Journal of Mechanical Engineering 2014 12 by Darko Svetak

http://www.sv-jme.eu

60 (2014) 12

Since 1955

Papers

769

Matjaž Fleisinger, Matej Vesenjak, Matjaž Hriberšek: Flow Driven Analysis of a Darrieus Water Turbine

777

Ivan Dunđerski: Managing Vehicle Acceleration Properties by Programming Functions for Engine Torque Control

789

Primož Potočnik, Tomaž Berlec, Alojz Sluga, Edvard Govekar: Hybrid Self-Organization Based Facility Layout Planning

797

Nataša Vujica Herzog, Stefano Tonchia: An Instrument for Measuring the Degree of Lean Implementation in Manufacturing

804

Marek Boryga: Trajectory Planning of an End-Effector for Path with Loop

815

Santiago Ruiz-Arenas, Imre Horváth, Ricardo Mejía-Gutiérrez, Eliab Z. Opiyo: Towards the Maintenance Principles of Cyber-Physical Systems

832

Zafer Barlas, Murat Çolak: Evaluation of the Influence of Upset Stage on Joint Properties of Friction Welded Dissimilar Aluminum-Copper Cast Alloys

Journal of Mechanical Engineering - Strojniški vestnik

Contents

12 year 2014 volume 60 no.

Strojniški vestnik Journal of Mechanical Engineering

Strojniški vestnik – Journal of Mechanical Engineering (SV-JME) Aim and Scope The international journal publishes original and (mini)review articles covering the concepts of materials science, mechanics, kinematics, thermodynamics, energy and environment, mechatronics and robotics, fluid mechanics, tribology, cybernetics, industrial engineering and structural analysis. The journal follows new trends and progress proven practice in the mechanical engineering and also in the closely related sciences as are electrical, civil and process engineering, medicine, microbiology, ecology, agriculture, transport systems, aviation, and others, thus creating a unique forum for interdisciplinary or multidisciplinary dialogue. The international conferences selected papers are welcome for publishing as a special issue of SV-JME with invited co-editor(s). Editor in Chief Vincenc Butala University of Ljubljana, Faculty of Mechanical Engineering, Slovenia

Technical Editor Pika Škraba University of Ljubljana, Faculty of Mechanical Engineering, Slovenia

Founding Editor Bojan Kraut University of Ljubljana, Faculty of Mechanical Engineering, Slovenia

Editorial Office University of Ljubljana, Faculty of Mechanical Engineering SV-JME, Aškerčeva 6, SI-1000 Ljubljana, Slovenia Phone: 386 (0)1 4771 137 Fax: 386 (0)1 2518 567 info@sv-jme.eu, http://www.sv-jme.eu Print: Littera Picta, printed in 400 copies Founders and Publishers University of Ljubljana, Faculty of Mechanical Engineering, Slovenia University of Maribor, Faculty of Mechanical Engineering, Slovenia Association of Mechanical Engineers of Slovenia Chamber of Commerce and Industry of Slovenia, Metal Processing Industry Association President of Publishing Council Branko Širok

University of Ljubljana, Faculty of Mechanical Engineering, Slovenia

Vice-President of Publishing Council Jože Balič University of Maribor, Faculty of Mechanical Engineering, Slovenia

Cover: Hydrokinetic Darrieus turbine with horizontal axis of rotation (middle) that uses kinetic energy of water current without impoundment or side channels and thus enables a quick, cost effective installation on a shallow riverbeds for electricity production with minimal environmental impact. The CFD simulation was performed with a flow driven approach, which better corresponds to the real operating conditions. Simulation results for two different turbine positions are shown above and below. Courtesy: Matjaž Fleisinger, University of Maribor, Faculty of Mechanical Engineering, Slovenia

Institutional prices include print & online access: institutional subscription price and foreign subscription €100,00 (the price of a single issue is €10,00); general public subscription and student subscription €50,00 (the price of a single issue is €5,00). Prices are exclusive of tax. Delivery is included in the price. The recipient is responsible for paying any import duties or taxes. Legal title passes to the customer on dispatch by our distributor. Single issues from current and recent volumes are available at the current single-issue price. To order the journal, please complete the form on our website. For submissions, subscriptions and all other information please visit: http://en.sv-jme.eu/. You can advertise on the inner and outer side of the back cover of the magazine. The authors of the published papers are invited to send photos or pictures with short explanation for cover content. We would like to thank the reviewers who have taken part in the peerreview process.

ISSN 0039-2480 © 2014 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 Research Agency. Strojniški vestnik - Journal of Mechanical Engineering is also available on http://www.sv-jme.eu, where you access also to papers’ supplements, such as simulations, etc.

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

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

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12 Contents

Contents Strojniški vestnik - Journal of Mechanical Engineering volume 60, (2014), number 12 Ljubljana, December 2014 ISSN 0039-2480 Published monthly

Papers Matjaž Fleisinger, Matej Vesenjak, Matjaž Hriberšek: Flow Driven Analysis of a Darrieus Water Turbine Ivan Dunđerski: Managing Vehicle Acceleration Properties by Programming Functions for Engine Torque Control Primož Potočnik, Tomaž Berlec, Alojz Sluga, Edvard Govekar: Hybrid Self-Organization Based Facility Layout Planning Nataša Vujica Herzog, Stefano Tonchia: An Instrument for Measuring the Degree of Lean Implementation in Manufacturing Marek Boryga: Trajectory Planning of an End-Effector for Path with Loop Santiago Ruiz-Arenas, Imre Horváth, Ricardo Mejía-Gutiérrez, Eliab Z. Opiyo: Towards the Maintenance Principles of Cyber-Physical Systems Zafer Barlas, Murat Çolak: Evaluation of the Influence of Upset Stage on Joint Properties of Friction Welded Dissimilar Aluminum-Copper Cast Alloys

769 777 789 797 804 815 832

Received for review: 2014-02-03 Received revised form: 2014-06-28 Accepted for publication: 2014-09-19

Flow Driven Analysis of a Darrieus Water Turbine Fleisinger, M. – Vesenjak, M. – Hriberšek, M. Matjaž Fleisinger* – Matej Vesenjak – Matjaž Hriberšek

University of Maribor, Faculty of Mechanical Engineering, Slovenia The paper discusses the development of transient sliding grid based flow driven simulation procedure for a Darrieus turbine analysis. A new computational procedure is developed within the Ansys-CFX software that allows fast solutions with good accuracy of the results. With this approach it is possible to determine the turbine power curve over its whole operating range at a given flow velocity in a single simulation. This enables comparison of several different geometry parameters to find the one that extracts the most of energy of water current. The proposed procedure allows a start-up analysis of a turbine and is also suitable to run in combination with fluid-structure interaction (FSI) analysis to analyze the mechanical behavior of the turbine blades. Keywords: hydrokinetic turbine, computational fluid dynamics (CFD) simulation, flow-driven approach

0 INTRODUCTION Energy of waterways is mostly extracted by means of conventional power plants. They require some kind of a dam that creates an artificial water head, which should be large enough to propel a water turbine. A few years ago, the tidal energy systems were adapted for river energy extraction, which have their origins in wind energy extraction technologies [1]. Such systems can be placed in a free stream in the same way that a wind turbine is in wind, without the need for a dam or a channel. Therefore, they extract kinetic energy of a current, where the major differences to the wind energy extraction are about 800 times greater density of water than air and slower current velocities, that are however more stable and predictable than wind. In order to improve the technology for energy extraction, the numerical simulations provide a valuable tool for numerous virtual experiments prior to the real prototypes being manufactured. The computational fluid dynamics (CFD) simulations used in this work use flow-driven approach that can consider more realistic conditions in simulation. In such simulation, the turbine rotation is governed by stream velocity from standstill onwards, which better corresponds to the real operating conditions. Because of a complex flow field around turbine blades, the use of CFD simulations is essential for better insight into flow conditions around the hydrofoils in order to improve the turbine operation, as well as for accurate performance prediction of turbine rotors. 1 MATERIALS AND METHODS 1.1 Hydrokinetic Turbines and Darrieus Turbine Hydrokinetic turbines are designed to be installed in a stream or current, extracting kinetic energy from

the flow of water to power an electric generator without impounding or diverting the flow of the water. Conceptually, this is similar to the way wind energy conversion devices work. Considering that hydrokinetic turbines can be deployed in any water resource having sufficient velocity to drive them (between 1 and 2 m/s or even less [2]), their energy generation potential is enormous. Water resources that could be harnessed include natural streams, tidal currents, ocean currents and constructed waterways such as channels. Installation of such systems is much simpler, because they do not need dams and they can be easily moved to another location or entirely removed from the waterway. During their operation or at rest, they also do not prevent the migration of water organisms.

Fig. 1. Darrieus water turbine with horizontal axis placed in a shallow riverbed

The Darrieus turbine (Fig. 1) was invented by Georges Jean Marie Darrieus and was patented in USA in 1931 [3] and [4]. Its original intent was extraction of wind energy, so because its rotational axis was positioned vertically, in literature it is often regarded as vertical axis wind turbine (VAWT). In recent years it has been adapted for water energy extraction, where

*Corr. Author’s Address: University of Maribor, Faculty of Mechanical Engineering, Smetanova ulica 17, 2000 Maribor, Slovenia, matjaz.fleisinger@amis.net

769

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 769-776

its axis could be positioned vertically or horizontally and in both cases orthogonal to the incoming water stream. Both positions have their own benefits: in vertical position the electric generator can be mounted above water level; in horizontal position on Fig. 1 it is possible to have only one generator for the whole width of the channel. Because of the rectangular cross section of its rotor, the water Darrieus turbines can fill a larger cross-sectional area in shallow water than horizontal axis turbines with a circular crosssection. Its advantage in tidal current extraction is also the ability to extract current flowing in any direction without the need for a turn. The Darrieus turbine is also beneficial regarding blocking of the river/channel cross section, because its structure is relatively open compared to propeller turbines. The latter mainly uses an adapted Kaplan turbine rotor that requires guiding vanes or ducts for increased efficiency. The weaknesses of the Darrieus turbines on the other hand are relatively poor starting torque and its high oscillations during [3]. To avoid these disadvantages different airfoil profiles, helical blades configuration, increased number of blades or blades with variable pitch might be used. The Darrieus turbine compared to propeller turbines has very complex flow around its blades because of its crossflow configuration, where the blades cross the global current twice every turn. The flow is undisturbed on upstream side, but on the downstream side it is already highly effected due to previous blade crossing. The angle of attack of each blade changes constantly during operation but the databases of available airfoil data (lift and drag coefficients) are limited between +15 and –15° [5] due to unstable operation outside this limits. Therefore the performance of a turbine cannot be estimated using simple lift and drag force equations. There are also some analytical approaches for VAWT performance prediction using different aerodynamic models (potential flow theory, vortex theory [3], [6] and [7]), but they also require airfoil coefficient data and their results are only estimations. Therefore, the computational fluid dynamics was employed allowing the study of the flow field around the turbine and to overcome the lack of previously mentioned methods. Furthermore, the CFD allows analyzing transient simulations with ability to run them as a flow driven problem. This means that the turbine starts to move and rotates due to current flow acting on its blades. The simulations of flow driven Darrieus turbine have not been shown in such detail yet. The parametric numerical simulations also allow analyzing many different turbine design parameters, 770

providing an optimal configuration for a given set of design parameters. The turbine design parameters were chosen following work by Shiono et al. [8], where the turbine was experimentally tested with several different solidity parameters and current velocities. Based on previous research [9] and work by Shiono et al. the helical blade configuration was not chosen for further investigation, due to its reduced power coefficient and the difficult manufacturing procedure of such blades. The disadvantage of common computational approach for turbomachinery simulations is the need to prescribe a set of operating parameters first, such as rotational and flow velocity [10] to [13]. The main scope of this work is to validate the computational approach with a self-developed Fortran [14] routine with experimental data in a way similar to how Howell et al. [15] performed their comparison of experimental and computational results. Our simulation approach employs flow driven turbine simulation which better corresponds to real turbine operating conditions. 1.2 CFD and Flow Driven Motion In order to study the performance of the Darrieus turbine computationally, the numerical model has to account for the flow phenomena in the fluid and the hydrodynamic load on the turbine blades. To obtain the flow dependent rotation of the turbine, a physically correct coupling algorithm between the fluid motion and the solid body motion has to be applied. From the physical viewpoint, the equations describing fluid flows and heat and mass transfer are simply versions of the conservation laws of physics, namely: conservation of mass in Eq. (1) and conservation of momentum in Eq. (2) [9]:   ∇ ⋅ v = 0, (1)      Dv ρ = −∇p + µ∇ 2 v + S M . (2) Dt 1.2.1 Flow Driven Motion with Six Degrees of Freedom (DOF) Solver A rigid body is a solid object that moves through a fluid, while its shape remains undeformed. Its motion is dictated by the fluid forces and torques acting upon it and external forces (lift, gravity, friction) and external torques. In Ansys CFX [16] to [20] the computation of position and orientation of a rigid body is performed using equations of motion. The equations of motion of a rigid body are written as:

Fleisinger, M. – Vesenjak, M. – Hriberšek, M.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 769-776

   dK  dL = F = , M . (3) dt dt

These equations state that for a rigid body undergoing translation and rotation, the rateof change  momentum of its linear K and angular L is equal   to the applied force F and torque M acting on the body. The equation of motion for a translating center  nd of a mass x can be expressed  using 2 Newton law, where m is the mass and F represents the sum of all forces. It includes hydrodynamic force, weight of rigid body, spring and explicit external force. Expanding the sum of all forces may be written as:        ∑F = m ⋅ a = Fhydro + m ⋅ g − kspring ⋅ ( x − xSO ) + Fext , (4) whereis hydrodynamic force, g is gravity coefficient,  kspring is the linear spring constant, xSO is initial position coordinate and Fext are all other external forces acting on the body (buoyancy, friction). The flow force Fhydro is the total component of the flow field acting on the body. It is determined from the RANS equations by integrating viscous wall shear stresses and pressure field over the body’s surfaces:    Fhydro = ∑ ( − pi ⋅ ni + τ i ) ⋅ Si . (5)

the angular velocity of a previous time step in external .txt file and another, which calculates the new angular velocity based on old one and difference in torque on turbine blades in the previous timestep. This result is then used as the rotating domain boundary condition. The equation for new angular velocity ωnew is as follows:

ωnew = ωold +

(T − M load ) ⋅ ∆t , (6) J

where ωold is an angular velocity of a previous time step, T is the torque on turbine blades in a previous timestep, Mload is the loading moment, Δt is the timestep value and J is mass moment of inertia of a turbine rotor. This approach enabled us to perform flow driven simulation of Darrieus turbine using transient sliding grid procedure. The angular velocity of a turbine was updated in every time step, based on difference in torque on turbine rotor, loading moment and rotor moment of inertia. In such simulation, the turbine starts to turn due to the torque acting on a turbine rotor in every previous timestep as a consequence of a water flow.

The pi denotes the pressure acting on the surface  of a control volume whilst ni is the normal vector of the individual control volume face. The viscous  stresses are denoted by τ i and the surface of the control volume face is Si. The rigid body motion solution allows the turbine simulation with a flow driven approach, which means it can start to rotate from standstill due to water flow acting on its blades. Such an approach is convenient when we want to investigate whole turbine operation not just some constant conditions, where we would have to predict and define all the other operating parameters (e.g. angular velocity at a certain load). 1.2.2 Flow Driven Motion in Transient Sliding Grid Simulation with User Routine The flow driven motion approach by using 6DOF solver shows relatively long computational times, but its major flaw is that it cannot be used in combination with FSI simulation. Therefore we developed our own user routine which enabled us to perform flow driven simulations using a less computational intensive but faster transient sliding grid method. Our routine was written using the Fortran programming language. There were two user routines needed, one that keeps

Fig. 2. Flow chart of a CFX simulation with rigid body solution and user CFX expression language (CEL) routine

From the simulation flowchart on Fig. 2, it can be observed that the user routine is called only once per timestep while CFX establishes a solution through iteration. Initially this was also the reason for instabilities in the early phase of simulation, because there is relatively large change of blade position in every timestep. This causes large pressure changes that produce high torque on the blades which is used

Flow Driven Analysis of a Darrieus Water Turbine

771

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 769-776

in further calculation of angular velocity, which moves the blades even faster. This means that there is self-amplifying effect, which drives the instability and can be reduced with certain measures. They are supposed to reduce large changes of results between timesteps, by applying smaller timesteps, starting a solution with a pre-simulated flow field around the blades, or decreasing change of angular velocity between time steps by applying a damp coefficient. For further use the simulation procedure has to be verified and validated with experimental results. For this purpose the results from experimental testing of water Darrieus turbine done by Shiono et al. [8] was used. 2 RESULTS AND DISCUSSION

[5], with chord length of 126 mm which is projected on a circle with the diameter of 300 mm. The solidity, σ, states a relation between the blade area and the turbine swept area. For a Darrieus turbine it is defined as [8]:

σ=

where n represents the number of blades, C is the chord length of blade profile, and D is the turbine diameter. The solidity coefficient σ of model is 0.4 which corresponds to geometry tested by Shiono et al. [8]. The fluid model consisted of water at 10 °C and the turbulence model is two-equation k–ω SST (shear stress transport) of Menter [17] and [18], with turbulence equations written in tensor form:

2.1 Simulation Setup

∂ ( ρk )

Simulations with both flow driven approaches were performed on the same model in Ansys-CFX 13 [10]. The model of the Darrieus turbine with horizontal axis has three blades with the length of 0.2 and 0.3 m in diameter and corresponds to the experimental model used by Shiono et al. [8]. The rotating domain around the turbine is 0.45 m in diameter. It is modelled in a non-structured manner with 5,100,000 elements, which was determined by a preliminary mesh independence study, where results of present mesh differed from finer one by 2%. The upper and lower blade surfaces were modelled with 52 elements, with the smallest positioned around the leading and trailing edge, as can be seen on Fig. 3.

n ⋅C , (7) π ⋅D

∂t

= P − β ' ρω k

(

∂ ρv j k ∂x j

∂  ∂k   ( µ + σ k µt )  , (8) ∂x j  ∂x j 

∂ ( ρω ) ∂t

(

∂ ρ v jω ∂x j

 ∂ω  ( µ + σ ω µt ) + ∂x j   ρσ ∂k ∂ω +2 (1− − F1 ) ω 2 . (9) ω ∂x j ∂x j

γ ∂ P − βρω 2 νt ∂x j

Therefore, we used inflation modelling with 10 layers around each blade, which is required by that model [10].

Fig. 3. Mesh distribution and boundary layer modeled around blade surface

The stationary domain around the turbine consists of 700,000 elements and extends one length of the diameter upstream, above and below the turbine and 2 diameters downstream in the wake field of a turbine, while its sides extend half of the blade length to the left and right side of the turbine. The aerodynamic profile of the blades is modified Naca 633018 profile 772

Fig. 4. Computational model of Darrieus water turbine and employed boundary conditions

Fleisinger, M. – Vesenjak, M. – Hriberšek, M.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 769-776

The turbine surface is modelled as the ‘non slip wall’. The top and the bottom surface of the test channel are prescribed as ‘free slip wall’ while the side surfaces has a ‘symmetry’ boundary condition. The water flow velocity is 1.2 m/s towards the turbine from the inlet on the front surface of stationary domain and the opposite surface, where the water exits, is modelled as ‘outlet’ with 0 Pa relative pressure (Fig. 4). In order to acquire complete turbine power curve in a single simulation, simulation begins with water starting to flow and the turbine starts to rotate because of the water current acting on its blades. From the start of simulation there is a loading torque applied to a turbine rotor, which slowly increased from zero value with a rate specified by expression: Tload = 0.4 [Nm/s]·t, where t [s] is the simulation time. Several different rates were previously tested, so that turbine inertia does not affect the overall result. With such setup the turbine first reaches its maximum angular velocity, which slowly starts to decrease due to the constantly increasing load. As the simulation proceeds the load slowly becomes large enough to stop the turbine. With this simulation approach, it is possible to analyse the turbine efficiency over its whole operating range without the need to set several constant levels of loading. To compare several results of different analyses the results has to be transformed into dimensionless values. The angular velocity is usually represented as a tip speed ratio (TSR), which is a relationship between the turbine blade velocity and the current velocity [3]:

TSR =

ω⋅R v , (10) = U∞ U∞

where ω represents turbine angular velocity, R the turbine radius and U∞ is the water current velocity. The available power of the water current Pavail can be calculated using following equation:

Pavail =

1 ⋅ ρ ⋅ A ⋅ U ∞3 , (11) 2

where ρ is the water density and A is the free flow cross sectional area, which in our case corresponds to a turbine rotor swept area. The power of a hydrokinetic turbine Pturb is defined as:

Pturb = ω ⋅ T , (12)

where T is the torque on a turbine shaft. The power coefficient CP of a turbine which also represents its efficiency can be calculated as:

CP =

Pturb . (13) Pavail

The equation represents the relationship between power extracted from water current and power available in the current flowing through the same area as projected by the turbine. Hydrokinetic turbines extract energy from the fluid in a way that reduces the flow velocity with little or no pressure reduction as the fluid passes through the turbine rotor. Therefore, the current streamlines must expand to maintain continuity but they cannot expand indefinitely, so there is a theoretical limit to the amount of kinetic energy that can be extracted from the whole energy of a fluid. This limit has been determined by Betz [21] to be 59.3% for a surface through which energy is extracted. 2 RESULTS We can observe the highly turbulent flow conditions [19] on the downstream side of a turbine, which are caused by its rotation. On Fig. 5 there are contours of fluid vx velocities on inlet and outlet plane for two turbine positions (50 and 110°) with flow streamlines as well as corresponding flow fields in the middle section of the turbine rotor. On the downstream side there is strong downward vertical velocity, caused by blade traveling against the flow. The average value of turbulence kinetic energy k on the upstream side of the turbine is 4.465∙10–3 J/kg and on the downstream side it is 2.596∙10–2 J/kg. The dimensionless wall distance parameter y+ value during turbine operation was between 70 and 190, which is within the range recommended by best practice for turbomachinery simulation [16]. Both flow driven approaches has been used for turbine simulation. In both cases the turbine started to rotate due to water current from standstill, where in simulation with routine based flow-driven approach the torque on turbine blades and consecutive angular velocity oscillated strongly after it commenced, which can be seen on Fig. 6. Simulation became stable after two seconds and proceeded in a periodical pattern, similar to 6DOF-based simulation. The torque on the blades slowly increased and angular velocity decreased due to increasing load, where the angular velocity in routine-based simulation remains on a slightly higher level and has ~30% larger amplitude than in 6DOF-based simulation. On Fig. 7, the torque on the turbine is shown for a single turn of a turbine during stable regular

Flow Driven Analysis of a Darrieus Water Turbine

773

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 769-776

Fig. 5. The contours of flow vx velocities on inlet and outlet of model and streamlines for two turbine positions, a) 50° and b) 110° and corresponding velocity vectors on a middle plane c) and d)

operation. The turbine has the same blade in its upmost position at the beginning and at the end of an interval. Therefore, the turbine positions in which its blades produce the most torque can be identified. It can be seen that the upmost blade starts producing torque from ~50° onwards until it reaches ~100° and that pattern is then repeated periodically for every 120° for a three bladed turbine. The comparison of performance curves on Fig. 8 shows that routine-based flow-driven simulation over-predicts the experimental results by 11.3% while 6DOF-based simulation gives 3.1% higher result at peak tip speed ratio. The routine-based results however were following the trend of experimental results at nearly the same distance over the large part of the curve, while the 6DOF-based results were close to the experiment only at the peak value. The difference becomes larger in other areas. The main reason for difference in results between simulations and experiment is in the method by which Shiono 774

et al. [8] performed measurements. Their turbine was gradually loaded by electromagnetic brake and in each constant load level the rotating velocity and torque were measured, where the result represented an average value of a 30-second interval. In our simulations the turbine is loaded with constantly increasing torque, meaning that the turbine does not stabilize in a steady regular manner. Also the correct data of the test turbine, such as material and its mass moment of inertia as well as surface roughness were not available. Computationally, the difference between each simulation approach occurred because of differing flow-driven approaches. The 6DOF solver is involved in internal CFX iteration loop, so the solution of angular velocity is determined iteratively upon equilibrium, while in the approach with user routine, the angular velocity is updated only once per timestep. Therefore it represents certain disturbance, which is dependent on the rate of change of torque on the blades in a timestep. This is at its greatest at

Fleisinger, M. – Vesenjak, M. – Hriberšek, M.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 769-776

Fig. 6. Comparison of simulation results for 6DOF and user routine flow driven simulation approach

the beginning of the simulation, when the turbine accelerates from standstill to its highest angular velocity, therefore in first two seconds of simulation the large instability appears.

Fig. 8. Validation of computational model for both flow-driven approaches, compared to published experimental results

3 CONCLUSIONS Fig. 7. Torque on a turbine rotor during its single turn, starting with one blade at θ = 0°

The inertia in fluid field gathered in this period has certain influence on torque and angular velocity in the early phase of simulation. Also the turbine geometry used in the experiment has some influence on overall results of numerical simulation, due to its cross section with aspect ratio of 0.66. This means that the turbine blade length is one-third smaller than turbine diameter, so consequently the effect of turbine blades is reduced and the effect of supporting structure is increased.

A new approach for fluid driven turbomachinery simulation was developed, enabling the analysis of a whole turbine operation range in a single simulation. The turbine simulations were performed with two approaches for flow-driven operation, where the turbine rotates due to the water stream acting on its blades. To capture its whole operation range, the turbine at the beginning of a simulation starts to rotate freely, after that, a gradually increasing load is applied, which slows the turbine rotation until it stops. Simulation results gathered with this procedure were well aligned to experimental results from the literature. The developed flow driven simulation approach is based on transient sliding mesh procedure

Flow Driven Analysis of a Darrieus Water Turbine

775

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 769-776

with an adaptive user routine, therefore it can be used in combination with strongly coupled FSI simulations. In this way in our further research the water current and the turbine could mutually interact, making it possible to analyse the influence of blade deformation on a turbine performance. For successful method application we would have to solve some issues regarding large instabilities at the start, which could cause some problems in FSI simulation due to large deformations under very high rotation velocities. 4 ACKNOWLEDGEMENTS This research was funded by SPIRIT Slovenia - Public Agency of the Republic of Slovenia for the Promotion of Entrepreneurship, Innovation, Development, Investment and Tourism. 5 REFERENCES [1] Lago, L.I., Ponta, F.L., Chen, L. (2010). Advances and trends in hydrokinetic turbine systems. Energy for Sustainable Development, vol. 14, no. 4, p. 287-296, DOI:10.1016/j.esd.2010.09.004. [2] Yavuz, T., Kilkis, B., Akpinar, H., Erol, O. (2011). Performance analysis of a hydrofoil with and without leading edge slat. Proceedings of the 10th International Conference on Machine Learning and Applications and Workshops, (Volume 2), p. 281-285, DOI:10.1109/ ICMLA.2011.113. [3] Paraschivoiu, I. (2002). Wind Turbine Design with Emphasis on Darrieus concept. Polytechnic International Press, Montreal. [4] Darrieus, G.J.M. (1926). Turbine Having Its Rotating Shaft Transverse to the Flow of the Current. Patent 1835018. U.S. Patent and Trademark Office, Washington D.C. [5] Krauss, T. (2012). Airfoil Investigation Database, from http://www.worldofkrauss.com, accessed on 2012-0920. [6] Islam, M., Ting, D.S.K. Fartaj, A. (2008). Aerodynamic models for Darrieus-type straight-bladed vertical axis wind turbines. Renewable and Sustainable Energy Reviews, vol. 12, no. 4, p. 1087-1109, DOI:10.1016/j. rser.2006.10.023. [7] Brahimi, M.T., Allet, A., Paraschivoiu, I. (1995). Aerodynamic analysis models for vertical-axis wind turbines. International Journal of Rotating Machinery, vol. 2, no. 1, p. 15-21, DOI:10.1155/ S1023621X95000169. [8] Shiono, M., Suzuki, K. Kiho, S. (2000). An experimental study of the characteristics of a Darrieus turbine for tidal power generation. Electrical Engineering in Japan, vol. 132, no. 3, p. 38-47, DOI:10.1002/15206416(200008)132:3<38::AID-EEJ6>3.0.CO;2-E.

776

[9] Fleisinger, M., Zadravec, M., Vesenjak, M., Hriberšek, M., Udovičič, K. (2012). Comparison of CFD simulation of Darrieus water turbine using the rigid body solver and the MFR method. Kuhljevi dnevi, Slovensko društvo za mehaniko, Ljubljana. (in Slovene) [10] Malipeddi, A.R., Chatterjee, D. (2012). Influence of duct geometry on the performance of Darrieus hydroturbine. Renewable Energy, vol. 43, p. 292-300, DOI:10.1016/j.renene.2011.12.008. [11] McTavish, S., Feszty, D. Sankar, T. (2012). Steady and rotating computational fluid dynamics simulations of a novel vertical axis wind turbine for small-scale power generation. Renewable Energy, vol. 41, p. 171-179, DOI:10.1016/j.renene.2011.10.018. [12] Yang, B., Lawn, C. (2011). Fluid dynamic performance of a vertical axis turbine for tidal currents. Renewable Energy, vol. 36, no. 12, p. 3355-3366, DOI:10.1016/j. renene.2011.05.014. [13] Rossetti, A., Pavesi, G. (2013). Comparison of different numerical approaches to the study of the H-Darrieus turbines start-up. Renewable Energy, vol. 50, p. 7-19, DOI:10.1016/j.renene.2012.06.025. [14] Intel Corporation (2010). Intel® Fortran Compiler 11.1 User and Reference Guides, from http://nf.nci. org.au/facilities/software/FORTRAN/Intel11.1/en_US/ compiler_f/main_for/index.htm, accessed on 2014-0203. [15] Howell, R., Qin, N., Edwards, J., Durrani, N. (2010). Wind tunnel and numerical study of a small vertical axis wind turbine. Renewable Energy, vol. 35, no.2, p. 412-422, DOI:10.1016/j.renene.2009.07.025. [16] Ansys (2010). Ansys CFX - Solver Theory Guide. Ansys, Inc., Canonsburg. [17] Hriberšek, M., Škerget, L., Poredoš, A. (2005). Process Engineering. Part 1, Basics, Mixing, Drying. University of Maribor, Faculty of Mechanical Engineering, Maribor. (in Slovene) [18] Menter, F., Langtry, R., Völker, S., Huang, P. (2005). Transition Modelling for General Purpose CFD Codes. Engineering Turbulence Modelling and Experiments 6: Procedings of the International Symposium on Engineering Turbulence Modelling and Measurements, Sardinia, p. 31-48, DOI:10.1016/B978-0080445441/50003-0. [19] Kraemer, K. (1961). Flügelprofile im kritischen Reynoldszahl Bereich. Aerodynamische Versuchsanstalt, vol. 27, no. 2, p. 33-46. [20] Simo, J.C., Wong, K.K. (1991). Unconditionally stable algorithms for rigid body dynamics that exactly preserve energy and momentum. International Journal for Numerical Methods in Engineering, vol. 31, no. 1, p. 19-52., DOI:10.1002/nme.1620310103. [21] Betz, A. (1920). Das Maximum der theoretisch möglichen Ausnützung des Windes durch Windmotoren. Zeitscrift für das gesamte Turbinenwesen, vol. 26.

Fleisinger, M. – Vesenjak, M. – Hriberšek, M.

