect to lowering wear rate of pure polymer [6] and [8]. In general, ship propulsion system design is directed to safety and functionality, where the assembly of line shafts is fundamental. In this system the shaft transfers torque of the propulsion engine to screw propeller and propulsion force from the propeller back to the thrust bearing. The most sensitive component in the propulsion shafting system is the aft stern tube bearing, which is exposed to heavy static and dynamic propeller loads exerted to the bearing surface by the propeller shaft [12]. Early arrangements used bearing materials such as lignum vitae (a very 26

dense sort of timber) which were lubricated by seawater. Currently, most of ocean-going ships use a propeller shaft typically supported by oil lubricated white metal bearings, with forward and aft shaft seals confining the oil within the stern tube. Seals are arranged to prevent the entry of seawater and also the loss of lubricating oil from stern tube bearing [13]. The Marine Environment Protection Committee in its document MEPC 58/INF.22 reported that, regarding seal proper function, seal manufacturers indicate the seal must leak at the shaft/seal interface (aft-to-sea, forward-to-bilges) in order for the seal to function properly. In addition, debris such as rope caught on a ship’s rotating shaft can also damage the aft seal, allowing stern tube oil to flow into the sea. Typical ocean-going ship stern tubes contain about 1500 liters of oil so, even at a conservative leakage rate of 6 l/day, stern tube oil pollution from normal operations can be estimated to be over 80 million liters annually [13]. Environmentally friendly materials and practices are other important issues to be taken into consideration. The International Maritime Organization (IMO) by the International Convention for the Prevention of Pollution from Ships (MARPOL) Annex I - Regulations for the prevention of pollution by oil, revised in 2003, covers prevention of pollution by oil from operational measures as well as from accidental discharges. Also, the recent amendments to MARPOL Annex VI Regulations for the prevention of air pollution from ships, make mandatory the Energy Efficiency Design Index (EEDI), for new ships, and the Ship Energy Efficiency Management Plan (SEEMP) for all ships. The regulations apply to all ships of 400 gross tonnage and above and are planned to enter into force on 1st January 2013 [14] to [16]. From the environmental aspect there is no doub that the use of modern polymer bearings, rather than the usual white metal ones, eliminates the possibility of pollution due to oil leakage. Therefore, the present study examines practicability and advantages of polymer propeller shaft bearings compared to conventional white metal bearings investigated on real ships. 1 MATERIALS AND METHODS Five different materials currently used for ship journal bearings were considered for the models calculation: One tin based alloy was used as comparative metallic material towards four different types of polymers. Table 1 presents the materials elastic modulus and Poisson’s ratio obtained from literature [2] and [8].

Roldo, L. – Komar, I. – Vulić, N.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 25-31

Table 1. Mechanical properties of materials used in ships stern tube journal bearing design White Metal Material Tin-base babbitt (White Metal) Polymeric Material Thermoplastic polyether polyurethane elastomer (polyether PU) Carbon reinforced composite with an epoxy matrix (CRE) Flexible-chain polyamide type 6 (PA type 6) Polytetrafluoroethylene unfilled (PTFE)

Young’s Modulus Poisson’s [MPa] ratio 55,000 0.35 253

0.467

10,000

0.25

3,530

0.36

552

0.46

where µ is coefficient of friction, Ft friction force [N], W bearing radial load [N] and τ shear stress [Pa]. Power loss in the bearing can be calculated by equation:

According to Jocanović et al. [17] in the process of hydraulics systems design various software simulation systems are used. However, the increase of efficiency of the designed hydraulic systems can be achieved in two ways: by making design modifications based on reliability theory or based on monitoring of system operating parameters. In this case, to determine the properties of the white metal bearing oil lubricant film and its load Reynolds differential equation is used, which is solved, for the real conditions, by the numerical method of finite difference [18]. This numerical method relies on the fact that a function can be represented with a sufficient accuracy over a small range by a quadratic expression. To solve the Reynolds equation (Eq. (1)) (which is expressed in terms of lubricant film thickness h, pressure p, journal velocity vj and lubricant dynamic viscosity η) by finite difference method the equation is transformed into its dimensionless form. Assuming that η = constant for a given temperature, the Reynolds equation describes lubricant pressure distribution as a function of journal speed, bearing geometry, and lubricant viscosity in stationary hydrodynamic lubrication in journal bearing:

∂  3 ∂p  ∂  3 ∂p  ∂h  = 6v jη . (1) h + h ∂x  ∂x  ∂y  ∂y  x

In journal bearings the friction coefficient is the ratio of circumferential friction force divided by the load. L 2π RB

µ=

Ft = W

∫ ∫

τ dxdy

∫ ∫

pdxdy

0 0 L 2π RB 0

, (2)

[W ] , (3)

where vj is journal velocity [m/s]. In another way by frictional power in a bearing or the amount of heat generated which is used for model validation measuring bearing working temperature on a real ship:

1.1 Calculation Model

Ploss = µ ⋅ W ⋅ v j ,

Pth = ρ ⋅ cv ⋅ Q ⋅ (Tout − Tin ) ,

[ W ] , (4)

where Tin is bearing inlet lubricant temperature [°C], Tout bearing outlet lubricant temperature [°C], Ρ density of lubricant [kg/m3], cv specific heat capacity of the lubricant [J/kgK] and Q total lubricant flow rate [m3/s]. Unlike white metal, polymer bearing works in EHL regime, in which the elastic deformation of contact surfaces has a significant impact on the minimum thickness of the created lubricant film. This lubrication regime is described by Hamrock and Dowson expressions for calculating the minimum and central thickness of the lubricant film under the conditions of the bearing EHL regime, as follows:

2 H min = 8.7GE 0.67 (1 − 0.85e −0.31k ) (U red / W ) , (5)

2 H c = 11.15GE 0.67 (1 − 0.72e −0.28k ) (U red / W ) , (6)

where H min is nondimensional minimal film thickness, H c nondimensional central film thickness, Ured nondimensional speed, W nondimensional load parameter, GE nondimensional elasticity andk nondimensional ellipticity. Starting from the theory available in the literature for modeling HL and EHL lubrication, an own computer program has been developed for the calculation of HL and EHL bearings lubrication as well as lubricant flow through the bearing. Calculation methods and models have been validated comparing with collected data from three types of the ships in service. 1.2 Calculation Parameters For the actual power loss [kW] calculations software composed of two modules S11partialRJB and S11isoviscRJB was developed [19]. Computational models have been verified experimentally based upon

Design and Materials Selection for Environmentally Friendly Ship Propulsion System

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 25-31

data of stern tube bearing operating temperatures obtained from three types of ships (bulk carrier, container ship and RO-RO catamaran passenger ship). Ship engines speed (revolutions per minute) were within interval of 30 to 121 rpm with a step of 10 for bulk carrier, 22 to 109 rpm with a step of 10 for container ship and 100 to 608 rpm with a step of 100 for catamaran. For the white metal oil lubricated bearing a hydrodynamic lubrication model was used – S11partialRJB [19]. The core of numerical calculations (finite difference methods for the numerical solution of Reynold’s equation) was based upon Matlab program partial taken from Stachowiak and Batchelor [18]. The polymeric water lubricated bearings, due to elastic deformation of the lubricated surfaces, needed a different approach: i.e. an elastohydrodynamic lubrication (EHL) program. S11isoviscRJB calculates polymer bearing elastic deflection based on Hamrock and Schmid [2] elastohydrodynamic lubrication analysis of isoviscous-elastic body lubrication regimes [18]. Validation of these models is based upon the actual bearing temperatures for different driving regimes obtained from real ships in service. These ships were a bulk carrier of 50,000 deadweight tonnage (DWT), a container ship of 11,000 twentyfoot equivalent units (TEU) and a catamaran of 496 gross tonnage (GT). Parameters of the lubricants to ships are: 1. Oil density ρ = 910 kg/m3, kinematic viscosity at 30 °C is ν = 175 mm2/s and specific heat capacity 1922 J/kgK. 2. Average seawater density r = 1025 kg/m3 at average temperature of 15 °C, considering trading area of the actual ships. At this condition the seawater as lubricant has kinematic viscosity of 1.1843 mm2/s and dynamic viscosity of 1.21·10-3 Pa·s. Table 2 presents relevant ship design parameters for power loss calculation using white metal bearing and polymer bearing of the aft stern tube for bulk carrier, container ship and catamaran respectively.

Additionally, based upon the results of the calculation the elastic line of shafting, the bulk carrier bearing was under a constant load of 225 kN, the container ship constant bearing load was 1325 kN and catamaran bearing under constant load of 3.6 kN with an arc bearing angle of 360°. Further analysis of polymer stern tube bearings installation instead one of white metal is based on the analysis of fuel saving within the specified period of exploitation of ships of twenty years, and based on consumption of fuel. Fuel consumption (Cf) of these items is based on the data from IMO of the average period of exploitation of the ship of 330 days per year and can be calculated by the formula [13] and [19]:

C f = Ploss ⋅ ms ⋅

24texp 1000

[ kg/year ] ,

where Ploss is power loss in stern tube bearings due to hydrodynamic friction [kW], ms specific fuel oil consumption (SFOC) of the propulsion engine [g/kWh], and texp number of days of ship exploitation per year. 2 RESULTS AND DISCUSSION 2.1 Comparative Parameters Results for White Metal and Polymer Bearing Figs. 1 to 3 show the comparison of power losses in the aft stern tube bearing in case of white metal and polymer applications as bearing material for bulk carrier, container ship and catamaran ship respectively. As shown in Fig. 1 the bulk carrier power loss due to friction of the polymeric bearings is approximately 6 (polyether PU) to 9 times (CRE) smaller than the power loss in the bearing of the white metal at a maximum speed of the propeller shaft of 121 rpm. Fig. 2 shows that the container ship power loss due to friction in the polyether PU bearing is about 6 times less and the carbon reinforced epoxy is nearly 8.5 times less than the power loss in the bearing of the

Table 2. Design parameters for the aft stern tube white metal and polymer bearing of the bulk carrier, container ship and catamaran Description Bearing nominal diameter Bearing length Bearing diametral clearance Journal diameter

Parameters

Dimensions

DB L Z Di

mm mm mm mm

Values Bulk carrier 469.8 950 0.8 469

White metal bearing Values Values Container Catamaran ship 991.2 120.22 2030 240 1.2 0.22 990 120

Roldo, L. – Komar, I. – Vulić, N.

Values Bulk carrier 516.59 950 1.59 469

Polymer bearing Values Values Container Catamaran ship 1,072.92 120.36 2030 240 2.92 0.36 990 120

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 25-31

white metal at a maximum speed of the propeller shaft of 109 rpm. As presented in Fig. 3 that the catamaran ship power loss due to friction in the polyether PU bearing is about 36 times less and the carbon reinforced epoxy is nearly 73 times less than the power loss in the bearing of the white metal at a maximum speed of the propeller shaft of 608 rpm. 2.2 Comparative Results of Energy Efficiency for White Metal and Polymer Bearing

Fig. 1. Comparison of bulk carrier power loss of stern tube WM bearing versus various polymer bearings at propeller shaft revolution from 30 to 121 rpm

Table 3 presents the overall comparative analysis of fuel consumption due to power loss because of hydrodynamic friction in the stern tube bearings, and fuel savings in the case of application of polymer bearings instead of white metal. The average specific fuel consumption of the propulsion engines based on the manufacturer data for bulk carrier and container ship is 180 g/kWh, and for catamaran is 210 g/kWh. Table 3. Comparative analysis of fuel consumption and power loss for white metal (WM) and polymer (Pol.) bearings

Fig. 2. Comparison of container ship power loss of stern tube WM bearing versus various polymer bearings at propeller shaft revolution from 22 to 109 rpm

Ship Type Data of power loss in stern tube bearing due to friction (W) Bulk Container Bearing material Catamaran carrier ship White metal – WM 5875 59617 1923 Polymer – Pol. 970 11103 53 Power loss difference 4905 48514 1870 WM – Pol. Data of fuel consumption based on the difference in power loss (WM) Bulk Container Catamaran carrier ship SFOC [g/kWh] 180 180 210 Daily fuel consumption [kg] 21.2 209.6 9.4 Exploitation days per year 330 330 330 Yearly fuel consumption [t] 7 69 3 Fuel consumption per 20 140 1380 60 years *[t] * Represents a saving of fuel in the case of application of polymer instead of white metal stern tube bearing.

Fig. 3. Comparison of catamaran power loss of stern tube WM bearing versus various polymer bearings at propeller shaft revolution from 100 to 608 rpm

The power loss is caused by hydrodynamic friction in the stern tube bearings for analyzed ships (fuel savings per item) using polyether PU instead of the conventional white metal bearings. Table 4 presents the projecting results of fuel saving on a sample of 1000 ships per type over 20 years of exploitation for the application of polymer bearing. Owing to projecting results for the application of polymer sterntube bearings at the global level on

Design and Materials Selection for Environmentally Friendly Ship Propulsion System

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 25-31

Table 4. Projecting results of fuel saving for the application of polymer stern tube bearings at global level on a sample of 1000 ships per type in 20 years of exploitation Fuel consumption per 20 years per ship [t] 140 1,380 60

Ship type Bulk carrier Container ship Catamaran

Fuel consumption per 20 years per 1000 ships [t] 140,000 1,380,000 60,000

Total [t] 1,580,000

Table 5. Projected results of oil leakage from the seal in normal operation of ocean-going vessels equipped with WM stern tube bearings of a sample of 1000 ships per type within the period of 20 years Ship type Bulk carrier Container ship Catamaran

Stern tube oil leakage rate per day [liter] 6 12 6

Stern tube oil leakage rate per ship per 20 years of exploitation [liter] 39,600 79,200 39,600

a sample of 3000 ships in the exploitation period of 20 years, it can be observed from Table 4 that the use of polymer sterntube bearings realises significant savings in fuel and oil consumption in all three ships taken into consideration. Table 5 shows the projected results of oil leakage during normal operation of ships equipped with white metal stern tube bearings at global level using a sample of 3000 ships in 20 years of exploitation. Oil consumption from the journal seal is controlled to the minimum acceptable to maintain the propulsion system functional, however by design it is essential to have oil at the mating surfaces. Oil consumption is always lost directly to the sea thus contaminating the environment. As per Lloyd’s Register Class Society Seal Type Approvals data, sterntube oil leakage rate amounts to 6 liters/day from normal operations based on seals in a laboratory condition running in clean and controlled environments [20]. In addition to fuel efficiency oil leakage into the seawater is also an important issue to be taken into consideration. Tabel 5 shows that the use of water lubricated polymer bearings eliminates sea contamination caused by propulsion system seal leakage. 3 CONCLUSIONS The analysis focuses on the value of effective power loss in sterntube bearings due to viscous friction in the lubricant film comparing white metal bearings with polymer ones. Therefore, reduction of the system power loss is achieved using lubricant with low viscosity such as seawater in case of polymer bearing. Comparing the power loss of the sterntube polymeric materials with the white metal and taking 30

Stern tube oil leakage rate per 1000 ship per 20 years of exploitation [liter] 39,600,000 79,200,000 39,600,000

Total stern tube oil leakage [liter] 158,400,000

into consideration the bulk carrier, container ship and catamaran ship, it is possible to verify that the actual power loss is, at least 6 times less for bulk carrier and container ship and 36 times less for catamaran (due to high nominal shaft rpm) using sterntube polymer bearings. Among polymeric bearing materials it was observed that the carbon reinforced composite with an epoxy matrix works more efficiently than other calculated polymer applications. Another important finding is that the overall bulk carrier power loss is 10 times smaller compared to the container ship. This fact is attributed to the sterntube system design as the bulk carrier has only one aft sterntube bearing, while container ship has aft and forward sterntube bearings due to bigger shaft dimension and respectively higher bearing load. Projecting the results of the economic reason for the application of polymer sterntube bearings at the global level on a sample of 3000 ships (1000 of each type) in the exploitation period of 20 years, 1,580,000 tonnes of fuel and 158,400,000 liters of lubricant oil may be saved. Therefore, the use of polymer sterntube bearings shows the possibility of significant savings in fuel and oil consumption. This would achieve not only significant financial savings but also improved energy efficiency of the ship. Therefore, it would also contribute to meeting the IMO ship environment requirements and related energy efficiency design index. Implementation of these models and a materials selection approach can lead even to a solid basis to propose a different design approach to ship designers: polymer bearings instead of white metal bearings. There may be some drawbacks to this solution, such as the need for careful machining of shafts

Roldo, L. – Komar, I. – Vulić, N.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 25-31

and the bearings, necessity of proper preparation (e.g. filtration) of sea-water, etc. that have not been considered within the scope of this paper. The use of polymer sterntube bearings shows the possibility of significant savings in fuel and oil consumption and contributes the environment ship requirements and related energy efficiency design index. Furthermore, the analysis defined the optimisation point, i.e. the time period after which the payment of higher initial investment for installation of polymer sterntube bearings will show to be reasonable. This gives the possibility to the ship-owners to take a decision whether and when it is worthwhile mounted polymer sterntube bearing instead of the white metal, based on the scientifically proven methodology. 4 ACKNOWLEDGEMENTS Liane Roldo thanks the support from Brazilian Research Agency CNPq – Conselho Nacional de Desenvolvimento Tecnológico. 5 REFERENCES [1] ASTM B 23-00. Standard Specification for White Metal Bearing Alloys 2010, ASTM International, West Conshohocken. [2] Hamrock, B.J., Schmid, S.R., Jacobson, B.O. (2004). Fundamentals of Fluid Film Lubrication. 2nd ed. Marcel Dekker, New York, Basel, DOI:10.1201/9780203021187. [3] Ünlü, B.S., Atik, E. (2007). Determination of friction coefficient in journal bearings. Materials & Design, vol. 28, no. 3. p. 973-977, DOI:10.1016/j. matdes.2005.09.022. [4] Rules for the Classification of Ships 2009, part 25. Croatian Register of Shipping, Split. [5] Feyzullahoǧlu, E., Şakiroǧlu, N. (2010). The wear of aluminium-based journal bearing materials under lubrication. Materials & Design, vol. 31, no. 5, p. 2532-2539, DOI:10.1016/j.matdes.2009.11.037. [6] Ginzburg, B.M., Tochil’ikov, G.D. Bakhareva, V.E., Anisimov, A.V., Kireenko, O.F. (2006). Polymeric Materials for Water-Lubricated Plain Bearings. Russian Journal of Applied Chemistry, vol. 79, no. 5, p. 695706, DOI:10.1134/S1070427206050016. [7] Sedlaček, M., Vilhena, L.M.S., Podgornik, B., Vižintin, J. (2011). Surface Topography Modelling for Reduced Friction. Strojniški vestnik - Journal of Mechanical Engineering, vol. 57, no. 9, p. 674-680, DOI:10.5545/ sv-jme.2010.140.

[8] Harper, C.A. (2002). Handbook of Plastics, Elastomers, and Composites. 4th ed. McGraw-Hill Handbook, New York. [9] Ünlü, B.S., Atik, E., Koksal, S. (2009) Tribological properties of polymer-based journal bearings. Materials & Design, vol. 30, no. 7, p. 2618-2622, DOI:10.1016/j. matdes.2008.11.018. [10] Bielinski, D.M., Glab, P., Slusarski, L. (2006). New approach to study tribological properties of polymer materials. A case of car windshield wipers. Journal of Achievements in Materials and Manufacturing Engineering, vol. 15, no. 1-2, p. 71-78. [11] Jia, J.H., Zhou, H.D., Gao, S.Q., Chen, J.M. (2003). A comparative investigation of the friction and wear behavior of polyimide composites under dry sliding and water-lubricated condition. Materials Science and Engineering: A, vol. 356, no. 1-2, p. 48-53, DOI:10.1016/S0921-5093(03)00052-2. [12] Komar, I., Vulić. N., Antonić, R. (2009). Specifics of shafting alignment for ships in service. Promet Traffic&Transportation, vol. 121, no. 5, p. 349-357, DOI:10.7307/ptt.v21i5.250. [13] Use of seawater lubricated tube bearings to eliminate sterntube oil pollution from ships Resolution MEPC 58/INF 22. IMO 2008. (2008). Marine Environment Protection Committee, International Maritime Organization, London. [14] International Convention for the Prevention of Pollution from Ships. Annex I Regulations for the prevention of pollution by oil (2003). International Maritime Organization, London. [15] Annex VI Regulations for the prevention of air pollution from ships (2005). International Maritime Organization, London. [16] Energy Efficiency Design Index. Resolution MEPC.203(62) (2011). International Maritime Organization, London. [17] Jocanović, M., Šević, D., Karanović, V., Beker, I., Dudić, S. (2012). Increased efficiency of hydraulic systems through reliability theory and monitoring of system operating parameters. Strojniški vestnik Journal of Mechanical Engineering, vol. 58, no. 4, p. 281-288, DOI:10.5545/sv-jme.2011.084. [18] Stachowiak, G.W., Batchelor, A.W. (2005). Engineering Tribology, 3rd ed. Elsevier Butterworth-Heinemann, Oxford, p. 728-736. [19] Komar, I. (2012). Contribution to the selection methodology of the most convenient marine propulsion stern tube bearings, Ph.D Thesis, Faculty of Maritime Studies, Rijeka. [20] Higgenbottom, A. (2003). Coastguard non-polluting sterntube sealing system. RINA International Conference for the Design and Operation of Container Ships, London, p. 53-60.

Design and Materials Selection for Environmentally Friendly Ship Propulsion System

Received for review: 2012-06-11 Received revised form: 2012-10-03 Accepted for publishing: 2012-11-12

Measurement Uncertainty Assessment in Remote Object Geolocation

Kuščer, L. – Diaci, J. Lovro Kuščer* – Janez Diaci University of Ljubljana, Faculty of Mechanical Engineering, Slovenia Remote object geolocation is a process of determining the geographic location of distant objects, which is often required in geology, military applications, surveying and elsewhere. Unlike in the case of widely used satellite localization and navigation solutions, there are very few studies of the measurement uncertainties and their propagation in the case of geolocating distant, non-cooperative objects. To investigate this specific segment of navigation we developed an experimental system built entirely of COTS (Commercial Of-The-Shelf) components. The aim of this work is to study the possibility for assessing the measurement characteristics of such integrated systems, based on the specified uncertainties of the individual measuring devices. Keywords: remote object geolocation, measurement uncertainty, Monte Carlo simulation

0 INTRODUCTION Global navigation satellite systems (GNSS) provide the means for fast determination of the geographic location, speed and time. Combined with the modern satellite receivers they enable high positioning precision and reliability in various environments. However, it is sometimes not sufficient to precisely establish only the position of the receiver. In some circumstances it is necessary to obtain the position of a distant, inaccessible or hazardous object. That is where the demand for the remote object geolocation arises. The capability of the remote geolocation is often required in geology [1], military applications [2] and [3], surveying [4], agriculture, forestry [5] and elsewhere. Many research works investigated different techniques for acquiring accurate remote object (or target) positions from ground or unmanned air vehicles (UAV) [6] and [7]. In order to obtain the target geolocation the UAV systems commonly employ a gimbal camera, a geo-referenced terrain database and a navigation system. Some advanced aerial systems employ SAR (Synthetic Aperture Radar) which allows high resolution imaging at long stand-off ranges [3]. In the case of target geolocation from the ground, laser rangefinders combined with satellite and inertial navigation systems [8] to [11] are often utilized. The main challenge in remote object localization is to precisely measure the position and attitude of the measuring system. While the estimation of position (using satellite navigation systems) is generally straightforward, the attitude determination with sufficient accuracy can be more difficult. The associated measurement uncertainties of the position [12] and especially the attitude measurements [13] 32

contribute significantly to the evaluated remote object’s position uncertainty. To study the options for remote object geolocation we designed and built an experimental measuring system using only COTS (Commercial Of-TheShelf) components. In this article we investigate to what extent it is possible to predict the measurement uncertainty of the integrated measuring system based on the manufacturer specifications for each employed COTS device. We also present the experimental work aimed at determining the actual measuring characteristics and examine the opportunities for enhancing the overall performance at measuring distances up to 20 km. 1 GEOLOCATING REMOTE OBJECTS The following section presents a brief overview of the method used for determining the geographic location of a remote object in the World Geodetic System 1984 (WGS 84). The WGS 84 coordinate system [14] was selected since it is currently used as the reference coordinate system for the Global Positioning System (GPS). The first step is the determination of the geographic location of the measuring system (origin) which is given by its latitude, longitude and altitude (φo, λo, ho) in WGS 84 coordinate system. Next, the relative position of the remote object with respect to the origin is obtained. The relative position is given by the distance (lor), azimuth (αor) and elevation (βor) in a spherical coordinate system (Fig. 1). Since the input coordinates (origin, relative position of the remote object) are given in two different coordinate systems, we transform them to a common coordinate system in order to calculate the geographic location of the remote object. For this

*Corr. Author’s Address: Univerza v Ljubljani, Fakulteta za strojništvo, Aškerčeva 6, 1000 Ljubljana, Slovenija, lovro.kuscer@fs.uni-lj.si

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 32-40

purpose the WGS 84 Cartesian coordinate system is used. This is a right-handed Cartesian coordinate system with the origin at the Earth’s centre of mass, the Z-axis in the direction of the IERS (International Earth Rotation and Reference Systems Service) reference pole, the X-axis as the intersection of the IERS reference meridian and the plane through the origin normal to the Z-axis, and the Y-axis completing the right-handed orthogonal coordinate system (Fig. 2a). The transformation of the WGS 84 coordinates from latitude, longitude and altitude to the Cartesian coordinates is given by Eq. (1) [15] where xo, yo and zo are the Cartesian coordinates of the origin, N is the radius of curvature in the prime vertical, e is the first eccentricity, and a and b are the semimajor and semiminor axis of the WGS 84 reference ellipsoid, respectively.

