journal of mechanical engineering 2012 9

Strojniški vestnikJournal of Mechanical Engineering

Since 1955

Contents Papers TomažFinkšt,JurijF.Tasič,MarjetaTerčelj-Zorman,MatejZajc:501 AutofluorescenceBronchoscopyImageProcessingintheSelectedColour Spaces DianaPopescu,CătălinGheorgheAmza,DanLăptoiu,GheorgheAmza:509 CompetitiveHopfieldNeuralNetworkModelforEvaluatingPedicleScrew PlacementAccuracy FrédéricVignat,DinhSonNguyen,DanielBrissaud:517 AMethodtoDeterminetheImpactofGeometricalDeviationsonProduct Performance BingLi,JimengLi,JiyongTan,ZhengjiaHe:527 AdSRBasedFaultDiagnosisforThree-AxisBoringandMillingMachine LidijaRihar,JanezKušar,StaneGorenc,MarkoStarbek:534 TeamworkintheSimultaneousProductRealisation FerhatDikmen,MeralBayraktar,RahmiGuclu:545 RailwayAxleAnalyses:FatigueDamageandLifeAnalysisofRailVehicleAxle IvanaIlić,ZlatkoPetrovic,MirkoMaksimović,SlobodanStupar, DragiStamenković:553 ComputationMethodinFailureAnalysisofMechanicallyFastenedJoints atLayeredComposites

no. 9year 2012volume58Jo

urna

lofM

echa

nicalE

nginee

ring

-Strojniškivestnik

58 (2

012)

9

http://www.sv-jme.eu

Strojniški vestnik – Journal of Mechanical Engineering (SV-JME)

Aim and ScopeThe 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 ChiefVincenc ButalaUniversity of Ljubljana Faculty of Mechanical Engineering, Slovenia

Technical EditorPika ŠkrabaUniversity of Ljubljana Faculty of Mechanical Engineering, Slovenia

Editorial OfficeUniversity of Ljubljana (UL)Faculty of Mechanical EngineeringSV-JMEAškerčeva 6, SI-1000 Ljubljana, SloveniaPhone: 386-(0)1-4771 137Fax: 386-(0)1-2518 567E-mail: [email protected], http://www.sv-jme.eu

PrintTiskarna Knjigoveznica Radovljica, printed in 480 copies

Founders and PublishersUniversity of Ljubljana (UL)Faculty of Mechanical Engineering, Slovenia

University of Maribor (UM)Faculty of Mechanical Engineering, Slovenia

Association of Mechanical Engineers of Slovenia

Chamber of Commerce and Industry of SloveniaMetal Processing Industry Association

International Editorial BoardKoshi Adachi, Graduate School of Engineering,Tohoku University, JapanBikramjit Basu, Indian Institute of Technology, Kanpur, IndiaAnton Bergant, Litostroj Power, Slovenia Franci Čuš, UM, Faculty of Mech. Engineering, SloveniaNarendra B. Dahotre, University of Tennessee, Knoxville, USAMatija Fajdiga, UL, Faculty of Mech. Engineering, SloveniaImre Felde, Bay Zoltan Inst. for Mater. Sci. and Techn., HungaryJože Flašker, UM, Faculty of Mech. Engineering, SloveniaBernard Franković, Faculty of Engineering Rijeka, CroatiaJanez Grum, UL, Faculty of Mech. Engineering, SloveniaImre Horvath, Delft University of Technology, NetherlandsJulius Kaplunov, Brunel University, West London, UKMilan Kljajin, J.J. Strossmayer University of Osijek, CroatiaJanez Kopač, UL, Faculty of Mech. Engineering, SloveniaFranc Kosel, UL, Faculty of Mech. Engineering, SloveniaThomas Lübben, University of Bremen, GermanyJanez Možina, UL, Faculty of Mech. Engineering, SloveniaMiroslav Plančak, University of Novi Sad, SerbiaBrian Prasad, California Institute of Technology, Pasadena, USABernd Sauer, University of Kaiserlautern, GermanyBrane Širok, UL, Faculty of Mech. Engineering, SloveniaLeopold Škerget, UM, Faculty of Mech. Engineering, SloveniaGeorge E. Totten, Portland State University, USANikos C. Tsourveloudis, Technical University of Crete, GreeceToma Udiljak, University of Zagreb, CroatiaArkady Voloshin, Lehigh University, Bethlehem, USA

President of Publishing CouncilJože DuhovnikUL, Faculty of Mechanical Engineering, Slovenia

General informationStrojniški 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 peer-review process.

ISSN 0039-2480

Cover: Medical images that are obtained by the autofluorescence bronchoscopy, are segmented into the potentially cancerous and healthy areas: - Images 1.1 and 2.2: manual segmentation by a medical doctor who is specialized in image reading.- Images 1.2 and 3.1: segmentation by the machine algorithm in the RGB space. - Images 2.1 and 3.2: segmentation by the machine algorithm in the HSV space. Use of the HSV space is preferred in this type of machine diagnostics.Image Courtesy: Laboratory LDSE, Faculty of Mechanical Engineering, University of Ljubljana

© 2011 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 AuthorsAll 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/.

Announcement:The authors are kindly invited to submitt the paper through our web

site: http://ojs.sv-jme.eu. 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 MANUSCRIPTThe 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.

REFERENCESA 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 cross-linked 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 ABSTRACTBy 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/.

COPYRIGHTAuthors 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.sv-jme.eu/.

PUBLICATION FEEFor 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 220.00 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.


Since 1955



urna

lofM

echa

nicalE

nginee

ring

-Strojniškivestnik

58 (2

012)

9


http://en.sv-jme.eu/data/upload/2011/07_08/01_2011_013_Hackenschmidt_04.pdf

Strojniški vestnik - Journal of Mechanical Engineering 58(2012)9Contents

Contents

Strojniški vestnik - Journal of Mechanical Engineeringvolume 58, (2012), number 9Ljubljana, September 2012

ISSN 0039-2480

Published monthly

Papers

Tomaž Finkšt, Jurij F. Tasič, Marjeta Terčelj-Zorman, Matej Zajc: Autofluorescence Bronchoscopy Image Processing in the Selected Colour Spaces 501

Diana Popescu, Cătălin Gheorghe Amza, Dan Lăptoiu, Gheorghe Amza: Competitive Hopfield Neural Network Model for Evaluating Pedicle Screw Placement Accuracy 509

Frédéric Vignat, Dinh Son Nguyen, Daniel Brissaud: A Method to Determine the Impact of Geometrical Deviations on Product Performance 517

Bing Li, Jimeng Li, Jiyong Tan, Zhengjia He: AdSR Based Fault Diagnosis for Three-Axis Boring and Milling Machine 527

Lidija Rihar, Janez Kušar, Stane Gorenc, Marko Starbek: Teamwork in the Simultaneous Product Realisation 534

Ferhat Dikmen, Meral Bayraktar, Rahmi Guclu: Railway Axle Analyses: Fatigue Damage and Life Analysis of Rail Vehicle Axle 545

Ivana Ilić, Zlatko Petrovic, Mirko Maksimović, Slobodan Stupar, Dragi Stamenković: Computation Method in Failure Analysis of Mechanically Fastened Joints at Layered Composites 553

*Corr. Author’s Address: University of Ljubljana, Faculty of Mechanical Engineering, Aškerčeva 6, 1000 Ljubljana, Slovenia, [email protected] 501

Strojniški vestnik - Journal of Mechanical Engineering 58(2012)9, 501-508 Paper received: 2012-02-10, paper accepted: 2012-06-07DOI:10.5545/sv-jme.2012.350 © 2012 Journal of Mechanical Engineering. All rights reserved.

0 INTRODUCTION

Reading medical images is one of the key diagnostic methods of modern medicine. The information deduced from these images is essential for the detection and understanding of pathological conditions. Automatic extraction, analysis and understanding of the visual content consist of many image processing operations. The most used are image acquisition, segmentation, compression, registration and quantitative analyses [1]. Enduring research in medical image processing is important for the improvement of existing analyses of medical images, which leads to better medical care.

Imaging revolutionized medicine. The imaging revolution started with the invention of X-rays. Ultrasound diagnostics followed. Progress in computer technologies provided the means for the invention of computerised tomography (CT). Research in physics resulted in magnetic resonance imaging (MRI) [2]. Auto fluorescence bronchoscopy (AFB) completes the set of modern non-invasive technologies for medical imaging.

The information content of colour medical images is richer than the information content of black and white images. Efficient image analysis implies the use of colour images whenever applicable (e.g., X-ray imaging produces only different levels of grey).

The information of a colour image resides in a specific colour space (e.g., red-green-blue (RGB) or hue-saturation-value (HSV)). Colour features of the image are distinguished by their sensitivity to the scene illumination and spectral distribution of light wavelengths and by human perception [3].

Conversions between different colour spaces are either linear or nonlinear. Spaces obtained by

nonlinear conversion can cause problems in pattern definition since colours in some areas easily have luminance and/or colour saturation that is too low in order for an effective readout to be possible. As a result, colour intensity can be used in pattern recognition processing [3].

The most important step in the segmentation of colour images is the definition of criteria and methods. Segmentation of colour images can be based on the analysis of pixel space features, homogeneity of adjacent pixels and physical properties of surfaces making the image and on edge detection between areas of different colours [3].

0.1 Auto Fluorescence Bronchoscopy Imaging

Bronchoscopy is one of the oldest methods of investigation in respiratory medicine [4]. A medical expert uses the method to determine the extent of pathological tissue changes. When the area is properly visualized and the consequential biopsy is carried out skillfully, it is usually possible to accurately state the diagnosis of the disease.

Using white light instead of blue makes detection of the earliest changes in bronchial mucosa more difficult, if not impossible (Fig. 1a). To improve the sensitivity of bronchoscopy performed with white-light illumination, the AFB [5] has been introduced. AFB relies on the fact that the deteriorated bronchial mucosa fluoresces less than the healthy mucosa when irradiated by the wavelength of about 440 nm. A helium-cadmium laser produces a blue light with the wavelength of 442 nm. Consequentially, it is used as a source of illumination in the AFB [6] (Fig. 1b).

Autofluorescence Bronchoscopy Image Processing in the Selected Colour Spaces

Finkšt, T. – Tasič, J.F. – Terčelj-Zorman, M. – Zajc, M.Tomaž Finkšt1,* – Jurij F. Tasič2 – Marjeta Terčelj-Zorman3 – Matej Zajc2

1 University of Ljubljana, Faculty of Mechanical Engineering, Slovenia 2 University of Ljubljana, Faculty of Electrical Engineering, Slovenia

3 University Medical Centre, Department of Pulmonary Diseases and Allergy, Slovenia

Reading diagnostic medical images usually requires the expertise of a specialist physician. To aid physicians we have developed an algorithm that deduces medical information by analysing colour nuances of an image obtained by bronchoscopy. The goal is to ensure a high probability of detecting bronchial cancer. Autofluorescent bronchoscopy images are analysed by the proposed algorithm. The machine-made diagnoses of early cancer stages are highly correlated with the diagnoses made by a medical expert. Reading the image using a specialized apparatus and producing a pre-diagnosis by image-recognition software and a special set of rules has the potential to produce automated second opinions for most cases of the disease. Keywords: colour spaces, image processing, image acquisition, image segmentation, edge detection, autofluorescence bronchoscopy

Strojniški vestnik - Journal of Mechanical Engineering 58(2012)9, 501-508

502 Finkšt, T. – Tasič, J.F. – Terčelj-Zorman, M. – Zajc, M.

There are several studies in the literature aiming at improving AFB sensitivity (sensitivity = number of true positives / (number of true positives + number of false negatives)) and specificity (specificity = number of true negatives / (number of true negatives + number of false positives)). To improve the diagnostics, the existing AFB techniques can be combined with other methods. Kusunoki et al. [6], Bard et al. [7], Zeng et al. [8] and Terčelj et al. [9] describe how the optical spectroscopy can add the information context to the AFB image-acquisition system. However, this approach involves the use of a spectrometer and prolongs the bronchoscopy procedure. Another approach used to improve the specificity of the AFB technique is presented by Goujon et al. [10] and Qu et al. [11] who employed off-line spectral image analysis and threshold algorithms to classify positive and false results, but with limited success.

0.2 Machine Supported AFB for a Teleconsultation

The diagnostic system that we use for image acquisition has been installed in a hospital specializing in the bronchoscopy examination. Our goal is to develop an image analysis system that can be used for tele-consultation by specialists [12].

In order to identify deteriorated, i.e., potentially cancerous areas of the bronchial mucosa, we use the image-processing method that finds edges of changed surfaces in the bronchoscopy image. Such surfaces can be quite small at the onset of the disease. The initial sensitivity set up for edge recognition is set by the operator. The algorithm then autonomously searches for the edges that define deteriorated surfaces, i.e., image segments. Poor initial setting can result in a significant error since the image segmentation algorithm involves iterative processing where inefficiencies are adding up.

In this paper we compare the impact of two colour spaces on the segmentation of the autofluorescence images. We evaluate the RGB and the HSV colour space [3]. We expect to get better results with the HSV space since it is closer to the human perception of colours than the RGB space that fits better to the properties of the imaging system hardware.

1 ACQUISITION OF THE AUTOFLUORESCENCE IMAGE

The set up for AFB [13] image acquisition is presented in Fig. 1. A combination of the selected light source, image acquisition apparatus and image display hardware characterize a particular AFB.

a)

Xsenonlamp

Lightsource

RGB CCD camera

Image acquisition

Workstation monitor

Image display

b)

Helium-cadmium laser

Lightsource

Fluorescence CCD camera

Image acquisition

Image display

Workstation monitor

Fig. 1. Image-acquisition process in the bronchoscopy;a) white light, b) blue light and the autofluorescence mode

1.1 The LIFE System

The AFB images that we investigate were acquired by the light imaging fluorescence endoscope (LIFE), produced by Xillix Technology, Vancouver, with the resolution of 800×600 pixels for both white-light and autofluorescence mode.

The LIFE system consists of: • Dual wavelength (red and green) camera with a

built-in amplifier.• White-light colour RGB camera.• Blue laser light source.• White-light source.• Switching mechanism to select a white or blue

light imaging mode.• Imaging workstation, the suite of the AFB

imaging applications and a computer monitor.Both images (white-light and autofluorescence)

are digitized simultaneously in real-time (i. e., at video rates of 30 frames/s) by the imaging board.

In the white-light imaging mode, an RGB camera is used to capture the reflected light. This image is digitized and displayed on the monitor. In both the fluorescence and white-light modes, the digital representation of the image is stored in the computer by taking a snapshot of the image.

In the autofluorescence imaging mode, a selected wavelength of the blue light (at 442 nm) is employed to excite the bronchial tissue and cause it to fluoresce. The light is delivered through a fibre-optic illumination channel of a bronchoscope. The reflected fluorescent light is captured by an image-intensifying CCD camera via a fibre-optic imaging bundle. Two spectral wavelength images are captured by the CCD camera: one in the green spectrum (480 to 520 nm) and the other in the red spectrum (more than 630 nm). A mathematical transformation of the red and green colour intensities produces a pseudo (false colour) video image for real time viewing: the normal tissue


503Autofluorescence Bronchoscopy Image Processing in the Selected Colour Spaces

appears green and the abnormal areas appear reddish-brown.

2 IMAGE PROCESSING FOR EDGE DETECTION

This section presents the algorithm, which defines the edges in the AFB image. To ensure the universality of the algorithm no assumption is made about the shapes of objects at the algorithm initialization, i.e., when setting up its’ segmentation stage.

The most common application in the field of digital image processing is the one which searches for specific shapes and patterns. The steps of this process are: image acquisition, removal of noise, image segmentation and image classification.

Different digital filters are used for noise removal. The result is an improved image histogram, which is needed for subsequent image processing.

The result of the image segmentation phase is image structuring, i.e., segmentation of the image into patterns, i.e., segments. Each segment is described by a set of features (centre of gravity, area, colour histogram and others).

The segmentation algorithm can be relatively complex since many threads can run simultaneously, each for its own region of the image. The patterns are correlated to the object of the model, which encapsulates the search rules and data on different search patterns. The result of the classification phase is confirmation or rejection of the search object presence.

2.1 Image Pre-Processing

The objective of the pre-processing phase is to properly prepare the image for further analysis. In the proposed approach, images are first adjusted by normalization of their colour histograms. Then, the image contrast is manipulated since the changes in the AFB images appear as nuances of brown-reddish colour.

The colour image is then transformed to a grey scale image for further processing [14]. Illumination of bronchia tissue with a blue light of 442 nm wavelength results in reflected light having wavelengths in red and green spectrum. It is the bio-chemical processes in the epitelium of the bronchia tissue that produce red and green reflection when illuminated with a blue light [6]. The level of grey in the grey image is defined as a quotient between intensities of red and green colours [5]. The gray image is analysed for groups of pixels with the same level of grey. The result of this phase is a histogram that presents sums of pixels for individual

levels of grey. We analyse the obtained histogram of grey level frequencies in each image.

Individual histograms are produced for each of the parameters that make the particular colour space (RGB and HSV). A colour filter is applied to produce an image with different intensities of the filtered parameter only. The resulting image is processed further by the same procedure as the grey scale image. The results of this phase are intensity histograms for each parameter of the selected colour space.

2.2 The Selected Colour Spaces

RGB and HSV colour spaces are used to study bronchoscopy and microscopy images [15]. Attributes of both colour spaces meet the requirements of the study. We processed the bronchoscopy images in both colour spaces.

2.2.1 Properties of the RGB Colour Space

RGB colour space is a three dimensional space. Any point in this space is presented by the three linearly independent vectors that make the Cartesian coordinates, namely red (R), green (G) and blue (B). Fig. 2a presents the space and position of some colours in the space. Any colour is presented as a vector defined by a triple of the RGB intensities.

RGB intensities are calculated by the Eq. (1) to the Eq. (3) for each of the colours. Wavelengths are integrated from 300 to 830 nm [4]. S(λ) represents the illumination light. R(λ), G(λ) and B(λ) are RGB components of the light that reflects from the object. The three components of the reflected light are captured by sensors that are sensitive to each of the RGB colours only [2].

R S R d= ∫ ( ) ( ) ,λ λ λ300

830 (1)

G S G d= ∫ ( ) ( ) ,λ λ λ300

830 (2)

B S B d= ∫ ( ) ( ) .λ λ λ300

830 (3)

Transformation from the spectral power distribution to a three-dimensional vector is a powerful compression technique. Elaborations on the compression details are given in [16]. A high compression ratio, as 10:1, or higher can be achieved by the transformation [16]. A side effect is the loss of information which introduces metamers, i.e., colours with the same R, G, and B values but with different



spectra. Such colours are perceived as a single colour since the human eye contains only three types of cone cells, i.e., colour receptors, which means that all colours are reduced to the three sensory quantities [3].

The RGB values depend on the spectrum of the light source and on the sensitivity of the capturing device. Differences in light sources and/or capturing devices can result in different triplets of values for the colour of the examined material. The RGB colour space is device dependent.

The use of the RGB space is convenient since images can be taken from capturing devices without transformation (RGB is the native space in the majority of the image oriented devices). Consequently, the RGB space is extensively used in current imaging systems; it is wide-spread in the image processing industry.

Hue

Value

B

G

R

Saturation

Magenta White

CyanBlue

BlackGreen

YellowRed

a)

b)

(1,0,0)

R

Graysca

le

(0,1,0)G

(0,0,1)

B

Fig. 2. Colour spaces: a) an RGB colour space where vertices of the cube represent the primary colours, b) an HSV colour space in

the cylindrical coordinates

Reservations against the use of the RGB space are:• Components of the space are highly correlated

when processing images with natural colours [3].• The correlation between the perceived difference

of two colours and their distance in the Euclidian RGB space is low.

• In order to be identified, the objects have to differ from the background to a higher extent than in some other colour spaces. Consequently, additional information is needed for effective processing.

• Changes of brightness and/or hue need a relatively high amount of calculations.

• Image processing requires significant amounts of computer memory since 256 levels of colour intensity result in 24 bits per pixel for the three colours of the space.

2.2.2 Properties of the HSV Colour Space

The most basic constraint in image processing is intuitive work with colours. For this purpose, properties of the colour space are to match human perception of colours to a high extent.

HSV, hue-saturation-intensity (HSI) and hue-lightness-saturation (HLS) colour spaces were invented in order to achieve a high correlation between the attributes of the colour space and the human perception of colours. The common denominator of these spaces is the inclusion of the hue (H) and saturation (S) attributes [3].

The H attribute is about the perceived colour (e.g., blue, red); the S attribute distinguishes a vivid from a pale colour.

Value (V), intensity (I), and lightness (L) attributes are about brightness of the colour. A brighter colour has higher V, I, or L value.

The HSV colour space has a shape of a single cone; HLS and HSI are encapsulated into a shape of a double cone. The difference is due to the different definitions of brightness in the three colour spaces (Fig. 2b).

Human perception of hue is more subtle than perception of colour saturation and brightness [5]. Consequently, usually 6 bits are used for the H value; S and V values are defined with 5 bits each. Corresponding H, S, and V resolutions have 64, 32, and 32 steps. Only 16 bits (6+5+5) are needed for the pixel definition.

HSV, HLS and HIS are perceptually non uniform colour spaces. Consequently, these spaces or the RGB space are not used for the calculation of distances among different colours.

The use of the HSV space is convenient since less computing and less computer memory is involved in colour manipulation than in the RGB space; colour manipulation in the HSV space feels most interactive (even small changes in H, S, and V values are perceived as a colour change). Image saturation and



brightness are manipulated most easily since they are intrinsic parameters of the HSV space. Consequently, the HSV space is very useful in the image processing industry. Even though the HSV space was introduced after the invention of the RGB space it uses only 16 bits per pixel compared to 24 bits in the RGB space.

Computational effectiveness and small memory footprint make the HSV space the optimal choice when designing embedded systems for imaging applications. The use of embedded systems results in high processing speed, compared to running imaging programs on workstations, but limitations of the resources that build an embedded system [17] and [18] need to be dealt with.

Reservations against the use of the HSV space are:• Conversions are needed since most image

oriented devices are designed for the RGB space.• The HSV space is perceptionally non uniform.• Sensitivity for non-saturated colours (close to

white and black) is perceptionally marginal.• The hue parameter has singularity at H = 0 and

H= 255.

2.3 Segmentation

The images have been captured by the LIFE system in the RGB colour space. We mapped them to the HSV space. We designed the image segmentation process in line with the work of Cheng and Sun [19], who obtained the data for segmentation from study of the image histogram.

The steps of our segmentation process for images having colours in the RGB or HSV colour space are:

First, a Standard Deviation of colours in the vicinity of the individual Pixel (SDPij) is calculated by Eq. (4) [19].

SDPd

gij pq ij

q jd

jd

p id

id

= −= −

−

+−

= −−

+−

∑∑12

2

1

2

1

2

1

2

1

2

( ) ,µ (4)

where d is distance of interest from the pixel (i,j), gpq is the colour intensity attribute of a pixel Ppq at the location (p,q). μij is the mean of the grey levels within rectangular window around the pixel (i,j). μij is calculated by Eq. (5):

µij pq

q j d

j d

p i d

i d

dg=

= −−

+−

= −−

+−

∑∑12

12

12

12

12

. (5)

Same as in Eg. (4), gpq represents the colour intensity attribute of a pixel Ppq at the location (p,q). Based on the SDP the image is categorized into homogenous and non-homogeneous areas.

Secondly, colour histograms are calculated for each of the areas, defined in the first step. Based on the histogram data and on the pre-set threshold values, the areas are further divided into the sub areas.

Thirdly, a custom gauge on colour matching is used to merge the sub areas that meet the rule. The Euclidian distance between colours is a measure of the colour match.

Colour variability of the adjacent pixels is calculated with edge detecting operators. The Canny operator [14] is applied in the algorithm.

3 RESULTS

A medical doctor, a specialist in reading the bronchoscopy images, investigated 30 AFB images. He found no pathological changes in 5 images; in 25 images there were pathological changes to bronchial mucosa. The specialist marked the areas with changes.

Fig. 4 shows the manually segmented images: a), d), g), and j); images segmented in the RGB space: b), e), h), and k); and images segmented in the HSV space: c), f), i), and l).

Image acquisition in RGB

Equalization of histograms

Mapping RGB -> HSV

SDP

Edge detection

Fig. 3. Segmentation algorithm in the RGB and HSV spaces

A visual match of pathologic areas, which are outlined by the specialist and by the image processing algorithm, is the most basic requirement in the



evaluation of the algorithm efficiency. Additional criteria need to be applied to underpin the evaluation of the algorithm [20]:a) False positive (FP) is the percentage of pixels that

are defined as pathological by the algorithm and not by the specialist.

b) False negative (FN) is the percentage of pixels that are defined as pathological by the specialist and not by the algorithm.

c) The relative-agreement index (Aagr) with a value between 0 and 1, defined by Eq. (6):

AN

N NagrA A

A A

man alg

man alg

=+∩2 . (6)

d) The relative-disagreement index (Adis) with a value between 0 and 1, defined by Eg. (7):

AN N

N NdisA A A A

A A

man alg man alg

man alg

=−

+∪ ∩2 , (7)

where Aman is the area defined by the specialist, and Aalg is the area defined by the algorithm, NA Aman alg∩ is the number of pixels in the cross-section of the two images and NA Aman alg∪ is the number of pixels in the union of the two images. Eq. (6) and (7) are in line with the work that is presented in [20]. Matching criteria (machine segmentation vs.

specialists’ segmentation) for the 30 AFB images are presented in Table 1. Based on the numbers, the quality of the matching can be judged as suboptimal. For better evaluation of the matching between human and machine reading, the variation in human detection of pathological areas needs assessment.

Fig. 4. Segmented images; a), d), g), and j) manually segmented images, b), e), h), and k) segmented in the RGB space and images c), f), i), and l) segmented in the HSV space



Table 1. Numerical results for segmented images in Fig. 4 in the RGB and HSV spaces

Segmented Images in Fig. 4

FP [%]

FN [%]

Aagr AdisColour Space

b) 6.6 9.3 0.85 0.25 RGBc) 6.9 11.8 0.82 0.25 HSVe) 0 0 1.00 0 RGBf) 0 0 1.00 0 HSVh) 3.7 5.9 0.93 0.18 RGBi) 3.2 5.1 0.95 0.15 HSVk) 5.7 8.4 0.88 0.20 RGBl) 5.9 7.3 0.89 0.22 HSV

The specialists’ segmentation is understood as the most correct. His work represents the norm for the evaluation of the segmentation algorithm. It takes years of specialization for a medical doctor to gain expert knowledge of a competent subject. Some variation in human detection of pathological areas can still exist since it is the fine grain of colour nuances that make the borderline difference between healthy and cancerous tissue.

The match between the experts’ segmentation and the segmentation obtained by the algorithm in the RGB colour space, is 84%. The corresponding match for the HSV colour space is 88%. The 4 percentage points higher level of matching in the HSV space is significant in medical diagnostics. Segmentation in the HSV space has an advantage over segmentation in the RGB space.

For autonomous use, the segmentation algorithm will need to pass meticulously selected verification requirements since there is no plan for the specialists’ segmentation being available in advance.

4 DISCUSSION

Studying the images in Fig. 4, one finds small areas that were not identified as morphological even though they significantly differ in colour from their surroundings. Since these areas are small enough they do not significantly influence the overall detection efficiency. These areas can be joined to the morphological or to background segments without harm. It is larger morphological segments that are significant for the assessment of matching to the experts’ definition.

The algorithmic segmentation is a hierarchical process that proceeds through sequential phases with well-defined interfaces among them, where decisions are made. The experts’ control in real time could add to the quality of assessment on pathological changes.

In the development phase, statistical data about the experts’ interaction with the system would give further insight into the diagnostic process. Such data would be most useful for the algorithm enhancements.

Reading, i.e., the analysis of a medical image is a complex process. It should not be overlooked that it takes years of training for a medical doctor to become an expert in imaging diagnostics. The implication is that relying only on context-independent segmentation methods cannot be the most comprehensive strategy for the development of a machine-based diagnostic system. Additional expert knowledge is needed in image pre-processing and in the segmentation phase.

5 CONCLUSIONS

The developed algorithm for the image segmentation was evaluated in the RGB and HSV colour spaces. The latter proved as a better option for the AFB image segmentation. Matching of context-independent segmentation to the experts’ work is remarkably high. The addition of built-in expert knowledge has the potential to bring the product to a level where it can be used autonomously for the image pre-assessment.

Alternatively, the product can be used as an interactive helper to the physician. In both cases, the doctors could spend less time reading the images and have more time for other work. Additional machine-based image evaluation adds to diagnose certainty.

As it could be expected, colour images are more suitable for segmentation than the grayscale images. Interestingly, working with the hue channel only can result in diagnostically useful image segmentation. Low intensity makes segmentation unstable, i.e., not working.

The development of a context-independent algorithm is the first stage in the development of a machine-based expert system. The conclusion of this phase is required before the development on more intelligent algorithms can begin.

A literature search shows that the authors of published segmentation algorithms only rarely compare the efficiency of newly developed algorithms to the existing ones. A thorough elaboration on the implementation complexity is more of an exception than a rule. Images are analysed in different colour spaces. None of the spaces has a far-reaching advantage for processing over the other colour spaces.

The evaluation results of the context-independent segmentation algorithm are a sound base for further work. More AFB images are needed; more specialists have to get involved into image analysis. Each AFB image needs to be analysed by more specialists;



a reference segmented image needs to be defined from their work. Biopsy results need to be taken into account in reading the reference images.

At this point the working of the algorithm has proved to be effective enough so that additional efforts can bring the product to a level which is needed for daily use in a working environment. There, the product works in a specific context, additional expert knowledge can be added, real time input of the operator is possible at all times – efficiency outperforms the one of the context independent, i.e., generic image segmentation product.

6 REFERENCES

[1] Bračun, D., Perdan, B., Diaci, J. (2011). Surface Defect Detection on Power Transmission Belts Using Laser Profilometry. Strojniski vestnik - Journal of Mechanical Engineering, vol. 57, no. 3, p. 257-266, DOI:10.5545/sv-jme.2010.176.

[2] Suetens, P. (2009). Fundamentals of Medical Imaging. Cambridge University Press, New York, DOI:10.1017/CBO9780511596803.

[3] Koschan, A., Abidi, M. (2008). Digital Color Image Processing. John Wiley & Sons, Hoboken, DOI:10.1002/9780470230367.

[4] Wang, K.P., Mehta, A.C., Turner, J.F. (2004). Flexible Bronchoscopy. Wiley-Blackwell, Massachusetts.

[5] Chen, W., Gao, X., Tian, Q., Chen, L. (2011). A comparison of autofluorescence bronchoscopy and white light bronchoscopy in detection of lung cancer and preneoplastic lesions: A meta-analysis. Lung Cancer, vol. 73, no. 2, p. 183-188, DOI:10.1016/j.lungcan.2010.12.002.

[6] Kusunoki, Y., Imamura, F., Uda, H., Mano, M., Horai, T. (2000). Early detection of lung cancer with laser-induced fluorescence endoscopy and spectrofluorometry. Chest, vol. 118, no. 6, p. 1776-1782, DOI:10.1378/chest.118.6.1776.