Received for review: 2014-04-08 Received revised form: 2014-06-17 Accepted for publication: 2014-07-08

Managing Vehicle Acceleration Properties by Programming Functions for Engine Torque Control Dunđerski, I. Ivan Dunđerski

University of Belgrade, School of Electrical Engineering and Computer Science Applied Studies, Serbia A new vehicle acceleration properties control system has been developed to improve passenger comfort. Nonlinearity of acceleration influences sensitive passengers, who experience nausea while in a moving vehicle. This control system implements advanced engine torque management for control of the pulling force, which defines vehicle acceleration dynamics with regard to passenger comfort. A theoretical model for the linear acceleration of the vehicle was developed. The function was defined to control acceleration intensity, and it determines conditions for acceleration properties to be linear. Programming code was written to manage engine charge according to acceleration function. A controller has been developed to be the executive device of the programming code, and to manage engine charge. In the experiments, the variable acceleration is converted to constant with the choice of acceleration intensity value, as the driving parameter. Keywords: vehicle acceleration function, mass jerk, engine torque programming control

0 INTRODUCTION Vehicle motion dynamics is a highly developed field. Safety is steadily increasing, fuel consumption is being reduced, and drive comfort is improving. In pursuing vehicle safety, many physical and mathematical models to control the behaviour of vehicle at variable speed, on straight and curved paths, on dry or slippery roads, have been developed. The stability and handling of the vehicle are related to the vector of the friction force between the wheel and the road [1]. The skidding of driving wheels during acceleration is prevented by traction control systems (TCS). Wheel locking while braking is prevented by the anti-lock braking system (ABS). Traction control with all-wheel drive is more complex than the control of two-wheel drive, because the information about the magnitude of traction cannot be obtained on the basis of differences in the rotation speed of the driving and non-driving wheels [2]. While moving through a curve, the lateral sliding of vehicle and vehicle spin around a vertical axis is controlled by an electronic stability program (ESP). By rule, the system integrates both TCS and ABS systems [3]. It is possible to control driving wheel skidding during vehicle acceleration via engine torque and braking the wheels. Control of torque eliminates the cause of skidding [4]. The feedback quickly changes the engine charge for engine torque to follow the nature of the skidding. The consequence is the unstable operation of the engine and the increased toxicity of exhaust gases [5]. Traction control by braking driving wheels eliminates that consequence by “cutting off” excess torque. The system comes into effect only after skidding occurs. Operation of

the system creates noise and vibrations, and wears the friction pairs of the braking system. Most of the applied models have feedback, and they reduce engine torque while simultaneously braking driving wheels. Excess torque keeps the system in operation on the verge of skidding, and cutting off the excess torque is minimal. A reduction of fuel consumption is achieved by reducing engine torque and delaying responses to the gas pedal [6]. The original program maps for the engine charge were “remapped”. The driver is influenced by monitoring his driving style and receiving advice for vehicle movement, especially for the way of acceleration [7]. According to the classic laws of physics, the total increase (change) of vehicle kinetic energy does not depend on the way the vehicle accelerates. In fact, by reducing the torque, engine unsteadiness is reduced, because smaller torque yields smaller acceleration of the vehicle. With minor changes of speed, the engine consumes less fuel. To summarise, engine torque is a common factor to all systems of controlling vehicle dynamics. The driving comfort of passengers has, thus far, been improved by controlling vehicle oscillations on the road. Active suspension systems have been developed to reduce the oscillation amplitudes, extend the period of oscillations and to reduce the vehicle inclination in a curve. Such systems are complex and expensive, because they are based on programming management and energy-demanding devices, with strong amplification, acting on the suspension system. Optional three-level systems have also been developed. The driver selects the characteristics of oscillations, depending on the road and his wishes. The vehicle body height above the road and the intensity

*Corr. Author’s Address: The School of Electrical Engineering and Computer Science Applied Studies, Vojvode Stepe 283, Belgrade, Serbia, ivand@viser.edu.rs

777

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 777-788

of oscillation damping are changed. Along with safety, comfort is improved by reducing vibrations and noise and by air-conditioning the passenger cabin [8]. Improving the driving comfort via vehicle dynamics is a relatively new technique. At first glance, it is difficult to connect driving comfort to the engine torque and find a way to achieve it. The most comfortable drive is along a straight line is at a constant vehicle speed, when there are no inertia effects on passengers. In reality, such driving is possible only on a highway over a longer period of time. The vehicle’s velocity vector changes direction when turning, and intensity while decelerating or accelerating. Moving through the curves and decelerating (braking) is basically a question of vehicle dynamics safety, not comfort. The remaining area where driving comfort can be improved, according to the velocity vector, is in the accelerated motion of the vehicle along a straight line. In that sense, it is possible to control the intensity of the acceleration through its properties, which are the magnitude and rate of change. The intensity of the acceleration determines the magnitude of inertia effect on the passengers. Minimal intensity is defined by law of the minimal dynamic characteristic of the vehicle. Change of acceleration intensity causes the vehicle to jerk, which affects the passengers. Variable acceleration creates Mass Jerk–Yank. In technical terms, mass jerk is well known. It is always taken into account when determining the motion dynamics of robot manipulation arms, tools for computer controlled machines (CNC), crank mechanisms, cam couples, etc. Jerk and mass jerk also occur during accelerated vehicle motion due to variable engine torque. When the engine speed increases, the torque rises, reaches maximum and then decreases. Vehicle transmission reflects variable engine torque into the variable pulling force of the vehicle. As the pulling force increases and decreases, the acceleration intensity changes along with it, producing jerk and mass jerk of a variable sign. Jerk creates effects similar to the effect of a ship tilting on lateral waves, also with a variable sign. Sensitive passengers experience nausea, called “seasickness” [9]. In order to reduce the intensity and jerk direction change, passenger ships are equipped with gyroscopic masses. The gyroscope diverts the acting force so that the lateral oscillations of the ship around the longitudinal axis are transferred into longitudinal oscillations around the lateral axis. 778

Ship inertia around the lateral axis is significantly greater than inertia around the longitudinal axis, so the amplitude of the oscillation decreases while the period of the ship oscillations increases. In that way, the comfort of cruising is also improved for the passengers who are not sensitive to the seasickness. In order to improve driving comfort during accelerated vehicle motion, the jerk should be eliminated. Nonlinear vehicle acceleration should be converted into a linear (constant) one. For this purpose, part of the vehicle pulling force, which remains available after overcoming moving resistance, should be constant along the vehicle acceleration path. Controlling the vehicle movement dynamics by controlling engine torque eliminates jerk and improves passenger comfort. In order to control engine torque during vehicle acceleration, the engine torque curve family for acceleration is required. This family differs from the stationary state torque family. Part of the engine torque, which accelerates the vehicle, is reduced in relation to the stationary state due to the inertia of masses that move in a straight line (translate) and rotate, which are not present while the vehicle is moving at a constant speed. In addition to the influence of inertia, during accelerated vehicle motion, the engine works in an unsteady state, and the engine efficiency is decreased. There are test benches at which the engine torque family is obtained for accelerated vehicle movement. Resistances to accelerated vehicle movement are simulated on the bench by means of programming management [10]. However, accurate results can be obtained only by recording the vehicle motion in real conditions. Therefore, the engine torque family in this research is obtained by recording acceleration data of the experimental vehicle. For the model of the accelerated vehicle motion dynamics, university textbooks were used [11] and [12]. 1 PHYSICAL – MATHEMATICAL MODEL 1.1 Balance of Forces Equations and Resistance to Motion During accelerated vehicle movement on the road, vehicle pulling force Fpull acts against road resistance R and inertia resistance Rin. The balance equation of vehicle pulling force with movement resistance during acceleration is expressed by a differential equation [11]: G dv dv 1 Fpull = KAv 2 + Gf cos α + G sin α ± ±J , g dt dt rd2

Dunđerski, I.

Fpull = R + Rin .

(1)

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 777-788

Road resistances are often referred to as “outside resistance”, because they act on the vehicle from “outside”. Pulling force Fpull is produced by engine torque T, which is conducted through the transmission to the vehicle drive wheel [11]: Fpull

are:

T = ηtr i ⋅ . (2) rd

Road resistances R during the vehicle movement R = FD + Frr + Fcl , R = ηtr i ⋅

TR . (3) rd

Road resistance elements are: FD = 0.5 ρ ⋅ cx ⋅ A ⋅ v 2 ,

With the path being horizontal, α = 0, the equation of balance of the forces and torques for vehicle on the horizontal road is:

Equations of balance of forces and torques of the front wheel and the equation of balance of forces and torques of the rear wheel are:

where FD and K = 0.5 ρ cx are given in [11]. Vehicle weight G is substituted as mg. Resistances R of the road exist during constant speed and accelerated vehicle movement. A vehicle accelerates, overcoming inertial resistances Rin by force F, which is part of the pulling force. The force F is generated from a portion of engine torque ΔT that remains available after overcoming road resistance R to movement. Rin = m

dv dω ∆T +J , F = ηtr i ⋅ , ∆T = T − TR . (5) dt dt rd

Resistances Rin to acceleration exist only during accelerated movement of the vehicle. The first member Rin in Eq. (5) represents the inertia of translating mechanical elements, and the second member represents the inertia of rotating mechanical elements. Fig. 1 shows the physical model of the vehicle movement, while accelerating along the straight path (the often used “bicycle” model for 1/2 of the vehicle).

X r rd = J rω r . (7)

T f − ( X f − X r ) rd = J f ω f + J rω r . (8)

Eliminating the tangent reactions (Xf – Xr) from Eq. (8), using Eq. (6), gives: M ⋅ rd ω f + R =

1 T f − ( J f ω f + J rω r )  , a = rd ω f . (9)  rd 

In experiments, the angular speed of each wheel was measured via sensors for wheel turning, which showed that the angular speeds of the front and rear wheels were almost identical. Therefore, the mathematical model can be simplified:

ω f = ω r = ω . (10) Substituting Eq. (10) into Eq. (9) gives:

( Mrd ω + R ) rd

= T f − J ω , J = J f + J r . (11)

From Eq. (11), torque equation for the driving wheel of the vehicle is obtained:

T f − T fR = ( Mrd2 + J ) ω , T fR = R ⋅ rd . (12)

1.2 Engine Speed during Accelerated Motion Engine torque T and road resistance torque TR are obtained from engine torque Tf and resistance torque TfR acting on the driving wheel, using overall transmission gear ratio i. Acceleration dn/dt of engine is obtained from the acceleration dω/dt of the driving wheel using the overall transmission gear ratio i and relation of angular speed to frequency ω=2πn (n is an integer).

Fig. 1. Forces and torques acting on the vehicle during accelerated motion

T f − X f rd = J f ω f ,

Summing equation (Eq. (7)) gives:

Frr = mg ⋅ f ⋅ cos α , Fcl = mg ⋅ sin α , (4)

M ⋅ a = X f − X r − R. (6)

T f = iT , TR = iT fR , i ⋅ ω = 2π

dn . (13) dt

Substituting the value from Eq. (13) to Eq. (12) gives:

Managing Vehicle Acceleration Properties by Programming Functions for Engine Torque Control

779

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 777-788

i (T − TR ) = ( Mrd2 + J )

2π dn ⋅ . (14) i dt

From Eq. (14) arises the equation for engine speed rate of change dn/dt during vehicle acceleration as a function of engine torque T and road resistance torque TR on the engine:

dn i2 = ⋅ (T − TR ) . (15) dt 2π ( Mrd2 + J )

In Eq. (15), the difference (T – TR) between engine torque and road resistance torque on the engine is torque ΔT that remains available to accelerate the vehicle after overcoming the road resistance R.

∆T = T − TR , k =

i2 . (16) 2π ( Mrd2 + J )

The constant k determines the influence of overall transmission gear ratio i and the inertia of masses M that translate and masses J that rotate during vehicle acceleration. 1.3 Engine Torque Curve Family From Eq. (15), engine torque T, required to overcome the road resistance torque TR and the inertia k–1·(dn/dt) during accelerated vehicle movement, is calculated: T=

2π ( M ⋅ rd2 + J ) dn dn ⋅ + TR , T = k −1 ⋅ + TR . (17) 2 i dt dt

Eq. (17) is used to obtain a family of torque curves for accelerated vehicle moving. Engine speed n during acceleration of the vehicle is taken from the speed sensor of the crankshaft. With engine speed n in time t of vehicle acceleration being known, the dependence n = f(t) is also known, and it is used to calculate the rate of change dn/dt in time t. Torque TR of resistance is calculated according to Eqs. (3) and (4). Digital data on values of engine family torques T, in relation to flap angle θ (engine charge) and the engine speed n, are stored in the matrix (Eq. (18)), (lookup table for the torque family), from which they are read by the controlling program during vehicle acceleration. Fig. 2 shows a graphical representation of the engine torque family. For this representation, the curve families are drawn “smoothly” through the points of the torque family from the matrix (Eq. (18)), by a special algorithm for data processing. 780

To define the gas pedal position and thereby the step of engine charge (load) per torque family, at an earlier stage, a mechanical spacer was used, working on the principle of the screw and nut. The electronic throttle control (ETC) pedal was physically limited to a certain height from vehicle floor. At a later stage, a function for the controller was written to determine gas pedal position as a percentage of the deflection angle (% DA) of ETC. The minimal step value is 1 as an integer. Data on the family are formed in steps of 2% DA throughout the entire range, except for the two curves with a minimum charge, 25 and 28% DA, where the step is 3% DA because of poor engine performance. The maximum physical position of the accelerator pedal position (full throttle) corresponds to 80% DA. This correspondence comes from the fact that the ETC deflection angle must be greater than the accelerator pedal position at full throttle. Otherwise, the ETC DA sensor could be damaged. Consequently, full throttle on Fig. 2 is given as 78% DA and represents an “outside engine torque characteristic”, which is the maximal torque that the engine can produce throughout the engine speed range. The curves were recorded multiple times in order to verify them. The vehicle acceleration path is a straight line, the road is horizontal, and the surface is hard (asphalt) and dry.

Fig. 2. Engine torque curves family

Conclusions from Fig. 2 about the shapes of curves are that above 50% DA, torque curves are similar in their shape/nature. Full engine performance is not changed for safety reasons, which are a priority. Naturally aspirated Otto engines, such as the one in the experimental vehicle, with an increase in engine speed, have an increase, a peak and a decrease in value of torque (also inflection points).

Dunđerski, I.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 777-788

Below 50% DA, torque curves differ in their shape/nature. Torque maximums are shifted to lower engine speeds, and the curves have two maximums. The increase in torque at lower engine speed allows moving the engine operation points from higher speeds to lower speed intervals. Engine operation with a smaller number of cycles saves fuel because the piston friction path is shortened, and the engine life is extended. The increase in torque (and power) at lower speeds is a feature of turbocharged engines, here achieved by remapping fuel maps of the engine control unit (ECU). Conclusions from Fig. 2 about the density of curves are: 1. Above 50% DA, engine performances are maximal, because the fuel is not saved at the expense of vehicle safety. The curves’ density is such that virtually all is “full throttle”. 2. Below 50% DA, at which the engine performance differs, the curve density is lower. This allows more accurate control of engine charge via the gas pedal. 1.4 Engine Torque and Acceleration Function Values of torques T of the engine family torque curves in the matrix (look-up table) are defined by:

 T11 … T1q      , (18) T T  pq   p1

where Tij is torque of the engine, for engine speed ni and flap angle θj of engine aspiration system. There is no explicit mathematical relation between engine torque and engine speed. The flap angle θ, needed to obtain the desired torque T at number n of engine speed, is calculated with numerical methods, from the matrix Eq. (18).

The programming model of the acceleration function θ(n) is stored in the controller module as an adaptive torque generator (ATG), shown in Fig. 3. The programming model code is connected with the executable code of the controller. The executive code of the controller is stored in the controller module for flap angle calculation, called the throttle position calculator (TPC). The ATG program module sends signals for torque values to the calculation module TPC TATG (desired torque), in each controller cycle. The values of the desired torques are calculated according to acceleration function θ(n). The TPC module calculates the flap angle θATG on the basis of value of the TATG torque, obtained from ATG and the current engine speed n. Signals carrying the value of the angle θATG are relayed by the flap angle calculator to the ECU. In response to these signals, the ECU sets the flap to angle θe = θATG and thereby determines the charge (load) of the engine. The controller is inserted between ETC and ECU, which were installed by vehicle manufacturer, and connected to the common “drive by wire” control system. At the start of acceleration, the driver determines desired final speed of the vehicle by using the gas pedal. The controller stores the driver’s desire and controls acceleration until the desired vehicle speed is reached. Once the desired speed is reached, the controller returns control to the driver. During acceleration the ECU “does not know” that it receives information about the value of the accelerator pedal position from the controller, not from the driver.

T = θ ( n ) . (19)

The function θ(n) of acceleration in Eq. (19) determines the desired torque T for a given engine speed n. The choice of acceleration properties gives the acceleration function curve θ(n), which generally intersects the family curves T(n;θ) of the engine torques. The way in which the acceleration function curve will intersect the engine torque curve family is determined by the chosen properties of the acceleration.

Fig. 3. Block diagram of Torque controller

The driver sets the desired ETC position; the ETC transmits the signal to the controller; the controller determines engine torque TATG and its corresponding

Managing Vehicle Acceleration Properties by Programming Functions for Engine Torque Control

781

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 777-788

throttle valve angle ƟATG and sends it to the engine ECU; the engine ECU sets the throttle valve angle Ɵe to the calculated value ƟATG; the engine produces engine torque Te whose value is TATG, and delivers it to the transmission. The current engine speed n is monitored and sent back to ATG controller. In Fig. 4, the image of the controller is shown.

Mass jerk Y causes non-constant force F, which accelerates mass m, i.e. vehicle. Mass jerk Y can be obtained starting with second Newton’s law F = ma and applying derivative by time: (d/dt)F = (d/dt)ma = m(da/dt), Y = dF/dt = mj. Jerk j on a vehicle arises from changing the intensity d2ω/dt2 of angular acceleration of the driving wheels. The angular speed ω is related to the speed v of vehicle movement on a straight path, with expression v = rdω, where rd is the dynamic radius of the wheel. Using this in Eq. (20) gives the jerk of the vehicle: j=

Fig. 4. Controller

Both the program model with its managing code and the controller with its executable code have been developed specifically for this research. This research explores function of acceleration with property that the differences between engine torque T and torque TR of resistance on the engine are constant. The vehicle accelerates without jerk with the desired intensity of acceleration. The constant difference of torques yields a constant difference F = const part of force Fpull that pulls the vehicle, which is available for vehicle acceleration after overcoming resistance R to motion: F = Fpull – R. 1.5 The Function of Torque Constant Differences Jerk j is defined as the third derivative of the position vector with respect to time, which is the rate of change of the acceleration a. For the acceleration a along the x-axis, jerk j is (scalar) [13]:

da d 2 v d 3 x = j = = = , j a= v=  x . (20) dt dt 2 dt 3

Mass jerk Y (Yank), [13] which is caused by jerk j during variable acceleration of the mass m is by definition: da d = ( ma ) , dt dt dF d . (21) F = m⋅a ⇒ Y = (F ) = dt dt Y = m⋅ j = m

782

2 d 2 v d ( rd ω ) d 2ω = = rd 2 = rd ω, v = rd ω. (22) 2 2 dt dt dt

Change of intensity d2ω/dt2 of the angular acceleration of the driving wheels occurs by changing the intensity d2n/dt2 of the angular acceleration of the crankshaft. Jerk je of the angular acceleration on the engine crankshaft is obtained through the overall transmission gear ratio i and angular speed to circular frequency relation ω = 2πn for the whole number n of revolutions.

je = ηtr ⋅ i ⋅ j = 2π rdηtr i ⋅ n, ω = 2π n. (23)

Mass jerk YF on vehicle depends on the intensity dF/dn of change of the force vector F, Fig. 5, which accelerates the vehicle along the x-axis and causes a jerk je on the engine. According to Eq. (21) it follows:

YF =

dF dF dn = ⋅ , dt dn dt

dF YF = . (24) dn n

In Eq. (24), the change dn/dt of engine speed is determined by Eq. (15). The force F which accelerates the vehicle is part of the force Fpull (Eq. (2)) of pulling, which remains available for acceleration of the vehicle after overcoming resistance R to motion (Eq. (3)):

F = Fpull − R. (25)

The force F accelerates the vehicle overcoming inertia while accelerating (Eq. (5)). Fig. 5a shows curves of force Fpull = f(n) of the vehicle pulling in dependence to engine speed n according to Eq. (2) and forces R = g(n) of resistance to movement of the vehicle Eq. (3), where the engine speed n = ωi/2π is expressed by the angular speed ω of the wheel, the overall transmission gear ratio i and the relation of the angular speed to the circular frequency. In Fig. 5b, vertical segments Fq represent the difference Fpull – R between the vehicle pulling force

Dunđerski, I.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 777-788

and movement resistance force. These segments of pulling force remain available for vehicle acceleration, after overcoming resistances R = g(n), and they represent values of force curve F = h(n). Values Fq in the top diagram are equal to values Fq in the bottom diagram for each arbitrary q. In the theoretical diagrams in Fig. 5, value F1 corresponds to engine speed n1 = nmin which is the minimal engine speed required for the engine to operate in practice. Depending on the engine’s construction, nmin varies from (800/60) to (1100/60) s–1. The intersection of functions f(n) and g(n) corresponds to the maximal engine speed achievable for given external resistances. For that engine speed value, the engine does not have the pulling force needed to accelerate the vehicle further.

Fig. 5. a) Force Fpull = f(n) of pulling and resistance R=g(n) force; and b) force F = h(n) which accelerates the vehicle

Fig. 5b shows that as the engine speed n increases, the force F, which accelerates the vehicle at first rises; dF/dn > 0, reaches maximum, dF/dn = 0, and then falls, dF/dn < 0. Mass jerk, YF created by changing intensity of force F, which accelerates the vehicle, acts in the direction of movement. The condition for mass jerk elimination is obtained from Eq. (23), by putting that je = 0:

je = 2π rdηtr i ⋅ n = 0, dn = 0, n = const. (26) dt

From Eq. (26), it follows that there is no mass jerk when the change of angular speed during vehicle acceleration is constant, dn/dt = const. According to Eqs. (16), (17) and (26), it follows:

2π ( M ⋅ rd2 + J ) dn ∆T = T − TR = ⋅ , i2 dt (27) ∆T = k −1 ⋅ n = const.

where k is given in Eq. (16). From Eq. (27), it follows for engine torque T of:

T = TR + ∆T , n = const. (28)

Torque T is guided along acceleration function θΔT(n), so according to Eq. (28) the acceleration function is: T = θ ∆T ( n ) , θ ∆T ( n ) = TR ( n ) + k −1 ⋅ n ,

n = const.

(29)

When the condition dn/dt = const is fulfilled, the vehicle accelerates by function θΔT(n) (Eq. (29)), with the property that acceleration is without jerk. Eq. (29) represents acceleration function θΔT(n), which has the property that during vehicle acceleration, the difference ΔT between engine torque T and resistance to motion torque TR on the engine is constant. The value of torque difference ΔT that accelerates the vehicle in Eq. (27) is a variable arbitrary parameter that defines the intensity of the acceleration of the vehicle. Theoretically, to drive with the best comfort, the minimal value of the arbitrary parameter should be taken, because when ΔT→0 acceleration is minimal, and the impact of inertia is minimal. In practice, the parameter value should satisfy legal limitations for the dynamic vehicle characteristic needed for safety during acceleration. Higher values of the chosen parameter increase vehicle acceleration, resulting in a greater influence of inertia on passengers, which decreases the comfort of driving. Substituting into Eq. (2) that Fpull = FθΔT(n) and that T = θΔT(n) it follows:

Fθ∆T ( n ) = ηtr itr

θ ∆T ( n ) . (30) rd

The graphical representation of curve FθΔT(n) of vehicle pulling force, which corresponds to the function θΔT(n) acceleration without jerk, is given in Fig. 6a, and the graphical representation of curve F = h(n) force that accelerates the vehicle without jerk is given in Fig. 6b. Each point of the function curve FθΔT(n) is located at the same distance Fq = const along the ordinate, above the curve of function R = g(n) (Fig. 6a). This

Managing Vehicle Acceleration Properties by Programming Functions for Engine Torque Control

783

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 777-788

distance represents intensity of part of pulling force, that accelerates the vehicle, (Fig. 6b). Therefore, the mass jerk YF is eliminated along the largest part of the acceleration path. At both the start and the end of the acceleration, there is a short-term jerk, and variable force F along with it, which accelerates the vehicle, because it is not possible to transit from constant speed movement to accelerated movement without jerk (similar to the movement of a numerical machine tool). However, the swinging effect is avoided.

of engine torque, above the curve TR(n) of resistance torque of the road, over a real family of engine torque curves of the experimental vehicle.

Fig. 7. Curve of constant difference of torques, computer

Fig. 6. Function FθΔT(n)of acceleration with constant difference Fq = const in relation to resistance a) R = g(n) of the road and b) force F = h(n) which accelerate vehicle

When braking, the vehicle braking force is constant on the largest part of the path [14] and [8] so the diagram of braking force is very similar to the acceleration diagram in Fig. 6b, along with the negative sign (deceleration of the vehicle). This fact implies that the nonlinear acceleration of the vehicle is the primary cause for the effect of swinging during the drive. 2 EXPERIMENTS 2.1 Computer Simulation In the early stage of research, the computer used Lyderman’s formula [11] and [12], which gives an explicit mathematical relation of engine power to engine speed. Lyderman’s formula for power was previously transformed into the formula for the torque. At a later stage of research, the computer used a real torque family, obtained by recording on the vehicle and applying the Eq. (17). Fig. 7 shows a computer simulation of acceleration with a constant difference ΔT = const 784

The curve θΔT(n) comp. sim. of engine torque lies above the curve TR(n) of the function of resistance torque of the road. Vertical segments between the two curves are equal by value across the entire range of engine speeds. The two curves correspond to the force F that accelerates the vehicle, according to Eqs. (29) and (30). Curve θmax(n) in Fig. 7 connects maximal torques of partial engine charges, illustrating the possibility of programming other acceleration functions. The characteristic “jagged” shape of θΔT(n) comp. sim. originates from the discretization of the ETC flap angle. The vertical segments of the curve represent the discreet change of the Δθe angle, through which the engine charge (load) is changed. Sloped segments between vertical segments are for the same valve angle, θe = const. Across sloped segments, engine speed n is changing along with engine torque, with the same flap angle (load). The value of discrete change of the flap angle Δθe programming step is an arbitrary parameter, defined by the programming model. The minimal value of the arbitrary parameter for the vehicle is limited to the minimal step value of the mechatronic system that controls the flap angle. 2.2 Vehicle Experiments were conducted on the VW Polo 1.2 laboratory vehicle with Simos N9.1 ECU Except for the installed controller, the vehicle is completely serial.

Dunđerski, I.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 777-788

2.2.1 Experimental Setup The values of the physical parameters used in Eq. (4) related to vehicle and road during experiments are: ρ = 1.226 kg/m3, cx = 0.30, A = 1.75 m2, m = 1200 kg (vehicle mass with two passengers), g = 9.81 m/s2, f = 0.018, α = 0. For practical calculations, the programming model uses a modified formula for k, given in Eq. (16):

i2 J , δ M = M + 2 , (31) rd 2π ⋅ rd2 ⋅ δ M

where coefficient δ is calculated according to δ = 1 + δ1·(ig)2 + δ2, δ1 = 0.040, δ2 = 0.035 are adopted as mean values from [11]. The estimation of the coefficient δ is applied because of difficulties noted to determine δ by theoretical formula [11]. The value of coefficient δ = 1+0.04·1.392+0.035 = 1.11. The value δM = 1.11·600 = 666 kg. The value J itself is not needed for calculation, but it can be obtained as J = (δ 1·(i g) 2 + δ 2)M(rd) 2 = (0.04·1.392+0.035)·600· ·0.2832 = 5.39 kgm2. The J value is small in respect to M, and thus has no significant influence to coefficient k in practice. Coefficient k influences the magnitude of torque T (Eq. (17)) but does not affect the nature of torque T curve. Values of physical parameters i, M, rd used in Eq. (31) related to vehicle during experiments are: M = 1/2 m, i = ωe / ωf, and ig = i / i0 = i / 4.19. The programming model calculates overall transmission ratio i for any transmission gear, using data from sensors for engine speed ωe and front wheels’ angular speed ωf. This simplifies the calculation algorithm by eliminating the need for additional input related to the current transmission gear and makes the programming model portable to any contemporary vehicle. In addition, no gearbox sensors are needed to inform the controller which gear is currently coupled. Value of rd was calculated once according to Eq (32):

rd =

v i⋅v i ⋅v = = , (32) ω f ωe 2π n

where v = 41 km/h = 11.39 m/s measured using navigation device (GPS), n = 2240 rpm = 37.33 s–1 measured using Bosch KTS 640 scanner, and i =

i0 to i3 = 4.19 to 1.39 = 5.82 adopted from vehicle manufacturer specification. According to Eq. (32) and given values, rd is 0.283 m. 2.2.2 Experimental Results Fig. 9 shows the curve θΔT(n) of the acceleration function, obtained experimentally by recording on the vehicle. Aperiodic oscillations in the curve are due to the noise accompanying the recording, since the flap changes its angle in discrete steps. Between two changes, the flap is standing still.

Fig. 8. Curve of constant torque difference, vehicle

The value of the optional parameter ΔT = const, (Eq. (27)) is available for the driver to set, by pressing the gas pedal. The curve in Fig. 9 was recorded with a value of 60%DA. Optional parameter ΔT corresponds to the force FθΔT(n), which accelerates the vehicle (Eq. (30)), by the function θΔT(n) of acceleration (Eq. (29)). At the beginning of the acceleration path, the vehicle is moving at low speed in the third gear of the transmission. In this gear, the vehicle has a top speed over 100 km/h, at which the influence of air resistance is visible in the diagram of the experimentation results. In the first transmission gear, the vehicle’s top speed is too low to produce air resistance sufficiently significant to be visible on the diagram. At the starting moment, Point 1, on Fig. 8, the gas pedal is rapidly pressed and held down until the end of the acceleration, Point 4. At Point 1, the starting of the acceleration curve, engine speed is ~1200 revolutions per minute (RPM), and the torque value is ~33 Nm. With the pressing of the gas pedal, the engine charge (load) rises and a small increase of torque to ~38 Nm follows, along with a small increase of engine speed to ~1300 rpm, point 2. In Point 2, the control begins. The controller modifies current movement mode of the vehicle

Managing Vehicle Acceleration Properties by Programming Functions for Engine Torque Control

785

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 777-788

to achieve the desired movement according to acceleration function θΔT(n). The convergence of the actual torque curve to the desired torque curve, defined by acceleration function, takes place gradually during the transitional period between Points 2 and 3. The controller calculates engine torque T from the current engine speed n, flap angle θ and resistance torque TR of the road, depending on current vehicle speed. In each cycle, the desired torque TATG is calculated so that its value is between engine torque T and the value of acceleration function θΔT(n). In that way, the actual acceleration curve approaches the theoretical one θΔT(n), as engine speed n increases in the interval between Point 2 and 3 (Fig. 8). This procedure avoids the jerk that would arise by rapid fall of engine torque T to the value of the function θΔT(n). Rapid and complete gas seizure would cause the engine to go from pull to brake, reversing the pulling force direction, which would generate jerk. From Point 3 to 4, the difference between torque T and torque value of acceleration function θΔT(n) are small enough that desired torque TATG matches the value of torque by function θΔT(n). In this way, the acceleration takes place by θΔT(n) a function of acceleration, with a ΔT(n) = const, according to Eqs. (27) and (29). At Point 4, the vehicle acceleration ends by function θΔT(n) of acceleration (electronically limited maximum engine speed is reached with a sharp power drop). From Fig. 8, it is clear that during vehicle acceleration, between Points 3 and 4, the vertical distances between the curves θΔT(n) and TR(n) n are practically the same. These distances represent the intensity of part of the engine torque ΔT = const that accelerates the vehicle, or remains available for increasing the vehicle speed after overcoming road resistance torque. Control of torque between Points 3 and 4 is achieved by the programming model of the controller, and the influence of the driver on the vehicle movement is excluded. 3 CONCLUSIONS 3.1 Introduced Concept This paper shows the concept for excluding mass jerk that affects vehicle during acceleration, which improves passenger driving comfort. A new system for engine torque control has been developed. The system eliminates mass jerk on the 786

largest portion of the vehicle acceleration path by maintaining constant vehicle acceleration intensity. This principle can be applied to model and control other acceleration properties, such as engine operation efficiency, fuel consumption or driving wheel slip. 3.2 Methodology Applied The physical and mathematical model is introduced to determine forces and torques acting on the vehicle during acceleration. According to the model, the pulling force is the sum of two parts. The first part overcomes resistance to vehicle motion on the road, while second part overcomes the inertia of rotating and translating masses. Pulling force is maintained so that its part that overcomes inertia is of constant intensity. In this manner, vehicle mass jerk is eliminated during acceleration. The acceleration function, programming model, and the controller were developed in order to control pulling force. The programming model instructs the controller to set the engine charge in respect to acceleration function. Managing the charge is discrete and iterative. The torque curve family for the experimental vehicle was recorded in actual vehicle acceleration conditions on the road. The data regarding the torque family is arranged in matrix (lookup table). The acceleration function evaluates throttle angle that produces the desired torque in real time from the matrix for current engine speed. Acceleration control is automatic. The driver sets the intensity of the acceleration by pressing the accelerator pedal only at the start of acceleration. 3.3 Results of Research In the first phase of research, a special programming module was written to simulate vehicle acceleration. The module requires input for vehicle mass, current gear ratio, starting engine speed and desired acceleration intensity. The main focus of output is a torque curve, showing engine states through time. Other various data for current simulation state are shown, such as current power, vehicle speed, elapsed time, distance travelled, etc. In the second phase of the research, the experiments were conducted while driving the vehicle on the road. During vehicle acceleration, the engine speed in time was recorded. Analysis of data obtained by both simulation and experimentation on the vehicle shows good results for

Dunđerski, I.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 777-788

the control of pulling force intensity. On the output diagram, the engine torque curve is located above road resistance torque curve. The difference of torque values for these two curves is practically the same at any given engine speed. Therefore, the part of pulling force that accelerates vehicle is considered to be constant, eliminating mass jerk during acceleration. 4 NOMENCLATURE A frontal area of the vehicle [m2] a vehicle acceleration [m/s2] air drag coefficient [-] cx f rolling resistance coefficient [-] G vehicle weight [N] g gravity acceleration [m/s2] F vehicle acceleration force [N] climbing resistance force [N] Fcl air resistance (drag) force [N] FD rolling resistance force [N] Frr vehicle pulling force [N] Fpull i overall transmission gear ratio [-] main transmission gear ratio [-] i0 gth gear transmission ratio [-] ig J moment inertia of rotating masses [kgm2] moment inertia of rotating masses of front Jf wheel, transmission elements, engine and clutch [kgm2] moment inertia of rear wheel [kgm2] Jr j jerk [m/s3] M physical model mass [kg] m vehicle mass [kg] n engine speed [s–1] R road resistances [N] inertia resistances [N] Rin dynamic radius of the driving wheel [m] rd T engine torque [Nm] engine torque transmitted to front driving Tf wheel [Nm] road resistance acting on front wheel as TfR torque [Nm] road resistance acting on engine as torque TR [Nm] v vehicle speed [m/s] road tangent reaction to front wheel [N] Xf road tangent reaction to rear wheel [N] Xr Y mass jerk [kgm/s3] vehicle mass jerk [kgm/s3] YF road normal reaction to front wheel [N] Zf road normal reaction to rear wheel [N] Zr α road inclination angle [°] δ inertia influence coefficient of rotating masses [-]

δ1 inertia influence coefficient of rotating masses of engine and clutch [-] δ2 inertia influence coefficient of rotating masses of drivetrain (wheels) [-] ηtr transmission efficiency coefficient [-] ρ air density [kg/m3] ω angular speed of wheels [rad/s] ωf angular speed of front wheels [rad/s] ωr angular speed of rear wheels [rad/s] 5 ABBREVIATIONS DA ECU ETC RPM TPC

deflection angle of ETC engine control unit electronic throttle control revolutions per minute throttle position calculator 6 REFERENCES

[1] Canudas, C., Tsiotras, P., Velenis, E., Basset, M., Gissinger, G. (2003). Dynamic friction models for road/tire longitudinal interaction. Vehicle System Dynamics, vol. 39, no. 3, p. 189-226, DOI:10.1076/ vesd.39.3.189.14152. [2] Deur, J., Pavkovic, D., Burgio, G., Hrovat, D. (2011). A model-based traction control strategy non-reliant on wheel slip information. Vehicle System Dynamics, vol. 49, no. 8, p. 1245-1265, DOI:10.1080/00423114.2010. 511675. [3] Masuno, K., Nitta, R., Inoue, K., Ichikaw, K., Hiwatashi, Y. (2000). Development of a new all-wheel drive control system. Proceedings of 2000 FISITA World Automotive Congress, Seoul, Technical paper. no. F2000G347. [4] Jae-Book, S., Byong-Cheol, K., Dong-Chul, S. (1999). Development of TCS slip control logic based on engine throttle control. KSME International Journal of Mechanical Science and Technology, vol. 13, no. 1, p. 74-81, DOI:10.1007/BF02946125. [5] Shin, M., Han, J., Youn, J., Sunwoo, M. (2006). Design of a network-based traction control system using a formalized design procedure. Proceedings of Yokohama FISITA World Automotive Congress, Yokohama, Technical paper no. F2006SC22. [6] Fuji Heavy Industries Ltd. (2007). Subaru intelligent drive (SI drive), from http://drive2.subaru.com/ Summer07_whatmakes.htm, accessed on 2013-02-06. [7] Ford motor company newsroom (2011). All-new ford focus features ecomode to help drivers perfect ecodriving techniques, from http://media.ford.com/article_ display.cfm?article_id=33965, accessed on 2013-0208. [8] Kost, F., Post, W. (eds.) (2006). Safety, Comfort and Convenience Systems. Robert Bosch GmbH, Cambridge.