Cartesian coordinate system. By adding this offset to the Cartesian coordinates of the origin, the coordinates of the remote object are obtained (Eq. (3)).  xr   xo       yr  =  yo  + z  z   r  o (3)  xloc sin ϕo cos λo − yloc sin λo + zloc cos ϕo cos λo    +  xloc sin ϕo sin λo + yloc cos λo + zloc cos ϕo sin λo  .   zloc sin ϕo − xloc cos ϕo  

   xo   ( N + ho ) cos ϕo cos λo       yo  =  ( N + ho ) cos ϕo sin λo  , z    o N 1 − e 2 + ho sin ϕo   

( (

a 1 − e 2 sin 2 ϕo

)

, e=

a 2 − b 2 (1) . a

Fig. 1. Determination of the geographic location of a remote object

Next, the Eq. (2) is used to transform the relative position of the remote object (given in spherical coordinates) to the Cartesian coordinates (xloc, yloc, zloc) in a local coordinate system (Fig. 2b).

 xloc   −lor cos β or cos α or       yloc  =  lor cos β or sin α or  . (2) z    lor sin β or  loc   

The coordinates in the local Cartesian coordinate system are then converted to the offset in the WGS 84

b) Fig. 2. Coordinate systems in remote object geolocation; a) WGS 84 coordinate system, b) local coordinate system

Finally, the Cartesian coordinates of the remote object (xr, yr, zr) are transformed back to the latitude, longitude and altitude (φr, λr, hr) in the WGS 84 coordinate system. This task is solved numerically by utilizing the Hirvonen and Moritz iterative method [16]. The described procedure yields the remote object’s coordinates in the WGS 84 coordinate system. 2 EXPERIMENTAL SYSTEM The experimental system is designed for use on land vehicles and consists of an inertially stabilized pan-tilt

Measurement Uncertainty Assessment in Remote Object Geolocation

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 32-40

unit, a laser rangefinder, a GNSS receiver/compass, an electronic inclinometer and a CCTV video camera (Fig. 3). The system is controlled using a portable computer with custom software.

cos (α )  R y (α ) =  0  sin (α )   cos (α )  Rz (α ) =  − sin (α )  0 

0 − sin (α )   1 0  , (5) 0 cos (α )  sin (α ) 0   cos (α ) 0  . 0 1 

In Eq. (4), the Rx, Ry and Rz are the rotation matrices for the coordinate system rotation about x, y and z, respectively and v’ is the representation of the unit vector v in the transformed coordinate system. From the vector v’, the azimuth and elevation are obtained using Eq. (6):

Fig. 3. Experimental system: (1) the laser rangefinder and video camera unit, (2) the pan-tilt unit, (3) the GNSS antenna, (4) the electronic inclinometer, (5) the communication and power interface, (6) the portable computer/controller

The laser rangefinder with a measuring range of 50 m to 20 km and the video camera are mounted on the pan-tilt unit, whereas the GNSS receiver/compass antennas and electronic inclinometer are attached to the platform base. Due to this setup, the distance (lor) and origin geographic location (φo, λo, ho) are obtained directly from the rangefinder and the GNSS receiver readings, while the azimuth (αor) and elevation (βor) are calculated employing the pan and tilt angles of the pan-tilt unit and the platform base orientation. The later is measured with the GNSS compass (yaw) and the electronic inclinometer (pitch and roll). To carry out the calculation of the azimuth and elevation from the measured quantities, we define a right-handed Cartesian coordinate system with the X-axis in the direction of the rangefinder laser beam and the Y-axis parallel with the tilt axis of the pan-tilt unit. In this coordinate system, we define a unit vector v that points in the direction of the positive X-axis. Next, we perform five consecutive coordinate system rotations to account for the tilt (θ), pan (δ), roll (γ), pitch (β), and yaw (α) by utilizing Eq. (4):

v ' = Rz (α ) Ry ( β ) Rx ( γ ) Rz (δ ) Ry (θ ) v , (4)

where:

1 0 0    Rx (α ) = 0 cos (α ) sin (α )  , 0 − sin (α ) cos (α )   

 vy '  α or = arctan  ,  vx '  (6) β or = arcsin ( vz ' ) ,

where the arctan is the four quadrant arctangent and vx’, vy’ and vz’ are the coordinates of the vector v’. 3 MEASUREMENT CHARACTERISTIC The measurement performance of the system depends on the measurement uncertainties of the employed devices. The manufacturer specification of the measuring equipment in the experimental system is presented in Table 1. Table 1. Manufacturer specification for measuring equipment GNSS Laser GNSS positioning rangefinder compass module Containment 0.60 m ±3 m ±0.15° limits (DGPS) Containment 95% 95% 68% probability Latitude: 10-5 ' Longitude: Resolution 2m 0.01° 10-5 ' Altitude: 0.1 m Measuring equipment

Electronic inclinometer (pitch and roll) ±0.5° 95%

0.05°

The uncertainty of each individual measurement contributes to the derived position uncertainty of the remote object in a specific manner. There are several methods for evaluating the combined effect of the individual measurement uncertainties on the result. The law of propagation of variances is most commonly used when the result can be calculated with a closed form expression. In the case of nonlinear

Kuščer, L. – Diaci, J.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 32-40

relations, the system equations are first linearized using the Taylor series [17]. However, in this case, a numerical solution (Hirvonen and Moritz iterative method) is employed to determine the position of the remote object. Hence, we use the Monte Carlo simulation [18] to [21] to assess the propagation of the measurement uncertainties. The term Monte Carlo denotes a set of stochastic computational techniques that are widely used for computer simulations in various fields of science and engineering. In the case of probabilistic uncertainty analysis, a probability density function is assigned to each measuring device. Then, repeated numerical calculations of the remote object’s position using randomly selected values from the assigned distributions are performed. This basically simulates the execution of multiple measurements in the presence of measurement uncertainties. The calculated positions of the remote object form a distribution which represents the combined effect of each measuring device and its associated uncertainty. The statistical analysis of the obtained distribution yields an estimation of the measurement uncertainty in the remote object’s position. 4 SIMULATION To perform the simulation the uncertainty estimates of the individual measuring devices need to be obtained. According to [22] this can be achieved by two different approaches. The first one (Type A) is based on retrieving relevant statistical information form repeated measurements, while the second (Type B) relies on manufacturer specification, past experience, or other sources. Since we are investigating the possibility of predicting, and not measuring the uncertainty of an integrated system, we utilize Type B estimate. The manufacturers often specify the uncertainty of their measuring equipment by the error containment limits and containment probability, while the underlying distribution is usually not provided. Commonly, the normal distribution is assumed applicable for the combined error of the measuring equipment with a central tendency and symmetrical containment limits. Since there is no indication that would encourage the use of a different probability distribution, we assume the normal distribution of the combined error for all measuring devices. When the error is normally distributed, the uncertainty (u) is obtained from Eq. (7) [23] where ±L represents the containment limits, p the containment

probability, and Φ-1() the inverse normal distribution function.

L . (7) −1  1 + p  Φ    2 

By applying Eq. (7) to the manufacturer specifications for error containment limits and containment probability (Table 1), the standard uncertainties of 1.5 m for distance measurements, 0.3 m for position of the origin, 0.15° for azimuth, and 0.26° for elevation are obtained. In order to illustrate the key characteristics of the results of Monte Carlo simulation Table 2 presents measurement uncertainties at four measuring distances (100 m, 1, 10 and 20 km) and 0.0° elevation angle. The uncertainties have been determined using 10,000 simulation runs for each distance. To facilitate the discussion the combined 2D uncertainty is presented along with its constituents in two separate directions denoting as longitudinal and lateral the direction of the rangefinder laser beam and its orthogonal, respectively. According to [23] the presented uncertainties are equivalent to the standard deviations of the error distributions and are evaluated by analyzing the simulation points. From the reported results it is evident that the increase in the measuring distance significantly influences the uncertainty in the lateral direction, while the one in longitudinal direction remains virtually constant for the simulated distances. The stated combined uncertainty enables the evaluation of confidence limits and confidence levels at certain measurement distance. For example, by multiplying the combined uncertainty of 3.1 m (object at 1 km distance) with the coverage factor of 2, we obtain the expanded uncertainty of ±6.2 m with a confidence level of approximately 95% (assuming the errors are normally distributed). In other words, 95% of all measurements performed at 1 km stand-off range are expected to fall within a circle with a 6.2 m radius around the true position of the distant object. The expected impact of the elevation uncertainty on the 2D combined uncertainty is minimal at the 0.0° elevation angle. When increasing the elevation angle, its uncertainty increasingly affects the resulting uncertainty in longitudinal direction. This effect is demonstrated in Fig. 4a, which shows the dependence of the uncertainty in longitudinal direction on the measuring distance and elevation angle. Despite the evident influence of the elevation angle value on its uncertainty propagation, this

Measurement Uncertainty Assessment in Remote Object Geolocation

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 32-40

a) b) Fig. 4. a) Dependence of the standard uncertainty in longitudinal direction on the measurement distance at different elevation angles, and b) dependence of the relative uncertainty contributions of the simulated quantities on the measurement distance

only contributes a rather small part to the combined uncertainty of the remote object’s position. This is clearly seen in Fig. 4b, where the relative contributions of the uncertainties in the simulated quantities to the overall measurement uncertainty are presented. The contributions are calculated for the elevation angle of 2.5° and are normalized to the sum of the resulting uncertainties of the individual quantities. It has been noticed that at distances below 1 km the most significant uncertainty source is attributed to distance measurement, while at greater distances the azimuth uncertainty prevails. Furthermore, the amount of the elevation uncertainty contribution, which highly depends on the elevation angle value (Fig. 4a), becomes the second most important uncertainty source at distances above 10 km.

closest two objects were set up for the needs of the experiments (1×1 m targets), whereas two existing objects were used for long distance measurements. The positions of the closest two objects were determined using a high performance GNSS receiver also exploiting the corrections from the EGNOS (European Geostationary Navigation Overlay Service) satellite based augmentation system (SBAS).

Table 2. Results of the simulation-based uncertainty analysis Distance Elevation Longitudinal Standard uncertainty [m] Lateral Combined uncertainty [m]

100 m 0.0° 1.6 0.4 1.7

1 km 0.0° 1.6 2.6 3.1

10 km 0.0° 1.6 26.1 26.2

20 km 0.0° 1.7 52.2 52.2

5 EXPERIMENTAL RESULTS AND COMPARISON WITH SIMULATION A series of field tests were performed in order to verify the simulation results. An open field with an unobstructed view of the sky was used as the test polygon (Fig. 5) with four distant objects located at different distances from the measuring system. The 36

Fig. 5. Layout of the test polygon

The measurements were performed over a period of 30 minutes with an unobstructed view of the sky. According to the manufacturer specification of the GNSS receiver and the available data on EGNOS system, it is reasonable to assume that the standard uncertainty of the measured 2D positions is about 1 m. The positions of the existing distant objects were obtained from the digital orthophoto images (DOF050) [24] with the specified standard uncertainty of ±1 m. Once the existing objects were identified in the digital image, their WGS 84 coordinates were determined employing the geographic information system.

Kuščer, L. – Diaci, J.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 32-40

a) b) Fig. 6. Distributions of errors obtained by; a) Monte Carlo simulation and b) field tests

The distant objects were positioned at distances of 130 m (object A), 1.0 km (object B), 7.4 km (object C) and 18.5 km (object D) from the measuring system. The corresponding azimuths of the objects A, B, C, and D (observed from the origin) were 85.5, 198.2, 139.8, and 214.3°, respectively. Since the aim is to investigate the uncertainty in instantaneous geolocation of distant objects, all measurements were performed by employing only the current readings from the sensors without averaging over time. The period of measurements (limited by the repetition rate of the laser rangefinder) was about 6 s. Approximately 200 measurements were performed for each distant object over a period of 20 minutes. Moreover, we performed a Monte Carlo simulation with the distances, azimuths and elevations of the four objects in the test polygon. The simulation results along with the field test results are displayed in Fig. 6. In addition, the simulation output was compared with the results obtained by the use of the law of propagation of variances. We notice that the simulation and measurement points for objects A and B in Fig. 6 are not distributed evenly but are rather grouped in several clusters. This is caused by the 2 m resolution of the range measurements which is also considered in the simulation. However, this effect is not observed for the objects C and D because of larger distances and the elevation angles of 1.8 and 2.9°. Due to the increased distance and elevation angle the measurement uncertainty of the electronic inclinometer becomes more pronounced (Fig. 4a) and overrides the effect of low range measurement resolution. When comparing the simulation and measurement results in Fig.

6, it is evident that a considerable measurement bias is present in the longitudinal direction. The observed discrepancies originate from the distance measurements that tend to be smaller than the actual distances for objects A, B, and C and larger for object D. However, these discrepancies appear to be systematic and can therefore be corrected for by the calibration over the entire measuring range in order to obtain the distance measurement characteristic that is consistent with the manufacturer specification. Another distinct difference between the measurements and simulation outputs is the scatter in longitudinal direction for objects A and B, which is considerably larger in the simulation. This is manifested through the occurrence of more separate clusters of simulation points and in larger sizes of individual clusters. Such differences imply that the actual standard deviations of range measurements and origin position measurements are smaller than the assessment based on the manufacturer specifications. Notwithstanding the fact that the discussed discrepancies in longitudinal direction are obvious, they only contribute a small part to the combined uncertainty at larger measuring distances. In these cases, the uncertainty in azimuth measurements gives rise to large errors in lateral direction as can be seen in Fig. 6. The uncertainty analysis was also performed employing the law of propagation of variances. The results of this approach are displayed in Fig. 7 as error ellipses with 95% confidence level. Usually, we assume that the underlying error distribution represented by the error ellipse is normal. In the case of objects C and D (Fig. 7b) this assumption seems

Measurement Uncertainty Assessment in Remote Object Geolocation

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 32-40

a) b) Fig. 7. Distributions of errors obtained by Monte Carlo simulation and error ellipses (95% confidence level) obtained by the law of propagation of variances; a) objects A and B, b) objects C and D

justified. However, the errors in Fig. 7a are not normally distributed and might therefore require a different representation. To describe and examine the distributions obtained by means of simulation and measurements, it is convenient to use standard deviation as a measure of dispersion and root mean square error (RMSE) as a measure of the overall performance. For the simulation, the standard deviations are equivalent to the standard uncertainties reported in Table 2. Furthermore, since no bias is incorporated in the simulation, the RMSE is equivalent to the combined uncertainty, which is in fact the standard deviation of the combined error distribution. A comparison of the simulation and the field test results (Table 3) indicates that the standard deviations of the measurements in longitudinal direction are significantly smaller than the simulated one for all objects. This implies that the random component of the range measurement error is actually smaller than expected. On the other hand, the systematic component associated with range measurements has a considerable influence on the overall position accuracy, especially at small stand-off ranges (object A). When comparing the standard deviations in lateral direction it can be noticed that the measurements and the simulations yield similar results only for smaller distances. At longer distances the azimuth measurement uncertainty, which is affected by many factors, including the number of GNSS satellites in view, satellite geometry and ionospheric activity, causes less predictable results. Nevertheless, the 38

difference between the simulated and measured RMSE is less than 5 m for objects A, B and C. Although the errors in longitudinal direction may have very little impact at long distances, they still limit the system’s performance at closer ranges. This is demonstrated in Fig. 8, where we notice that the smallest mean error is not observed at the closest object A but at object B. This is attributed primarily to the aforementioned bias in the distance measurements. However, in general, the measured positions of the more distant objects exhibit larger variations and mean errors compared to the closer ones. This is especially evident when comparing measurements of objects C and D. Such results are expected considering the dependence of the azimuth and elevation uncertainty propagation on the measuring distance. The presented results suggest that at smaller stand-off ranges (a few hundred meters) the distance measurement uncertainty dominantly affects the overall system performance, while at longer ranges the azimuth uncertainty prevails. In order to perform accurate measurements over longer distances, it is necessary to reduce the azimuth uncertainty. This is achievable either with a high accuracy azimuth measuring device (gyrocompass), which would significantly increase the price of the measuring system, or with a slightly modified use of the GNSS receiver/compass. To determine the orientation of the measuring system a calibration with a known distant object can be performed. Since the positions of the measuring system and the distant object are known, the true azimuth of the object with respect to the measuring system can be determined. The obtained azimuth

Kuščer, L. – Diaci, J.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 32-40

Table 3. Comparison of measurements and simulation

Std. deviation [m]

A Distance: 130 m Azimuth: 85.5° Elevation: 0.2° Sim. Meas. 1.6 0.7 0.4 0.4 1.7 6.3

Longitudinal Lateral

RMSE [m]

B Distance: 1 km Azimuth: 198.2° Elevation: -0.2° Sim. Meas. 1.6 0.7 2.6 2.2 3.1 3.2

C Distance:7.4 km Azimuth: 139.8° Elevation: 1.8° Sim. Meas. 1.9 0.9 19.7 11.3 19.8 15.8

Fig. 8. Measurement errors; a) objects A and B, b) objects C and D

value is then used to calibrate the azimuth sensor readings. Such calibration is possible with a geographic information system and has been included in the current version of the control software of the experimental system. 6 CONCLUSION The measurement characteristics of the developed mobile system for remote object geolocation depend on the measurement performances of each employed COTS device. To estimate the overall measurement uncertainty of the measuring system we performed a Monte Carlo simulation with the manufacturer specifications for each COTS device as the simulation input. The comparison of the simulation outputs with the field test results shows that the simulation provides sufficiently accurate estimation of the measurement uncertainty at the distances of up to 7 km. However, at smaller distances of about 100 m there is a significant bias in the range measurements that limits the system performance. In order to provide better simulation

D Distance: 18.5 km Azimuth: 214.3° Elevation: 2.9° Sim. Meas. 4.7 2.3 48.3 54.5 48.5 62.9

results for shorter ranges, this error needs to be properly characterized and included in the simulation. At longer distances, the azimuth measurement uncertainty of the GNSS compass becomes the main uncertainty source. Additional set of experiments implied that lower azimuth uncertainties can be achieved using calibration landmarks with known positions. For this purpose it is possible to use existing objects that are represented in a geographic information system or custom calibration landmarks. With the employed GNSS receiver we were able to significantly reduce the azimuth uncertainty with the custom calibration landmark placed only 130 m away from the measuring system. The obtained results show that the presented simulation-based methodology for determining the measurement uncertainty of the integrated system is able to provide a satisfactory estimate of the measurement characteristic by relying solely on the manufacturer specification of the employed measuring equipment.

Measurement Uncertainty Assessment in Remote Object Geolocation

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 32-40

7 REFERENCES [1] Xu, X., Bhattacharya J.B., Davies, R.K., Aiken, C.L.V. (2001). Digital geologic mapping of the Ferron Sandstone, Muddy Creek, Utah, with GPS and reflectorless laser rangefinders. GPS Solutions, vol. 5, no. 1, p. 15-23, DOI:10.1007/PL00012872. [2] Madison, R., DeBitetto, P., Rocco Olean, A., Peebles, M. (2008). Target geolocation from a small unmanned aircraft system. IEEE Aerospace Conference, p. 1-19. [3] Pedlar, D.N., Blake, A.P. (2005). SAR target geolocation performance. IEEE International Radar Conference, p. 212-216, DOI:10.1109/RADAR.2005.1435821. [4] Lee, J.K., Jekeli, C. (2012). A Dual-IMU/GPS based geolocation system. Journal of Navigation, vol. 65, no. 1, p. 113-123, DOI:10.1017/S0373463311000567. [5] Hopkinson, C., Chasmer, L., Lim, K., Treitz, P., Creed, I. (2006). Towards a universal lidar canopy height indicator. Canadian Journal of Remote Sensing, vol. 32, no. 2, p. 1-14, DOI:10.5589/m06-006. [6] Han, K., DeSouza, G.N. (2011). Geolocation of multiple targets from airborne video without terrain data. Journal of Intelligent and Robotic Systems, vol. 62, no. 1, p. 159-183, DOI:10.1007/s10846-010-94427. [7] Barber, D.B., Redding, J.D., McLain, T.W., Beard, R.W., Taylor, C.N. (2006). Vision-based target geolocation using a fixed-wing miniature air vehicle. Journal of Intelligent & Robotic Systems, vol. 47, no. 4, p. 361-382, DOI:10.1007/s10846-006-9088-7. [8] Grejner-Brzezinska, D.A., Toth, C., Sun, H., Wang, X., Rizos, C. (2008). Novel geolocation technology for geophysical sensors for detection and discrimination of unexploded ordnance. IEEE/ION Position, p. 9931007. [9] Lee, J.K., Jekeli, C. (2011). Rao-Blackwellized unscented particle filter for a handheld unexploded ordnance geolocation system using IMU/GPS. Journal of Navigation, vol. 64, no. 2, p. 327-340, DOI:10.1017/ S0373463310000548. [10] Kansal, S., Cook, G. (2005). Use of fiducials and unsurveyed landmarks as geolocation tools in vehicular-based landmine search. IEEE Transactions on Geoscience and Remote Sensing, vol. 43, no. 6, p. 1432-1439, DOI:10.1109/TGRS.2005.846152. [11] Savage, C.O., La Scala, B.F. (2007). Accurate target geolocation using cooperative observers. Information, Decision and Control, p. 248-253, DOI:10.1109/ IDC.2007.374558.

[12] Karimi, H.A., Roongpiboonsopit, D., Kasemsuppakorn, P. (2011). Uncertainty in personal navigation services. Journal of Navigation, vol. 64, no. 2, p. 341-356, DOI:10.1017/S037346331000055X. [13] Bell, T. (2000). Error analysis of attitude measurement in robotic ground vehicle position determination. Navigation, vol. 47, no. 4, p. 289-296. [14] El-Rabbany, A. (2002). Introduction to GPS: The Global Positioning System. Artech House Publishers, Boston. [15] Xu, G. (2007). GPS: Theory, Algorithms and Applications. Springer, Berlin. [16] Burtch, R. (2006). A comparison of methods used in rectangular to geodetic coordinate transformations. American Congress on Surveying and Mapping, p. 1214. [17] Kutin, J., Bajsić, I. (2002). An analytical estimation of the Coriolis meter’s characteristics based on modal superposition. Flow Measurement and Instrumentation, vol. 12, no. 5, p. 345-351, DOI:10.1016/S09555986(02)00006-7. [18] Dieck, R.H. (2002). Measurement Uncertainty: Methods and Applications, ISA – The Instrumentation, Systems, and Automation Society, Research Triangle Park. [19] Gusell, A., Ačko, B., Mudronja, V. (2009). Measurement uncertainty in calibration of measurement surface plates flatness. Strojniški vestnik – Journal of Mechanical Engineering, vol. 55, no. 5, p. 286-292. [20] Lugarić, L., Majdandžić, L., Škrlec, D. (2010) Countrywide Positioning of Domestic Solar Water Heating Systems using Risk Analysis and Geographical Information System. Strojniški vestnik – Journal of Mechanical Engineering, vol. 56, no. 1, p. 3-17. [21] Vignat, F., Nguyen, D.S., Brissaud, D. (2012). A Method to Determine the Impact of Geometrical Deviations on Product Performance. Strojniški vestnik – Journal of Mechanical Engineering, vol. 58, no. 9, p. 517-526, DOI:http://dx.doi.org/10.5545/sv-jme.2011.268 [22] JCGM 100:2008 (2008). Evaluation of Measurement Data – Guide to the expression of uncertainty in measurement. Joint Committee for Guides in Metrology, International Organization for Standardization, Geneva. [23] Castrup, S., Castrup, H.T. (2010). Measurement Uncertainty Analysis Principles and Methods, National Aeronautics and Space Administration, Washington DC. [24] Geopedia (2011). from http://www.geopedia.si, accessed on 2011-05-07.