[7] Bard, M.P.L., Amelink, A., Skurichina, M., Den Bakker, M., Burgers, S.A., Van Meerbeeck, J.P., Duin, R.P.W., Aerts, J.G.J.V., Hoogsteden, H.C., Sterenborg, H.J.C.M. (2005). Improving the specificity of fluorescence bronchoscopy for the analysis of neoplastic lesions of the bronchial tree by combination with optical spectroscopy: preliminary communication. Lung Cancer, vol. 47, no. 1, p. 41-47, DOI:10.1016/j.lungcan.2004.06.009.

[8] Zeng, H., Petek, M., Zorman, M.T., McWilliams, A., Palcic, B., Lam, S. (2004). Integrated endoscopy system for simultaneous imaging and spectroscopy for early lung cancer detection. Optics Letters, vol. 29, no. 6, p. 587-589, DOI:10.1364/OL.29.000587.

[9] Tercelj, M., Zeng, H., Petek, M., Rott, T., Palcic, B. (2005). Acquisition of fluorescence and reflectance spectra during routine bronchoscopy examinations

using the ClearVu Elite device: pilot study. Lung Cancer, vol. 50, no. 1, p. 35-42, DOI:10.1016/j.lungcan.2005.05.028.

[10] Goujon, D., Zellweger, M., Radu, A., Grosjean, P., Weber, B.C., Van Den Bergh, H., Monnier, P., Wagnières, G. (2003). In vivo autofluorescence imaging of early cancers in the human tracheobronchial tree with a spectrally optimized system. Journal of Biomedical Optics, vol. 8, no. 1, p. 17-25, DOI:10.1117/1.1528594.

[11] Qu, J.Y., Chang, H., Xiong, S. (2002). Fluorescence spectral imaging for characterization of tissue based on multivariate statistical analysis. Journal of the Optical Society of America, vol. 19, no. 9, p. 1823-1831, DOI:10.1364/JOSAA.19.001823.

[12] Meža, M., Breskvar, M., Košir, A., Bricl, I., Tasič, J., Rozman, P. (2007). Telemedicine in the blood transfusion laboratory: Remote interpretation of pre-transfusion tests. Journal of Telemedicine and Telecare, vol. 13, no. 7, p. 357-362, DOI:10.1258/135763307782215370.

[13] Takehana, S., Kaneko, M., Mizuno, H. (1999). Endoscopic diagnostic system using autofluorescence. Diagnostic and Therapeutic Endoscopy, vol. 5, no. 2, p. 59-63, DOI:10.1155/DTE.5.59.

[14] Russ, J.C. (2011). The Image Processing Handbook. CRC Press, Boca Raton.

[15] Bountris, P., Apostolou, A., Haritou, M., Passalidou, E., Koutsouris, D. (2009). Combined texture features for improved classification of suspicious areas in autofluorescence bronchoscopy. Proceedings of the 9th

International Conference on Information Technology and Applications in Biomedicine, ITAB, Larnaca, p. 1-4.

[16] Poynton, C. (1995). A guided tour of color space. Physics Today, vol. 2, p. 1-14.

[17] Jenko, M. (2010). Ratiometric measurement for long term precision, reasoning and case study. Information MIDEM, vol. 40, no. 2, p. 124-130.

[18] Jenko, M., Medjeral, N., Butala, P. (2001). Component-based software as a framework for concurrent design of programs and platforms. Microprocessors and Microsystems, vol. 25, no. 6, p. 287-296, DOI:10.1016/S0141-9331(01)00120-X.

[19] Cheng, H.D., Sun, Y. (2000). A hierarchical approach to color image segmentation using homogeneity. IEEE Transactions on Image Processing, vol. 9, no. 12, p. 2071-2082, DOI:10.1109/83.887975.

[20] Klemenčič, J. (2000). Segmentation and volumetry of hippocampus from magnetic resonance images using active curves. MSc. Thesis, University of Ljubljana, Ljubljana.

[21] Garnavi, R., Aldeen, M., Celebi, M.E., Bhuiyan, A., Dolianitis, C., Varigos, G. (2010). Automatic segmentation of dermoscopy images using histogram thresholding on optimal color channels. International Journal of Medicine and Medical Sciences, vol. 1, no. 2, p. 126-134.

http://en.sv-jme.eu/scripts/download.php?file=/data/upload/2011/03/09_2010_176_Bracun_03.pdf

http://en.sv-jme.eu/scripts/download.php?file=/data/upload/2011/03/09_2010_176_Bracun_03.pdf

http://dx.doi.org/10.1017/CBO9780511596803

http://dx.doi.org/10.1017/CBO9780511596803

http://dx.doi.org/10.1002/9780470230367

http://dx.doi.org/10.1016/j.lungcan.2010.12.002


http://dx.doi.org/10.1378/chest.118.6.1776



http://dx.doi.org/10.1364/OL.29.000587



http://dx.doi.org/10.1117/1.1528594

http://dx.doi.org/10.1364/JOSAA.19.001823

http://dx.doi.org/10.1258/135763307782215370

http://dx.doi.org/10.1155/DTE.5.59

http://dx.doi.org/10.1016/S0141-9331(01)00120-X

http://dx.doi.org/10.1016/S0141-9331(01)00120-X

http://dx.doi.org/10.1109/83.887975

*Corr. Author’s Address: University Politehnica of Bucharest, Splaiul Independentei, 313, sector 6, 060042, Bucharest, Romania, [email protected] 509


0 INTRODUCTION

Posterior stabilization of spine is required in different types of pathologies and can be accomplished by inserting screws in vertebral pedicles and connecting them with rods in order to determine a long-term biological fusion by holding together different segments of the spine. The pedicle screw insertion procedure poses specific problems to spine surgeons due to the vicinity of nerve roots, the particular shape of the pedicle and its patient-dependent dimensions and spatial orientation, and requires preoperative studies based on medical imaging for choosing the proper screw type, diameter and length.

Clinical experience and biomedical studies [1] to [3] recommend the screw to be inserted along the pedicle axis starting from an entry point (Fig. 1) established by surgeon considering the bone quality and pedicle anatomy and orientation. Usually, in the free hand techniques, the entry point for pedicle screw is visually chosen by the surgeon and the insertion trajectory is determined by palpation and by mentally reconstructing the position of the vertebra according to the CT/MRI information gathered during the planning stage.

However, despite the use, during implantation, of advanced medical imagistic techniques, such as C-arm fluoroscopy, literature [2] and [4] reports screws misplacements up to 8 to 13%, hence the importance of developing training systems for surgeons. Such a training system is described in Fig. 2 and it is based on the use of a Competitive Hopfield Neural Network (CHNN) algorithm for performing X-ray image segmentation required for the automated assessment of the pedicle screw insertion precision.

The development of this training system and its interface is part of an on-going interdisciplinary research project of the authors and gathers expertise from different fields such as mechanical engineering, image processing, medical modelling and programming.

According to the method presented in Fig. 2, the surgeon establishes the entry point, chooses pedicle screw diameter and length and inserts the screw in the test vertebra manufactured from polyurethane foam based on different CT/MRI real patient data.

Fig. 1. Anatomy of a lumbar vertebra (L3)

In order to mimic the real surgical conditions as much as possible, the screw is inserted in a vertebra placed in a sand box so that only several landmarks are visible to the trainee. In the next stages, the vertebra is placed on a conveyor and brought into the X-ray inspection unit, and images of the vertebra and screw, in two planes, are acquired using a standard protocol. Image segmentation is performed using a CHNN based algorithm, the vertebra and the screw are extracted as individual objects and calculations are

Competitive Hopfield Neural Network Model for Evaluating Pedicle Screw Placement Accuracy

Popescu, D. – Amza, C. Gh. – Lăptoiu, D. – Amza, Gh.Diana Popescu1,* – Cătălin Gheorghe Amza1 – Dan Lăptoiu2 – Gheorghe Amza1

1 University Politehnica of Bucharest, Romania2 Colentina Clinical Hospital, Romania

In this paper, the application of an X-ray image segmentation algorithm based on a Competitive Hopfield Neural Network (CHNN) model for evaluating the insertion accuracy of pedicle screws is presented.

In practice, the evaluation of pedicle screw insertion accuracy is made visually in two planes and is based on postoperative computer tomography scans or radiography. In order to increase the reliability of the assessment, this research proposes a new approach that automates this process and can be used for developing a training system for pedicle screw implantation. The proposed approach implements a training method which allows extracting features of the pedicle screw from X-ray images segmented using a modified HNN algorithm, and compares them with values from a knowledge database.Keywords: medical imaging, pedicle screw, Hopfield neural network, lumbar vertebra


510 Popescu, D. – Amza, C. Gh. – Lăptoiu, D. – Amza, Gh.

made for determining the deviation of the screw axis from the pedicle axis – considered as ideal trajectory. The value of the deviation is compared to the safety area limits [4] to [6] and evaluation messages are displayed accordingly.

1 FRAMEWORK FOR AN INTELLIGENT X-RAY TRAINING SYSTEM FOR PEDICLE SCREW INSERTION

1.1 X-Ray Images Processing and Analysis

In practice, in most cases, the evaluation of the pedicle screw insertion accuracy is based on postoperative radiography/CT scans took in two planes (transverse and sagittal) and visually analyzed by surgeons, which raises questions regarding the reliability of this assessment method [5]. Therefore, this paper presents a different innovative approach in which the subjectivity is eliminated by using an intelligent segmentation algorithm for X-ray images analysis. Based on this algorithm, dedicated software is developed within a training system for automatically determining the deviation of the screw insertion trajectory from the pedicle axis.

Artificial neural networks have applications not only in many engineering fields [7] to [9], but they are also used in the segmentation of medical images, as presented in [10] to [13] for HNN algorithms. In this sense, a literature survey presented in [14] shows that HNN are mostly used in image reconstruction, feed forward neural networks are used for image segmentation, while back propagation neural networks are used for object recognition. However,

automatic medical images processing and analysis is still a challenging task related mainly to: (i) the high differences between patients’ anatomical structures which localization and shape are hard to be interpreted by computer software, and (ii) the difficulty to find a proper segmentation method for processing images with high noise determined by the human tissue and bones characteristics [15]. Literature reports solutions related to spinal cord automatic detection from CT images [16], virtual endoscopy [17] or mammography [18].

No references about an automated training system based on intelligent X-ray image processing and analysis applied to pedicle screw implantation accuracy are presented in literature, to the best of the authors’ knowledge. Regarding advanced training system for pedicle screw implantation, the literature reports a computer-assisted system based on X-ray [19] and several simulators based on augmented or virtual reality [20] and [21]. The system presented in [19] uses markers-based imaging registration and spatial geometric transformation for guiding the trajectory of a 3D digitizer. A senior surgeon indicates the entry point and the optimal trajectory, and then the system guides the trainee in placing the screw within the safety area.

For the application presented in this paper, the segmentation algorithm has to be manageable from the point of view of computations involved, therefore suitable for real time use, and it has to be independent of the size of the image. First, classical thresholding algorithms used for image segmentation were comparatively analysed, then segmentation algorithms

Digital modelling of a lumbar vertebra obtained from CT/

RMN data

Manufacturing process of a lumbar vertebra

test model

Choosing the entry point and pedicle screw

diameter and length

Drilling the vertebra and inserting the screw

Acquiring X-ray images in transverse and

sagittal planes

Applying a HNN image segmentation algorithm for extracting pedicle

screw attributes

Displaying deviation value and messages of information/evaluation

Fig. 2. Training method for pedicle screw insertion based of intelligent processing of X-ray images


511Competitive Hopfield Neural Network Model for Evaluating Pedicle Screw Placement Accuracy

based on artificial intelligence were applied to the analyzed images and the results were compared. Finally, a competitive HNN algorithm is proposed, developed and implemented in software.

1.2 X-Ray Training System for Screw Insertion in Vertebral Pedicle

In the case of pedicle screw insertion, the evaluation process implies taking an X-ray image of a test vertebra. This image is then subsequently automatically analyzed by a computer algorithm and the insertion precision is measured. The proposed system – an implementation of a general pattern recognition system, comprises the following units/systems (Fig. 3):1. Image acquisition system for the acquisition of a

dual-band energy image of the vertebra;2. Image pre-processing system for enhancing

the X-ray images for intermediate level image processing (contrast enhancement, background removal, noise removal etc.);

3. Image segmentation system for partitioning the X-ray image into meaningful classes for further higher level inspection using a HNN module. This system makes possible to automatically extract the “pedicle screw” as a separate object from the X-ray image;

4. High level detection system – for calculating the deviation from an “ideal” position of a pedicle screw. This module first extracts the segmented “pedicle” and then it compares its position to an ideal position that is incorporated within the system. The object is extracted by using simple back-tracking algorithm (considering an area of same pixel values as resulted from the segmentation process). Then, only for visual

purposes for trainees, the original image can be superimposed over the segmented image.The following stages are part of the general X-ray

image analysis process:1. Acquire the X-ray image/images of the vertebra

test model.2. Low-level image processing of the resulting

image/images: a) X-ray image or images pre-processing –

background removal, contrast enhancement and removal of possible noise;

b) X-ray image segmentation using an algorithm based on Competitive Hopfield Neural Network (CHNN);

c) Pedicle screw extraction for the obtained segmented image using a simple backtracking algorithm.

3. High-level detection of the insertion precision of pedicle screw:

a) Feature extraction for the pedicle screw – position, rotation angle, size, etc.;

b) High-level detection of the deviation from an ideal position of the pedicle screw.Human experience has to be incorporated into

the design of such a training system. This knowledge is gathered into a database which contains general data about the pedicle screw (such as physical and chemical characteristics, types, dimensions), vertebra morphological data for different populations, data about the possible errors that may appear (such as vertebral wall penetrations), safety limits expressed in grades according to classification used in practice and any other information directly or indirectly related to the training process (such as entry points for pedicle screws, type of instrumentations used in practice, surgical approaches, etc.).

Fig. 3. Intelligent X-ray based system for pedicle screw insertion accuracy evaluation



2 X-RAY IMAGE SEGMENTATION ALGORITHM BASED ON A HNN MODEL

2.1 X-Ray Image Segmentation Using Classic and Artificial Intelligence Based Algorithms

Image segmentation subdivides a digital image into multiple continuous, disconnected and nonempty subsets (or segments) with uniform and homogenous characteristics, which provides meaningful regions for a certain application, allowing the extraction of the image attributes. These segments should correspond to structural units (“objects”) in the scene.

There is no universal method that can be successfully applied to all types of images due to a lack of a general mathematical model. Literature reports different categories of image segmentation algorithms: edge-based (such as Canny technique), region-based (Otsu, region growing etc.) and special theory-based (Fuzzy).

Fig. 4. The original X-ray image

Fig. 5. Otsu thresholding segmentation algorithm

Several classical segmentation algorithms and edge detection techniques (Sobel, Canny, Prewwit, Robert, Laplace, Otsu thresholding, Watershed,

SIOX, Shanbhag, Huang, gradient etc.) available in dedicated image processing and analysis software (such as the open-source programs: ImageJ [22], ITK [23] or Creaseg [24]) or in other commercial software such as Matlab (Dipimage toolbox), were analyzed to determine if they are suitable for use in segmenting X-ray image of vertebra real/test models (Fig. 4). Using ImageJ different algorithms and edge detection operators were applied to the original X-ray image of the test vertebra, several results being presented in Figs. 5 to 8.

Fig. 6. Shanbhag segmentation algorithm

Fig. 7. Prewit segmentation algorithm

Fig. 8. Gradient segmentation algorithm



Fig. 9. Fuzzy segmentation algorithm 3 classes

Fig. 10. Fuzzy clustering segmentation algorithm

Fig. 11. Hopfield Koss segmentation algorithm

This analysis showed that simple thresholding and multi-thresholding algorithms are not good enough for a correct segmentation of vertebra X-ray images due to the images fuzziness and to the fact that the vertebra has variable dimensions/thickness. Also, the use of artificial intelligence image segmentation algorithms (Fuzzy C-means, Fuzzy clustering, Hopfield-Koss etc.) proved unsuitable for the application (see Figs. 9 to 11).

2.2 X-ray Image Segmentation Using HNN – Proposed Algorithm and Implementation

HNN was proposed in 1985 by Hopfield as a way of solving optimisation problems. The network for the optimisation application tends to relax into stable states that minimize an energy function of a Lyapunov form [25].

The segmentation process can also be seen as a constraint optimisation problem. The constraints, in this case, are based on the fact that objects extracted from the image need to be homogenous and different from each other. Spatial constraints can also be introduced, i.e. objects over the edges of the image are not important. Starting with a random selection of objects, a HNN should be able to reach a stable state in which all the segmentation constraints are satisfied.

Thus, the usual strategy for segmenting X-ray images using HNN comprises two steps [25]: 1. Find a binary representation for the segmentation

solution, so that it can be mapped into a HNN stable state.

2. Define the energy function whose minimization will lead to an optimum solution to the problem.The problem of segmenting an image of n×n

pixels into k classes is to choose a suitable architecture for the HNN. For this research, considering the above mentioned requirements, the approach presented in [26] was applied, using a grid of N by k neurons, where N is the number of grey-level values found in the input image. The number of neurons is N×k. Using this binary mapping, the segmentation constraints can be summarized as follows: only one neuron per row can be active (output is 1); this puts each grey-level into one class (left term of Eq. (1)); the sum of outputs of all neurons in one row is 1, this ensuring the fact that each grey-level belongs to only and only one class (right term of Eq. (1)), where α and β are constant values:

E v v vsystactic xii

k

xi xjjj i

k

i

k

x

N

x= −

+

= =≠

===∑ ∑∑∑α β1

1

2

1111

NN

∑ . (1)

By minimizing the semantic energy (defined in this case as the sum of square distances from each grey-level to the centre of its class), these distances decrease to a minimum leading to a solution for the segmentation.

For each vertebra, two images are taken, one high-energy X-ray and one low-energy X-ray image, thus a semantic energy for both images can be defined as follows:



Ehl v

v DIS hl vsemantic

y yiy

N xi xy yi

k

y

N

x

N

yi1

11 1

1

111= +

+

=

=== ∑∑∑∑ϑ

δ11

1

111

22

hh vv DIS hh v

y yiy

N xi xy yi

k

y

N

x

N

=

=== ∑∑∑∑ ,

(2)

where N1, N2 are the number of grey-levels present in the low-energy respectively high-energy image, ϑ and δ are constants and hly and hhy are the histogram values of the y grey-level for the low-energy band and high-energy band image respectively and DISxy represent the distance between grey-level lx and grey-level ly.

The formula used for calculating DISxy is the following:

DIS d l lx y l l x yx y, , .= = −( )2 (3)

Using the above mentioned formulas, the energy can be expressed as presented in Eq. (4):

E E E

v V V

syntactic semantic

xii

k

x

N

xi xjj

= + =

= −

+

== =∑∑α β1

1

2

1 1jj i

k

i

k

x

N

y y yiy

N xi x y y yi hl hh v

v l l hl hh

≠==

=

∑∑∑

∑

+

++

− +

11

1

21η

( )( ) ( )

====∑∑∑

111

k

y

N

x

N

yiv ,

(4)

where N = max(N1, N2).

A simplification of the energy equation can be done using a Winner Take All (WTA) scheme transforming HNN into a competitive architecture (CHNN). The input-output function for a neuron is modelled as to satisfy the constraints of the energy function. For every row, only one neuron can be active. The neuron that receives the maximum input from all other neurons is declared a winner and its output is set to 1; the output of the rest of neurons for the same row is set to 0:

Vu v

x ix Ni k

x i x ii k,

,...,,...,

, ,,...,

, max==

===

11

11 if

00,.

otherwise

(5)

In other words, only one neuron is assigned 100% to a class. This satisfies the syntactic energy terms, therefore the energy Eq. (4) can be simplified to:

Ehl hh V

V DIS hl hh Vy y yi

y

N xi xy y yi

k

y

N

x

N

yi=+

+

=

=== ∑∑∑∑ 1

1

111 ( )( ) . (6)

Comparing Eq. (5) with the definition of the Lyapunov energy we can compute the updated equation for the interconnection weights when no bias or threshold is present, where Vxi and Vyi are the binary values for the output of neurons (x,i) and (y,i):

whl hh V

V DIx i y jx y Ni j k y y yi

y

N xi( , )( , ), ,...,, ,..., ( )==

=

= −+∑1

11

1 SS hl hh Vxy y y yi( ) .+ (7)

Fig. 12. Implementation of CHNN segmentation algorithm



The equation for the total input to the neuron (x,i):

vhl hh V

DIS hl hh Vxi

y y yiy

N xy y y yiy

N

= −+

+

=

=∑∑1

1

1( )( ) . (8)

An algorithm was designed and implemented in Borland DelphiTM using the above equations (Fig. 12). Figs. 13 and 14 present the results of applying this segmentation algorithm for 4 classes to the analyzed X-ray image of the test lumbar vertebra.

The research focused on a set of 34 different images obtained from various vertebra models with an image resolution of 760×520. The quality of the images was considered high, the images being obtained using a digital X-ray Siemens apparatus from Colentina Clinical Hospital Romania.

The deviation value between pedicle axis and screw axis is calculated after applying several simple geometrical computations to the image segmented using CHNN. First, the vertebra is extracted as a single object using a simple backtracking algorithm (as explained in section 1.2). Its centre of gravity is then computed [25] and two axes are then “drawn” from that position. The same type of computations is applied to the screws themselves.

The evaluation criteria is based on the difference in position between the axis of the vertebrae and the screw axis, which is compared with theoretically accepted values (grades) based on the safety area limits established from the literature data [4] to [6]. These grades are: I. penetration less than 2 mm (acceptable), II. penetration between 2 to 4 mm (requires screw repositioning), III. penetration more than 4 mm (unacceptable).

Fig. 13. CHNN proposed segmentation algorithm – 4 classes

Fig. 14. Extraction of vertebra and screw axis

3 FURTHER RESEARCH

Further research will consider the design of a user-friendly interface for the training system. This interface will display, for each trainee, the X-ray images of the vertebra and screws, both in sagittal and transverse planes, the segmented images, the deviation values, as well as evaluation messages and scores. All this information can be used for assessing the trainees’ performances and their evolution in time.

4 ACKNOWLEDGEMENT

This work has been co-funded by the Sectoral Operational Programme Human Resources Development 2007-2013 of the Romanian Ministry of Labour, Family and Social Protection through the Financial Agreement POSDRU/89/1.5/S/62557.

5 REFERENCES

[1] Foley, K.T., Gupta, S.K. (2002). Percutaneous pedicle fixation of the lumbar spine: preliminary clinical results. Journal of Neurosurgery: Spine, vol. 97, no. 1 (suppl.), p. 7-12, DOI:10.3171/spi.2002.97.1.0007.

[2] Amiot, L.P., Lang, K., Putzier, M, Zippel, H., Labelle, H. (2000). Comparative results between conventional and computer assisted pedicle screw installation in the thoracic, lumbar, and sacral spine. Spine, vol. 25, no. 5, p. 606-614, DOI:10.1097/00007632-200003010-00012.

[3] Cook, S.D., Salkeld, S.L., Stanley, T., Faciane, A., Miller, S.D. (2004). Biomechanical study of pedicle screw fixation in severely osteoporotic bone. Spine Journal, vol. 4, no. 4, p. 402-408, DOI:10.1016/j.spinee.2003.11.010.

[4] Lien, S.B., Liou, N.H., Wu, S.S. (2007). Analysis of anatomic morphometry of the pedicles and the safe zone for through-pedicle procedures in the thoracic and lumbar spine. European Spine Journal, vol. 16, no. 8, p. 1215-1222, DOI:10.1007/s00586-006-0245-2.

[5] Zdichavsky, M., Blauth, M., Knop, C., Graessner, M., Herrmann, H., Krettek, C., Bastian, L. (2004). Accuracy of pedicle screw placement in thoracic spine Fractures. Part I inter- and intra-observer reliability of the scoring system. European Journal of Trauma, vol. 30, no. 4, p. 234-240, DOI:10.1007/s00068-004-1422-9.

[6] Mirza, S.K., Wiggins, G.C., York, J.E., Bellabarba, C., Knonodi, M.A., Chapman, J.R., Shaffrey, C.I. (2003). Accuracy of thoracic vertebral body screw placement using standard fluoroscopy, fluoroscopic image guidance, and computed tomographic image guidance: A cadaver study. Spine, vol. 28, no. 4, p. 402-413, DOI:10.1097/01.BRS.0000048461.51308.CD.

[7] Polanecka, I., Korošec, M., Kopač, J. (2007). Drilling-force forecasting using neural networks. Strojniški

http://dx.doi.org/10.3171/spi.2002.97.1.0007

http://dx.doi.org/10.1097/00007632-200003010-00012

http://dx.doi.org/10.1097/00007632-200003010-00012

http://www.ncbi.nlm.nih.gov/pubmed?term=%22Salkeld SL%22%5BAuthor%5D

http://www.ncbi.nlm.nih.gov/pubmed?term=%22Stanley T%22%5BAuthor%5D

http://www.ncbi.nlm.nih.gov/pubmed?term=%22Faciane A%22%5BAuthor%5D

http://www.ncbi.nlm.nih.gov/pubmed?term=%22Miller SD%22%5BAuthor%5D

http://dx.doi.org/10.1016/j.spinee.2003.11.010

http://dx.doi.org/10.1016/j.spinee.2003.11.010

http://dx.doi.org/10.1007/s00586-006-0245-2

http://dx.doi.org/10.1007/s00068-004-1422-9

http://dx.doi.org/10.1097/01.BRS.0000048461.51308.CD



vestnik - Journal of Mechanical Engineering, vol. 53, no. 11, p. 771-783.

[8] Marinković, V. (2009). Application of artificial neural network for modelling the flash land dimensions in the forging dies. Strojniški vestnik - Journal of Mechanical Engineering, vol. 55, no. 1, p. 64-75.

[9] Donizete, V., de Almeida, L.F., Mathias, M.H. (2010). Wear particle classifier system based on an artificial neural network. Strojniški vestnik - Journal of Mechanical Engineering, vol. 56, no. 4, p. 284-288.

[10] Cheng, K.S., Lin, J.S., Mao, C.W. (1996). The application of competitive Hopfield neural network to medical image segmentation. IEEE Transactions on Medical Imaging, vol. 15, no. 4, p. 560-567, DOI:10.1109/42.511759.

[11] Poli, R., Valli, G. (1997). Hopfield neural networks for the optimum segmentation of medical images. Handbook of Neural Computation, Oxford University Press, Oxford, ch. G5.5, p. 1-10.

[12] Koss, J.E., Newman, F.D., Johnson, T.K., Kirch, D.L. (1999). Abdominal organ segmentation using texture transforms and a Hopfield neural network. IEEE Transactions on Medical Imaging, vol. 18, no. 7, p. 640-648, DOI:10.1109/42.790463.

[13] Chang, C.Y. (2003). Contextual Hopfield neural networks for medical image edge detection. 16th IPPR Conference on Computer Vision, Graphics and Image Processing, p. 161-167.

[14] Shi, Z., He, L. (2010). Application of neural networks in medical image processing. Proceedings of the 2nd International Symposium on Networking and Network Security, p. 23-26.

[15] Scholl, I., Aach, T., Deserno, T.M., Kuhlne, T. (2011). Challenges of medical image processing. Computer Science – Research and Development, vol. 26, no. 1-2, p. 5-13.

[16] Archip, N., Erard, P.J., Egmont-Pteresen, M., Haefliger, J.M., Germond, J.F. (2002). A knowledge-based approach to automatic detection of the spinal cord in CT images. IEEE Transactions on medical imaging, vol. 21, no. 12, p. 1504-1516, DOI:10.1109/TMI.2002.806578.

[17] Szilaghi, L., Szilaghi, S., Benyo, Z. (2007). Automated medical image processing methods for virtual endoscopy. World Congress on Medical Physics and Biomedical Engineering 2006. IFMBE Proceedings, vol. 14, part 15, p. 2383-2387.

[18] Chaabani, A.C., Boujelben, A., Mahfoudhi, A., Abid, M. (2010). An automatic-pre-processing method for mammographic images. International Journal of Digital Content Technology and Its Applications, vol. 4, no. 3, p. 190-200.

[19] Fang, J.J., Yang, C.Y., Lin, R.M. (2005). A computer-aided training system for pedicle screw implantation. International Congress Series. Computer Assisted Radiology and Surgery, vol. 1281, p. 661-666.

[20] Morris, D. (2006). Haptics and physical simulation for virtual bone surgery. PhD thesis, University of Stanford, Stanford.

[21] Kellerman, K., Salah, Z., Monch, Z. (2011). Improved spine surgery and intervention with virtual training and augmented reality. International Workshop on Digital Engineering, p. 8-15.

[22] ImageJ – Image Processing and Analysis in Java from http://rsb.info.nih.gov/ij/, accessed on 2010-12-12.

[23] Insight Segmentation and Registration Toolkit from http://www.itk.org/ ITK, accessed on 2011-02-10.

[24] Dietenbeck, T., Alessandrini, M., Friboulet, D., Bernard, O. (2010). CREASEG: A free software for the evaluation of image segmentation algorithms based on level-set. IEEE International Conference on Image Processing, p. 665-668, DOI:10.1109/ICIP.2010.5652991.

[25] Hopfield, J.J. (1984). Neurons with graded response have collective computational properties like those of two-state neurons. Biophysics, Proceedings of the National Academy of Sciences, vol. 81, p. 3088-3092, DOI:10.1073/pnas.81.10.3088.

[26] Cheng, K.S., Lin, J.S., Mao, C.W. (1996). The application of competitive Hopfield neural network to medical image segmentation. IEEE Transactions on Medical Imaging, vol. 15, no. 4, p. 560-567, DOI:10.1109/42.511759.

http://dx.doi.org/10.1109/42.511759

http://dx.doi.org/10.1109/42.790463

http://dx.doi.org/10.1109/TMI.2002.806578

http://dx.doi.org/10.1109/TMI.2002.806578

http://rsb.info.nih.gov/ij/

http://www.itk.org/

http://dx.doi.org/10.1109/ICIP.2010.5652991

http://dx.doi.org/10.1109/ICIP.2010.5652991

http://dx.doi.org/10.1073/pnas.81.10.3088

http://dx.doi.org/10.1109/42.511759

*Corr. Author’s Address: Danang University of Technology, Danang, Vietnam, [email protected] 517


0 INTRODUCTION

Robustness is a key factor of product design under uncertainty from material properties, manufacturing operations and practical environment. Actually, from designer’s brain to users’ hands, the product must pass through many stages of its life cycle. The variability generated at each stage obviously has an influence on the performances of the product. It can be the cause of the designed product not fully meeting the requirements of the customers and users.

Each part of making up the product is manufactured from raw material in the manufacturing stage using processes such as forging, cutting or grinding. Geometrical deviations are generated and accumulated on each part over the successive set-up of the manufacturing process due to inherent imperfections of raw material, tooling and machine. Then, the parts with deviations are assembled at the assembly stage. The deviations of the surfaces of the assembled parts generated at manufacturing stage affect the assemblability and the final geometry of the product. The geometry of the final product is, therefore, different from the nominal one at the end of these two production stages. On the other hand, the current product modelling technology is not capable of taking into account these deviations. Most of the simulations performed to predict the behaviour and the performance of a product (kinematics, dynamics, aerodynamics, etc.) are based on the nominal model of the product. Since this model cannot deal with geometrical deviations generated throughout the product life cycle, the variation of product behaviour and performance cannot be predicted. Thus, the “real” performance of the product, which is different from the designed one (nominal performance), cannot be

verified. The risk is then that the designed product fails to fully meet customers’ and users’ requirements in which situation, the product-process design has to be considered as not good or at least not robust.