Managing Vehicle Acceleration Properties by Programming Functions for Engine Torque Control

787

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 777-788

[9] Wikipedia. (2014). Motion Sickness, from http:// en.wikipedia.org/wiki/Motion_sickness, accessed on 2014-02-19. [10] Joly, P., Duboc, S. (2000). Common rail mapping on power train test bench. Proceedings of Seoul 2000 FISITA World Automotive Congress, Seoul, Paper no. F2000A054. [11] Janković, D., Ivanović, G., Todorović, J., Rakicević, B. (2001). The Theory of the Movement of Motor Vehicles.

788

Faculty of Mechanical Engineering, Belgrade. (in Serbian) [12] Simić, D. (1977). Motor Vehicles. Science book, Belgrade. (in Serbian) [13] Wikipedia. (2014). Jerk physics, from http:// en.wikipedia.org/wiki/Jerk_(physics), accessed on 2014-02-19. [14] Todorović, B. (1988). Braking of Motor Vehicles. Department of textbooks and teaching aids, Belgrade. (in Serbian)

Dunđerski, I.

Received for review: 2014-02-14 Received revised form: 2014-05-30 Accepted for publication: 2014-07-22

Hybrid Self-Organization Based Facility Layout Planning Potočnik, P. – Berlec, T. – Sluga, A. – Govekar, E. Primož Potočnik* – Tomaž Berlec – Alojz Sluga – Edvard Govekar University of Ljubljana, Faculty of Mechanical Engineering, Slovenia A novel hybrid two-stage method of facility layout planning, based on self-organized clustering, is presented. In the first stage, a self-organizing map (SOM) is applied in order to organize the production process into production cells which encapsulate products with similar properties and similar machining requirements. In the second stage, the internal layout of each cell is optimized by an expert operator, taking into account various local restrictions, technological specifications, and methods of transportation. The method combines the advantages of both algorithmic and manual expert-based approaches to layout planning. The proposed method was applied in a real production company environment with promising results indicating a 42% reduction in intensity-length efficiency measure with respect to the current layout. Keywords: facility layout problem, self-organization, hybrid layout method, cellular manufacturing

0 INTRODUCTION Optimization of production costs in modern production can be accomplished by methods such as lean production [1] or the implementation of cellular manufacturing systems [2]. Various cell formation methods have been proposed [3] to [6], and promising results have been reported by using a self-organizing map (SOM) as a clustering tool [7] and [8]. The self-organization principle has also been proposed for facility layout planning [9] and the organization of distributed manufacturing systems [10]. Other proposed approaches for solving the cell formation and facility layout problems include partitional clustering [11], the correlation analysis approach [12], and various evolutionary, genetic and ant colony optimization algorithms [13] to [17]. An extended approach to layout planning based on the self-organization principle, which includes machines, products and the relevant attributes of products was proposed in [18]. The algorithmic approaches referenced above rely on a mathematical formulation of the facility layout problem, and provide solutions that usually do not take into account the variety of local placement restrictions with respect to size, weight, installations, technological specifications, or transportation. The algorithmic approaches thus only support human operators in creating the final layout [15]. An alternative, widely accepted approach is to apply suitable software for the manual creation and management of layout solutions [19]. Only in the case of restricted small-scale layout problems, a solution has been proposed that combines both automatic and manual modules for facility layout planning [20]. In order to combine the algorithmic formation of production cells and the manual expert-based layout planning approach, a novel hybrid two-stage facility

layout planning method is proposed in this paper. The method consists of two stages: 1) the automatic selforganized formation of production cells, and 2) expert operator based fine layout planning, which finalizes the layout details. The proposed method uses the algorithmic approach in order to decompose the initial large-scale layout problem into smaller sub-problems that can be efficiently managed and solved by expert operators. The method is demonstrated on production data obtained from a manufacturing company KGL d.o.o. The paper is organized as follows: the production data applied in this study are discussed in Section 1, whereas Section 2 presents the self-organized method for the formation of production cells, and Section 3 describes the proposed fine layout planning method. Results and discussion are presented in Section 4, and the conclusions are summarized in Section 5. 1 PRODUCTION DATA The production line comprises mechanical services on CNC lathes and machining centres, high-pressure aluminium die casting and cast processing, the pressing of sheet metal, the fabrication of cylinders for gasoline engines, and the assembly of parts manufactured in blanks. Operational data from the company comprise 252 products with a description of the operations required to manufacture each product, such as: band cutting, thermal cutting (laser, plasma), broaching, drilling, 3-axis CNC machining, brushing, welding, etc., and additional properties of each product as follows: materials, shape, dimensional accuracy, appearance of the product, request for examination, need for the protection of parts, weight, volume, number of assembly parts, number of operations, number of possible variants, and value. Based on the

*Corr. Author’s Address: University of Ljubljana, Faculty of Mechanical Engineering, Aškerčeva 6, 1000 Ljubljana, Slovenia, primoz.potocnik@fs.uni-lj.si

789

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 789-796

required operations and the additional properties, each manufactured product can be described as shown, by means of an example, in Table 1. Table 1. An example of product description with required operations and properties Product ID: 147035 Operations 1. Band cutting 2. CNC turning 3. Service-dip galvanizing 4. Assembly 5. Testing 6. Progressive compression 7. Packaging

Properties Material: Weight: Volume: Shape: No. of parts: No. of operations: Dimens. accuracy: Appearance: Examination: Protection: Value: Quantity:

Fe 1.7 kg 0.64 dm3 round 5 8 0.01 very important functional anticorrosion 7.81 €/part 18,000 per year

good properties of a SOM [21]. Compared to hierarchical clustering and k-means clustering, SOM clustering yields good clustering quality (expressed by silhouette values), and its neighbouring property guarantees the optimal arrangement of cells. For the production data described in Section 1, hexagonal 2‐dimensional SOM topology is appropriate because it corresponds to a 2‐dimensional architectural layout. The decision to apply 2-dimensional topology instead of 1-diemensional one is based on the assumption that the SOM clusters should correspond to architectural arrangement of the production processes. Therefore 2-dimensional topology is better suited to floor planning. The proposed approach applies a 2-dimensional grid with Nc = 6 cells (this was a recommendation from the company), arranged in 2 rows of 3 elements each, as shown in Fig. 1.

1.1 Data Pre-Processing The product descriptions in terms of properties and required operations include various types of variables (Boolean, numerical and categorical), so the following data encoding approach was applied in order to represent the data in a numerical format suitable for automated clustering solutions: • Boolean variables, such as the presence of a specific operation, were encoded as –1 (false) or 1 (true). • Categorical variables (materials, form, etc.) were encoded by 1-of-C coding that introduces additional dummy variables for each category. • Numeric variables (weight, volume, etc.) were rescaled into a [–1, 1] interval. The applied data pre-processing results in a description of each product with 58 attributes (37 operations and 21 product properties). The prepared data form a basis for self-organized cell formation and subsequent fine layout planning. 2 SELF-ORGANIZED FORMATION OF PRODUCTION CELLS An important objective of the cellular organization of production is the minimization of work and material flow, and consequently the minimization of production costs. The formation of production cells can be accomplished by clustering the products according to their required operations and properties into organizational units (cells) that share similar resources. Research into clustering approaches for the organization of production cells [18] has demonstrated 790

Fig. 1. Hexagonal SOM topology representing a cellular layout with 6 cells

The self-organized clustering procedure is performed by constructing the SOM clusters using both product properties and required operations. The data (properties and operations) are encoded as described in section 1.1 which results in a set of 58 attributes describing each product. In the next step, the SOM algorithm is applied to construct a mapping from this 58-dimensional space into a 2-dimensional grid representing the initial layout. The result of this stage is the initial cellular layout (Fig. 1) with products distributed in cells according to the SOM clustering algorithm. The products in each cell share similar properties and also required operations, so the next step is to arrange the required machines into each cell. The SOM algorithm only defines which product is assigned to which cell therefore the additional interpretation of SOM results is required to define machines for each cell. The interpretation of SOM clustering results can be accomplished by examining which machines are required by products assigned to each cell Ci; i = 1, 2, ..., 6. For each operation in a particular cell, we can provide a percentage

Potočnik, P. – Berlec, T. – Sluga, A. – Govekar, E.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 789-796

of products in this cell, which should require this operation in order for this operation to be included into the production cell. Various percentile margins can be defined, such as: p = {50th, 75th, 100th}. For each percentile margin, the particular operation should be included into the production cell if at least (100 – p)% of products in a particular SOM cluster require this operation. The percentile p can be considered as an open layout parameter that regulates the machine population density, i.e. the ratio of machines available in each cell and the machines that have to be borrowed in neighbouring cells. In this study p = 75th percentile was applied as a suitable compromise, and the result is shown in Fig. 2. For each cell Ci; i = 1, 2, ..., 6, the assigned operations are marked in black (the columns from 1 to 37) and the remaining columns (38 to 58) represent the product properties. Product properties show that the products are clustered into cells also according to the similar properties, not only according to the required machine operations. Therefore the properties (columns 38 to 58) are involved in the self-organized cell formation but they do not affect the distribution of machines as described above. The subsequent steps in arranging the SOMbased layout may include the economic optimization of each cell, and the assignment of rare operations. Economic optimization gives priority to products with high economic impact (value × quantity). Consequently, the initial placement shown in Fig. 2 can be adjusted in order to support high impact products with more required machines. The economic

optimization optimizes unified percentile parameter p into pi; i = 1, 2, ..., 6, assigning different machine population densities for each cell. As first, relative economic importance of each cell is estimated by summing the economic impacts of products (value × quantity) in each cell. Then, the percentile parameter pi is increased in cells with high economic importance, and decreased in cells with low economic importance. The implementation of this rule for p = 75th can be accomplished by increasing the most important cell’s percentile to pi = 100th, and decreasing the least important cell’s percentile to pi = 50th (and in other cells linearly accordingly within this range). The result of economic optimization is shown in Fig. 3 as slightly reordered placement of machines (black operations). Rare operations that are required to complete some products are initially not placed in the layout, so these machines are finally included in those cells with the highest requirement for such operations. The final version of the interpreted self-organized cell formation is shown in Fig. 3 with several rare operations (blue markers) included in the appropriate cells. It can be observed that several cells require the same type of machine due to frequently applied operations. When the self-organized cell formation shown in Fig. 3 is reordered into the hexagonal floor layout shown in Fig. 1, the final version of the self-organized layout can be obtained. Fig. 4 shows such a layout with 2-dimensional cell formation, and the recommended machines in each cell. This is the result of the first stage of the proposed facility layout planning method,

Operations (1 to 37) Properties (38 to 58) Fig. 2. Initial interpretation of SOM clusters assigning required machines to each production cell

Operations (1 to 37) Properties (38 to 58) Fig. 3. Finalized interpretation of SOM clusters with an economic interpretation of the products Hybrid Self-Organization Based Facility Layout Planning

791

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 789-796

i.e. automatic self-organized formation of production cells. In the next section this layout is further developed with respect to the positioning of machines within each cell.

The indices in Eqs. (1) and (2) denote: n is the total number of paths, lj the transport length of the jth path, ij the intensity on the jth path. The total transport length can be obtained (measured) from the layout considering the sequence of operations, and then the intensity-length product can be expressed by applying the intensity values for each path. The ILP measure is a weighted version of TTL because it assigns different weights to each path depending on the amount of transported items or units. When the new layout is obtained, the improvement can be measured by a relative decrease in both efficiency measures, TTL and ILP. 3.2 Material Flow Diagram and From-To Matrix

Fig. 4. Layout solution based on self-organized clustering and the economic evaluation of products

3 FINE LAYOUT PLANNING Fine layout planning section describes the second stage of the proposed hybrid facility layout planning method. This stage consists of expert operator based fine layout planning that finalizes the layout details. The method proposed in this section summarizes various known approaches for layout planning but is specifically unique and tailored to complete the proposed self-organized cell formation. Because the cellular layout is already defined by self-organization, fine layout planning only considers the positioning of machines within the cells. The specificity of the proposed approach is to simultaneously optimize internal and external material flow because the selforganized cells are small but also share substantial resources with neighbouring cells in hexagonal SOM topology. An initially large-scale layout problem is decomposed into several small problems that can be efficiently solved by expert operators as described below. 3.1 Layout Efficiency Measures In order to evaluate the efficiency of a layout, we propose two measures, namely total transport length (TTL) [m], and the product of intensity and length (ILP) [part×m]:

TTL = ∑ j =1 l j [m],

ILP = ∑ j =1 i j ⋅ l j [part×m].

792

(

)

Fine layout optimization should consider both the technology and the sequence of operations. An important tool in fine layout optimization is the material flow diagram (MFD), which can be constructed based on information about quantity, technology and the sequence of operations. The MFD shows a path of the product in a production line with the quantity transported and produced on this path. MFD is usually represented in a graphical form such as that shown in Fig. 5 for the case study presented in this paper. Another tool in fine layout optimization is the so-called from-to matrix (FTM), which can be extracted from the MFD and presents the quantity data more clearly compared to the latter. In the FTM, all incoming pieces for each operation are collected from the MFD, which reveals how many parts pass from one operation to another. 3.3 Graphical Layout Representation Layout optimization procedure usually requires a graphical representation that corresponds to the architectural layout of the production hall, and also calculates the corresponding efficiency measures for each tested layout. In our study, Vistable software was applied, although any appropriate software [19] can be used, too, for this purpose. For the case study presented in this paper, the current layout which represents the existing fine layout plan in the company is shown in Fig. 6. The efficiency of the current layout is expressed by the following values:

(1)

TTLcurrent = 5933 [m],

(2)

ILPcurrent = 307×106 [part×m].

Potočnik, P. – Berlec, T. – Sluga, A. – Govekar, E.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 789-796

3.4 Layout Creation Procedure Beside the described tools (layout efficiency measures, MFD, FTM, and graphical layout representation), the following additional information is required to proceed to fine layout creation: • the current layout, • restrictions on placement in the production hall (building restrictions, workplace restrictions, installations, technological specifications, standards and laws, etc.), • the type of transportation, • the standard transport routes. The layout optimization should consider the complete workplaces and not only isolated machines, as each machine requires additional manipulating surfaces and may impose various placement restrictions with respect to size, weight, installations, technological specifications, etc. Because of the

huge number of specific restrictions usually present in production environments, a mathematical model describing all the requirements and restrictions would be very difficult to construct. Consequently, based on the tools described above, we propose an expert operator based fine layout planning approach for the finalization of the layout details. The method is applied to each cell by following the steps described below. Each step, including the optimization steps (3 and 4) can be efficiently solved by expert operator as the initial layout problem has already been decomposed into several simplified small problems. 1. Placement of cells in the production hall. The virtual space of constructed self-organized cells should be positioned within the available production space. The layout shown in Fig. 4. can be adjusted in order to fit within the production environment, and the cell orientation should

Fig. 5. Material flow diagram Hybrid Self-Organization Based Facility Layout Planning

793

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 789-796

Fig. 6. Current (existing) layout

Fig. 7. The resulting layout

794

Potočnik, P. – Berlec, T. – Sluga, A. – Govekar, E.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 789-796

5. 6.

minimize paths from/to the input warehouse, the packaging department, and the output warehouse. Organization of workplaces within the cell. Experience showed that the optimal placement is a U-shaped cell, so an initial intra-cell layout follows a U-shaped format, with the main entrance to the cell positioned at the top of the U-shape. Optimizing external material flow. From the workplaces assigned to a cell, find the two with the highest external material flow (to other cells, the warehouse, etc.). Both workplaces are positioned on opposite sides of the cell entrance, thus minimizing the external material flow to and from this cell. Optimizing internal material flow. From the remaining workplaces, select the one with the highest internal material flow with respect to already positioned workplaces. Position the selected workplace according to the U-shape close to the already positioned workplaces so that the internal material flow is minimized. Repeat step 4 until all the workplaces have been placed in the cell. Finalize the internal layout. The expert operator should consider all the additional information (restrictions, installations, technological specifications, types of transportation, standard transport routes, etc.) and finalize the fine layout accordingly, and with respect to minimal TTL and ILP layout efficiency measures. 4 RESULTS AND DISCUSSION

The proposed layout creation procedure was applied to the production data considered in this study, and the resulting layout with cells marked in red is presented in Fig. 7. Evaluation of the resulting layout yielded the following efficiency:

TTLresulting = 3540 [m],

ILPresulting = 178×106 [part×m].

In relative terms, the TTL and ILP measures are decreased (improved) by 40 and 42%. In the presented case study, the hexagonal SOM topology with Nc = 6 cells was chosen based on company demands, although Nc can be considered as an open design parameter that may result in different layout solutions. This means that the proposed method can be adjusted to a wide range of layout planning problems with considerable flexibility. Nc can be selected either according to the architectural conditions and/or according to the rule

that the number of workplaces within each cell should be small enough to simplify the expert operator based fine layout planning. Besides the proposed hexagonal SOM topology, other topologies can be applied if they can better represent the architectural environment. 5 CONCLUSIONS The paper presents a novel hybrid two-stage method for facility layout planning based on the automatic selforganized formation of production cells, and expert operator based fine layout planning. The proposed method is suitable for small and medium enterprises (SMEs) that are characterized by individual and small batch production with many different products in their production range. The method combines the advantages of both the algorithmic and the manual expert-based approaches to layout planning, and is particularly suitable for the rearrangement of existing layouts due to numerous environmental constraints. The method effectively minimizes the work and material flow, and consequently reduces production costs. The proposed approach was tested on a case study based on data provided by a real manufacturing company. The fine layout planning results show a considerable improvement in both efficiency measures, with a 40% reduction in TTL and a 42% reduction in ILP. 6 REFERENCES [1] Hines, P., Tayor, D. (2000). Going Lean. Lean Enterprise Research Centre, Cardiff Business School, Cardiff. [2] Shishir Bhat, B.N. (2008). Cellular Manufacturing – The Heart of Lean Manufacturing. Advances in Production Engineering & Management, vol. 3, no. 4, p. 171-180. [3] Adenso-Díaz, B., Lozano, S., Racero J., Guerrero, F. (2001). Machine cell formation in generalized group technology. Computers & Industrial Engineering, vol. 41, no. 2, p. 227-240, DOI:10.1016/S03608352(01)00056-0. [4] Starbek, M., Menart, D. (2000). The optimization of material flow in production. International Journal of Machine Tools & Manufacture, vol. 40, no. 9, p. 12991310, DOI:10.1016/S0890-6955(99)00126-1. [5] Fung, R.Y.K., Liang, F., Jiang, Z., Wong, T.N. (2008). A multi-stage methodology for virtual cell formation oriented agile manufacturing. International Journal of Advanced Manufacturing Technologies, vol. 36, no. 7-8, p. 798-810, DOI:10.1007/s00170-006-0871-1. [6] Fan, Z.P., Chen, Y., Ma, J., Zhu, Y. (2009). Decision support for proposal grouping: a hybrid approach using knowledge rules and genetic algorithms. Expert

Hybrid Self-Organization Based Facility Layout Planning

795

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 789-796

Systems with Applications, vol. 36, no. 2, p. 1004-1013, DOI:10.1016/j.eswa.2007.11.011. [7] Kiang, M.Y., Kulkarni, U.R., Tam, K.Y. (1995). Selforganizing map network as an interactive clustering tool – An application to group technology. Decision Support Systems, vol. 15, no. 4, p. 351-374, DOI:10.1016/01679236(94)00046-1. [8] Guerrero, F., Lozano, S., Smith, K.A., Canca, D., Kwok, T. (2002). Manufacturing cell formation using a new self-organizing neural network. Computers & Industrial Engineering, vol. 42, no. 2-4, p. 377-382, DOI:10.1016/S0360-8352(02)00039-6. [9] Ueda, K., Fujii, N., Hatono, I., Kobayashi, M. (2002). Facility layout planning using self-organization method. CIRP Annals - Manufacturing Technology, vol. 51, no. 1, p. 399-402, DOI:10.1016/S0007-8506(07)61546-7. [10] Sluga, A., Butala, P., Peklenik, J. (2001). Selforganization in a distributed manufacturing system based on constraint logic programming. CIRP Annals - Manufacturing Technology, vol. 50, no. 1, p. 323-326, DOI:10.1016/S0007-8506(07)62131-3. [11] Silva, H.B., Brito, P., da Costa, J.P. (2006). A partitional clustering algorithm validated by a clustering tendency index based on graph theory. Pattern Recognition, vol. 39, no. 5, p. 776-788, DOI:10.1016/j. patcog.2005.10.027. [12] Hachicha, W., Masmoudi, F., Haddar, M. (2008). Formation of machine groups and part families in cellular manufacturing systems using a correlation analysis approach. The Industrial Journal of Advanced Manufacturing Technology, vol. 36, no. 11-12, p. 11571169, DOI:10.1007/s00170-007-0928-9. [13] Chang, M., Ohkura, K., Ueda, K., Sugiyama, M. (2002). A symbiotic evolutionary algorithm for dynamic facility layout problem. Proceedings of the 2002 Congress on Evolutionary Computation, IEEE Computer Society Washington, DC, p. 1745-1750.

796

[14] Brintrup, A.M., Takagi, H., Ramsden, J. (2006). Evaluation of sequential, multi-objective, and parallel interactive genetic algorithms for multi-objective floor plan optimisation. Lecture Notes in Computer Science, vol. 3907, p. 586-598, DOI:10.1007/11732242_56. [15] Ficko, M., Palčič, I. (2013). Designing a layout using the modified triangle method, and genetic algorithms. International Journal of Simulation Modelling, vol. 12, no. 4, p. 237-251, DOI:10.2507/IJSIMM12(4)3.244. [16] Kia, R., Khaksar-Haghanib, F., Javadianc, N., Tavakkoli-Moghaddamd, R. (2014). Solving a multifloor layout design model of a dynamic cellular manufacturing system by an efficient genetic algorithm. Journal of Manufacturing Systems, vol. 33, no. 1, p. 218-232, DOI:10.1016/j.jmsy.2013.12.005. [17] Mladineo, M., Veža, I., Čorkalo, A. (2011). Optimization of the selection of competence cells in regional production network. Tehnički vjesnik – Technical Gazette, vol. 18, no. 4, p. 581-588. [18] Potočnik, P., Berlec, T., Starbek, M., Govekar, E. (2013). Self-organizing neural network-based clustering and organization of production cells. Neural Computing and Applications, vol. 22, no. 1, p. 113-124, DOI:10.1007/s00521-012-0938-x. [19] Shariatzadeh, N., Sivard, G., Chen, D. (2012). Software evaluation criteria for rapid factory layout planning, design and simulation, Procedia CIRP, vol. 3, p. 299304, DOI:10.1016/j.procir.2012.07.052. [20] Jiang, S., Nee, A.Y.C. (2013). A novel facility layout planning and optimization methodology. CIRP Annals – Manufacturing Technology, vol. 62, no. 1, p. 483486, DOI:10.1016/j.cirp.2013.03.133. [21] Kohonen, T. (1997). Self-Organizing Maps, 2nd ed. Springer-Verlag, Berlin, DOI:10.1007/978-3-64297966-8.

Potočnik, P. – Berlec, T. – Sluga, A. – Govekar, E.

Received for review: 2014-04-11 Received revised form: 2014-09-25 Accepted for publication: 2014-10-07

An Instrument for Measuring the Degree of Lean Implementation in Manufacturing Vujica Herzog, N. – Tonchia, S. Natasa Vujica Herzog1 – Stefano Tonchia2

2Universitiy

1University of Maribor, Faculty of Mechanical Engineering, Slovenia of Udine, Department of Electrical, Managerial and Mechanical Engineering, Italy

Despite lean thinking being a broadly accepted approach, there is still some confusion within present terminology regarding ‘lean’ and its issues. This paper presents the results of a research survey carried out within 72 medium and large-sized Slovenian manufacturing companies. The sample selection criteria adopted, together with the research items investigated, ensured a generic framework for our research. Eight crucial areas were identified based on a synthesis of ‘lean’ literature for assessing and measuring the degree of lean implementation within existing manufacturing systems: value concept and customers, value stream mapping (VSM), pull/kanban and flow, waste elimination, productive maintenance, just-in-time (JIT), employee involvement and the development of excellent suppliers (lean suppliers). Variables were constructed within these areas using Likert scales, and statistical validity and reliability analyses. For example, when measuring the developments of excellent suppliers the focus should be on three variables: on time deliveries, supplier relationships, and a skilled and loyal supplier. The results show that the developed variables can be important both for understanding ‘lean’ and measuring the degree of lean implementation within existing manufacturing systems. Keywords: lean manufacturing, lean implementation, survey research, reliability, validity, new variable design

0 INTRODUCTION According to available literature it could be said that lean concepts are on the agenda again [1] to [3], particularly because of high quality requirements, generally known as six sigma as launched in the Motorola company and later expanded in General Electric and world-wide that urge companies to focus their knowledge and activities on higher operational excellence. Lean thinking [4] is broadly accepted as an approach linked to superior performance (excellence), and for its ability to provide competitive advantage. Despite its broad acceptance there is still some confusion within present terminology and a lack of common conceptual definitions regarding lean and its issues. In 1996, the Slovenian economist Ursic [5] determined that ‘Slovenian companies poorly understand and master those procedures, approaches, tools and methods that could enable greater competitiveness. In the following year the same author, together with his colleagues, published further research [6] exposing prevalent management methods within Slovenian companies’ strategic management, benchmarking, such as the TQM, ISO 9000 and 20 Keys methods. What about ‘lean’ or ‘six-sigma’ or even ‘lean six sigma’ and ‘design for six- sigma’? Knowledge about lean tools and techniques is available in university textbooks or can be acquired at several conferences and external education institutions that offer their services daily. But do companies practise

these methods? As there is no evidence about the presence of lean concepts within Slovenian companies the above questions were the basic motivators for the presented research. The first reflections from the performed research were positive and showed the presence of lean concepts within Slovenian companies. Experiences with six-sigma were however rarer and mostly limited to companies with foreign ownership. Therefore further analyses were made for only those research aspects dealing with ‘lean’. Based on a survey’s research results, an attempt was made to address the confusion within present terminology that leads to certain difficulties when measuring the level of lean implementation. In reviewing the literature, the following major issues important for ‘lean’ could be identified: 1. The definition of ‘lean’. An attempt was made to review the more important aspects of ‘lean’ through existing definitions. 2. Tools and techniques. A short overview is presented of the essential tools and techniques for ‘lean’. 3. Pull/kanban and flow. Managing flow is at the heart of ‘lean’, based on a pull-system and operationalised using kanban. 4. Waste elimination. The elimination of waste is central to lean approaches. 5. Employee involvement. Motivation, education and above all responsiveness are discussed as the conditions for being ‘lean’.