Kuščer, L. – Diaci, J.

Received in review: 2012-02-21 Received revised form: 2012-08-03 Accepted for publishing: 2012-09-04

Numerical and Experimental Study of Frictional Behavior in Bending Under Tension Test Lemu, H.G. – Trzepieciński, T. Hirpa G. Lemu1,* – Tomasz Trzepieciński2

1 University

of Stavanger, Department of Mechanical and Structural Engineering and Materials Technology, Norway University of Technology, Department of Materials Forming and Processing, Poland

2 Rzeszow

This paper presents the results of frictional resistance research for three different types of sheet metals, namely from steel, brass and aluminum alloy. The research on friction behavior of these alloys was carried out by using bending under tension test. The influence of the amount of plastic deformation on friction coefficient value was investigated in dry and lubrication conditions. The material data of the sheet metals were determined from the tensile tests. Numerical simulations using the finite element method in MSC.Marc 2007r1 computer program are conducted by taking into account the strain hardening phenomenon of the sheet metals. The results show that there are minor differences between the experimental and numerical results. These may be due to the simplification where isotropic material properties are assumed and influences of structural defects are not accounted for in the numerical model. Keywords: bending under tension, stamping, friction, friction coefficient, surface roughness

0 INTRODUCTION Sheet metal forming processes like draw bending or deep drawing lead to a large plastic deformation of the material. During stamping process of drawpieces the material is drawn over a radius experiencing bending and back bending. This results in large plastic strains that lead to flow anisotropy. The induced anisotropic behavior manifests itself in the case of a strain change by different stress-strain responses depending on the type of the strain path change. While many metals exhibit a drop of the yield stress after a load reversal, some metals show an increase of the yield stress after orthogonal strain path change. The reason for this induced flow anisotropy is the development of persistent dislocation structures during large deformations. Friction behavior in sheet metal forming process depends on several parameters such as contact pressure, sliding velocity, sheet metal surface roughness, tool surface roughness, tool material and lubricant conditions [1] and [2]. Moreover, frictional resistance depends on physical and chemical factors acting on the contact surface, dynamics of the loads and temperature [1]. Studies also show that friction and material characteristics have a direct influence on the process and are sensitive to each other [3]. Furthermore, recent studies such as [4] and [5] show that the topography of a surface influences the frictional behavior of a contact surface and hence its wear. Hence, there is a need to better understand the role of friction and to find reliable methods to quantitatively determine the friction coefficient values in metal forming. In parallel, there is a growing trend to use computer simulation based research tools [6] *Corr. Author’s Address: University of Stavanger, 4036 Stavanger, Norway, Hirpa.g.lemu@uis.no

and other advanced modeling techniques [7] of sheet metal forming operations. Furthermore, understanding the precise coefficient of friction and the surface qualities requires sufficient knowledge of the tribological behavior at the interface between tool and workpiece. Two contact conditions are observed at this interface: (1) the sliding condition under compression and (2) the sliding condition under tension bending. To describe the friction in sheet metal forming by simulations, a model that quantifies friction coefficients is needed. This is complicated by the fact that any of a variety of lubrications regimes may co-exist in the sheet-tooling interface. The realistic friction models must also treat the influence of roughness and surface topography on the lubricant flow and on the asperity contact [8]. Friction coefficient is normally obtained by experiment under certain assumptions and must be obtained in a single experiment. Nowadays, there are many kinds of friction tests that are modified and developed by several researchers. The tensile strip test developed by Duncan et al. [9] is widely used. In this test a strip specimen of sheet metal was pulled over the cylindrical surfaces of pins to simulate stretching and drawing processes. The pulling force on one side of the pin was measured along with the strain in a section of the test specimen on the other side. The strip force on the second side of the pin was calculated from the measured strain using the stress–strain characteristics of the test material. In subsequent studies, Wang et al. [10] showed that the coefficient of friction increased with strip sliding distance and that increasing the pin radius resulted in a small increase in measured coefficient of friction. Weinmann et al. [11] measured the coefficient of friction in the sliding condition under 41

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 41-49

real stamping process. The tests were conducted using a modified measurement system specially designed for previous friction research [22]. The selected steel, brass and aluminum sheet metal in different temper states were tested. Tensile test was carried out in a universal testing machine to determine mechanical parameters of the samples including hardening properties (Table 1). The extensive specimens were cut under angle of 0 and 90º with respect to the rolling direction of the sheet metal. The thickness of the aluminum sheet is 0.8 mm while that of brass and steel is 1 mm. Surface roughness was measured using Taylor Hobson Subtronic 3+ instrument and the arithmetic average (Ra) of filtered roughness along both the rolling (Ra0) and transverse directions (Ra90) were registered.

tension bending developed by Littlowod and Wallace [12]. To account for severe deformation conditions, the researchers [11] suggested use of friction factor to describe tool–work friction. In the expression for the friction factor in terms of measured strain, the pin radius appears explicitly and experimental results show that friction factor decreases with increasing the pin radius. This work did not account for plastic deformation of specimen. He et al. [13] developed a new bending under tension (BUT) test, but this method only focused on the state of friction in the sheet steel bend-forming process. Many BUT tests have contributed to the knowledge about sheet forming tribology [14] to [20]. The traditional way of performing these BUT tests is by differential measurements carrying out two tests after each other, one by drawing over a fixed circular cylindrical tool-pin, the other over a freely rotating pin, implying that no sliding takes place. The difference in front tension measured in two tests gives an estimate of the friction. A drawback of this method is stochastic variations, which may cause large scatter, and the fact that steady-state conditions must be present when measuring. Weinmann and Kernovsky’s [21] design allows accurate measurement of both front tension and back tension, though this still does not allow direct measurement due to the contributions from bending and unbending friction. The above review of recent literature clearly shows that there is little information available on the friction and lubrication of the drawing process of sheet metals, thus further research is important and necessary. The objective of the research reported in this article is to make a comparative study of the friction behavior in sheet metal forming of steel, brass and aluminum alloys using both experimental and numerical approaches.

2 EXPERIMENTAL PROCEDURES The schematic view of the test device is shown in Fig 1. A test strip was held at one end in a grip supported by a load cell. The specimen (No. 3 in the Fig 1) is wrapped around a cylindrical fixed roll with diameter of 20 mm and loaded in a tensile testing machine ensuring contact over an angle of approximately 90°. The application of fixed pin allows setting up the rolls in four positions to utilize full circumference of the roll. The test was carried out using four rolls made of X165CrV12 tool steel with different average surface roughness qualities Ra = 2.5, 1.25, 0.63 and 0.32 µm measured parallel with the roll axis. The average roughness parameter Ra was selected in this research because it is widely known and universally used, though recent research [5] claims that Ra parameter lacks information on the wavelength and is not sensitive to small changes in profile compared with, for example, the root mean square deviation parameter (Rq). The tensile forces F1 and F2 were measured simultaneously during the test. A major advantage of this test apparatus is that strain does not have to

1 MATERIALS AND EQUIPMENT The introduced bending under tension test allows determining frictional resistances on the punch edge in

Table 1. Selected mechanical properties and roughness parameters of tested sheet metals Material Aluminium (AA5754 H24) Steel (DDQ*) Brass CuZn20 r

Sample orientation [°]

Yield point R0.2 [MPa]

Material constant C [MPa]

Strain hardening exponent, n

0 90 0 90 0 90

151 153 170 160 120 121

494 475 385 369 594 593

0.22 0.21 0.16 0.15 0.37 0.37

*DDQ - Deep Drawing Quality steel

Lemu, H.G. – Trzepieciński, T.

Roughness parameters Ra [μm] Rt [μm] 1.64 11.3 1.79 11.9 0.22 1.8 0.24 3.5 0.14 1.4 0.16 1.9

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 41-49

be measured to determine coefficient of friction. For some tests the effect of strain on coefficient of friction may be of interest. In other cases the use of an extensometer may not be reasonable or warranted. It is expected in most cases that the uniform deformation region is of interest only when measurement of specimen elongation is needed to calculate specimen strain away from the grip and pin regions. Constant stretching speed, i.e. the speed of the test machine head, is equal to 0.3 mm/s. While executing the BUT test, strain was measured and strain rate was calculated for the section of the test strip between the machine head and pin assuming that strain was uniform over this length of test material. With increasing upper grip displacement increases the sample deformation until fracture. Specimens were carefully prepared to assure constant width of 10 mm. To realize dry conditions both rolls and sheet specimens were degreased using acetone, and for lubricant conditions conventional machine oil was used. The friction coefficient value determined is averaged for the whole contact area. The BUT test allows determining the friction coefficient and also its changes during stretching process of the sample. These changes may be related with changes of surface topography and strain hardening phenomenon of the sample material. The occurrence of frictional resistance between the sheet metal and roll causes that F1 > F2. The deformation of the strip in the zone where the force F1 acts is not only due to the condition of sliding distance of the sheet metal around the pin, but also corresponds to limit deformation of the sheet metal [23]. Assuming that there is a constant friction coefficient µ in the contact region and the wrap angle γ (Fig. 2) is constant during the test according to the equilibrium of all forces acting on an elemental cut of the strip dγ, it can be shown that:

F + q µ wRd γ − ( F + dF ) = 0 , (1) qwRd γ − F sin

dγ dγ − ( F + dF ) sin = 0 , (2) 2 2

where q is unit normal contact pressure and w is the width of the strip. For a very small dγ one can assume that dγ dγ sin ≈ and dF << F. Thus, combining Eq. (1) 2 2 and Eq. (2) gives:

µ dγ =

dF . (3) F

Fig. 1. Schematic view of testing device; 1 – machine base, 2 – device frame, 3 – specimen, 4 and 5 – tension members, 6 – working roll, 7 - fixing pin, 8 and 9 – extensometer

Fig. 2. Forces acting on an elemental cut of the strip

Integrating Eq. (3) and taking into account γ = π the coefficient of friction is determined to be: 2 2 F  µ = ln  1  . (4) π  F2  The average contact stress, the unit contact pressure q in this case, is determined from the following equation:

F1 + F2 . (5) 2 wR

The values of tensile forces include the deformation resistance related with bending force of the sheet metal around the roll so that Eq. (4) does not include explicitly the bending force. In the other test variant [24] where the friction phenomenon on die edge is modeled, the bending force is determined by performing the test when the roll is unlocked. The extension of the sheet in the zone where F1 force

Numerical and Experimental Study of Frictional Behavior in Bending Under Tension Test

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 41-49

acts is an important parameter because it not only determines the amount of sliding of the sheet over the pins but it also represents the deformation limit of a specimen. 3 NUMERICAL MODELLING The simulation of the BUT test was conducted using MSC.Marc + MENTAT 2007r1 program. Both dry friction and oil lubrication were considered in the numerical simulation of friction test for sheet metal made of aluminum AA5754 H24. The roll was modeled as perfectly rigid and suitable boundary conditions that allow measuring the tensile forces were applied to both ends of the sample. The geometric model of the blank consists of 5472 isoparametric brick elements that are recommended by the program [25] to analyze contact conditions. The assumed strain formulation was applied to improve the bending characteristics of the elements. This can substantially improve the accuracy of the solution though the computational costs of assembling the stiffness matrix increase. An elasto-plastic material model approach was implemented. The plastic behavior of the metal was described by the von Mises yield criterion with isotropic work hardening. To describe contact conditions the Coulomb friction law was assumed as in Eq. (6):

ft = µ f n

 vr 2 arctan   π  RVCNST

  T , (6) 

where ft is tangential (friction) force, µ friction coefficient, fn normal force, ||νr|| relative sliding velocity, RVCNST value of the relative velocity below which sticking occurs and T tangential vector in the direction of the relative velocity. The value of RVCNST determines how closely the mathematical friction model represents the step function given as:

f t = µ f n sign ( vr ) . (7)

A very large value of RVCNST results in a reduced value of the effective friction. On the other hand, a very small value may result in poor convergence of contact algorithm. It is thus recommended that the value be 1 to 10% of a typical relative sliding velocity. The value of friction coefficient changes in accordance with the displacement of the upper grip of tensile machine. Simulations of friction tests were performed for roll with surface roughness value of Ra = 0.32 µm. Knowledge of the grip velocity allows 44

finding dependence of changes of friction coefficient versus time (t) in dry friction µd(t) and oil lubrication µo(t) given by: for t ≤ 0.69 : µ d ( t ) = 0.042 ln ( t ) + 0.005,

for t > 0.69 : µ d ( t ) = −3 ⋅10

−12 4

t + 4 ⋅10−9 t 3 − 2 ⋅10−6 t 2 + 0.0001t + 0.132 ,

for t ≤ 2.07 : µo ( t ) = 0.031 ln ( t ) − 0.011,

for t > 2.07 : µo ( t ) = 2 ⋅10−7 t 2 − 0.0001t + 0.129. To model the process of sample rupture the Cockroft-Latham damage criterion that was implemented into MSC.Marc [25] was used. In agreement with this criterion the moment of the fracture depends on energy accumulated by tensile stress only. The Cockroft-Latham indicator, Eq. (8), is a postprocessing value to indicate a possible damage area. σ max ∫ σ ε dt ≥ C , (8) where σ is the effective von Mises stress, σmax is the maximum principal stress, ε is the effective plastic strain rate and C is material constant threshold for damage. The critical value C has been defined as a workpiece material constant that does not depend on the working operation. The critical value is evaluated by a tensile test. If the critical value of the indicator was reached elements were deleted by the algorithm implemented in MSC.Marc. 4 RESULTS AND DISCUSSIONS 4.1 Influence of Surface Roughness of Rolls Plots of variations of friction coefficient obtained from tests on aluminum AA5754 H24 sheet metal as a function of selected surface roughness values are shown in Fig. 3. The plots reveal that friction coefficient values change depending on surface roughness of the rolls and there exist some significant differences of friction behavior based on the friction conditions. The case of dry friction (Fig. 3a) show that the friction coefficient at high roughness values (high Ra surface parameter) decreases as the sample is more deformed. This may be as a result of changes of sheet metal surface topography under the deformation process which causes the real contact area to increase simultaneously with the normal pressure. The real

Lemu, H.G. – Trzepieciński, T.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 41-49

contact area becomes also less than nominal area and depends on, for instance, roughness parameters of the sheet metal and the tools, inclination to strain hardening of roughness picks and the geometry of contact surface. This makes it difficult to generalize and interpret the obtained results of friction coefficient variation. Results from other researches [26] indicate that aluminum sheet metal made of AA1100 H14 with high tendency to seizing reveals a reduction of frictional resistance with the sheet metal deformation, while AA3104 sheet metal with low tendency to seizing shows increasing friction coefficient value under the influence of deformation. The reduction of friction with the strain in contact is apparently due to a decrease in contact area associated with roughening of the strip by plastic deformation. The results also indicate that theoretical predictions of the variation of the real area of contact with strain show excellent agreement with experiments using model asperities in rolling.

friction coefficient is stable during the friction test, i.e. it indicates a somewhat similar variation pattern. The lowest values of the friction coefficient of aluminum (Fig. 3) and brass (Fig. 4) sheet metal for roll characterized by surface roughness of 0.63 µm may be explained by surface topography. This is because sheet metals with high surface roughness create large amounts of oil pockets at mixed lubrication condition. The mixed lubrication regime is the intermediate zone between the boundary lubrication regime and the elasto-hydrodynamic lubrication regime, where the applied load is partly carried by the interacting asperities and the remaining part by the fluid film. In these conditions the suitable lubricant viscosity plays a key role [27]. Accordingly, the surface roughness of rolls applied in oil lubrication regime reduces the value of friction coefficient approximately by 25 to 40%.

Fig. 4. Variation of friction coefficient of CuZn20r brass sheet as a function of roll surface roughness for; a) dray friction, and b) lubricated conditions Fig. 3. Variation of friction coefficient of AA5754 H24 aluminum sheet as a function of roll surface roughness for; a) dry friction, and b) lubricated conditions

As the plot in Fig. 3a shows the result from roll with surface roughness value (Ra) of 0.32 µm, after some initial instability, is almost constant. For the rolls with surface roughness value of 0.63 and 1.25 µm in both friction conditions (Figs. 3a and b), the change of

The variation of friction coefficient value for brass sheet metal in dry friction condition is stable during the friction test (Fig. 4a). Brass in soft temper state hardens very strongly under the influence of deformation. Thus, a reduction of sample width caused by sample elongation is compensated by increased yield point of material. Application of lubricant between the contact surfaces (Fig. 4b) changes the

Numerical and Experimental Study of Frictional Behavior in Bending Under Tension Test

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 41-49

character of frictional resistance variation during the test. Even identical influence of surface roughness of the rolls on the friction coefficient value is observed. The lowest value of friction coefficient is observed for roll with surface roughness value (Ra) of 0.63 and the friction coefficient value increases for rolls with roughness value 0.32, 0.63 and 1.25 µm in sequence. This phenomenon is observed in both friction conditions – dry and lubricated. The degree of reduction measured at the necking stage varies from 40% for rolls with roughness value of 2.5 µm to above 50% for rolls with roughness value of 0.32 µm. The variation trend of the test results for steel DDQ sheet metal is similar to those received for the above discussed materials. The values of the friction coefficient increased in sequence for rolls Ra = 0.63, 0.32, 1.25 and 2.5 µm.

Fig. 5. Maximum contact pressure as necking occurs recorded for rolls with different roughness

4.2 Contact Pressure at Necking The level of the average contact pressure at necking has been studied for both dry and lubricated contact (Fig. 5). The plots do not show a typical trend or variation as a function of surface quality and friction condition. A general observation is, however, that the contact pressure tends to drop with increasing roughness value. Furthermore, using oil lubrication tends to raise the contact pressure for steel sheet at roughness value of Ra = 0.32 µm and 0.63 µm. This does not have to mean a proportional increase of frictional resistances because in the lubrication regime there occurs a higher sample elongation in F1 force acting zone and the width of contact area of roll and the sheet metal is reduced. Among the tested materials, the lowest contact pressure is registered for AA5754 H24 sheet metal and the DDQ (Deep Drawing Quality) steel sheet has the strongest hardening capacity. As depicted in Fig. 6, the value of the tensile forces in dry friction increases after sample yielding and then remains at a stable level. Furthermore, the relation between forces and the friction condition does not change. For materials with little hardening capacity follow fast increase of friction coefficient at relatively small pressure [1]. No significant increase in real contact area occurs with increasing normal contact pressure. In addition, the shear stress on contact surface does not increase and thus friction force remains constant. The application of a lubricant makes material flow easy from the bottom of the drawpiece to the wall. This leads to the fact that the forces increase uniformly during the whole friction contact process until the sample fractures. 46

Fig. 6. Values of forces during friction tests of AA5754 H24 sheet metal determined experimentally (heavy lines) and numerically (fine lines) in dry and lubricated friction conditions

4.3 Equivalent Total Plastic Strain Figs. 7 and 8 depict distribution of equivalent total plastic strain on samples of the same material for lubricated and dry friction conditions respectively. The equivalent strains are plotted as a function of different values of a total length strain (ε1) of the samples calculated for a 10 mm wide sample using Eq. (9).

εl = ∫ l0

dl , (9) l

where l0 = length of sample before deformation, and l = length of sample after deformation. In all the analyzed cases, a collapse of the free ends of the sample loaded by F1 force caused by braking resistances of outside surface that is in contact with roll was observed. This phenomenon corresponds to necking punch surface by the drawpiece particularly at high friction value.

Lemu, H.G. – Trzepieciński, T.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 41-49

From a technological point of view this phenomenon is advantageous because it increases the drawpiece ripping force. Comparing the lubricated and dry friction condition in Figs. 7a and 8a, there essentially exist differences in distribution of strains but very low. The small differences in equivalent plastic strain distribution result from small differences between the values of friction coefficients in dry and lubrication conditions. Lubricated surface contact results in more uniform strain distribution on the contact surface, but only in the range of small total strains. In case of large contact pressure, for instance, as in the case of Figs. 7c and 8c, the influence of strain distribution in contact zone is mainly dominated by geometry, and minor influence of lubrication conditions is observed. The most loaded section is the place where strip loaded by the F1 force leaves contact with the roll. Total strain of the simulated sample at the instant when fracture takes place is about 8% less than the value determined by experiment.

Moreover, in the numerical models, parameters result from real polycrystal structure, for instance impurity and structure defects could not directly be taken into account. 4.4 Distribution of Contact Friction and Effective Strain As depicted in Fig. 9, the distribution of contact friction stress on the sheet-roll contact surface is not uniform along the contact surface.

Fig. 7. Equivalent plastic strain distribution in lubrication conditions and under total length strain of; a) εl = 0.024, b) εl = 0.048, c) εl = 0.072 and d) after fracture

Fig. 9. Distribution of contact friction stress for different total length strain of the samples for; a) dry contact and, b) lubricated contact condition Fig. 8. Equivalent plastic strain distribution in dry friction conditions and under total length strain equals; a) εl = 0.024, b) εl = 0.048 and c) εl = 0.072

This difference is visible as decreasing simulated forces (Fig. 8) that produced strain localization leads to fracture of the sheet metal. The difference between experimental and numerical results may be due to the simplification as a result of the assumption that the material has isotropic mechanical properties.