Many methods and tools are proposed in the academic research in order to manage the effects of geometrical variability on a product design. However, only mentioned manufacturing or assembly stage of the product life cycle is mentioned. In [1], the authors addressed the impact of the manufacturing errors on the performance of the product. They defined the Manufacturing Variation Pattern (MVP) to represent the manufacturing characteristics and investigated its effects on the performance of the product. In [2], the authors presented the theory that offers an analytical and geometrical description of the performance sensitivity distribution of a product in the variation space. The theory can be applied to find the robust design less sensitive to the dimensional variation due to manufacturing errors or product wear. The authors in [3] proposed a new Probabilistic Sensitivity Analysis (PSA) approach for the design under uncertainty based on the concept of relative entropy. This approach allows providing the valuable information about the impact of the design variables on the performance of the product and the whole range or a partial range of the performance distribution. In [4], the authors proposed a statistical approach in order to evaluate the impact of geometrical variations on the angular rotational velocity between two bevel gears. The Monte-Carlo simulation method is used to consider the geometrical behaviour simulation and tooth contact analysis. The authors in [5] proposed a methodology for quantifying the kinematic position errors due to manufacturing and assembly tolerances.

A Method to Determine the Impact of Geometrical Deviations on Product Performance

Vignat, F. – Nguyen, D.S. – Brissaud, D.Frédéric Vignat1 – Dinh Son Nguyen2 – Daniel Brissaud1

1 University of Grenoble, G-SCOP Laboratory, France 2 Danang University of Technology, The University of Danang, Vietnam

Robustness is the key to successful product design when many variation sources exist throughout the product lifecycle. Variations are of many sources such as material defects, machining errors, and use conditions of the product. Most of product performance simulations are traditionally carried out using the numerical model created in the CAD system. This model only represents the nominal information about the product. Thus, it is difficult to take these variations into account in product performance prediction. A method is proposed in this paper to allow integrating the effect of these variation sources into the product performance simulation. This method is based on a random design of experiment method. As a result, an image of the “real” performance of the product is determined.Keywords: product life cycle, product performance, geometrical deviations


518 Vignat, F. – Nguyen, D.S. – Brissaud, D.

Based on this method, a kinematic amplitude variation for bodies’ position is calculated.

In [6], the authors proposed to integrate material and manufacturing process uncertainties in the design in order to consider their impacts on the performance of the product. They developed a procedure for uncertainty propagation from the material random field to the end product performance based on the product finite-element mesh. In mechanical assembly, there are several statistical approaches to determine the effect of geometrical deviation introduced in [7]. The authors in [8] proposed a tolerancing model, called Proportioned Assembly Clearance Volume (PACV), based on the Small Displacements Torsor (SDT) concept. This model aims to determine the effect of geometrical deviation of surfaces on an assembly. In [9], the authors proposed a new calculation method to analyse the geometrical deviations stack-up in the assembly line (parallel and serial assembly line) on the product designed. The authors in [10] introduced a model of geometrical product specifications for product life cycle. This model allows the communication of geometrical information which can come from design, manufacturing or inspection.

These studies examine the impact of geometrical variations in manufacturing or assembly stage on the designed product. However, the effects of the variation sources during the product life cycle on the performance of the product are not mentioned. Especially, the mathematical relationship between the performance of the product and the parameters of variation sources is unknown. In many cases, this relationship is not established and numerical resolutions using tools such as finite elements used to determine the performance for one specific set of values of the product parameters. One numerical solution thus determines one point on the response surface of the relationship between the performance of the product and the product parameters. In order to establish an approximation of the required relationship, this paper proposes to use a set of numerical solution combined with a design of experiment.

In order to study the effect of geometrical variability on product performance, there are important issues that have to be considered:• How to establish the relationship between the

performance and the geometrical deviations of the product?

• How to manage the causes and consequences of these deviations at design stage?This paper proposes, as an answer to the first

question, a method that allows establishing the relationship between performance and geometrical

deviations. The geometrical deviations generated and accumulated are modelled by the geometrical deviation model presented in [11] and reminded at the beginning of Section 3. A partial answer to the second question has been proposed in [12] for identification and classification of the influence of the deviation parameters.

1 INCLUDING GEOMETRICAL DEVIATIONS IN PRODUCT PERFORMANCE SIMULATION

There are many kinds of disturbances, during the product lifecycle, which may influence product quality and functionality [13]. In order to investigate their effect on the product performance, a method that allows taking them into account in the product performance simulation is proposed in this section. The overview of the proposed method is shown in Fig. 1.

Fig. 1. Performance simulation of the product with geometrical deviations

In the manufacturing and assembly stages, the geometrical deviations of each surface of the final product are caused by the effects of variation sources, such as tooling deformation, thermal deformation, material property of the part, etc. These deviations are modelled by the geometrical deviation model (GDM) that will be detailed in the next section. Monte-Carlo simulation method is then used to create an image of the “real” product population with geometrical deviations. The integration of these deviations into the product performance simulation is based on a design of the experiment (DOE) method. This DOE establishes the mathematical relationship between the product performance and the parameters of deviation.


519A Method to Determine the Impact of Geometrical Deviations on Product Performance

This relationship is used to generate an image of the product population performance from the “real” product population with geometrical deviations.

1.1 Geometrical Deviation Model

As presented above, the nominal model of a product created in CAD/CAM systems can only represent nominal product information and cannot handle geometrical deviations generated and accumulated during the product life cycle and especially at manufacturing and assembly stage. The GDM, as presented in [11], can model them based on small displacement torsor. A small displacement torsor T at a point O in the Cartesian coordinate system (O, X, Y, Z) is described by rotational vector R and translational vector D as shown in equation as follows:

T R DO X Y Z

= , ., , ,

A surface deviation torsor is a small displacement torsor describing the deviation between an associated surface and a nominal surface. The associated surface is an ideal surface associated with the real surface using a minimum distance criterion such as the least square. For example, the deviation torsor of the associated plane relative to its nominal position is described by the SDT TSurface at a point O in the local coordinate system (O, X, Y, Z), as shown in equation as follows:

Trxrytz

Surface

O X Y Z

=

0

00

, , ,

,

where rx, ry are rotational and tz translational components regarding X, Y, Z axis, respectively. The plan, in this case, has three degrees of freedom, so three positioning deviations of the plan are invariant (i.e. cannot be measured due to the surface class) relative to their nominal position and their values are arbitrarily fixed. Thus, the 0 value is chosen in order to hide the notion of invariance.

The GDM model, for the manufacturing stage, is based on the model of the manufactured part (MMP) proposed by [14]. The geometrical deviations generated by a manufacturing process are considered to be the result of two independent phenomena: positioning and machining, and are accumulated over the successive set-ups. The manufactured deviations of the part surfaces are expressed relative to their nominal position by a SDT T

P Piji,.

T T TP P Sj P Sj Pi

ji i

ji, , ,,= − + (1)

where TSj Pi, models the positioning deviation of

workpiece in set-up Sj. This deviation is a function of the MMP surfaces deviation generated by the previous set-ups, the part-holder surfaces deviations and the links part-holder/part surfaces.

TSj Pj

i, models the deviation of the machined

surface j realised in set-up Sj. This deviation is expressed relatively to the nominal machine. This torsor merges deviations of the surface swept by the tool and cutting local deformations.

A product is made up of parts assembled by the way of connections. Each part has already passed through the manufacturing stage where geometrical deviations were generated. Then, the product passes through assembly stage of its life cycle. The assembly stage brings its share of deviations to the product. The GDM for the assembly stage is based on the model of the assembled part (MAP). The positioning deviations of each part relative to its nominal position in the global coordinate system of the product are modelled by a SDT T

P Pi,.

T T T T TP P P P P P P P P Pi k k

nk

nk

mi i

mi, , , , ,,= + + + (2)

where TP Pnk

mi, is the link torsor between surface m of

MMP i (part i) and surface n of MMP k (part k). TP Pk,

is the positioning deviation torsor of part k, it models the positioning deviation of the part k (a subassembled part coming from the previous set-up of the assembly process) relative to their nominal position in the global frame of the product.

The GDM establishes the mathematical relation between the deviation sources from the manufacturing and assembly stage and product surfaces deviations. A Monte-Carlo simulation method is then used to create a set of M products with geometrical deviations (M generally chosen between 10,000 and 1 million). As a result, the product designers can be aware of the distribution of the product surfaces deviation [15].

1.2 Design of Experiment

Simulations to predict product performance (kinematics, dynamics, behaviour, failure, etc.) are usually carried out based on the numerical model of the product. This numerical model is created using the current product modelling technology such as CAD software and the simulation, which are performed using current simulation technologies such



as finite element, computing fluid dynamics, etc. For a complex product system, it is, however, difficult to create M models of the product with geometrical deviations and to calculate the performance of M products because performance simulations can be time consuming (several hours for one simulation). Thus, it is not possible to perform simulation for the M products. In order to overcome this limitation, a DOE approach is proposed to determine an approximated mathematical relationship between the performance of the product and the product parameters deviation. Then, it is possible to determine the performance of the M products by using this relationship.

Considering that each simulation is CPU time intensive, it is necessary to limit the number of simulations to be performed. The number of simulations increases with the number of factors taken into account for the design of experiment. It is thus necessary to limit this number of factors. These factors are geometrical deviation parameters and are defined based on expert knowledge. These key parameters are measured on the virtual product and are functions of the elementary deviation parameters. The value of these factors can be calculated by using the result of Monte-Carlo simulation of the M products.

The design of the experiment requires a defining level for these factors. The number of levels for these factors is chosen based on a compromise between the desired precision and the calculation time. The values of the factor levels are then calculated according to the factor range of variation. Depending on the number of factors n, the number of levels k and the chosen strategy, the number of deviated products N are determined. The corresponding N values of the n factors are gathered in a matrix P named design matrix, as shown in Eq. 3.

P

p p pp p p

p p p

n

n

N N nN

=

11 21 1

12 22 2

1 2

...

...... ... ... ...

...

. (3)

Next, simulation tools as FEA, CFD, etc., are used to calculate the performance of the N products. A set of N deviated models of the product has to be created in the CAD system. Each deviated model of the product j is modelled according to the value pij of the factors. The performance of each product is then calculated and finally, the performance of the N products is gathered in a vector R called response vector as expressed in Eq. (4).

R

rr

rN

=

1

2

.... (4)

The relationship between the performance of the product and the selected factors p1, p2, p3, ..., pn is established using a regression model and can be expressed by Eq. (5).

Performance = f (p1, p2, p3, ..., pn). (5)

For example, the linear least square fit model [16] can be used to establish the relationship in the case of n factors and 2 levels. From the result of 2n simulations, the relationship is expressed by a function as given in Eq. (6).

f p= ⋅ +β ε , (6)

where ε is the residual vector, p = p1, p2, p3, ..., pnT is a vector gathering the n factors, β is a coefficient vector of the model. It is calculated by Eq. (7).

β = ⋅ ⋅ ⋅−( ) ,P P P RT T1 (7)

where R = r1, r2, r3, ..., rNT is a response vector including N simulation responses.

The performance for the population of M products is then calculated by replacing the value of the selected factors p1, p2, p3, ..., pn with the collected data from the Monte-Carlo simulation into Eq. (5).

1.2.1 Factorial Design

As mentioned before, the number of simulations to be performed depends on the number of factors, the number of levels and the chosen strategies. In this paper, three strategies are proposed and can be chosen depending on the required precision, expert knowledge, the number of factors and calculation time. Two strategies, full factorial and Taguchi design, are commonly used while the last one, while random design is original.

A full factorial design of the experiment is an experiment that takes on all possible combinations of levels across all factors. The number of experimental to run is thus equal to kn for n factors and k levels. To limit the number of factors, key geometrical parameters are defined based on expert knowledge. Then, the number of levels for these factors has to be defined depending on a compromise between the desired precision and the calculation time. The values of the key parameters are measured on the



virtual product and are functions of the elementary deviation parameters. The variation range of these factors can be determined based on the M results from the Monte-Carlo simulation. The value assigned to the levels of these factors is determined according to these ranges of variation. The kn values of the n parameters are then gathered in the design matrix P. An extended description of the use of full factorial design of experiment has been presented in [18]. With full factorial design of the experiment, the number of numerical simulations increases dramatically with the numbers of selected factors and levels. For example, 23=8 simulations are necessary in the case of three factors and two levels (see Fig. 2). It is then necessary to find some alternatives to this method.

Fig. 2. Full factorial design with three factors, two levels

1.2.2 Taguchi Design

One alternative method is Taguchi’s orthogonal arrays. This method has been designed in order to reduce the number of simulations to be performed.

Table 1. Table of Taguchi design

Num

ber o

f lev

els

(k) Number of factors (n)

2 3 4 5 6 7 8 9 10 11 122 L4 L4 L8 L8 L8 L8 L12 L12 L12 L12 L163 L9 L9 L9 L18 L18 L18 L18 L27 L27 L27 L274 L16 L16 L16 L16 L32 L32 L32 L32 L32 5 L25 L25 L25 L25 L25 L50 L50 L50 L50 L50 L50

Taguchi’s orthogonal arrays are highly fractional orthogonal designs proposed by Taguchi. They are used to estimate the main effects with only a few experiments. This approach is also suitable to apply for certain mixed level experiments where the factors included do not have the same number of levels.

The number of factors is selected from expert knowledge in order to eliminate the factors that have few effects on the performance of the product. The value associated with the levels of each factor is also

defined from the results of the geometrical deviation Monte-Carlo simulation. Then, the number of experimental to run is selected based on the Taguchi’s orthogonal arrays, as shown in Table 1. In the case of n factors and k levels, there are Ln experimental runs that must be realized according to Taguchi table. For example, it is necessary to create 4 deviated models of the product for the performance simulation in the case of 3 factors and 2 levels. The design matrix P in this case is different from the one from factorial design. It is defined according to Taguchi table for each factor and level, as given in Eq. (8).

P =

++−−

+−+−

−++−

1111

1111

1111

. (8)

An extended description of the use of Taguchi’s orthogonal arrays has been presented in [17] and [18].

1.2.3 Random Design

In case of increasing complexity alongside with the number of factors, the number of necessary simulations to determine the relationship between performance and deviations can become too large and thus time consuming even when using Taguchi design. Moreover, the use of expert knowledge to determine key factors filters the deviation sources and can reach the list of some influential factors. Factorial design and Taguchi’s orthogonal arrays are not effective in this case. Thus, a random design of the experiment method is proposed in order to address these issues. All product geometrical deviations parameters, called factors pi (i = 1, ..., n) that have small effects on the performance of the product are taken into account.

The random design approach is realized in four steps (see Fig. 3):

Fig. 3. Algorithm diagram of random design

• Step 1. Draw randomly a product with geometrical deviations in the set of product collected from the Monte-Carlo simulation.



A kth product with geometrical deviations is randomly drawn in a set of M products collected from results of geometrical deviation simulation. The value of each factor ( 1,.., )ip i n= is calculated based on the drawn product deviation parameters values. The set of values of the factors pi (i = 1, ..., n) is added into the kth row of the design matrix P.

• Step 2. Create the deviated CAD model. The deviated model of kth product will be created in the CAD software corresponding to the value of each geometrical deviation parameters pi (i = 1, ..., n).

• Step 3. Simulate the performance of the product. The deviated model created in Step 2 is used to simulate the performance of the product in order to determine the performance of the representative product. The result is appended into the response vector R.

• Step 4. Eliminate the drawn product in the set of M products.

Repeat from Step 1 until the number of the drawn product reach the desired number N.

The main limitation of this method is the definition of the number N of simulations to be run. The number N of products drawn is defined considering the balance between the required precision and the expected calculation time. However, the problem is similar to that of every stochastic method were the size of the sample is a key parameter.

2 A CASE STUDY

The DOE method to integrate geometrical deviation in product performance simulation has already been applied to the complex example of a centrifugal pump [19]. In this application Factorial and Taguchi strategies have been used. In the present paper, the three strategies are applied to the same case for the purpose of comparison and evaluation. Thus, a simple example, a harmonic oscillator, is proposed in order to save calculation time (less than 1 ms per simulation) and to compare the three strategies to the analytical result which is known in the present case.

2.1 A harmonic Oscillator

The harmonic oscillator system includes a spring and a mass (see Fig. 4). The performance considered in this case is the natural frequency of the oscillator. The natural frequency of the oscillator is expressed by Eq. (9).

f km

= ⋅1

2π, (9)

where m is the mass of the load which depends on its density ρ and its volume. Thus, the load mass deviation depends on the geometrical deviations of the load surfaces. For example, the deviations of the cylinder radius and of the position of the two boundary planes (rotation and translation) affect the mass m. The mass is calculated by Eq. (10).

m = ρ · V, (10)

k is the spring constant or spring stiffness and it is calculated by Eq. (11).

k GdnD

=4

38, (11)

G E=

+2 1( ),

υ (12)

where E is Young’s modulus, d spring wire diameter, D spring mean diameter, n number of active windings and n a Poisson ratio.

From Eqs. (10) to (12), the variation of the frequency is obviously affected by geometrical deviations of the surfaces of the mass and the spring. Deviation torsor parameters of each surface, given in Table 2, influence the natural frequency of the mass-spring system.

Table 2. Geometrical deviation paramaters

Component GeometryDeviation

parametersDescription

Mass

Plan 1 Rx1, Ry1, Tz1Parameters of deviation

torsor of plan P1

Cylinder 2Rx2, Ry2 Tx2, Ty2

Parameters of deviation torsor of cylinder C2

drRadius deviation of cylinder C2

Plan 3 Rx3, Ry3, Tz3Parameters of deviation

torsor of plan P3

Spring Spire ∆, δ Deviation of spring outer and wire diameter

The geometrical deviations of each surface of the mass-spring system are modeled by a GDM as presented in Section 1 and the Monte-Carlo simulation method is then used to give an image of the “real” production of the mass-spring systems.

For example, the distributions of the cylinder’s deviations of 100,000 virtually produced loads are described in Fig. 5. The histogram of the translational



deviation (Tx2, Ty2) of cylinder C2 relative to X and Y axis are shown in Figs. 6a and b. These deviations can be included in the volume V calculation and by using Eq. (10) the mass deviation can be determined as shown in Fig. 6c.

Fig. 4. Mass-spring system

From the expert knowledge presented above, the relationship between the frequency of the spring system and its geometrical deviations can be determined and expressed by Eq. (13).

f G d

n Dh R h dr RR h T R h Tz z

=+

+− + + +

+ + + +

14 2 2

4

32 2

21

2

πδ

ρπ π

π π

( )

( )( )

( ) (∆

33 )

,

(13)

where h, R and r are respectively nominal height, radius and density of cylinder C2 of the load.

In fact, the frequency simulation of this system can be realized using the Eq. (13) directly. A population of 100,000 frequencies is generated from the results of the Monte-Carlo simulation mentioned in [15]. The histogram of the 100,000 frequencies calculated is shown in Fig. 6. An approximated relationship between the frequency and the geometrical deviation parameters is then established

Fig. 5. Monte-Carlo simulation results

using a linear regression model with the 100,000 frequencies data. This equation given in Eq. (14) is the most precise possible linear regression approximation. It will be used for the purpose of comparison with the other approximations.

Fig. 6. The distribution of frequency

f = 4.79048 ‒ 0.0547296Tz1 ‒ 0.155346dr ‒ ‒ 0.0545993Tz3 + 2.48972δ ‒ 0.149673∆. (14)

The three other DOE strategies will then be applied to this simple example to compare accuracy among them. The results will also be compared to this first approach that will be used as a benchmark.

2.1.1 Factorial Design

From expert knowledge as given in Eq. (13), geometrical deviation parameters of the load and the spring that have a strong influence on the natural frequency of the oscillator are defined as factors. These factors are selected as follows for the factorial design study: • Tz1, Tz3: translational deviation of the two planes

of the load P1, P3,• D,d: deviation of the spring mean and wire

diameter, • dr: radius deviation of the load cylinder.



The number of levels for the five factors is chosen as two and the number of runs is thus 32. The levels for the factors are calculated from the results of the Monte-Carlo simulation. The low and high levels are respectively selected at ‒3s and +3s of each factor distribution. The relationship between the frequency and selected factors is established by using a linear regression model from the 32 runs results. This function is expressed by Eq. (15).

f = 4.82605 ‒ 0.0560401Tz1 ‒ 0.16062dr ‒ ‒ 0.0560471Tz3 + 2.51307d ‒ 0.152286D . (15)

Fig. 7. The distribution of frequency by Factorial Design

The population of the frequency is generated from the results of Monte-Carlo simulation and Eq. (15). The distribution of the frequency is shown in Fig. 7.

2.1.2 Taguchi Design

The selected factors include all parameters of geometrical deviations of the spring system. There are thus 15 parameters as follows:• 6 parameters of deviation torsor of two plans of

the load,• 4 parameters of deviation torsor of the load’s

cylinder,• 2 parameters of deviation of the spring’s wire and

mean diameter,• 3 parameters of dimensional deviation of the base

plate (obviously not to influence the frequency).The number of levels is chosen as two and,

as indicated in the table of Taguchi’s orthogonal arrays, it is necessary to make 16 runs in this case. The relationship between the frequency and selected factors is similarly established by using the linear regression model from 32 runs results. The mathematical relationship is described by Eq. (16).

f = 4.91592 ‒ 0.0387945Tz1 ‒ 0.201746dr ‒ ‒ 0.0448108Tz3 + 2.53885δ ‒ 0.1232∆ . (16)

A population of 100,000 frequencies is generated by the collected data in the Monte-Carlo simulation and the Eq. (16). The histogram of the population is represented in Fig. 8.

Fig. 8. The distribution of frequency by Taguchi Design

2.1.3 Random Design

This method is realized by the four steps as presented in Section 2. Ten oscillators are randomly drawn from the 100,000 virtually produced oscillators. Then, the relationship between the frequency and the geometrical deviation parameters is established from 10 runs results. This relationship is expressed by Eq. (17).

f = 4.91592 ‒ 0.0387945Tz1 ‒ 0.201746dr ‒ ‒ 0.0448108Tz3 + 2.53885δ ‒ 0.1232∆ . (17)

Similarly, a population of 100000 frequencies is produced by using the result of the Monte-Carlo simulation and Eq. (17). The histogram of the population is shown in Fig. 9.

Fig. 9. The distribution of frequency by Random Design

2.1.4 Comparison

In order to compare the three DOE strategies accuracy, three points of view are proposed. First, the arithmetic difference between the frequency calculated from the approximated relationship obtained by each of the three DOE strategies and the exact result is



considered. The error distribution for each approach (factorial, Taguchi, random design) is shown in Fig. 10. The three strategies are precise and the maximum error is equal to 0.16 to be compared with the mean natural frequency which is equal to 4.79. Among the three strategies, the random design approach is the less accurate. However, the number of runs is minimal for this strategy and the number of factors is maximal. Random design approach is the most effective when the product system is complex and not well known and when the performance calculation is time-consuming. On the other hand, Factorial design is the most accurate approach but supposes a good expertise about the system behavior.

The second point of view, the main statistics descriptors of the natural frequency distribution obtained from the three proposed approaches are compared as shown in Table 3. The results for the three strategies and the first approach are very similar. The three approaches are then accurate and similar in terms of statistical results.

Table 3. Summary of the proposed methods

Factorial design

Taguchi design

Random Design

Exact model

Mean of frequency 5.03582 5.0719 5.00409 4.99506Standard deviation of frequency

0.594517 0.59592 0.61488 0.589002

Deviation parameters

5 All All All

Number of runs 32 32 10 100,000

The third point of view, the coefficient of the relationship between the natural frequency and the geometrical deviation parameters approximated by factorial, Taguchi and random design and the first approach are compared. These coefficients are given in Table 4. There is not much difference between factorial, Taguchi and random design approach and the first approach regarding these coefficients except for Tz1 coefficient of Random Design. The three strategies are then accurate for the determination

of an approximate relationship between deviation parameters and performance except for the low influence coefficient in the case of random design.

Table 4. Coefficient comparison among proposed methods

VariablesFactorialdesign

Taguchidesign

RandomDesign

Exactmodel

Constant 4.82605 4.91592 4.82709 4.79048Tz1 -0.05604 -0.03879 0.02266 -0.05473dr -0.16062 -0.20174 -0.18498 -0.15535Tz3 -0.05605 -0.04481 -0.11745 -0.05459δ 2.51307 2.53885 2.61698 2.48972∆ -0.152286 -0.1232 -0.09892 -0.14967

The selection among the three approaches to establish the relationship between the performance and the geometrical deviations of a product depends on the requirements concerning accuracy, time and cost. The factorial and Taguchi design are chosen when the expert knowledge is effective and the number of factors is not too large. The random design is chosen when it is difficult to determine the factors that have a strong influence on the product performance, and the number of factors is then considerably large and when performance calculation is complex and time consuming. Thus, random design is a new approach for performance simulation of complex product system, taking into account geometrical deviations generated during its lifecycle that can be used as the first approach to increase designer’s knowledge about the designed product behavior ad robustness. The three strategies are better than the first (full) approach in terms of time and cost and are necessary when the calculation time per simulation becomes long.

3 CONCLUSION

In this paper, a design of the experiment method is proposed to take into account the effects of geometrical deviations generated throughout the product lifecycle into product performance simulation. Three different

Factorial design Taguchi design Random designFig. 10. The distribution of error frequency



strategies: factorial, Taguchi and random design, are proposed for use depending on the complexity of the product considered. Factorial and Taguchi design of experiment are already well known but a novel strategy is proposed based on random choice of the design sample. This random design strategy is effective when the number of factors and levels and the time needed for performance simulation are becoming high.

The relationship between the performance of the product and the geometrical deviation parameters is established by using one of the three different strategies. An image of the population of the manufactured product is calculated by the Monte-Carlo simulation method, from a GDM. From the result of the Monte Carlo simulation, using the established relationship, an image of the performance of the population of products virtually manufactured is calculated. Then, the product designer can identify and classify the effect of each parameter of variation source on the product performance based on the result of the Monte-Carlo simulation and the corresponding performance for each virtual product.

In future work, the variance of product performance relative to variation of the deviation parameters can be determined. As a result, a robust design can then be found by minimizing the variance of the performance variation.

4 REFERENCES

[1] Yu, J.-C., Ishii, K. (1998). Design for robustness based on manufacturing variation patterns. Journal of Mechanical Design, vol. 120, no. 2, p. 196-202, DOI:10.1115/1.2826959.

[2] Zhu, J., Ting, K. (2001). Performance distribution analysis and robust design. Journal of Mechanical Design, vol. 123, no. 1, p. 11-17, DOI:10.1115/1.2826959.

[3] Liu, H., Chen, W. (2006). Relative entropy based method for probabilistic sensitivity analysis in engineering design. Journal of Mechanical Design, vol. 128, no. 2, p. 326-337, DOI:10.1115/1.2159025.

[4] Bruyère, J., Dantan, J.-Y., Bigot, R., Martin, P. (2007). Statistical tolerance analysis of bevel gear by tooth contact analysis and Monte Carlo simulation. Mechanism and Machine Theory, vol. 42, p. 1326-1351, DOI:10.1016/j.mechmachtheory.2006.11.003.

[5] Flores, P. (2011). A methodology for quantifying the kinematic position errors due to manufacturing and assembly tolerances. Strojniški vestnik - Journal of Mechanical Engineering, vol. 57, no. 6, p. 457-467, DOI:10.5545/sv-jme.2009.159.

[6] Yin, X., Lee, S., Chen, W., Liu, W.K., Horstemeyer, M.F. (2009). Efficient random field uncertainty

propagation in design using multiscale analysis. Journal of Mechanical Design, vol. 131, no. 2, p. 021006-021010, DOI:10.1115/1.3042159.

[7] Nigam, S.D., Turner, J.U. (1995). Review of statistical approaches to tolerance analysis. Computer-Aided Design, vol. 27, no. 1, p. 6-15, DOI:10.1016/0010-4485(95)90748-5.

[8] Teissandier, D., Couétard, Y., Gérard, A. (1999). A computer aided tolerancing model: proportioned assembly clearance volume. Computer-Aided Design, vol. 31, p. 805-817, DOI:10.1016/S0010-4485(99)00055-X.

[9] Mansuy, M., Giordano, M., Hernandez, P. (2011). A new calculation method for the worst case tolerance analysis and synthesis in stack-type assemblies. Computer-Aided Design, vol. 43, no. 9, p. 1118-1125, DOI:10.1016/j.cad.2011.04.010.

[10] Dantan, J.-Y., Ballu, A., Mathieu, L. (2008). Geometrical product specifications – model for product life cycle. Computer-Aided Design, vol. 40, p. 493-501, DOI:10.1016/j.cad.2008.01.004.

[11] Nguyen, D.S., Vignat, F., Brissaud, D. (2011). Geometrical Deviation Model of product throughout its life cycle. International Journal of Manufacturing Research, vol. 6, no. 3, p. 236-255, DOI:10.1504/IJMR.2011.041128.

[12] Nguyen, D.S., Vignat, F., Brissaud, D. (2010). Integration of multiphysical phenomena in robust design methodology. Proceedings of the 20th CIRP International Conference on Design, Springer, Berlin, p. 167-182.

[13] Kimura, F.. Matoba, Y.. Mitsui, K. (2007). Designing product reliability based on total product lifecycle modelling. Annals of the CIRP, vol. 56, p. 163-166, DOI:10.1016/j.cirp.2007.05.039.

[14] Vignat, F., Villeneuve, F. (2007). Simulation of the manufacturing process, generation of a model of the manufactured parts. Digital Enterprise Technology, Springer, p. 545-552.

[15] Nguyen, D.S., Vignat, F. , Brissaud, D. (2009). Applying Monte-Carlo methods to geometric deviations simulation within product life cycle. Proceedings of the 11th CIRP International Conference on Computer-Aided Tolerancing, Annecy, p. 1-10.

[16] Rao, C. R.; Toutenburg, H (1995). Linear Models: Least squares and Alternatives. Springer, New York.

[17] Korkut, I. Kucuk, Y. (2010). Experimental analysis of the deviation from circularity of bored hole based on the Taguchi method. Strojniški vestnik - Journal of Mechanical Engineering, vol 56, no. 5, p. 340-346.

[18] Nguyen, D.S., Vignat, F., Brissaud, D. (2010). Integration of geometrical deviations thougout product lifecycle into performance simulation. Proceedings of the 10th Global Congres on Manufacturing and Management, Bangkok, p. 624-631.