*Corr. Author’s Address: 1University of Maribor, Faculty of Mechanical Engineering, Slovenia, Smetanoa 17, 2000 Maribor, Slovenia, natasa.vujica@um.si

797

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 797-803

6. Lean suppliers and lean design. The importance of external issues is examined. This paper aims at clarifying these lean-areas and issues using a wider-range of items in respect to previous studies. This instrument can be useful, simple, and precise for accessing and measuring the degrees of lean implementation within existing productive systems or in connection with new lean programmes. 1 LITERATURE REVIEW Lean manufacturing goes back as far as 1978 when Ohno (1978) wrote his book on the Toyota Production System (TPS) in Japanese. It could be said that the ‘lean’ principles resulted from the broader community outside Japan, as a respond to the mass-production system that was practised in most American and European companies after the Second World War. The first ideas of TPS were focused mainly on waste elimination through the simplification of manufacturing processes [7] to [8]. The basic idea of TPS is to produce the kind of units needed, at the time needed and in the quantities needed. These goals can be achieved through different concepts such as JIT, automation, flexible workforce, work standardisation, links between suppliers and customers, and many others. 1.1 The Definition of ‘Lean’ Lean manufacturing was, for a long time, equated with JIT and thus it is difficult to make a clear distinction between lean and JIT. Similarly to its origin JIT lean aims to meet demand instantaneously, with perfect quality and no waste. Several authors have provided different interpretations of lean. Starting with Womack [9] lean manufacturing is defined as an integrated set of socio-technical practices aimed at eliminating waste along the whole value chain within and across companies. On the other hand, lean can also be seen from the practical perspective as a set of management practices, tools, or techniques for effective lean management [10] and [11]. 1.2 Tools and Techniques In general, lean manufacturing is described from two points of view, either from a philosophical perspective related to guiding principles and goals [4] and [10], or from the practical perspective as a set of management practices, tools, or techniques that can be observed directly [12]. 798

1.3 Pull/Kanban and Flow In the pull system, typical for lean manufacturing, the job is pulled to successive workstations instead of being pushed by its preceding workstation. In other words, in a pull system the material is only moved when the next stage requires it. The flow of parts throughout the production line can be controlled by kanban cards. The primary advantage of the pull system is the reduced inventory and therefore the associated cost of inventory reduction [13]. 1.4 Waste Elimination Identifying waste is the first step towards eliminating it. It could be said that waste is anything that does not directly add value to the final product or contribute to the product’s transformation [14]. Toyota identified seven types of waste, which have been found to apply in many different types of operations – both service and production – and which form the core of lean philosophy: over-production, waiting time, transport, processing, inventory, motion, and defectives. 1.5 Employee Involvement Various studies have concluded that without the total commitment of senior management, a company-wide project or change of programme could never succeed. Top management commitment with the active cooperation of all employees can be expounded as the more important success factor. As any change in operations usually presents a certain level of stress for employees, training about the roles of cooperation and preparedness for changes is the next essential element for success. 1.6 Lean suppliers and lean design The main focus of lean enterprise is to reduce waste and simultaneously increase value to the customer. Nowadays the customer is a ‘king’ as he will buy only the products that satisfy his needs and wants [15]. As the production of a high percentage of valueadded components during most manufacturing – and non- manufacturing organisations are outsourced, it is insufficient to be just the most efficient firm without having the most efficient network. A collaborative relationship between organisations and suppliers should be established in order to reach this goal. The same concept should be used to establish lean design [16] to [18]. Two-way communication and cooperation from product design

Vujica Herzog, N. – Tonchia, S.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 797-803

to delivery to the end-user is essential for effective lean manufacturing. 2 PREVIOUSLY PERFORMED SURVEYS In order to obtain a transparent set of issues (dimensions or variables) that must be practised for achieving effective lean manufacturing within a business unit, the results of previous studies were examined that dealt with the systematisation of ‘lean’. Due to space limitation only those selected representative writings from the last two decades are briefly summarised below. Karlsson and Ahlstrom [19] tried to assess those changes towards lean production. Following the ultimate goal of implementing lean production within an operational (increase productivity, enhance quality, shorten lead times, reduce cost etc.), they developed a model for operationalising the determinants of a lean production system (actions taken, the principles implemented, and the changes made to achieve the desired performance). The principles of ‘lean’ were presented through nine determinants as follows: elimination of waste, continuous improvement, zero defects, JIT, pull, multifunctional teams, decentralised responsibilities, integrated functions, and vertical information systems. Through a multiple case-study approach Panizzolo [20] explored how the lean production model was adopted by 27 excellent firms operating throughout international markets. He defined six characteristic areas of a ‘lean’ company: process and equipment, manufacturing planning and control, human resources, product design, supplier relationships, and customer relationships. Sanchez and Perez [21] developed and tested an integrated check-list for assessing manufacturing changes towards lean production. Within the lean production model they combined six groups of indicators from common basic lean production practices: elimination of zero-value activities, continuous improvement, multifunctional teams, JIT production and delivery, supplier integration, and a flexible information system. Shah and Ward [12] tried to establish a distinction between the system and its components through a literature review from the earliest publications relating to the Toyota Production System to the more recent. They realised that many descriptions exist about lean production and its underlying components, with a few conceptual definitions. Following the results of an extended survey research was performed at 280 companies by developing ten distinct dimensions

of a lean system: supplier feedback, JIT delivery, developing suppliers, involved customers, pull, flow, low set-up, controlled processes, productive maintenance, and involved employees. Based on a synthesis of literature reviews and available resources, nine lean issues were designed for our investigations (Table 1), listed in the first column on the left. Nine lean issues were checked for reliability and validity – this will be discussed later in Chapter 4, ‘Results analyses and variable construction’. Whilst it is certainly true that other sets of critical factors could be developed, this set appears to capture most of the important aspects of effective ‘lean’. Table 1. Internal consistency analysis results for the critical factors of lean manufacturing Lean issues 1. The value concept + customers 2. VSM 3. Pull/kanban + flow 4. Waste elimination 5. Productive maintenance 6. Just-in-time 7. Employee involvement 8. Lean suppliers

Cronbach α 0.546 0.691 0.768 0.760 0.670 0.667 0.800 0.642

3 METHODOLOGY An exploratory survey research methodology was adopted for considering the presented problem [22] to [24]. The research was divided into the following phases ( Fig 1): • an analysis of existing literature was made to determine the major dimensions of lean manufacturing; • a questionnaire was designed, pre-tested on experts and pilot-firms (as suggested by Dillman, [25]). The questionnaire contained 59 items, designed according to the Likert scales, ranging from ‘strongly disagree’ to ‘strongly agree’; • the resulting data were examined through reliability and validity analyses, and then analysed using uni- and multi-variate statistical techniques. The unit of analysis was the individual company and specifically the lean projects within the individual company. 3.1 Data Collection and Measurement Analyses The research was carried out at 72 Slovenian companies within the mechanical, electro-mechanical,

An Instrument for Measuring the Degree of Lean Implementation in Manufacturing

799

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 797-803

and some other industries. The criterion for the choice of sample was the size of the company. It was limited to medium (from 50 to 249 employees and revenue from 8.8 to 35 million), and larger sized companies (from 250 upwards employees and revenue from 35 million EUR upwards).

When determining the measurement properties of the constructs used within the statistical analysis, reliability and construct validity were assessed [28], using Cronbach’s alpha and principal components analysis (PCA) respectively. 3.2 Reliability Reliability is a statistical measurement of how reproducible the survey instrument’s data are [29] to [31]. Reliability is commonly assessed in three forms: test-retest, alternate-form and internal consistency. Internal consistency reliability is the commonly used psychometric measure in assessing survey instruments and scales and it is an indicator of how well the different items measure the same issue. As variables were being developed for the first time Cronbach’s alpha for measuring internal consistency was used. According to Nunnally and Bernstein [30] the newly-developed measurements can be accepted with α ≥ 0.6, otherwise α ≥ 0.7 should be the threshold. With α ≥ 0.8 the measurement is very reliable. In our research all of the multi-item variables had a Cronbach’s alpha of at least 0.6, thus well exceeding the guidelines set for the development of new variables. 3.3 Validity

Fig. 1. New variable construction process

The response rate was very good for the postcontact methodology (18.6%), and showed the firms interest in the subject. Included in the firms that responded to the survey were some of the more successful Slovenian companies [26]. The subsequent statistical analysis was, therefore, carried out on the results from those 72 companies that returned the questionnaires correctly filled in. A five- point Likert scale [27] was used to indicate the degree or extent of each item, ranging from ‘strongly disagree’ to ‘strongly agree’. For each of the eight crucial areas statements were designed with positive or negative connotations (to keep the attention). For the Value concept + Customer 10 statements were designed, e.g. as follows: 1. The value of a product can be measured in terms of customer satisfaction, 2. Higher product quality caused higher satisfaction, 3. The ratio between quality and costs does not essentially influence customer satisfaction,… 800

Besides determining survey reliability we must also assess the validity of a measurement. It refers to the extent to which it measures what it was intended to measure [31] and [32]. Three different types of validity are typically measured: content validity, criterion related validity, and construct validity. Content validity is a subjective measurement of how appropriate the items are. Content validity was derived from several extended reviews of recent literature about lean manufacturing [12], [19] to [21]. In order to establish criterion validity, each item of the questionnaire was reviewed and also by three general managers from different manufacturing companies. Following the pre-tests of the items, 65 items remained appropriate for conducting research. Construct validity was checked through the use of PCA. PCA was carried out in order to uncover the underlying dimensions, eliminate problems of multicollinearity and reduce the number of variables to a limited number of orthogonal factors. Each multiitem variable was factor-analysed separately: for the items loaded on more than one factor, the items responsible for the other factors beyond the first were eliminated (or considered in another variable) and

Vujica Herzog, N. – Tonchia, S.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 797-803

Cronbach’s alpha was re-calculated. The presented variables are all in their final versions. The same procedure was then adopted to group several variables in order to obtain a more manageable set of variables. Rotation was applied to aid interpretation. 4 RESULTS ANALYSES AND VARIABLE CONSTRUCTION An internal consistency analysis was performed separately, using the PASW Statistics 18 program package for the items of each eight scales on Lean manufacturing (Table 1). This table shows that the reliability coefficients or Cronbach’s α ranged from 0.546 to 0.800. According to instructions [30] measurements for the most critical factors (except for ‘Value concept + Customers’) are very reliable. The initial reliability value of ‘Value concept + Customers’ stayed on 0.546 even though several combinations of the defined items’ eliminations were tried for improving reliability. As this was the only criteria that didn’t fulfill reliability conditions completely, it was decided to continue with PCA for all eight scales. For value concept + customers the three factor solution was generated with eigenvalues 1.7, 0.9 and 0.8 explaining 88.8% of variance (Table 2). Table 2. Value concept + customers Items on Value concept + customers High product quality Product quality for customer satisfaction Warranty Quality in perceiving customer needs and demands Eigenvalue Proportion of variance explained [%] Cumulative variance explained [%] Re-calculated Cronbach α Name given to the new variable

1st factor 0.814

2nd factor 0.194

3rd factor 0.209

0.899

-0.043

0.037

0.148

0.069

0.983

0.078

0.988

0.068

1.779 44.466 44.466

0.952 23.799 68.265

0.805 20.129 88.394

0.666

VAR 1

VAR 2

VAR 3

The first variable consists of ‘high product quality’ and ‘product quality for customer satisfaction’, with re-calculated Cronbach’s α 0.666. The first variable, VAR 1 was named ‘Customer satisfaction’ and the other two ‘Perceiving customers demands’ and ‘Warranty’. similar procedure as used for new variable construction as ‘Value concept + customers was then used for all eight scales on Lean manufacturing. The final set of lean variables constructed is presented in

Table 3. As reported in Table 3, the 24 Lean variables proposed are grouped in 8 issues or areas. Table 3. Summary of lean variables constructed Lean issues

New variables

New Cronbach α 0.666

1. Customer satisfaction 2. Perceiving customers demands 3. Warranty 4. Process mapping Value stream 5. Waste evidence mapping 6. Cost reduction 7. Early information on customer needs 8. Customer involvement during Pull/kanban + product design 9. Flexible response on customers’ demands Flow 10. Planning and control 11. Parts standardization 12. Inventory management Waste 13. Capacity utilization and working elimination conditions 14. Total preventive maintenance Productive maintenance 15. First-pass quality 16. On-time deliveries 17. Cooperation with suppliers Just in time 18. Parts reduction 19. Order and cleanliness in the plant 20. Employee cooperation Employee involvement 21. Team working 22. On-time deliveries by the suppliers Development of excellent (lean) 23. Supplier relationships suppliers 24. A skilled and loyal supplier The value concept + Customers

0.763

0.730 0.713 0.676 0.633 0.670 0.627

0.833 0.775

The value concept is represented by three variables: customer satisfaction level, degrees of warranties, and capacity to perceive the customers’ demands. Another group of variables concerns Value Stream mapping (VSM) i.e. the visualisation of value during the firm’s processes; also here three variables have been developed by the presenters for covering this theme: the presence of process mapping, the evidence of waste, and the cost reduction. Five variables regarding the Womack’s principles of flow & pull: early information on customer needs (the starting point of each process, the client being external or internal), customer involvement since product design, flexibility in responding to customers, parts standardisation/modular products (which evidently allow flow and pull), an adequate planning and control system [33]. The waste elimination for perfection, which is a dogma of the lean, is realised through the variables of

An Instrument for Measuring the Degree of Lean Implementation in Manufacturing

801

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 797-803

inventory management (concept of “standard workin-progress” i.e. the only material needed by a pulled pipeline) and capacity utilisation. Productive maintenance being the sustainability of a model is realised by two other variables: Total Productive Maintenance -TPM and first-pass quality (the latter meaning that the system should be able to guarantee an acceptable production at the first shot). Just-in-time is another issue typically ascribed to lean and would result in measurements by four variables: the existence of on-time deliveries, cooperation with the suppliers, a reduced number of parts, and order cleanliness inside the plant. The other two variables (employee cooperation and teamwork) state that a crucial factor for lean success is the employee involvement i.e. lies in the human resources of a company. For development of excellent suppliers on-time deliveries are crucial along with good relationships and a skilled and loyal supplier. 5 DISCUSSION AND CONCLUSIONS Lean management has become the main managerial initiative for improving firms’ performances, leading to excellence within the context of sustainable competitive advantage. Due to the present markets’ crises and especially for companies with production systems in higher cost areas, Lean management is central and crucial. This paper has covered a wider range of items in respect of previous studies, collating many of them. Starting from 59 items 24 ‘lean’ variables were constructed and tested. A rigorous method for obtaining or confirming these variables was the other main contribution of the presented work. The substantial questionnaire utilised was based on literature reviews and experts’ interviews, and covered, in our opinion, all the most representative “lean” aspects at the moment expect lean design. Lean design was initially a part of the questionnaire but since we received too different answers we estimated that specific investigation would be of benefit in the future. Table 3 assumes that the variables proposed, for each of them had been constructed and justified according to the described statistical methodology, so this presentation of Lean is quite unique in its rigour in respect of many other contributions that do not adopt a statistical survey method such as this. 802

Although the survey was performed in Slovenian companies the results of the survey can be generally used. Besides the consideration that the presented work can be useful for studies aimed at a more “objective” approach to lean management, this wide and rigorous contribution has led to a concrete managerial instrument for usage. 6 REFERENCES [1] Eroglu, C., Hofer, C. (2011). Lean, leaner, too lean? The inventory-performance link revisited. Journal of Operations Management, vol. 29 no. 4, p. 354-369, DOI:10.1016/j.jom.2010.05.002. [2] Moyano-Fuentes, J., Sacristan-Diaz, M. (2012). Learning on lean: a review of thinking and research. International Journal of Operations and Production Management, vol. 32, no. 5, p. 551-582, DOI:10.1108/01443571211226498. [3] Alsmadi, M., Almani, A., Jerisat, R. (2012). A comparative analysis of Lean practices and performance in UK manufacturing and service sector firms. Total Quality Management & Business Excellence, vol. 23 no. 3-4, p. 381-396, DOI:10.1080/14783363.2012.669 993. [4] Womack, J.P., Jones, D.T. (1996). Lean Thinking: Banish Waste and Create Wealth in Your Corporation, Simon & Schuster, New York. [5] Ursic, D. (1996). Innovation of Enterprise, Studio Linea, Maribor. Monden, Y. (1983). Toyota Production System: A Practical Approach to Production Management. Industrial Engineers and Management Press, Norcross. [6] Ursic, D, Anteric, S., Mulej, M. (2005). Business process reengineering in practice – an example of a medium-sized Slovenian company in difficulties. Systemic Practice and Action Research, vol. 18 no. 1, p. 89-117, DOI:10.1007/s11213-005-2461-9. [7] Monden, Y. (1983). Toyota Production System: A Practical Approach to Production Management. Industrial Engineers and Management Press, Norcross. [8] Ohno, T. (1988). Toyota Production System: Beyond Large Scale Production. Productivity Press, Cambridge. [9] Womack, J.P., Jones, D.T., Roos, D. (1990). The Machine That Changed the World. Rawson Associates, Macmillan, New York. [10] Shah, R., Ward, P.T. (2003). Lean manufacturing: context, practice bundles, and performance. Journal of Operations Management, vol. 21, no. 2, p. 129-149, DOI:10.1016/S0272-6963(02)00108-0. [11] Buchmeister, B., Friscic, D., Palcic, I. (2013). Impact of demand changes and supply chain’s level constraints on bullwhip effect. Advances in Production Engineering & Management, vol. 8, no. 4, 199-208, DOI:10.14743/ apem2013.4.167.

Vujica Herzog, N. – Tonchia, S.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 797-803

[12] Shah, R., Ward, P.T. (2007). Defining and developing measures on lean production. Journal of Operations Management, vol. 25, no. 4, p. 785-805, DOI:10.1016/j. jom.2007.01.019. [13] Gracanin D., Lalic B., Beker I., Lalic D., Buchmeister B. (2013). Cost-time profile simulation for job shop scheduling decisions. International Journal of Simulation Modelling, vol. 12, no. 4, p. 213-224, DOI:10.2507/IJSIMM12(4)1.237. [14] Ayadi M., Costa Affonso R., Cheutet V., Masmoudi F., Riviere A., Haddar M. (2013). Conceptual model for management of digital factory simulation information. International Journal of Simulation Modelling, vol. 12, no. 2, p. 107-119, DOI:10.2507/IJSIMM12(2)4.233. [15] Kusar, J., Duhovnik, J., Tomazevic, R., Starbek, M. (2007). Finding and evaluating customers needs in the product-development process. Strojniški vestnik Journal of Mechanical Engineering, vol. 53, no. 2, p. 78-104. [16] Ward, A.C. (2007). Lean Product and Process Development. Lean Enterprise Institute, Cambridge. [17] Morgan, J.M., Liker, J.K. (2006). The Toyota Product Development System: Integrating People, Process, and Technology. Productivity Press, New York. [18] Rihar, L., Kušar, J., Gorenc, S., Starbek, M. (2012). Teamwork in the simultaneous product realisation. Strojniški vestnik - Journal of Mechanical Engineering, vol. 58, no. 9, p. 534-544, DOI:10.5545/svjme.2012.420. [19] Karlsson, C., Ahlstrom, P. (1996). Assessing changes towards lean production. International Journal of Operations and Production Management, vol. 16, no. 2, p. 24-41, DOI:10.1108/01443579610109820. [20] Panizzolo, R. (1998). Applying the lessons learned from 27 lean manufacturers. International Journal of Production Economics, vol. 55, no. 3, p. 223-240, DOI:10.1016/S0925-5273(98)00066-8. [21] Sanchez, A.M., Perez, M.P. (2001). Lean indicators and manufacturing strategies. International Journal of Operations and Productions Management, vol. 21 no. 11, p. 1433-1451, DOI:10.1108/01443570110407436. [22] Vujica-Herzog, N., Tonchia, S., Polajnar, A. (2009). Linkages between manufacturing strategy,

benchmarking, performance measurement and business process reengineering. Computers & Industrial Engineering, vol. 57, no. 3, p. 963-975, DOI:10.1016/j. cie.2009.03.015. [23] Vujica Herzog, N., Polajnar, A., Tonchia, S. (2007). Development and validation of business process reengineering (BPR) variables: a survey research in Slovenian companies. International Journal of Production Research, vol. 45, no. 24, p. 5811-5834, DOI:10.1080/00207540600854992. [24] Dalgobind, M., Anjani, K. (2009). The effect of lean manufacturing on product quality and industrial productivity: An empirical survey. Advances in Production Engineering & Management, vol. 4, no. 4, p. 221-232. [25] Dillman, D.A. (1978). Mail and Telephone Surveys: The Total Design Method, John Wiley & Sons, New York. [26] Palčič, I., Buchmeister, B., Polajnar, A. (2010). Analysis of innovation concepts in Slovenian manufacturing companies. Strojniški vestnik - Journal of Mechanical Engineering, vol. 56, no. 12, p. 803-810. [27] Rossi, P.H., Wright, J.D., Anderson, A.B. (1983). Handbook of Survey Research, Academic Press, New York. [28] Dick, W., Hagerty, N. (1971). Topics in Measurement: Reliability and Validity, McGraw-Hill, New York. [29] Cronbach, L.J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, vol. 16, no. 3, p. 297-334, DOI:10.1007/BF02310555. [30] Nunnally, J.C., Bernstein, I.H. (1994). Psychometrics Theory, 3rd ed., McGraw Hill, New York. [31] Litwin, M.S. (1995). How to Measure Survey Reliability and Validity. Sage Publication, Thousand Oaks, London, New Delhi. [32] Perme, T. (2011). Modelling and discrete simulation for the sustainable management of production and logistics issues. Transactions of FAMENA, vol. 35, no. 1, p. 83-90. [33] Tan, Y., Takakuwa, S. (2011). Use of simulation in a factory for business continuity planning. International Journal of Simulation Modelling, vol. 10, no. 1, p. 1727, DOI:10.2507/IJSIMM10(1)2.172.

An Instrument for Measuring the Degree of Lean Implementation in Manufacturing

803

Received for review: 2014-05-20 Received revised form: 2014-08-06 Accepted for publication: 2014-09-19

Trajectory Planning of an End-Effector for Path with Loop Boryga, M. Marek Boryga

University of Life Sciences in Lublin, Faculty of Production Engineering, Poland This paper presents an algorithm for rectilinear-arc trajectory planning whose path is composed of two rectilinear segments connected with a loop-shaped arc. The algorithm can be used in solutions with high speed cornering applications. During the realization of the given trajectory, the end-effector passed through the corner twice and the time difference is, simultaneously, the time of loop tracking. On the rectilinear segments, end-effector acceleration was described by a 7th degree polynomial, whereas the velocity while moving through a loop was a constant value. The results of the trajectory planning are presented as courses of displacements, speeds and accelerations of the end-effector and displacements, speeds and accelerations in kinematic pairs of the three manipulators. Keywords: trajectory planning, sharp corner, path with loop, polynomial acceleration profile, manipulator

0 INTRODUCTION Planning of smooth trajectories has been a very active area of research, hence a number of scientific reports address the problem. The publications concern both the generation of manipulator end-effector trajectories and designing the motion of tools of machine tools. With regard to manipulators, smooth trajectory planning is mostly imposed by the given tasks (assembly, transport of fragile objects, carrying open containers, gluing and painting) as well as through attempts to reduce the wear of manipulator components (decreased driving torque or limited vibration level caused by a resonance frequency). Regarding machine tools, smooth trajectory planning can facilitate a full utilization of the tools’ dynamic capabilities, with high-performance maintenance. Boryga and Grabos [1] presented a planning mode of trajectory for a seriallink manipulator with higher-degree polynomials application. The linear acceleration profiles of endeffectors were planned as polynomials of degree 9th, 7th and 5th. Chettibi [2] proposed a method to plan minimum cost movements for robotic manipulators along prescribed geometric paths while taking into account various kinodynamic constraints. Numerical examples using genetic algorithms are presented to illustrate the effectiveness of the proposed approach. Constantinescu and Croft [3] obtained the smooth trajectory of manipulators by considering torque rate and jerk limits in trajectory planning problem which was solved using flexible tolerance method. Elnagar and Hussein [4] studied acceleration-based optimal piecewise trajectory planning. The acceleration and curvature were used to generate the minimumenergy trajectory. Gasparetto and Zanotto [5] and [6] presented a method for smooth and optimal trajectory planning of robot manipulators. They worked out an objective function containing a term proportional 804

to the integral of the squared jerk and a second term proportional to the total execution time. The algorithm has been tested in simulation, yielding good results. Machmudah et al. [7] described a point-to-point of an arm robot motion planning in complex geometrical obstacle utilizing a 6th degree polynomial as the joint angle path. A planar robot will be utilized to simulate the proposed approach. Perumaal and Jawahar [8] proposed an approach to generate a synchronized jerkbounded trigonometric S-curve trajectory for a robotic manipulator. The results of simulations show that the proposed approach is able to generate a synchronized, smooth trajectory with minimum execution time. Tian and Collins [9] formulated a constraint by keeping the end-effector trajectory for a manipulator away from the obstacle. A genetic algorithm using a floating point representation is proposed to search for the optimal trajectory. In the work of Bu et al. [10], the complicated robotic task is decomposed into two kinds of subprocesses. In the free motion process, the kinematic models of the quasi trapezoidal and quasi triangular waveform are proposed with the dynamic limits of maximum velocities, accelerations and jerks of robotic joints. In the constrained motion process, the mathematical presentation of the task paths is extracted from the CAD models of the work pieces. Chen et al. [11] presented a smooth S-curve feed-rate profiling generation algorithm that produces continuous feed-rate, acceleration, and jerk profiles. The proposed algorithm ensures the automated machinery’s motion smoothness. Du et al. [12] presented an adaptive NURBS (Non-Uniform Rational B-Spline) interpolator taking into consideration an acceleration-deceleration control. A real-time flexible acceleration-deceleration control scheme was introduced to solve the sudden feed-rate change around corners with large curvature in their method. Emami and Arezoo [13] introduced look-ahead

*Corr. Author’s Address: University of Life Sciences in Lublin, Faculty of Production Engineering, ul. Głęboka 28, 20-612 Lublin, Poland, marek.boryga@up.lublin.pl

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 804-814

trajectory generation to determine the deceleration stage according to the fast estimated arc length and the reverse interpolation of each curve at every sampling time. Erkorkmaz and Altintas [14] presented a quintic spline trajectory generation algorithm that produces continuous position, velocity, and acceleration profiles. The proposed trajectory generation algorithm has been tested in machining on a three-axis milling machine. Farouki et al. [15] proposed an approach that takes into account limiting constraints such as the maximal of motor torque and power available to each axis of the machine tool in the calculation of the tool path feed-rate. Nam and Yang [16] developed a recursive trajectory generation method to estimate and determine the final deceleration stage according to the distance left to travel, resulting in exact feed-rate trajectory generation through jerk-limited acceleration profiles for the parametric curves. Stori and Wright [17] proposed an algorithm for convex contours, where an offset tool path is modified so that the engagement is kept constant. Zheng et al. [18] proposed two modified nonlinear tracking differentiators to generate the smooth trajectory for industrial mechatronic system from a rough reference, with bounded velocity and acceleration. Kovacic and Balic [19] proposed an autonomous, intelligent programming system for the cutting device controller based on an evolutionary genetics algorithm. Some research reports discuss the planning of motion, in which a tool moves over the path with a so-called sharp corner. Lloyd and Hayward [20] proposed the use of a 5th degree polynomial to connect the successive motions. To adjust the spatial shape of the transition curve, the authors defined two parameters that can be set for the various values in the interval (0,1). As an example, they presented a simple algorithm for the trajectory generation. Macfarlane and Croft [21] described a method, which uses a concatenation of quintic polynomials to provide a smooth trajectory between two points. The authors presented the experimental results and simulations on an industrial robot. Erkorkmaz et al. [22] put forward a path planning strategy for maintaining a high positioning accuracy in high speed cornering applications. The authors developed two spline-fitting strategies for smoothing sharp corners (the under-corner and over-corner approach). The obtained cornering accuracy was verified in the experiments. To eliminate jerk constraints and remove discontinuities in the acceleration profile, Dong et al. [23] added an acceleration-continuation procedure to the feed-rate optimization algorithm. In order to verify the effectiveness of this approach, the authors presented some application examples and the

research results obtained. Tsai et al. [24] proposed an interpolation algorithm ILD (Integrated Look-ahead Dynamics-based) that considers geometric and servo errors simultaneously. The study results indicate that the proposed approach significantly improves tracking and contour accuracy. Tseng et al. [25], aiming at jerk limitation, suggested a parametric interpolator composed of a look-ahead stage and a real-time sampling stage. The algorithm ensures that chord error as well as maximum acceleration and jerk are within the allowable limits. In the work of Wang et al. [26], aiming to minimize contour errors and feed rate fluctuations, a NURBS (Non-Uniform Rational B-Spline) interpolation algorithm was developed that adjusts adaptively to the machine dynamics and kinematics. The simulation test results confirm the effectiveness of the proposed interpolator for machining curved paths. Grabos and Boryga [27] presented an algorithm PCM (Polynomial Cross Method) for planning motion of a manipulator endeffector, whose path was composed of two rectilinear segments. The paper proposes the algorithm of the method and research results in the form of runs of velocities, acceleration and jerk for the prescribed motion trajectory. The objective of the present study was to develop a trajectory-planning algorithm that could automatically reconcile two essential requirements, i.e. trajectory smoothing and end-effector moving along the sharp corner. The most commonly used method consists in smoothing the corner. This method prevents a transition of the end-effector through the sharp corner. One solution, which simultaneously provides trajectory smoothing and transition through the corner, is using a loop. The path of the end-effector comprises two rectilinear segments, connected with a loop. The end-effector passes through the corner twice and the time difference is, simultaneously, the time of loop tracking. The work is organized as follows: Section 1 depicting a trajectory planning technique with the 7th degree polynomial application; Section 2 presents the algorithm for planning rectilinear-arc trajectory with a loop; Section 3 depicts the numerical example, while the simulation results are given in Section 4. The final section includes the concluding remarks. 1 TRAJECTORY PLANNING WITH 7th DEGREE POLYNOMIAL USE The planning of the robot end-effector trajectory can be accomplished by using higher-degree polynomials that facilitate the acceleration profile development. In a

Trajectory Planning of an End-Effector for Path with Loop

805

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 804-814

study by Boryga and Grabos [1] to build a polynomial form, the properties of the root’s multiplicity were utilized. This approach to polynomial form structure necessitates the determination of only one polynomial coefficient, irrespective of its order. Their work also shows the 5th, 7th and 9th degree polynomials describing the acceleration profile; the lowest values of linear and angular jerks were reported for the 7th degree polynomial. Therefore, in this paper, the acceleration profile on rectilinear segments is described with the 7th degree polynomial in the form of:

where p is coefficient of polynomial 7th degree, and tf time of motion end. The polynomials describing the profiles of velocity and displacement are as following: 1 1 19 5 v(t ) = − p ( t 8 − t f t 7 + t 2f t 6 − t 3f t 5 + 8 2 24 8 1 1 (2) + t 4f t 4 − t 5f t 3 ), 4 24

v(t ) = vmax . (5)

In order to determine the time of movement tf and polynomial coefficient p for a given path Δs it was set to the maximum polynomial value, which defines the profile of the displacement. Then, a system of equations was created,

a (t ) = − p ⋅ (t ) 2 (t − 0.5t f )3 (t − t f ) 2 , (1)

∃

t∈< 0,t f >

 p ⋅ t 8f = vmax   6144 , (6)  9  p ⋅tf  10080 = ∆s

whose solution is the time of movement tf and polynomial coefficient p. 2 ALGORITHM FOR TRAJECTORY PLANNING 2.1 Algorithm Description Fig. 2 shows the path of the end-effector’s movement with an implemented, local coordinate system ξηζ.

1 9 1 8 19 2 7 5 3 6 t − tft + tft − tft + 72 16 168 48 1 4 5 1 5 4 + t f t − t f t ). (3) 20 96

s (t ) = − p (

Sample profiles of acceleration, velocity and displacement are shown in Fig. 1.

Fig. 2. Path of motion with the introduced local coordinate system

Fig. 1. The profiles of acceleration, velocity and displacement for the acceleration profile described by a 7th degree polynomial

It was assumed that the velocity profile should realize the following conditions:

806

∀ − vmax ≤ v(t ) ≤ vmax , (4)

t∈< 0,t f >

The acceleration profile on a straight BM section is described by a 7th degree polynomial being used only at the first part of the profile, which represents the start-up phase (a solid line in Fig. 1). In the initial point B and final M, the acceleration profile is tangential to the time axis, which eliminates any undesirable jerk effect in these points. Over the MT1 segment, the endeffector moves at a constant velocity, just like over the arc connecting the points T1 and T2 as well as along the T2M segment. The acceleration profile on the ME segment was described, like the BM segment, by a 7th degree polynomial. The second phase of the profile was used and it is the braking phase (broken lines in

Boryga, M.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 804-814

Fig. 1). The maximum linear velocity on the BM and ME segments is νmax. This velocity is also the endeffector velocity on the loop. 2.2 Calculations of Path Geometry Flowchart in Fig. 3 presents the calculations of path geometry step by step. Assume the following: the coordinates of the initial point B, the coordinates of the final point E, the coordinates of the point M, and distance from the point M to the center of a circle Δ. It is also possible to assume the radius R, calculate the angle BME and introduce into the algorithm the distance Δ calculated using sine function of half angle BME.

will be a = a / a and d = d / d , respectively. The equation of the straight line on which the angle bisector BME lies can be established using the coordinates of the point M and direction vector b, which is the sum of the unit vectors a and d . Normalize vector b and the obtained unit vector b multiplied by Δ value. The coordinates of the vector calculated in this way should be added to the coordinates of the point M to obtain the coordinates of the O1 loop arc center. The radius R of a circle is determined as the distance from the center of the circle O1 to the line going through the points B and M (or the line going through the points E and M). Calculate the distance between the tangency points and point M using the Pythagorean theorem. Multiply the calculated distance value by the unit vector a and add to the coordinates of the point M to get the coordinates of the point T1, then multiply by the unit vector d and add to the coordinates of the point M to finally obtain the coordinates of point T2. The angle β can be determined by adding to the angle π the angle between vectors a and d . Path increments on the rectilinear segments are calculated using the coordinates of the starts and ends of the appropriate segments. 2.3 Calculations for the First Rectilinear Segment For calculating the time of movement on the BM section it is necessary to use a solution of a system of equations and use substitutions tf = 2tBM and Δs = 2 ΔsBM. Finally:

t BM =

105 ∆s BM , (7) 64 vmax

where νmax is linear velocity in the motion through the loop, which is simultaneously the maximum velocity on the rectilinear segment BM. As for the motion along the MT1 segment, the following dependency should be utilized:

Fig. 3. Flowchart for calculation of path geometry Let a denote a vector of the start point B and end point M, while d is a vector starting at the point E and ending at the point M. Unit vectors of these vectors

t MT1 =

∆s MT1 . (8) vmax

The motion time on the BT1 segment is equal to the sum of tBM and t MT1 times. The polynomial coefficient for the BM section can be calculated from the dependency:

Trajectory Planning of an End-Effector for Path with Loop

p BM =

315 ⋅ ∆s BM . (9) 8 ⋅ (t BM )9 807

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 804-814

In order to determine the acceleration, velocity and dislocation profiles on the BM section, it is necessary to use dependencies (Eqs. (1) to (3)) and use substitutions tf = 2tBM and p = pBM. 2.4 Calculations for the arc

tA =

β ⋅R . (10) vmax BT1

BT1

The angular position for t ≤ t ≤ t + t is obtained from the following dependency considering the initial angular position βb:

β (t ) = β b +

Motion time over the BT1T2E trajectory is calculated, summing up the motion times on the rectilinear segments and the motion time over the arc

vmax ⋅ (t − t BT 1 ), (11) R

The linear path on the BT1T2E trajectory is calculated as the sum of the path on the rectilinear segments and the path over the arc. 3 NUMERICAL EXAMPLE The coordinates of the points for the prescribed endeffector trajectory B, M, E, the center of loop O1 and tangent points T1, T2 are presented in Table 1. Table 1. Coordinates of characteristic points of trajectory Point denotation

where β b = π − ( β / 2). Coordinates of the position vector SA in the motion on the arc in the basic coordinate system xyz are calculated using the transformation matrix T between the local coordinate system ξηζ and basic system xyz:

and

Ξ A = [ R ⋅ cos β (t ), R ⋅ sin β (t ), 0, 1] .