The distribution of contact friction stress in dry and lubricated conditions is approximately uniform until about 60° of contact period. The distribution in this range shows that the values of contact friction stress for dry friction are about five times greater than those for lubricated contact condition. Local peaks of contact friction stress are also observed at the start of contact, in both analyzed friction conditions, for total

Numerical and Experimental Study of Frictional Behavior in Bending Under Tension Test

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 41-49

strains of sample equal to 0.044 and 0.066. This may be explained by different material flow characteristic at higher total strains of the sample. Analogous to the distribution of contact friction stress in the middle part of the sample section, the distribution of effective strain is not uniform along the contact surface. The highest gradient of effective strains exists near the entrance and exit of sheet from sheet-roll contact (Fig. 10). For the same value of total strain of the sample, increasing friction coefficient value causes the effective strain of the sample to increase. Furthermore, the existence of friction on one side of the sheet in BUT test causes local flexure of the sheet. The higher value of friction coefficient in case of dry friction conditions corresponds to an increase of flexure in the sample (Fig. 10b). This flexure influences the characteristic of the contact along the sheet-roll contact surface and is visible as a local decrease of contact normal stress before reaching the maximum value.

coefficient value to higher degree for higher roughness values. Use of tools with low surface roughness value to reduce the frictional resistance is unfounded because the increased real contact area increases the interatomic interaction of surfaces. This phenomenon increases frictional resistance. The characteristics of the changes depend on friction conditions, i.e., dry friction and lubricated. Application of the finite element method in this research allows a simulation of the material flow of the sheet metal by taking the complex friction phenomenon into account. It further allows a better understanding of the contact conditions that occur at the punch radii of sheet metal stamping processes. Moreover, on the basis of numerical results we can forecast the value and distribution of local deformations in real sheet metal forming operations. The results will assist future research into developing friction tests and the possibility of determining the friction coefficient values. 6 ACKNOWLEDGEMENT This research was realized with financial support provided by Island, Liechtenstein and Norway and was co-financed by European Economic Area and Norwegian Financial Mechanism under the Scholarship and Training Fund. The authors would like to appreciate this financial support. 7 REFERENCES

Fig. 10. Distribution of effective strain for sample total length strain equal to 0.066 for; a) lubrication and, b) dry friction conditions

5 CONCLUSIONS A study of friction behavior of sheet metal forming in bending under tension test is presented in this paper. The study focused on the influence of lubrication condition and the variations of friction behavior with surface roughness. It has been observed that the application of lubricant during tests of all of the sheet metal samples causes reduction of friction coefficient value. Furthermore, the results of this study indicate that lubricated contact condition reduces the friction 48

[1] Gierzyńska, M. (1983). Friction and lubrication in plastic forming of metalsi. WNT, Warszawa. (in Polish) [2] Matuszak, A. (2000). Factors influencing friction in steel sheet forming. Journal of Material Process Technology, vol. 106, p. 250-253, DOI:10.1016/S09240136(00)00625-7. [3] Volk, M., Nardin, B., Dolšak, B. (2011). Application of numerical simulations in deep-drawing process and holding system with segments’ inserts. Strojniški vestnik - Journal of Mechanical Engineering, vol. 57, no. 9, p. 697-703, DOI:10.5545/sv-jme.2010.258. [4] Sedlaček, M., Vilhena, L.M.S, Vižintin, J. (2011). Surface topography modeling for reduced friction. Strojniški vestnik - Journal of Mechanical Engineering, vol. 57, no. 9, p. 674-680, DOI:10.5545/svjme.2010.140. [5] Sedlaček, M., Podgornik, B., Vižintin, J. (2009). Influence of surface preparation on roughness parameters, friction and wear. Wear, vol. 266, p. 482487, DOI:10.1016/j.wear.2008.04.017. [6] Petek, A., Kuzman, A. (2012). Backward hole-flanging technology using an incremental approach. Strojniški

Lemu, H.G. – Trzepieciński, T.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 41-49

vestnik - Journal of Mechanical Engineering, vol. 58, no. 2, p. 73-80, DOI:10.5545/sv-jme.2011.194. [7] Hol, J., Cid Alfaro, M.V., de Rooij, M.B., Meinders, T. (2011). Advanced friction modeling for sheet metal forming. Wear, vol. 286-287, p. 66-78, DOI:10.1016/j. wear.2011.04.004. [8] Zdravecka, E., Suchanek, J., Lubiščak, J. Greškovič, F., Spišak, E. (2002). Examination of surface characteristics in tribological system by AFM. Tribologia, vol. 33, p. 1713-1722. [9] Duncan, J.L., Shabel, B.S., Filho, J.G. (1978). A tensile strip test for evaluating friction in sheet metal forming. SAE Technical Paper, no. 780391, doi: 10.4271/ 780391. [10] Wang, X.J., Duncan, J.L., Devenpeck, M.L. (1983). Punch friction tests for sheet metal forming. Journal of Applied Metalworking, vol. 3, p. 3-11, DOI:10.1007/ BF02833871. [11] Weinmann, K.J., Bhonsle, S.R., Gersten-berger, J. (1990). On the determination of the coefficient of friction and the friction factor by the strip–tension friction test. CIRP Annals, vol. 30, p. 263-266, DOI:10.1016/S0007-8506(07)61049-X. [12] Littlewood, M., Wallace, J.F. (1964). The effect of surface finish and lubrication on the friction variation involved in the sheet-metal-forming process. Sheet Metal Industries, vol. 41, p. 925-930. [13] He, D.N., Yin, X.F. Tao, H.Z. (2000). Research on evaluation method of friction and lubrication in deep drawing, Acta Metallurgica Sinica, vol. 13, p. 439-445. [14] Wihlborg, A, Gunnarsson, L. (2000). A frictional study of uncoated EBT steel sheets in a bending under tension friction test. Wear, vol. 237, p. 129-136, DOI:10.1016/ S0043-1648(99)00319-1. [15] Alinger, M.J., Van Tyne, C.J. (2001). Evolution of the tribological characteristics of several forming die materials. Journal Material Processing Technology, vol. 111, p. 20-24, DOI:10.1016/S0924-0136(01)00512-X. [16] Kim, H.Y., Hwang, B.C., Bae, W.B. (2002). An experimental study on forming characteristics of precoated sheet metals. Journal Material Processing Technology, vol. 120, p. 290-295, DOI:10.1016/S09240136(01)01088-3. [17] Garza, L.G., Van Tyne, C.J. (2007). Friction and formability of galvannealed interstitial free sheet steel.

Journal Material Processing Technology, vol. 187-188, p. 164-168, DOI:10.1016/j.jmatprotec.2006.11.062. [18] Moona, Y.H., Kima, D.W., Van Tyneb, C.J. (2008). Analytical model for prediction of sidewall curl during stretch-bend sheet metal forming. International Journal of Mechanical Sciences, vol. 50, p. 666-675, DOI:10.1016/j.ijmecsci.2008.01.003. [19] Hilditch, T.B., Matlock, D.K., Levy, B.S., Siekirk, J.F. (2005). Experimental evaluation of curl and tensile properties of advanced high strength sheet steels. SAE Transactions, Journal Materials and. Manufacturing, vol. 5, p. 457-466. [20] Hilditch, T.B., Speer, J.G., Matlock, D.K. (2007). Influence of low-strain deformation characteristics of high strength sheet steel on curl and springback in bend-under-tension tests. Journal Material Processing Technology, vol. 182, p. 84-94, DOI:10.1016/j. jmatprotec.2006.06.020. [21] Weinmann, K.J, Kernovsky, S.K. (1996). Friction studies in sheet metal forming based on a unique die shoulder force transducer for sheet metal forming research. CIRP Annals, vol. 15, p. 269-272, DOI:10.1016/S0007-8506(07)63061-3. [22] Stachowicz, F., Trzepieciński, T. (2004). ANN application for determination of frictional characteristics of brass sheet metal. Journal of Artificial Intelligence, vol. 1, p. 81-90. [23] Lovell, M.R., Deng, Z. (2002). Characterization of interfacial friction in coated sheet steels: influence of stamping process parameters and wear mechanisms. Tribology International, vol. 35, p. 85-95, DOI:10.1016/ S0301-679X(01)00097-4. [24] Han, S.S. (1997). The influence of tool geometry and friction behavior in sheet metal forming. Journal Material Processing Technology, vol. 63, p. 129-133, DOI:10.1016/S0924-0136(96)02612-X. [25] MSC.Marc Volume B: Element Library, MSC.Software Corporation 2007. [26] Saha, P.K., Wilson, W.R.D. (1994). Influence of plastic strain on friction in sheet metal forming. Wear, vol. 172, p. 167-173, DOI:10.1016/0043-1648(94)90284-4. [27] Lowell, M.R., Khonsari, M.M., Marangoni R.D. (1993). The response of balls undergoing oscillatory motion: crossing from boundary to mixed lubrication regimes. ASME Journal Tribology, vol. 115, p. 261266, DOI:10.1115/1.2921000.

Numerical and Experimental Study of Frictional Behavior in Bending Under Tension Test

Received for review: 2012-06-22 Received revised form: 2012-09-12 Accepted for publication: 2012-11-29

Characterizing Effective d31 Values for PZT from the Nonlinear Oscillations of Clamped-Clamped Micro- Resonators Dick, A.J. Andrew J. Dick

Rice University, Department of Mechanical Engineering and Materials Science, Nonlinear Phenomena Laboratory, USA In order to accurately predict the performance of micro-electromechanical systems which use piezoelectric material, precise knowledge of the piezoelectric coefficients is critical. Current material characterization methods rely on either simple structures restricted to small amplitude, linear oscillations or consider the piezoelectric material separate from the specific micro-scale device. A method is proposed for the characterization of the effective transverse piezoelectric coefficient d31 of lead zirconate titanate in a clamped-clamped micro-beam resonator experiencing nonlinear oscillations. Parameter trends identified by using a parametric identification scheme are analyzed and an approach is presented to calculate the linear piezoelectric coefficient. This method utilizes the relationship between a DC bias added to the excitation signal and the frequency shift experienced by the nonlinear response behavior. Through an additional numerical study, the sensitivity of the results to changes in the device length is identified and all data sets provide the same coefficient value when a length variation of less than 2% is allowed. Keywords: Piezoelectric material, micro-beam resonator, nonlinear oscillations

0 INTRODUCTION Piezoelectric material is attractive for the development of a wide range of micro-electromechanical systems (MEMS). With the ability to transform strain into an electric current through the direct piezoelectric effect and, by way of the converse piezoelectric effect, convert an applied electric field into stress, piezoelectric material is used to provide both sensing and actuation capabilities. Some examples of MEMS devices which utilize piezoelectric material that have recently been developed include contour-mode microresonators [1], shunt-type ohmic RF MEMS switches [2], MEMS generators [3], nano-robotics [4], and three-dimensional valveless micro-pumps [5]. In the design of piezoelectric MEMS devices, it is important to know the properties of the piezoelectric material in order to accurately predict actuation and sensing performance. One of the most important properties in many MEMS devices is the transverse piezoelectric coefficient d31. Due to the different methods employed in the fabrication of these devices, the effective properties of these materials may vary significantly. In order to address this issue, it is important to have methods to successfully characterize these materials. This need has led to the development of various methods for characterizing piezoelectric materials. These studies have utilized micro-scale cantilevered structures as well as different styles of micro-scale diaphragms and membranes in the characterization of thin film piezoelectric materials. Working with these structures, the properties of the piezoelectric materials were determined by analyzing 50

static deformations [6] to [8] and resonance frequency shifts [7] and [9]. Dynamic responses were also used to characterize piezoelectric materials, often at very low frequencies to avoid large amplitude oscillations associated with resonance and the potential nonlinear behavior which may result [9] to [12]. While these methods are effective for low amplitude oscillations, the use of linear modeling techniques, such as those employed by many finite element models, significantly limit the effective operation range of these methods. Due to the influence of scaling on these structures, their behavior has been found to become significantly nonlinear more readily than equivalent macro-scale structures [13] to [15]. When characterizing the effective properties of piezoelectric materials from the oscillations of a MEMS device, the linear range can be limiting and monitoring these “small” amplitude oscillations may require the use of high precision measurement equipment. When nonlinear properties are introduced, system dynamics can become complicated and specific nonlinear analysis techniques are required (e.g. [16]).

Fig. 1. Diagram of the side view of a clamped-clamped beam resonator

*Corr. Author’s Address: Rice University, Department of Mechanical Engineering and Materials Science, 6100 Main Street, Houston, Texas, USA, andrew.j.dick@rice.edu

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 50-55

For excitation magnitudes above the levels which produce linear behavior, micro-scale structures can exhibit nonlinear frequency-response behavior. Nonlinear Düffing-like hard-ening behavior has been observed in frequency-response data collected from clamped-clamped style beam resonators for harmonic excitation with constant amplitude and a range of bias voltage levels. A diagram of the device config-uration is presented in Fig. 1. The drive and sense electrodes are each one-quarter of the device length to provide the best electro-mechanical coupling [17]. The frequency-response data collected from these devices can be analyzed by using a parametric identification scheme devel-oped by the author [18]. Parametric identification techniques rely on an accurate model of the system under investigation in order to calculate specific system parameters (e.g. [19]). This parametric identification scheme is used to calculate the parameter values associated with a discretized system model from nonlinear frequencyresponse data. For the clamped-clamped beam-like multi-segment and multi-layered micro-structure considered, a nonlinear partial integro-differential beam equation is used. This equation, presented as Eq. (1), includes the standard terms of an EulerBernoulli beam model as well as a term for an applied axial force and an integral term to account for axial stretching which results from large displacements. The segmenting of the top platinum electrode layer is modeled as three beams in series with the subscript n for n = 1, 2, and 3. The three sections correspond to 0 < x < x1, x1 < x < x2, x2 < x < L. This allows for the model to accurately represent the decrease in the stiffness of the structure caused by the absence of the middle segment of the top platinum electrode and its subsequent effect on the characteristic frequencies and mode shapes.

ρ An wn ,tt + cn wn ,t + EI n wn ,xxxx − 3

−∑  12  m =1 

EAm L

∫xm−1 ( wm ,x ) dx  wn ,xx − P0 wn ,xx = M n ,xx . (1) xm

Subscripts following commas indicate partial derivatives. The parameters w, ρA, c, EI, EA, L, P0, and M corresponds to the transverse displacement, mass per length, viscous damping, flexural rigidity, axial stiffness, length, axial force, and applied moment, respectively. The applied moment is produced by the axial force from the actuated piezoelectric layer and the offset between the piezoelectric layer and the neutral axis of composite structure. The values of ρA, EI, and EA are averaged across the three or four layers in each of the sections [20].

In order to complete the model, four boundary conditions and eight compatibility conditions are defined. The boundary conditions correspond to a clamped-clamped configuration. The compatibility conditions ensure that position, slope, moment, and shear force are balanced at positions x1 and x2. By using linear mode shapes calculated with a linear form of Eq. (1), the boundary conditions, and the compatibility conditions, Eq. (1) is discretized with the Galerkin method to produce a single mode approximation, defined in Eq. (2). The use of the linear mode shapes is based on the assumption that the structure only exhibits weakly nonlinear behavior.

m q1 + c q1 + k q1 + α 3 q13 = F0 cos (ω t ) . (2)

The variables q1, m , c , k, a3, F0, ω, and t correspond to the first mode response, modal mass, modal damping, stiffness, nonlinear stiffness, excitation magnitude, excitation frequency, and time. By using the first order approximate analytical solution to Eq. (2) calculated by using the method of multiple scales [21], the equations for an analytical frequencyresponse curve are derived. The assumption of weakly nonlinear behavior is also required in order to apply the method of multiple scales to Eq. (2). By tuning the parameter values in order to match the analytical frequency-response curve with the experimental data, the values of the system parameters in Eq. (2) are identified. An additional step in this process is used to identify the value of the axial force in order to match the identified effective linear natural frequency with the frequency from an analytical model. Through an iterative process, the parameter values are identified to the desired level of precision. A complete description of the parametric identification process is presented in reference [18]. Table 1. Resonator dimensions Dimension Width Thickness, SiO2 Thickness, Bottom Pt Thickness, PZT Thickness, Top Pt

Value 20 µm 1.06 µm 135 nm 530 nm 200 nm

The remainder of the paper is organized in the following manner. The data analyzed in this study is discussed in section one. In section two, the method proposed for calculating the effective transverse piezoelectric coefficient is presented. Results obtained from the application of the proposed method are

Characterizing Effective d31 Values for PZT from the Nonlinear Oscillations of Clamped-Clamped Micro- Resonators

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 50-55

the sol-gel PZT, specifically the PZ52T48 used within the devices [22]. Here the method to characterize the properties of PZT in MEMS devices is presented. The method for determining the effective transverse piezoelectric coefficient is based upon the relationship between the additional DC bias voltage added to the harmonic input signal and resulting axial force produced in the structure. However, it is not possible to measure the axial force in the microresonator directly. This information is determined indirectly from the effective linear natural frequency which is calculated from the identified values of the linear stiffness k and modal mass m measured with the parametric identification scheme. In order to calculate a value for the axial force versus DC bias voltage, the frequency versus axial force and frequency versus DC bias voltage are first calculated. When examining the effective linear natural frequency versus DC bias data, the influence of the hysteretic properties of the piezoelectric material is clearly observed. This property can be seen in the representative plot of frequency versus DC bias for a 200 µm device in Fig. 3. The hysteretic characteristics of the frequency versus DC bias are addressed by fitting a linear approximation to the identified parameter values in a least-squares sense. In Fig. 3, the joined data points correspond to the identified effective linear natural frequency values for a range of DC bias voltage values. The linear approximation is represented in the figure by the dashed line. While the hysteretic properties of the response are not captured by the linear approximation, it does provide an effective representation of the general frequencyvoltage relationship, especially for decreasing DC bias voltage values.

presented in section three. Concluding remarks and comments on the direction of future work are gathered in the section four. 1 PIEZOELECTRIC MICRO-RESONATORS The devices examined within this study consists of a silicon dioxide base layer, a bottom platinum electrode layer, a layer of sol-gel lead zirconate titanate (PZT), and a segmented top platinum electrode layer [17]. The dimensions of the devices studied are listed in Table 1. Data collected from devices with lengths of 100, 200 and 400 µm is examined. In order to characterize the piezoelectric material in these devices, multiple data sets of frequencyresponse data are collected from piezoelectrically actuated clamped-clamped beam-style micro-scale devices. A sweep-sine signal with an added DC bias was applied to the drive electrodes and a laser Doppler vibrometer focused onto the center of the device was used to measure the device’s response at room temperature and pressure. Bias voltage levels ranged from 0 to 4 volts corresponding to electric field strength values of up to 7.55×106 volts per meter were applied. Representative data collected by using this method is presented in Fig. 2. The dashed lines indicate where the response of the resonator abruptly changed from high amplitude to low amplitude responses for increasing frequency sweeps due to the nonlinear stiffening property of the structure.

Fig. 2. Representative nonlinear frequency-response data for a range of bias voltage levels

2 CHARACTERIZATION METHOD

Fig. 3. Representative plot of linear natural frequency of a 200 mm piezoelectric micro-resonator versus DC bias (joined data points) and linear approximation (dashed line)

By analyzing the experimental data, the effective linear transverse piezoelectric coefficient is calculated for 52

Dick, A.J.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 50-55

In order to calculate the relationship between the axial force and the effective linear natural frequency, an Euler-Bernoulli beam model with an axial force term is used, as defined by Eq. (3).

ρ An wn ,tt + EI n wn ,xxxx − FΛ wn ,xx = 0. (3)

The tensile axial force induced by the piezoelectric layer is represented by FΛ. This force is defined by using the block force model in terms of the free-strain, material properties, and geometry of the layer of piezoelectric material by the formula presented in Eq. (4).

FΛ = − EPZT bPZT Λ13t PZT . (4)

The simplification used by the block force model results in an over estimation of the predicted force. The negative sign indicates that a positive bias voltage will result in a compressive axial force under the drive electrode. The other two segments of the microstructure are subjected to equal and opposite axial loading which results in an increase in the device’s fundamental frequency. The free-strain Λ13 is defined in terms of the transverse piezoelectric coefficient d31, the applied DC voltage VDC, and the thickness of the piezoelectric material tPZT.

Fundamental frequency values drop to zero when the device undergoes buckling. Post-buckling conditions are not considered in this study. The high accuracy linear approximation for the relationship between the fundamental frequency and the axial force (kHz/FΛ) and the linear approximation of the fundamental frequency with the bias voltage (kHz/VDC) are used to calculate a relationship between the axial force and the bias voltage (FΛ/VDC). This information is used with Eq. (6), which is derived from Eqs. (4) and (5), to calculate a value for the transverse piezoelectric coefficient. F 1 d31 = − Λ . (6) VDC EPZT bPZT Table 2. Fundamental frequency ranges Length [µm]

Min. [kHz]

Max. [kHz]

100 200 400

847 291 121

877 320 146

Table 3. Linear approximation details Length [µm] 100 200 400

Slope [kHz/mN] Intercept [kHz] 96.98 774.01 62.45 218.28 32.98 78.69

R2 0.99999 0.99989 0.99963

3 RESULTS

Fig. 4. Fundamental frequency for a range of axial force values with linear trendlines (thick) over ranges of relevant frequency values

Λ13 = d31 (VDC t PZT ) . (5)

The nonlinear relationship between the fundamental frequency of the micro-structure and the applied axial load calculated by using Eq. (3) and clamped-clamped boundary conditions is illustrated in Fig. 4 for three device lengths: 100 µm (solid), 200 µm (dashed), and 400 µm (dot-dashed).

By using the frequency-voltage slope values calculated from seven data sets, the frequency-force slope values listed in Table 3, and the material properties and geometry, transverse piezoelectric coefficient values are calculated. These values are listed in Table 4 along with the corresponding force-voltage information. By assuming each device is of nominal length, transverse piezoelectric coefficient values range from –119.13 to –141.52 pm/V. This corresponds to a mean value of –129.67 pm/V with a standard deviation of less than 6% of the mean. Based on the use of the block force model, these values represent an upper bounds for the transverse piezoelectric coefficient values. While the values presented in Table 4 do not suggest any trends associated with device length, the fabrication methods used generally results in undercutting which will cause the device length to deviate from the nominal value. Although the device length does not directly affect the d31 value in Eq. (6), it does affect the value of the kHz/FΛ slope. In order to study this influence, kHz/FΛ slope values are calculated for length variations of ±7.5% for each of the three nominal

Characterizing Effective d31 Values for PZT from the Nonlinear Oscillations of Clamped-Clamped Micro- Resonators

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 50-55

device lengths. The slope values of the linear approximations calculated over this range have R2 values that deviate from unity by less than one-tenth of a percent. This data is plotted in Fig. 5 along with a second order polynomial fit to the data. The R2 values for the three polynomial fits are within three-tenths of a percent of unity and provide a highly accurate representation of the relationship between the kHz/FΛ slope and the effective device length.

4 CONCLUDING REMARKS

Table 4. Calculated d31 values Nom. L [µm] 100 200 200 200 400 400 400

FΛ/VDC [µN/V] 70.76 59.56 63.21 68.06 62.08 63.91 66.26

devices used. However, the values calculated are significantly greater than those of other efforts [6] to [9] and [11]. In these studies, a Young’s modulus value of about 100 GPa was used for the PZT, which is significantly greater than the value used in this study. Due to the nature of Eq. (6), an increase in the value of the Young’s modulus would result in a decrease in the value of the calculated transverse piezoelectric coefficient.

d31 [pm/V] –141.52 –119.12 –126.42 –136.12 –124.16 –127.82 –132.52

In this study, the nonlinear oscillations of a clampedclamped beam piezoelectric micro-scale resonator have been analyzed to calculate effective transverse piezoelectric coefficient values. By using the shift in the effective natural frequency of the nonlinear oscillator along with the analytical relationship between the axial force and this frequency, the transverse piezoelectric coefficient values have been calculated. The influence of variations in the length has been studied and the value of d31 has been determined to be sensitive to length. When allowing for an average length variation of less than 2%, a value of d31 = –127.84 pm/V for the lead zirconate titanate material in the devices from which the seven data sets were collected. The large value is attributed to the smaller value of the Young’s modulus which had been identified for the devices and which has been used in this study. 5 ACKNOWLEDGEMENTS The author would like to thank Balakumar Balachandran, C. Daniel Mote, Jr. and Donald L. DeVoe for their helpful discussion. Brett Piekarski and ARL Adelphi are thanked for providing the piezoelectric resonators used in the study. 6 REFERENCES

Fig. 5. Calculated slope values (points) for a range of length values around the nominal values and a second order polynomial fit (curve) to the data

[1] Zuo, C., Sinha, N., Piazza, G. (2010). Very high frequency channel-select MEMS filters based on selfcoupled piezoelectric AlN contour-mode resonators. Sensors and Actuators A: Physical, vol. 160, no. 1-2, p. 132-140, DOI:10.1016/j.sna.2010.04.011. [2] Lee, H.-C., Park, J.-H., Park, Y.-H. (2007). Development of shunt type ohmic RF MEMS switches actuated by piezoelectric cantilever. Sensors and Actuators A: Physical, vol. 136, no. 1, p. 282-290, DOI:10.1016/j.sna.2006.10.050. [3] Lee, B.S., Lin, S.C., Wu, W.J., Wang, X.Y., Chang, P.Z., Lee, C.K. (2009). Piezoelectric MEMS generators fabricated with an aerosol deposition PZT thin film.