[19] Nguyen, D.S., Vignat, F., Brissaud, D. (2011). Taking into account geometrical variation effect on product performance. International Journal of Product Lifecycle Management, vol. 5, no. 2/3/4, p. 102-121.

http://dx.doi.org/10.1115/1.2826959

http://dx.doi.org/10.1115/1.2826959

http://dx.doi.org/10.1115/1.2159025

http://dx.doi.org/10.1016/j.mechmachtheory.2006.11.003

http://en.sv-jme.eu/scripts/download.php?file=/data/upload/2011/06/02_2009_159_Flores_04.pdf

http://dx.doi.org/10.1115/1.3042159

http://dx.doi.org/10.1016/0010-4485(95)90748-5

http://dx.doi.org/10.1016/0010-4485(95)90748-5

http://dx.doi.org/10.1016/S0010-4485(99)00055-X

http://dx.doi.org/10.1016/S0010-4485(99)00055-X

http://dx.doi.org/10.1016/j.cad.2011.04.010

http://dx.doi.org/10.1016/j.cad.2008.01.004

http://dx.doi.org/10.1504/IJMR.2011.041128

http://dx.doi.org/10.1504/IJMR.2011.041128

http://dx.doi.org/10.1016/j.cirp.2007.05.039

*Corr. Author’s Address: State Key Laboratory for Manufacturing Systems Engineering, Xi’an Jiaotong University, Xi’an, China, [email protected] 527


0 INTRODUCTION

The three-axis boring and milling machine are key devices in the modern manufacturing industry. Under operation, the components’ faults degrade the machining accuracy. It is difficult to detect fault features because the structures of three-axis boring and milling machines are complex. The factors such as the influence of transmission path, the transmission medium, the ambient environment, etc., degrade the measured signals. They lower the signal-to-noise ratio. In an extreme case useful information is buried in the noise so we can hardly recover it [1]. Therefore, it is necessary to study the fault diagnosis methods of this kind of machines [2] to [5].

Recently the detection of the incipient, weak fault has attracted more and more attention. Almost all conventional methods in weak signal processing are applied to filter or mask noise [6] and [7] so that while the noise is reduced, the useful signal may be weakened or even destroyed. Different from the traditional signal processing methods, stochastic resonance (SR) as a novel signal processing method, can achieve the effect of detecting a signal by utilizing noise to amplify weak signals in nonlinear dynamical systems instead of eliminating noise. Due to the characteristics of using noise to enhance signals, SR has extensively drawn attention in wide fields, especially in the weak signals detection [8] to [10].

During the past two decades, there have been many theoretical developments of SR in bistable systems [11] to [14]. Based on adiabatic approximation theory, the classical SR is only applicable to the small parameters object, namely the driving force frequency

and amplitude and noise intensity are far less than 1 [15]. But large parameter problems (driving force frequency and/or amplitude and/or noise intensity can be much larger than 1) may usually be involved in fault diagnosis of the mechanical systems. Therefore, the study of large parameter stochastic resonance methods become necessary, and in fact several achievements have been obtained during the past few years, such as modulated stochastic resonance (MSR) [16], re-scaling frequency stochastic resonance (RFSR) [17], frequency-shifted and re-scaling stochastic resonance (FRSR) [18] and so on. All of these non-classical SR methods have greatly enlarged its application areas.

The occurrence of SR needs strict conditions, that is, the periodic signal, noise and the nonlinear system must satisfy certain matching relations. However, based on the research theory at present, qualitative analysis of this matching relation can be only obtained [19]. So far, the engineering application of SR, which mainly depends on researchers’ experience of the and a large number of experiments has been limited to a large extent. With the development of computer technology, adaptive signal processing has been developed. Adaptive stochastic resonance (AdSR) was firstly proposed by Sanya Mitamin [20] who observed stochastic resonance with tuning the noise level. Adaptive stochastic resonance algorithms mainly contain two points: one is a study of the search rules and the other is the selection of the optimization index which is highly significant because it can determine whether the adaptive SR algorithm is valid or not. So far, among all of the measurement indexes that can evaluate the detection effect of SR, weighted kurtosis (WK) [21] and weighted signal-to-noise ratio

AdSR Based Fault Diagnosis for Three-Axis Boring and Milling MachineLi, B. – Li, J. – Tan, J.Y. – He, Z.J.

Bing Li1,* – Jimeng Li1 – Jiyong Tan2 – Zhengjia He1

1 State Key Laboratory for Manufacturing Systems Engineering, Xi’an Jiaotong University, China 2 The 29th Institute of China Electronics Technology Group Corporation, China

This paper introduced an adaptive stochastic resonance (AdSR) signal processing technique to extract fault feature of machining accuracy decay in boring and milling machine providing a vibration time-frequency distribution with adaptable precision. The AdSR uses a correlation coefficient of the input signals and noise as a weight to construct the weighted kurtosis (WK) index. The influence of high frequency noise is alleviated and the index used in traditional SR is improved accordingly. The AdSR with WK can obtain optimal parameters adaptively. In addition, through the secondary utilization of noise, AdSR makes the signal output waveform smoother and the fluctuation period more obvious. It has been found that AdSR appears to be a better tool compared to fast Fourier transform for fault characterization extraction in boring and milling machine in experiment case. It has been concluded that AdSR based signal processing technology successfully diagnosis the fault of machining accuracy decay in three-axis boring and milling machine.Keywords: stochastic resonance, fault diagnosis, boring and milling machine


528 Li, B. – Li, J. – Tan, J.Y. – He, Z.J.

(WSNR) [22] have been used most widely in signal processing and fault diagnosis.

By SR methods, the weak signals can be enhanced to a certain extent. However, when the signal-noise-ratio is too low, the detection effect is not satisfactory. In order to further improve the detection effect of the weak signals, stochastic resonance enhancement methods have been studied, such as cascade stochastic resonance [23], coupled stochastic resonance [13] and so on. With the intercoupling of adjacent resonance units, coupled SR can interrelate all of resonance units to improve the output signal-noise-ratio appropriately. And in a sense, SR based on a mechanism of energy transition from high-frequency area to low-frequency area to amplify low-frequency signal gradually can be regarded as a special low-pass filter and its filtering effect is better than the conventional low-pass filter [23]. Cascade SR, two bistable systems connected in series can weaken high-frequency dithering and make the output time domain waveform more smooth. The cascade SR can achieve higher signal-noise-ratio than single SR. Therefore, in processing the weak signals, the cascade SR has more advantages.

In the present work, the classical SR theory is introduced in brief in section 1.1, and an adaptive stochastic resonance algorithm is introduced in section 1.2. Finally, AdSR is applied to fault diagnosis for boring and milling machine in section 2. The effectiveness of the proposed method is confirmed by the application result.

1 STOCHASTIC RESONANCE

1.1 Basic Theory of Stochastic Resonance

SR, introduced by Benzi et al. [24], is a physical phenomenon. Here, for reasons of a convenient description the overdamped motion of a Brownian particle in a bistable potential in the presence of noise and periodic forcing is considered:

x t U x A t t( ) ( ) cos( ) ( ),'= − + + +0 Ω ϕ ξ (1)

where U(x) denotes the reflection-symmetric quadratic potential function:

U x a x b x( ) ,= − +2 4

2 4 (2)

where the barrier parameters a and b are positive real parameters.

Then Eq. (1) can be written as:

x t ax bx A t t( ) cos( ) ( ).= − + + +30 Ω ϕ ξ (3)

In Eq. (3), A0 is the periodic input signal amplitude, Ω ( = 2 π f0) is the driving frequency, ξ(t) denotes a zero-mean, Gaussian white noise, i.e.,

⟨ ⟩ =ξ ( ) ,t 0 (4)

⟨ + ⟩ =ξ ξ τ δ τ( ) ( ) ( ),t t D2 (5)

here, D is the noise intensity, ∗ stands for the statistical mean value calculation. According to the Eq.(2), there are two stable fixed points at x a b= ± and one quasi-stable fixed point at x = 0 which are the local minima and local maximum of the potential function U(x) respectively. The height of the potential barrier of the two minima is ΔU = a2 / 4b. The potential function U(x) is shown in Fig. 1.

In the absence of the periodic signal and noise, the position of the Brownian particle is determined by the initial conditions and is never changed. Only in the presence of the periodic signal, the Brownian particle moves in one of the two potential wells and can not cross the barrier, however, in the assistance of noise, the particle will accumulate enough energy to cross the potential barrier. When the noise intensity is large enough, the particle will continue to cross the barrier and jump between the two potential wells back and forth. When the transition rate between the two potential wells caused by noise, namely Kramers rate rK, matches the period of the input signal, the periodic signal is enhanced and stochastic resonance takes place.

r a UDK = −

2π

exp ,∆ (6)

and the requirements of the time matching in stochastic resonance:

2 1

0r fK

= . (7)

-2 -1 0 1 2-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

U∆

a b− a b+

0x =

x

U(x)

Fig. 1. The curve of the quadratic bistable potential function U(x)

When the amplitude of the periodic signal is smaller than the noise intensity, detection effect of the


529AdSR Based Fault Diagnosis for Three-Axis Boring and Milling Machine

input signal using single stochastic resonance is not satisfactory, and the output signal still contains a certain amount of noise and the feature of the useful signal is not significant. To further improve the detection effect of the weak signal, stochastic resonance enhancement methods have been studied. Cascade stochastic resonance, single bistable system connected in series, which not only makes the amplitude of characteristic frequency more outstanding in the frequency domain, but also the output waveform becomes more smooth in the time domain. The framework of cascade stochastic resonance, two bistable systems connected in series, is shown in Fig. 2.

Input signalp(t)=s(t)+n(t)

Output signalx(t)

The first bistable systemU(x) = –ax2/2 + bx4/4

The second bistable systemU(x) = –ax2/2 + bx4/4

Fig. 2. The framework of cascade stochastic resonance system

1.2 Adaptive Stochastic Resonance

Generally, an adaptive algorithm contains two aspects of contents: one is the study of search rules and the other is the selection of optimization index, which is highly significant as it can determine whether or not the adaptive algorithm is valid. In terms of SR, in all measurement indexes that can evaluate the detection effect of stochastic resonance, the signal-noise-ratio has been used most widely. The algorithm [21] flow of the adaptive stochastic resonance is shown in Fig. 3.

Algorithm:(i) The initial values of the barrier parameters, a, b,

theirs searching range and step size are set.(ii) The original input signal is input into the bistable

system and the output can be calculated.(iii) Aiming at the barrier parameters in each group,

WK of the output signal will be calculated.(iv) The maximum of the WK will be searched in the

searching range of a and b. When a and b exceed the searching range, the maximum of the WK and its corresponding barrier parameters a and b will be saved. However, if a and b do not exceed the searching range, the values of a and b will be changed and the algorithm will return to step (iii).

(v) The optimal detection result of the original input signal can be achieved by the cascade stochastic resonance by using the optimal parameters.

Fig. 3. The algorithm flow of the adaptive stochastic resonance

2 FAULT DIAGNOSIS OF BORING AND MILLING MACHINE USING AdSR

Fig. 4a is a picture of a three-axis boring and milling machine. The device was first used in 1993 when it was used to machine parts with high precision. Now, its machining precision decreases quickly. It can only serve as semi-precision machining equipment. Fig. 4b shows a bomb case of a tank which is processed by this machine.

The structural sketch of a bomb case is shown in Fig. 5. The main machining procedures include: (i) milling end face A; (ii) boring hole 1; (iii) rotating the bomb case with 180º, and milling end

face B;(iv) boring hole 2.

We found that coaxiality of two holes during a set of workpieces was poor. The capacity factor of procedure, Cp is 0.95, which belongs to third lever. After adjustment of machining parameters, the coaxiality still can not approach accuracy requirement. The milled bottom face was used as allocation



character of boring hole, so there is an error transfer effect of two procedures. The experimental results show that the location error of the milling procedure is key factor that causes coaxiality error along x direction of the two holes in the boring procedure. The procedures were adjusted to another boring machine, the errors were in acceptable domain, and the capacity factor of procedure, Cp is approach to second lever. The results indicate that the reason for error of the three-axis boring and milling machine is self-accuracy decay. We doubt that the key factor is turning accuracy decay of machine operating platform.

a)

b) Fig. 4. Machine and it’s part; a) boring and milling machine, b) a

bomb case of tank

a)

b) Fig. 5. Structural figure of bomb case; a) front view, b) top view

The turning operation platform is shown in Fig. 6. The platform includes mechanisms for secondary

change-speed. The transmission ration of the belt is 2.5. The number of teeth on worm, z1 is 1, and the number of teeth on worm wheel, z2 is 72. Worm wheel shaft is an output one to drive the turning operation platform. Thus, the accuracy of turning operation platform depends on the accuracy of the worm couple.

The vibration signals include abundant information and are easy to be picked, so these kinds of signals are used widely in advanced measurements. In this paper, we chose the vibration data as the target detection signals. The piezoelectric accelerometer with three directions was installed on the platform. The location of sensor is shown in Fig. 7. The concrete parameters of the applied acceleration sensor are shown in Table 1.

Table 1. Detailed information of the used sensor

Parameters ValueSensor Acceleration sensor

Sensitivity 100 mv/gMeasurement range ±50g pk

Sensor model LC0010Working frequency range 0.2 Hz to 5 KHz

Resolution 0.0002 g

Fig. 6. The turing operationg platform

Fig. 7. Location of sensor

Sony Ex data acquisition system were utilized in this test. The speed of turning operation platform is 2000 rpm. The rotating frequency of the small belt wheel is 33.33 Hz, and the big belt wheel and worm 13.33 Hz. The rotating frequency of the worm wheel,



output shaft are 0.19 Hz. The sample frequency is set 12.8 kHz, and the length of the data is 16384 in signal processing. Fig. 8 is x direction vibration signal of turning operation platform. It can be seen that the signal of time domain is in disorder. Almost no useful information can be found. The spectrum peaks of 857.80 Hz, 1523.00 Hz in frequency domain waveform have no relation with characteristic frequencies of transmission mechanism. The detail signal under 100 Hz is zoomed in and likewise there is no useful information. In Fig. 9, it can be seen that the frequency components are very complicated and still concentrate on the medium and high frequency band, which is not consistent with the fault characteristic information to be identified. Although the low frequency components can also be found in the low frequency band, the fault characteristics frequency is not obvious enough and is still polluted by the interference noise. Consequently, it is difficult to draw a final conclusion.

Fig. 8. Signal of platform in time and frequency domain

Fig. 9. Hilbert envelope spectrum of original signal

The AdSR are used to analyze the signal. The parameters of AdSR are set as below: a is [0.1, 5], step is 0.01, b is 0.1, and the compressibility of variable metric R is 200.

By AdSR, the optimum parameter a = 4.14 can be obtained. The optimum result of resonance domain is shown in Fig. 10. From Fig. 10, it can be seen that amplitude modulation is obvious. There are 17 waveforms of amplitude modulation with obvious periodic property. In order to get a better detection effect, we construct dipole cascade SR is constructed. Fig. 11 is the result of dipole cascade SR. After comparison of Fig. 10 and Fig. 11, it can be seen that

the phenomenon of amplitude modulation in Fig. 10 is more obvious.

In order to diagnose the fault of turning operation platform, the signal in Fig. 7 is dealt with Hilbert envelope demodulation. Fig. 12 is the result of envelope demodulation. It can be seen that the periodic property is clear in Fig. 12 and corresponding frequency is 13.49 Hz.

Fig. 10. Output signal of SR

Fig. 11. Output signal of dipole cascade SR

Fig. 12. Hilbert envelope demodulation of dipole cascade SR

The turning operation platform is mechanism for secondary change-speed. The connecting form of the motor and worm is a flexible belt connector driving, so the fault of the motor can not influence the turning accuracy of output shaft. In addition, there is no operating frequency of motor in frequency spectrum and envelope spectrum. While the periodic envelope signal is obvious in AdSR and dipole cascade SR. We have concluded that the reason for machining error came from two aspects. The actual operation frequency of the motor is not a stable value, 2000 rpm, but that it fluctuates. On the other hand, the error is caused by frequency distinguish ability. Due to the fact that the connector of the big belt wheel and belt is flexible contact, the fault of the wheel and gear will not generate catastrophe burst. The 13.49 Hz frequency peak can be seen in Fig. 12 as the worm’s



characteristic frequency. The worm wear fault of the turning operation platform is concluded to be the main reason generating accuracy decay of the three-axis boring and milling machine.

3 CONCLUSIONS

This paper attempted to findfault features of machining accuracy decay for a boring and milling machine using the adaptive stochastic resonance method. The AdSR is found to be a better tool for extracting the fault features compaed with the Fourier analysis alleviating the influence of high frequency noise consisting primarily in the machining vibration signals. Through the secondary utilization of noise, AdSR makes the output waveform smoother and the fluctuation period more obvious, the signal-noise-ratio is further improved, and realizes the enhancement of the fault feature. Research is being continued to explore the changing regularities of the machine fault diagnosis using the AdSR to monitor continuous machining procedure of boring and milling machine.

4 ACKNOWLEDGEMENTS

This work was supported by the National Basic Research Program of China (“973” Program) (Grant No. 2011CB706805) and the National Natural Science Foundation of China (Grant No. 11176024, 51075033).

5 REFERENCES

[1] Hu, N.Q., Chen, M., Wen, X.S. (2003). The application of stochastic resonance theory for early detecting rub-impact fault of rotor system. Mechanical Systems and Signal Processing, vol. 17, no. 4, p. 883-895, DOI:10.1006/mssp.2002.1470.

[2] Čuš, F., Župerl, U. (2011). Real-Time cutting tool conditon monitoring in milling. Strojniški vestnik – Journal of Mechanical Engineering, vol. 57, no. 2, p. 142-150, DOI:10.5545/sv-jme.2010.079.

[3] Tuan, D.V., Pil, C.U. (2011). Signal model based fault detection and diagnosis for induction motors using features of vibration signal in two dimension domain. Strojniski vestnik – Journal of Mechanical Engineering, vol. 57, no. 9, p. 655-666.

[4] Chaari, R., Abdennadher, M., Louati, J., Haddar, M. (2011). Modelling of the 3D Machining Geometric Defects Accounting for Workpiece Vibratory Behaviour. International Journal of Simulation Modelling, vol. 10, no. 2, p. 66-77, DOI:10.2507/IJSIMM10(2)2.173.

[5] Roy, S.S. (2010). Modelling of Tool Life, Torque and Thrust Force in Drilling: a Neuro-Fuzzy Approach. International Journal of Simulation Modelling, vol. 9, no. 2, p. 74-85, DOI:10.2507/IJSIMM09(2)2.149.

[6] Jiang, H.K., He, Z.J., Duan C.D. (2006). Gearbox fault diagnosis using adaptive redundant lifting scheme. Mechanical Systems and Signal Processing, vol. 20, no. 8, p. 1992-2006, DOI:10.1016/j.ymssp.2005.06.001.

[7] Li, B., Chen, X.F., He, Z.J. (2005). Detection of crack location and size in structures using wavelet finite element methods. Journal of Sound and Vibration, vol. 285, no. 4-5, p. 767-782, DOI:10.1016/j.jsv.2004.08.040.

[8] Hou, Z.F., Yang, J., Wang, Y.P. (2008). Weak signal detection based on stochastic resonance combining with genetic algorithm. 11th IEEE Singapore International Conference on Communication Systems, IEEE ICCS, p. 484-488.

[9] Zhao, W.L., Wang, Z.G., Huang, Z.Q. (2011). A new model of stochastic resonance used in weak signal detection. Applied Mechanics and Materials, vol. 43, p. 229-232, DOI:10.4028/www.scientific.net/AMM.43.229.

[10] Chen, M., Hu, N.Q., Qin, G.J. (2008). A study on additional-signal-enhanced stochastic resonance in detecting weak signals. IEEE International Conference on Networking, Sensing and Control, IEEE ICNSC, p. 1636-1640.

[11] Ye, Q.H., Huang, H.N., He, X.Y. (2003). A study on the parameters of bistable stochastic resonance systems and adaptive stochastic resonance. Proceeding of the IEEE International Conference on Robotics, Intelligent Systems and Signal, p. 484-488.

[12] Guo, W.N., Obermayer, K. (2003). Activity driven adaptive stochastic resonance. Physical Review Letters, vol. 90, p. 120602/1-120602/4.

[13] Kenfack, A., Kamal, P.S. (2010). Stochastic resonance in coupled underdamped bistable systems. Physical Review E – Statistical, Nonlinear, and Soft Matter Physics, vol. 82, no. 4, p. 046224/1-046624/5.

[14] Zhou, Y.R. (2011). Stochastic resonance in a time-delayed mono-stable system with correlated multiplicative and additive white noise. Chinese Physics B, vol. 1, no. 20, p. 010501/1-010501/6.

[15] Leng, Y.G., Wang, T.Y., Qin, X.D. (2004). Power spectrum research of twice sampling stochastic resonance in a bistable system. Acta physical Sinica, vol. 53, p. 717-723.

[16] Ye, Q.H., Huang, H.N., He, X.Y. (2003). Improved bearing estimates of weak signals using stochastic resonance and frequency shift techniques. MTS/IEEE Oceans Conference Record, vol. 5, no. 3, p. 2410-2413.

[17] Leng, Y.G., Wang, T.Y., Guo, Y. (2007). Engineering signal processing based on bistble stochastic resonance. Mechanical Systems and Signal Processing, vol. 21, no. 1, p. 138-150, DOI:10.1016/j.ymssp.2005.08.002.

[18] Tan, J.Y., Chen, X.F., Wang, J.Y. (2009). Study of frequency-shifted and re-scaling stochastic resonance and its application to fault diagnosis. Mechanical systems and signal processing, vol. 23, no. 3, p. 811-822, DOI:10.1016/j.ymssp.2008.07.011.

http://dx.doi.org/10.1006/mssp.2002.1470

http://en.sv-jme.eu/scripts/download.php?file=/data/upload/2011/02/07_2010_079_Cus.pdf

http://dx.doi.org/10.2507/IJSIMM10(2)2.173

http://dx.doi.org/10.2507/IJSIMM09(2)2.149

http://dx.doi.org/10.1016/j.ymssp.2005.06.001

http://dx.doi.org/10.1016/j.jsv.2004.08.040

http://dx.doi.org/10.1016/j.jsv.2004.08.040

http://dx.doi.org/10.4028/www.scientific.net/AMM.43.229

http://dx.doi.org/10.4028/www.scientific.net/AMM.43.229





[19] Xu, B.H., Duan, F.B., Bao, R.H. (2002). Stochastic resonance with tuning system parameters: the application of bistable system in signal processing. Chaos Solitons Fractals, vol. 13, no. 4, p. 633-644, DOI:10.1016/S0960-0779(00)00266-6.

[20] Mitamin, S., Kosko, B. (1998). Adaptive stochastic resonance. Proceedings of IEEE, vol. 86, no. 11, p. 2152-2183, DOI:10.1109/5.726785.

[21] Tan, J.Y., Chen, X.F., He, Z.J. (2010). Impact signal detection method with adaptive stochastic resonance. Journal of Mechanical Engineering, vol. 46, no. 23, p. 61-67, DOI:10.3901/JME.2010.23.061. (In Chinese)

[22] Li, B., Li, J.M., He, Z.J. (2011). Fault feature enhancement of gearbox in combined machining center by using adaptive cascade stochastic resonance. Science China Technological Sciences, vol. 54, no. 12, p. 3203-3210, DOI:10.1007/s11431-011-4612-9.

[23] He, H.L., Wang, T.Y., Leng, Y.G. (2007). Study on non-linear filter characteristic and engineering application of cascaded bistable stochastic resonance system. Mechanical systems and signal processing, vol. 7, no, 21, p. 2740-2749, DOI:10.1016/j.ymssp.2007.02.004.

[24] Benzi, R., Sutera, A., Vulpiana, A. (1981). The mechanism of stochastic resonance. Physical A, vol. 14, no. 11, p. 453-457.

http://dx.doi.org/10.1016/S0960-0779(00)00266-6

http://dx.doi.org/10.1109/5.726785

http://dx.doi.org/10.3901/JME.2010.23.061

http://dx.doi.org/10.1007/s11431-011-4612-9



*Corr. Author’s Address: University of Ljubljana, Faculty of Mechanical Engineering, Aškerčeva 6, 1000 Ljubljana, Slovenia, [email protected]

Teamwork in the Simultaneous Product RealisationRihar, L. – Kušar, J. – Gorenc, S. – Starbek, M.

Lidija Rihar1 – Janez Kušar1,* – Stane Gorenc2 – Marko Starbek1 1 University of Ljubljana, Faculty of Mechanical Engineering, Slovenia

2 PLASTA d.o.o, Slovenia

The paper presents a transition from sequential to simultaneous product realisation. Such a transition is not possible without prior well-organised teamwork or virtual teamwork. The article demonstrates a two-level team structure for simultaneous product realisation with a core team on the first level and several project teams of simultaneous product realisation loops on the second level and process for forming team or virtual team. Track and loop process and well-organised teamwork or virtual teamwork of the core team and project teams of simultaneous product realisation loops allows the savings in time and costs achieved by a transition from sequential to simultaneous product realisation. The results of organising teamwork and virtual teamwork are shown on a case study of simultaneous realisation of component for automotive industry.Keywords: product realisation, track and loop process, virtual team, communication tools, communication matrix

A transition from sequential to simultaneous product realisation considerably reduces the time and costs of product realisation [5] and [6], as shown in Fig. 2.

It can be seen from Fig. 2 that product definition costs rise uniformly in sequential product realisation, because of sequential execution of product definition activities (marketing, product draft, product development, elaboration of design documentation, material management), while production costs rise rapidly, due to long iteration loops for carrying out changes or eliminating errors.

The cost of product definition is much higher in simultaneous product realisation due to the parallel execution of activities (more work is done during this stage), while production costs are much lower than in sequential realisation, due to short iteration loops for carrying out changes and eliminating errors.

In simultaneous product realisation, there are interactions between individual stages of the product realisation process. Track-and-loop technology has been developed for executing these interactions [1]. The type of loop defines the type of co-operation between the overlapping stages of the simultaneous product realisation process. Winner [7] suggests that 3-T loops should be used where interactions exist between three levels of a simultaneous product realisation process.

A transformation of input into output is made in every loop on the basis of requirements and restrictions [4] and [6].

In small companies, a two-level team structure is planned for the execution of 3-T loops of a simultaneous product realisation process [5], [6] and [8], with a variable structure of core and project teams, as shown in Fig. 3. The task of the core team is process support and control, while the task of (virtual)

0 INTRODUCTION

The essence of modern production is to make a product that a customer needs as quickly and cheaply as possible.

Under these conditions, only a company that can provide customers with the right products in terms of functionality and quality, produced at the right time, at the right location, of the required quality and at an acceptable price, can expect global market success. A product that is not produced in accordance with the wishes and requirements of customers, hits the market too late and/or is too expensive, will not survive competitive pressure [1] to [3]. The customer should therefore participate in the process of simultaneous realisation of a product as early as possible. They can participate by expressing their wishes and requirements regarding project definition. The customer should be a temporary member of project teams in simultaneous product realisation loops.

The main feature of sequential product realisation is the sequential execution of stages in the product realisation process [4] and [5]. The observed stage of the product realisation process can only begin after the preceding stage has been completed. Data on the observed process stage are built gradually and are completed at the end of the stage–the data are then forwarded to the next stage (Fig. 1). In contrast with sequential product realisation, the main feature of simultaneous product realisation is the concurrent execution of stages in the product realisation process [4] and [6]. In this case, the observed stage can begin before the preceding stage has been completed. Data on the observed process stage are collected gradually and are forwarded continuously to the next stage (Fig. 1).


535Teamwork in the Simultaneous Product Realisation

Fig. 1. Sequential and simultaneous product realisation

Fig. 2. Time and costs of sequential and simultaneous product realisation

project teams is the execution of the tasks defined within the simultaneous product realisation process. It is obvious that simultaneous product realisation is not possible without well-organised teamwork or

virtual teamwork, which is the means for organisation integration. It incorporates:• the formation of a core team, project teams or

virtual project teams in product realisation loops,


536 Rihar, L. – Kušar, J. – Gorenc, S. – Starbek, M.

• the selection of communication tools for the core team, project teams or virtual project teams,

• the definition of a communication matrix.

1 TEAMWORK IN SIMULTANEOUS PRODUCT REALISATION

Teamwork is a precondition for a transition to simultaneous product realisation.

1.1 Forming Teams or Virtual Teams for Product Realisation

Analysis of teams in small companies led the authors to the conclusion that simultaneous product realisation required a shift from the terms “team” and “teamwork” to “virtual team” and “virtual teamwork” [9] and [10] when forming project teams.

A team is defined as a small group of people with complementary abilities that are activated in order to achieve the common goal for which they are all responsible. Team members are at the same location and in the same room. A virtual team is defined as a team consisting of members that are located in various buildings, countries or states and their cooperation is not limited by distance, organisation or national borders. Virtual teams are formed to carry out a specific project. The teams are disbanded when

the project is finished. A geographically dispersed virtual team allows a company to select the best team members, regardless of their locations. There are also substantial savings in time and costs of virtual team operation. Moreover, a virtual team may often have short meetings (if needed), which is physically difficult to achieve with a »classical« team.

Experience in solving problems related to forming teams or virtual teams [11] to [13] led the laboratory researchers to the conclusion that a virtual team should be formed in the following steps:Step 1: Identifying the need for a virtual team. Step 2: Definition of virtual team tasks.Step 3: Definition of procedures and processes for

achieving the common goal.Step 4: Selection of virtual team members.Step 5: Appointment of a virtual team leader.

The success of a virtual team leader depends on their skills, tools, techniques and strategies in a virtual environment. Because of many different forms of expert knowledge and leadership abilities, it is possible to rotate the virtual team leader–various members of a virtual team can undertake the role of team leader at various stages of the product realisation process.

Table 1. Advantages and drawbacks of tools for (virtual) teamwork

Communication tool Features Advantages Drawbacks

TEAM MEETINGon one locationSuitable for:TEAMWORK

Best tool for real-time communication because of personal contact and visual & verbal communication between team members.Meetings can be formal or informal.

Visual and verbal communication.Personal contacts between team members.All team members know each other.Participants can prepare for a meeting.

All team members must have time to attend the meeting.Much time needed for travel.High travel costs.

VIDEO CONFERENCE

Suitable for:VIRTUAL TEAM

Good tool for real-time communication because of visual and verbal communication and the possibility of interactions between team members.No direct personal contacts between team members.

Visual and verbal communication.Indirect personal contact.Prompt communication.No expensive travel.Saving in time.Team members can prepare for a meeting if they know its purpose and agenda in advance.The use of audio/video equipment.

All team members must be in the video conference room at the same time.Preparation in advance is required.Time delay of video due to distance.High costs of hiring communication channels.

AUDIO CONFERENCE

Suitable for:VIRTUAL TEAM

Good tool for real-time communication.Verbal communication and the possibility of interactions between team members.Functions in the Internet environment.

Reliable and always available communication tool.Participants are on various locations.Participants only need the Internet connection.Low cost of use.

Only verbal communication.Participants must be simultaneously present in the communication network.

VOICE MAILSuitable for:VIRTUAL TEAM

Tool for impersonal communication.For urgent messages only.

Message is sent to the recipient regardless of his presence.Recipient has time to prepare an answer.

Impersonal communication.Suitable for urgent, short messages.

E-MAILSuitable for:VIRTUAL TEAM

Impersonal communication without visual and verbal communication.No interactions between team members.

Useful for sending text messages and documents.Return receipt.

Impersonal communication.Limited size- documents to be sent.

GROUPWARESuitable for:VIRTUAL TEAM

Allows verbal communication between team members.Exchange if information in real-time.Simultaneous communication between several team members.During task execution the system allows simultaneous work of several participants on various locations.Common databases.Communication process must be defined in advance.