Velocities VA and accelerations AA are derived as the successive derivatives of the position vector SA. 2.4 Calculations for the Second Rectilinear Segment The motion time on the T2M segment is the same as on the MT1 segment. In order to obtain motion time over the ME segment, the dependency (Eq. (7)) should be used with the path increment ΔsME substitution. Motion time on the T2E segment is equal to the sum of t T2 M and tME times. For the calculation of a polynomial coefficient on a segment, use the Eq. (9) and substitute path increment ΔsME and time tME in the formula. A profile of acceleration, velocity, and position on the ME segment can be derived in an analogous manner as with the BM segment, and each of them needs translation in the time by τ:

808

B M E T1 T2 O1

S A = T ⋅ Ξ A , (12)

where S A =  s xA (t ), s yA (t ), s zA (t ), 1  

te = t BT1 + t A + t T2 E . (14)

Motion time over the T1T2 arc is calculated from the formula:

2.5 Final Calculations

τ = t BT1 + t A + t T2 M − t ME . (13)

Point coordinates [m]

0.5 0.54 0.54 0.554 0.54 0.549

0.5 0.53 0.5 0.541 0.548 0.548

1 1 1 1 1 1

The assumed distance between the center of loop O1 and point M was Δ = 0.02 m. A calculated radius arc circle was R = 8.944×10–3 m, while the angle β = 4.069 rad. The path increments on the rectilinear segments are as follows: ∆sBM = 0.05 m, ∆s MT1 = ∆sT2 M = 0.0179 m and ∆sME = 0.03 m. There was an assumed maximum velocity on the rectilinear segments νmax = 0.25 m/s. The motion times over the rectilinear segments BM and MT1 are tBM = 0.328 s and t MT1 = 0.072 s, respectively, whereas for the coefficient of the polynomial profile of acceleration, the velocity and position for the segment BM was BM p232 = 4.46156 × 104 m/s9. The motion time through the loop arc tA = 0.146 s and the initial angle value was βp = 1.107 rad. The motion times on the rectilinear segments T2M and ME were as follows: t T2 M =0.072 s and tME = 0.197 s, while the coefficient of the polynomial profile of acceleration, velocity and position for the segment ME was pME = 2.65844×106 m/s9. The translation in time was τ = 0.42 s, whereas the total motion time te = 0.815 s. The total path along the motion trajectory was Δs = 0.152 m, while on the loop ΔsA = 0.036 m. Fig. 4 presents the planned path of movement in a coordinate system xyz.

Boryga, M.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 804-814

acceleration is proportional to the square of the velocity and is inversely proportional to the radius of the arc. The changing direction of centripetal acceleration results in a change in components in the direction x and y. The period of movement on the MT1T2M loop is marked in addition.

Fig. 4. The assumed end-effector’s path in a coordinate system xyz

4 SIMULATION RESULTS Fig. 5a gives the course of end-effector positions along the axes x and y, whereas Fig. 5b is the course of end-effector positions along the prescribed trajectory. Fig. 6 presents the courses of velocity components of the end-effector toward the axes x and y, and the run of the resultant velocity. During the motion through the loop (on the rectilinear segments MT1 and T2M and arc T1T2), the resultant velocity does not change and is equal to the preset one νmax = 0.25 m/s, while on the segments BM and ME it changes in accordance with the assumed profile. The components of velocity along axes x and y are obtained through the projection of the resultant velocity vector onto the directions of the coordinate system xyz. The changes in the components on the segments BM and ME are associated with changes in the resultant velocity, while the change in velocity components on the loop are associated with changes in the direction of a constant vector of resultant velocity. Both the runs of the components and the resultant velocity run are continuous functions considering the value. Fig. 7 presents the course of acceleration components in the x, y direction, and the course of resultant acceleration. A profile of resultant acceleration and acceleration components at points B, M, E, are tangential to the time axis and their value is equal to zero. The acceleration value is also equal to zero on the segments MT1 and T2M. The jump in the acceleration value on the arc T1T2 results from the motion on curved path. Tangential acceleration in motion along an arc equals zero, while centripetal

Fig. 5. The courses of end-effector position a) on x and y coordinates, b) along the path

Fig. 6. The courses of linear velocity on x, y coordinates and resultant velocity

The suggested method of trajectory planning is used to simulate a motion of three manipulators. For each of the manipulators, the following are determined: • coordinates of the end-effector in the coordinate system xyz,

Trajectory Planning of an End-Effector for Path with Loop

809

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 804-814

•

configuration coordinates (inverse kinematic task), velocity and acceleration of links.

•

Fig. 8. The anthropomorphic manipulator Fig. 7. The courses of acceleration on x, y coordinates, and resultant acceleration

Fig. 8 presents a manipulator with an anthropomorphic structure. The coordinates of the end-effector’s location in the coordinate system xyz are written below:

sx = c1 (l2a c2 + l3a c23 ), (15)

s y = s1 (l2a c2 + l3a c23 ), (16)

sz = λ1a + l2a s2 + l3a s23 , (17)

where λ1a , l2a , l3a are the lengths of links and cij = cos(θia + θ ja ) , ci = cos(θia ) , si = sin(θia ) , a a sij = sin(θi + θ j ) . The assumed lengths of respective kinematic chain links were as following λ1a =0.33 m, l2a =0.42 m, l3a =0.36 m. As a result of a system of Eqs. (15) to (17) due to the configuration coordinates is:

θ1a = arctg(

θ 2a = arctg where

sy sx

), (18)

A C 2 − ( B + A2 − D ) 2 − arcsin , (19) E 2 ( B + A2 ) B

θ3a = arccos( A = s z − λ1a ,

B + A2 − D ), (20) C B = s x2 + s 2y ,

C = 2l2a l3a ,

D = (l2a ) 2 + (l3a ) 2 and E = 2l2a . Courses of links’ angular displacements for the generated trajectory are presented in Fig. 9. 810

Boryga, M.

Fig. 9. The courses of angular displacement for an anthropomorphic manipulator a) link 1, b) link 2, c) link 3

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 804-814

The course of velocities and accelerations of links is presented on Figs. 10 and 11.

where λ1s , λ2s , l1s , l2s are the lengths of the links, while ci = cos(θis ) , si = sin(θis ) , cij = cos(θis + θ js ) , sij = sin(θis + θ js ) . The assumed lengths of particular links of respective kinematic chain were as following λ1s =0.3 m, λ2s =0.065 m, l1s =0.54 m, l2s =0.42 m. As a consequence of the equation system (Eqs. (21) to (23)), and due to configuration coordinates, the below was obtained:

Fig. 10. The courses of angular velocity of links for an anthropomorphic manipulator

) − arcsin(

C 2 − ( B − D)2 ), (24) E2B

θ 2a = −arccos(

B−D ), (25) C

θ1a = arctg (

λ3s = λ1s + λ2s − sz , (26)

B = sx2 + s y2 ,

where s 2 1

C = 2l1s l2s ,

E = 2l1s

and

s 2 2

D = (l ) + (l ) . The courses of angular accelerations in the kinematic pairs of the SCARA manipulator for a planned trajectory are presented in Fig. 13 (coordinate λ3s was skipped because it has a constant value). The velocities and link angular accelerations are presented in Figs. 14 and 15. Fig. 11. The courses of angular acceleration of links for an anthropomorphic manipulator

Fig. 12 shows the SCARA manipulator.

Fig. 12. The SCARA manipulator

Coordinates of the end-effector in the coordinate xyz system are defined by the following dependencies:

sx = l1s c1 + l2s c12 , (21)

s y = l1s s1 + l2s s12 , (22)

sz = λ1s + λ2s − λ3s (23)

Fig. 13. The courses of angular displacement for the SCARA manipulator: a) link 1, b) link 2

Trajectory Planning of an End-Effector for Path with Loop

811

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 804-814

where l2c is the length of manipulator’s link and l2c =0.24 m. As a result of the equation system solution (Eqs. (27) to (29)), and due to configuration coordinates, the following is obtained:

Fig. 14. The courses of angular velocity for the SCARA manipulator

λ1c = sz − l2c , (30)

λ2c = s y , (31)

λ3c = − sx . (32)

The courses of displacement of links for the kinematic chain studied for a planned trajectory are presented in Fig. 17 ( λ1c coordinate was skipped because it has a constant value).

Fig. 15. The courses of angular acceleration for the SCARA manipulator

Fig. 16 shows a manipulator with a Cartesian structure.

Fig. 17. The courses of linear displacement for the Cartesian manipulator a) link 2, b) link 3 Fig. 16. The Cartesian manipulator

The coordinates of end-effector position in the xyz coordinate system are:

sx = −λ3c , (27)

s y = λ2c , (28)

sz = λ1c + l2c , (29)

812

Velocities and linear accelerations of the links are presented in Figs. 18 and 19. For all manipulators, the courses of movements and velocities in kinematic pairs are continuous. The courses of accelerations in kinematic pairs are discontinuous. Discontinuity of acceleration results from centripetal acceleration, which occurs during the motion of end-effector on a curved path.

Boryga, M.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 804-814

position, velocity and acceleration, it is sufficient to calculate a polynomial coefficient and motion time. 4. The simulation results indicate the applicability of the proposed method in the analysis stage as well as the design of manipulators and machine tools. Further studies will be needed to determine the effect of loop size on trajectory mapping accuracy, taking into account the deformability of manipulator kinematic chains. Fig. 18. The courses of linear velocity for the Cartesian manipulator

Fig. 19. The courses of linear acceleration for the Cartesian manipulator

5 CONCLUSIONS In order to obtain high accuracy mapping of a trajectory with concomitant full utilization of machine dynamic capabilities, it is necessary to generate smooth trajectories with minimum jerk constraints, acceleration or velocity. This paper proposes a method for planning a rectilinear-arc trajectory in which two opposite requirements meet, i.e. trajectory smoothing and simultaneous passing through a sharp corner. The simulation tests performed allowed for the formulation of the following final remarks: 1. For the reason that a tool passes through the point M twice, the generated motion trajectory can be used both, as a whole – BT1T2E or in part – BME. In the latter case, the on-loop motion MT1T2M can be treated as the tool output. 2. At the characteristic points B, M, E of the trajectory, the acceleration profile is tangential to the time axis which causes zero jerk value at these points. 3. The algorithm is effective for calculations. The most laborious prove to be the calculations of trajectory geometry, whereas to obtain a profile of

6 REFERENCES [1] Boryga, M., Graboś, A. (2009). Planning of manipulator motion trajectory with higher-degree polynomials use. Mechanism and Machine Theory, vol. 44, no. 7, p. 14001419, DOI:10.1016/j.mechmachtheory.2008.11.003. [2] Chettibi, T. (2006). Synthesis of dynamic motions for robotic manipulators with geometric path constraints. Mechatronics, vol. 16, no. 9, p. 547-563, DOI:10.1016/j.mechatronics.2006.03.012. [3] Constantinescu, D., Croft, E.A. (2000). Smooth and time-optimal trajectory planning for industrial manipulators along specified path. Journal of Robotic Systems, vol. 17, no. 5, p. 233-249, DOI:10.1002/ (SICI)1097-4563(200005)17:5<233::AIDROB1>3.0.CO;2-Y. [4] Elnagar, A., Hussein, A. (2000). On optimal constrained trajectory planning in 3D environments. Robotics and Autonomous Systems, vol. 33, no. 4, p. 195-206, DOI:10.1016/S0921-8890(00)00095-6. [5] Gasparetto, A., Zanotto, V. (2007). A new method for smooth trajectory planning of robot manipulators. Mechanism and Machine Theory, vol. 42, no. 4, p. 455471, DOI:10.1016/j.mechmachtheory.2006.04.002. [6] Gasparetto, A., Zanotto, V. (2008). A technique for time-jerk optimal planning of robot trajectories. Robotics and Computer-Integrated Manufacturing, vol. 24, no. 3, p. 415-426, DOI:10.1016/j.rcim.2007.04.001. [7] Machmudah, A., Parman, S., Zainuddin, A., Chacko, S. (2012). Polynomial joint angle arm robot motion planning in complex geometrical obstacles. Applied Soft Computing, vol. 13, no. 2, p. 1099-1109, DOI:10.1016/j.asoc.2012.09.025. [8] Perumaal, S., Jawahar, N. (2012). Synchronized trigonometric S-curve trajectory for jerk-bounded timeoptimal pick and place operation. International Journal of Robotics & Automation, vol. 27, no. 4, p. 385-395, DOI:10.2316/Journal.206.2012.4.206-3780. [9] Tian, L.F., Collins, C. (2003). Motion planning for redundant manipulators using a floating point genetic algorithm. Journal of Intelligent & Robotic Systems, vol. 38, no. 3-4, p. 297-312, DOI:10.1023/ B:JINT.0000004973.29102.33. [10] Bu, W., Liu Z., Tan, J. (2010). Motion planning for industrial robots based on task process decomposition.

Trajectory Planning of an End-Effector for Path with Loop

813

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 804-814

Chinese Journal of Mechanical Engineering, vol. 23, no. 2, p. 135-141, DOI:10.3901/CJME.2010.02.135. [11] Chen, Y., Wei, H., Sun, K., Liu, M., Wang, T. (2011). Algorithm for smooth S-curve feedrate profiling generation. Chinese Journal of Mechanical Engineering, vol. 24, no. 2, p. 237-247, DOI: 10.3901/ CJME.2011.02.237. [12] Du, D., Liu, Y., Guo, X., Yamazaki, K., Fujishima, M. (2010). An accurate adaptive NURBS curve interpolator with real-time flexible acceleration/ deceleration control. Robotics and ComputerIntegrated Manufacturing, vol. 26, no. 4, p. 273-281, DOI:10.1016/j.rcim.2009.09.001. [13] Emami, M.M., Arezoo, B. (2010). A look-ahead command generator with control over trajectory and chord error for NURBS curve with unknown arc length. Computer-Aided Design, vol. 42, no. 7, p. 625632, DOI:10.1016/j.cad.2010.04.001. [14] Erkorkmaz, K., Altintas, Y. (2001). High speed CNC system design. Part I: jerk limited trajectory generation and quintic spline interpolation. International Journal of Machine Tools & Manufacture, vol. 41, no. 9, p. 1323-1345, DOI:10.1016/S0890-6955(01)00002-5. [15] Farouki, R.T., Tsai, Y.F., Wilson, C.S. (2000). Physical constraints on feedrates and feed accelerations along curved tool paths. Computer Aided Geometric Design, vol. 17, no. 4, p. 337-359, DOI:10.1016/S01678396(00)00004-2. [16] Nam, S.H., Yang, M.Y. (2004). A study on a generalized parametric interpolator with real-time jerk-limited acceleration. Computer-Aided Design, vol. 36, no. 1, p. 27-36, DOI:10.1016/S0010-4485(03)00066-6. [17] Stori, J.A., Wright, P.K. (2000). Constant engagement tool path generation for convex geometries. Journal of Manufacturing Systems, vol. 19, no. 3, p. 172-184, DOI:10.1016/S0278-6125(00)80010-2. [18] Zheng, C., Su, Y., Mueller, P.C. (2009). Simple online smooth trajectory generations for industrial systems. Mechatronics, vol. 19, no. 4, p. 571-576, DOI:10.1016/j.mechatronics.2008.11.017. [19] Kovacic, M., Balic, J. (2003). Evolutionary programming of a CNC cutting machine. International

814

Journal of Advanced Manufacturing Technology, vol. 22, no.1-2, p. 118-124, DOI:10.1007/s00170-0021450-8. [20] Lloyd, J., Hayward, V. (1993). Trajectory generation for sensor-driven and time-varying tasks. International Journal of Robotics Research, vol. 12, no. 4, p. 380393, DOI:10.1177/027836499301200405. [21] Macfarlane, S., Croft, E.A. (2003). Jerk-bounded manipulator trajectory planning: design for real-time applications. IEEE Transactions on Robotics and Automation, vol. 19, no. 1, p. 42-52, DOI:10.1109/ TRA.2002.807548. [22] Erkorkmaz, K., Yeung, C.H., Altintas, Y. (2006). Virtual CNC system. Part II. High speed contouring application. International Journal of Machine Tools & Manufacture, vol. 46, no. 10, p. 1124-1138, DOI:10.1016/j.ijmachtools.2005.08.001. [23] Dong, J., Ferreira, P.M., Stori, J.A. (2007). Feedrate optimization with jerk constraints for generating minimum-time trajectories. International Journal of Machine Tools & Manufacture, vol. 47, no. 12-13, p. 1941-1955, DOI:10.1016/j.ijmachtools.2007.03.006. [24] Tsai, M.S., Nien, H.W, Yau, H.T. (2008). Development of an integrated look-ahead dynamics-based NURBS interpolator for high precision machinery. ComputerAided Design, vol. 40, no. 5, p. 554-566, DOI:10.1016/j. cad.2008.01.015. [25] Tseng, S.J., Lin, K.Y., Lai, J.Y., Ueng, W.D. (2009). Nurbs curve interpolator with jerk-limited trajectory planning. Journal of the Chinese Institute of Engineers, vol. 32, no. 2, p. 215-228, DOI:10.1080/02533839.200 9.9671498. [26] Wang, X., Wang, J., Rao, Z. (2010). An adaptive parametric interpolator for trajectory planning. Advances in Engineering Software, vol. 41, no. 2, p. 180-187, DOI:10.1016/j.advengsoft.2009.09.010. [27] Grabos, A., Boryga, M. (2013). Trajectory planning of end-effector with intermediate point. (Eksploatacja i Niezawodnosc) Maintenance and Reliability, vol. 15, no. 2, p. 182-187.

Boryga, M.

Received for review: 2013-11-20 Received revised form: 2014-03-25 Accepted for publication: 2014-04-25

Towards the Maintenance Principles of Cyber-Physical Systems Ruiz-Arenas, S. – Horváth, I. – Mejía-Gutiérrez, R. – Z. Opiyo, E. Santiago Ruiz-Arenas1,2,* – Imre Horváth1 – Ricardo Mejía-Gutiérrez2 – Eliab Z. Opiyo1 1Delft

University of Technology, Faculty of Industrial Design Engineering, the Netherlands 2Universidad EAFIT, Design Engineering Research Group, Colombia

Cyber-physical systems (CPSs) are rapidly proliferating in different applications. Their system features significantly differ from those of linear complex systems (LCSs). Consequently, they pose novel challenges with regard to ensuring the dependability of system operation. Maintenance of CPSs raises new theoretical and practical issues. To guarantee a high level of dependability, new and efficient system maintenance principles should be explored and operationalized in various contexts. This paper reports on the first results of the authors’ work in this direction. A comprehensive literature review has been conducted with the objective of identifying the specific features of LCSs and CPSs. We analysed the major maintenance principles and approaches currently applied to complex systems to see how they can be applied to CPSs. We found that the existing maintenance principles have various relationships with CPSs: (i) some of them cannot be considered in the context of CPSs due to incongruent system features, (ii) some of them can be adapted due to certain partial congruencies, and (iii) some of them can be applied directly due to the congruency of some system features of LCSs with CPSs. It was also found and demonstrated through a number of practical examples that many specific maintenance principles need to be developed for CPSs. We assert that the system features of CPSs without parallel in LCSs primarily reveal what sort of new maintenance principles and approaches are needed. The ultimate goal of our on-going research is to define and test these new maintenance principles. In this paper, we identify and define these principles, starting from the unique system features of CPSs and aiming to develop a maintenance advisory system. Keywords: complex systems, cyber-physical systems, maintenance principles, failures, maintenance advisory system

0 INTRODUCTION Technical maintenance is an important multifaceted set of activities performed to preserve the operation of the system in a dependable and optimal state. It involves activities such as inspection, adjustment, replacement, repair, overhaul, and renewal. Maintenance increases the useful life and reliability of systems, reduces the size, scale and number of repairs, and the need for emergency repairs as well as the overall costs while increasing safety and security. Various policies have been conceptualized, which are operationalized through various maintenance principles that have been widely studied in the context of conventional engineering systems. However, maintenance becomes a challenging issue as the heterogeneity and complexity of systems increase. In general terms, the higher the amount of uncontrollable conditions, the more uncertain the physical world becomes [1]. The overall objective of our research is to address the challenge of maintenance of cyber-physical systems (CPSs), and to define a possible set of generic principles that can be applied in the development of system-specific maintenance plans and actions. We presumed that some maintenance principles of linear complex systems (LCSs) may be considered, but also that many new maintenance principles will likely be needed due to the distinct system features of CPSs. The literature suggests that the maintenance principles of LCSs have been derived by considering two complementary maintenance strategies, i.e.

preventive maintenance and corrective maintenance. The principles associated with them lent themselves to the development and application of various design methods, such as redundancy, that are nowadays commonly used to enhance reliability, fault tolerant operation, and ease of repair [2]. In the framework of preventive maintenance, time-based plans are generated for periodic and systematic testing and the replacement of fault-prone elements to prevent sudden failures. Modern predictive maintenance intends to apply sensing technologies to monitor the status of the physical system components in real time, and to initiate the necessary maintenance actions. It also envisages equipping systems with reasoning capabilities to support automated decision making on the necessity of maintenance. With the goal of restoring its intended operation, corrective maintenance is carried out after the malfunctioning, failure, or breakdown of system component has been detected. As mentioned above, the known principles of systematic maintenance have been developed and applied in the case of LCSs, whose behaviour is linear and remains so even under intensively varying operational circumstances. This is an important issue because CPSs typically operate as highly dynamic systems, while LCSs operate as steady-state systems. CPSs may even sometimes operate under unpredictably varying environmental conditions [3] and their performance may be mainly influenced by the effects of such external factors [1]. The resulting non-linear operation makes it difficult to

*Corr. Author’s Address: Delft University of Technology, Faculty of Industrial Design Engineering, Landbergstraat 15, Delft, the Netherlands, s.ruizarenas@tudelft.nl

815

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 815-831

predict momentary system behaviour and to ensure permanent system availability. Our literature review explored a significant knowledge gap in the field of the maintenance of non-linear systems. The methods and tools currently used for controlling the physical part and the cyber part of CPSs are very different and often do not fit adequately [4]. To this end, systems science attempts to integrate knowledge from different engineering disciplines [5] and to facilitate a concurrent management of the computational part and the physical part of cyber-physical systems [6]. Should the research begin out of the knowledge of complex systems science and/or out of the currently evolving science of CPSs in order to derive these particular maintenance principles, or can some of the known maintenance principles of LCSs be reused? This is the central research question of this paper. The reasons this question has both theoretical and practical significance are that (i) many research activities have concentrated on transferring the fault tolerance-related system features of LCSs to CPSs, (ii) only a few of them have focused on the systematic development of maintenance principles that could provide efficient preventive or corrective maintenance solutions for CPSs, and (iii) the distinctive system features of CPSs have not been sufficiently studied for their influences on maintenance. Therefore, we first systematically surveyed and analysed the major system features and the currently applied maintenance principles of LCSs. Then, we investigated the applicability of the maintenance principles of LCSs to CPSs based on a comparison of the system features of LCSs and CPSs. We note here that, for our purposes, system features have been perceived to consist of functional, structural, operational, interaction, application and behavioural attributes and characteristics that differentiate artefacts and service combinations. A first observation has been that the strong interdependency of the components of CPSs will probably make them more vulnerable to errors, attacks and failures [7]. We also observed that the available methods used in the context of maintenance of CPSs are premature and suffer from some fundamental limitations, such as their limited ability to deal with uncertain situations [8]. The reasoning model shown in Fig. 1 has been used in our explorative investigation. Our objective was to establish relationships between the system features of CPSs and the maintenance principles applied for LCSs. First, we investigated in which sense the system features of LCSs and CPSs differ from each other (Arrow A). Next, we considered which maintenance principles are commonly used in 816

LCSs (Arrow B). Following that, we analysed what maintenance principles can be considered for CPSs, taking into account the differences between the system features of LCSs and CPSs (Arrow C). As the title of this paper suggests, the ultimate objective of our research is to explore specific maintenance principles for high-end CPSs (Arrow D).

Fig. 1. Reasoning model used in knowledge aggregation

Therefore, the paper has been structured as follows: Section 1 presents the analysis of the system features of ordinary and complex systems, and Section 2 those of CPSs. This created a knowledge platform for the follow-up reasoning about the relevance of the maintenance principles of LCSs to CPSs. Section 3 analyses the principles currently applied in the maintenance of LCSs. It discusses various maintenance policies commonly applied to physical devices, as well as various failure management mechanisms that have been used in embedded systems. Section 4 projects the known maintenance principles of LCSs to generic CPSs. Section 5 discusses in which manner the maintenance principles with application potentials can be considered in developing general maintenance principles for CPSs. Section 6 discusses the need for dedicated additional maintenance principles for CPSs. Section 7 presents some examples that illustrate the consequences of applying the identified maintenance principles in a cyber-physical greenhouse. Section 8 evaluates the findings and proposes further research activities. 1 EXPOSITION OF SYSTEM FEATURES OF LINEAR COMPLEX SYSTEMS Science differentiates systems based on complexity and behaviour. The complexity of systems is a measure influenced by factors, such as the (i) the number of components included in the system, (ii) the type of components that constitute the system, (iii)

Ruiz-Arenas, S. – Horváth, I. – Mejía-Gutiérrez, R. – Z. Opiyo, E.

StrojniĹĄki vestnik - Journal of Mechanical Engineering 60(2014)12, 815-831

the number of sub-systems of different scales, (iv) the interconnections among co-located components, (v) the communications among geographically dislocated components, (vi) the interactions of the system with stakeholders, and (vii) the connections of the system with its environments [2]. In addition to these, (viii) the heterogeneity of the components, and (ix) the distinct material, energy and information flows within the system are also factors influencing system complexity. From an engineering point of view, system operation or behaviour can be linear or non-linear. Therefore, in our interpretation, systems can be categorized as simple and complex and as linear or non-linear. A system is linear if: (i) it is functionally and structurally reductionist, (ii) its output is directly proportional to its input, and (iii) it satisfies the superposition principle. Simple and compound reductionist engineering systems belong to this category. CPSs can also be linear systems, but the overwhelming majority of them fall into the category of non-linear complex systems. As such, they can be found in several alternative forms, e.g. as complicated, adaptive, evolving and replicating complex systems. Systems belonging to the category of non-linear complex systems have some sort of learning and selforganizing capabilities. We had to be pragmatic in our research because adaptive, evolving and replicating complex systems are still in their infancy from an implementation point of view, and thus the knowledge about their behaviour and maintenance needs remains limited. In addition, there are many open theoretical and practical issues concerning the design, implementation and operation of these systems. Consequently, their utilization into the practical life remains rather limited. This explains why we have made the choice to focus only on complicated cyber-physical systems (C-CPSs) in the first phase of our research. Furthermore, in theory non-linear complex systems (NLC) CPSs are able to optimize their overall performance in cases of largely varying environmental conditions, changing internal relationships, and operational discontinuities. Likewise, we do not intend to address the issues related to so-called self-healing systems, which supposedly have the capability of automatically regaining functionality when components break down, or signiďŹ cant perturbations occur in the system. Ordinary systems (OSs) are, usually, simple small-scale systems in which a single cause produces a single effect, which makes them reducible, composable and predictable in modelling and design. The basic assumption is that a small change in the input implies a small change in the output [9].

Examples of OSs are electro-mechanical systems, such as a coffee maker or a refrigerator, which have pre-programmed or adjustable control devices, and operate under steady-state conditions. These types of systems usually have only one energy source, one integrated functional unit, and one interface unit. As a consequence of their pre-programmed nature, no changes or updates are possible in their embedded software after their release to the market, and they cannot manage emergent real-time data [10]. LCSs are complex in the sense that they are composed of a diverse set of interconnected components, but do not have any capability to reorganize their structure, or change their designed functionality. The overall operation of LCSs is a union of the operation of their components. In other words, the aggregated functionality of the components determines the operation of the system as a whole, and no emergent behaviour occurs due to the interaction among the components or within the environment in which the system is embedded. They are closed systems with centralized architectures and control functions, which are aligned to the tendency of the so-called disappearing computer (that hides software components in a physical device) [11]. Therefore, these systems are controlled by a microprocessorenabled embedded software system, and they typically perform (much) more complex operations than OSs [12]. The control function is realized through multiple feedback loops through which the software monitors and controls the whole system and its components in an optimized way [2]. The functional components of LCSs intensively interact with each other and the surrounding environment [11]. This type of system may be geographically distributed and decentralized, equipped with multiple energy sources, may have repetitions in the functional units, and the components may communicate by using wireless technologies. LCSs are widely used, for instance, in the automotive, electronics, avionics, railways, telecommunication, health, and security sectors [11]. In these systems, the maintenance of the physical components may be carried out by using preventive and corrective maintenance procedures. 2 SYSTEM FEATURES OF CYBER-PHYSICAL SYSTEMS CPSs came about as a result of the emergence of faster computer processors, the miniaturization of electronic components, broader communication bandwidths, and seamless integration of networked computing with everyday systems [13]. They blend

Towards the Maintenance Principles of Cyber-Physical Systems

817

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 815-831

physical technologies, software and middleware technologies, and cyber technologies. Future systems will make more extensive use of synergic technologies, which integrate hardware and cyber technologies [14]. Physical technologies enable the implementation of artefacts that can be recognized, located, operated, and/or controlled in the physical world [15]. Cyber technologies are used for capturing, analysing and processing sensed signals and data produced in the physical world for decision-making. Synergic technologies enable not only a borderless interoperation between physical and cyber elements, but also a holistic operation of the whole system. The design of the physical and computational aspects is becoming an integrated activity [16]. As mentioned above, CPSs link the physical world with the cyber world through the use of multiple sensor and actuator networks integrated under an intelligent decision system [17]. In other words, CPSs combine sensing and actuation with computation, networking, reasoning, decision making, and the supervision of physical processes [18]. With a view to their emergent nature, it seems expedient to differentiate low-end and high-end implementations of CPSs based on the extensiveness and sophistication of the resultant integrity [14]. Low-end implementations are linearly complex, closed architected, distributed and networked, sensing and reasoning enabled, smart and proactive, (often embedded and feedback controlled) collaborative systems. High-end implementations are non-linearly complex, open and decentralized, heterogeneous and multi-scale, intelligent and partly autonomous, self-learning and context-aware systems. The systems belonging to the latter class of CPSs display organization without any predefined organizing principle and change their functionality, structure and behaviour by self-learning, self-adaption, or self-evolving. The previously mentioned C-CPSs are low-end implementations because they are not supposed to change their functionality or architecture, but to optimize their behaviour, for instance, energy efficiency (e.g., due to the necessity to operate during an extended period of time) [19], while operating under dynamically changing operating conditions or unforeseen circumstances. Some of these systems should operate in real-time applications and provide a precisely timed behaviour [20] as well as achieving a synergic interaction between the physical and the cyber worlds by integrating computational and physical processes [21]. The cyber and physical parts of the systems are interconnected and affect each other through information flows [22]. Due to this functional 818

synergy, the overall system performance is of higher value than the total of the individual components [23]. This synergy is particularly important in the case of high-end CPSs, which exhibit properties such as selforganization [24]. In general, CPSs strive toward a natural human-machine interaction that also extends to the human cognitive domain [25]. These kinds of systems are also capable of exhibiting extensive remote collaboration [26]. Unlike LCSs, CPSs also work on non-dedicated networks [27]. CPSs are often connected in a hierarchical manner, as systems of systems, in which one system monitors, coordinates, controls and integrates the operation of other systems [28]. For this reason, they can be considered to be multi-dimensional complex systems [29]. Based on their functionality and characteristics, high-end CPSs can be used in areas such as transportation, health care, and manufacturing [28].