By using these relationships, it is possible to explore how variations in the length will affect the calculated values of the transverse piezoelectric coefficient. The d31 values for all seven of the data sets are equal to –127.84 pm/V when the average length variation of less than 2%. This reveals that the value of the transverse piezoelectric is sensitive to variations in length for the clamped-clamped style 54

Dick, A.J.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 50-55

Journal of Micromechanics and Microengineering, vol. 19, no. 6, art. no. 065014, DOI:10.1088/09601317/19/6/065014. [4] Škorc, G., Cas, J., Brezovnik, S., Šafaric, R. (2011). Position control with parameter adaptation for a nanorobotic cell. Strojniški vestnik - Journal of Mechanical Engineering, vol. 57, no. 4, p. 313-322, DOI:10.5545/ sv-jme.2009.017. [5] Cheng, Y.-L., Lin, J.-H. (2007). Manufacture of three-dimensional valveless micropump. Journal of Materials Processing Technology, vol. 192-193, p. 229236, DOI:10.1016/j.jmatprotec.2007.04.055. [6] Shepard, J.F., Moses, P.J., Trolier-McKinstry, S. (1998). The wafer flexure technique for the determination of the transverse piezoelectric coefficient (d31) of PZT thin films. Sensors and Actuators A: Physical, vol. 71, no. 1-2, p. 133-138, DOI:10.1016/S0924-4247(98)001617. [7] Guirardel, M., Bergaud, C., Cattan, E., Remiens, D., Belier, B., Petitgrand, S., Bosseboeuf, A. (2004). PZT polarization voltage effects on off-centered PZT patch actuating silicon membrane. Sensors and Actuators A: Physical, vol. 110, no. 3, p. 385-389, DOI:10.1016/j. sna.2003.08.016. [8] Wong, C.W., Jeon, Y., Barbastathis, G., Kim, S.-G. (2004). Analog piezoelectric-driven gratings with nanometer resolution. Journal of Microelectromechanical Systems, vol. 31, no. 6, p. 9981004, DOI:10.1109/JMEMS.2004.839592. [9] Muralt, P., Kholkin, A., Kohli, M., Maeder, T., Setter, N. (1995). Characterization of PZT thin films for micromotors. Microelectrical Engineering, vol. 29, no. 1-4, p. 67-70, DOI:10.1016/0167-9317(95)00116-6. [10] Verardi, P., Craciun, F., Dinescu, M. (1997). Characterization of PZT thin film transducers obtained by pulsed laser deposition. Proceedings of IEEE Ultrasonics Symposium, vol. 1, p. 569-572. [11] Kanno, I., Kotera, H., Wasa, K. (2003). Measurement of transverse piezoelectric properties of PZT thin films. Sensors and Actuators A: Physical, vol. 107, no. 1, p. 68-74, DOI:10.1016/S0924-4247(03)00234-6. [12] Huang, Z., Leighton, G., Wright, R., Duval, F., Chung. H. C., Kirby, P., Whatmore, R. W. (2007). Determination of piezoelectric coefficients and elastic constant of thin films by laser scanning vibrometry

techniques. Sensors and Actuators A: Physical, vol. 135, no. 2, p. 660-665, DOI:10.1016/j.sna.2006.10.002. [13] Balachandran, B., Li, H. (2003). Nonlinear phenomena in microelectromechanical resonators. Proceedings of the IUTAM Symposium on Chaotic Dynamics and Control of Systems and Processes in Mechanics, Rome. [14] Li, H., Preidikman, S., Balachandran, B., Mote, Jr., C.D. (2006). Nonlinear free and forced oscillations of piezoelectric microresonators. Journal of Micromechanics and Microengineering, vol. 16, no. 2, p. 356-367, DOI:10.1088/0960-1317/16/2/021. [15] Li, H., Balachandran, B. (2006). Buckling and free oscillations of composite microresonators. Journal of Microelectromechanical Systems, vol. 15, no. 1, p. 4251, DOI:10.1109/JMEMS.2005.863598. [16] Vesenjak, M., Ren, Z. (2003). Dynamic analysis of a road-restraint system’s deformation resulting from a vehicle impact. Strojniški vestnik – Journal of Mechanical Engineering, vol. 49, no. 12, p. 586-592. [17] DeVoe, D.L. (2001). Piezoelectric thin film micromachined beam resonators. Sensors and Actuators A: Physical, vol. 88, no. 3, p. 263-272, DOI:10.1016/ S0924-4247(00)00518-5. [18] Dick, A.J., Balachandran, B., DeVoe, D.L., Mote, Jr., C.D. (2006). Parametric identification of piezoelectric microscale resonators. Journal of Micromechanics and Microengineering, vol. 16, no. 8, p. 1593-1601, DOI:10.1088/0960-1317/16/8/021. [19] Jakšić, N., Boltežar, M. (2002). Parameter identification for single-degree-of-freedom dynamic systems. Strojniški vestnik – Journal of Mechanical Engineering, vol. 48, no. 6, p. 302-317. [20] Crawley, E.F., Anderson, E.H. (1990). Detailed models of piezoceramic actuation of beams. Journal of Intelligent Materials, Systems, and Structures, vol. 1, no. 1, p. 4-25, DOI:10.1177/1045389X9000100102. [21] Nayfeh, A.H., Balachandran, B. (2004). Applied Nonlinear Dynamics: Analytical, Computational, and Experimental Methods, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim. [22] Piekarski, B., DeVoe. D.L., Dubey, M, Kaul, R., Conrad, J. (2001). Surfaced micromachined piezoelectric resonant beam filters. Sensors and Actuators A: Physical, vol. 91, no. 3, p. 313-320, DOI:10.1016/S0924-4247(01)00601-X.

Characterizing Effective d31 Values for PZT from the Nonlinear Oscillations of Clamped-Clamped Micro- Resonators

Received for review: 2012-06-07 Received revised form: 2012-10-19 Accepted for publishing: 2012-11-21

Detection of Flooding and Drying inside a PEM Fuel Cell Stack Debenjak, A. – Gašperin, M. – Pregelj, B. – Atanasijević-Kunc, M. – Petrovčič, J. – Jovan, V. Andrej Debenjak1,* – Matej Gašperin1,2 – Boštjan Pregelj1,3 – Maja Atanasijević-Kunc4 – Janko Petrovčič1,3 – Vladimir Jovan1,3 1 Jožef Stefan Institute, Slovenia 2 University of West Bohemia, Faculty of Electrical Engineering/RICE, Plzeň, Czech Republic 3 Centre of Excellence for Low-carbon technologies – CO NOT, Slovenia 4 University of Ljubljana, Faculty of Electrical Engineering, Slovenia

Proton Exchange Membrane (PEM) fuel cells are currently seen as the most suitable choice for implementation into daily-use applications. However, the PEM technology does not yet fulfil all the necessary requirements that the mass-market demands and proper strides towards elimination of remaining issues have to be taken. Hence, in this paper, the focus is made on water management faults, i.e. flooding and drying. More precisely, it deals with detection of them with the use of Electrochemical Impedance Spectroscopy (EIS). The EIS was successfully applied as a diagnostic tool to a fuel cell stack consisted of 80 cells without usage of any special purpose measurement equipment, where, in addition to the stack current, only voltage of the complete stack was measured. The paper describes the modifications that were made on the EIS to make it capable of handling the diagnostics of fuel cell stacks. The results of the experimental study show that the approach is successful in detecting the flooding and drying faults and that for detection only excitation signals with frequencies between 30 and 300 Hz are required. Based on the experimental data and conclusions, a diagnostic decision algorithm is proposed. Keywords: PEM fuel cell system, electrochemical impedance spectroscopy, diagnostics, flooding, drying

0 INTRODUCTION Hydrogen fuel cells with Proton Exchange Membrane (PEM) [1] and [2] present a potential alternative to existing internal combustion engines [3]. PEM fuel cells are energy conversion devices that convert chemical energy of hydrogen fuel, which can be produced in different ways [4] and [5], directly into electricity with high efficiency [6]. In addition to the electricity as the main product, they also produce heat, water and no greenhouse gases. Further, PEM fuel cells operate at low operating temperatures, use oxygen from the air, have high specific power (kW/kg), and have short start-up and shut-down times. All this makes them suitable for various kinds of stationary and transportation applications. However, the successful introduction of the PEM fuel cell technology to everyday applications depends on the durability and reliability of this technology on one hand, and a reduction in the costs of production, operation, and maintenance on the other [7] and [8]. Research is done on various issues that affect PEM fuel cells performance [9] and [10]. One wide research field deals with material degradation, where the main role plays corrosion of the electrodes and degradation of the membranes [11] and [12]. Next, catalyst and membrane poisoning is an irreversible process influencing fuel cell life-time [13] and [14]. Another aspect is connected to temperature. It was shown that higher temperatures negatively affect the characteristics of the membranes [15]. On the other 56

hand, the PEM fuel cells are also sensitive to sub-zero temperatures [16]. Recent studies have shown that the durability and reliability of a PEM fuel cell system is vitally affected by water management faults [9] and [17]. A precise balance of liquid and vapour water inside each cell needs to be maintained to ensure the optimal performance of the stack. An inadequate water balance leads to two different conditions, which both have a direct effect on the performance of the stack. First, an insufficient removal of the produced water leads to flooding of the gas channels, and secondly, an excessive water removal or too dry inlet gases cause membrane drying. Furthermore, in case of severe and long-term drying, irreversible damage to the membrane may occur. In order to improve the reliability and durability the two faults have to be detected, classified, and finally, proper control action has to be taken to eliminate the faults and prevent the system from faulty operation and possible damage. Fuel cell flooding and membrane drying are faults that cannot be measured directly, therefore proper diagnostics method has to be applied. Electrochemical impedance spectroscopy (EIS) [18] is an effective method for diagnosing fuel cell flooding and drying [19] and [20]. In [21], the authors suggested parameter identification of proposed fuel cell model through the use of impedance measurements. The approach gains good identification results, but the identification works only with single cells. A stack similar to the one used in our study was tested in [22]. The authors used laboratory measuring equipment and reported they *Corr. Author’s Address: Jožef Stefan Institute, Jamova 39, Slovenia, andrej.debenjak@ijs.si

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 56-64

can detect drying, flooding and anode poisoning. Still, the used equipment is prohibitively expensive for industrial usage and in contrast to our work, voltage of all the cells in the stack have to be measured. Multi-sine test signals were also successfully applied to the fuel cells in order to perform the EIS [23]. The approach is useful, since it shortens the time needed for the measurements. However, the benefit comes with a price, i.e. more demanding computation and excitation equipment. This is why we used single-sine signals in spite of the multi-sine’s benefits. Two main advantages of the EIS method are its ability to distinguish between fuel cell flooding and drying, and non-invasive in-situ measurements. The EIS has been already proved as a successful tool for detecting flooding and drying in laboratory environment where only single fuel cells were tested, whereas stack level diagnostics has not been given much attention yet. In a recent work, the method was used on a laboratory stack consisting of 20 cells without using any dedicated EIS equipment [24]. However, the total equipment costs were still almost €20,000, which makes any proposed implementation unfeasible for commercial systems. This paper presents an attempt towards the use of the EIS as an on-line diagnostic tool on larger fuel cell systems without relying on special purpose measurement equipment, where only total stack impedance is measured. The purpose is to detect and classify the flooding and drying faults of commercial systems consisting of a few dozens of fuel cells. The main issue with such an approach is that the faults may occur only in a few cells, whereas only total impedance of the stack is measured, which makes the faults hard to detect. There were three main objectives within the study. The first was to build up a low cost measurement system in such a way that on the one hand, it will be sensitive enough to perform the EIS measurements, and will be costs-acceptable for implementation in commercial systems on the other. The second was to adapt the method's computation procedure to be effective in real-life measurements. And the third was to test and validate the method on a real-life system. 1 EXPERIMENTAL METHOD Electrochemical reactions involve many processes and parameters. In the PEM fuel cells, some of them are influenced by the hydration of membranes and presence of liquid water formed inside the cells. These influences can be detected by using EIS, which enables an insight into electrochemical reactions [18].

EIS basically measures the frequency response (impedance characteristic) of an electrochemical process and analyses it. The theoretical procedure for acquiring impedance at a specific frequency consists of three steps, namely exciting the process, measuring both excitation and response signals, and performing computation of the impedance. The process must be excited by a perturbation signal that consists of DC and AC (sinusoidal) components. The DC component determines the operational point of the process, and the AC component represents the perturbation signal. In case of the measurement in galvanostatic mode (i.e. applying current excitation signal) and assuming the system is linear near an operating point, the current excitation signal and the process' voltage response can be presented as functions of time:

i ( t ) = I dc + I 0 sin (ω0 t ) , (1)

u ( t ) = U dc + U 0 sin (ω0 t + ϕ ) , (2)

where t denotes time, ω0 the angular frequency, Idc and Udc the signal’s DC component, I0 and U0 the signal’s AC amplitudes, and φ the phase shift of the voltage response signal. For further computation the AC components of both signals are expressed as phasors at an angular frequency ω0. In this case, both signals can be written as: I = I 0 e jω0t , (3)

U = U 0 e j ( ω0t +ϕ ) . (4)

Finally, the impedance at an angular frequency ω0 can be calculated using Ohm’s law: U U e j ( ω0t +ϕ ) Z = = 0 jω0t = Z 0 e jΦ , (5) I I0e where Z0 is the impedance gain, and Φ is the impedance phase angle. By measuring the impedances at different frequencies of interest, the impedance characteristic is obtained and can further be used for analysis. 2 EXPERIMENTAL 2.1 Measurement Setup The measurement setup is schematically presented in Fig. 1. It consists of three main subgroups of equipment. The first is a fuel cell system, the second represents a load, and the third consists of sensors and data acquisition (DAQ) equipment.

Detection of Flooding and Drying inside a PEM Fuel Cell Stack

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 56-64

The experiments were performed on a commercial PEM fuel cell based power unit presented in Fig. 2. With a stack of 80 PEM fuel cells, the unit produces up to 8.5 kW of electric power. The unit’s detailed specifications are collected in Table 1. In addition to the stack, the unit consists of all the required subsystems for autonomous operation (including a hydrogen recirculation pump, an air delivery system with blower, a cooling system, and electronics). The important feature of the unit is that it does not include a humidifier for inlet gases. Since the aim is to measure the impedance characteristics of the unit in presence of flooding and drying inside the stack, this feature allows flooding and drying to be provoked by setting extreme conditions. Table 1. PEM fuel cell system specifications

+ –

ELECTRONIC LOAD

– + output

FUEL CELL SYSTEM

Value 200 cm2 80 8.5 kW 40 to 80 V 190 A 50%

reference – +

Parameter Surface Area No. of Cells Max Power Voltage Range Max Current Peak Net Efficiency

FUNCTION GENERATOR

CURRENT SENSOR VOLTAGE SENSOR

DAQ-BOARD

RELATIVE HUMIDITY AND TEMPERATURE SENSORS

Fig. 1. Measurement setup scheme

Fig. 2. PEM fuel cell system

An AMREL PLW9K-120-1000E electronic load (e-load) was used to simulate an ohmic load on the one hand, and to excite the power unit in the proper way on the other. The e-load can be externally controlled by a function generator. This allows setting a desired waveform of the e-load current that is drawn 58

from the power unit. For this purpose a Tektronix AFG 3101 function generator was used. In other words, this subsystem allows the desired waveform of a current to be applied to the fuel cell system by setting the function generator parameters (frequency, offset, amplitude). The final subgroup of the measurement setup consists of the sensors and the DAQ equipment. The components of this subgroup are further divided in two groups based on the fact if the component is to be used in a future diagnostic system, or it is not to be needed. The choice of the components in the first group was subjected to the desire of constructing a diagnostics system for commercial applications in the long run, based on the knowledge presented in this paper. Above all, components of such a system must be cost acceptable. Thus, in the first group are voltage and current sensors, which in future work will remain as they are. The voltage signals were measured using an in-house-developed voltage sensor comprising a galvanic isolator and an input bandpass filter with lower and upper cut-off frequencies at 3.7 Hz and 3.4 kHz, respectively. Due to its considerable influence, the frequency response of the voltage sensor is required (and it was recorded) for further compensation of the impedance results. The current signals were measured using a low-cost LEM LA 100-P Hall-Effect-based current transducer with a much higher upper cut-off frequency (i.e., 200 kHz) and, consequently, a negligible influence on a frequency response of current measurements. In the second group are sensors and DAQ equipment that will either not be needed or will be replaced with inhouse developed electronics in the future diagnostic system. These components are air temperature and air humidity sensors, a National Instruments USB6215 DAQ-board, and a PC with in-house designed LabView-based EIS measurement software. During the experiments four quantities (i.e. voltage, current, temperature and relative humidity) were measured with four different sensors. In Table 1, standard uncertainties and measuring ranges of all the sensors are presented. Table 2. Standard uncertainties and measuring range of the used sensors Sensor Voltage Current Temperature Relative Humidity

Standard Uncertainty 1.55 mV 0.11 A 0.06 °C 1.73 %RH

Debenjak, A. – Gašperin, M. – Pregelj, B. – Atanasijević-Kunc, M. – Petrovčič, J. – Jovan, V.

Range ±10 V ±150 A – 40 to 180°C 0 to 100 %RH

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 56-64

Measurement of air temperature and relative humidity were performed only to give a rough estimation of environmental parameters. The uncertainty of these two measurements does not affect the measurement of impedance, which is the main objective of this study. Since the air temperature and relative humidity were not under control (as this is a real-life situation), their estimated extended uncertainties are 1.5 °C and 5%RH, respectively. Because of frequency analyses, the voltage and current sensors measurement uncertainties negligibly affect final impedance result. The uncertainty of the impedance is therefore mainly affected by the changes in stack itself, which is taken into account in the decision algorithm. 2.2 Measurement Parameters Measurement parameters that have to be set in the right way are the operating point, AC amplitude, excitation frequencies, and sampling frequencies. All the measurements were performed at the same operating point at 40 A DC (Idc), since this is a working point where the fuel cell system operates in the area of its highest efficiency. The amplitude of the sinusoidal current perturbation signal was set to 1 A (I0). In this way, it represents approximately 2% of the DC current and is therefore small enough not to cause problems due to the non-linear nature of the real system. Due to the limitations set by our voltage sensor, the measurements were performed in the frequency range from 1 Hz to 1 kHz. The following frequencies of the perturbation signal were used: 1, 2, 5, 10, 12, 15, 20, 30, 40, 50, 100, 200, 300, 500, and 1000 Hz. The frequencies were chosen in such a way that the resulting points in the Nyquist diagram are evenly distributed and clearly describe the system’s impedance characteristic. At excitation frequencies up to 50 Hz the sampling frequency was 600 Hz, while at higher excitation frequencies it was set to 6 kHz. 2.3 Experiments Three sets of measurements were conducted at different environment conditions, i.e. sets in normal, dry, and moist environments. To perform appropriate experiments, imitations of dry and humid environments were prepared in order to provoke fuel cell system drying and flooding. The measurements under normal operating conditions were conducted in a normal (non-modified) environment. At an air temperature of 25±1.5 °C and a relative humidity of 30±5%RH, the values are within

the specified operating environment parameters provided by the manufacturer. In order to stimulate drying and flooding inside the fuel cells two different setups for providing inlet air were constructed. A dry environment was simulated by feeding the system with air from the pneumatic compressor, where air is dehumidified after compressed. An air pipe with a diameter wider than the inlet air filter diameter was used to lead the air to the inlet filter. In this way, dry air was blown over the inlet air filter and entered in the system, while the environmental air was blocked. The air parameters were a temperature of 25±1.5 °C and a relative humidity of 4±5%RH. The moist environment was simulated by using hot water. The moist air was captured right above the water level, where its relative humidity was very high and was led to the inlet air filter via the air pipe. The air parameters were a temperature of 55±1.5 °C and a relative humidity of 98±5%RH. 3 RESULTS 3.1 Computation of the Impedance The real fuel cell system operation is subjected to influences from the electrical subsystems and the environment, so the measurements are corrupted by a measurement noise. Furthermore, the real system is not linear, as assumed in the theoretical EIS background given in Section 1. Therefore, the acquired signals in reality do not have a pure sinusoidal waveform. Consequently, the impedance computation cannot be made by the straightforward application of Eqs. (1) through (5). Instead, the appropriate signal-processing techniques must be employed. The calculations of the impedance were carried out in the following procedure for each measurement at each excitation frequency: • applying the Hamming window to both the current and voltage datasets [25], • calculating the fast Fourier transform (FFT) of both datasets [25], • extracting the FFT values at the excitation frequency, • calculating the impedance by dividing the extracted FFT voltage by FFT current values, • compensating the result for the voltage sensor’s influence in order to obtain the final impedance.

Detection of Flooding and Drying inside a PEM Fuel Cell Stack

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 56-64

3.2 Impedance Characteristics As pointed out in Section 2.3 three different sets of measurements were conducted. With the use of the impedance computation, these sets yielded three impedance characteristics of the fuel cell system in case of normal operating conditions, and in presence of flooding and drying. The characteristics are presented in the form of Nyquist plots in Fig. 3 (normal operating conditions – black dashed line, drying – red dotted line, and flooding – blue dash-dot line). The points of these lines are calculated as the average of 15 consecutive measurements of the impedance (the three lines are only presented to better visualize the characteristics). Those impedance measurements are in Fig. 3 presented as crosses – normal, dots – drying, and circles – flooding. The results in Fig. 3 affirm the possibility to perform EIS diagnostics of a larger PEM fuel cell system. The impedance characteristics of the system in case of different operating conditions clearly differentiate from each other, which makes the diagnostics possible. However, it is clear that the results in low frequency region below 20 Hz overlap and therefore the diagnostics cannot be performed in this frequency region. This observation can be addressed to low-frequency disturbances (e.g., jumping of the voltage because of water collecting on the cathode and the voltage gain during a hydrogen purge), and not strictly stationary nature of the system. 0

With regard to the results, the appropriate frequency range for diagnostics is from 20 to 300 Hz. The results indicate that at lower frequencies up to 50 Hz only the real part of the impedance can be used for diagnostics because the imaginary part does not change much. In contrary, at higher frequencies above 100 Hz both parts can be used for diagnostics, as they both change substantially. At frequencies up to 50 Hz a decrease in the real part value indicates flooding, while an increase indicates drying. At higher frequencies above 100 Hz a decrease in the real part value together with an increase in the imaginary part value implies flooding, whereas vice versa situation occurs in the case of drying. 3.3 Decision Algorithm The results present a solid base for constructing a decision algorithm for diagnostics. The main goal of such algorithm is to classify input data (the impedance measurements) in to three classes – namely normal, flooding, and drying. As pointed out in section 3.2, diagnostics for case-study fuel cell system is possible in frequency area between 20 and 300 Hz. So, the first step in constructing the algorithm is the selection of appropriate frequencies at which measurements will be performed. From the algorithm complexity and data acquisition duration point of view, a lower number of chosen frequencies is preferred. On the other hand,

1 kHz

100 Hz 10 Hz −0.05

ℑ(Z) [Ω]

1 Hz

−0.1

normal drying flooding −0.15 0.05

0.1

0.15 ℜ(Z) [Ω]

0.2

0.25

Fig. 3. Nyquist plots of the system impedance at normal operation, in presence of flooding, and in presence of drying

Debenjak, A. – Gašperin, M. – Pregelj, B. – Atanasijević-Kunc, M. – Petrovčič, J. – Jovan, V.