Simultaneous cooperation of team participants on various locations.Concurrent exchange of data and information.Access to data on a common server.Video communication is possible with additional video equipment.Information can be sent to team members via voice mail.

High burden for computer communications.High data-transmission costs.

ELECTRONIC WHITE BOARDSuitable for:TEAMWORK andVIRTUAL TEAM

Portable or fixed board that allows electronic data acquisition, exchange and archiving.

Simple use.Intended for taking notes on results.Rapid electronic transfer of the board contents to other team members.

High investment cost.Expensive and complicated maintenance.



1.2 Communication Tools Used in Teams and Virtual Teams for Simultaneous Product Realisation

Members of (virtual) teams must constantly communicate in order successfully to perform their tasks and to achieve the common goal. This is possible by using the available hardware and software [14]. Hardware includes telephones, modems and communication links (Internet connections). These are used for data transfer and for video conferences. Software includes efficient programs, LAN, communication and other tools for holding meetings.

It is possible to achieve efficient communication between members of the core team and virtual project teams by using the Internet. Several communication tools exist for efficient communication among team members [11]:

• team meeting,• video conference,• audio conference,• voice mail,• e-mail,• groupware and• electronic white board.

The research group at the Laboratory for Manufacturing Systems at the Faculty of Mechanical Engineering in Ljubljana, Slovenia, decided to analyse the characteristics, advantages and drawbacks of communication tools required in (virtual) teamwork of simultaneous product realisation. On the basis of collected and verified data from vendors of (virtual) teamwork communication tools, every team member made a list of the features, advantages and drawbacks of these tools. The team leader then organized a

Table 2. Communication matrix in simultaneous product realisation loops

ID Input information–document Activity Output information–documentTools used

Information (document)

sent by

Information (document) received by

Communication tool

1 Input information of activity 1 ACTIVITY 1 Output information of activity 1 ... Sender 1 Receiver 1 Tool 12 ... ... ... ... ... ... ...3 ... ... ... ... ... ... ...

: : : : : : :n Input information of activity n ACTIVITY n Output information of activity n Sender n Receiver n Tool n

Fig. 3. Loops of simultaneous realisation of car component



creativity workshop to obtain a coordinated proposal of the features, advantages and drawbacks of available communication tools. The results of the creativity workshop are shown in Table 1.

It can be seen from Table 1 that only two types of communication tools are suitable for teamwork (team meeting and electronic white board), while other tools are suitable for virtual teamwork.

1.3 Communication Matrix in Product Realisation Loops

The communication matrix defines the method of exchanging information and documents in the execution of simultaneous product realisation activity loops.

A list (Table 2) must be made for every activity:• input information with required documents for

beginning the execution of the activity,• output information with required documents that

arise from execution of the activity,• tools for creating and storing information,• sender of the information or document,• receiver of the information or document,• communication tool used for information

exchange.

2 SIMULTANEOUS REALISATION OF A CAR COMPONENT

2.1 Organisation for Simultaneous Realisation of Car Component

A company decided to make a project plan for simultaneous realisation of a car component (pedal assembly) and to carry out this project. The goal of the project was to make a competitive car component, suitable in terms of quality, reliability, mass, price and realisation time. Simultaneous realisation of the car component was divided into six stages:Stage 1: Preparation of the car component.Stage 2: Development of the car component.Stage 3: Development of the realisation process.Stage 4: Test production of car component.Stage 5: Qualification of the realisation process.Stage 6: Regular production.

The WBS method [10] and [11] was used for a decomposition of the project into activities. There were 280 activities and five loops of simultaneousrealisation of the car component within the six stages of pedal assembly realisation:• order acquisition loop (3-T loop),• pedal assembly development loop (3-T loop),

Fig. 4. Structure of teams for simultaneous realisation of car component



• pedal assembly process loop (3-T loop),• pedal assembly qualification loop (3-T loop),• completion of the project of pedal assembly

realisation loop (2-T loop).Fig. 3 shows how the loops are formed, and the

type of cooperation within realisation stages.When linking activities in a project network

diagram, the principle of parallelism was used in such a way as to achieve maximum overlapping of activities [10] and [11] where needed due to continuous information flow between people who execute activities.

After seeing the presentation of two- and three-level structures of (virtual) teams in product realisation loops [5] and [8] the company management selected a two-level team structure, whereby the core team is on the first level and five virtual project teams are on the second level (Fig. 4).

2.2 Forming Teams / Virtual Teams for Simultaneous Realisation of Car Component

The core team for simultaneous realisation of the car component will monitor the whole project, solve organisational issues and coordinate the strategy of performing tasks [15] and [16]. The company management decided that the following people would be members of the core team:• project manager (PM)–permanent member,• project team leader of a particular loop (VPL)–

non-permanent member,

• head of supply department (external supply and sales of investment funds–PUR+SIF)–permanent member,

• head of sales and sales logistics department (S+LD)–permanent member,

• head of development department (DEV)–permanent member,

• head of industrialisation and development of manufacturing technology department (IND+MTD)–permanent member,

• head of manufacturing planning and supply, maintenance and manufacturing centre (MP+MNT+MC)–permanent member,

• head of quality control department (Q)–permanent member,

• head of suppliers (SUP)–permanent member,• head of customers (CUS)–permanent member.

Fig. 5 shows the structure of the core team for simultane ous realisation of the car component.

The core team members (with the exception of the project manager) will work on the project for someof their working time, while the rest of the time they will perform tasks in their departments.

The project team manager will be outside their department throughout the project duration and will work full time on the project. When the project is finished, the project team manager will return to their department.

As shown in Fig. 3, there will be five virtual project teams in loops of simultaneous realisation of the car component.

Fig. 5. Core team structure



Members of virtual teams will be experts from 14 company departments and two representatives from strategic suppliers and customers, depending on the level of assigned responsibility for execution of activities within a particular loop.

When the company obtains an offer, loop 1 activities (Order acquisition loop) are started; its three stages are: project preparation, the development of the pedal assembly and the development of the pedal assembly process. This loop is executed when the sales department considers that it is sensible to make an offer for the realisation of the pedal assembly.

Loop 1 is followed by loops 2 to 5.The project manager decided (in agreement

with the company management) that the intensity of responsibility of each virtual team member during the execution of activities would be marked by a 1-3-9 method, as shown in Table 3.

A creativity workshop [17] was organised with 14 representatives from company departments, as well as representatives from suppliers and customers.

The goal of the workshop was to score the intensity of the responsibility of virtual team members when executing the activities of the five loops in simultaneous realisation of the car component.

Part of the results of scoring the intensity of responsibility of virtual team members during the execution of the first loop of simultaneous realisation of the car component are presented in Table 3.

The responsibilities of each virtual team member for the execution of activities in the first loop of car component realisation can be seen from Table 3.

The procedure of scoring the intensity of the responsibility of virtual team members was also carried out for the other loops.

From the sum of points assigned to the ith team member during execution of activity in the jth loop, a factor of total intensity of responsibility of the ith member in the jth loop can be calculated as:

FTISMPSAPi j

i j

j,

, ,=

where FTIi,j is a factor of total intensity of responsibility of the ith team member in the jth loop, SMPi,j sum of the points assigned to the ith member in the jth loop and SAPj sum of all points assigned in the jth loop.

A part of the results of the calculation of the total intensity of the virtual project team members’ responsibility factor during the execution of activities in all five loops of simultaneous realisation of pedal assembly are shown in Table 4.

After they had made an overview of the total intensity of responsibility factors of virtual team members during the execution of activities in the loops of pedal assembly realisation, the creativity workshop participants reached the following conclusions:• the ith member of the virtual project team (VPT)

of the jth loop of realisation of the pedal assembly, with the maximum factor of total intensity of responsibility, would be appointed as team leader of the jth loop of PTL,

• representatives from departments with a total intensity of responsibility factor above 5% would

Table 3. Scoring the intensity of responsibility of virtual team members in the “Order acquisition loop”

Name MNG S PM DEV IND Q MTD SIF PUR MC MP MNT AD LD SUP CUSSIMULTANEOUS REALISATION OF PEDAL ASSEMBLYOffer preparation project activities

Preparation of the project Inquiry reviewReceipt of inquiries and determination of their status

9 3

Review of requirements for completeness

9 3

Opening of the project 9 1 1 1 1 1 1 1Preparation of inquiry 9 1 1 1 1 1 1 1Definition of project team 9 1 1 1 1 1 1 1Repartition draftInitial meeting 1 (inquiry) 9 3 3 3 3 3 3 3

etc.,



Table 4. Factors of total intensity of responsibility of virtual project team members during the execution of loops of “Order acquisition loop”

Realisation of pedal assembly loops

Virtual team members

MNG S PM DEV IND Q MTD SIF PUR MC MP MNT AD LD SUP CUS SUM

Loop 1:Preparation of order

Scoring of individual team members in Loop 1

3 165 19 163 104 61 9 84 87 9 10 0 0 11 10 30 765

Intensity factor of individualteam member

0.39 21.5 2.48 21.3 13.6 7.97 1.19 11.2 11.5 1.19 1.33 0 0 1.46 1.33 3.98 100

Selected team members in Loop 1

165 163 104 61 84 87 30 694

Intensity factor of the selected team member

0 23.8 23.4 14.9 8.79 12.1 12.5 4.32 100

Fig. 6. Virtual project teams in the loops of simultaneous realisation of the car component



Table 5. Communication matrix for execution of “Order acquisition loop” activities

ID Input information – document Activity Output information – document

Tools used during

execution of activity

Information (document)

sent by

Information (document) received by

Communi- cation tool

0 SIMULTANEOUS REALISATION OF PEDAL ASSEMBLY

1 Offer preparation project activities2 Preparation of the project

3 Inquiry review

4 Inquiry Receipt of inquiries and determination of their status

PP document Customer P e-mail

5 PP document Review of requirements for completeness

Check of data (first sieve)

S

6 Check of data (first sieve) Opening of the project Design of implementation project

SAP S All departments

7 Design of implementation project Preparation of inquiry Message about opening of inquiry

SAP S All departments

e-mail;GW

8 PP document Definition of project team

Decision about temporary project group

S DEV, IND, PUR, Q, PM

e-mail;GW

: : : : : : : :

64 3D model, customer requirements; 2D drawings; Minutes of the meeting about product draft; Known required quantities, manufacturing deadlines and price

Technological process design

Technological process design

IND DEV, Q, IND SAP;SMT

65 Known required quantities, manufacturing deadlines and price; Message on opening of inquiry; Design of implementation project

Process planning, synoptics, process feasibility

Definition of plan, synoptics and process feasibility

SAP;SMT;e- mail

66 Customer requirements regarding packaging; 3D model, customer requirements; 2D drawings

Design of packaging Design of packaging

IND S SVP; EWB

67 3D model, customer requirements; Minutes of the meeting about product draft, Repartition draft; 2D drawings; Technological process design

QM plan elaboration QM plan elaboration

IND DEV,Q, IND, Calculation

SAP;SMT

68 Table of tolerances, Special requirements for a particular technology; 2D drawings; Test validation report

Checking of feasibility, Reminder 01

Reminder 01 IND DEV,Q, IND SAP; SMT; meeting

69 Reminder 01 PKU according to reminder01

PKU according to reminder01

IND, Q DEV, IND meeting

70 Technological process design; QM plan elaboration

Design of the necessary KMPO

Report on KMPO IND, Q DEV, IND SAP;SMT,EWB

also be included in the jth loop of pedal assembly realisation,

• representatives of suppliers and customers would also be included in the j-th loop of pedal assembly realisation, regardless of their total intensity of responsibility factor, in order to avoid misunderstanding suppliers’ and customers’ requirements.

Fig. 6 presents the structure of virtual project teams of five loops in simultaneous realisation of the pedal assembly.

2.3 Forming the Communication Matrix

A creativity workshop was organised with 14 representatives from company departments, as well as representatives from suppliers and customers.



The goal of the workshop was to define the following for every activity in the loops of simultaneous realisation of the car component:• input information with required documents for

beginning execution of an individual activity,• output information with required documents that

arise from execution of an individual activity,• tools for creation and storage of information,• senders of information or documents,• receivers of information or documents, and• the mode of sending the information or

documents.Table 5 shows some results of the creativity

workshop regarding the formation of the communication matrix for execution of activities of the “Order acquisition loop”. The communication matrix defines in advance the mode of information exchange and communication tools required.

3 CONCLUSION

The paper emphasises that simultaneous product realisation is not possible without well-organised teamwork or virtual teamwork.

A two-level team structure of a track-and-loop process of simultaneous product realisation, suitable for small companies, is presented. An overview is given of available communication tools for teamwork/virtual teamwork, with the advantages and drawbacks of individual tools.

When making virtual project teams in the loops of simultaneous product realisation, a method of calculating a responsibility intensity factor of team members with respect to loop activities was used. The content of the communication matrix of simultaneous product realisation is formed, defining the exchange of information/documents in the execution of simultaneous product realisation activity loops.

The suggested methodology of forming teams or virtual teams and communication matrix of simultaneous product realisation was tested on a study case of a pedal assembly. Altogether, there were 41 people in the core team and in the teams of the concurrent realisation loops. Because of the use of simultaneous product realisation strategies (parallelism, standardisation and integration) [10], [18] and [19] and the tools for continuous communication between team members, the execution time for realisation of pedal assembly project was reduced by 42%.

Further work on solving simultaneous product realisation problems will be focused on making a catalogue of the entire simultaneous product

realisation process using ARIS–a tool for process modelling and re-engineering [20].

4 REFERENCES

[1] Dickman, P. (2009). Schlanker Materialfluss. Springer Verlag, Berlin - Heidelberg.

[2] Palčič,I.,Buchmeister,B.,Polajnar,A.(2010).Analysisof innovation concepts in Slovenian manufacturing companies. Strojniški vestnik - Journal of Mechanical Engineering, vol. 56, no. 12, p. 803-810.

[3] Anišić,Z.,Krsmanović,C.(2008).Assemblyinitiatedproduction as a prerequisite for mass customization and effective manufacturing. Strojniški vestnik - Journal of Mechanical Engineering, vol. 54, no. 9, p. 607-618.

[4] Prasad, B. (1996). Concurrent Engineering Fundamentals, Volume I: Integrated Product and Proces Organization, Prentice Hall, New Jersey, p. 216-276.

[5] Kušar, J., Duhovnik, J., Grum, J., Starbek, M. (2004). How to reduce new product development time. Robotics and Computer-Integrated Manufacturing, vol. 20, no. 1, p. 1-15.

[6] Rihar, L., Kušar, J., Duhovnik, J., Starbek, M. (2010). Teamwork as a precondition for simultaneous product realisation. Concurrent Engineering: Research and Applications, vol. 18, no. 4, p. 261-273, DOI:10.1177/1063293X10389789.

[7] Winner, R.I. (1988). The Role of Concurrent Engineering in Weapons System Acquisition, IDA Report R-338, Institut for Defence Analysis, Alexandria.

[8] Duhovnik, J., Starbek, M., Dwivedi, S.N., Prasad, B. (2001). Development of New Products in Small Companies. Concurrent Engineering: Research and Applications, vol. 9, no. 3, p. 191-210.

[9] Rad, P.F., Levin, G. (2003). Achieving Project Management Success using Virtual teams. J. Ross Publishing, Boca Raton.

[10] Duhovnik, J., Žargi, U., Kušar, J., Starbek,M. (2009). Project-driven concurrent product development. Concurrent Engineering: Research and Applications, vol. 17, no. 3, p. 225-236, DOI: 10.1177/1063293X09343823.

[11] Kušar, J., Bradeško, L., Duhovnik, J., Starbek, M. (2008). Project management of product development. Strojniški vestnik - Journal of Mechanical Engineering, vol. 54, no. 9, p. 588-606.

[12] Žargi, U., Kušar, J., Berlec, T., Starbek, M. (2009).A company’s readiness for concurrent product and process development. Strojniški vestnik - Journal of Mechanical Engineering, vol. 55, no. 7/8, p. 427-437

[13] Vujica-Hecog,N.,Tuppinger,J.,Polajnar,A.,Palčič,I.(2007). Modern management concepts in Slovenian and Austrian companies, an empirical research. Advanced Production Engineering management, vol. 2, no. 1, p. 28-36.



[14] Duarte, D.L., Snyder, N.T. (2006). Mastering Virtual Teams. Jossey-Bass, cop., San Francisco.

[15] Fain, N., Moes, N.., Duhovnik, J. (2010). The role of the user and the society in new product. Strojniški vestnik - Journal of Mechanical Engineering, vol. 56, no. 7/8, p. 521-530.

[16] Fain, N., Kline, M., Duhovnik, J. (2011). Integrating R&D and marketing in new product development. Strojniški vestnik - Journal of Mechanical Engineering, vol. 57, no. 7/8, p. 599-609, DOI:10.5545/sv-jme.2011.004.

[17] Scheer, J. (2007). Kreativitätstechniken. GABAL Verlag, Offenbach.

[18] Kušar, J., Berlec, T., Žefran, F., Starbek, M. (2010).Reduction of machine setup time. Strojniški vestnik - Journal of Mechanical Engineering, vol. 56, no. 12, p. 833-845.

[19] Polajnar, A., Leber, M., Vujica-Herzog, N. (2010). Muscular-skeletal diseases require scientifically designed sewing workstations. Strojniški vestnik - Journal of Mechanical Engineering, vol. 56, no. 1, p. 31-40.

[20] Scheer, A.W. (1999). ARIS – Business Process Modeling. Springer-Verlag, Berlin – Heidelberg.

http://en.sv-jme.eu/data/upload/2011/07_08/08_2011_004_Fain_04.pdf

http://en.sv-jme.eu/data/upload/2011/07_08/08_2011_004_Fain_04.pdf

*Corr. Author’s Address: Yildiz Technical University, Mechanical Engineering Department, Barbaros Bulvari, 34349, Besiktas, Istanbul, Turkey, [email protected] 545


0 INTRODUCTION

The railway components are usually designed for infinite life based on the endurance limit or fatigue limit of the material. While this is in general sufficient, a comparatively small number of failures occur in practice, a fact that is due to limitations and uncertainties of the concept such as the number of loading cycles railway components such as axles and wheels experience over their service time, which is usually a multiple of the 106 to 107 cycles realized in a common S–N test. For a duty of 400,000 km per year, the number of load cycles of axles and wheels is about 2×108 [1] which refers to the range of the so-called giga-cyle fatigue [2] and [3]. Also, an introduction to railway applications such as axle, wheel and rail of fracture mechanics was given in the review paper of [4]. Additionally, it is possible to see the work steps of a damage tolerance analysis of a railway axle in [5].

Fatigue failures in railway axles are rare. Benyon and Watson [6] report on one to two failures per year on the United Kingdom railway network. Smith [1] specifies this number to 1.6 axles per year over the last 25 years out of a population of 180,000 axles. The rejection of 6,800 axles due to flaws in Russia in 1993 is reported in [7]. For a total number of about 2,000,000 to 2,500,000 axles this referred to an amount of 0.3% [8]. Although railway axles do not fail in North America freight service, they are known as critical components. Dedmon et al. present the results of stress analysis calculations performed for various different North American freight railway axle designs. Also, the authors propose a standard axle stress analysis method [9]. In order to stress control in axle-assembly, Okorn et al. [10] have obtained the

dynamic forces on the wheel in the case of straight ride, ride over obstacle and shock braking by using the coefficients from diagrams [11]. The stresses were calculated with conventional static equations and the FEM. The safety factor calculation has also been performed according to DIN 743 explaining shaft and axle calculations, in both cases.

Also, Bayraktar et al. [12] studied life analysis of light rail vehicle axles. Axles exposed to different loads have been analyzed and logarithmic life equations have been obtained due to equivalent stress which is calculated by cumulative damage theory called Palmgren-Miner [13] to [15]. The results obtained by analytical calculations have been compared with real broken values of the axles. Bayraktar [16] also improved these logarithmic life equations by inserting the effect of vibration of the axle. In the study, measured dynamic vibrations of the axle during traveling of the vehicle have been used to obtain the equations for life analysis. The results are very interesting as these calculated life values are nearly the same with damaged axle lives. This study has revealed a negative effect of vibration on rail vehicle axle.

In the present study, the wagon axle of the wagon of TCDD (Turkish Republic State Railways) with serial number 8000 suburban train travelling Sirkeci-Halkali route has been examined. It is observed that it is the axle of trailing wagon which is subjected to the most forcing. Related to the static load and the dynamic forces, which are functions of speed, critical section of the axle is determined by calculating the minimum safety factor along the axle. Therefore, the location of the fracture is the area between the wheel and the gear in which safety factor is less than

Railway Axle Analyses: Fatigue Damage and Life Analysis of Rail Vehicle Axle

Dikmen, F. – Bayraktar, M. – Guclu, R.Ferhat Dikmen – Meral Bayraktar* – Rahmi Guclu

Department of Mechanical Engineering, Yildiz Technical University, Turkey

In this study, failures in axles of rail vehicles have been examined. The axle failure in the paper is a classic fatigue problem with high magnitude bending stresses which alternate between tension and compression. The scope of this paper is to address life value of axle related to reliability and compare it with the realized life value up to fracture. The activity firstly deals with the definition of critical section in the axle. The location of the fracture is the section between the wheel and gear. The research then addresses determination of the wagon loading cases based on statistical data related to the number of passengers. First, minimum life value of the axle was determined considering full load, then effective life values were calculated by using Palmgren- Miner’s theorem as a cumulative failure theorem for real loading conditions in the case of different distributions. It is apparent from data found by calculations are in good agreement with practical damage values. Finally, changing of effective life values related to different working conditions is presented.Keywords: rail vehicle axle, fatigue, cumulative failure, life, reliability


546 Dikmen, F. – Bayraktar, M. – Guclu, R.

1. The determination of safety factor is performed by conventional strength calculations including Soderberg equations. Also, the photograph of the broken section is shown in Fig. 1. In order to find critical section having the minimum safety factor, it should be noted that the strength analysis is performed by calculating the dynamic forces due to the speed traveling at 20 m/s. And the effective stress occurred in this section is 315 N/mm2 [17].

Fig. 1. The broken section area of the axle [17]

1 LIFE ANALYSIS

The study focuses on which conditions fracture occurs and the life of the axle up to fracture. It is a fact that the life of a machine element depends on the material of which it is made and the working conditions. Moreover, the same materials, which are under the same conditions fail in the different times. This can be explained by Wohler diagrams. As it is known Wohler diagrams are obtained for every kind of materials based on experiments on laboratory fatigue specimens. It is clear that the laboratory conditions and the real working conditions of the machine element can not be the same. For this reason, effective stress should be considered for the life analysis performed by the help of Wohler diagrams. Also, the reliability of this life value should be stated.

It is stated that ultimate strength should be more than 650 N/mm2 for the considered axle material according to the specification. The materials used are 25CrMo4 and C60 which proper to DIN standards and provide mentioned provision. The material of failed axles is 25CrMo4. Therefore, the life analysis is performed by considering 25CrMo4 in the study. The Wohler diagram of 25CrMo4 with 50% reliability which is used in life analysis is given in Appendix. Unfortunately, the diagram is not suitable for usage since it is plotted for ultimate strength (σK = 800 N/mm2) and only in the form of 50% reliability. According to DIN 17200 standards, when radius increases, ultimate strength decreases in these types

of material. In particular, in the case of 100<d<250 [mm], ultimate strength (σK) is 650 N/mm2.

Since the Wohler diagrams for other reliability values could not be assured, the diagram is revised in the paper as explained below [18]:

In the considered diagram, ultimate strength (σK) is 800 N/mm2 and continual strength (σeD) is 400 N/mm2. It is known that σeD = 0.5×σK. Therefore, for σK = 650 N/mm2; σeD = 325 N/mm2. So, the diagram given for 50% reliability is replotted by replacing the diagram as the ratio of (325:400 = 0.8125). In addition, the standard deviation values of other materials similar with 25CrMo4 (for example: 34CrMo4) are calculated due to their Wohler diagrams plotted by considering various reliability values (Appendix). Then, by accepting that 25CrMo4 has the same standard deviation, the Wohler diagram of other reliability values is obtained for 25CrMo4 and given in Fig. 2.

In life analysis, firstly, it is assumed that the wagon travels under full load and then by considering the real load case, effective life will be found.

1.1 Life Analysis in Case of Full Load

The aim in this part is to find minimum life of the axle and to compare it with effective life. In the previous part, it is explained that in the case of full load (full passenger) with 20 m/s speed of wagon, the effective stress at the critical section is calculated as σef = 315 N/mm2 [17]. The life value corresponding to stated stress due to 10, 50 and 90% reliability related to Wohler diagram given in the Appendix is presented in Table 1. It can be seen that the life is infinite for the values of fewer than 50% reliability. However, it should be noted that it is not proper to realize the life analysis considering 50% reliability, for a machine element which has a vital importance. For this reason, the analysis will be performed by considering 90% reliability in this paper. It is also possible to determine life values by using 90 to 99.9% reliability. However, these values become more important only in the case of competition of firms [19] to [20].

Table 1. The life valued corresponding to σef = 315 N/mm2

Damage probability [%] Reliability [%] Life [Load cycle]10 90 2.5×106

50 50 infinite90 10 infinite

It is known that the mileage of the motor train is about 90,000 km/year. When the horizontal, vertical and angular irregularities are examined, it is clear that


547Railway Axle Analyses: Fatigue Damage and Life Analysis of Rail Vehicle Axle

dynamic forces acting the axle get the maximum value at each 50 meters (Appendix). The life is given in Eq. (1):

Lkm = load cycle distance × load cycle repetition . (1)

So, the life with 90% reliability can be found as mentioned below;

Lkm = 0.05×2.5×106 = 125,000 km

Lyear = 125.000/90.000 ≈ 1.39 year

Fig. 2. The new revised Wohler diagram plotted due to different reliability values [17] and [18]

It is clear that axle life is too short as 1.39 year in the case of traveling under full load with the speed of 20 m/s. However, the fact is not as seen practically.

Since it is changeable, passenger circulation for each station cause complex loading case and different way conditions cause various speed cases. Therefore, the forces acting the axle change continuously among the stations. In order to calculate the life of the axle sensitively, loading case should be known exactly. For this reason, statistical identifications are performed for determining daily real wagon load given in Figs. 3 and 4.

1.2 Statistical Distributions

The diagrams shown below are used for determining wagon load. In the light of these diagrams, the amounts of percentage at various fullness ratios of the traveled distance between stations 1 and 18 are determined and then, if the ratio of traveling time of the wagon with full passenger (100%) for the same distance to total traveling time is determined, the distributions shown in Table 2 can be obtained.

Table 2. The distributions due to fullness, distance, passenger number and time

Fullness [%]

Percentage of total distance

Number of passenger

Percentage of total time

20 3.8 10 2.6330 3.8 35 7.9040 7.6 60 10.5250 9.5 80 10.5260 9.5 100 10.5270 11.4 125 13.1580 16.4 150 15.8090 15.2 175 13.1595 11.4 200 10.52

100 11.4 220 5.26

By means of Table 2, for each fullness value, the real passenger numbers and the percentage of distance traveled by the mentioned number of passenger in total distance are calculated. It is be easy to explain the approach considered in Table 2 by giving an example:The distance which is traveled by 60% fullness is 9.5% of total distance. And, in 100% fullness, the time which is traveled by 125 passengers is 13.5% of total time. Therefore, in 60% fullness, the real passenger number is found as 0.60 × 125 = 75. And with these numbers of passengers, the percentage of traveled distance in

a) b) Fig. 3. Fullness ratio of wagon; a) departure, b) arrival



total distance is obtained by 0.095 × 0.1315 = 0.0125 (1.25%). This approach is realized for each fullness value as given in Table 3.

Additionally, Table 4 is used for approximately determining loading case. In the table, the number of passengers is grouped and the percentages of each group are presented by adding.

a) b) Fig. 4. Number of passengers versus day time a) departure, b) arrival

Table 3. Statistical load distribution for fullness

Fullness 20% Fullness 30% Fullness 40% Fullness 50% Fullness 60%Number of passengers

[%]Number of passengers




[%]

12 0.001 3 0.001 4 0.002 5 0.002 6 0.0027 0.003 10 0.003 14 0.006 18 0.007 21 0.00712 0.004 18 0.004 24 0.008 30 0.010 36 0.01016 0.004 24 0.004 32 0.008 40 0.010 48 0.01020 0.004 30 0.004 40 0.008 50 0.010 60 0.01025 0.005 37 0.005 50 0.010 63 0.012 75 0.01230 0.006 45 0.006 60 0.012 75 0.015 90 0.01535 0.005 52 0.005 70 0.010 88 0.012 105 0.01240 0.004 60 0.004 80 0.008 100 0.010 120 0.01044 0.002 66 0.002 88 0.004 110 0.005 132 0.005

Fullness 70% Fullness 80% Fullness 90% Fullness 95% Fullness 100%Number of passengers





[%]

7 0.003 8 0.004 9 0.004 10 0.003 10 0.00324 0.009 28 0.013 32 0.012 33 0.009 35 0.00942 0.012 48 0.017 54 0.016 57 0.012 60 0.01256 0.012 64 0.017 72 0.016 76 0.012 80 0.01270 0.012 80 0.017 90 0.016 95 0.012 100 0.01287 0.015 100 0.021 112 0.020 119 0.015 125 0.015

105 0.018 120 0.026 135 0.024 146 0.018 150 0.018122 0.015 140 0.021 157 0.020 167 0.015 175 0.015140 0.012 160 0.017 180 0.016 190 0.012 200 0.012154 0.006 176 0.009 198 0.008 209 0.006 220 0.006

Table 4. Statistical distribution

Group ofpassenger numbers

Average value[passenger]

Percentage related tototal length [%]

1 to 35 15 1736 to 65 50 2066 to 95 80 1996 to 125 110 18

126 to 152 140 10153 to 177 165 8178 to 199 190 5200 to 220 210 3



1.3 Loading Condition and Effective Life

For each loading case, the effective stresses occurred in the critical section and the corresponding life values as load cycles obtained from Wohler diagram are given in Table 5. It should be noted that this life calculation is realized for the speed of 20 m/s. Also, the variation of axle life versus other wagon speeds is given in Fig. 5.

In order to transform found life values for each loading case to effective life values, Palmgren-Miner cumulative damage theory will be used.

2 PALMGREN MINER CUMULATIVE DAMAGE THEORY AND AXLE LIFE

Since the variations of loading speed and road conditions are variable, the stresses occurring on the axle change permanently. These variations cause cumulative damage on the axle. Palmgren-Miner damage theory helps to determine an equivalent stress which corresponds to all these stresses [12], [14], [16] and [17].

CN

CN

CN Nn

n ef

1

1

2

2

1+ + + =... , (2)

C C Cnn ef

1

1

2

2

1σ σ σ σ

+ + + =... , (3)

where Ci is the percentage of statistical distribution of the forcing, Ni the life that corresponds to Ci and Nef effective life (number of cycles).

Therefore, if the life values found for each loading case, are applied in Eq. (2):

Nef = 2.187 × 107 [load cycle] ,

Lkm = 1,093,488 [km] ,

Lyear = 12.1 [year] .

By assuming that there is a mistake made during the determination of loading case distribution in Table 4, different distributions are proposed in Table 6. So, effective life values are recalculated by using these different distributions (Table 7).