Fig. 2. Clarification of the main terms: A) relationships of the terms, and B) interpretation of the terms

Some CPSs are mission critical systems (MCSs) because their correct functioning is critical to the success of a mission, provisioning an essential supply, or safeguarding security and well-being [30] and [31]. These are the systems that ensure proper and continuous operation of (for example) nuclear plants, automated robot control systems, and automatic landing systems for aircraft [32]. Any failure in MCSs can lead to loss of human life and to damage to the environment, and may cause losses in terms of supply and cost [33]. However, their operation is always characterized by the presence of uncertainty. This introduces challenges from the point of view of the dependability, maintenance and repair of mission critical non-linear cyber-physical systems [14]. In

Ruiz-Arenas, S. – Horváth, I. – Mejía-Gutiérrez, R. – Z. Opiyo, E.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 815-831

the long run, it is crucial to comprehensively analyse what the maintenance of these systems theoretically, methodologically, and practically means, and how it can be implemented in different systems.

which operate with a lower level of synergy [37]. The maintenance of LCSs is based on this classic view and, consequently, its maintenance principles are also based on it (Fig. 3).

3 OVERVIEW OF THE MAINTENANCE PRINCIPLES APPLIED TO LCSs The terminologies related to system maintenance do not seem to be uniform in the literature. Terms such as “maintenance strategy”, “principles” and “policy” are used with various interpretations and meanings, often even interchangeably or confusingly. In general, a policy is defined as a collection of rules that, depending on the most essential state variables, precisely specifies what to do in a particular situation [34]. From a managerial point of view, “strategy” is described as the definition of long-term goals, objectives and courses of action for a company, and the allocation of resources for the achievement of such objectives [35]. There are some other basic terms used in literature whose definition is often taken for granted. These are “principle”, “method”, “rule” and “action”. We adopted the Oxford dictionary definitions [36]. Therefore, a principle is interpreted as “a fundamental source or basis of something”; a method as a “particular procedure for accomplishing or approaching something” in a systematic way, a rule as a set of explicit understood regulations, and an action as the logically separable procedural element of doing something. The application of these terms in our maintenance context is presented in Fig. 2A. This figure shows the interrelationships between the above-defined significant terms and separates them according to whether they are of epistemological (knowing) or praxiological (executional) flavour. Fig. 2B summarizes the above interpretation of the two groups. Maintenance seeks to ensure the permanent availability of a system through the application of its basic principles. Consequently, these principles should be applicable to any system, including CPSs. However, it has been recognized that this claim is not apparent with regards to CPSs, because these systems should be considered differently from a maintenance perspective due to the inherent heterogeneity of their physical, software and cyberware components. The high level of synergy makes the maintenance of the three basic kinds of components inseparable from each other. This is in contrast with the classic view of the abstract machine architecture in which systems are composed of hardware and software components,

Fig. 3. Articulation of maintenance strategies for LCSs

The maintenance principles of LCSs are applicable from both physical (artefactual systems) and cyber (software systems) perspectives. The physical part maintenance is based on the assumption that every component of the system has a limited life cycle and, therefore, may be subjected to wear or breakdown. Therefore, for the physical part, the main (global) principles have been maintaining system availability and doing so in a cost-effective way. Based on this principle, two main approaches have been developed: preventive maintenance (PM) and corrective maintenance (CM). Principles of PM aim to avoid failures before they occur through preventive actions, such as revisions, exchanging components and repairs. The principles of CM allow a system to operate up to the occurrence of failures if the consequences of failures are not critical, or do not have an effect during a particular operation period [38]. PM may be conducted according to the principle of time-based maintenance (TBM) or of condition-based maintenance (CBM). The TBM principle (P1) entails scheduling maintenance actions [39]. Therefore, knowledge management techniques should be applied in order to determine a schedule for conducting revisions, exchanging of components and repairs, while the CBM principle (P2) is based on the completion of inspection activities by which maintenance actions will be initiated and completed. This principle can be applied to component that do not exhibit failure predictability or fail randomly, while scheduled maintenance principle may be applied to those components that show evident signs of wearing [38]. In contrast, CM can be implemented by following the principles of (i) failure-based maintenance (FBM), (ii) opportunistic-based maintenance (OBM), and (iii) design-out maintenance (DOM). OBM and DOM

Towards the Maintenance Principles of Cyber-Physical Systems

819

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 815-831

belong to this group of principles because both of them assume failures to occur in order to be identified. Therefore, the FBM principle (P3) considers repair and making changing components once a failure had occurred [40], it deals with maintenance only if failures or breakdown occurs [41]. The principle of OBM (P4) suggests completing general inspection of all of the components when any of them fails. It has been reported that combining PM activities can lead to savings in terms of system cost [42]. Therefore, the principle of OBM states that there is an opportunity to conduct general maintenance when a maintenance intervention is required for other components [43]. Finally, the principle of DOM (P5) aims to use redesign to avoid the causes of failure. This principle is usually applied when breakdowns frequently occur [44]. The application of one or another principle depends on how likely the components exhibit wear characteristics and how random components fails. In the context of maintenance of software (and knowledge) intensive information systems, the primary assumption is that there are obviously no physical (e.g. wearing) processes. Therefore, the primary maintenance principle for software systems is that failures should be self-avoided and self-managed by the system. Several principles may be applied in the case of software system for a proper fault management [45] and [8]. These may be based on the consideration of: (i) fault prevention, (ii) fault removal, (iii) fault detection and isolation, (iv) fault forecasting, (v) fault tolerance, and (vi) fault reporting. The principle of fault prevention (P6) seeks to avoid the occurrence of faults through preventive actions [46], while the principle of fault detection and isolation (P10) aims to detect and determine whether a fault occurred in a particular system, by attempting to autonomously detect these faults and to isolate the affected component [47]. Having a different objective, the principle of fault removal (P7) seeks to reduce the number of faults and their severity [48]. The principle of fault forecasting (P9) allows predicting failures and their impact, based on the fault records [49]. It entails estimating the incidence and consequences of faults, based on the present number of faults. The principle of fault tolerance (P8) aims to assure the continuity of system operation, despite the presence of faults, errors or attacks [50]. The principle of fault reporting (P11) is based on alerting the user or operator in case there is a fault in order to allow actions to be taken [51]. All of the aforementioned principles for the software (and cyber) side are focused on autonomously taking actions such as identification, diagnosis, isolation, repair and/or reporting. 820

In addition to the abovementioned principles, the principle of e-maintenance and the principle of vaccination, which are preferred and commonly implemented in auto-immune systems (AIS), have also been identified and worked out. The principle of e-maintenance is based on the exploitation of particular ICT affordances for enhancing the effectiveness of maintenance decisions [52]. It entails making use of information technologies to exploit data required in decision-making. This principle is mostly applied in manufacturing plants where full system availability is required. The principle of vaccination has a natural analogy. In the context of human beings, the principle of vaccination seeks to create immunity to any particular disease by introducing a soft version of the disease in the body, and to generate a memory of the pathogens. This natural principle has been extended to software systems, and now it allows the adaptation of the system behaviour against new and evolving attacks [53]. It is usually applied to systems whose complexity levels are higher than of OSs. Since we have focused in our research on systems that need external management of maintenance, rather than taking care of it by themselves, we will not deal with the principles of e-maintenance and vaccination. As for the current state, it is apparent that an extensive set of maintenance approaches are available for LCSs, regardless of whether purely physical or purely software systems are considered. It can be argued that cyber-physical systems need some sort of combination or even a blending of these in order to be able to provide system dependability. As our survey and analysis has revealed, from the physical perspective, maintenance is conducted to avoid general system failures, or to reduce their probability, based on repairs, spare changes, and revision activities. From the software (and cyber) perspective, maintenance is orientated to the control functions of systems and they are usually kept operational through fault management. In the latter case, the intension is to assure system operation even in the presence of failures, or when facing any type of faults. Our other observation has been that both hardware and software systems’ possible failures are addressed by maintenance principles that have been developed for LCSs. The combination of the hardware- and the software-systems oriented maintenance principles works properly in LCSs systems. However, when the level of complexity of the target system increases and the operation of the system becomes non-linear, these changes cause a higher level of unpredictability. This has consequences on the applicability of the maintenance principles. In the case of complicated

Ruiz-Arenas, S. – Horváth, I. – Mejía-Gutiérrez, R. – Z. Opiyo, E.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 815-831

CPSs, it is also important to analyse how the interactions among the system components happen under varying operational conditions, and how they may affect the operation of the system as a whole. In other words, it is necessary to investigate how maintenance principles should be adapted to meet the functional requirements of cyber-physical systems. 4 PROJECTING THE MAINTENANCE PRINCIPLES OF LCSs TO CPSs A high-level, three-tier structure is proposed in [54] as a reasoning model for the maintenance of nonlinear complex systems such as CPSs; it includes an environmental tier, a service tier, and a control tier. The environmental tier is related to the physical devices, the service tier is a typical computing environment with services in a service-oriented architecture (SOA), and the control tier is for decision making. This reasoning model clearly differentiates the methods and techniques that can be applied for artefactual systems and for software systems (Fig. 3). This differentiation is significant from the aspect of applying the traditional maintenance principles of LCSs in CPSs. Since we have decided to focus on complicated systems, this reasoning model has many limitations. Consequently, we have based our study on the model proposed by [14] which identifies three generic constituents: (i) physical technologies, (ii) cyber technologies, and (iii) synergic technologies. This model specifies that the maintenance policy of CPSs should consider these three constituents in their synergy (Fig. 4.) We use this reasoning model to facilitate the simultaneous consideration of the hardware, software and information content-related issues and principles of maintenance.

the analysis of maintenance principles of LCSs in the previous chapter, provided a basis for us to determine which maintenance principles are transferable to CPSs. We investigate each of the maintenance principles from the aspect of transferability below. In the assessment, we take into account the similarities of the system features of the two kinds of systems, as well as the importance of the functions that they perform. • Schedule maintenance actions (P1) This maintenance principle is appropriate for LCSs that operate continuously. In order to avoid system failures, the different common working cycles of the physical components are taken into consideration, together with signals concerning their state of wear. The states of the components and their criticality together determine when a maintenance action should be executed. As for the transferability of this principle to CPSs, we can argue that CPSs are dynamic systems whose actual operations cannot always be predicted with a high degree of probability. This dynamic operation affects the common working cycles of the physical components and thus the frequency of maintenance may be different for each of them. In the context of our exploration, it means that although this principle cannot be neglected it needs adaptation to be adequate for CPSs. •

Support maintenance actions by monitoring activities (P2) Even in the case of traditional LCSs, implementation of this maintenance principle requires augmentation with agents for operation monitoring. In principle, these agents can be embedded in CPSs that normally have a set of physical sensors, sensor networks, or software sensors. For this reason, this principle can be transferred to CPSs. The necessary maintenance actions of the system will be determined based on permanent monitoring, which can be applied even in the case of a non-linear behaviour of CPSs. Conduct maintenance actions once a failure has occurred (P3) Application of this principle to CPSs is far from straightforward, in particular when mission critical systems are considered. In the case of MCSs, continuous availability is not negotiable, and risk in the operation is usually not tolerable. It implies that general system failures, as well as cascade failures, should be avoided through engineering actions, or by dedicated system functions. This however implies that this maintenance principle should not be considered in the case of mission critical CPSs. •

Fig. 4. Doctrine of integral maintenance for CPSs

The analysis concerning the congruencies of the system features of LCSs and CPSs, extended with

Towards the Maintenance Principles of Cyber-Physical Systems

821

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 815-831

•

Conduct general maintenance once any of the system components fail (P4) This principle is not associated with any particular system feature of LCSs and C-CPSs, but considers the entire system. For this reason, its applicability raises a concern regarding the fact that the overwhelming majority of C-CPSs are complex, decentralized systems, whose subsystems and modules may be characterized by some level of autonomy and operation profile. In other words, they may have and operate according to their own maintenance scenarios. Consequently, they may not need to go through general maintenance when failure in other subsystems or modules occurs. It has to be noted that this principle may be relevant to complex components.

• Redesign to avoid the cause of failures (P5) This principle is also not related to any particular system feature of LCSs or CPSs. The application of this principle entails re-designing the components and features of the system if they prone to be the source of recurrent failures [39]. Redesigning may be needed or be advantageous because recurrent failures can significantly affect the overall availability of CPSs and can increase the costs of operation. In the case of CPSs, further consideration is needed if the redesign is to focus on the hardware, software, or the information contents constituents, or any combination of them. •

Avoid failures in the system by preventing the occurrence of faults (P6) This principle works best in the case of systems such as LCSs whose operations are not highly dynamic. The reason is that the application of this principle assumes the conducting of reliability tests during the system development and installation stages. Consequently, possible failures and failure modes that may negatively affect the operation of the systems can be explored or predicted. However, CPSs are dynamic and highly complex systems, and their testing before fullscale operation cannot be exhaustive. The currently used testing approaches cannot cover all aspects of the operation of CPSs. Therefore, transferring this principle to CPSs necessitates adaptation. •

Reduce the amounts of faults and their severity (P7) This principle is applied during the design stage of the system with the objective of avoiding functional and structural failures. Although this principle in theory can be considered applicable to CPSs, it should be adapted to their system features. Multi-aspect fault propagation prevention methods and failure 822

interaction evaluation methods will certainly be needed to make this principle applicable and efficient. •

Assure continuity of system operation despite the presence of faults (P8) This principle can be transferred to CPSs because the intelligence (i.e. sensing, reasoning and actuator capabilities) embedded in these systems can support its implementation by detecting the faults of physical components and the malfunctioning of software components, and activating protection mechanisms. Decentralization of the system operation and control also allows conducting an adaptive resource management. In addition, the application of preventive and corrective measures such as redundancy, reconfiguration and replacement, may be used to avoid complete system failure. However, the large possible number of functional connections complicates the identification of affected components and the prevention of fault propagation. Other advanced characteristics of CPSs, such as selforganization and self-adaptation, can take care of assuring the continuity of system operation, and in general, facilitate the application of this principle. Therefore, high-end CPSs will be able to transfer tasks from failed components to components that are working properly while the fault is eliminated. As a result, this principle can be the main principle for the maintenance strategy of CPSs. • Predicting failures on the system (P9) The main objective of applying this principle is to forecast faults and failures and systematically avoid them. It is the most effective principle for systems with limited complexity and operational linearity, such as LCSs. Incongruities of the system features of CPSs and LCSs affect its applicability to CPSs, as the currently applied predictive and/or probabilistic models developed for LCSs are not appropriate for CPSs. However, the self-diagnosis and self-adaptation capabilities of CPSs may contribute to the effective application of this principle. • Detect and isolate faults (P10) Traditionally, this principle is operationalized rather “manually”. However, the intelligence and autonomy of CPSs may significantly influence the application of this principle. CPSs may be equipped with capabilities to detect fault events autonomously, and may analyse the consequences of emergent faults. Furthermore, the interactions among components allow the extraction of information for different devices to conduct performance tests, which contribute to the detection of whether the system operates properly.

Ruiz-Arenas, S. – Horváth, I. – Mejía-Gutiérrez, R. – Z. Opiyo, E.

StrojniĹĄki vestnik - Journal of Mechanical Engineering 60(2014)12, 815-831

Decentralization makes it possible to properly manage resources during the execution of these tests. This resource management will avoid system overloads and, therefore, the occurrence of faults or errors in processing. As a result, we argue that P10 can directly be applied to CPSs. â&#x20AC;˘ Alerting the operator in case of fault (P11) This principle can be transferred to CPSs because its implementation only entails application of information technologies in physical devices as long as LCSs and CPSs have physical features. The differences between system features do not affect the applicability of this principle. 5 OPERATIONALIZATION OF RELEVANT MAINTENANCE PRINCIPLES FOR CPSs The above analysis shows that there are different relationships between the generic system features of CPSs and the maintenance principles that have been used in LCSs. Four categories of relationships can be identified: (i) non-applicable principles (Px), (ii) adaptable principles (Pa), (iii) exportable principles (Pe), and (iv) additional (Pn) principles (Fig. 5). Two maintenance principles, i.e. P3 and P4, belong to the category of non-applicable principles (i.e. Px = (P3,P4 )). These seem to be problematic in the context of maintenance of CPSs due to their probable criticality and decentralization. Conducting maintenance actions once a failure has occurred in mission critical CPSs is not logical. Likewise, conducting general maintenance on a system that is capable of managing the consequences of failures in their separate or autonomous parts is also illogical.

Fig. 5. Roadmap towards maintenance principles for CPSs

The rest of the principles seem to be applicable but in different ways. There are principles that

can be applied without any modifications. These have been named exportable principles. They are: Pe = (P2, P5, P8, P10, P11). They can be used without modifications, but the way of applying these depends on the CPS in question. The remaining group of principles can be applied only after a purposeful adaptation. They are: Pa = (P1, P6, P7, P9). Our observation has been that certain system features of CPSs will require additional (not yet specified) maintenance principles, because they cannot be addressed by the principles known to be applicable to LCSs. Since we primarily focused on the reusability of maintenance principles of LCSs in the context of CPSs, these additional novel principles have not yet been explored in our study. In the next section, we will discuss the essence of these additional principles. As mentioned above, the group (Pa) comprises those principles that should be and can be adapted. Their adaptation needs further considerations of the system features. In the following paragraphs, we consider the adaptations that should be made. For instance, the transfer of principle P1 to CPSs needs knowledge about the lifecycle of components, their failure modes and effects, as well as about the specific forms and opportunities for automation of maintenance activities, such as revision, repairs, and spare-part changes. The very reason this principle needs adaptation is that the abovementioned activities greatly differ from those associated with LCSs. Some enabling methods, such as failure mode and effect analysis (FMEA) [55], fault tree analysis (FTA) [56], hazard and operability study (HAZOP) [57], and component fault tree [58] can be applied to conduct specific failure analyses. However, the use of these methods requires a large amount of data and information about operation of complex systems, which may be difficult to obtain [59]. Furthermore, since the use of this principle in LCSs requires a high-level of human involvement, some sort of automation of the scheduled revision activities seems to be necessary. The adaptation should consider the self-monitoring and self-repair potentials of CPSs. It is important to note that P1 was originally developed for physical components that provide observable indication (signals) of wear. Further studies are needed to investigate how this principle can be applied to electronic components, which are normally subjected to random failures only [60]. Principle P6 also requires adaptation in order to provide optimal results for CPSs. The adaptation should consider new different types of tests, which take into account the effects of unexpected external

Towards the Maintenance Principles of Cyber-Physical Systems

823

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 815-831

and internal events. Currently, there are limitations in terms of what can be tested through functional or performance simulations and runtime tests. They should be able to deal with unique faults and failures of CPSs, which do not occur in LCSs. It seems to be necessary to design new protocols for behavioural and performance tests in order to determine how they should be conducted, which values are expected as the key performance indicators, and how to aggregate these in distributed and decentralized systems. Principle P7 implies the consideration and inclusion of fault avoidance and system maintenance at the design stage of the product development process. While various methodologies have been elaborated for LCSs, they do not seem to be directly applicable in the design processes of CPSs. It is necessary to include new design criteria based on the system features of CPSs, as well as quality standards and test procedures in the design processes of CPSs, which involves hardware, software and information platforms design. Further research is needed to develop comprehensive verification and validation methodologies for CPSs and subsystems that can be applied in the early phase of the design process. It is also imperative to investigate how the designed systems will respond to faults and what the impacts and consequences of the potential faults may be. Principle P9 places emphasis on the run-time prediction of possible failures and black outs of CPSs. This principle assumes predictive and/or probabilistic system models that are actualized in run-time and can prognosticate system operation based on the evaluation of subsequent system states. The predictive models currently applied for LCSs are not transferable to CPSs due to the dynamic nature and operational conditions of these systems. Relevant predictive models should be able to capture the internal dynamics of the systems, the dynamic interaction of the systems together with their environment, and the dynamics of the embedding environments. To effectively apply this principle, forecasting mechanisms are needed that are capable of forecasting future faults and failures of CPSs based on operation or application history information. This may be enabled by information provided by networked sensors, tracking the frequency, and amount of failures reported in the system, and even learnt from the conditions under which these faults occurred. Group (Pe) comprises those principles that can be used in the maintenance of CPSs without adaption. However, it has to be mentioned that while these principles do not need reinterpretation or redefinition, the way of operationalizing them in the case of CPSs 824

may be different from the way they are applied in the case of LCSs. Principle P2 can be directly (without adaptation) applied to CPSs due to the availability of the enabling technologies, such as sensing, monitoring, information processing, fault diagnosis, and failure prognosis algorithms [61]. The application process is essentially the same as in the case of LCSs, which typically involves the use of FMEA, FTA, HAZOP, Markov chain [62] and Petri-net [63] methods, and Bayesian models for failure analysis. Methods such as FTA and Petri-nets can also be used for failure propagation analysis [64]. Which signals are to be considered as indicators of faults, how they can be sensed in real time, and with which frequency they have to be sensed and evaluated has to be carefully determined. Similarly, the monitoring frequency for each component needs careful consideration and harmonization. It is, however, acknowledged in the literature that the introduction of a high-level of automation usually results in more complex and costly maintenance actions [65]. The process of applying principle P5 to CPSs is practically the same as to LCSs. It involves analysing the criticality of component failures, as well as the urgency of response and repair actions. The objective of this analysis is to determine the components that are prone to failures, with the highest probability and causes the highest risk levels; and to make decisions on the strategy of redesigning and on better solutions. Decisions can also be made on which failures can be managed by the CPSs themselves, and which need immediate availability of maintenance plans and involvement of personnel. These depend on the forecasted occurrence frequency of component failures. Having considered these influencing factors, whether applying structural redundancy, more resilient components, functional re-configuration, or more robust system architecture can be a better solution can also be determined, taking into consideration the associated costs and extra efforts [2]. The application of principle P8 can be considered as a design challenge. It concerns not only the design decisions and solutions in the design process of CPSs, but the preventive and corrective actions that can be taken during the operation of a system. In other words, the application of this principle requires concurrent elaboration of both a preventive maintenance strategy and a corrective maintenance plan. In the context of CPSs, the objective of principle P10 is to detect and isolate faults through a collaborative strategy that involves the actions of both the maintenance experts and the self-adaptive system.

Ruiz-Arenas, S. – Horváth, I. – Mejía-Gutiérrez, R. – Z. Opiyo, E.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 815-831

The information platform required by the latter can be generated by continuous monitoring of the system, reflective real-time modification of detection algorithms, introducing changes in the system arrangement, and planning the response actions. Once faults are known, they can be prioritized based on the probability of occurrence, as well as on their criticality with regards to system operation. Finally, the reason principle P11 can be applied to CPSs without adaptation is that the system functionality and the technologies used can alert the operator if there is a fault. This involves diagnosisbased report generation, ubiquitous communication of failure information, decision making on the suspension or continuation of system operation, proposals for maintenance or repair, identification of replaceable parts, determination of resources and tool demands, and capacity and activity planning. The system should also determine what information should be delivered to which stakeholders. The three most important pieces of information that should be delivered to the operator are the description of the failure, its place in the system, its neighbourhood, and its criticality. 6 SOME SUGGESTIONS ON SPECIFIC MAINTENANCE PRINCIPLES FOR CPSs Which new principles are needed for a particular family of CPS? It is obvious that due to the complex functionality, structure, and operation of CPSs, they need additional maintenance principles that are not necessary for LCSs. The sought after dedicated principles are especially important for high-end CPSs, which are open, dynamic, decentralized, intelligent and self- organizing systems. Their intense interaction with the natural and engineered environments and penetration into the social and cognitive domains of stakeholders require further investigations, because of the increasing exposure to the environment and humans. The primary CPS features that makes them require novel maintenance principles are: (i) nonlinearity (interaction, circumstances, and behaviour), (ii) applications in dynamic and harsh environments, and (iii) growing level of automation. The non-linearity of CPSs has many sources and forms of manifestation. Open decentralized systems may have the capability to dynamically change their system boundaries. They may also adapt their operation to the actual operation circumstances. In general, the change in the components and the change of behaviour of the components complicates both the forecasting and the correction of the failures. These systems are not predictable as they frequently and

intensively move between many discrete states and transitions [66]. To cope with these characteristics, maintenance principles dedicated to dynamic complex systems are needed. It is imperative that they must address fault management and elimination in hardware, software and information systems in an integrated way. Evidently, eliminating the sharp boundary between analogue and discrete physical components and the software and information system components is a fundamental challenge. Currently, biological analogies, such as the human immune system, are dealt with in some research to understand which features, behaviours and architecture result in perpetual corrective behaviour that emerges from local detection and interventions. In terms of interoperating software, some researchers have dealt with the notion of the fractionated CPS that goes beyond the conventional definition of a softwarecontrolled hardware system that is interacting with the physical world [31]. The operation of CPSs in unpredictable and harsh environmental elements, such as chemical reagents and humidity, also imply the need for new maintenance principles. These and similar operating conditions invalidate the traditional forecasting models, as these conditions will most likely affect the hardware component’s lifecycle, and increase the chances of malfunctioning. Researchers are engaged in finding theories and technological solutions for inherently fault-tolerant dynamic architectures, as well as non-model-based zero-delay monitoring and proactive detection solutions. Another domain of research interest can be vague forecasting based on incomplete and localized bodies of knowledge. Both CPSs and their components are reaching a high level of autonomy. This is enabled by their increasing smartness or intelligence, which is a result of wide-ranging information elicitation, reasoning and inference, and the “agentialization” of system operation. System intelligence also supports moving decision making and preparation of maintenance from the design phase to the runtime phase of the system’s lifecycle. The automation of maintenance not only has positive technical outcomes, but also reduces the required human efforts, intervention, costs and safety, and improves servicing capabilities [67]. It seems to be necessary to include maintenance-related aspects in model-based design of CPSs and to be able to detect near failure states in operation. Finally, there is a need to develop selfmaintenance principles for various families of CPSs. As discussed earlier in this paper, some of the current maintenance principles can be considered in the

Towards the Maintenance Principles of Cyber-Physical Systems

825

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 815-831

case of systems with self-detection (self-diagnosis) and failure prevention capabilities. State sensors built into physical components, smart materials, and emergent behaviour analysers are already used in current CPSs. These principles should thoroughly cover the maintenance process in such a way that human involvement is reduced considerably. It follows that currently used self-diagnosis and failure detection methods should be combined with new techniques for monitoring, changing and repairing parts during system operation. Lee et al. argue that self-maintenance techniques should enable awareness of the changing operational regimes to dynamically select prognostic models in order to ensure accurate prediction [68]. We can say that there are advantages of combining the implementation of this concept with the implementation of response actions through automated actuators. 7 DEMONSTRATION OF THE APPLICABILITY OF MAINTENANCE PRINCIPLES TO CYBER-PHYSICAL GREENHOUSES We use a case of a cyber-physical greenhouse (CPGH) to demonstrate the applicability of the maintenance principles presented and discussed in the previous Sections. A CPGH is considered to be a cyber-physical augmentation of the traditional greenhouse in order to make it capable of providing new services. Actual examples are used to explain how a maintenance principle can or cannot be operationalized in a CPGH, and what types of adaptations may be necessary. 7.1 Non-Applicable Principles •

Conduct maintenance actions once a failure has occurred Because CPGHs are naturally mission critical systems, it is necessary to prevent any system failure, rather than to eliminate the effects of failure, and recover from occurred failures and malfunctioning. What follows from this requirement is that the principles of the strategy of corrective maintenance simply cannot be applied in this case. Instead of these, the operationalization of an extended set of preventive maintenance principles is needed. This need is evident from the following practical challenge: in a CPGH, parameters such as temperature, humidity and CO2 are typically controlled through ventilation. The plantmonitoring sub-system should be able to measure the transpiration and temperature of the plant. If, for example, due to the lack of maintenance, this subsystem fails, the actuators (such as heaters or fans) 826

will not be able to react, or will erroneously respond, and this will cause a serious damage of the plant. •

Conduct general maintenance once any of the system components fail This maintenance principle has no relevance in the context of mission critical CPGH systems. As in the above explained case, if maintenance activities are done only when a component fails, both the risk of plant damage and the hazard of the lack of availability of the entire system prevail. Consequently, only maintenance principles that stimulate preventive maintenance activities should be operationalized. As a practical situation, one can argue that some crops, such as roses, are highly sensitive to changes in temperature. Suppose that no failure has occurred in the entire system until a given point in time. If the above maintenance principle is applied, then even the critical system components are not maintained. Should there be lack of maintenance, for instance, not only may the temperature control sub-system break down, but also other critical components of the greenhouse, such as the boiler, and this may lead to a complete failure of the CPGH system. Likewise, the malfunctioning of the boiler may cause damage to roses during cold seasons, and this may not only seriously affect their quality, but may also cause losses to the grower.

7.2 Exportable Principles •

Support maintenance actions by monitoring activities This principle suggests a continuous monitoring of a system in order to be able to explore the need for maintenance, and to reduce the chance of failure over time. The principle is not only operationalizable, but also very useful in the context of CPGH systems, which should typically feature multiple wireless sensor networks. We can illustrate possible practical utilization in this regard. Let us consider, for example that definition of the so-called “set points” is currently done in CPGHs based on monitoring temperature, humidity and CO2 levels. Any unexpected variation in these sensed parameters with respect to the “set points” may lead to improper operation. Furthermore, the observed variations of the parameters can be used as alerting signals of failure. It can also be the case that variations in the physiological parameters of plants may also cause failures in the sensors and/or actuators.