0.3

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 56-64

when reliability and robustness is under consideration, more input information normally results in a more reliable and robust algorithm. Considerring both sides, three frequencies were selected, i.e. 30, 100, and 300 Hz. These frequencies were chosen because they cover the entire interesting frequency range and because they are evenly spread across the frequency domain in terms of logarithmic scale.The three chosen impedances consist of six numerical values, but because the real part of the impedance at 30 Hz does not provide any additional information, only the remaining five are selected as algorithm’s input numerical features. Therefore, in total there are three real part values and two imaginary ones. For the decision algorithm, borders that separate the three classes must be constructed. For this purpose the following procedure is proposed, where the borders are derived from the performed measurement based on their mean and standard deviation values. In general, at each frequency two borders (i.e. normalflooding and normal-drying) have to be computed separately for the real and imaginary parts. Therefore, four border values are computed in total for each frequency. However, only two border values are computed in the case of 30 Hz, since only real part of the impedance is considered relevant. The borders at each frequency are computed based on the measurement data in such a way that the following equality equations hold true:

{ }

ℜ {Z no } − k1σ ℜ ,no = ℜ Z fl + k1σ ℜ , fl , (6)

ℜ {Z no } + k2σ ℜ ,no = ℜ {Z dr } − k2σ ℜ ,dr , (7)

ℑ{Z no } + k3σ ℑ ,no = ℑ Z fl − k3σ ℑ , fl , (8)

ℑ{Z no } − k4σ ℑ ,no = ℑ{Z dr } + k4σ ℑ ,dr . (9)

{ }

In Eqs. (6) through (9), ℜ {Z no } , ℜ Z fl and ℜ {Z dr } represent the mean value of the impedance real parts. Similarly, ℑ{Z no } , ℑ Z fl and ℑ{Z dr } are the mean value of the imaginary parts. σ ℜ,no , σ ℜ, fl and σ ℜ,dr are standard deviations of the real parts, σ ℑ,no , σ ℑ, fl and σ ℑ,dr are standard deviations of the imaginary parts, and k1, …, k4 are search variables. From Eqs. (6) through (9) all four border equations are derived:

{ }

Z ℜ ,no − fl = ℜ {Z no } −

{ }

ℜ {Z no } − ℜ Z fl

σ ℜ ,no + σ ℜ , fl

σ ℜ ,no , (10)

Z ℜ ,no − dr = ℜ {Z no } +

Z ℑ ,no − fl = ℑ{Z no } +

Z ℑ ,no − dr = ℑ{Z no } −

ℜ {Z no } − ℜ {Z dr }

σ ℜ ,no + σ ℜ ,dr

{ }

ℑ{Z no } − ℑ Z fl

σ ℑ ,no + σ ℑ , fl ℑ{Z no } − ℑ{Z dr }

σ ℑ ,no + σ ℑ ,dr

σ ℜ ,no , (11)

σ ℑ ,no , (12)

σ ℑ ,no . (13)

Table 3 and Fig. 4 illustrate the proposed computation procedure, where the Table 3 outlines the numerical values of the computed borders and Fig. 4 graphically presents them. In addition, 2σ areas (95% confidence) of the measurements from which the borders were derived are presented in the figure. Once the borders are computed, the decision algorithm is straight forward. The algorithm simply compares the input numerical features with the belonging border values and decides to which class the input feature belongs to. In other words, the algorithm tags each input feature with normal, flooding or drying tag. Since there are five input features and they can be in general tagged with different tags, the algorithm has to merge the intermediate result into one final. Therefore, the algorithm defines the final result by choosing the one tag, which is tagged to the majority of the features (e.g. three features are tagged with normal and the other two with drying, the final result is normal). Table 3. Border values computed based on the presented measurements Border Value [mΩ] normal-flooding normal-drying

f [Hz]

Part

ℜ

78.24

86.43

ℜ

51.15

56.61

ℑ

-24.78

-26.78

ℜ

44.72

47.38

ℑ

-12.06

-13.79

100

300

4 DISCUSSION The diagnostics of a commercial fuel cell system consisting of a few dozen cells, where only the impedance of the entire stack is measured, is a challenging task. Since not all the cells inside the

Detection of Flooding and Drying inside a PEM Fuel Cell Stack

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 56-64

ℑ(Z) [mΩ]

30 Hz

300 Hz

100 Hz −24 −12

−54

−58

−14

−28 75

ℜ(Z) [mΩ] normal

ℜ(Z) [mΩ] flooding

drying

48 ℜ(Z) [mΩ]

ℜ border

ℑ border

Fig. 4. Graphically presented borders

stack get flooded or dried, the total impedance of the system is not affected as much as in case of a shortstack system, and thus the change is hard to detect. In contrast to short-stack measurements there is yet another important difference, which is non-negligible influences of the system’s electrical subsystems. In other words, these influences can be neglected in case of short-stack measurements, but with commercial systems, these influences significantly affect the measurement data, so steps to exclude them have to be taken. The successfully performed measurements and results prove a few important things. Firstly, the used low-cost voltage and current sensors are appropriate for such measurements. Consequently, in future work these sensors will be used for building up a new diagnostic system. Secondly, the signal processing, impedance computation, and sensor compensation are done in the right way. Finally, the diagnostics can be carried out successfully, since the impedance characteristic of the system in the presence of different faults differentiate from each other. The conclusion is that the EIS measurements and diagnostics can also be performed on a larger commercially available system without using special EIS measurement equipment. The main advantage of the proposed border computation and decision algorithm is simplicity, which is an important feature when the algorithm has to work with limited computation resources (i.e. embedded systems) and in real time. However, to increase the robustness of the algorithm some additional decision rules can be added, especially when the real and imaginary value of the impedance at the same frequency are labelled with different tags. 62

5 CONCLUSION The work presented shows that the EIS based diagnostics of flooding and drying inside a larger commercially available PEM fuel cell system is possible. It also proves the suitability of the chosen voltage and current sensors, and the effectiveness of the entire proposed computation of the impedance. The next stage of work will deal with the implementation of the EIS into complete diagnostic system, which will be appropriate for implementation into commercial systems as on-board diagnostic system. 6 ACKNOWLEDGEMENT The authors thank the Centre of Excellence for Low-carbon Technologies – CO NOT, financed by the Slovenian Ministry of Higher Education, Science and Technology and co-financed by the European Regional Development Fund, Slovenian Research Agency (P2–0001) and the project EXLIZ – CZ.1.07/2.3.00/30.0013, which is co-financed by the European Social Fund and the state budget of the Czech Republic for their financial support. 7 NOMENCLATURE Symbols i [A] I [A] I0 [A] Idc [A] j k1,..., k4 t [s] u [V] U [V] U0 [V]

current current phasor current amplitude current DC component imaginary unit search variables time voltage voltage phasor voltage amplitude

Debenjak, A. – Gašperin, M. – Pregelj, B. – Atanasijević-Kunc, M. – Petrovčič, J. – Jovan, V.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 56-64

Udc [V] Z [Ω] Z0 [Ω] ℜ (Z) [Ω] ℑ (Z) [Ω] σ [Ω] Φ [rad] φ [rad] ω0 [rad/s]

voltage DC component impedance impedance amplitude real part of impedance imaginary part of impedance standard deviation impedance phase shift phase shift angular frequency

Subscripts no fl dr no-fl no-dr

normal flooding drying border between normal and flooding border between normal and drying 8 REFERENCES

[1] Barbir, F. (2005). PEM Fuel Cells: Theory and Practice. Elsevier, Amsterdam. [2] Srinivasan, S. (2006). Fuel Cells - From Fundamentals to Applications. Springer Science-Business Media LLC, New York. [3] McGuiness, P. (2008). Fuelling the Car of the Future. Strojniški vestnik – Journal of Mechanical Engineering, vol. 54, no. 5, p. 356-363. [4] Lotrič, A., Sekavčnik, M., Kunze, C., Spliethoff, H. (2011). Simulation of water-gas shift membrane reactor for integrated gasification combined cycle plant with CO2 capture. Strojniški vestnik - Journal of Mechanical Engineering, vol. 57, no. 12, p. 911-926, DOI:10.5545/ sv-jme.2011.100. [5] Pirc, A., Sekavčnik, M., Mori, M. (2012). Universal model of a biomass gasifier for different syngas compositions. Strojniški vestnik - Journal of Mechanical Engineering, vol. 58, no. 5, p. 294-299, DOI:10.5545/sv-jme.2011.101. [6] Jovan, V., Perne, M., Petrovčič, J. (2010). An assessment of the energetic flows in a commercial PEM fuel-cell system. Energy Conversion and Management, vol. 51, no. 12, p. 2467-2472, DOI:10.1016/j. enconman.2010.04.014. [7] Wu, J., Yuan, X.Z., Martin, J.J., Wang, H., Zhang, J., Shen, J., Wu, S., Merida, W. (2008). A review of PEM fuel cell durability: Degradation mechanisms and mitigation strategies. Journal of Power Sources, vol. 184, no. 1, p. 104-119, DOI:10.1016/j. jpowsour.2008.06.006. [8] Gerbec, M., Jovan, V., Petrovčič, J. (2008). Operational and safety analyses of a commercial PEMFC system. International Journal of Hydrogen Energy, vol. 33, no. 15, p. 4147-4160, DOI:10.1016/j.ijhydene.2008.04.063. [9] Schmittinger, W., Vahidi, A. (2008). A review of the main parameters influencing long-term performance and durability of PEM fuel cells. Journal of Power

Sources, vol. 180, no. 1, p. 1-14, DOI:10.1016/j. jpowsour.2008.01.070. [10] Pregelj, B., Vrečko, D., Jovan, V. (2011). Improving the operation of a fuel-cell power unit with supervision control – A simulation study. Journal of Power Sources, vol. 196, no. 22, p. 9419-9428, DOI:10.1016/j. jpowsour.2011.06.077. [11] Borup, R.L., Davey, J.R., Garzon, F.H., Wood, D.L., Inbody, M.A. (2006). PEM fuel cell electrocatalyst durability measurements. Journal of Power Sources, vol. 163, no. 1 (Spec. iss.), p. 76-81, DOI:10.1016/j. jpowsour.2006.03.009. [12] Borup, R., Meyers, J., Pivovar, B., Kim, Y.S., Mukundan, R., Garland, N., Myers, D., Wilson, M., Garzon, F., Wood, D., Zelenay, P., More, K., Stroh, K., Zawodzinski, T., Boncella, J., McGrath, J.E., Inaba, M., Miyatake, K., Hori, M., Ota, K., Ogumi, Z., Miyata, S., Nishikata, A., Siroma, Z., Uchimoto, Y., Yasuda, K., Kimijima, K.-I., Iwashita, N. (2007). Scientific aspects of polymer electrolyte fuel cell durability and degradation. Chemical Reviews, vol. 107, no. 10, p. 3904-3951, DOI:10.1021/cr050182l. [13] Farrell, C., Gardner, C., Ternan, M. (2007). Experimental and modelling studies of CO poisoning in PEM fuel cells. Journal of Power Sources, vol. 171, no. 2, p. 282-293, DOI:10.1016/j.jpowsour.2007.07.006. [14] Cheng, X., Shi, Z., Glass, N., Zhang, L., Zhang, J., Song, D., Liu, Z.-S., Wang, H., Shen, J. (2007). A review of PEM hydrogen fuel cell contamination: Impacts, mechanisms, and mitigation. Journal of Power Sources, vol. 165, no. 2, p. 739-756, DOI:10.1016/j. jpowsour.2006.12.012. [15] Song, Y., Xu, H., Wei, Y., Kunz, H.R., Bonville, L.J., Fenton, J.M. (2006). Dependence of high-temperature PEM fuel cell performance on Nafion content. Journal of Power Sources. vol. 154, no. 1, p. 138-144, DOI:10.1016/j.jpowsour.2005.04.001. [16] Yan, Q., Toghiani, H., Lee, Y.-W., Liang, K., Causey, H. (2006). Effect of sub-freezing temperatures on a PEM fuel cell performance, startup and fuel cell components. Journal of Power Sources, vol. 160, no. 2, p. 12421250, DOI:10.1016/j.jpowsour.2006.02.075. [17] Li, H., Tang, Y., Wang, Z., Shi, Z., Wu, S., Song, D., Zhang, J., Fatih, K., Zhang, J., Wang, H., Liu, Z., Abouatallah, R., Mazza, A. (2008). A review of water flooding issues in the proton exchange membrane fuel cell. Journal of Power Sources, vol. 178, no. 1, p. 103117, DOI:10.1016/j.jpowsour.2007.12.068. [18] Yuan, X.Z., Song, C., Wang, H., Zhang, J. (2010). Electrochemical Impedance Spectroscopy in PEM Fuel Cells, Fundamentals and Applications, Springer, London, DOI:10.1007/978-1-84882-846-9. [19] Yuan, X., Wang, H., Sun, J.C., Zhang, J. (2007). AC impedance technique in PEM fuel cell diagnosis – A review. International Journal of Hydrogen Energy, vol. 32, no. 17, p. 4365-4380, DOI:10.1016/j. ijhydene.2007.05.036.

Detection of Flooding and Drying inside a PEM Fuel Cell Stack

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 56-64

[20] Miller, M., Bazylak, A. (2011). A review of polymer electrolyte membrane fuel cell stack testing. Journal of Power Sources, vol. 196, no. 2, p. 601-613, DOI:10.1016/j.jpowsour.2010.07.072. [21] Fouquet, N., Doulet, C., Nouillant, C., Dauphin-Tanguy, G., Ould-Bouamama, B. (2006). Model based PEM fuel cell state-of-health monitoring via ac impedance measurements. Journal of Power Sources, vol. 159, no. 2, p. 905-913, DOI:10.1016/j.jpowsour.2005.11.035. [22] Le Canut, J.-M., Abouatallah, R.M., Harrington, D.A. (2006). Detection of Membrane Drying, Fuel Cell Flooding, and Anode Catalyst Poisoning on PEMFC Stacks by Electrochemical Impedance Spectroscopy.

Journal of The Electrochemical Society, vol. 153, no. 5, p. A857-A864, DOI:10.1149/1.2179200. [23] Brunetto, C., Moschetto, A., Tina, G. (2009). PEM fuel cell testing by electrochemical impedance spectroscopy. Electric Power Systems Research, vol. 79, no. 1, p. 1726, DOI:10.1016/j.epsr.2008.05.012. [24] Wasterlain, S., Candusso, D., Harel, F., Hissel, D., François, X. (2011). Development of new test instruments and protocols for the diagnostic of fuel cell stacks. Journal of Power Sources. vol. 196, no. 12, pp. 5325-5333, DOI:10.1016/j.jpowsour.2010.08.029. [25] Oppenheim, A., Schafer, R. (1998). Discrete-Time Signal Processing, Prentice Hall, New Jersey.

Debenjak, A. – Gašperin, M. – Pregelj, B. – Atanasijević-Kunc, M. – Petrovčič, J. – Jovan, V.

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 65-67 List of reviewers

List of reviewers who reviewed manuscripts in 2012 Abellan-Nebot Jose Vincente, Spain Abmrožič Vanja, Slovenia Adin Hamit, Turkey Ahmadi Iman, Iran Ambrož Miha, Slovenia Baer Sebastian, Germany Baldan Ahmed, Turkey Balič Jože, Slovenia Baragetti Sergio, Italy Barbieri Renato, Brazil Bartušek Karel, Czech Republic Beg Darko, Slovenia Beji Lofti, France Benz Michaela, Germany Bergada Josep M., Spain Bergant Anton, Slovenia Bergant Zoran, Slovenia Berlec Tomaž, Slovenia Bilik Igal, USA Birolini Alessandro, Switzerland Bobovnik Gregor, Slovenia Boltežar Miha, Slovenia Bombač Andrej, Slovenia Borkowski Przemyslaw, Poland Bošnjak Srdjan, Serbia Boy J. J., France Brajlih Tomaž, Slovenia Brezovnik Simon, Slovenia Broek Johan, The Netherlands Bruzzone Agostino, Italy Buchmeister Borut, Slovenia Calise Francesco, Italy Carlsson Bo, Sweden Casavola Alessandro, Italy Celent Luka, Croatia Chandrashekhara K., USA Chazal Claude, France Cheng Gang, China Cortés Pablo, Spain Courbon Cedric, France Croccolo Dario, Italy Četina Matjaž, Slovenia Čudina Mirko, Slovenia

Čuš Franci, Slovenia D`Urso Gianluca, Italy da Silva A.A.M., Spain Daneshmand Saeed, Canada Davim J. Paulo, Portugal Delgosha Olivier Coutier, France Demšar Ivan, Slovenia Depcik Christoper, USA DeSouza Guilherme, USA Devadula Sivasrinivasu, Sweden Devaraju Ayyannan, India Diaz Rafael, USA Dick Andrew J., USA Dikici Burak, Turkey Doležel Ivo, Chech Republic Dolinšek Slavko, Slovenia Donevski Božin, Macedonia Dragan Dejan, Slovenia Du Sha, China Dular Matevž, Slovenia Dumas Claire, France Emri Igor, Slovenia Erdönmez Cengiz, Turkey Essert Mario, Croatia Esterline Albert C., USA Fajdiga Matija, Slovenia Fetvaci Cuneyt, Turkey Ficko Mirko, Slovenia Filipič Bogdan, Slovenia Flajs Rado, Slovenia Flašker Jože, Slovenia Flores Paulo, Portugal Fung Rong-Fong, PR China Fuschi Paolo, Italy Gambarotta Agostino, Italy Gantar Tine, Slovenia Gatti Gianluca, Italy Geasin Savio S., India Ghani Jahrat Bt. A., Malaysa Ghoreishy Mir Hamid Reza, Iran Goettlich Emil, Austria Gotlih Karl, Slovenia Greiner David, Spain 65

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 65-67

Grimberg Raimond, Romania Grum Janez, Slovenia Guleren Melih K., Turkey Gusel Leo, Slovenia Halilovič Miroslav, Slovenia Hamilton Benjamin Carter, USA Hardell Jens, Sweden Harl Boštjan, Slovenia Herakovič Niko, Slovenia Hočevar Marko, Slovenia Hoecherl Johan, Germany Hoes Csaba, Hungary Horvat Darja, Slovenia Horvath Imre, The Netherlands Ibaraki Soichi, Japan Ihalainen Petri, Finland Iljaž Jurij, Slovenia Jankowski Alan, USA Jantunen Erkki, Finland Jezeršek Matija, Slovenia Johnson Eric, USA Jošt Dragica, Slovenia Kabele Karel, Czech Republic Kaplunov Julius, UK Kariž Zoran, Slovenia Kartnig George, Austria Kasper Roland, Germany Katrašnik Tomaž, Slovenia Kegl Marko, Slovenia Khader Iyas, Germany Klančar Gregor, Slovenia Klemenc Jernej, Slovenia Kljajin Milan, Croatia Klobčar Damjan, Slovenia Kokalj Filip, Slovenia Kolota Jakub, Poland Komkin A. I., Russia Kopač Janez, Slovenia Korkut Ishan, Turkey Kosel Tadej, Slovenia Krajnik Peter, Slovenia Kramar Davorin, Slovenia Krella Alicja, Poland Krishnaiah J., India Kržan Boris, Slovenia Kušar Janez, Slovenia Kuzman Karl, Slovenia Kyratsis Panagiotis, Greece 66

Lankarani Hamid, USA Lebar Andrej, Slovenia Leciejewski Zbigniew Kazimierz, Poland Lee Paul, Taiwan Li H., China Li Qing-Kui, China Lisjak Dragutin, Croatia Litwin Wojciech, Poland Liu Huibin, China Liu Zhiqiang, China Lojen Gorazd, Slovenia López Javier, Spain Lovrec Darko, Slovenia Lovrin Neven, Croatia Lu Hongbing, USA Magudeeswaran G., India Managuli Ravi, USA Marini M., Italy Medved Sašo, Slovenia Meneghetti Giovanni, Italy Mester Gyula, Hungary Meža Marko, Slovenia Minsaas Atle, Norway Mok Swee, USA Mole Nikolaj, Slovenia Moon Daniel, Sweden Morales-Acevedo Arturo, Mexico Mou Jun Min, PR China Mùjica Mota Miguel, Spain Munih Marko, Slovenia Nagode Marko, Slovenia Nikas George K., Greece Nobile Enrico, Italy Nykänen Arne, Sweden Ocana Jose Luis, Spain Okorn Ivan, Slovenia Olafsen Linda J., USA Oliveira A. Virgílio M., Portugal Oman Simon, Slovenia Padalier Cedric, Australia Palčič Iztok, Slovenia Páscoa Jose, Portugal Pavletić Duško, Croatia Peer Angelika, Germany Pereira Cândida, Portugal Perkovič Marko, Slovenia Perman Mihael, Slovenia Pettit Stephen, UK

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, 65-67

Philipp Ulrich, Germany Pillai Anju S., India Piotrowski Jozef, Poland Pittalà Gaetano M., Italy Pogrebnyak Aleksander D., Ukraine Pokhmurska Hanna, Germany Popović Vladimir, Serbia Potočnik Primož, Slovenia Precup Radu-Emil, Romania Prek Matjaž, Slovenia Pušavec Franci, Slovenia Ramji B. R., India Rauchs Gaston, Luxembourg Razfar Mohammad Reza, Iran Rek Zlatko, Slovenia Rossi Rene M., Switzerland Rozumek Dariusz, Poland Sagade Atul, India Salacinski Tadeusz, Poland Salimi Hamid Reza, Iran Samtaş Gürcan, Turkey Santos Rafael M., Belgium Sauer Bernd, Germany Scheidl Rudolf, Austria Schmidt Dietrich, Germany Schweiker Marcel, Germany Sedlaček Marko, Slovenia Selvam Panner, USA Senegačnik Andrej, Slovenia Silva Renato S. Brasil Simani Silvio, Italy Sinigoj Anton Rafael, Slovenia Slavič Janko, Slovenia Slemnik Mojca, Slovenia Sluga Alojzij, Slovenia Stanković Tino, Switzerland Starbek Marko, Slovenia Stojanović Blaža, Serbia Stopar Bojan, Slovenia Stroud Ian, Switzerland Sućeska Muhamed, Croatia Sun Guifang, USA Sznitman Joss, USA Šafarič Riko, Slovenia Šajn Viktor, Slovenia

Šeruga Domen, Slovenia Štorga Mario, Croatia Šturm Roman, Slovenia Taha Zahari, Malaysia Taher Fatma, UAE Tasič Jurij F., Slovenia Tavčar Jože, Slovenia Theodossiades Stephanos, UK Thomas John, Canada Toibero Marcos, Argentina Tušek Jaka, Slovenia Tušek Janez, Slovenia Ubeyli Mustafa, Turkey Ulaga Samo, Slovenia Ulbin Miran, Slovenia Valentičič Joško, Slovenia Velkavrh Igor, Slovenia Veža Ivica, Croatia Videnič Tomaž, Slovenia Vidmar Peter, Slovenia Vintro-Sanchez Carla, Spain Vuherer Tomaž, Slovenia Vujica Herzog Nataša, Slovenia Vukašinović Nikola, Slovenia Wang Z., China Weber Gerhard-Wilhelm, Turkey Winczek Jerzy, Poland Wu Xi, USA Xiong Liangshan, China Xu S. P., China Yang Ronggen, China Yu T. X., China Zadravec Matej, Slovenia Zaera Ramon Pólo, Spain Zhang Dan, USA Zhang Jie, UK Zhang Mingxing, Australia Zhang Qin He, China Zmitrowicz Alfred, Poland Zoppi Matteo, Italy Zupanič Franc, Slovenia Žerovnik Janez, Slovenia Živanović Zlatomir, Serbia Župerl Uroš, Slovenia

The Editorial would like to thank all the reviewers in participating in reviewing process. We appreciate the time and effort and greatly value the assistance as a manuscript reviewer for Strojniški vestnik – Journal of Mechanical Engineering. 67

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1 Vsebina

Vsebina Strojniški vestnik - Journal of Mechanical Engineering letnik 59, (2013), številka 1 Ljubljana, januar 2013 ISSN 0039-2480 Izhaja mesečno

Razširjeni povzetki člankov Blaž Florjanič, Edvard Govekar, Karl Kuzman: Uporaba nevronske mreže za podporo ekspertne ocene projektov v proizvodnji orodij Yikai Chen, Jie He, Mark King, Wuwei Chen, Changjun Wang, Weihua Zhang: Razvoj modela in analiza dinamične porazdelitve obremenitev pri vzdolžno povezanem zračnem vzmetenju Liane Roldo, Ivan Komar, Nenad Vulić: Zasnova in izbira materialov za okolju prijazen ladijski pogonski sistem Lovro Kuščer, Janez Diaci: Merilna negotovost pri določanju geografske lokacije oddaljenih objektov Hirpa G. Lemu, Tomasz Trzepieciński: Numerična in eksperimentalna študija tornega obnašanja pri natezno-upogibnem preizkusu Andrew J. Dick: Karakterizacija efektivnih vrednosti d31 za PZT na osnovi nelinearnih nihanj dvostransko vpetih mikroresonatorjev Andrej Debenjak, Matej Gašperin, Boštjan Pregelj, Maja Atanasijević-Kunc, Janko Petrovčič, Vladimir Jovan: Detekcija poplavljanja in izsuševanje sklada PEM gorivnih celic Osebne vesti Doktorske disertacije, znanstvena magistrska dela, diplomske naloge

SI 3 SI 4 SI 5 SI 6 SI 7 SI 8 SI 9 SI 10

Prejeto v recenzijo: 2011-07-12 Prejeto popravljeno: 2012-08-22 Odobreno za objavo: 2012-10-25