Table 6. Distribution percentage of different configurations related to number of passengers

Number of passengers

1st distribution [%]

2nd distribution [%]

3rd distribution [%]

15 17 18 1550 20 22 1880 19 21 17

110 18 19 17140 10 8 12165 8 6 10190 5 4 7210 3 2 4

It seems that 2nd and 3rd distributions are similar with 1st distribution. However, these values are determined for total time. Also, the difference is the amount of probable error resulting from predictions.

Table 7. Life values of axles related to distribution percentages

Distributions [%] Nef Lkm Lyear

1st 2.187 × 107 1,093,488 12.12nd 2.848 × 107 1,424,062 15.83rd 1.706 × 107 853,045 9.5

The life due to light loading case will be more than in heavy loading case. The results given in Table 7 prove how the probable errors in distributions affect the life analysis. Consequently, it can be said that effective life of the axle is between 9.5 to16 years or about 12 years. Moreover, it is possible to see how the effective values change under different working conditions in Figs. 5 and 6.

Variation of axle life versus wagon load for different wagon speeds is presented in Fig. 5. The life

Table 5. Distribution of load, stress and life

Number of passengers Body weight [kN] The effective stres [N/mm2] The effective percentage Life [Number of cycles]15 300 281.6 17 infinite50 325 287.2 20 infinite80 348 292.3 19 infinite110 370 297.4 18 61.15×106

140 392 302.4 10 14.64×106

165 412 307 8 6.46×106

190 430 311.2 5 4.39×106

210 448 315.3 3 2.47×106



is infinite when the wagon speed is less than 13 m/s even with full loading. However, for the speed of 40 m/s, even with no passenger, the life is not less than 2 years. Since the examined wagons do not exceed the speed of 22 m/s, the critical limit occurs with 70 passengers.

Fig. 5. Variation of axle life versus wagon load for different wagon speeds

In Fig. 6, it can be seen the number of years of the effective life value calculated by using Eq. (2) for different travel speeds. For the speed of less than 15 m/s, life is infinite while the life is less than 1 year for about 32 m/s. For the speed of 22 m/s which is not exceeded in practically, it is 6.5 years and it is more than 50 years for the speed of 17 m/s.

Fig. 6. Variation of effective axle life related to travel speed

Table 8. The statistics for damaged axles

Axle no. Damage Approximately life [year]058.597 Broken 12848.793 Crack 11838.583 Crack 11.5

The life values of failed axles of considered wagons are given in Table 8. It is clear that the calculated values are approximate to the values given in Table 8. It should be noted that failure probability

is 10% since; the calculation is performed by considering 90% reliability. This means that at the end of the mentioned year value, all of the axle will not be failed immediately.

3 CONCLUDING REMARKS

This study has focused on the life analysis of the axle which has a vital important by considering 90% reliability. First, life analysis has been performed by supposing that wagon has been traveling with full load. Therefore, the minimum life value has been obtained as 1.4 year. Then, in the light of statically data and different distributions, the analysis has been realized by considering real loading conditions. And by using Palmgren-Miner cumulative damage theory, the effective life values are calculated as 12.1and 15.8 and 9.5 years related to different distributions. It is clear that the life value in the case of heavy loading is less than the life in the case of light loading. As a result, an average value as 12 years has been determined.

Additionally, it is clear that working conditions affect the effective life. The life is infinite when the wagon speed is less than 13 m/s even with full loading. However, for the speed of 40 m/s, even with no passengers, the life is not less than 2 years. Also, it is clear that how many years the effective life value calculated by using Eq. (2) will be for different travel speeds. For the speed of less than 15 m/s, life is infinite, while life is less than 1 year for about 32 m/s. For the speed of 22 m/s, which is not exceeded in practically, it is 6.5 years and it is more than 50 years for the speed of 17 m/s. As a result, the calculation of life values has been compared with real damaged axle life value and a good agreement has been revealed.

Also, these findings suggest that new constructive modifications should be realized. It is possible to obtain infinite life by increasing the diameter of axle from 171 to 173 mm and by changing the surface quality of the axle. The research has also shown that other materials which have more strength, such as 42Cr Mo4 or C60 can be used for axle material.

4 REFERENCES

[1] Smith, A. (2000). Fatigue of railway axles: a classic problem revisited. Proceeding of 13th European Conference on Fracture, p. 173-181.

[2] Bathias, C., Miller, K.J., Stanz-Tschegg, S. Ed. (1999). Special issue on gigacycle fatigue. Fatigue & Fracture of Engineering Materials & Structures, vol. 22, p. 543-728.



[3] Stanz-Tschegg, S., Mayer, H. (2001). Proceeding of international conference on fatigue in the very highcycleregime, University of Agriculture, Science Press.

[4] Zerbst, U., Madler, K., Hintze, H. (2005). Fracture mechanics in railway applications. An overview. Engineering Fracture Mechanics, vol. 72, p. 163-194.

[5] Zerbst, U., Vormwald, M., Andersch, C., Madler, K., Pfuff, M. (2005). The development of a damage tolerance concept for railway components and its demonstration for a railway axle. Engineering Fracture Mechanics, vol. 72, p. 209-239.

[6] Benyon, J.A.,Watson, A.S. (2001). The use of Monte-Carlo analysis to increase axle inspection interval. Proceeding of the 13th International Wheelset Congress.

[7] Hirakawa, K., Masanobu, K. (1998). On the fatigue design metod for high speed railway axles. Proceedings of the 12th International Wheelset Congress, p. 447-482.

[8] Ishizuka, H., Toyama, K., Kubota, M. (1999). Probability of improvement in routine inspection work of Shinkansen vehicle axles. Quaterly Report of Railway Technical Research Institute, vol. 40, no. 2, p. 70-73.

[9] Dedmon, S.L., Pilch, J.M., Lonsdale, C.P. (2001). A comparison of railroad axle stress results using different design sizes, loading criteria and analysis methods. ASME International Mechanical Engineering Congress and Exposition, p. 11-16.

[10] Okorn, I., Bester, T., Orbanic, P., Fajdiga, M. (2006). The difference between a front-axle stress calculation using the finite-element method and the same calculation cccording to DIN743. Strojniški vestnik - Journal of Mechanical Engineering, vol. 52, no. 1, p. 41-51.

[11] Goljar, M. (1977). Motor Vehicles. University of Ljubljana, Faculty of Mechanical Engineering, Ljubljana, p. 195-200 & 212-228.

[12] Bayraktar, M., Tahrali, N., Guclu, R. (2010). Reliability and fatigue life evaluation of railway axles. Journal of Mechanical Science and Technology, vol. 24, no. 3, p. 671-679.

[13] Miner, MA. (1945). Cumulative damage in fatigue. Transactions of ASME, Journal of Applied Mechanics, A159.

[14] Tahrali, N., Dikmen, F. (1995). Reliability and life analysis in construction elements. Yildiz Technical University Press, Istanbul. (in Turkish)

[15] Saatci, G.E., Tahrali, N. (2003). Cumulative damage theory and application to transmission

element. Journal of Aeronautics and Space Technologies, vol. 1, no. 1, p. 21-30.

[16] Bayraktar, M. (2010). Dynamic analysis of rail vehicle axles. PhD Thesis, Yildiz Technical University, Istanbul. (in Turkish)

[17] Dikmen, F. (1989). Axle fractures and precaution in rail vehicles: computer based dynamic analysis and constructive approach, PhD Thesis, Yildiz Technical University, Istanbul. (in Turkish)

[18] Niemann, G. (1975). Maschinelemente. Band I, 2. Auflage. Springer-Verlag, Hidelberg, New York.

[19] Gnilke, W. (1980). Lebensdauerberechnung der Machinelemente. Carl Hanser Verlag München, Wien, p. 119-124.

[20] Winter, H., Hirt, M. (1973). Die Zahnfussfestigkeitals Problem der Allgemeinen Festigkeitsberechnung. VDI-Berichte, no. 195, p. 119-134.

5 APPENDIX

Parameters of irregularities: Rail horizontal, vertical and angular irregularities

are explained by the Eqs. (A1) to (A3) respectively:

y y xy= ⋅ ⋅0 cos( ),Ω

(A1)

z z xz= ⋅ ⋅0 cos( ),Ω (A2)

θ θ θ= ⋅ ⋅0 cos( ),Ω x (A3)where, yi, zi, θi are rail road excitation inputs (I = 1.4). The maximum forces occur when these irregularities are maximum at the same time as shown in the Figure given below and as explained by Eq. (A4).

y z A A x

x

yz

yz

+ + = + + ( ) ⋅⋅ +( ) ⋅

θ θ θ θ

θ θ

02

02

0 5

cos .

cos . ,

∆Ω

Ω ∆Ω (A4)

where

y z y z y z x

x

yz

z yz

+ = + + ( ) ⋅⋅ +( ) ⋅

02

02

0 02

0 5

cos .

cos . ,

∆Ω

Ω ∆Ω

and A y z y z yz= + + ( )0

202

0 02 cos .∆Ω

Table A1. Amplitude, wave length and angular speed

Amplitude [m;rad] Wave length [m] Angular speed [1/m]Y0=0.006+0.012×v/50 Ly = 12.3 Ωy = 0.5108280Z0=0.006+0.004×v/50 Lz = 13 Ωz = 0.4833219θ0=0.010+0.020×v/50 Lθ = 17 Ωθ = 0.3695991



Fig. A1. Total rail irregularity

Fig. A2. The Wohler diagram of 25CrMo4

Fig. A3. The Wohler diagram of 34CrMo4 due to different reliability values

*Corr. Author’s Address: Termoelektro d.o.o. Uralska 9, 11060 Belgrade, Serbia, [email protected] 553


0 INTRODUCTION

Fiber reinforced composite materials have been gaining wide application in joining composites either to composites or to metal. Most commonly, joints are formed using mechanical fasteners. Composite structural components are generally connected with other components by means of bolted joints because of low cost, simplicity and ease to assemble or disassemble. Therefore, special attention must be paid to design of the bolted joints. Mechanically fastened joints in composite structures are commonly used in aerospace vehicles. Due to the anisotropic and heterogeneous nature, joint problems in composite structures are more difficult to analyze than those in isotropic materials. Due to the significance of the problem, many investigators have studied the strength of mechanically fastened joints in composite structures, [1] to [7]. Factors such as joint geometry and fiber orientation are important parameters for the mechanically fastened joints in composite plates.

One of the main prediction methods is Chang`s model, [8]. In this model the joint is taken to have failed when certain combined stresses have exceeded a prescribed value in any of the plies along the characteristic curve. In many previous investigations Yamada–Sun failure criterion, [8] to [13] was used. Icten et al. [14] established the behavior of mechanically fastened joints in woven glass-epoxy composites with [(0/90)3]s and [(±45)3]s material configurations. The failure analysis based on Hashin and Hoffman criteria was performed and

compared with experimental results. Okutan and Karakuzu [15] studied on the response of pin loaded laminated E/glass-epoxy composites for two different ply orientations such as [0/ ± 45]s and [90/ ± 45]s. The objective of this work is to study the behavior of graphite-epoxy pin loaded joints numerically, with particular attention paid to the sensitivity of the model to different geometric dimensions. The two-dimensional finite element method was used to obtain stress distribution at the composite lugs. To determine the failure load and failure mode initial damage prediction model was selected with Tsai-Wu failure criteria.

In this analysis, based on the Chang et al. strength prediction model, [8], the point stress failure criterion will be used to evaluate the characteristic lengths in tension and compression and a two-dimensional contact pin/lug finite element analysis used to evaluate the stress distribution in the vicinity of the joint. In many previous investigations Yamada-Sun failure criterion [8] to [12] was then used. In this paper Tsai Wu failure criterion was used to evaluate joint failure and the results compared with available experimental data [11] and correlation observed.

1 CHARACTERISTIC LENGTH METHOD

When a laminate is loaded through a fastener such as a pin or a bolt, both sides of the fastener hole are subjected to high tensile stress due to stress concentration. On the other hand, the front-area of the fastener hole experiences high compressive

Computation Method in Failure Analysis of Mechanically Fastened Joints at Layered Composites

Ilić, I. – Petrovic, Z. – Maksimović, M. – Stupar, S. – Stamenković, D.Ivana Ilić1 – Zlatko Petrovic2 – Mirko Maksimović3 – Slobodan Stupar2 – Dragi Stamenković4,*

1Military Technical Institute, Serbia 2Faculty of Mechanical Engineering in Belgrade, Serbia

3 Water Supplies, Serbia 4Termoelektro, Serbia

This paper considers a computation method in failure analysis of layered composites containing pin-loaded holes. The investigation is focused on developing a reliable computation procedure to analyze initial failure load for pin-loaded holes at layered composite structures. Finite element method (FEM) is used to determine stress distribution around the fastener hole. Combining Chang-Scott-Springer characteristic curve model and Tsai-Wu initial failure criterion are used to determine joint failure. Special attention in this work is paid to pin-load distributions and its effect on the load level of failure and its location. In previous work initial failure analysis was carried out using cosine distribution between pin/lug mechanically fastened joint. Here contact finite element pin/lug model is analysed. The influence of stacking sequences of layered composites containing pin-loaded holes is also investigated. Special attention is paid to failure load and mode analyses in composites with stacking sequence [0/(±45)3/903]S. The computation results are compared with available experimental results. Good correlations between computation and experimental results are obtained.Keywords: Composites, Failure analysis, FEM, Contact problem, Pin-loaded joints, Failure Index


554 Ilić, I. – Petrovic, Z. – Maksimović, M. – Stupar, S. – Stamenković, D.

stress. Furthermore, as the applied load increases and laminate deforms the contact surface between the fastener and the laminate changes.

A practical method considered to predict the failure load of composite joints with the least amount of testing is the characteristic length method. This method was proposed by Whitney and Nuismer [16], and has been developed by Chang et al. [10] and [11]. It is still used for the failure analysis of composite joints, [17]. In the characteristic length method, two parameters, i.e. compressive and tensile characteristic length should be determined by the stress analysis associated with the results of bearing and tensile tests on the laminates with and without hole. Once the characteristic lengths are determined, an artificial curve connecting the compressive and tensile characteristic lengths named characteristic curve is assumed, [8]. Failure of a joint is evaluated on the characteristic curve and not on the edge of the fastener hole. In this method the joint is taken to have failed when certain combined stresses have exceeded a prescribed value in any of the plies along the characteristic curve.

In order to evaluate the strength of composite pinned joints, Fig. 1, the stress distribution along a characteristic dimension around the hole must first be evaluated. The conditions for failure can then be predicted with the aid of an appropriate failure criterion. In this investigation the Tsai-Wu failure criterion was used for this analysis. This criterion can be written as [18]:

( . )

,F I F F F F

F F F= + + + +

+ + +1 1 2 2 6 6 11 1

2

22 22

66 62

12 1 22

σ σ σ σ

σ σ σ σ

FX X

FX Xt c t c

1 111 1 1

= + = −, ,

FY Y

FYYt c t c

2 221 1 1

= + =, - , (1)

F FS

F6 66 2 120 1 0= = =, , ,

where F.I is failure index, σi (i = 1, 2, 6) are stress components with respect to material principal axes and Xt,c, Yt,c are longitudinal and transverse tensile/compressive strengths of a unidirectional lamina and S is the ply shear strength. In this model, failure is expected to occur when the value of F.I is greater than or equal to unity.

The characteristic curve is an artificial curve made of compressive and tensile characteristic lengths.

Since the characteristic lengths are determined just for pure compression and tension, other combined failure modes are evaluated on the characteristic curve.

A popular method to construct the characteristic curve is proposed by Chang, Scott and Springer [8]. The characteristic curve, Fig. 2, is expressed as follows:

r D R R Rc ot oc tθ θ( ) = + + −( )/ cos ,2 0 (2)

where Roc and Rot are compressive and tensile characteristic lengths, respectively. The angle θ is measured counterclockwise or clockwise from the loaded direction toward the sides of the fastener hole as shown in Fig. 2.

Fig. 1. Geometry of the composite plate with a circular hole, subjected to pin

Fig. 2. Characteristic curve schematic diagram

The ultimate failure of a joint is generally divided into three modes depending on the failure location, θf [8].

0 ≤ θf ≤ 15° Bearing mode, 30 ≤ θf ≤ 60° Shear-out mode, (3) 75 ≤ θf ≤ 90° Net-tension mode.

There are in general three main failure modes: net tension (N-T), shear out (S) and bearing (B) as shown in Fig. 3. Net tension and shear out modes are catastrophic and result from excessive tensile and shear stresses. The bearing mode is local failure


555Computation Method in Failure Analysis of Mechanically Fastened Joints at Layered Composites

and progressive, and related to compressive failure. Net tension and shear out modes can be avoided by increasing the end distance (e) and width (W) of the structural part for a given thickness but bearing failure cannot be avoided by any modification of the geometry.

Fig. 3. Illustration of three basic failure modes

2 FAILURE LOADS OF MECHANICALLY FASTENED JOINTS

To determine failure load of mechanically fastened joint the procedure is composed of stress analysis and failure analysis using adequate initial failure criteria along the characteristic curve.

The strategy for the finite element modeling of the joints is the same as in the finite element model of the laminate for bearing tests shown in Fig. 5. Nonlinear finite element analysis for the joints composite structural components [19] is conducted by MSC/NASTRAN, [20]. Interface between fasteners and laminates is modeled by the slide line contact element provided by the software. The slide line element in MSC/NASTRAN was adopted to simulate the contact between the pins and the laminates. The pin and the laminate were modeled using CQUAD4 shell elements.

Force was applied to the pin as a uniformly distributed load. A typical finite element model of the mechanically fastened joint is shown in Fig. 4.

Fig. 4. Finite element model of the mechanical fastened joint

In this paper the problem of mechanically fastened joints of a laminated composite plate with frictional contact conditions is analyzed. Coulomb friction law is used and the contact constraints are

handled by extended interior penalty methods. The perturbed variation principle is adopted to treat the non-differential term due to the Coulomb friction. The computed results by our formulations are compared with the results of the experiment [21] to [23]. Here fatigue life estimation of these structural components can be included using procedure [24].

3 NUMERICAL VERIFICATION

To validate computation procedure of mechanically fastened joints numerical examples are included. Geometry properties of a mechanically fastened joint at composite structure are shown in Fig. 5. The finite element model of the contact problem of pin-loaded joint is shown in Fig. 6. Lug and pin are made from CFC composite and steel materials [25], respectively. Mechanical properties of these materials are given in Tables 1 and 2. For the purpose of comparison, failure analysis of mechanically fastened joints is carried out also using cosine load distribution. Here, specimens made from carbon type composite material T300/1034-C, with stacking sequence [0/(±45)3/903]s are considered.

In this investigation, Tsai Wu failure criteria are used in numerical analyses for predicting the failure behavior. The results of this analysis are shown in Tables 3 and 4 and Figs. 8 to 15. In these tables, it is seen that numerical and experimental results are similar.

Figs. 8 to 15 illustrates in detail, the failure mode of the composite material in Case 1, (Table 3). In this case net-tension (N-T) failure mode is evident.

Fig. 5. Geometry properties of mechanically fastened joints at composites



Fig. 6. Geometrical model of contact problem of pin-loaded joint

The geometric factors, clearances and friction, play important roles in determining contact stress. With a variation of these factors, it can be found that each point takes a different magnitude of pin loading, and extended parametric studies on these factors

may be needed for design consideration. Mechanical properties of carbon composite material T300/1034-C, Table 1, are obtained from standard tests.

Fig. 7. Finite element model of contact problem of pin-loaded joint

Table 4 shows comparisons between present comutation and experimental results for failure analyses of mechanically fastened joints. As it was previously mentioned, there are two different computation procedures considered in this work.

Table 1. Mechanical properties of CFC composite material T300/1034-C [4]

Longitudinal Young’s Modulus [N/mm2]

Transverse Young’s Modulus [N/mm2]

Shear Modulus [N/mm2]

Pooission’s Ratio

Longitudina Tensile and

Compressive Strength [N/mm2]

Transverse Tensile and

Compressive Strength [N/mm2]

Rail Shear Strength [N/mm2]

One Layer Thickness

[mm]

146860 11720 6180 0.30 1730 1380 134 0.139

Table 2. Mechanical properties of pin

Young’s Modulus[N/mm2]

Shear Moduluss[N/mm2]

Poisson’s RatioUltimate Tensile

Strength[N/mm2]

Ultimate Shear Strength[N/mm2]

Static Friction Coefficient

210000 81400 0.29 1250 800 0.25

Table 3. Comparisons computation with experimental results

CaseStacking sequence [0/(±45)3 / 903]S

d[mm]

e[mm]

w[mm]

L[mm]

Rot[mm]

Roc[mm]

Fexp

[N]FI cont FI cos Failure

mode1 3.175 9.525 9.525 46 0.457 1.778 5670 1.03 0.92 N-T2 6.35 19.05 19.05 93 0.938 2.245 10340 0.9 0.85 N-T3 6.35 19.05 31.75 93 1.166 2.395 13110 0.97 0.87 B

Table 4. Comparisons computation with experimental results

Case

Failure load (F.I=1) Difference Fexp and F present solutionsDifference

Fexp and Fcos [4] [%]

Contact FEM CosineCosine [4] Contact

[%]Cosine

[%]present solution

Fcont [N] Fcos [N] Fcos [N]1 5550 5900 5909 -2 +4 +4.22 10900 11200 9300 +5 +8 +103 13200 14000 11390 +0.7 +7 +13.1



Fig. 8. Distributions of F.I at composite lug for experimental failure load (Case 1, N-T failure mode, contact (F=567 daN)

Fig. 9. Distributions of F.I at composite lug for experimental failure load (Case 1, N-T failure mode, cosine (F= 567 daN)

Fig. 10. Distribution of F.I along characteristic curve for experimental failure load (Case 1), contact (F= 567 daN)

Fig. 11. Distribution of F.I along characteristic curve for experimental failure load (Case 1), cosine (F= 567 daN)

Fig. 12. Distributions of F.I at composite lug for failure load (Case 1: N-T failure mode), contact (F= 555 daN)

Fig. 13. Distributions of F.I at composite lug for failure load (Case 1: N-T failure mode), cosine (F= 590 daN)

Fig. 14. Distribution of F.I along characteristic curve for failure load (Case 1), contact (F= 555 daN)

Fig. 15. Distribution of F.I along characteristic curve for failure load (Case 1), cosine (F= 590 daN)



These two methods are based on: (1) combining contact finite element method for stress analysis in conjunction with Tsai-Wu failure criteria for initial failure analysis and (2) combining cosine load distribution method with Tsai-Wu failure criteria for initial failure analysis.

Clearly, the present contact finite element method in conjunction with Tsai-Wu failure criteria gives the best match to experimental results, (Table 4), while the present cosine load distribution method with Tsai Wu criterion is comparable to results from reference [11]. Results in reference [11] are obtained combining cosine load distribution method in conjunction with Yamada-Sun failure criteria.

Coulomb friction law is used and the contact constraints are handled by extended interior penalty methods. The perturbed variation principle is adopted to treat the non-differential term due to the coulomb friction. The computed results by the previous formulation are compared with experimental results. Good agreement between computation and experimental results is obtained.

4 CONCLUSIONS

In this work a numerical study on the failure load and failure mode investigations of pin loaded composite joints are presented. In the numerical study, a Tsai-Wu failure criterion is used to predict the failure load and failure mode. Special attention is paid to failure load and mode analyses in composites with stacking sequence [0/(±45)3/903]S. For the verification of the proposed computation method, composite laminated joints with various ratios of width-to-hole diameter examined and compared with the test results. It can be seen that the results obtained numerically and experimentally are close to each other. Failure loads of the joints with bearing (B) or net-tension (N-T) modes are the same in both, the present numerical method and the test-based conventional method. Summarizing the results, it is shown that the proposed numerical method in this work based on combining the contact finite element method for stress analysis in conjunction with Tsai-Wu failure criteria for the initial failure analysis predicts the strength of composite joints under pin loading within the maximum of 5% difference from the test results.

The computed results by considered formulations are compared with experimental results in the case of a single hole and agree well.

5 ACKNOWLEDGMENTS

This work was financially supported by the Ministry of Science and Technological Developments of Serbia under Projects OI-174001 and TR-34028.

6 REFERENCES

[1] Dano, M.L., Gendron, G., Piccard, A. (2000). Stress and failure analysis of mechanically fastened joints in composite laminates. Composite Structures, vol. 50, no. 3, p. 287.296.

[2] Lessard, L.B., Shokrieh, M.M. (1995). Two-dimensional modeling of composite pinned-joint failure. Journal of Composite Materials, vol. 29, no. 5, p. 671-697, DOI:10.1177/002199839502900507.

[3] Aktas, A., Karakuzu, R. (1999). Failure analysis of two-dimensional carbon-epoxy composite plate pinned joint, Mechanics of Composite Materials and Structures. vol. 6, no. 4, p. 347-361.

[4] Icten, B.M., Karakuzu, R. (2002). Progressive failure analysis of pin-loaded carbon-epoxy woven composite plates. Composites Science and Technology, vol. 62, no. 9, p. 1259-1271, DOI:10.1016/S0266-3538(02)00071-4.

[5] Camanho P.P., Matthews, F.L. (1999). A progressive damage model for mechanically fastened joints in composite laminates. Journal of Composite Materials, vol. 33, no. 24, p. 2248-2280, DOI:10.1177/002199839903302402.

[6] Okutan, B., Karakuzu, R. (2002). The failure strength of pin-loaded multidirectional fiber-glass reinforced epoxy laminate. Journal of Composite Materials, vol. 36, no. 24, p. 2695-2712, DOI:10.1177/002199802761675502.

[7] Lin, H.J., Tsai, C.C. (1995). Failure analysis of bolted connections of composites with drilled and moulded in hole. Composite Structures, vol. 30, no. 3, p. 159-168, DOI:10.1016/0263-8223(94)00040-9.

[8] Chang, F.K., Scott, R.A., Springer, G.S. (1982). Strength of mechanically fastened composite joints.Journal of Composite Materials, vol. 16, no. 6, p. 470-494, DOI:10.1177/002199838201600603.

[9] Maksimovic S. (2007). Post-buckling and failure analysis of axially compressed composite panels using FEM. Scientific Technical Review, vol. 57, no. 3-4, p. 43-48.

[10] Chang, F.K. (1986). The effect of pin load distribution on the strength of pin loaded holes in laminated composites. Journal of Composite Materials, vol. 20, no. 4, p. 401-408, DOI:10.1177/002199838602000407.

[11] Chang, F.K., Scott, R.A., Springer, G.S. (1984). Failure of composite laminates containing pin loaded holes-method of solution. Journal of Composite Materials, vol. 18, no. 3, p. 255-278, DOI:10.1177/002199838401800305.

[12] Maksimović, S. (1990). Some computational and experimental aspects of the optimal design process of

http://dx.doi.org/10.1177/002199839502900507

http://dx.doi.org/10.1016/S0266-3538(02)00071-4

http://dx.doi.org/10.1016/S0266-3538(02)00071-4

http://dx.doi.org/10.1177/002199839903302402

http://dx.doi.org/10.1177/002199802761675502

http://dx.doi.org/10.1016/0263-8223(94)00040-9

http://dx.doi.org/10.1177/002199838201600603

http://dx.doi.org/10.1177/002199838602000407

http://dx.doi.org/10.1177/002199838401800305



composite structures. Composite Structures, vol. 16. no. 1-3, p. 237-258, DOI:10.1016/0263-8223(90)90074-O.

[13] Wu, T.J., Hahn, T. (1998). The bearing strength of E-glass/vinyl-ester composites fabricated by Vartm.Composites Science and Technology, vol. 58, no. 9, p. 1519-1529, DOI:10.1016/S0266-3538(97)00180-2.

[14] Icten, B.M., Okutan, B., Karakuzu, R. (2003). Failure strength of woven glass fiber-epoxy composite pinned joint. Journal of Composite Materials, vol. 37, no. 15, p. 1337-1351, DOI:10.1177/0021998303034307.

[15] Okutan, B., Karakuzu, R. (2003). The strength of pinned joints in laminated composite plates. Composites Science and Technology, vol. 63, no. 6, p. 893-905, DOI:10.1016/S0266-3538(02)00313-5.

[16] Whitney, J.M., Nuismer, R.J. (1974). Stress fracture criteria for laminated composites containing stress concentration. Journal of Composite Materials, vol. 8, no. 3, p. 253-265, DOI:10.1177/002199837400800303.

[17] Whitworth, H.A., Othieno, M., Barton, O. (2003). Failure analysis of composite pin loaded joints. Composite Structures, vol. 59, no. 2, p. 261-266, DOI:10.1016/S0263-8223(02)00056-9.

[18] Okutan, B. (2002). The effects of geometric parameters on the failure strength for pin-loaded multi-directional fiber-glass reinforced epoxy laminate, Composites Part B: Engineering, vol. 33, no. 8, p. 567-578, DOI: 10.1016/S1359-8368(02)00054-9.

[19] Djordjevic Z., Maksimovic S., Ilic I., (2011). Dynamics Analysis of Hybrid Aluminum/Composite Shafts, Scientific Technical Review, vol. 58, no. 2, pp. 3-7.

[20] MSC/NASTRAN, Theoretical Manuals (2000), MSC Software, Santa Ana.

[21] Pierron, F., Cerisier, F. (2000). A numerical and experimental study of woven composite pin-joints, Journal of Composite Materials, vol. 34, no. 12, p. 1028-1054.

[22] Maikuma, H., Kobomura, K. (1993). Bearing strength and damage progress for PAN-based and pitch based carbon fiber composites, Journal of Composite Materials, vol. 27, no. 18, p. 1739-1761, DOI:10.1177/002199839302701803.

[23] Okutan, B., Aslan, Z., Karakuzu, R. (2001). A study of the effects of various geometric parameters on the failure strength of pin-loaded wovenglass- fiber reinforced epoxy laminate, Composites Science and Technology, vol. 61, no. 10, p. 1491-1497, DOI: 10.1016/S0266-3538(01)00043-4..

[24] Stamenković D., Maksimović K., Nikolić-Stanojević V., Maksimović S., Stupar S., Vasović I. (2010). Fatigue Life Estimation of Notched Structural Components, Strojniški vestnik - Journal of Mechanical Engineering, vol. 56, no. 12, p. 846-852.