Ruiz-Arenas, S. – Horváth, I. – Mejía-Gutiérrez, R. – Z. Opiyo, E.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 815-831

• Redesign to avoid the causes of failure To discuss the reusability of this maintenance principle, let us use a practical example of the sensors used for measuring plant transpiration and temperature. It can occur in both traditional and cyber-physical greenhouses that sensors are recurrently suffering from failures due to the effects of humidity and chemical corrosion within the greenhouse. Considering the recurring nature of this important sub-system, this part of the whole system calls for redesign and replacement; otherwise, the grower will face substantial operational costs. To facilitate redesign, the most influential factors and the weak points have to be identified. This requires a comprehensive and systematic analysis because, in the case of CPGHs, failure can be caused by disruption of hardware, malfunction of software, the loss of cyber content, or all of these together. •

Assure continuity of system operation despite the presence of faults One widely-used approach to assure the continuity of operation is the building of various types of redundancies into a system. Sub-system or component multiplication is seen as an effective approach to increasing the dependability of mission critical CPGHs. For example, the use of more than one fertilizer injector machine in an irrigation process can guarantee the availability of fertilizers even if one of them breaks down or malfunctions. If this principle is considered in the design process of CPGHs, the foreseeable operational deficiencies can be eliminated, or the number of their occurrence can be reduced. This principle also entails taking measures to make sure that a failure in the sub-system will not affect other sub-systems. For instance, failure of the reasoning engine of the irrigation sub-system will not influence the performance of the rest of the CPGH system if each of its intelligent sub-systems has a reasoning engine on its own. • Detect and isolate faults This principle can be applied straightaway in the case of CPGH systems. Owing to their component-based implementation, it can simply be the identification of which components have broken down, or are not working properly. Component-based implementation of CPGHs also facilitates the isolation of erroneous components and helps sustain the operation of the rest of the systems. For instance, the behaviour of a sensor that measures the temperature of plants can be tested by making control measures, by comparing the temperature needed locally in the greenhouse

to the measurements taken on close to the plants. If the differences are above the margin of error, it can be concluded that a sensor fault is in development. These measurements should be incorporated in the troubleshooting algorithms of CPGHs and in any other post-processing procedures. • Alerting the operator in case of fault CPSs allow both direct (co-located) and indirect (dislocated or remote) interaction with users (both sub-systems and humans) through dedicated interfaces. Communication with external agent sub-systems can be used for alerting and requiring intervention beyond the level of reliability that is it usually achievable with human users and supervisors in the case of LCSs. For instance, automatically sending a message to a supervisory agent sub-system as well as to the greenhouse operator using mobile devices such as tablets and smartphones can shorten the reaction time, and may lead to more knowledgeintensive decision making. This duality in alerting is particularly necessary if any failure occurs that cannot be managed by the system. 7.3 Principles that Require Adaptation • Schedule maintenance actions CPGHs are dynamic systems subjected to unpredictable situations. A situation may have various influence on the life cycle of the involved individual physical components. This differs from the way of operation and from the operational situations that are typical in the maintenance of LCSs. The dynamic (taskand environment-influenced) operation of CPGHs makes the planning and execution of systematic maintenance somewhat difficult. Meanwhile, the increased opportunity of sensor- and smart reasoningbased automated monitoring makes it possible to combine the principle of scheduled maintenance with comprehensive, continuous monitoring. Efficient use of sensors in measuring the most informative parameters of plants can lead to a context-sensitive surveillance and the control of the CPGH system. Let us take the example of using artificial light. If lighting components are used less often, this can be taken into consideration in their scheduled replacement and, in addition, the planned visual/instrumented checking activities can also be done less frequently. Avoid failures in the system by preventing the occurrence of faults In the case of LCSs, the operating conditions are usually known in the design stage. The system

•

Towards the Maintenance Principles of Cyber-Physical Systems

827

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 815-831

operation and behaviour can be pre-tested based on virtual or testable physical prototypes. However, this is hardly possible in the case of CPGHs working in unforeseeable dynamic circumstances. If the deficiencies in behaviour cannot be explored and eliminated through prototype testing, then the objective is to prevent the occurrence of faults. To this end, the development of new testing approaches that are able to evaluate the capability of the subsystems in unexpected situations, considering the negotiation processes between the subsystems as well, is necessary. For instance, if the automated cooling/heating sub-system decides that natural airing is to be done (i.e. windows should be opened) due to a sudden change in the climate situation in the greenhouse (plant’s temperature), this decision may also affect the CO2 regulation but the direction of opening of the vents may not be appropriate. In this particular situation, a negotiation process between the reasoning engines of both systems is required. This means that the negotiation capability of the concerned sub-systems should be tested during the design and prototyping stage. • Reduce the amounts of faults and their severity CPGHs are more complex systems than traditional greenhouses; consequently, many more different possibilities are there for both component failures and system break downs. The number of faults and reducing their severity requires redefinition and reinterpretation of the above maintenance principle. While the overall goal should be kept, the way of achieving it should be adapted to the complexity of CPGH systems and the multitude of functional interactions among the components. Consider the fact that the natural horticultural system constituents (such as plants and climate, CO2, humidity, and lighting sub-systems) and the constituents of cyber-physical augmentation (including sensing technologies, reasoning engines, data transmitters, and smart actuators) should be seamlessly blended and operating. Formulation of all the relevant and most appropriate maintenance principles requires further research, in particular if reducing the severity and impacts of the failures is also a major objective. • Predicting failures on the system Typically, the system models currently used in the analysis and operation simulation of LCSs do not take into account the occurrence of the unexpected situations to which CPSs are often subjected. Therefore, the results of these traditional prediction models and software tools may not be entirely reliable 828

in the context of CPSs. In the case of model-based maintenance, sufficiently comprehensive (modelling the dynamics of the CPGH systems as well as the dynamics of the embedding environment) and articulated (covering both the natural horticultural system constituents and the cyber-physical augmentation constituents) prediction models are needed. Such models would capture information about the growth of the plants and their effect on the system performance. In fact, the ultimate objective of developing such kind of prediction tools is to reduce the amount of unexpected situations through a deep analysis of the effects of variations of the internal and external parameters of CPGH systems. Research in this direction is still at its infancy; therefore, these desirable new maintenance principles are not yet known. 8 CONCLUSIONS We have reviewed the maintenance principles currently applied in LCSs with the intention of determining if they are relevant to the maintenance of CPSs. Due to the proliferation of CPSs and their applications, including in mission critical areas, there has been a growing need to analyse how the maintenance of these systems should be conducted and to identify maintenance principles that can be successfully applied to them. CPSs are complicated complex systems, which nevertheless have some similarities with LCSs. For instance, both integrate information technologies into physical devices, are geographically distributed, have multiple energy sources, functional units, and intense interactions with human stakeholders and the embedded environment. High-end CPSs are, however, non-linear systems, which feature a multitude of functional connections among the components, exhibit a high level of automation and intelligence, and are developed to operate in dynamic or harsh environments. There is also a great dissimilarity between their system features. These facts inspired us to analyse which generic maintenance principles of LCSs could be transferred to CPSs. In the work presented in this paper, we identified the four groups of principles presented and discussed in the previous sections. We have argued and explained why certain principles can be applied directly, and why certain principles need adaptation. In our analysis, we established that some features of CPSs cannot be addressed by exportable maintenance principles. Novel maintenance principles should, therefore, be developed for these features. The reported work,

Ruiz-Arenas, S. – Horváth, I. – Mejía-Gutiérrez, R. – Z. Opiyo, E.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 815-831

however, is just the first step, and further research is expected to provide a deeper understanding through more accurate and focused analyses aimed at identifying the appropriate maintenance principles for CPSs. Future study will also include the identification of influential factors and causalities. We should probably not expect to develop overly generic maintenance principles that are equally and broadly applicable to all CPSs, including high-end CPSs that are, for example, capable of reorganizing themselves. The maintainability of high-end CPSs depends on multiple external factors that dynamically influence their operation. The aforementioned examples illustrate how the known maintenance principles of LCSs can be considered in the case of CPGH systems. Also indicated is the need for extensive further research as well as for real-world environment-based studies to reveal what new maintenance principles are needed, and how they can be operationalized in future cyber-physical greenhouse systems. Therefore, in addition to defining new maintenance principles, our future research will also concentrate on the development of a complex troubleshooting model and a maintenance advisory system for CPGHs. 9 REFERENCES [1] Chun, I., Kim, J., Kim, W.T., Lee, E. (2011). Selfmanaged system development method for cyberphysical systems. Control and Automation, and Energy System Engineering, vol. 256, p. 191-194, DOI:10.1007/978-3-642-26010-0_23. [2] Colnaric, M., Verber, D., Halang, W. A. (2008). RealTime Characteristics And Safety Of Embedded Systems. Distributed Embedded Control Systems, Springer, London, p. 3-28. [3] Sierla, S., O’Halloran, B.M., Karhela, T., Papakonstantinou, N., Tumer, I.Y. (2013). Common cause failure analysis of cyber–physical systems situated in constructed environments. Research in Engineering Design, vol. 24, no. 4, p 375-394, DOI:10.1007/s00163-013-0156-2. [4] Frazzon, E.M., Hartmann, J., Makuschewitz, T., Scholz-Reiter, B. (2013). Towards socio-cyber-physical systems in production networks. Procedia CIRP, vol. 7, p. 49-54, DOI:10.1016/j.procir.2013.05.009. [5] Parvin, S., Hussain, F.K., Hussain, O.K., Thein, T., Park, J.S. (2012). Multi-cyber framework for availability enhancement of cyber physical systems. Computing, vol. 95, no. 10-11, p 927-948, DOI:10.1007/s00607012-0227-7. [6] Sha, L., Gopalakrishnan, S., Liu, X., Wang, Q. (2009). Cyber-Physical Systems: A New Frontier. Tsai, J.J.P.,

Yu, P.S. (eds.), Machine Learning in Cyber Trust, Springer, p. 3-13, DOI:10.1007/978-0-387-88735-7_1. [7] Yagan, O., Qian, D., Zhang, J., Cochran, D. (2012). Optimal Allocation of Interconnecting Links in Cyber-Physical Systems: Interdependence, Cascading Failures, and Robustness. IEEE Transactions On Parallel And Distributed Systems, vol. 23, no. 9, p. 1708-1721, DOI:10.1109/TPDS.2012.62. [8] Miclea, L., Sanislav, T. (2011). About dependability in cyber-physical systems. 9th East-West Design & Test Symposium, Sevastopol. [9] Érdi, P. (2008). Complex Systems: The Intellectual Landscape. Complexity Explained. Springer, Berlin, Heidelber, p. 1-23. [10] Qian, K., den Haring, D., Cao, L. (2009). Introduction to Embedded Systems. Embedded Software Development, Springer, p. 1-37, DOI:10.1007/978-1-4419-0606-9_1. [11] Marwedel, P. (2011). Introduction. Embedded System Design. Springer Netherlands, Amsterdam, p. 1-19, DOI:10.1007/978-94-007-0257-8_1. [12] Reddy, P.M. (2002). Embedded systems. Resonance, vol. 7, no. 12, p. 20-30, DOI:10.1007/BF02834526. [13] Barolli, L., Takizawa, M., Hussain, F. (2011). Special issue on emerging trends in cyber-physical systems. Journal of Ambient Intelligence and Humanized Computing, vol. 2, no. 4, p. 249-250, DOI:10.1007/ s12652-011-0062-2. [14] Horváth, I., Gerritsen, B. (2012). Cyber-Physical Systems: Concepts, technologies and implementation principles. 9th Tools and Methods of Competitive Engineering, Karlsruhe. [15] Kumara, S., Cui, L., Zhang, J. (2011). Sensors, networks and internet of things: research challenges in health care. 8th International Workshop on Information Integration on the Web, New York. [16] Karsai, G., Sztipanovits, J. (2008). Model-Integrated Development of Cyber-Physical Systems. Brinkschulte, U., Givargis, T., Russo, S. (eds.), Software Technologies for Embedded and Ubiquitous Systems, Springer, Berlin, Heidelberg, p. 46-54, DOI:10.1007/978-3-54087785-1_5. [17] Wu, F.J., Kao, Y.F., Tseng, Y.C. (2011). From wireless sensor networks towards cyber physical systems. Pervasive and Mobile Computing, vol. 7, no. 4, p. 397413, DOI:10.1016/j.pmcj.2011.03.003. [18] Song, Z., Chen, Y., Sastry, C. R., Tas, N.C. (2009). Optimal Observation for Cyber-physical Systems. Springer, London, p. 1-25, DOI:10.1007/978-1-84882656-4_1. [19] Beck, A.C.S., Lisbôa, C.A.L., Carro, L., Nazar, G.L., Pereira, M.M., Ferreira, R.R. (2013). Adaptability: the key for future embedded systems. in Adaptable Embedded Systems, Beck, A.C.S., Lisbôa, C.A.L., Carro, L. (eds.). Springer, New York, p. 1-12, DOI:10.1007/978-1-4614-1746-0_1. [20] Santos, R.M., Santos, J., Orozco, J.D. (2009). Power saving and fault-tolerance in real-time critical embedded

Towards the Maintenance Principles of Cyber-Physical Systems

829

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 815-831

systems. Journal of Systems Architecture, vol. 55, no. 2, p. 90-101, DOI:10.1016/j.sysarc.2008.09.001. [21] Baheti, R., Gill, H. (2011). Cyber-physical systems, Samad, T., Annaswamy. A. (eds.), The Impact of Control Technology, IEEE Control System Society, Munich, p. 161-166. [22] Hu, L., Xie, N., Kuang, Z., Zhao, K. (2012). Review of Cyber-Physical System Architecture. IEEE 15th International Symposium on Object/Component/ Service-Oriented Real-Time Distributed Computing Workshops, p. 25-30, DOI:10.1109/ISORCW.2012.15. [23] Gerritsen, B., Horváth, I. (2012). Current drivers and obstacles of synergy in Cyber-Physical Systems Design. ASME International Design Engineering Technical Conference & Computer and Information in Engineering Conference, Chicago. [24] Guinand, F., Pigné, Y. (2006). Problem Solving and Complex Systems. Aziz-Alaoui, P.M.A., Bertelle, P.C. (eds.), Emergent Properties in Natural and Artificial Dynamical Systems, Understanding Complex Systems. Springer, Berlin, Heidelberg, p. 53-85, DOI:10.1007/3540-34824-7_3. [25] Wang, Y., Tan, G., Wang, Y., Yin, Y. (2012). Perceptual control architecture for cyber–physical systems in traffic incident management. Journal of Systems Architecture, vol. 58, no. 10, p. 398-411, DOI:10.1016/j. sysarc.2012.06.004. [26] Lun, Y., Cheng, L. (2011). The research on the model of the context-aware for reliable sensing and explanation in cyber-physical system, Procedia Engineering, vol. 15, p. 1753-1757. DOI:10.1016/j.proeng.2011.08.327. [27] Talcott, C. (2008). Cyber-physical systems and events. Wirsing, M., Banatre, J.-P., Hölzl, M., Rauschmayer, A. (eds.), Software-Intensive Systems and New Computing Paradigms, vol. 5380, Springer, Berlin, Heidelberg, p. 101-115, DOI:10.1007/978-3-540-89437-7_6. [28] Rajkumar, R. Lee, I., Sha, L., Stankovic, J. (2010). Cyber-physical systems: the next computing revolution. 47th Design Automation Conference, New York, DOI:10.1145/1837274.1837461. [29] Li, P., Xu, Q., Wang, N. (2013). Evolution course and analysis of internet of things. International Conference on Information Engineering and Applications, Springer, London, p. 221-228. [30] Perraju, T.S., Uma, G. (2005). Mission Critical Intelligent Systems. Leondes, C.T. (ed.), Intelligent Knowledge-Based Systems. Springer, p. 1184-1210, DOI:10.1007/978-1-4020-7829-3_34. [31] Stehr, M.O., Talcott, C., Rushby, J., Lincoln, P., Kim, M., Cheung, S., Poggio, A. (2011). Fractionated Software for Networked Cyber-Physical Systems: Research Directions and Long-Term Vision. Agha, G., Danvy, O., Meseguer, J. (eds.), Formal Modeling: Actors, Open Systems, Biological Systems, Springer, Berlin, Heidelberg, p. 110-143. [32] Wedde, H.F., Lind, J. A. (1997). Building Large, Complex, Distributed Safety-Critical Operating

830

Systems. Real-Time Systems, vol. 13, no. 3, p. 277-302, DOI:10.1023/A:1007915628098. [33] Ponsard, C., Massonet, P., Rifaut, A., Molderez, J.F., van Lamsweerde, A., Van Tran, H. (2005). Early verification and validation of mission critical systems. Electronic Notes in Theoretical Computer Science, vol. 133, p. 237-254, DOI:10.1007/s10703-006-0028-8. [34] Waeyenbergh, G., Pintelon, L. (2004). Maintenance concept development: A case study. International Journal of Production Economics, vol. 89, no. 3, p. 395-405, DOI:10.1016/j.ijpe.2003.09.008. [35] Moesgaard, M., Froholdt, M., Poulfelt, F. (2010). Return on Strategy: How to Achieve it!, Routledge, New York, London. [36] Stevenson, A. (2010). Oxford Dictionary of English, 3rd ed., Oxford University Press, Oxford. [37] Xie, F., Yang, G., Song, X. (2006). Component-based hardware/software co-verification. 4th ACM and IEEE International Conference on Formal Methods and Models for Co-Design, Proceedings, p. 27-36. [38] Kelly, A. (2006). Strategic Maintenance Planning. Butterworth-Heinemann, Oxford. [39] Starr, A., Al-Najjar, B., Holmberg, K., Jantunen, E., Bellew, J., Albarbar, A. (2010). Maintenance Today and Future Trends. Holmberg, K., Adgar, A., Arnaiz, A., Jantunen, E., Mascolo, J., Mekid, S. (eds.), E-maintenance. Springer, London, p. 5-37, DOI:10.1007/978-1-84996-205-6_2. [40] Burhanuddin, M.A., Halawani, S.M., Ahmad, A.R. (2011). An efficient failure-based maintenance decision support system for smalland medium industries. Jao, C. (ed.), Efficient Decision Support Systems - Practice and Challenges From Current to Future. InTech, Rijeka, p. 195-211. [41] Pintelon L.M., Gelders, L.F. (1992). Maintenance management decision making. European Journal of operational Research, vol. 58, no. 3, p. 301-317, DOI:10.1016/0377-2217(92)90062-E. [42] Zhou, X., Xi, L., Lee, J. (2009). Opportunistic preventive maintenance scheduling for a multiunit series system based on dynamic programming. International Journal of Production Economics, vol. 118, no. 2, p. 361-366, DOI:10.1016/j.ijpe.2008.09.012. [43] Pintelon L., Muchiri, P.N. (2009). Safety and maintenance. Ben-Daya, M., Duffuaa, S.O., Raouf, A., Knezevic, J., Ait-Kadi, D. (eds.), Handbook of Maintenance Management and Engineering, Springer, London, p. 613-648, DOI:10.1007/978-1-84882-4720_22. [44] Labib A.W., Yuniarto, M.N. (2009). Maintenance strategies for changeable manufacturing. ElMaraghy, H.A. (ed.), Changeable and Reconfigurable Manufacturing Systems. Springer, London, p. 337-351, DOI:10.1007/978-1-84882-067-8_19. [45] Johnson, D.M. (1996). A review of fault management techniques used in safety-critical avionic systems. Progress in Aerospace Sciences, vol. 32, no. 5, p. 415431. DOI:10.1016/0376-0421(96)82785-0.

Ruiz-Arenas, S. – Horváth, I. – Mejía-Gutiérrez, R. – Z. Opiyo, E.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 815-831

[46] Xu, S. (1999). On dependability of computing systems. Journal of Computer Science and Technology, vol. 14, no. 2, p. 116-128, DOI:10.1007/BF02946517. [47] Dubrova, E. (2013). Fundamentals of Dependability. Fault-Tolerant Design, Springer, New York, p. 5-20, DOI:10.1007/978-1-4614-2113-9_2. [48] Avižienis, A., Laprie, J.C., Randell, B. (2004). Dependability and its threats: a taxonomy. Jacquart, R. (ed.), Building the Information Society, Springer, New York, p. 91-120, DOI:10.1007/978-1-4020-8157-6_13. [49] Jhumka, A. (2010). Dependability in service-oriented computing. Griffiths, N., Chao, K.M. (eds.), AgentBased Service-Oriented Computing. Springer, London, p. 141-160, DOI:10.1007/978-1-84996-041-0_6. [50] Kontoleon, J. (2008). Status and trends in the performance assessment of fault tolerant systems. Misra, P.K.B. (ed.), Handbook of Performability Engineering, Springer, London, p. 1087-1106, DOI:10.1007/978-1-84800-131-2_66. [51] Birman K.P., Glade, B.B. (1995). Reliability through consistency. IEEE Software, vol. 12, no. 3, p. 29-41. DOI:10.1109/52.382182. [52] Verma, A. Srividya, A. Ramesh, P. (2010). A systemic approach to integrated e-maintenance of large engineering plants. International Journal of Automation and Computing, vol. 7, no. 2, p. 173-179, DOI:10.1007/ s11633-010-0173-9. [53] Sobh, T.S., Mostafa, W.M. (2011). A cooperative immunological approach for detecting network anomaly. Applied Soft Computing, vol. 11, no. 1, p. 1275-1283, DOI: 10.1016/j.asoc.2010.03.004. [54] La, H.J., Kim, S.D. (2010). A Service-Based Approach to Designing Cyber Physical Systems. IEEE/ACIS 9th International Conference on Computer and Information Science, p. 895-900. [55] Ben-Daya, M. (2009). Failure mode and effect analysis. Ben-Daya, M.S., Duffuaa, O., Raouf, A., Knezevic, J., Ait-Kadi, D. (eds.), Handbook of Maintenance Management and Engineering, Springer, London, p. 75-90, DOI:10.1007/978-1-84882-472-0_4. [56] Čepin, M. (2011). Fault Tree Analysis. Assessment of Power System Reliability. Springer, London, p. 61-87, DOI:10.1007/978-0-85729-688-7_5. [57] Stapelberg, R.F. (2009). Safety and risk in engineering design. Handbook of Reliability, Availability, Maintainability and Safety in Engineering Design. Springer, London, p. 529-798.

[58] Xu, T., Liu, Z., Tang, T., Zheng, W., Zhao, L. (2012). Component based design of fault tolerant devices in cyber physical system. 15th IEEE International Symposium on Object/Component/Service-Oriented Real-Time Distributed Computing Workshops- p. 3742, DOI:10.1109/ISORCW.2012.17. [59] Garg, H., Sharma, S.P. (2012). Stochastic behavior analysis of complex repairable industrial systems utilizing uncertain data. ISA Transactions, vol. 51, no. 6, p. 752-762, DOI:10.1016/j.isatra.2012.06.012. [60] Bencsik, A.L., Gati, J., Kartyas, G. (2012). Maintenance of complex automated systems. 7th IEEE International Symposium on Applied Computational Intelligence and Informatics, p. 223-227. [61] Chen, C., Brown, D., Sconyers, C., Zhang, B., Vachtsevanos, G., Orchard, M.E. (2012). An integrated architecture for fault diagnosis and failure prognosis of complex engineering systems. Expert Systems with Applications, vol. 39, no. 10, p. 9031-9040, DOI:10.1016/j.eswa.2012.02.050. [62] Ching ,W.K., Ng, M.K. (2006). Markov Chains: Models, Algorithms and Applications. Springer, New Yourk. [63] Narahari, Y. (1996). Petri nets. Resonance, vol. 4, no. 8, p. 58-69, DOI:10.1007/BF02837068. [64] Cao, R., Chen, Y., Kang, R. (2012). Critical review of system failure behavior analysis method. IEEE Conference on Prognostics and System Health Management, p. 1-10. [65] Pinjala, S.K., Pintelon, L., Vereecke, A. (2006). An empirical investigation on the relationship between business and maintenance strategies. International Journal of Production Economics, vol. 104, no. 1, p. 214-229, DOI:10.1016/j.ijpe.2004.12.024. [66] McDermid, J., Thomas, M., Redmill, F. (2009). Professional issues in system safety engineering. Dale, C., Anderson, T. (eds.) Safety-Critical Systems: Problems, Process and Practice. Springer, London, p. 135-145. [67] Hanemann, A., Sailer, M., Schmitz, D. (2004). Assured service quality by improved fault management. 2nd International Conference on Service Oriented Computing, New York, p. 183-192. [68] Lee, J., Ghaffari, M., Elmeligy, S. (2011). Selfmaintenance and engineering immune systems: Towards smarter machines and manufacturing systems. Annual Reviews in Control, vol. 35, no. 1, p. 111-122. DOI:10.1016/j.arcontrol.2011.03.007.

Towards the Maintenance Principles of Cyber-Physical Systems

831

Short Scientific Paper

Received for review: 2014-04-04 Received revised form: 2014-09-08 Accepted for publication: 2014-09-25

Evaluation of the Influence of Upset Stage on Joint Properties of Friction Welded Dissimilar Aluminum-Copper Cast Alloys Barlas, Z. – Çolak, M. Zafer Barlas* – Murat Çolak

Sakarya University, Technology Faculty, Turkey Friction welding was applied to dissimilar aluminum cast bars alloyed with 2 and 8% copper and the effects of upset time on joint properties were evaluated in this study. The welding process was carried out by a vertical milling machine. Tensile test results show that the ultimate tensile strength increased with an increase in upset time and it reached 88% for Al-2%Cu and 74% for Al-8%Cu base metals. The heat affected zone, thermo-mechanical affected zone, and weld metal were detected in this joint by optical examination, in addition to the base metals. A peak temperature of 436 °C was measured in the weld zone. Hardness values also varied according to microstructural changes. The highest hardness values were detected in the weld metal and its average hardness was 217 Brinell. Keywords: friction welding, Al-Cu cast alloy, upset stage, tensile strength, microstructure, weld temperature

0 INTRODUCTION Aluminum-copper (Al-Cu) cast alloys (2XX.X series) are widely used in various industrial applications due to their high strength and hardness properties in the ascast state in spite of their low corrosion resistance, low fluidity and ductility. However, joining of Al-Cu cast alloys by conventional welding processes is difficult or not recommended because of its susceptibility to stress-corrosion cracking and hot cracks [1] to [3]. A literature survey shows that there is also a lack of research on the weldability of Al-Cu cast alloys by solid-state welding techniques. Friction welding (FW) is one kind of solid-state welding process where the heat required for joining is produced by mechanical friction at the interface of the work-pieces. The work-pieces to be joined are first prepared so as to have smooth, square-cut surfaces. While one of the work-pieces remains stationary, the other is rotated against the first piece at high speed under applied pressure. As soon as the interface temperature has reached an optimum value for plastic deformation, the rotation is stopped and the forging pressure is further increased to complete the weld [4] to [6]. Some process parameters have significant effects on producing sound welds. These are: (i) rotational speed, (ii) friction pressure, (iii) friction time, (iv) forging (upset) pressure, and (v) forging time. The other parameters such as feed rate, upset delay time, and brake delay time should be considered for a good weld [7] to [9]. In addition to similar metals, FW is also suitable for welding dissimilar metals with different melting points and physical properties. Therefore, many papers have been reported about the effects of these FW parameters on the weld quality in joining of similar and dissimilar 832

metal groups. According to a study reported by Rafi et al. [10], relatively high friction pressure, high rotation speed and lower forging pressure should be used to create high joint strength in AA7075-T6 aluminum alloy. Sathiya et al. [11] looked at the effects of FW parameters in joining ferritic stainless steel pieces. They found out that friction time should be kept as short as possible, while friction and forging pressures should be as high as possible in order to obtain high joint strength. For FW in AZ31 magnesium alloys, higher upset pressure was caused the grain refinement and increasing hardness near the weld interface [12]. Kurt et al. [5] focused on friction welded dissimilar AISI 1010 mild steel to ASTM B22 copper bronze joints properties at various friction pressures and upset pressures, as well as upset time, under a constant rotation speed and friction time conditions. The upset and friction pressures and times are key parameters according to the authors. They observed that the tensile strengths of the joints increased up to a certain value with an increase in friction pressure and that the hardness generally also increased with increasing friction and upset pressures, but that hardness values decreased with increasing upset time. A similar study about dissimilar 6063 Al alloy/ austenitic stainless steel was carried out by Sammaiah et al. [13]. They recommended low friction pressure and high upset pressure in order to obtain high joint strength. Manideep and Balachandar [14] examined the microstructure and hardness distribution of FW parameters for joining of dissimilar stainless steels. They explained that high upset pressure resulted in a fine grain structure in HAZ and high hardness, while higher friction pressure leads to grain coarsening for FW of AISI 321 to AISI 430 stainless steel. Another study about friction welding of 6061-T6 aluminum

*Corr. Author’s Address: Sakarya University,Esentepe Campus,Serdivan,Sakarya,Turkey, barlas@sakarya.edu.tr

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 832-837

and AISI 1018 steel by Taban et al. [15] revealed that higher forge pressure led to higher tensile strength. Sahin [16] studied the effect of friction stage on dissimilar Al/Cu joint strength and observed that if friction time and pressure are increased, tensile strength increases up to a certain peak point, then decreases with higher friction time and pressure. Similar results were presented by Ratkovic et al. [17] in a study on FW of dissimilar Al and Cu. They found that the tensile strength of the joints increases up to a certain peak value and then slightly decreases with higher friction time. The main purpose of the present study is to evaluate the effects of upset stage including pressure and time in friction welded Al cast alloy bars with contents of 2 and 8% Cu by a vertical milling machine. 1 MATERIALS AND METHOD In the study, dissimilar Al cast alloy bars containing of 2 and 8% Cu (in wt.) in dimensions of 20 mm (diameter) × 90 mm (length) were joined by friction welding. Alloys were poured into a sand mold. Casting alloys were melted in a SiC crucible using a resistance melting furnace with an 8 kg capacity. The chemical compositions of the Al alloys used in the experimental studies are shown in Table 1. The friction welding trials were performed using a vertical milling machine. Al alloy bar with 8% Cu (Al-8%Cu) was rotated (rotation side, RS) while the Al alloy bar with 2% Cu (Al-2%Cu) was fixed (stationary side, SS) via a clamp on the worktable equipped with load-cells. Table 2 shows the welding parameters used. As shown in Table 2, the effects of upset stage (pressure and time) have been studied. Upset forces of 7.5 and 10 kN and upset times of 20, 50, and 80 s were employed under a constant friction force of 2.5 kN, friction time of 60 s and friction speed of 1500 rpm. These applied forces were divided into the cross-sectional area of the aluminum bar thus these values were presented as the friction pressure and upset pressure, respectively. Increasing temperature during the welding processes was measured 3 mm below the abutted surface and at a depth of 2.5 mm from the outer surface of the Al-2%Cu bar using a K-type thermocouple. Tensile testing was carried out on a Shimadzu tester (model AG-IC) with a 50 kN capacity at a cross-head speed of 2 mm×min–1 in order to evaluate the friction weld trials and base alloys strength. The geometry of the tensile test specimens is shown in Fig. 1 [18]. Brinell and Vickers hardness test methods were carried out in order to observe the hardness distribution and features of the joint having the highest ultimate tensile strength

(UTS) value. The Brinell hardness test was performed using a 2.5 mm diameter ball and a 612.9 N indentation force on across the cross-section of the weld zone and near the weld interface. A load of 200 g and dwell time of 10 s were employed in the Vickers test. The crosssection perpendicular to the weld interface of the weld zones was polished and then etched with Keller’s reagent (1.5 ml hydrochloric acid, 2.5 ml nitric acid, 1 ml hydrofluoric acid and 95 ml water). A Nikon Eclipse L150A optical microscope with computer assisted image analysis (Clemex Vision Lite) was used for microstructural examination. A scanning electron microscope (SEM) equipped with an energy dispersive X-ray spectroscopy (EDS) apparatus and X-ray diffraction (XRD) analysis were also used on the weld zone and tensile fracture surface of the joint having the highest UTS value. Table 1. Chemical composition of Al alloys (wt.%) Alloy Al-2%Cu Al-8%Cu

Si 0.11 0.13

Fe 0.21 0.12

Cu 2.04 8.11

Mn 0.023 0.021

Mg 0.012 0.011

Al bal. bal.

Table 2. Experimental parameters and tensile test results for joining of dissimilar Al alloys by a friction welding process Sample S1 S2 S3 S4 S5 S6

Upset pressure [MPa] 24 24 24 32 32 32

Upset time [s] 20 50 80 20 50 80

Joint efficiency [%]

UTS [MPa]

Al-2%Cu

Al-8%Cu

58 83 94 115 121 142

36 51 58 71 75 88

30 43 49 60 63 74

Notes: UTS of Al-2%Cu and Al-8%Cu base metals are 162 and 193 MPa, respectively. The joint efficiency was calculated by dividing the tensile strength of the welded sample by that of the base metals.

Fig. 1. Geometry of tensile test specimen (units in mm)

2 RESULTS AND DISCUSSION The overviews and cross-section images of the friction welded S1 having the lowest UTS and S6 having the highest UTS are shown in Fig. 2. Firstly, the weld trials exhibited weld flash due to the upset

Evaluation of the Influence of Upset Stage on Joint Properties of Friction Welded Dissimilar Aluminum-Copper Cast Alloys

833

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 832-837

pressure and elevated temperature. But the flashes and the axial shortening occurred relatively more in the Al-2%Cu alloy, because it is softer and has a higher thermal conductivity than the Al-8%Cu alloy [5], [10] and [19]. Using a higher upset pressure and time also increased the weld flash amount and caused axial shortening. According to the macro-images for these samples in Figs. 2c and d, FW trials displayed distinguishable weld zone appearances. A linearlike weld interface occurred at the lowest upset time and pressure condition (Fig. 2c), whereas an arc-like weld interface was seen in more plasticized Al-2%Cu at the highest upset time and pressure (S6) caused by its different mechanical and thermal properties (Fig. 2d). The weld metal area increased with higher upset pressure and time and a symmetrical weld metal occurred in S6 taking into account the axes of the bars. It is clearly seen that inadequate upset time and pressure resulted in a lack of bonding through the abutted surfaces for S1 (Fig. 2c). However, S6 exhibited better bonding despite the local lack of bonding defects indicated by the arrows in Fig. 2d.