Uporaba nevronske mreže za podporo ekspertne ocene projektov v proizvodnji orodij Florjanič, B. – Govekar, E. – Kuzman, K. Blaž Florjanič1,* – Edvard Govekar2 – Karl Kuzman2 1 iMold d.o.o., Slovenija 2 Univerza

v Ljubljani, Fakulteta za strojništvo, Slovenija

Proizvodnja orodij je individualen proizvodni proces, za katerega je značilen projektni pristop. Proizvodnja orodij zajema zelo specifičen segment v življenjskem ciklusu izdelka. Nahaja se med razvojem izdelka in serijsko proizvodnjo. Vsak projekt v orodjarstvu je unikaten, neponovljiv. Ena ključnih aktivnosti je zgodnja faza ocene virov, potrebnih za izvedbo projekta. Ta faza določa konkurenčnost proizvodnje. Pravilne ocene v zgodnjih fazah projekta so ključne za njegovo uspešno realizacijo. Namen raziskav je zgraditi zanesljiv in časovno učinkovit ocenitveni model za napovedovanje obsega izdelovalnih časov pri izdelavi orodij za injekcijsko brizganje termoplastov. Obseg izdelovalnih časov je seštevek vseh strojnih proizvodnih ur (časov obdelav), ki so potrebne za izdelavo orodja. Za proizvodnjo orodij je značilno, da se v naročilo realizira manj kot 10% vseh povpraševanj, v rabi pa je pretežno še vedno intuitivni ali hevristični pristop za reševanje predmetnega problema. Namen prispevka je predstaviti cenitveni model za oceno obsegov izdelovalnih časov ob zadovoljivi natančnosti, kar hkrati predstavlja tudi orodje za ovrednotenje ekspertnih ocen. Pri tem se kot podpora pri oceni obsega izdelovalnih časov uporabi model, ki temelji na računalniški simulaciji nevronske mreže. Raziskovalne metode temeljijo na metodah empiričnega opisa in modeliranja. Osrednja naloga raziskave je kvantifikacija vhodnih faktorjev (parametrov) za ustrezen in dovolj zanesljiv opis kompleksnosti geometrije izdelka in modeliranje vpliva kompleksnosti geometrije na obseg izdelovalnih časov. Pri tem se za modeliranje povezav med vhodnimi faktorji in obsegom izdelovalnih časov, kar za eksperta predstavlja kompleksnejši induktivni miselni proces, uporabijo nevronske mreže. Na nevronskih mrežah temelječi predstavljeni model zagotavlja povečano natančnost, objektivnost in ponovljivost ocenitvenega procesa. Za model je pomembna pravilna opredelitev ustreznih vhodnih parametrov, ki so značilno korelirani z obsegom izdelovalnih časov. Prav tako se skrajša čas, potreben za sam cenitveni proces. Uporaba takšnega modela predstavlja zanesljiv temelj za oceno proizvodnih stroškov. Predstavljena metodologija in model sta osnova za vse skupine orodij. V našem primeru se model omejuje na specifično skupino tako imenovanih 1+1-gnezdnih orodij za injekcijsko brizganje termoplastov, značilnih za avtomobilsko industrijo. Navedena specifična skupina zadošča za preverbo metodologije in postavljenega modela. S tem smo se v raziskavi izognili pridobivanju obsežne baze ustreznih primerov za posamezno skupino orodij, ki jih mora uporabnik za svoj obravnavani primer pridobiti iz lastne baze znanj. Ključni doprinos članka je oblikovanje sistematičnega ekspertnega ocenitvenega procesa, podprtega z uporabo modela, temelječega na nevronskih mrežah. Modeliranje z nevronsko mrežo je umeščeno v zgodnjo fazo razvoja izdelka, ko se oblikujejo ocene izdelovalnih časov. Podan je sistematičen ocenitveni postopek, katerega uporabnost je predstavljena na podlagi praktičnega primera, v katerem so podani napotki za doseganje ustrezne stopnje zanesljivosti rezultatov. Model služi tudi kot osnova za nadaljnji razvoj in razširitev cenitvenih metod pri ostalih družinah orodij za injekcijsko brizganje termoplastov. Ključne besede: orodjarstvo, proizvodnja, nevronske mreže, cenitveni procesi

*Naslov avtorja za dopisovanje: iMold d.o.o., Cesta v Pečale 33, 1231 Ljubljana-Črnuče, Slovenija, blaz.florjanic@imold.si

SI 3

Prejeto v recenzijo: 2012-08-23 Prejeto popravljeno: 2012-10-06 Sprejeto v objavo: 2012-11-16

Razvoj modela in analiza dinamične porazdelitve obremenitev pri vzdolžno povezanem zračnem vzmetenju Chen, Y. – He, J. – King, M. – Chen, W. – Wang, C. – Zhang, W. Yikai Chen1,5,* – Jie He2 – Mark King3 – Wuwei Chen4 – Changjun Wang5 – Weihua Zhang1 1 Tehniška

univerza Hefei, Fakulteta za transport, Kitajska univerza, Fakulteta za transport, Kitajska 3 Tehniška univerza v Queenslandu, Center za raziskavo nesreč in varnost v cestnem prometu, Avstralija 4 Tehniška univerza Hefei, Fakulteta za strojništvo in avtomobilsko tehniko, Kitajska 5 Prometni raziskovalni inštitut pri Ministrstvu za javno varnost, Kitajska 2 Jugovzhodna

V zadnjih desetletjih je bilo opravljenih mnogo študij »cesti prijaznih« tovornih vozil, ki so bile namenjene zmanjšanju poškodb cest in povečanju nosilnosti vozil. Zmožnost porazdeljevanja obremenitev med osmi večosnih tovornih vozil, ki je povezana tudi s prijaznostjo do ceste, pa do zdaj še ni bila dovolj preučena. Vzdolžno povezano zračno vzmetenje je učinkovito pri zagotavljanju enakomernega porazdeljevanja obremenitev med osmi. Zaradi omejitev laboratorijske opreme pa je bila večina preizkusov osredotočena samo na hitrost vozil in na omejeno število povezav zračnega vzmetenja. Natančnost rezultatov pri ostalih študijah na osnovi matematičnih modelov pa je omejena zaradi poenostavitev nelinearnosti v modelih. Cilj raziskave je bil preučiti vpliv ključnih parametrov zasnove vzmetenja (statični tlak v zračnih blazinah, premer zračnih vodov in priključkov) na dinamično porazdelitev obremenitve v vzdolžno povezanem zračnem vzmetenju triosnega polpriklopnika s pomočjo natančnih matematičnih modelov. Predstavljen je nov nelinearni polovični model večosnega polpriklopnika z vzdolžno povezanim zračnim vzmetenjem na osnovi mehanike fluidov in termodinamike. Razvit je bil tudi model vzbujanja s cestnimi ravninami na osnovi PSD (spektra gostote moči). Integrirani model vozila in ceste je bil preizkušen in preverjen na različnih cestnih odsekih v okviru skupnega projekta Vzmetenje tovornih vozil – preizkušanje in analize Tehniške univerze v Queenslandu (QUT) ter Oddelka za transport in glavne ceste v Queenslandu (TMR). Nato je bil analiziran vpliv parametrov vzmetenja na dinamično porazdeljevanje obremenitev in prijaznost polpriklopnika do ceste. Glavne ugotovitve so: (1) Metrika prijaznosti do ceste DLC se v splošnem ujema z metriko porazdelitve obremenitve DLSC. (2). Ko se poveča statična višina ali statični tlak, se razmerje optimizacije DLSC zmanjša. Razlog je v tem, da se statična višina povečuje s statičnim tlakom, Vs10 se poveča in prostornina zračnega voda / Vs10 se zmanjša. Učinek uporabe večjega zračnega voda in priključkov je tako manj pomemben (3). Če privzamemo, da je premer zračnega voda vedno večji od premera priključka, je vpliv premera zračnega voda večji od vpliva premera priključka. Če polpriklopnik vozi s hitrostjo 20 m/s po standardni cesti razreda B in je fiksni premer priključka 10 mm, se DLSC pri povečanju premera zračnega voda iz 10 na 100 mm zmanjša samo za 1%. Če pa je fiksni premer zračnega voda 100 mm, je zmanjšanje ob povečanju premera priključka iz 10 na 30 mm 11-odstotno. Vrednost DLSC s povečevanjem premera priključka zelo počasi pada in se nato ustali na konstantni vrednosti. Glavni prispevek tega članka je v razvoju natančnega modela vzdolžno povezanega zračnega vzmetenja ter v analizi vpliva parametrov vzmetenja na dinamično porazdeljevanje obremenitev. Raziskava upošteva samo vpliv parametrov vzmetenja na porazdeljevanje obremenitev. Predlagani model bo osnova za prihodnje raziskave vpliva voznih pogojev (vozna hitrost, neravnine na cestišču) ter integriranih metod krmiljenja (krmiljenje LQG, krmiljenje z mehkimi nevronskimi mrežami) vzdolžno povezanega zračnega vzmetenja in polaktivnih amortizerjev na osnovi porazdeljevanja obremenitev. Ključne besede: modeliranje, vozni pogoji, dinamično porazdeljevanje obremenitev, vzdolžna povezanost, zračno vzmetenje, tovorna vozila

SI 4

*Naslov avtorja za dopisovanje: Tehniška univerza Hefei, Fakulteta za transport, 193 # Tunxi Road, Hefei, Kitajska, leochen079307@hotmail.com

Prejeto v recenzijo: 2012-05-21 Prejeto popravljeno: 2012-10-19 Odobreno za objavo: 2012-11-06

Zasnova in izbira materialov za okolju prijazen ladijski pogonski sistem Roldo, L. – Komar, I. – Vulić, N. Liane Roldo1,* – Ivan Komar2 – Nenad Vulić3

1 Zvezna

univerza Rio Grande do Sul, Materials Department, Brazil

2 Univerza v Splitu, Fakulteta za pomorstvo, Hrvaška 3

Hrvaški ladijski register, Hrvaška

Za posebno zasnovo uležajenja grodnične cevi za osi plovil je ob upoštevanju sistema oljnega ali vodnega mazanja in posledične lekaže maziva ključnega pomena izbira materiala. Grodnična cev za os značilne oceanske ladje vsebuje približno 1500 litrov olja. Zato je tudi pri konzervativni vrednosti lekaže 6 l/dan letno onesnaženje z oljem iz grodničnih cevi zaradi ladijskega prometa na globalni ravni (svetovna flota trgovskih ladij obsega približno 40000 ladij z nosilnostjo nad 1000 DWT) ocenjeno na več kot 80 milijonov litrov. Predstavljena študija zato raziskuje izvedljivost ter prednosti uporabe vodno mazanih polimernih ležajev namesto običajnih oljno mazanih ležajev iz bele kovine. Metoda raziskave vključuje numerični model in programsko aplikacijo na osnovi metode končnih razlik in izoviskoznega modela. Za oljno mazano uležajenje iz bele kovine je bil izbran hidrodinamični model mazanja. Jedro za numerične izračune (metoda končnih razlik za numerično reševanje Reynoldsovih enačb) je bilo izdelano v Matlabovem programu Partial. Pri vodno mazanih polimernih ležajih je bil zaradi elastičnih deformacij mazanih površin potreben drugačen pristop, oz. program elastohidrodinamičnega mazanja (EHL). Program izračunava elastične deformacije polimernih ležajev na osnovi analize elastohidrodinamičnega mazanja izoviskoznih režimov mazanja elastičnih teles. Pri izračunavanju modelov je bilo vključenih pet različnih materialov, ki se trenutno uporabljajo za izdelavo ležajev ladijskih osi: poleg štirih različnih vrst polimerov je bila kot primerjalni kovinski material uporabljena kositrova zlitina. Za validacijo modelov so bile uporabljene dejanske temperature ležajev za različne pogonske režime značilnih ladij – ladje za prevoz razsutih tovorov s 50000 DWT, kontejnerske ladje z 11000 TEU in katamarana s 496 GT. Analiza zbranih podatkov za tri različne vrste ladij je pokazala, da so izgube moči pri polimernih ležajih od 6- do 36-krat manjše kot pri običajnih ležajih iz bele kovine. Zato je mogoče sklepati, da so polimerni ležaji bistveno energijsko učinkovitejši in okolju prijaznejši. Ob upoštevanju napovedi ekonomskih učinkov uporabe polimernih ležajev grodničnih cevi za osi na globalni ravni, ki je bila narejena za vzorec 3000 ladij (1000 ladij vsake vrste) in obdobje uporabe 20 let, bi bilo mogoče prihraniti 1580000 ton goriva ter 158400000 litrov mazalnega olja. Z uporabo polimernih ležajev grodničnih cevi za osi bi bilo torej mogoče doseči pomembne prihranke goriva in olja. Razen pomembnih finančnih prihrankov pa bi dosegli tudi boljšo energijsko učinkovitost ladij. Tako bi prispevali k izpolnjevanju okoljskih zahtev Mednarodne pomorske organizacije ter indeksa energijske učinkovitosti zasnove. Uveljavitev teh modelov in pristopa k izbiri materialov daje tudi trdno osnovo za predlog drugačnega pristopa za ladijske konstruktorje, oziroma nadomestitev ležajev iz bele kovine s polimernimi ležaji. Rešitev ima tudi določene pomanjkljivosti, kot je potreba po natančni obdelavi osi in ležajev, pravilni pripravi (npr. filtraciji) morske vode idr., ki pa v tem članku niso bile obravnavane. Z uporabo polimernih ležajev grodničnih cevi za osi bi bilo mogoče doseči pomembne prihranke goriva in olja, kakor tudi prispevati k izpolnjevanju okoljskih zahtev za ladje in povezanega indeksa energijske učinkovitosti zasnove. V analizi je podana tudi točka optimizacije oz. obdobje odplačila višje začetne naložbe v vgradnjo polimernih ležajev grodnične cevi. Lastniki ladij se lahko tako na podlagi znanstveno preizkušene metodologije lažje odločijo o smiselnosti zamenjave ležajev grodnične cevi iz bele kovine s polimernimi ležaji. Ključne besede: ladijski pogonski sistem, izbira materialov, polimer, Babbittova kovina, ležaj osi, programska aplikacija

*Naslov avtorja za dopisovanje: Zvezna univerza Rio Grande do Sul, Av. Osvaldo Aranha, 99/604, Porto Alegre - RS, 90035-190 Brazilija liane.roldo@ufrgs.br

SI 5

Prejeto v recenzijo: 2012-06-11 Prejeto popravljeno: 2012-10-03 Odobreno za objavo: 2012-11-12

Merilna negotovost pri določanju geografske lokacije oddaljenih objektov Kuščer, L. – Diaci, J. Lovro Kuščer* – Janez Diaci Univerza v Ljubljani, Fakulteta za strojništvo, Slovenija

Natančno določanje položaja na zemeljski površini je v sodobnem času postalo nepogrešljivo na področju transporta, geologije, kmetijstva, reševanja v nujnih primerih, pa tudi pri okoljskem nadzoru, vojaških aplikacijah in drugje. V zadnjih desetletjih je bil opazen skokovit napredek v razvoju navigacijskih sistemov in naprav, s čimer so se odprle možnosti njihove uporabe na novih področjih. Pri tem velja izpostaviti predvsem satelitske navigacijske sisteme, ki v kombinaciji s sodobnimi sprejemniki omogočajo natančno in zanesljivo določitev položaja v najrazličnejših okoljih. Vendar pa natančno poznavanje zgolj lastnega položaja v določenih primerih ne zadostuje. Pogosto želimo hitro in enostavno pridobiti informacije o položaju nekega oddaljenega objekta, ki je lahko nedostopen, nevaren ali zgolj preveč oddaljen. S takšnimi zahtevami smo pogosto soočeni zlasti na področju vojaških aplikacij, okoljskega nadzora in geologije. Za izvedbo analize merilnih negotovosti pri določanju položaja oddaljenih objektov smo v okviru predstavljene raziskave razvili integriran merilni sistem, ki je sestavljen iz komercialno dostopnih gradnikov. Merilni sistem je prirejen za namestitev na vozilo in vsebuje naslednje merilne naprave: laserski razdaljemer, GNSS sprejemnik/ kompas in dvoosni elektronski merilnik naklona. Temu je za potrebe dokumentiranja meritev dodana še CCTVvideokamera, ki je skupaj z laserskim razdaljemerom nameščena na dvoosnem stabiliziranem vrtljivem podnožju. Merilni sistem krmili prenosni osebni računalnik z namensko razvito programsko opremo, ki vključuje tudi dostop do geografskega informacijskega sistema za takojšen prikaz izvedenih meritev na tridimenzionalnem zemljevidu. V članku predstavljamo pristop k določanju merilnih karakteristik integriranega merilnega sistema na osnovi specifikacij proizvajalca posamezne merilne naprave ter primerjavo dobljenih rezultatov s terenskimi meritvami. Pri tem je bila ocena merilne negotovosti izmerjenega položaja pridobljena z uporabo simulacij Monte Carlo ter z zakonom o prenosu varianc in kovarianc. Na osnovi pridobljenih rezultatov so bili določeni prispevki posameznih merilnih naprav k skupni negotovosti izmerjenega položaja oddaljenega objekta. Izkazalo se je, da ima pri manjših oddaljenostih negotovost meritve razdalje največji vpliv na skupno merilno negotovost, medtem ko pri večjih razdaljah prevlada vpliv negotovosti pri meritvi azimuta. Poleg teoretične analize na osnovi specifikacij proizvajalcev merilnih naprav je bila izvedena tudi eksperimentalna analiza merilnih karakteristik na terenu, kjer so bili uporabljeni namensko postavljeni objekti (tarče) na razdaljah do 1 km in obstoječi objekti na razdaljah do 20 km. Rezultati meritev so sicer pokazali določene razlike med izmerjenimi in specificiranimi merilnimi karakteristikami, vendar so opažena odstopanja sistematična in jih je zato mogoče odpraviti z ustreznim umerjanjem. Nadalje se je izkazalo, da največjo omejitev razvitega merilnega sistema, ki je v osnovi namenjen meritvam na večjih razdaljah, predstavlja negotovost pri merjenju azimuta. Zato smo sistemu dodali možnost umerjanja z znanimi oddaljenimi objekti, ki jih najdemo v geografskem informacijskem sistemu. Primerjava rezultatov terenskih meritev s simulacijami je pokazala, da je z izvedbo simulacij mogoče pridobiti zadovoljivo oceno merilnih karakteristik predstavljenega integriranega sistema zgolj na osnovi specifikacij proizvajalcev posameznih merilnih naprav. Ključne besede: lokalizacija oddaljenih objektov, merilna negotovost, simulacija Monte Carlo, laserski razdaljemer, GNSS kompas, geografski informacijski sistem

SI 6

*Naslov avtorja za dopisovanje: Univerza v Ljubljani, Fakulteta za strojništvo, Aškerčeva 6, 1000 Ljubljana, Slovenija, lovro.kuscer@fs.uni-lj.si

Prejeto v recenzijo: 2012-02-21 Prejeto popravljeno: 2012-08-03 Odobreno za objavo: 2012-09-04

Numerična in eksperimentalna študija tornega obnašanja pri natezno-upogibnem preizkusu Hirpa G. Lemu1,* – Tomasz Trzepieciński2 1 Univerza

v Stavangerju, Oddelek za strojništvo, konstrukcije in materiale, Norveška univerza v Rzeszowu, Katedra za preoblikovanje, Poljska

2 Tehniška

Torno obnašanje pri preoblikovanju pločevine je odvisno od več parametrov, med katerimi sta tudi površinska hrapavost in mazanje. Študije kažejo, da torne in materialne lastnosti neposredno vplivajo na proces in so tudi medsebojno odvisne Topografija površine ima velik vpliv tudi na torno obnašanje kontaktne površine in njeno obrabo. Zato obstaja potreba po opredelitvi vloge trenja in zanesljivih metodah za kvantifikacijo koeficienta trenja pri preoblikovanju kovin. V članku je predstavljena primerjalna študija tornega obnašanja pri natezno-upogibnem preizkusu (BUT) pločevine iz jekla, medenine in aluminijevih zlitin z eksperimentalnim in numeričnim pristopom. Zasnovana je bila eksperimentalna študija za karakterizacijo vpliva mazanja in površinske hrapavosti na torne lastnosti. Uporabljena je bila tudi metoda končnih elementov (MKE), ki omogoča simulacijo tečenja materiala pločevine ob upoštevanju kompleksnega pojava trenja. Eksperimentalni preizkus je bil izveden s preizkusnim trakom, ki je bil na eni strani vpet v glavo z merilno celico. Trak je bil ovit okrog fiksnega valja premera 20 mm in obremenjen v napravi za natezni preizkus, ki zagotavlja približno 90-stopinjski stik. Preizkus je bil opravljen s štirimi koluti orodnega jekla X165CrV12 z različnimi vrednostmi površinske hrapavosti. Med izvedbo nateznoupogibnega preizkusa so bile merjene deformacije, na osnovi tega pa je bila izračunana hitrost preoblikovanja po prerezu traku. Vpliv plastičnih deformacij na koeficient trenja je bil preučevan v pogojih brez mazanja in z mazanjem. Simulacija je bila opravljena s programom MSC.Marc + MENTAT 2007r1. Kolut je bil modeliran kot popolnoma tog elastoplastičen material, lastnosti pa so bile popisane z von Misesovim kriterijem tečenja z izotropnim preoblikovalnim utrjevanjem. Rezultati: (1) Vpliv površinske hrapavosti kolutov: Torni preizkusi s pločevino AA5754 H24 so pokazali, da se koeficient trenja spreminja s površinsko hrapavostjo kolutov in, da obstajajo značilne razlike pri tornem obnašanju. Pri preizkusih suhega trenja se koeficient trenja pri višjih vrednostih hrapavosti zmanjšuje z deformacijami preizkušanca. Vzrok za to so lahko spremembe topografije površine pločevine v deformiranem stanju. Ugotovljeno je bilo tudi, da površinska hrapavost kolutov v mazanem območju zmanjša koeficient trenja za približno 25 do 40 %. Medtem ko je koeficient suhega trenja pri medeninasti pločevini stabilen, pa mazanje spremeni značaj sprememb tornega upora, pri čemer je bila najnižja vrednost ugotovljena pri kolutih s površinsko hrapavostjo Ra = 0,63 mm. Trend rezultatov za jekleno pločevino kakovosti za globoki vlek je podoben kot pri ostalih dveh materialih in koeficient trenja se povečuje s površinsko hrapavostjo. (2) Kontaktni tlak pri lokalnem zoženju: Rezultati v tem primeru ne kažejo tipičnega trenda ali spremenljivosti kot funkcije kakovosti površine in tornega stanja. Splošna ugotovitev je, da kontaktni tlak pada s povečevanjem vrednosti hrapavosti. Najmanjši kontaktni tlak med preizkušanimi materiali je bil ugotovljen pri AA5754 H24, jeklo za globoki vlek pa ima največjo zmogljivost utrjevanja. (3) Ekvivalentna celotna plastična deformacija: V vseh primerih je bil ugotovljen kolaps prostih koncev preizkušancev, obremenjenih z zavorno silo upora. Pojav ustreza lokalnemu zoženju preseka zlasti pri visokih vrednostih trenja. Primerjava pogojev suhega trenja in mazanja kaže majhne, a značilne razlike v porazdelitvi deformacij. Vpliv porazdelitve deformacij pri velikem kontaktnem tlaku v kontaktnem območju je npr. pretežno odvisen od geometrije, vpliv mazanja pa je majhen. Določene opažene razlike med eksperimentalnimi in numeričnimi rezultati so verjetno posledica poenostavitev. (4) Porazdelitev kontaktnega upora in efektivnih deformacij: Ugotovljeno je bilo, da porazdelitev napetosti kontaktnega trenja po kontaktni površini pločevinastega koluta ni enakomerna, trend porazdelitve pa je podoben v suhem in v mazanem stanju do približno 60-stopinjskega kontakta. Porazdelitev v tem območju kaže, da so vrednosti napetosti kontaktnega trenja pri suhem trenju približno petkrat večje kot pri mazanem stiku. Na začetku kontakta so bili v obeh primerih opaženi tudi lokalni vrhovi napetosti kontaktnega trenja, vzrok temu pa je lahko v različnih lastnostih tečenja materiala pri višjih celotnih deformacijah preizkušanca. Članek obravnava predvsem vpliv mazanja in površinske hrapavosti. Uporaba MKE na tem področju omogoča simulacijo tečenja materiala ob upoštevanju pojava kompleksnega trenja ter podaja boljše razumevanje kontaktnih pogojev za sprejemljivo ceno. Ključne besede: upogibanje in nateg, preoblikovanje, trenje, koeficient trenja, površinska hrapavost, kontaktni tlak *Naslov avtorja za dopisovanje: Univerza v Stavangerju, Oddelek za strojništvo, konstrukcije in materiale, 4036 Stavanger, Norveška, Hirpa.g.lemu@uis.no

SI 7

Prejeto v recenzijo: 2012-06-22 Prejeto popravljeno: 2012-09-12 Odobreno za objavo: 2012-11-29

Karakterizacija efektivnih vrednosti d31 za PZT na osnovi nelinearnih nihanj dvostransko vpetih mikroresonatorjev Dick, A.J. Andrew J. Dick