[25] Chen, D.-C., Chen, W.-J., Lin, J.-Y., Jheng, M.-W., Chen, J.-M. (2010). Finite Element Analysis of Superplastic Blow-Forming of Ti-6Al-4V Sheet into Closed Ellip-Cylindrical Die. International Journal of Simulation Modelling, vol. 9, no. 1, p. 17-27, DOI:10.2507/IJSIMM09(1)2.137.

http://dx.doi.org/10.1016/0263-8223(90)90074-O

http://dx.doi.org/10.1016/S0266-3538(97)00180-2

http://dx.doi.org/10.1177/0021998303034307

http://dx.doi.org/10.1016/S0266-3538(02)00313-5

http://dx.doi.org/10.1177/002199837400800303

http://dx.doi.org/10.1016/S0263-8223(02)00056-9

http://dx.doi.org/10.1016/S1359-8368(02)00054-9

http://dx.doi.org/10.1016/S1359-8368(02)00054-9

http://dx.doi.org/10.1177/002199839302701803

http://dx.doi.org/10.1016/S0266-3538(01)00043-4

http://dx.doi.org/10.1016/S0266-3538(01)00043-4

http://dx.doi.org/10.2507/IJSIMM09%281%292.137

Strojniški vestnik - Journal of Mechanical Engineering 58(2012)9Vsebina

Vsebina

Strojniški vestnik - Journal of Mechanical Engineeringletnik 58, (2012), številka 9Ljubljana, september 2012

ISSN 0039-2480

Izhaja mesečno

Razširjeni povzetki člankov

Tomaž Finkšt, Jurij F. Tasič, Marjeta Terčelj-Zorman, Matej Zajc: Obdelava avtofluorescenčnih bronhoskopskih slik v izbranih barvnih prostorih SI 103

Diana Popescu, Cătălin Gheorghe Amza, Dan Lăptoiu, Gheorghe Amza: Model kompetitivne Hopfieldove nevronske mreže za vrednotenje natančnosti vsaditve pedikularnega vijaka SI 104

Frédéric Vignat, Dinh Son Nguyen, Daniel Brissaud: Metoda za določanje vpliva geometrijskih odstopanj na zmogljivost izdelkov SI 105

Bing Li, Jimeng Li, Jiyong Tan, Zhengjia He: Diagnosticiranje napak za triosne vrtalne in rezkalne stroje na osnovi adaptivne stohastične resonance SI 106

Lidija Rihar, Janez Kušar, Stane Gorenc, Marko Starbek: Timsko delo pri sočasnem osvajanju izdelka SI 107Ferhat Dikmen, Meral Bayraktar, Rahmi Guclu: Analiza osi železniških vozil: utrujenostne poškodbe

ter analiza življenjske dobe SI 108Ivana Ilić, Zlatko Petrovic, Mirko Maksimović, Slobodan Stupar, Dragi Stamenković: Računska

metoda za analizo odpovedi mehanskih spojev plastnih kompozitov SI 109

Osebne vestiDoktorske disertacije, diplomske naloge SI 110

*Naslov avtorje za dopisovanje: Univerza v Ljubljani, Fakulteta za strojništvo, Aškerčeva 6, 1000 Ljubljana, Slovenija, [email protected] SI 103

Strojniški vestnik - Journal of Mechanical Engineering 58(2012)9, SI 103 Prejeto: 2012-02-10, sprejeto: 2012-06-07 © 2012 Strojniški vestnik. Vse pravice pridržane.

Obdelava avtofluorescenčnih bronhoskopskih slik v izbranih barvnih prostorih

Finkšt, T. – Tasič, J.F. – Terčelj-Zorman, M. – Zajc, M.Tomaž Finkšt1,* – Jurij F. Tasič2 – Marjeta Terčelj-Zorman3 – Matej Zajc2

1 Univerza v Ljubljani, Fakulteta za strojništvo, Slovenija 2 Univerza v Ljubljani, Fakulteta za elektrotehniko, Slovenija

3 Univerzitetni klinični center Ljubljana, Klinični oddelek za pljučne bolezni in alergijo, Slovenija

Na področju obdelave slik v zadnjih letih za razpoznavanje objektov vse pogosteje uporabljamo barvno informacijo. Pri obdelavi slik je razčlenjevanje ključen postopek, ki je pomemben za pravilno izločanje, analizo in interpretacijo slikovne vsebine. Informacijska vsebina medicinskih slik je ključnega pomena za odkrivanje in razumevanje normalnih in bolezenskih stanj človeškega organizma. Kakovost slikovno podprtih medicinskih preiskav je v veliki meri odvisna od tehnike zajema slik, interpretacije vsebine slik ter od raziskovalnega in medicinskega okolja, ki vzpodbujata zajemanje slik in njihovo uporabo.

Članek obravnava uporabo razčlenjevanja pri analizi avtofluorescenčnih medicinskih slik v dveh barvnih prostorih. Podrobno seznanja s problemi, na katere naletimo pri načrtovanju takšnih algoritmov. Načrtovanje od konteksta odvisnih medicinskih razčlenjevalnih algoritmov ni možno brez tesnega sodelovanja specialistov, kar se kaže predvsem pri vrednotenju dobljenih rezultatov. Načrtovanje takšnih algoritmov ima predpisano obliko, in sicer se je treba najprej seznaniti z medicinsko domeno, nato izbrati ustrezno razčlenjevalno metodo, končno pa je treba dobljene razčlenjene rezultate še ovrednotiti.

Glede na rezultate nas je zanimalo, kako uspešna je izbrana razčlenjevalna metoda pri tem realnem problemu. Pri izračunu smo upoštevali dobljena področja neposredno po razčlenjevanju, področja pa so bile razpoznana kot spremembe na osnovi kriterijev o presekih. Te rezultate smo nato primerjali s pravilnimi rezultati, ki nam jih je posredoval specialist, ter tako izračunali učinkovitost našega sistema.

Predlagani algoritem se je izkazal za primernega, saj sta se površina ročno razčlenjene slike in slike, razčlenjene v prostoru HSV, v povprečju ujemali 88-odstotno, v prostoru RGB pa je bilo ugotovljeno 84-odstotno ujemanje. Sledi, da je prostor HSV v našem primeru primernejši za razčlenjevanje avtofluorescenčnih slik. Predstavljeni postopek razčlenjevanja fluorescenčnih slik v bronhoskopiji na osnovi upoštevanja barvnih komponent v različnih barvnih prostorih se je izkazal za dobrega. Barvne slike omogočajo zanesljivejše razčlenjevanje prizorov kot sivinske. Uporaba značilke odtenka barve H pogosto zadošča za dobro razčlenjevanje, lahko pa se pojavijo težave z njeno stabilnostjo pri majhnih intenzitetah.

Z omenjenimi postopki lahko olajšamo delo zdravnikom specialistom pri razbiranju ustrezne informacije iz avtofluorescenčnih slik. V nadaljevanju bo treba pridobiti več različnih slik pacientov, tako zdravih kot bolnih, in seveda povratno informacijo od zdravnikov specialistov. Najbolje bi bilo, da bi vsi specialisti razčlenili isto medicinsko sliko in bi iz teh rezultatov izračunali povprečje, ki bi ga vzeli za referenčni podatek pri primerjavi rezultatov. S takšnim pristopom se minimizira variabilnost, ki jo vnašajo specialisti. Vsekakor pa bi morali za vsako zajeto sliko dobiti tudi diagnozo o vzetem vzorcu, ki jo zajamejo za vsako AFB sliko posebej.Keywords: barvni prostori, obdelava slik, zajem slik, razčlenjevanje slik, ugotavljanje obrisov, avtofluorescenčna bronhoskopija

Strojniški vestnik - Journal of Mechanical Engineering 58(2012)9, SI 104 Prejeto: 2011-10-17, sprejeto: 2012-04-04 ©2012Strojniškivestnik.Vsepravicepridržane.

*Naslovavtorjezadopisovanje:PolitehničnauniverzavBukarešti,SplaiulIndependentei,313,sektor6,060042,Bukarešta,Romunija,[email protected] 104

Model kompetitivne Hopfieldove nevronske mreže za vrednotenje natančnosti vsaditve pedikularnega vijaka

Popescu, D. – Amza, C. Gh. – Lăptoiu, D. – Amza, Gh.Diana Popescu1,* – Cătălin Gheorghe Amza1 – Dan Lăptoiu2 – Gheorghe Amza1

1 Politehnična univerza v Bukarešti, Romunija2 Colentina klinični center, Romunija

V članku je predstavljena uporaba algoritma za segmentacijo rentgenskih posnetkov na osnovi modela kompetitivne Hopfieldove nevronske mreže (CHNN) pri vrednotenju natančnosti vstavitve pedikularnih vijakov v lumbalno hrbtenico.

Natančnost vstavitve pedikularnih vijakov se v praksi izvaja vizualno v dveh ravninah (sagitalni in transverzalni) s pomočjo posnetkov pooperativne računalniške tomografije ali radiografije. V literaturi pa kljub uporabi naprednih medicinskih tehnik slikanja poročajo o napakah pri vsaditvi vijakov v višini do 13 %, zaradi česar je pomemben razvoj sistema za usposabljanje kirurgov. Razvoj sistema za usposabljanje, ki je opisan v tem članku, je novost na tem področju, predlagani pristop pa je zasnovan na uporabi spremenjenega algoritma CHNN za segmentacijo rentgenskih posnetkov, ki avtomatizira vrednotenje natančnosti vsaditve pedikularnega vijaka.

Raziskava je bila osredotočena na nabor 34 različnih posnetkov več modelov lumbalnih vretenc z ločljivostjo slike 760×520 slikovnih točk. Algoritem za segmentacijo mora biti za aplikacijo, ki je opisana v tem članku, obvladljiv z vidika računskih operacij in neodvisen od velikosti posnetka. Analiziranih je bilo več klasičnih algoritmov za segmentacijo in tehnik zaznavanja robov (Sobel, Canny, Prewwit, Robert, Laplace, določitev pragov Otsu, Watershed, SIOX, Shanbhag, Huang, gradientna itd.) z namenom ugotavljanja, ali so primerni za segmentacijo rentgenskih posnetkov modelov vretenc. Analiza je pokazala, da enostavni in večkratni algoritmi določitve pragov ne zadostujejo za pravilno segmentacijo rentgenskih posnetkov vretenc. Kot neprimerni za aplikacijo so se izkazali tudi algoritmi segmentacije posnetkov na osnovi umetne inteligence (fuzzy C-means, mehko gručenje, Hopfield-Koss itd.), to pa je privedlo do razvoja in programske implementacije algoritma CHNN za segmentacijo posnetkov.

Algoritem poišče vretenca in vijak kot posamezne objekte ter opravi izračune, s katerimi ugotovi odstopanje osi vijaka od osi pedikla kot idealne trajektorije. Vrednost odstopanja med osjo vijaka in osjo pedikla se izračuna po tem, ko se na posnetku, segmentiranem s CHNN, izvede več geometrijskih izračunov: vretenca se poiščejo kot posamezni objekti z algoritmom za povratno sledenje, nato se izračuna njihovo težišče in iz njega se izrišeta dve osi.

Enaki izračuni se opravijo tudi na samih vijakih.Predlagani sistem predstavlja implementacijo splošnega sistema za zaznavanje vzorcev ter vključuje naslednje

enote/sisteme: 1. sistem za zajem dvopasovnega energetskega posnetka vretenc, 2. sistem za predobdelavo posnetkov, ki rentgenske posnetke izboljša za vmesno obdelavo (povečanje kontrasta, odstranitev ozadja, odstranitev šuma itd.), 3. sistem za segmentacijo posnetkov, ki rentgenski posnetek razdeli v smiselne razrede za nadaljnje višjenivojske preglede z modulom HNN. Sistem omogoča avtomatsko izločitev pedikularnega vijaka kot posebnega objekta na rentgenski sliki, in 4. visokonivojski sistem zaznavanja, ki izračunava odstopanje pedikularnega vijaka od idealnega položaja.

Kriteriji vrednotenja so odvisni od razlike med položajem osi vretenca in osi vijaka, ki je primerjana s teoretično sprejemljivimi vrednostmi (razredi) na osnovi varnostnih meja iz literature. Ti razredi so: I. penetracija pod 2 mm (sprejemljivo), II. penetracija od 2 do 4 mm (zahteva nastavitev položaja vijaka), III. penetracija nad 4 mm (nesprejemljivo).Ključne besede: medicinski posnetki, pedikularni vijak, Hopfieldova nevronska mreža, lumbalna vretenca

*Naslov avtorja za dopisovanje: Tehnična univerza v Danangu, Danang, Vietnam, [email protected] SI 105


Metoda za določanje vpliva geometrijskih odstopanj na zmogljivost izdelkovVignat, F. – Nguyen, D.S. – Brissaud, D.

Frédéric Vignat1 – Dinh Son Nguyen2,* – Daniel Brissaud1

1 Laboratorij G-SCOP, Univerza v Grenoblu, Francija 2 Tehnična univerza v Danangu, Vietnam

Robustnost je ključni dejavnik snovanja izdelkov ob upoštevanju negotovosti pri lastnostih materiala, proizvodnih operacijah in delovnem okolju. Vsak izdelek potuje skozi številne faze življenjskega cikla, od konstruktorjevih možganov do uporabnikovih rok. Variabilnost, ki je prisotna v vsaki fazi življenjskega cikla izdelka, očitno vpliva na njegove zmogljivosti. Realnih zmogljivosti izdelka, ki se razlikujejo od načrtovanih, zato ni mogoče verificirati. Obstaja tveganje, da zasnovani izdelek ne bo v celoti izpolnjeval zahtev strank in končnih uporabnikov.

Plod akademskih raziskav so številne metode in orodja za upravljanje z vplivi geometrijske variabilnosti na zasnovo izdelkov. Te raziskave pa se ukvarjajo samo s fazami življenjskega cikla izdelka, ki imajo opraviti s proizvodnjo ali montažo. Vplivi virov variabilnosti na zmogljivosti izdelka med življenjskim ciklom izdelka niso upoštevani zlasti če niso znane matematične povezave med zmogljivostmi izdelka in parametri virov variabilnosti. Te povezave v mnogih primerih niso ugotovljene in za določanje zmogljivosti se uporabljajo numerične rešitve. V članku je za aproksimativno določanje zveze med zmogljivostmi izdelka in parametri izdelka predlagano numerično reševanje v kombinaciji z načrtovanjem eksperimenta.

Pri preučevanju vpliva geometrijske variabilnosti na zmogljivost izdelka je treba upoštevati pomembne vidike:• Kako ugotoviti razmerje med zmogljivostmi in geometrijskimi odstopanji izdelka?• Kako upravljati z vzroki in posledicami teh odstopanj v fazi konstrukcije?

V članku je kot odgovor na prvo vprašanje predlagana metoda, ki omogoča določanje razmerij med zmogljivostmi in geometrijskimi odstopanji. Nastala in zbrana geometrijska odstopanja so popisana z geometrijskim modelom odstopanj. Predlagan je tudi delni odgovor na drugo vprašanje za identifikacijo in razvrščanje vpliva parametrov odstopanj.

Predlagane so tri različne strategije načrtovanja eksperimentalne metode, odvisno od zahtevnosti obravnavanega problema. Faktorski in Taguchijev načrt eksperimenta sta že dobro znana, v članku pa je predlagana nova strategija na osnovi naključne izbire vzorca načrta.

Izbor med tremi pristopi za določanje povezav med zmogljivostmi in geometrijskimi odstopanji izdelka je odvisen od zahtev glede natančnosti, časa in stroškov. Faktorska in Taguchijeva metoda načrtovanja sta primerni, če je ekspertno znanje učinkovito in število faktorjev ni preveliko. Naključni načrt se izbere, ko je težko določiti dejavnike z velikim vplivom na zmogljivost izdelkov, je število dejavnikov precej veliko ter je izračun zmogljivosti zahteven in časovno zamuden.

Povezava med zmogljivostmi izdelka in parametri geometrijskih odstopanj je določena z eno od treh različnih strategij. Slika populacije izdelanega izdelka je izračunana iz geometrijskega modela odstopanj (GDM) po metodi simulacij Monte-Carlo. Na osnovi rezultatov simulacije Monte Carlo in določene povezave je izračunana slika zmogljivosti populacije virtualnih izdelkov. Konstruktor lahko nato na osnovi rezultatov simulacije Monte Carlo identificira in klasificira vpliv vsakega parametra vira variabilnosti na zmogljivost izdelka, kakor tudi pripadajočo zmogljivost vsakega virtualnega izdelka. Končno je mogoče določiti še varianco zmogljivosti izdelka glede na variabilnost parametrov odstopanj. Robustno zasnovo je torej mogoče določiti z iskanjem minimuma variance variabilnosti zmogljivosti.Ključne besede: življenjski cikel izdelka, zmogljivost izdelka, model geometrijskih odstopanj, simulacija proizvodnje, robustna zasnova, simulacija Monte Carlo


*Naslovavtorjazadopisovanje:Državnilaboratorijzainženiringproizvodnihsistemov,UniverzaXi’anJiaotong,Xi’an,Kitajska,[email protected] 106

Diagnosticiranje napak za triosne vrtalne in rezkalne stroje na osnovi adaptivne stohastične resonance

Li, B. – Li, J. – Tan, J.Y. – He, Z.J.Bing Li1,* – Jimeng Li1 – Jiyong Tan2 – Zhengjia He1

1 Državni laboratorij za inženiring proizvodnih sistemov, Univerza Xi’an Jiaotong, Kitajska 2 29. institut pri China Electronics Technology Group Corporation, Kitajska

Triosni vrtalni in rezkalni stroji so ključni del opreme v sodobni proizvodni industriji. Napake komponent, ki se pojavljajo med obratovanjem, zmanjšujejo natančnost obdelave. Zaradi vplivov prenosne poti, prenosnega medija in okolja pa je le težko zaznati značilne napake, ki so skrite v močnem šumu. Obstajajo številne konvencionalne metode za obdelavo šibkih signalov, vendar pa večina teh metod uporablja filtriranje ali prikrivanje šuma. Čeprav se šum na ta način zmanjša, pa lahko uporabni signal postane šibkejši ali pa je celo uničen. Stohastična resonanca (SR) šuma ne odpravlja, ampak omogoča zaznavanje signala z izkoriščanjem šuma za ojačenje šibkih signalov v nelinearnih dinamičnih sistemih.

V članku je predstavljena tehnika obdelave signalov s stohastično resonanco za iskanje značilnih napak, ki zmanjšujejo natančnost obdelave z vrtalnimi in rezkalnimi stroji. Tehnika daje časovno-frekvenčno porazdelitev vibracij s prilagodljivo natančnostjo. Čeprav daje stohastična resonanca boljše rezultate zaznave, je inženirska uporaba SR še vedno v veliki meri omejena z izbiro pravega merilnega indeksa in optimalnih parametrov sistema.

Za optimalno izbiro parametrov sistema je uveden uteženi indeks sploščenosti. Indeks sploščenosti kot brezdimenzijski indeks se pogosto uporablja za kvantitativno merjenje vplivnih komponent signala vibracij s pomočjo značilne občutljivosti vplivne komponente. Če se za vrednotenje resonančnega učinka SR za zaznavanje vplivnega signala uporablja iskanje maksimuma indeksa sploščenosti, pa ni mogoče zaznati nekaterih vplivnih komponent z majhnimi amplitudami. Koeficient korelacije lahko odraža podobnost dveh različnih signalov in je neodvisen od amplitude signalov. Podobnost dveh različnih signalov je zato mogoče kvantitativno okarakterizirati z absolutno vrednostjo korelacijskega koeficienta. Uteženi indeks sploščenosti (WK), ki združuje prednosti indeksa sploščenosti in korelacijskega koeficienta, ne prevzema le občutljivosti indeksa sploščenosti za vplivne komponente, temveč daje tudi kar največjo podobnost rezultata zaznave in originalnega signala ter zajame tudi manj vplivne komponente z majhnimi amplitudami. Uteženi indeks sploščenosti omogoča adaptivno iskanje optimalnih parametrov sistema. Metode SR omogočajo ojačenje šibkih signalov do določene mere, a če je razmerje med signalom in šumom premajhno, učinek zaznavanja enojne SR še vedno ni zadovoljiv. Kaskadna SR je izboljšana metoda SR, pri kateri dva zaporedna bistabilna sistema še dodatno oslabita visokofrekvenčno trepetanje in zgladita izhodno valovno obliko v časovni domeni za večje razmerje med signalom in šumom kot pri enojni SR. Kaskadna SR se uporablja za obdelavo signala vibracij, pridobljenega iz vrtalnega in rezkalnega stroja za še dodatno izboljšanje učinka zaznavanja SR.

Na osnovi uteženega indeksa sploščenosti in kaskadne SR je podan algoritem adaptivne stohastične resonance (AdSR). Metoda adaptivne stohastične resonance se uspešno uporablja za iskanje značilnih napak, ki zmanjšujejo natančnost obdelave z vrtalnimi in rezkalnimi stroji. Rezultati zaznave kažejo, da je AdSR boljše orodje za iskanje značilnih napak od Fourierjeve analize, saj zmanjšuje vpliv visokofrekvenčnega šuma zaradi vibracij pri obdelavi.

Predlagana metoda ne omogoča le adaptivne izbire optimalnih parametrov sistema z uteženim indeksom sploščenosti, temveč tudi gladi izhodno valovno obliko signala, perioda fluktuacij pa postane s sekundarno uporabo šuma očitnejša.

Zaključek obravnava, da je obraba polžastega zobnika vrtljive mize glavni razlog za poslabšanje natančnosti pri triosnih vrtalnih in rezkalnih strojih. Prihodnje raziskave bodo usmerjene v možnosti sprotnega diagnosticiranja napak med obratovanjem vrtalnih in rezkalnih strojev z metodo AdSR. Čeprav je uteženi indeks sploščenosti primernejši za vrednotenje učinka zaznave vplivnega signala, ga je treba dodatno preučiti in izboljšati, saj je izbor eksponenta v uteženem indeksu sploščenosti še vedno odvisen od izkušenj raziskovalcev.Ključne besede: stohastična resonanca, adaptivna, kaskada, uteženi indeks sploščenosti, diagnostika napak, vrtalni in rezkalni stroj

*Naslov avtorje za dopisovanje: Univerza v Ljubljani, Fakulteta za strojništvo, Aškerčeva 6, 1000 Ljubljana, Slovenija, [email protected] SI 107


Timsko delo pri sočasnem osvajanju izdelkaRihar, L. – Kušar, J. – Gorenc, S. – Starbek, M.

Lidija Rihar1 – Janez Kušar1,* – Stane Gorenc2 – Marko Starbek1 1 Univerza v Ljubljani, Fakulteta za strojništvo, Slovenija

2 PLASTA d.o.o, Slovenia

V članku so prikazani časovni in stroškovni prihranki, ki jih zagotavlja prehod iz sekvenčnega na sočasni način osvajanja izdelka. Prehod na sočasno osvajanje izdelka ni možen brez vnaprej dobro organiziranega timskega oziroma virtualnega timskega dela.

Za sekvenčno osvajanje izdelka je značilno zaporedno izvajanje stopenj procesa osvajanja izdelka. Opazovana stopnja procesa osvajanja izdelka se lahko prične šele, ko se predhodna stopnja v celoti konča. Informacije o opazovani stopnji procesa osvajanja izdelka se gradijo postopno in so ob koncu opazovane stopnje popolnoma izgrajene ter se posredujejo naslednji stopnji.

V nasprotju s sekvenčnim osvajanjem izdelka pa je za sočasno osvajanje izdelka značilno vzporedno izvajanje stopenj procesa osvajanja izdelka. Tu se opazovana stopnja procesa osvajanja izdelka lahko prične že pred koncem njej predhodne stopnje. Informacije o opazovani stopnji procesa osvajanja izdelka se gradijo postopno in se že med gradnjo posredujejo naslednji stopnji.

S prehodom iz sekvenčnega na sočasno osvajanje izdelka se dosežejo bistveno krajši čas in nižji stroški osvajanja izdelka, kakor tudi večja kakovost in vitka organiziranost izvedbe procesa sočasnega osvajanja izdelka.

Pri sekvenčnem osvajanju izdelka stroški definiranja izdelka zaradi zaporedne izvedbe aktivnosti definiranja izdelka (marketing, koncept izdelka, razvoj izdelka, izdelava konstrukcijske dokumentacije, gospodarjenje z materialom) naraščajo enakomerno, stroški proizvodnje pa naraščajo hitro zaradi dolgih iteracijskih zank izvedbe sprememb oziroma odstranjevanja napak in neskladnosti.

Pri sočasnem osvajanju izdelka pa so stroški definiranja izdelka zaradi vzporedne izvedbe aktivnosti definiranja izdelka zaradi intenzivnega vložka dela veliko večji kot pri sekvenčnem osvajanju, stroški proizvodnje pa so bistveno nižji kot pri sekvenčnem osvajanju zaradi kratkih iteracijskih zank izvedbe sprememb oziroma odstranjevanja napak in neskladnosti.

Rezultati predstavljene raziskave dokazujejo potrebo po različnih načinih oblikovanja strukture timov pri sočasnem osvajanju izdelka. Ugotovljeno je bilo, da je za manjša podjetja primerna dvoravninska struktura timov, kjer je na prvi ravnini jedrni tim, na drugi ravnini pa več projektnih timov zank sočasnega osvajanja izdelka. Predstavljen je postopek oblikovanja jedrnega in projektnih timov za izvedbo zank sočasnega osvajanja izdelka, ki temelji na matriki odgovornosti (odgovornosti so definirane po metodi 1-3-9) in izračunani skupni stopnji intenzitete odgovornosti i-tega člana tima v j-ti zanki sočasnega osvajanja izdelka.

Za zagotavljanje uspešnega dela članov jedrnega in projektnih timov so predlagana potrebna komunikacijska orodja ter oblikovana komunikacijska matrika, ki določa način posredovanja informacij za izvedbo aktivnosti projekta sočasnega osvajanja izdelka. Komunikacijska matrika je osnova za izbiro ustreznih komunikacijskih orodij za podporo timskemu oziroma virtualnemu timskemu delu, kakor tudi za določanje informacijskih povezav pri izbranem komunikacijskem sistemu podjetja.

Predlagana metodologija oblikovanja timov oziroma virtualnih timov ter komunikacijske matrike sočasnega osvajanja izdelka, ki predstavlja izvirno rešitev na področju projektnega vodenja, razširjenega z elementi sočasnega inženiringa, je bila preizkušena na primeru osvajanja komponente avtomobila.

Dobljeni rezultati so namenjeni tako raziskovalcem s področja osvajanja izdelkov, kot tudi projektnim timom, ki se ukvarjajo z osvajanjem izdelkov v realnem okolju.

Nadaljnje delo na problematiki sočasnega osvajanja izdelka bo usmerjeno v popis celotnega procesa sočasnega osvajanja izdelka s pomočjo orodja za modeliranje in prenovo procesov ARIS.Ključne besede: osvajanje izdelka, stezno-zankasti proces, virtualni tim, komunikacijska orodja, komunikacijska matrika


*Naslovavtorjezadopisovanje:Fakultetazastrojništvo,UniverzaYildiz,BarbarosBulvari,34349,Besiktas,Istanbul,Turčija,[email protected] 108

Analiza osi železniških vozil: utrujenostne poškodbe ter analiza življenjske dobe

Dikmen, F. – Bayraktar, M. – Guclu, R.Ferhat Dikmen – Meral Bayraktar* – Rahmi Guclu

Fakulteta za strojništvo, Univerza Yildiz, Turčija

Problem utrujenostnega zloma mehanskih komponent je pomemben za raziskovalce na področju konstrukcije v strojništvu. Pomen te problematike je še posebej velik pri železniških vozilih, zlasti pri komponentah, kjer lahko v primeru odpovedi pride do smrtonosnih posledic. Ta problem je bil zaradi nedavnih zlomov osi tirnih vozil pri turških železnicah (TCDD) postavljen v središče pozornosti in je tudi motivacija za to študijo.

Študija obravnava odpoved osi vagona mestnega vlaka TCDD ALS-THOM 8000, ki vozi na relaciji Sirkeci – Halkali. Zlom osi je klasičen problem utrujanja zaradi velikih upogibnih napetosti, ki se spreminjajo med nategom in tlakom. Cilj tega članka je obravnava zanesljivosti osi ter primerjava z dejansko življenjsko dobo osi do zloma. Obravnava se začne z določitvijo kritičnega prereza osi.

Za določitev pričakovanega mesta zloma osi so bile izračunane napetosti v vseh kritičnih prerezih z običajnimi trdnostnimi izračuni. Ugotovljeno je bilo, da je mesto zloma med kolesom in zobnikom. Mesto zloma kaže tudi slika polomljene osi. Nato je za najbolj kritično mesto določena varna življenjska doba pri različnih obratovalnih pogojih.

Raziskava se je nadaljevala z določitvijo obremenitvenih primerov za vagon na osnovi statističnih podatkov o številu potnikov. Najprej je bila ugotovljena najkrajša življenjska doba osi pri polni obremenitvi, nato pa so bile izračunane efektivne življenjske dobe po Palmgren-Minerjevem teoremu, ki upošteva kumulativno odpoved za realna obremenitvena stanja v primeru različnih porazdelitev.

Analiza kumulativnih utrujenostnih poškodb ima ključno vlogo pri napovedovanju življenjske dobe komponent in konstrukcij, ki so izpostavljene zgodovinam obremenitev. Teorijo kumulativnih poškodb je leta 1920 prvič uporabil Šved A. Palmgren za napovedovanje življenjske dobe valjčnih ležajev. Sledil mu je B. F. Langer s splošnim pristopom. Uporaba teorije pa se ni razširila, dokler se ni pojavila v študiji M. A. Minerja iz leta 1945. Obravnava kumulativnih utrujenostnih poškodb je bila od tedaj deležna vse več pozornosti. Ta linearna teorija je znana kot Palmgren-Minerjeva hipoteza oz. linearno pravilo poškodb.

Cilj članka je predstavitev približnih izračunov časa loma osi ter pogojev, ki vplivajo na lom. Do odpovedi lahko pride pri istem materialu ob različnih časih, kot je razvidno iz Wöhlerjevih diagramov. Ti diagrami so izdelani za raznovrstne materiale in različne vrednosti zanesljivosti pri različnih laboratorijskih pogojih. V okviru te študije je bil pridobljen Wöhlerjev diagram za material osi 25CrMo4, ki vključuje različne vrednosti zanesljivosti ob upoštevanju standardne deviacije.

Najkrajša življenjska doba je 1,4 leta, ob predpostavki, da vagon vozi pod polno obremenitvijo. Ob upoštevanju statističnih podatkov in različnih porazdelitev je bila opravljena analiza z realnimi pogoji obremenitve. Palmgren-Minerjeva teorija kumulativnih poškodb daje za različne porazdelitve efektivno življenjsko dobo 12,1, 15,8 in 9,5 let. Očitno je, da je življenjska doba pri zahtevnih obremenitvah krajša od življenjske dobe pri lažjih obremenitvah. Ugotovljena je bila povprečna vrednost 12 let.

Dejstvo je, da obratovalni pogoji vplivajo na dejansko življenjsko dobo. Če je hitrost vagona manjša do 13 m/s, je življenjska doba neskončna celo pri polni obremenitvi. Pri hitrosti 40 m/s pa življenjska doba ni daljša od dveh let, tudi če v vagonu ni potnikov. Očitno je tudi, da je efektivna življenjska doba različna za različne vozne hitrosti. Pri hitrosti, manjši od 15 m/s, je življenjska doba neskončna, pri hitrosti 32 m/s pa znaša približno eno leto. Pri hitrosti 22 m/s, ki v praksi ni nikoli presežena, je življenjska doba 6,5 leta, pri hitrosti 17 m/s pa več kot 50 let. Izračunane življenjske dobe so bile primerjane z življenjsko dobo realnih poškodovanih osi, pri čemer je bilo ugotovljeno dobro ujemanje.