Univerza Rice, Oddelek za strojništvo in materiale, Združene države Amerike

V članku je predstavljena metoda za izračunavanje približnih vrednosti transverzalnega piezoelektričnega koeficienta PZT na osnovi nelinearnih nihanj mikroresonatorja. V primeru, da mikroresonatorji v podatkih spektralnega odgovora izkazujejo značilno nelinearno vedenje, kot je npr. utrjevanje, uporaba linearnih metod karakterizacije namreč ni možna. Za karakterizacijo transverzalnega piezoelektričnega koeficienta materiala PZT na osnovi nelinearnega odgovora je bil izpeljan model nelinearnega mikroresonatorja in uporabljen za parametrično identifikacijo. Postopek parametrične identifikacije je uporabljen na podatkih frekvenčnega odgovora, zajetih iz mikroresonatorja pri vhodu, sestavljenem iz sinusnega signala z naraščajočo frekvenco in inkrementalno spreminjajoče se nosilne napetosti. Za identifikacijo odvisnosti med silami v materialu PZT zaradi uporabljene nosilne napetosti so bili na modelu preučeni premiki efektivne lastne frekvence, ugotovljeni s postopkom parametrične identifikacije. Izračunana povprečna vrednost transverzalnega piezoelektričnega koeficienta za obravnavane naprave je bila –129,7 pm/V, standardna deviacija pa je znašala manj kot 6 % te vrednosti. Dolžina naprav, uporabljenih v študiji, je bila 100, 200 in 400 mm. V nadaljnji analizi se je izkazalo, da je za vse naprave izračunana vrednost piezoelektričnega koeficienta –127,8 pm/V, ko se modelirana dolžina vsake naprave razlikuje od imenske vrednosti za manj kot 2 %. Raven geometrijske variabilnosti se ujema z natančnostjo postopkov mikroizdelave. Rezultati študije kažejo, da je metoda primerna za določanje efektivne vrednosti transverzalnega piezoelektričnega koeficienta materiala PZT, ki se uporablja kot izvršni člen pri mikroresonatorjih z dvostransko vpetim nosilcem, če je dinamično obnašanje značilno nelinearno. Izračunane vrednosti dajejo z uporabo blokovnega modela sile za popis piezoelektričnega materiala zgornjo mejo za izračunani koeficient. Študija je prav tako pokazala, da so natančni podatki o dolžini naprave pomembni, saj so izračunane vrednosti koeficientov občutljive na spremembe modelirane dolžine. Opisani postopek karakterizacije dodatno izboljšuje možnosti preučevanja piezoelektričnega vzbujanja na mikroravni ob prisotnosti značilnega nelinearnega vedenja, kakor tudi preučevanja vpliva različnih izdelovalnih postopkov na lastnosti piezoelektričnih materialov v teh napravah. Delo predstavlja orodje za razširitev trenutnih metod linearne karakterizacije materiala s karakterizacijo transverzalnega piezoelektričnega koeficienta na osnovi nelinearnega odgovora. Z vse manjšimi izmerami naprav se povečuje tudi možnost značilnih nelinearnih vplivov. Tehnike nelinearne analize, kot je metoda, opisana v članku, so pomembne za razvoj novih manjših naprav. Ključne besede: piezoelektrični material, resonator z mikronosilcem, nelinearna nihanja

SI 8

*Naslov avtorja za dopisovanje: Univerza Rice, Oddelek za strojništvo in materiale, 6100 Main Street – MS 321, Houston, Texas 77005, ZDA, andrew.j.dick@rice.edu

Prejeto v recenzijo: 2012-06-07 Prejeto popravljeno: 2012-10-19 Odobreno za objavo: 2012-11-21

Detekcija poplavljanja in izsuševanje sklada PEM gorivnih celic Andrej Debenjak1,* – Matej Gašperin1,2 – Boštjan Pregelj1,3 – Maja Atanasijević-Kunc4 – Janko Petrovčič1,3 – Vladimir Jovan1,3 1 Institut “Jožef Stefan”, Slovenija univerza Plzen, Fakulteta za elektrotehniko/RICE, Češka 3 Center odličnosti nizkoogljične tehnologije – CO NOT, Slovenija 4 Univerza v Ljubljani, Fakulteta za elektrotehniko, Slovenija

2 Zahodnočeška

Vodikove gorivne celice s protonsko prevodno membrano (angl. Proton Exchange Membrane – PEM) so elektrokemične naprave, ki z visokim izkoristkom pretvarjajo kemično energijo vodika v električno energijo. Edini produkt pretvorbe je voda, poleg električne energije pa se sprošča tudi toplota. Z okoljevarstvenega vidika so gorivne celice zanimive, ker med delovanjem ne proizvajajo onesnaževalcev okolja in toplogrednih plinov. V primerjavi z ostalimi tipi gorivnih celic odlikujejo PEM gorivne celice razmeroma nizke obratovalne temperature med 50 in 120 °C, visoka specifična moč (kg/kW), zmožnost delovanja s kisikom iz zraka ter kratki zagonski ter ustavljalni časi. Vse te lastnosti PEM gorivnih celic govorijo v prid njihovi uporabi v mnogih stacionarnih in transportnih aplikacijah ter tako predstavljajo možno alternativo današnjim motorjem z notranjim izgorevanjem. Kljub vsemu pa PEM gorivne celice še niso našle svojega mesta v vsakodnevnih aplikacijah. Vzroke za to gre iskati predvsem v nekaterih težavah, povezanih z zanesljivostjo in vzdržljivostjo te tehnologije, ki jih je nujno odpraviti za uspešno masovno uporabo. Študije so pokazale, da na kakovost delovanja PEM gorivnih celic močno vpliva upravljanje z vodo znotraj celic, kjer je treba vzdrževati natančno ravnovesje plinaste in utekočinjene vode. Neustrezno ravnovesje namreč vodi v neželene pojave, ki imajo neposreden negativen vpliv na delovanje. V primeru nezadostnega odvajanja nastale vode se le-ta začne kondenzirati v dovodnih plinskih kanalčkih, kar povzroči poplavljanje gorivnih celic ter onemogoči dovod reaktantov do mesta reakcije. V nasprotnem primeru pa se zaradi prekomernega odvajanja vode PEM membrana izsuši, to pa posledično poveča notranjo upornost celice. Z namenom izboljšanja zanesljivosti in vzdržljivosti je torej treba diagnosticirati in klasificirati ti dve napaki, nato pa izvesti ustrezno regulacijsko akcijo, ki odpravi napake in hkrati prepreči nastanek poškodb. V članku smo se posvetili diagnosticiranju poplavljanja in izsuševanja PEM gorivnih celic z uporabo Elektrokemične Impedančne Spektroskopije (EIS). Metoda je že bila uspešno uporabljena za diagnosticiranje posameznih laboratorijskih gorivnih celic, pri čemer je bila uporabljena namenska strojna in programska oprema. Naš namen je bil razširiti uporabnost metode na večje, komercialno zanimive sisteme, ki imajo sklad celic, sestavljen iz več deset PEM gorivnih celic, pri čemer morajo biti vse potrebne meritve in obdelava podatkov opravljene s splošno namensko opremo. Največja prednost takšnega pristopa je, da je treba meriti le tok in napetost celotnega sklada gorivnih celic. Hkrati pa ravno to prinaša tudi največje težave pri sami diagnostiki. Težava se namreč skriva v tem, da v danem trenutku niso vse celice v skladu podvržene določeni napaki, ampak se napaka pojavlja le v nekaterih celicah. Ker pa se meritve opravljajo na celotnem skladu, je napako mnogo težje zaznati, saj se napaka izredno šibko odraža v impedanci celotnega sklada. S študijo smo pokazali, da je z metodo EIS mogoče uspešno diagnosticirati poplavljanje in izsuševanje tudi znotraj večjega sklada PEM gorivnih celic. Impedančne meritve so bile opravljene v frekvenčnem področju od 1 Hz do 1 kHz. Izkazalo se je, da je reprezentativno frekvenčno območje, kjer je možna diagnostika, le med 30 in 300 Hz. V tem področju se impedanca sklada ob prisotnosti poplavljanja ali izsuševanja zaznavno spreminja, pri čemer je vpliv poplavljanja nekoliko izrazitejši od vpliva izsuševanja. Članek razen podrobnega opisa načina in parametrov meritve podaja tudi postopek obdelave podatkov ter odločitvenega algoritma za končno klasifikacijo napak. Le-ta je osnovan na podlagi statističnih lastnosti izmerjenih impedanc ter na podlagi vhodnih impedančnih podatkov določi, ali se v skladu dogaja poplavljanje ali izsuševanje. Nadaljnje delo bo usmerjeno predvsem v širitev splošnosti metode. Največjo oviro splošnosti predstavlja odvisnost impedančne karakteristike od bremenskega toka. To bomo skušali rešiti z opazovanjem relativnih sprememb impedance in ne absolutnih vrednosti, kot je izvedeno sedaj. Nadalje pa se bomo posvetili tudi možnosti kvantitativnega ocenjevanja vlažnosti znotraj sklada, saj trenutna zasnova omogoča le ugotavljanje ali je določena napaka prisotna ali ne. Dolgoročni cilj je usmerjen v zasnovo celostnega diagnostičnega sistema, ki bo v realnem času spremljal delovanje sklada PEM gorivnih celic. Ključne besede: PEM gorivne celice, elektrokemična impedančna spektroskopija, diagnostika, poplavljanje, izsuševanje *Naslov avtorja za dopisovanje: Institut “Jožef Stefan”, Jamova 39, 1000 Ljubljana, Slovenija, andrej.debenjak@ijs.si

SI 9

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, SI 10-12 Osebne objave

Doktorske disertacije, znanstvena magistrska dela, diplomske naloge

DOKTORSKE DISERTACIJE Na Fakulteti za strojništvo Univerze v Ljubljani sta obranila svojo doktorsko disertacijo: ● dne 10. decembra 2012 Blaž KRESE z naslovom: »Nelinearna analiza dinamike lasersko tvorjenih kapljic« (mentor: prof. dr. Edvard Govekar); V doktorskem delu obravnavamo dinamiko različnih režimov kapljanja, ki jih opazimo pri laserskem tvorjenju zaporedja kovinskih kapljic v odvisnosti od moči ločilnega bliska. Za opazovanje procesa tvorjenja kapljic smo uporabili hitro infrardečo kamero. Iz časovnega zaporedja termovizijskih posnetkov kamere smo tvorili skalarne časovne vrste, na podlagi katerih smo analizirali in okarakterizirali dinamiko kapljanja. Pri tem smo se osredotočili na analizo časovnih vrst značilnih primerov spontanega, mešanega in prisilnega kapljanja. Za analizo izbranih časovnih vrst smo uporabili metode nelinearne analize časovnih vrst, rekurentne diagrame in časovnofrekvenčno analizo z uporabo trenutne frekvence. Iz rezultatov analize smo ugotovili, da je proces laserskega tvorjenja zaporedja kapljic smiselno obravnavati kot deterministični nizkodimenzionalni dinamični sistem. Na podlagi izračunov Ljapunovih eksponentov in testa 0-1 za kaos smo za časovni vrsti spontanega in prisilnega kapljanja ugotovili, da sta odraz kaotičnega dinamičnega sistema. Nadalje smo z uporabo rekurentnih diagramov in njihove kvantitativne analize omenjeni rezultat potrdili ter dodatno okarakterizirali prehod med spontanim in prisilnim kapljanjem kot intermitenčni prehod kaos-kaos z vmesnim nestacionarnim režimom kapljanja. Nazadnje smo izbrane režime kapljanja okarakterizirali tudi v časovno-frekvenčnem prostoru z uporabo empirične modalne dekompozicije in trenutne frekvence; ● dne 20. decembra 2012 Karolj NEMEŠ z naslovom: »Optimizacija termičnega lečenja in izkoristka bliskovnega laserja Er:YAG« (mentor: prof. dr. Janez Možina); V tem delu je obravnavan vpliv termičnega lečenja laserja Er:YAG, delujočega v prosti generaciji in črpanega s polikromatično svetlobo ksenonove bliskavice, na širjenje laserskega snopa znotraj in zunaj laserskega resonatorja. Predstavljena je zgradba sodobnega Er:YAG laserja in analizirana sta dva različna načina črpanja laserja s polikromatično svetlobo ksenonove bliskavice z zvonastimi in pravokotnimi črpalnimi bliski. Idetificirani so in analizirani viri termičnega lečenja, ki izhajajo iz SI 10

kompleksne strukture energijskih stanj kristala Er:YAG in prenosa energije med temi v času črpanja s ksenonovo bliskavico. Vpliv termičnega lečenja na tvorbo laserskih bliskov je eksperimentalno izmerjen in podana je kvalitativna slika formiranja termične leče v času črpalnih bliskov ter razvoja povprečne termične leče v kristalu Er:YAG pri črpanju laserja s ponavljajočimi se črpalnimi bliski. Izmerjena je odvisnost goriščne razdalje termične leče od povprečne črpalne moči in določena funkcionalna odvisnost gorišča od črpalne moči. Izdelan je numerični model širjenja laserskega snopa z uporabo ABCD matrik z upoštevanjem termične leče, s katerim je možno načrtovati širjenje laserskega snopa znotraj in zunaj resonatorja. Model je eksperimentalno potrjen in uporabljen za optimizacijo resonatorja in izkoristka laserja Er:YAG. ZNANSTVENA MAGISTRSKA DELA Na Fakulteti za strojništvo Univerze v Ljubljani je z uspehom zagovarjal svoje magistrsko delo: ● dne 11. novembra 2012 Anže ČELIK z naslovom: »Identifikabilnost utrjevalnih lastnosti elasto-plastičnega gradiva na osnovi torzijskega obremenjevanja« (mentor: prof. dr. Boris Štok). DIPLOMIRALI SO Na Fakulteti za strojništvo Univerze v Ljubljani so pridobili naziv univerzitetni diplomirani inženir strojništva: dne 19. decembra 2012: Jure BRANK z naslovom: »Analiza poravnave keramično vezanih brusov iz kubičnega borovega nitrida« (mentor: doc. dr. Peter Krajnik Somentor: prof. dr. Janez Kopač); Jernej VRHOVEC z naslovom: »Kinematika toka v Francisovi turbini« (mentor: izr. prof. dr. Marko Hočevar, somentor: prof. dr. Branko Širok); dne 21. decembra 2012: Uroš JERIČ z naslovom: »Tehnološka in ekonomska primerjava varjenja MAG/MIG in CMT« (mentor: prof. dr. Janez Tušek, somentor: doc. dr. Damjan Klobčar); Jure MAVRI z naslovom: »Varjenje orodnih jekel pod praškom« (mentor: prof. dr. Janez Tušek);

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, SI 10-12

Marko PODGORELEC z naslovom: »Optimiranje pritrdilnih mest vibracijskega vpenjala žarometa« (mentor: izr. prof. dr. Jernej Klemenc); Tea UŠAJ z naslovom: »Ravni fotonapetostni - toplotni sprejemniki sončne energije« (mentor: doc. dr. Andrej Kitanovski Somentor: prof. dr. Alojz Poredoš): Gregor ZEVNIK z naslovom: »Analiza odgora elementov pri varjenju orodnih jekel po postopku TIG in pri laserskem varjenju« (mentor: prof. dr. Janez Tušek). * Na Fakulteti za strojništvo Univerze v Mariboru so pridobili naziv univerzitetni diplomirani inženir strojništva: dne 20. decembra 2012: Gregor FAJDIGA z naslovom: »Sledenje procesa izdelave s selektivnim laserskim sintranjem« (mentor: izr. prof. dr. Igor Drstvenšek, somentor: asist. mag. Tomaž Brajlih); Rok PANIKVAR z naslovom: »Numerična simulacija naravne konvekcije okoli vročega telesa« (mentor: doc. dr. Jure Ravnik, somentor: prof. dr. Leopold Škerget); Marko VURZER z naslovom: »Inženirsko oblikovanje nevrokirurške naprave “MICRODRIVE”« (mentor: izr. prof. Vojmir Pogačar). * Na Fakulteti za strojništvo Univerze v Ljubljani so pridobili naziv diplomirani inženir strojništva: dne 5. decembra 2012: Aleksander BERGANT z naslovom: »Preizkušanje izolacijske učinkovitosti požarnozaščitnih premazov« (mentor: izr. prof. dr. Ivan Bajsić); Kristian GRUDEN z naslovom: »Razvoj in izdelava preizkuševališča za simulacijo vpliva peska na delovanje krmilnega ventila za sanitarno toplo vodo« (mentor: izr. prof. dr. Ivan Bajsić); Iztok LAMBREŠČAK z naslovom: »Optimizacija razreza mineralne volne« (mentor: doc. dr. Davorin Kramar, somentor: prof. dr. Janez Kopač); Darja PEKOVEC z naslovom: »Lasersko osvetljevanje posadke civilnega zrakoplova« (mentor: viš. pred. mag. Aleksander Čičerov, somentor: prof. dr. Janez Možina); Peter REGOUC z naslovom: »Računalniško in eksperimentalno podprto konstruiranje svetlobnega vodnika« (mentor: izr. prof. dr. Jože Tavčar, somentor: prof. dr. Jožef Duhovnik);

Urban SIMONČIČ z naslovom: »Priprava tehnologije varjenja krom-molibdenovih jekel« (mentor: prof. dr. Janez Tušek); dne 7. decembra 2012: Marko BRCE z naslovom: »Energijska in stroškovna analiza fasadnega sistema v stanovanjskih stavbah« (mentor: prof. dr. Vincenc Butala, somentor: doc. dr. Matjaž Prek); Boštjan CEPIČ z naslovom: »Načrtovanje materialnih potreb« (mentor: izr. prof. dr. Janez Kušar, somentor: prof. dr. Marko Starbek); Luka DERŽAJ z naslovom: »Parametrična analiza vpliva ogrevalnih in prezračevalnih sistemov na toplotno okolje« (mentor: prof. dr. Vincenc Butala, somentor: doc. dr. Matjaž Prek); Marko JANČAR z naslovom: »Razvoj in optimizacija ohišja hidravličnega digitalnega ventila« (mentor: izr. prof. dr. Niko Herakovič); Tomaž PODOBNIK z naslovom: »Naprava za avtomatizirano sestavljanje konektorjev VVT 3.0« (mentor: izr. prof. dr. Peter Butala); dne 11. decembra 2012: Nuša KRAVANJA z naslovom: »Primerjava obdelave in uporabe bakrenih in grafitnih elektrod« (mentor: prof. dr. Janez Kopač); Anže MERHAR z naslovom: »Popravilo kompozitnih materialov pri jadralnih letalih« (mentor: izr. prof. dr. Tadej Kosel, somentor: izr. prof. dr. Roman Šturm); Martin MLINAR z naslovom: »Vpliv špranjskega toka na karakteristiko aksialnega ventilatorja« (mentor: prof. dr. Branko Širok); Jure NOVAK z naslovom: »Analiza poletov letala z vertikalnim vzletom in pristankom« (mentor: doc. dr. Viktor Šajn, somentor: pred. Miha Šorn); Urban SAJOVIC z naslovom: »Analiza porušitve vzgona pri vzletu letala« (mentor: doc. dr. Viktor Šajn, somentor: pred. Miha Šorn); Bojan ŠUBIC z naslovom: »Mobilna platforma za merjenje aerodimaničnih veličin« (mentor: izr. prof. dr. Tadej Kosel, somentor: doc. dr. Boris Jerman). * Na Fakulteti za strojništvo Univerze v Mariboru je pridobil naziv diplomirani inženir strojništva (UN): dne 20. decembra 2012: Sandi BROZ z naslovom: »Oblikovanje delovnega mesta izdelave “VRAT COMBI” z upoštevanjem antropometrije« (mentorica: doc. dr. Nataša Vujica Herzog, somentor: doc. dr. Marjan Leber); SI 11

Strojniški vestnik - Journal of Mechanical Engineering 59(2013)1, SI 10-12

* Na Fakulteti za strojništvo Univerze v Mariboru so pridobili naziv diplomirani inženir strojništva: dne 18. decembra 2012: Dušan TOPOLOVEC z naslovom: »Programiranje varilnega robota« (mentor: izr. prof. dr. Karl Gotlih, somentor: doc. dr. Tomaž Vuherer); Andrej SMERKE z naslovom: »Primerjava porabe energije za ogrevanje med pasivnimi, nizkoenergijskimi in klasičnimi hišami« (mentorica: doc. dr. Matjaž Ramšak, somentor: prof. dr. Aleš Hribernik);

SI 12

Nejc VRABIČ z naslovom: »Uporaba polimernih kompozitov pri izdelavi orodij« (mentorica: prof. dr. Ivan Anžel, somentor: izr. prof. dr. Ivan Pahole); * Na Fakulteti za strojništvo Univerze v Mariboru je pridobil naziv diplomirani inženir strojništva (VS): dne 20. decembra 2012: Martin FIJAVŽ z naslovom: »Izvedba eksoskeletona spodnjih okončin« (mentor: doc. dr. Aleš Belšak, somentor: izr. prof. dr. Miran Ulbin).

Strojniški vestnik – Journal of Mechanical Engineering (SV-JME) Aim and Scope The international journal publishes original and (mini)review articles covering the concepts of materials science, mechanics, kinematics, thermodynamics, energy and environment, mechatronics and robotics, fluid mechanics, tribology, cybernetics, industrial engineering and structural analysis. The journal follows new trends and progress proven practice in the mechanical engineering and also in the closely related sciences as are electrical, civil and process engineering, medicine, microbiology, ecology, agriculture, transport systems, aviation, and others, thus creating a unique forum for interdisciplinary or multidisciplinary dialogue. The international conferences selected papers are welcome for publishing as a special issue of SV-JME with invited co-editor(s). Editor in Chief Vincenc Butala University of Ljubljana Faculty of Mechanical Engineering, Slovenia Technical Editor Pika Škraba University of Ljubljana Faculty of Mechanical Engineering, Slovenia Editorial Office University of Ljubljana (UL) Faculty of Mechanical Engineering SV-JME Aškerčeva 6, SI-1000 Ljubljana, Slovenia Phone: 386-(0)1-4771 137 Fax: 386-(0)1-2518 567 E-mail: info@sv-jme.eu, http://www.sv-jme.eu Print DZS, printed in 450 copies Founders and Publishers University of Ljubljana (UL) Faculty of Mechanical Engineering, Slovenia University of Maribor (UM) Faculty of Mechanical Engineering, Slovenia Association of Mechanical Engineers of Slovenia

http://www.sv-jme.eu

59 (2013) 1

Chamber of Commerce and Industry of Slovenia Metal Processing Industry Association

Since 1955

Strojniški vestnik Journal of Mechanical Engineering

rd Govekar, Karl Kuzman: sed Model for Supporting the Expert Driven Project in Mold Manufacturing

Mark King, Wuwei Chen, Changjun Wang, Weihua Zhang: t and Dynamic Load-Sharing Analysis of ected Air Suspensions

omar, Nenad Vulić: ls Selection for Environmentally Friendly Ship

ctive d31 Values for PZT from the Nonlinear Oscillations ed Micro- Resonators

atej Gašperin, Boštjan Pregelj, Maja Atanasijević-Kunc, dimir Jovan: ng and Drying inside a PEM Fuel Cell Stack

Journal of Mechanical Engineering - Strojniški vestnik

Diaci: rtainty Assessment in Remote Object Geolocation

asz Trzepieciński: erimental Study of Frictional Behavior in Bending

year

no. 1 2013 59

Image Courtesy: iMold d.o.o., Slovenia and University of Ljubljana, Faculty of Mechanical Engineering, Slovenia

volume

http://www.sv-jme.eu

59 (2013) 1

Strojniški vestnik Journal of Mechanical Engineering

Since 1955

Papers

Blaž Florjanič, Edvard Govekar, Karl Kuzman: Neural Network-Based Model for Supporting the Expert Driven Project Estimation Process in Mold Manufacturing

Yikai Chen, Jie He, Mark King, Wuwei Chen, Changjun Wang, Weihua Zhang: Model Development and Dynamic Load-Sharing Analysis of Longitudinal-Connected Air Suspensions

Liane Roldo, Ivan Komar, Nenad Vulić: Design and Materials Selection for Environmentally Friendly Ship Propulsion System

Lovro Kuščer, Janez Diaci: Measurement Uncertainty Assessment in Remote Object Geolocation

Hirpa G. Lemu, Tomasz Trzepieciński: Numerical and Experimental Study of Frictional Behavior in Bending Under Tension Test

Andrew J. Dick: Characterizing Effective d31 Values for PZT from the Nonlinear Oscillations of Clamped-Clamped Micro- Resonators

Andrej Debenjak, Matej Gašperin, Boštjan Pregelj, Maja Atanasijević-Kunc, Janko Petrovčič, Vladimir Jovan: Detection of Flooding and Drying inside a PEM Fuel Cell Stack

Journal of Mechanical Engineering - Strojniški vestnik

Contents

1 year 2013 volume 59 no.