Izsledki raziskave med drugim kažejo, da so potrebne nove konstrukcijske spremembe. Neskončna življenjska doba bi bila dosegljiva s povečanjem premera osi iz 171 na 173 mm ter s spremembo površinske hrapavosti osi. Raziskava je pokazala, da bi bilo za izdelavo osi mogoče uporabiti tudi druge materiale z večjo trdnostjo, npr. 42CrMo4 ali C60.Ključne besede: os tirnega vozila, utrujanje, kumulativna odpoved, življenjska doba, zanesljivost

*Naslov avtorje za dopisovanje: Termoelektro d.o.o., Uralska 9, 11060 Beograd, Srbija, [email protected] SI 109


Računska metoda za analizo odpovedi mehanskih spojev plastnih kompozitov

Ilić, I. – Petrovic, Z. – Maksimović, M. – Stupar, S. – Stamenković, D.Ivana Ilić1 – Zlatko Petrovic2 – Mirko Maksimović3 – Slobodan Stupar2 – Dragi Stamenković4,*

1Vojno-tehnični institut, Srbija

2Fakulteta za strojništvo, Srbija

3Vodovod, Srbija

4Termoelektro, Srbija

Članek obravnava računsko metodo za analizo odpovedi plastnih kompozitov v spoju vrste luknja-čep. Raziskava je bila usmerjena v razvoj zanesljivega računskega postopka za analizo obremenitve inicialne odpovedi spojev vrste luknja-čep v plastnih kompozitnih konstrukcijah. Trdnost mehanskih spojev v kompozitnih konstrukcijah so preučevali številni raziskovalci. Trdnost takšnih spojev je odvisna od številnih dejavnikov, med drugim od geometrije spoja, orientacije vlaken, vrstnega reda zlaganja itd. Velik del raziskav, ki so bile opravljene na mehanskih spojih, je bil osredotočen na eksperimentalno in numerično ugotavljanje vpliva geometrijskih dejavnikov na trdnost spojev. Opravljena je bila numerična študija za določanje odpovedi mehanskih spojev laminiranih kompozitnih materialov, ojačenih z vlakni.

Cilj tega dela je numerična preučitev vedenja spojev luknja-čep pri kompozitih vrste grafit-epoksidna smola, pri čemer je bila posebna pozornost posvečena občutljivosti modela na različne geometrijske dimenzije. Za določanje porazdelitve napetosti v kompozitnih ušesih je bila uporabljena dvodimenzionalna kontaktna metoda končnih elementov. Za določanje obremenitve pri odpovedi in načina odpovedi je bil izbran model napovedovanja inicialnih poškodb s Tsai-Wujevimi odpovednimi pogoji. Analiza vključuje Chang-Scott-Springerjev model karakteristične krivulje in analizo po dvodimenzionalni kontaktni metodi končnih elementov za vrednotenje porazdelitve napetosti okrog luknje za pritrdilni element.

Za določanje odpovedi spoja je uporabljena kombinacija Chang-Scott-Springerjevega modela karakteristične krivulje in Tsai-Wujevega kriterija inicialne odpovedi. V delu je posebna pozornost posvečena porazdelitvi obremenitev na spoju vrste luknja-čep ter vplivu porazdelitve na raven obremenitev in mesto odpovedi. Analize inicialne odpovedi so bile pri prejšnjih raziskavah izvedene s kosinusno porazdelitvijo v mehanskem spoju ušes in čepa. V tem delu je analiziran model ušesa/čepa po kontaktni metodi končnih elementov.

V članku je podana analiza problema mehanskih spojev laminiranih kompozitnih plošč s tornimi kontaktnimi pogoji. Uporabljen je zakon Coulombovega trenja, omejitve kontakta pa so vključene z razširjenimi metodami notranjih kazni. Za obravnavo nediferencialnega člena zaradi Coulombovega trenja je bil uporabljen perturbirani variacijski princip. Rezultati izračunov so primerjani z rezultati eksperimentov.

Preučen je tudi vpliv vrstnega reda zlaganja plasti kompozitnega materiala z luknjami za čepe. V delu je posebna pozornost posvečena načinu odpovedi in modalni analizi pri kompozitih z vrstnim redom zlaganja [0/(±45)3/903]S. Rezultati izračunov so bili primerjani z razpoložljivimi rezultati eksperimentov in ugotovljena je bila dobra korelacija med rezultati izračunov in eksperimentov.

Rezultati kažejo, da lahko predlagana numerična metoda na osnovi kontaktne metode končnih elementov za analizo napetosti v povezavi s Tsai-Wujevim kriterijem odpovedi za analizo inicialne odpovedi napove trdnost kompozitnih spojev vrste luknja-čep z največ 5 odstotnim odstopanjem glede na rezultate preizkusov.Ključne besede: kompoziti, analiza odpovedi, MKE, kontaktni problem, spoj luknja-čep, indeks odpovedi

Strojniški vestnik - Journal of Mechanical Engineering 58(2012)9, SI 110-114Osebne objave

SI 110

Doktorske disertacije, magistrska dela, specialiostična dela, diplomske naloge

DOKTORSKE DISERTACIJE

Na Fakulteti za strojništvo Univerze v Ljubljani so z uspehom obranili svojo doktorsko disertacijo:

dne 2. julija 2012 Uroš LESKOVŠEK z naslovom: »Sočasni prenos toplote in snovi v vlaknasti toplotni izolaciji s tesnimi robnimi površinami« (prof. dr. Sašo Medved);

V izolacijskih gradnikih ovoja stavb s sendvič strukturo je vlaknasta toplotna izolacija obdana s slojema pločevine. Zaradi specifičnih razlogov se v vlaknasti toplotni izolaciji lahko pojavi voda, sušenje izolacijskega jedra pa zaradi tesnih robnih površin večinoma ni mogoče. V doktorskem delu so predstavljene stacionarne in nestacionarne raziskave sočasnega prenosa toplote in snovi v vlaknasti toplotni izolaciji, ki je obdana s tesnimi robnimi površinami. Prenos toplote in snovi smo obravnavali eksperimentalno ter z analitičnim in z numeričnim modelom. Nestacionarni numerični model, ki smo ga razvili, smo v širokem območju temperatur in ploskovnih gostot vode v izolacijskem jedru verificirali z laboratorijskimi ter z v naravnem okolju izvedenimi eksperimenti. Pri periodičnem spreminjanju robnih temperatur smo prenos toplote in snovi parametrično opredelili. Razvili smo neporušitveno metodo za določanje ploskovne gostote vode v vlaknasti toplotni izolaciji, za obravnavane navlažene gradnike pa smo nadgradili tudi standardizirano metodo vrednotenja dinamičnih toplotnih lastnosti elementov ovoja stavb;

dne 4. julija 2012 Jurij ŽUMER z naslovom: »Karakterizacija kinematičnih parametrov sistema togih teles z uporabo inercijskih mikro-elektro-mehanskih sistemov« (mentor: prof. dr. Miha Boltežar, somentor: doc. dr. Janko Slavič);

Raziskovalno delo obravnava popis kinematičnih parametrov sistema togih teles z uporabo inercijskih MEMS zaznaval. Delo se osredotoča na iskanje rešitev za zmanjšanje napake pri oceni parametrov, ki nastane zaradi determinističnih in naključnih nestabilnosti zaznaval. V tem okviru se medsebojno povezujejo kinematični modeli sistema togih teles in modeli odziva inercijskih zaznaval. Predstavljeni so tako prostorsko in časovno neomejeni kot omejeni primeri kinematike sistema togih teles. V okviru predlaganih rešitev je predstavljen popis nelinearnega napetostnega odziva nizkoamplitudnega troosnega MEMS pospeškomera za izboljšanje učinkovitosti rekonstrukcije kretenj z računalniškim pisalom. Delo analizira opazovanje cestnega vozila v prostoru in v okviru razširjenega Kalmanovega filtra predlaga

dodatne opazovalne korake za zmanjšanje napake velikosti hitrosti zgolj s pomočjo inercijskih zaznaval. V primeru prostorsko omejenega sistema togih teles s kinematičnimi vezmi je predstavljena formulacija uporabe poljubnega števila inercijskih enot na posameznem togem telesu za prikaz zmanjšanja napake, ki izvira iz nenatančnega pozicioniranja inercijskih enot. Vsi predstavljeni koraki so eksperimentalno validirani;

dne 29. avgusta 2012 Aleš HANČIČ z naslovom: »Določitev in optimizacija mehanskih lastnosti bio-polimernih kompozitnih materialov s pomočjo mikromehanskih metod« (mentor: prof. dr. Karl Kuzman, somentor: prof. dr. Igor Emri);

Napovedovanje mehanskih lastnosti bio-polimernih kompozitov z matematičnimi metodami je bilo do sedaj slabo raziskano. Predstavljeno doktorsko delo obravnava kompozitne materiale, ojačane z celuloznimi vlakni, njihove mehanske lastnosti ter analizira uporabnost matematičnih mikromehanskih metod. Pripravljeni so bili trije kompoziti z različno vsebnostjo celuloznih vlaken, in sicer z 20, 35 in 50% masnega deleža, ter blok- kopolimerom polipropilen (PP) kot matrico kompozita. Za določitev tehnološkega okna za nove kompozitne materiale smo preučili vpliv temperature taline. Ugotovljeno je bilo, da temperatura taline nima pomembnega vpliva na modul elastičnosti ter na maksimalno napetost.

Preučen je bil vpliv vsebnosti celuloznih vlaken, temperature okolice in hitrosti deformacije na modul elastičnosti, maksimalno napetost in raztezek ob pretrgu. Ugotovljeno je bilo, da lahko prisotnost celuloze poveča modul elastičnosti in maksimalno napetost do 300% od osnovne vrednosti. Temperatura linearno zmanjšuje modul elastičnosti ter maksimalno napetost.

Rezultate nateznih preizkusov smo primerjali z različnimi matematičnimi modeli. Za primerjavo sta bila uporabljena dva matematična analitična mikromehanska modela, in sicer Mori-Tanaka metoda (MTM) in Posplošena metoda celic (PMC). Poleg mikromehanskih modelov je bila za napovedovanje mehanskih lastnosti kompozitnih materialov uporabljena tudi skupina enačb, znana kot Pravilo zmesi (PZ). Ugotovljeno je bilo, da lahko s pomočjo vseh treh modelov predvidimo modul elastičnosti in maksimalno napetost z natančnostjo manj kot ±60% eksperimentalne vrednosti, medtem ko raztezka ob pretrgu ni mogoče napovedati.

Strojniški vestnik - Journal of Mechanical Engineering 58(2012)9, SI 110-114

SI 111

Na Fakulteti za strojništvo Univerze v Mariboru sta z uspehom obranila svojo doktorsko disertacijo:

dne 6. julija 2012 David GREIF z naslovom: »Kavitacijska erozija v vbrizgalnih komponentah z uporabo polidisperznega kavitacijskega modela« (mentorica: prof. dr. Breda Kegl);

Predložena doktorska disertacija se ukvarja z numeričnim modeliranjem erozijskih efektov kot posledico kavitacijskega toka. Tovrstni tokovi se pojavljajo v različnih industrijskih procesih in so še posebej pomembni v industriji, ki proizvaja ali uporablja komponente za dovod in vbrizg goriva pod visokimi tlaki. Zmožnost numeričnih simulacij tovrstnih notranjih tokov s pomočjo računalniške dinamike tekočin (CFD) omogoča učinkovito oblikovanje vbrizgalnih šob in drugih komponent za dovod goriva, kot so tlačilke, skupni vodi, različni ventili. Ustrezno fizikalno modeliranje samega kavitacijskega toka je predpogoj za določitev agresivnosti obravnavanih pogojev in erozijskih posledic kavitacije. Gonilna sila za nastanek erozijske škode so implozije parnih mehurčkov v kavitacijskem toku. Ponavljajoče se implozije parne faze na površini izpostavljenega materiala sčasoma povzročijo površinske poškodbe. Pravilna ocena nastanka in transporta parne faze v obliki mehurčkov je ključnega pomena. Polidisperzni kavitacijski model v komercialnem CFD programu AVL FIRE je bil uporabljen kot osnova predstavljene disertacije. Materialni erozijski model je bil modificiran in prilagojen v okvirje dinamike tekočin ter implementiran v CFD program. Predvideva, da material ne prenese obremenitev nad mejo plastičnosti materiala in ocenjuje agresivnost kavitacijskega toka kot funkcijo števila implozij mehurčkov na enoto površine in velikosti le-teh ob imploziji na površini. Model omogoča numerične simulacije erozijskih efektov, kar ima izredno vrednost pri oblikovanju komponent za dovod in vbrizg goriva. Kombinacija materialnega erozijskega modela z večfaznim modelom za dinamiko tekočin je unikatna in predstavlja jedro disertacije. Prikazanih je več rezultatov, ki kažejo na zmožnost modela za kvantitativno in kvalitativno oceno verjetnosti nastanka erozijskih poškodb materiala. Prikazane so tudi posledice erodirane površine izvrtin vbrizgalne šobe na širjenje curka goriva. To nakazuje na vpliv erozije na pripravo mešanice goriva in samo zgorevanje;

dne 13. julija 2012 Jurij ILJAŽ z naslovom: »Reševanje inverznih problemov prenosa toplote v tkivu z metodo robnih elementov« (mentor: prof. dr. Leopold Škerget);

Delo obravnava inverzni problem prenosa toplote v tkivu, tj. numerično določitev krajevno

odvisnega perfuzijskega pretoka v nehomogenem tkivu na osnovi neinvazivnih meritev temperature in toplotnega toka. Pri čemer obravnavani problem temelji na Pennesovem matematičnem modelu. Tako pridemo do realnejšega opisa problema oziroma stanja tkiva ter možnosti določitve perfuzijskega pretoka v posameznem tkivu, ki pa je še kako pomemben tako za diagnostiko kot tudi nekatere klinične aplikacije.

Problem je bil zaradi svoje zahtevnosti obravnavan numerično, pri čemer je bil postavljen nov numerični algoritem na osnovi MRE (Metode Robnih Elementov) in optimizacije. Pri tem je bila MRE uporabljena za reševanje direktnega problema prenosa toplote v nehomogenem tkivu ob predpostavljeni neznani spremenljivki ter s tem določitve namenske funkcije, ki jo želimo minimizirati s primerno optimizacijsko metodo. Izbrani sta bili dve optimizacijski metodi; BFGS (Broyden-Fletcher-Goldfarb-Shanno) in LM (Levenberg-Marquardt), ki smo ju zaradi nestabilnosti obravnavanega inverznega problema nadgradili z uporabo Tikhonove regularizacije prvega reda, pri tem pa uporabili L-Curve metodo določitve optimalnega regularizacijskega parametra.

Numerični algoritem je bil pri tem testiran za različne testne funkcije perfuzijskega koeficienta tako homogenega kot nehomogenega tkiva, pri čemer robni in začetni pogoji zadostijo enoličnosti inverznega problema. S tem je bil analiziran numerični algoritem kot tudi reševanje inverznega problema za primer eksaktnih meritev in meritev s šumom. Tako smo prišli do celovite analize primernosti optimizacijske metode, uporabljene regularizacije za reševanje tovrstnih primerov, kot tudi vpliva nehomogenosti, začetne vrednosti, šuma v meritvah in porazdelitev neznane funkcije, ki jo želimo rekonstruirati.

Rezultati pri tem kažejo, da je numerični algoritem z uporabo LM metode ter uporabljene regularizacije primernejši od BFGS metode za podani inverzni problem, saj je rešitev globalno stabilna, natančneje določi neznano funkcijo in tudi hitreje konvergira. Regularizacija prvega reda je primerna le za gladke monotone funkcije z nizko vrednostjo prvega odvoda, drugače je rekonstrukcija funkcije možna le v območju blizu meritev, pri čemer ima nehomogenost snovnih lastnosti velik vpliv, še zlasti ob prisotnosti tkiva z nižjo toplotno prevodnostjo. Rekonstrukcija je možna tudi ob prisotnosti nižje stopnje šuma, medtem ko je resolucija pri višji stopnji na račun regularizacije izgubljena, kar oteži reševanje problema.

Delo je usmerjeno le v numerični del problema in tako predstavlja osnovo za nadaljnji razvoj neinvazivnih metod določitve perfuzijskega pretoka krvi v realnem nehomogenem tkivu.


SI 112

MAGISTRSKA DELA

Na Fakulteti za strojništvo Univerze v Ljubljani je z uspehom zagovarjal svoje magistrsko delo:

dne 6. julija 2012 Robert ZAKRAJŠEK z naslovom: »Sodobni sistemi za preoblikovanje tankostenskih kovinskih izdelkov« (mentor: prof. dr. Karl Kuzman, somentor: doc. dr. Tomaž Pepelnjak);

*

Na Fakulteti za strojništvo Univerze v Mariboru sta z uspehom zagovarjala svoje magistrsko delo:

dne 2. julija 2012 Marko ŠVERKO z naslovom: »Uporaba načel vitke proizvodnje pri proizvodnji žarometov« (mentor: izr. prof. dr. Borut Buchmeister);

dne 5. julija 2012 Katja KOMPOLŠEK z naslovom: »Vpliv postopkov odstranjevanja hlapljivih organskih snovi na emisije pri proizvodnji izolacij« (mentor: prof. dr. Matjaž Hriberšek);

SPECIALISTIČNA DELA

Na Fakulteti za strojništvo Univerze v Ljubljani je z uspehom zagovarjal svoje specialistično delo:

dne 10. julija 2012 Gorazd JELENC z naslovom: »Moderni postopki odrezavanja težko obdelovalnih materialov s postopki zarezovanja« (mentor: prof. dr. Janez Kopač, somentor: prof. dr. Marko Starbek).

*

Na Fakulteti za strojništvo Univerze v Mariboru je z uspehom zagovarjal svoje specialistično delo:

dne 5. julija 2012 Aleš WEISS z naslovom: »Nestacionarni tok tekočine v cevi z loputo« (mentor: prof. dr. Leopold Škerget).

DIPLOMIRALI SO

Na Fakulteti za strojništvo Univerze v Ljubljani so pridobili naziv univerzitetni diplomirani inženir strojništva:

dne 28. avgusta 2012:Jure KOBAL z naslovom: »Obnašanje polietilena

nizke gostote pod vplivom udarnih obremenitev« (mentor: prof. dr. Igor Emri);

Alen OSELI z naslovom: »Razvoj in izdelava naprave za merjenje dinamične volumetrične voljnosti časovno odvisnih materialov« (mentor: prof. dr. Igor Emri);

dne 29. avgusta 2012:Žiga AHAČIČ z naslovom: »Izdelava prototipa

hidro-ciklona za separacijo algne biomase« (mentor:

izr. prof. dr. Slavko Dolinšek, somentor: prof. dr. Branko Širok);

David HOMAR z naslovom: »Hibridna izdelava orodja za brizganje polimerov« (mentor: prof. dr. Janez Kopač, somentor: izr. prof. dr. Slavko Dolinšek);

Dominik KOŠUTA z naslovom: »Obdelava lonca za stikalo zaganjalnika« (mentor: prof. dr. Janez Kopač);

Anton ŽNIDARČIČ z naslovom: »Kavitacija na ultrazvočni sondi« (mentor: izr. prof. dr. Matevž Dular, somentor: prof. dr. Branko Širok);

dne 31. avgusta 2012:Matija BRUMAT z naslovom: »Identifikacija

dušenja struktur na podlagi meritve specifičnih deformacij« (mentor: prof. dr. Miha Boltežar, somentor: doc. dr. Janko Slavič);

Simon STRNAD z naslovom: »Soproizvodnja toplote in električne energije s kombiniranim plinsko-parnim postrojenjem« (mentor: izr. prof. dr. Mihael Sekavčnik);

Simon VIPAVEC z naslovom: »Strukturna dinamika kinematično vzbujanih izdelkov « (mentor: prof. dr. Miha Boltežar, somentor: doc. dr. Janko Slavič);

Sandi VOLOVEC z naslovom: »Radialna plinska turbina v zaprtem postroju za sočasno proizvodnjo toplote in električne energije« (mentor: izr. prof. dr. Mihael Sekavčnik);

Urban ŽVAR BAŠKOVIČ z naslovom: »Simulacija adiabatnega toka zraka na referenčno geometrijo kabine prototipa dvosedežnega električnega vozila« (mentor: prof. dr. Vincenc Butala);

Miha SREBOT z naslovom: »Modeliranje, zasnova preizkuševališča in eksperimentalna analiza mehanske roke s pnevmatično mišico« (mentor: izr. prof. dr. Niko Herakovič);

Boštjan VIDIC z naslovom: »Sistem za pranje medicinskih pripomočkov« (mentor: prof. dr. Jožef Duhovnik);

Miha ŽAGAR z naslovom: »Parametrična CAD podpora za obvladovanje toleranc zavesnih fasadnih sistemov« (mentor: prof. dr. Jožef Duhovnik, somentor: asist. dr. Leon Kos).

*

Na Fakulteti za strojništvo Univerze v Mariboru je pridobil naziv univerzitetni diplomirani inženir strojništva:

dne 6. julija 2012:Jernej DROFELNIK z naslovom:

»Aerodinamična analiza nihajočega krila v energijsko-ekstrakcijskem režimu z uporabo stisljivih Navier-Stokesovih enačb z učinkovitim in natančnim


SI 113

nizko-hitrostnim predpogojevalnikom« (mentor: prof. dr. Leopold Škerget, somentor: dr. M. Sergio Campobasso).

*

Na Fakulteti za strojništvo Univerze v Mariboru sta pridobila naziv univerzitetni diplomirani gospodarski inženir:

dne 30. avgusta 2012:Marko TROJNER z naslovom: »Razvoj ročaja

hladilnika Gorenje po meri kupca« (mentor: doc. dr. Marjan Leber, somentorica: prof. dr. Majda Bastič);

Matjaž ŽUPAN z naslovom: »Model razvoja ortopedskega pripomočka z uporabo naprednih tehnologij« (mentor: prof. dr. Jože Balič, somentor: prof. dr. Duško Uršič).

*

Na Fakulteti za strojništvo Univerze v Mariboru sta pridobila naziv magister inženir strojništva:

dne 29. avgusta 2012:Marko DOBNIKAR z naslovom: »Pregled in

analiza postopkov preoblikovanja pločevine z mediji« (mentor: izr. prof. dr. Ivan Pahole, somentor: doc. dr. Mirko Ficko);

Mihael KRALJ z naslovom: »Konceptualna rešitev sistema za litje aluminijastih izdelkov« (mentor: izr. prof. dr. Ivan Pahole).

*

Na Fakulteti za strojništvo Univerze v Mariboru je pridobila naziv magister inženir tehniškega varstva okolja:

dne 29. avgusta 2012:Marijana LAKIĆ z naslovom: »Razvoj votlih

nanosfer za uporabo v okoljevarstvu« (mentorica: prof. dr. Aleksandra Lobnik).

*

Na Fakulteti za strojništvo Univerze v Mariboru so pridobili naziv diplomirani inženir strojništva (UN):

dne 30. avgusta 2012:Jernej DOMANJKO z naslovom: »Razvoj

pasivnega sončnega grelnika zraka« (mentor: prof. dr. Aleš Hribernik, somentor: doc. dr. Matjaž Ramšak);

Nejc GORINŠEK z naslovom: »Oblikovanje delovnega mesta za robkanje z upoštevanjem antropometrije« (mentorica: doc. dr. Nataša Vujica Herzog, somentorica: doc. dr. Simona Jevšnik);

Peter JURGEC z naslovom: »Integracija orodnega merilnega sistema v RIP« (mentor: prof. dr. Jože Balič);

Jernej KUHARIČ z naslovom: »Primerjava naprednih konceptov organizacije proizvodnje« (mentor: izr. prof. dr. Borut Buchmeister, somentorica: doc. dr. Nataša Vujica Herzog).

*

Na Fakulteti za strojništvo Univerze v Ljubljani so pridobili naziv diplomirani inženir strojništva:

dne 29. avgusta 2012:Tomaž JEROVŠEK z naslovom: »Razvoj

konstrukcije pozicionierne mize za vpenjanje varjencev« (mentor: izr. prof. dr. Jože Tavčar, somentor: prof. dr. Jožef Duhovnik);

Samo KOZODERC z naslovom: »Vpliv mehanskega recikliranja na časovno odvisne lastnosti polietilena nizke gostote (LDPE)« (mentor: prof. dr. Igor Emri);

Gregor OŽBOLT z naslovom: »Napoved delovanja fotonapetostnih sistemov na osnovi vremenske napovedi« (mentor: prof. dr. Sašo Medved);

Petra VERČIČ z naslovom: »Analiza oskrbe letala na Letališču Jožeta Pučnika in predlogi za izboljšavo« (mentor: doc. dr. Viktor Šajn, somentor: pred. Miha Šorn);

dne 31. avgusta 2012:Sebastjan ERČULJ z naslovom: »Odstranjevanje

odpadnega abraziva iz lovilnega bazena stroja za rezanje z abrazivnim vodnim curkom« (mentor: doc. dr. Henri Orbanić, somentor: prof. dr. Mihael Junkar);

Žiga PANTAR z naslovom: »Določanje temperaturnega stanja v klimatski komori« (mentor: izr. prof. dr. Ivan Bajsić);

Aleš ZUPANČIČ z naslovom: »Nadzor kakovosti procesa izdelave s pomočjo analize SPC« (mentor: doc. dr. Davorin Kramar, somentor: prof. dr. Janez Kopač).

*

Na Fakulteti za strojništvo Univerze v Mariboru so pridobili naziv diplomirani inženir strojništva:

dne 30. avgusta 2012:Tadej GMEINER z naslovom: »Načrtovanje

proizvodnje po naročilu v podjetju Tehcenter Ptuj« (mentor: izr. prof. dr. Borut Buchmeister, somentor: doc. dr. Iztok Palčič);

Jože HREN z naslovom: »Optimizacija proizvodnje na paletnem sistemu CNC strojih MAKINO A55« (mentor: prof. dr. Franci Čuš);


SI 114

Samo KOLOŠA z naslovom: »Transportni sistemi z vključeno akumulacijsko funkcijo« (mentor: prof. dr. Iztok Potrč, somentor: izr. prof. dr. Tone Lerher);

Mitja LIPUŠ z naslovom: »Obvladovanje kakovosti montaže pesta z metodo Poka Yoke« (mentor: prof. dr. Bojan Ačko, somentor: prof. dr. Nenad Gubeljak);

Peter NAVOTNIK z naslovom: »Konstruiranje naprave za prenos aluminijastih kolutov« (mentor: doc. dr. Janez Kramberger);

Darijan PLAZ z naslovom: »Ekonomska ocena stroškov ogrevanja poslovnega sistema s plinom ali s toplotno črpalko« (mentor: doc. dr. Matjaž Ramšak).

Na Fakulteti za strojništvo Univerze v Ljubljani so pridobili naziv diplomirani inženir strojništva (VS):

dne 29. avgusta 2012:Mateja BOŽIČ z naslovom: »Uporaba

termoplastičnega poliuretana (TPU) za dušenje mehanskih vibracij« (mentor: prof. dr. Igor Emri);

dne 31. avgusta 2012:Miha KLAVŽAR z naslovom: »Izboljšava

karakteristik zavornega ventila« (mentor: doc. dr. Jožef Pezdirnik);

Dominik RUPERT z naslovom: »Sledenje gibanja varilnega obloka v realnem času na osnovi stereo vida« (mentor: prof. dr. Alojzij Sluga, somentor: doc. dr. Drago Bračun).

Strojniški vestnik – Journal of Mechanical Engineering (SV-JME)

Aim and ScopeThe 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 ChiefVincenc ButalaUniversity of Ljubljana Faculty of Mechanical Engineering, Slovenia

Technical EditorPika ŠkrabaUniversity of Ljubljana Faculty of Mechanical Engineering, Slovenia

Editorial OfficeUniversity of Ljubljana (UL)Faculty of Mechanical EngineeringSV-JMEAškerčeva 6, SI-1000 Ljubljana, SloveniaPhone: 386-(0)1-4771 137Fax: 386-(0)1-2518 567E-mail: [email protected], http://www.sv-jme.eu

PrintTiskarna Knjigoveznica Radovljica, printed in 480 copies

Founders and PublishersUniversity of Ljubljana (UL)Faculty of Mechanical Engineering, Slovenia

University of Maribor (UM)Faculty of Mechanical Engineering, Slovenia

Association of Mechanical Engineers of Slovenia

Chamber of Commerce and Industry of SloveniaMetal Processing Industry Association

International Editorial BoardKoshi Adachi, Graduate School of Engineering,Tohoku University, JapanBikramjit Basu, Indian Institute of Technology, Kanpur, IndiaAnton Bergant, Litostroj Power, Slovenia Franci Čuš, UM, Faculty of Mech. Engineering, SloveniaNarendra B. Dahotre, University of Tennessee, Knoxville, USAMatija Fajdiga, UL, Faculty of Mech. Engineering, SloveniaImre Felde, Bay Zoltan Inst. for Mater. Sci. and Techn., HungaryJože Flašker, UM, Faculty of Mech. Engineering, SloveniaBernard Franković, Faculty of Engineering Rijeka, CroatiaJanez Grum, UL, Faculty of Mech. Engineering, SloveniaImre Horvath, Delft University of Technology, NetherlandsJulius Kaplunov, Brunel University, West London, UKMilan Kljajin, J.J. Strossmayer University of Osijek, CroatiaJanez Kopač, UL, Faculty of Mech. Engineering, SloveniaFranc Kosel, UL, Faculty of Mech. Engineering, SloveniaThomas Lübben, University of Bremen, GermanyJanez Možina, UL, Faculty of Mech. Engineering, SloveniaMiroslav Plančak, University of Novi Sad, SerbiaBrian Prasad, California Institute of Technology, Pasadena, USABernd Sauer, University of Kaiserlautern, GermanyBrane Širok, UL, Faculty of Mech. Engineering, SloveniaLeopold Škerget, UM, Faculty of Mech. Engineering, SloveniaGeorge E. Totten, Portland State University, USANikos C. Tsourveloudis, Technical University of Crete, GreeceToma Udiljak, University of Zagreb, CroatiaArkady Voloshin, Lehigh University, Bethlehem, USA

President of Publishing CouncilJože DuhovnikUL, Faculty of Mechanical Engineering, Slovenia

General informationStrojniški 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 peer-review process.

ISSN 0039-2480

Cover: Medical images that are obtained by the autofluorescence bronchoscopy, are segmented into the potentially cancerous and healthy areas: - Images 1.1 and 2.2: manual segmentation by a medical doctor who is specialized in image reading.- Images 1.2 and 3.1: segmentation by the machine algorithm in the RGB space. - Images 2.1 and 3.2: segmentation by the machine algorithm in the HSV space. Use of the HSV space is preferred in this type of machine diagnostics.Image Courtesy: Laboratory LDSE, Faculty of Mechanical Engineering, University of Ljubljana

© 2011 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 AuthorsAll 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/.

Announcement:The authors are kindly invited to submitt the paper through our web

site: http://ojs.sv-jme.eu. 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 MANUSCRIPTThe 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.

REFERENCESA 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 cross-linked 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 ABSTRACTBy 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/.

COPYRIGHTAuthors 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.sv-jme.eu/.

PUBLICATION FEEFor 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 220.00 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.


Since 1955



urna

lofM

echa

nicalE

nginee

ring

-Strojniškivestnik

58 (2

012)

9


http://en.sv-jme.eu/data/upload/2011/07_08/01_2011_013_Hackenschmidt_04.pdf


Since 1955



urna

lofM

echa

nicalE

nginee

ring

-Strojniškivestnik

58 (2

012)

9


journal of mechanical engineering 2012 9

Documents

industrial engineering

process engineering

faculty of engineering

graduate school of engineering

scopethe international

tohoku university

university of tennessee

international conferences