Fig. 2. Appearances and cross-sectional views a) and c) for S1 and b) and d) for S6

The UTS values of dissimilar friction weld trials dependent on upset pressure and time are given in Table 2. As can be clearly seen, the values increased with increasing upset pressure and time. The better bonding was obtained by a higher upset pressure and time. On the other hand, the upset pressure is more 834

influential than the upset time. The highest UTS value (142 MPa) can be achieved at an upset pressure of 32 MPa and an upset time of 80 s, represented as S6 in Table 2. S6 has a joint efficiency of about 88 and 74% compared to Al-2%Cu and Al-8%Cu base metals, respectively. The fracture for all tensile samples occurred close to the weld interface in the study. Fig. 3 presents SEM image and EDS analysis results, and Fig. 4 shows XRD results obtained from the fractured surface of S6. The SEM micrograph revealed that the brittle mode of fracture primarily took place in the weld zone. This result is akin to that of the as-casted Al-8%Cu base metal. EDS analysis and XRD results showed the presence of a brittle Al2Cu phase in the fractured surface. Therefore, it can be said that the brittle fracture of intermetallic Al2Cu dominated the fracture behavior of the FW joint. Fig. 3c revealed a micro-crack between the Al and relatively coarse Cu-rich particles in the WM. It is believed that the tensile test properties were negatively affected by the presence of a crack, and the lack of bonding defect at the outer periphery of the weld interface (Fig. 2d), in addition to the effect of presence of Al2Cu phase. At the same time, we think that the lack of bonding near the weld flash (see Fig. 2d) had no effect on these properties, because this zone was lathed when preparing the tensile test samples. The peak temperature reached 436 °C according to the measurement taken 3 mm below the abutted surface and at a depth of 2.5 mm from the outer surface of the Al-2%Cu bar by the K-type thermocouple. This temperature is below the eutectic temperature and melting point of aluminum, but it should be taken into account where the temperature of 436 °C was measured as explained above. Thus, the peak temperature may be slightly higher toward the mid-weld interface. Microstructures of the as-casted Al alloys are given Figs. 5a and d, respectively. The microstructures of the base metals (BMs) formed by the dendritic structure of the α-Al solid solution and Al2Cu eutectic mixture in the inter-dendritic regions exhibit a net-like feature. In addition, the different copper content led to the different microstructure formation in the BMs [2]. The average grain sizes of Al-8%Cu BM and Al2%Cu BM were about 82 and 108 mm, respectively. Some pores were also shown in the base metals under microstructural examination. The FW process under an upset pressure of 32 MPa and upset time of 80 s led to the occurrence of distinct microstructural zones, i.e. the heat affected zones (HAZ), the thermomechanical affected zones (TMAZ) and the weld metal (WM), in addition to the BM. HAZ at both

Barlas, Z. – Çolak, M.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 832-837

Intensity [counts]

Fig. 3. a) Fracture surface micrograph, b) EDS analysis results of marked by plus signs on the fracture surface, and c) image of a crack along interface between Al and Cu-rich particles in the WM

Two-Theta [deg]

Fig. 4. X-ray diffraction pattern of the fractured surface for S6

side of the weld interface were characterized by a dendrite growth and the average grain size of HAZs

for Al-8%Cu and Al-2%Cu were about 220 and 174 mm, respectively. A reduction in the voids in the inter-

Evaluation of the Influence of Upset Stage on Joint Properties of Friction Welded Dissimilar Aluminum-Copper Cast Alloys

835

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 832-837

The macrohardness of as-casted Al-2%Cu and Al-8%Cu is 46 HB and 95 HB, respectively. Fig. 6 exhibits macrohardness distribution in the weld zone for S6. It is obvious that the hardnesses increased in the weld zone. As seen in Fig. 6, the hardness trend of RS is also higher than that of SS as a result of the increasing Cu ratio. The highest macrohardness value, which is 220 HB, was measured in WM and this zone has an average hardness of 217 HB across the axis of the friction welded bar. The hardness was also measured near the weld interface and it was observed that the macrohardness changed from 210 to 220 HB. Increasing the macrohardness can be mainly attributed to the grain refinement in WM. Vickers microhardness measurements in the weld zone for S6 revealed an unhomogeneous distribution in contrast to the Brinell macrohardness test results, as shown in Fig. 6. Since the hardness values depended on the indenter location, making a clear definition in terms of hardness feature is relatively difficult for the weld zones. Therefore, several measurements were done at different locations apart from the distribution given in Fig. 6. According to the results, the main factor determining the microhardness value was the Al2Cu eutectic phase in addition to the grain refinement effect. That is to say, having more Al2Cu surrounding the indenter caused an increase in microhardness values (measured up to about 150 HV in WM). Conversely, if the indenter was located at the α-Al phase, the microhardness dropped down to about 60 HV. Moreover, it is believed that the existence of different Cu particle size and pores have effects on the hardness tallies with our observations.

Macrohardness [HB]

Microhardness [HV0.2]

dendrites was observed in HAZs (Figs. 5b and e) due to elevated temperature and upset pressure and taking into account BM features. Aluminum dendrites were elongated in an upward and downward flow within both TMAZ due to plastic deformation during the FW process (Figs. 5e and f). However, again, the different copper content led to a different TMAZ formation. It is observed that the deformed grains at the RS were distinctly finer than those on the SS. WM (bright region in Fig.2d) showed extremely fine grains and dispersed copper-rich particles (Figs. 5f and g). That is, after the FW process in this zone, a net-like phase and elongated grains transformed to copper-rich particles and fine grains. These features of the weld zone consequently led to a joint efficiency of about 74 to 88 percent with regard to BMs despite the lack of bonding defect at the weld interface.

Distance of the weld center [mm]

Fig. 6. Hardness profiles across the cross-section for S6

3 CONCLUSIONS

Fig. 5. Micrographs of the weld zone for S6 a) BM for Al-8%Cu, b) HAZ at RS, c) TMAZ and WM at RS, d) BM for Al-2%Cu, e) HAZ at SS, f) TMAZ at SS, and g) central region of WM

836

Al cast alloy bars with contents of 2 and 8% Cu were joined using a vertical milling machine by friction welding. The upset pressure is a more important parameter than the upset time for the joining of dissimilar Al-Cu cast alloys. The UTS of 142 MPa

Barlas, Z. – Çolak, M.

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, 832-837

can be achieved at an upset pressure of 32 MPa and upset time of 80 s. This value is lower by 12 and 26% than that of the Al-2%Cu and Al-8%Cu base metals, respectively. Typical microstructural zones were observed in the welding zone having the highest UTS. Hardness features in the weld zone were dominated by existence of an Al2Cu intermetallic phase and grain refinement. 4 NOMENCLATURE FW RS SS BM HAZ TMAZ WM UTS HB HV SEM EDS XRD

Friction welding Rotation side Stationary side Base metal Heat affected zone Thermo-mechanical affected zone Weld metal Ultimate tensile strength Hardness Brinell Hardness Vickers Scanning electron microscope Energy dispersive X-ray spectroscopy X-ray diffraction 5 REFERENCES

[1] Kaufman, J.G., Rooy, E.L. (2004). Aluminum Alloy Castings Properties Processes and Applications. ASM International, Materials Park. [2] Zlaticanin, B., Radonjic B., Filipovic, M. (2004). Characterization of Structure and Properties of As-cast AlCuMg Alloys. Materials Transactions, vol. 45, no. 2, p. 440-446, DOI:10.2320/matertrans.45.440. [3] Kearns, W.H. (Ed.) (1982). Welding Handbook, Metals and Their Weldability. American Welding Society, Miami. [4] Black, J.T., Kohser, R.A. (2008). DeGarmo’s Materials and Processes in Manufacturing. John Wiley & Sons, Hoboken. [5] Kurt, A., Uygur, I., Paylasan, U. (2011). Effect of Friction Welding Parameters on Mechanical and Microstructural Properties of Dissimilar AISI 1010 ASTM B22 Joints. Welding Journal, vol. 90, no. 5, p. 102-106. [6] Liu, W., Wang, F., Yang, X., Li, W. (2013). Upset Prediction in Friction Welding Using Radial Basis Function Neural Network. Advances in Materials Science and Engineering, vol. 2013, p. 1-9, DOI:10.1155/2013/196382. [7] Murthy, K.N., Raghupathy, V.P., Sethuram D. (eds.) (2011). Two Day Workshop on Friction Welding & Friction Stir Welding, Indian Welding Society, Kolkata. [8] Rich, T., Roberts, R. (1971). The Forge Phase of Friction Welding. Welding Journal, p. 137-145.

[9] Yilbas, B.S., Sahin, A.Z., Kahraman N., Al-Garni, A.Z. (1995). Friction Welding of St-Al and Al-Cu Materials. Journal of Materials Processing Technology, vol. 49, no. 3-4, p. 431-443, DOI:10.1016/09240136(94)01349-6. [10] Rafi, H.K., Ram, G.D.J., Phanikumar, G., Rao, K.P. (2010). Microstructure and Tensile Properties of Friction Welded Aluminum Alloy AA7075-T6. Materials and Design, vol. 31, no. 5, p. 2375-2380, DOI:10.1016/j.matdes.2009.11.065. [11] Sathiya, P., Aravindan, S., Noorul Haq, A. (2007). Effect of friction welding parameters on mechanical and metallurgical properties of ferritic stainless steel. International Journal of Advanced Manufacturing Technology, vol. 31, no. 11, p. 1076-1082, DOI:10.1007/s00170-005-0285-5. [12] Fukumoto, S., Tanaka, S., Ono, T., Tsubakino, H., Tomita, T., Aritoshi, M., Okita, K. (2006). Microstructural Development in Friction Welded AZ31 Magnesium Alloy. Materials Transactions, vol. 47, no. 4, p. 1071-1076, DOI:10.2320/matertrans.47.1071. [13] Sammaiah, P., Suresh, A., Tagore, G.R.N. (2010). Mechanical properties of friction welded 6063 aluminum alloy and austenitic stainless steel. Journal of Materials Science, vol. 45, no. 20, p. 5512-5521, DOI:10.1007/s10853-010-4609-y. [14] Manideep D., Balachandar, K. (2012). Welding Parameters-Metallurgical Properties Correlation of Friction Welding of Austenitic Stainless Steel and Ferritic Stainless Steel. Journal of Applied Sciences, vol. 12, no. 10, p. 1013-1019, DOI:10.3923/ jas.2012.1013.1019. [15] Taban, E., Gould, J.E., Lippold, J.C. (2010). Dissimilar Friction Welding of 6061-T6 Aluminum and AISI 1018 Steel: Properties and Microstructural Characterization. Materials and Design, vol. 31, no. 5, p. 2305-2311, DOI:10.1016/j.matdes.2009.12.010. [16] Sahin, M. (2010). Joining of Aluminium and Copper Materials with Friction Welding. International Journal of Advanced Manufacturing Technology, vol. 49, p. 527-534, DOI:10.1007/s00170-009-2443-7. [17] Ratković, N., Sedmak, A., Jovanović, M., Lazić, V., Nikolić, R., Krstić, B. (2009). Quality analysis of AlCu joint realized by friction welding. Tehnički Vjesnik - Technical Gazette, vol. 16, no. 3, p. 3-7. [18] Ochi, H., Yamamoto, Y., Yamazaki, T., Sawai, T., Kawai G., Ogawa, K. (2008). Evaluation of tensile strength and fatigue strength of commercial pure Aluminum / tough pitch copper friction-welded joints by deformation heat input. Materials Transactions, vol. 49, no. 12, p. 2786-2791, DOI:10.2320/matertrans.LMRA2008836. [19] Ho, C.Y., Ackerman, M.W., Wu, K.Y., Oh, S.G., Havill, T.N. (1978). Thermal Conductivity of Ten Selected Binary Alloy Systems. Journal of Physical and Chemical Reference Data, vol. 7, no. 3, p. 959-1177, DOI:10.1063/1.555583.

Evaluation of the Influence of Upset Stage on Joint Properties of Friction Welded Dissimilar Aluminum-Copper Cast Alloys

837

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12 Vsebina

Vsebina Strojniški vestnik - Journal of Mechanical Engineering letnik 60, (2014), številka 12 Ljubljana, december 2014 ISSN 0039-2480 Izhaja mesečno

Razširjeni povzetki Matjaž Fleisinger, Matej Vesenjak, Matjaž Hriberšek: Analiza s tokom gnane Darrieusove vodne turbine Ivan Dunđerski: Upravljanje lastnosti pospeševanja vozila s programiranjem funkcij za nadzor motornega navora Primož Potočnik, Tomaž Berlec, Alojz Sluga, Edvard Govekar: Hibridno načrtovanje postavitve proizvodnje na osnovi samo-organizacije Nataša Vujica Herzog, Stefano Tonchia: Merjenje stopnje izvedbe konceptov vitkosti v proizvodnji Marek Boryga: Načrtovanje trajektorije orodja na poti z zanko Santiago Ruiz-Arenas, Imre Horváth, Ricardo Mejía-Gutiérrez, Eliab Z. Opiyo: O načelih vzdrževanja kiberfizičnih sistemov Zafer Barlas, Murat Çolak: Vrednotenje vpliva faze povečanja osnega pritiska pri varjenju s trenjem na lastnosti spoja ulitkov iz različnih aluminijevo-bakrovih zlitin

SI 145 SI 146 SI 147 SI 148 SI 149 SI 150 SI 151

Osebne vesti Doktorske disertacije, diplomske naloge

SI 152

V spomin prof. dr. Antonu Kuhlju

SI 156

Prejeto v recenzijo: 2014-02-03 Prejeto popravljeno: 2014-06-28 Odobreno za objavo: 2014-09-19

Analiza s tokom gnane Darrieusove vodne turbine Fleisinger, M. – Vesenjak, M. – Hriberšek, M. Matjaž Fleisinger* – Matej Vesenjak – Matjaž Hriberšek Univerza v Mariboru, Fakulteta za strojništvo, Slovenija

V zadnjih letih se med tehnologije za izkoriščanje vodne energije pogosteje uvrščajo tudi vodi prilagojene tehnologije za izkoriščanje energije vetra, obojim pa je skupno dejstvo, da so postavljene v tok brez zajezitve. Darrieusova turbina je še posebej primerna za izkoriščanje energije vodotokov, saj je njen presek pravokotne oblike, tako lahko že posamezna turbina zajame večji del preseka plitvega kanala, medtem ko običajna propelerska turbina opisuje krožnico in šele v ohišju zajame večji del kvadratnega preseka, zato za tipičen pravokoten kanal potrebujemo več takšnih turbin. Uveljavljeni postopki za napovedovanje zmogljivosti hidrokinetičnih turbin zajemajo različne aerodinamične modele, medtem ko so se metode računalniške dinamike tekočin na tem področju pričele uporabljati šele pred kratkim. Tudi slednje se najpogosteje izvajajo z vnaprej predpisanimi parametri, med katere sodita vrtilna hitrost turbine in hitrost toka, ponavadi pa se kot rezultat opazuje tlačno in hitrostno polje okrog lopatic. V prispevku je predstavljen razvoj računalniških simulacij z novim pristopom s tokom gnane turbine, pri katerem se turbina v simulaciji vrti na podlagi delovanja sil toka na lopatice in zaviralnega momenta nanjo, kar predstavlja veliko bolj realne pogoje, kot so v običajnih simulacijah. S pristopom s tokom gnane turbine je mogoče s posamično simulacijo napovedati celotno obratovalno karakteristiko turbine za določeno hitrost toka. To je izvedeno s spremenljivo obremenitvijo turbine v obliki zavornega momenta, ki počasi narašča, dokler se turbina ne ustavi. Pri tem se vrtilna hitrost turbine od najvišjih vrtljajev na začetku simulacije počasi zmanjšuje do ustavitve, se pravi, da med simulacijo preide celotno območje vrtljajev. Procedura, ki omogoča uporabo pristopa s tokom gnane turbine v programskem paketu Ansys CFX,je model gibanja togega telesa. Ker želimo pri nadaljnjem delu razviti metodologijo, ki bo upoštevala tako pristop s tokom gnane turbine, kot tudi deformacijo lopatic v močno vezani simulaciji interakcije tekočine in strukture, smo razvili postopek za simulacijo s tokom gnane turbine, ki temelji na nestacionarni simulaciji z drsečo mrežo in uporabniško rutino. Takšna simulacija je združljiva s simulacijo interakcije tekočine in strukture. Na tak način bo mogoče izvajati simulacije, pri katerih se turbina vrti zaradi sil, ki delujejo na njene lopatice, hkrati pa bo mogoče upoštevati vpliv deformacije lopatic na delovanje turbine. V takšni simulaciji bo naenkrat lahko zajetih več parametrov, ki v realnih obratovalnih pogojih hkrati vplivajo na turbinske stroje, s čimer se poveča kompleksnost simulacije, hkrati pa zmanjša tveganje, da bi spregledali katerega izmed pomembnih dejavnikov. Na novo razvito simulacijo smo validirali z rezultati iz literature, kjer je bil model turbine preizkušen v umetnem vodnem kanalu, primerjali pa smo tudi rezultate simulacije z modelom gibanja togega telesa. Pri tem so odstopanja v rezultatih predvsem posledica dejstva, da so bile meritve turbin na preizkusu izvedene po korakih, pri čemer so na posamezni stopnji obremenitve, kot rezultat zajeli povprečno vrednost 30 s dolgega intervala, medtem ko smo v simulaciji obremenitev neprestano počasi zviševali. Dodaten vpliv je imela geometrija rotorja turbine, pri katerem znaša dolžina lopatic 2/3 premera rotorja, kar pomeni, da je zmanjšan vpliv aerodinamičnega profila, povečan pa je vpliv podporne konstrukcije. Pri razviti simulaciji smo opazili tudi nestabilnost v začetni fazi simulacije, ko prihaja do velikih sprememb v vrtilni hitrosti turbine in posledično navoru na lopatice, ki pa se kmalu ustalijo in simulacija poteka stabilno do konca, podobno kot z uporabo modela gibanja togega telesa. Ključne besede: hidrokinetične turbine, računalniška dinamika tekočin, pristop s tokom gnane turbine, model gibanja togega telesa

*Naslov avtorja za dopisovanje: Univerza v Mariboru, Fakulteta za strojništvo, Smetanova ulica 17, 2000 Maribor, Slovenija, matjaz.fleisinger@uni-mb.si

SI 145

Prejeto v recenzijo: 2014-04-08 Prejeto popravljeno: 2014-06-17 Odobreno za objavo: 2014-07-08

Upravljanje lastnosti pospeševanja vozila s programiranjem funkcij za nadzor motornega navora Dunđerski, I. Ivan Dunđerski

Visoka šola za elektrotehniko in računalništvo, Beograd, Srbija

Namen članka je izboljšanje udobja potnikov, ki med vožnjo trpijo za slabostjo (morsko boleznijo). Glavni vzrok slabosti potnikov med vožnjo so nihanja sistema vzmetenja vozila ter spremenljiv pospešek vozila. Spremenljivi pospešek je bil zato pretvorjen v konstanten pospešek na najdaljšem delu razdalje, ki jo vozilo prevozi med pospeševanjem. V raziskavi sta bila postavljena fizikalno-matematični model dinamike pospeševanja vozila in programski model za upravljanje s pospeševanjem vozila. V prvi fazi raziskave so bile opravljene računalniške simulacije skladno s programskim modelom. V drugi fazi raziskave so bili opravljeni eksperimenti z vozili v realnih pogojih. Intenziteta vektorja pospeška vozila se nadzoruje z vlečno silo pogonskih koles vozila. Funkcija pospeševanja vozila upravlja z motornim navorom in tako nadzorovano določa vlečno silo vozila. Del vlečne sile se porabi za premagovanje upora pri gibanju, ki je spremenljiv, preostanek vlečne sile pa pospešuje vozilo in je konstanten. Rezultat je konstantno pospeševanje vozila oz. pretvorba variabilnega pospeševanja vozila v konstantno. Konstanten pospešek odpravi sunke (ki povzročajo slabost) na razdalji, prevoženi med pospeševanjem. Rezultati eksperimentov kažejo, da je lastnosti pospeševanja mogoče upravljati. Z določitvijo pogojev pospeševanja se lahko vpliva na udobje med vožnjo, učinkovitost delovanja motorja, varnost gibanja itd. Nadaljnje raziskave bodo opredelile ostale lastnosti pospeševanja, ki so zanimive za upravljanje, kakor tudi njihove funkcije za vgradnjo v fizikalno-matematični model. Raziskave pospeševanja vozil v objavljenih člankih so bile usmerjene predvsem v porabo goriva. Pospeševanje vozila, ki vpliva na udobje potnikov med vožnjo, ni bilo deležno zadostne pozornosti. Izboljšanje udobja med vožnjo se doseže z ustrezno zasnovo sistema vzmetenja vozila in potniške kabine. Poraba goriva se najpogosteje omejuje z zmanjševanjem motornega navora in odloženim odzivom na nagel pritisk na stopalko za plin. Zmanjšanje porabe goriva je tako doseženo z manj intenzivnim pospeševanjem ter z optimizacijo odmerjanja goriva. Zmanjšanje intenzitete pospeševanja fizikalno ne vpliva na celotno spremembo gibalne količine oz. kinetične energije vozila od začetne do končne hitrosti. Z manj intenzivnim pospeševanjem se zmanjša tudi porast obratov motorja v času in verjetneje je, da bo motor deloval v stacionarnem kot nestacionarnem režimu. Pri toplotnih strojih, ki delujejo z Ottovim ali dizelskim ciklom, se poraba goriva zmanjša zahvaljujoč dejstvu, da je najmanjša prav pri delovanju v stacionarnem režimu. transki učinek zmanjšanja motornega navora, ki omogoča zmanjšanje porabe goriva, je tudi omejitev sunkov. Sunki pa ne izginejo povsem zaradi variabilne razlike med krivuljo motornega navora in navorom upora proti gibanju. Del vlečne sile, ki pospešuje vozilo, je zato variabilen. Članek opisuje možnosti izboljšanja udobja pri vožnji s pretvorbo pospeševanja s sunki v pospeševanje brez sunkov. To pomeni, da ostane naklon gladine tekočine v kozarcu v vozilu glede na smer gibanja vozila med pospeševanjem nespremenljiv. V ta namen je uporabljeno upravljanje lastnosti in intenzitete pospeševanja s programsko kodo za krmiljenje motornega navora. Posebej je bil razvit krmilnik (strojna oprema) za upravljanje s polnitvijo motorja ter pripadajoča programska koda (programska oprema). Z razvojem fizikalno-matematičnega modela je bil uresničen princip modeliranja pospeševanja pri danih lastnostih. Z razvojem programskega modela je bilo uresničeno upravljanje pospeševanja z danimi lastnostmi. Modeli omogočajo tudi vključitev drugih funkcij pospeševanja in njihovih lastnosti, saj tvorijo načelno osnovo za modeliranje lastnosti pospeševanja in programiranje upravljanja pospeševanja. Ključne besede: funkcija pospeševanja vozila, sunki mas, programsko krmiljenje motornega navora

SI 146

*Naslov avtorja za dopisovanje: Visoka šola za elektrotehniko in računalništvo, Vojvode Stepe 283, Beograd, Srbija, ivand@viser.edu.rs

Prejeto v recenzijo: 2014-02-14 Prejeto popravljeno: 2014-05-30 Sprejeto v objavo: 2014-07-22

Hibridno načrtovanje postavitve proizvodnje na osnovi samo-organizacije Potočnik, P. – Berlec, T. – Sluga, A. – Govekar, E. Primož Potočnik* – Tomaž Berlec – Alojz Sluga – Edvard Govekar Univerza v Ljubljani, Fakulteta za strojništvo, Slovenija

Namen raziskave je predstaviti novo hibridno metodo načrtovanja postavitve proizvodnje, ki združuje princip samo-organizacije in ekspertnega načrtovanja. Predlagana metoda rešuje problem načrtovanja postavitve proizvodnje z namenom optimiranja toka materiala med posameznimi proizvodnimi enotami. V literaturi lahko zasledimo algoritmične pristope k načrtovanju postavitve proizvodnje, ki temeljijo na matematični formulaciji problema, vendar težko upoštevajo številne lokalne omejitve, tehnične specifikacije, načine transporta, dimenzije, itd. Alternativni in najpogosteje uporabljeni pristop temelji na uporabi grafičnih programskih paketov za ročno načrtovanje proizvodnih enot. V tem primeru je mogoče natančno upoštevati vse lokalne omejitve, vendar je težko zagotoviti globalno optimirano proizvodno postavitev. Predlagana metoda združuje prednosti algoritmičnega in ročnega načrtovanja proizvodnje. Metoda povezuje princip samo-organizacije za avtomatizirano načrtovanje proizvodnih celic in nato lokalno optimizacijo s strani eksperta za končno postavitev proizvodnje, ki upošteva tudi vse lokalne omejitve, tehnične specifikacije in načine transporta. Za samo-organizirano formiranje heksagonalno razmeščenih proizvodnih celic je uporabljena nevronska mreža tipa SOM (self-organizing map), ki združi produkte s podobnimi lastnostmi in proizvodnimi zahtevami v dvo-dimenzionalno celično strukturo. Tako razmeščene proizvodne celice že predstavljajo izhodiščno globalno optimirano postavitev proizvodnje. V naslednjem koraku poteka ekspertna postavitev proizvodnih enot znotraj vsake celice. Ker vsaka celica vsebuje le manjše število proizvodnih enot, je postavitev enostavno rešljiva z uporabo grafičnega programskega orodja, kjer je mogoče upoštevati in načrtovati tudi vse lokalne omejitve. Predlagana metoda je bila uporabljena na podatkih podjetja KGL d.o.o., ki izdeluje različne izdelke za prvo vgradnjo iz črnih in barvnih kovin, ter že montirane sestavne dele za avtomobilsko industrijo. Podatkovno bazo z opisi 252 produktov, ki so vključevali vse zahtevane proizvodne operacije in tudi lastnosti produkta, smo uporabili za samo-organizirano formiranje proizvodnih celic, ter v naslednjem koraku za ročno grafično finalizacijo postavitve proizvodnje. Za vrednotenje dobljenih rezultatov smo uporabili dvoje mer, in sicer skupno transportno dolžino (TTL) in produkt transportne intenzitete in dolžine poti (ILP). Glede na izhodiščno stanje proizvodnje smo izkazali znatno izboljšanje organizacije proizvodnje, in sicer kar 40% oziroma 42% zmanjšanje TTL in ILP mer učinkovitosti. Predlagana metoda je primerna za manjša in srednje velika podjetja, za katera je značilna individualna in maloserijska proizvodnja s številnimi različnimi proizvodi. Metoda združuje prednosti algoritmičnega in ekspertnega načrtovanja proizvodnje, in je zlasti primerna za reorganizacijo obstoječe proizvodnje zaradi številnih lokalnih omejitev. Metoda učinkovito minimizira tok dela in materiala, in s tem posledično znižuje proizvodne stroške. Ključne besede: načrtovanje postavitve proizvodnje , samo-organizacija, nevronske mreže, hibridna metoda postavitve, celična proizvodnja, optimiranje proizvodnje

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

SI 147

Prejeto v recenzijo: 2014-04-11 Prejeto popravljeno: 2014-09-25 Odobreno za objavo: 2014-10-07

Merjenje stopnje izvedbe konceptov vitkosti v proizvodnji Vujica Herzog, N. – Tonchia, S. Natasa Vujica Herzog1 – Stefano Tonchia2

2Univerza

1Univerza v Mariboru, Fakulteta za strojništvo, Slovenija v Udinah, Oddelek za strojništvo, elektrotehniko, menedžment in mehaniko, Italija

Kljub temu, da velja prepričanje, da so koncepti vitke proizvodnje splošno znani in sprejeti, ostaja na področju razumevanja in terminologije o vitkosti še precej nedorečenosti. V članku so predstavljeni rezultati anketne raziskave, izvedene v 72-ih srednje velikih in velikih podjetjih slovenske kovinsko-predelovalne in elektro-strojne industrije. Izbira vzorca in izbira področij za raziskovanje zagotavlja splošen okvir za preučevanje predstavljene tematike. Na osnovi pregleda literature iz področja vitkosti smo izbrali 8 področij, kritičnih za oceno in merjenje stopnje vitkosti v podjetjih: koncept vrednosti in kupci; prikaz toka vrednosti – (value stream mapping) VSM; koncept vlečenja – pull, kanban in tok proizvodnje; odprava izgub; vzdrževanje; proizvodnja točno ob pravem času – just in time; sodelovanje zaposlenih in razvoj odličnih dobaviteljev. Znotraj teh področij smo z uporabo vprašalnika, zasnovanega po načelih Likertove skale in preverjanja zanesljivosti in veljavnosti izvedene ankete oblikovali spremenljivke za merjenje pokritosti posameznih področij. Npr. za merjenje stopnje razvoja odličnih dobaviteljev so pomembne naslednje veličine: pravočasne dobave, sodelovanje z dobavitelji in usposobljeni in zanesljivi dobavitelji. Za preučevanje predstavljene problematike smo izbrali metodologijo anketne raziskave in določili najpomembnejša področja vitke proizvodnje. Nato smo oblikovali vprašalnik in izvedli pilotsko testiranje. Nejasnosti, ki so se pokazale pri pilotskem testiranju, smo odpravili in dobili končni vprašalnik. Vprašalnik vsebuje 59 trditev, ovrednotenih po Likertovi lestvici. Dobljene rezultate smo preučili iz vidika zanesljivosti in veljavnosti. Ko govorimo o zanesljivosti, mislimo predvsem na stabilnost in doslednost meritve, medtem ko nas pri veljavnosti zanima predvsem ustreznost meritve. Z uporabo statističnega paketa PASW Statistics 18 smo izvedli analizo rezultatov. Vse več-točkovne spremenljivke smo najprej testirali z vidika zanesljivosti (Cronbachov koeficient a) in nadaljevali s projekcijo na prvo glavno komponento. Postopek oblikovanja novih spremenljivk smo ponavljali tako dolgo, dokler nismo dobili končnega niza spremenljivk. Končni niz spremenljivk obsega 23 kazalnikov. Za področje koncept vrednosti in kupci so najpomembnejši kazalniki stopnja zadovoljstva kupca, število reklamacij in sposobnost zaznavanja zahtev kupcev. Spremenljivke pomembne za področje prikaz toka vrednosti – VSM so uporaba oz. prisotnost metod za prikaz toka vrednosti, prepoznava izgub in zniževanje stroškov. Koncept vlečenja – pull, kanban in tok proizvodnje ponazarjajo zgodnje poznavanje zahtev kupca, sodelovanje kupca pri oblikovanju izdelka, fleksibilnost pri izpolnjevanju zahtev kupca, standardizacija in modularnost izdelkov ter ustrezno načrtovanje in vodenje proizvodnje. Koncept odprave izgub, ki velja za bistvo vitke proizvodnje se odraža skozi menedžment zalog in izkoriščenost kapacitet. Za področje vzdrževanja je pomemben celovit sistem vzdrževanja in prvo preverjanje kakovosti. Proizvodnja točno ob pravem času – just in time se kaže kot pravočasne dobave, sodelovanje zaposlenih, manjše število kosov in podatek o tem, kje v procesu proizvodnje je naročilo. Sodelovanje zaposlenih in timsko delo sta spremenljivki, ki sta pomembni za dobro delo zaposlenih. Za razvoj odličnih dobaviteljev so pomembne pravočasne dobave, dobri odnosi z dobavitelji ter izkušeni in zaupanja vredni dobavitelji. Glede na velikost in izbiro ciljne skupine je mogoče rezultate raziskave posplošiti in uporabiti v različnih podjetjih. V prihodnje bi bilo v zanimivo izvesti enako raziskavo v podjetjih drugih držav in primerjati rezultate. Rezultati so pomembni za razumevanje področja vitkosti in se lahko hkrati uporabljajo kot konkreten inštrument menedžmenta za merjenje prisotnosti konceptov vitkosti v proizvodnji. Ključne besede: vitka proizvodnja, merjenje izvedbe, anketne raziskave, veljavnost, zanesljivost, oblikovanje novih spremenljivk

SI 148

*Naslov avtorja za dopisovanje: Univerza v Mariboru, Fakulteta za strojništvo, Smetanova ulica 17, 2000 Maribor, Slovenija, natasa.vujica@um.si

Prejeto v recenzijo: 2014-05-20 Prejeto popravljeno: 2014-08-06 Odobreno za objavo: 2014-09-19

Načrtovanje trajektorije orodja na poti z zanko Boryga, M. Marek Boryga

Naravoslovna univerza v Lublinu, Fakulteta za proizvodno strojništvo, Poljska

Članek podaja nov pristop k načrtovanju gibanja orodja po poti s t. i. ostrim ovinkom. Predstavljeni algoritem samodejno usklajuje dve bistveni zahtevi: glajenje trajektorije in gibanje orodja v ostrem ovinku. Rešitev za istočasno glajenje trajektorije in prehod skozi ovinek je uporaba zanke. Za visokonatančno preslikavo trajektorij in polno izrabo dinamičnih zmogljivosti stroja je treba ustvariti gladke trajektorije z minimalnimi sunki pospeška in hitrosti. Pri načrtovanju gibanja orodja po poti s t. i. ostrim ovinkom se najpogosteje uporablja pristop glajenja ovinka, s katerim se izognemo prehodu orodja skozi oster ovinek. Predlagana pot orodja je sestavljena iz dveh premočrtnih segmentov, ki sta povezana z zanko. Orodje gre skozi vogal dvakrat in časovna razlika med prehodoma je enaka času gibanja po zanki. Profil pospeška na prvem premočrtnem segmentu, ki predstavlja fazo zagona, je v prvem delu opisan s polinomom 7. reda. Profil pospeška v začetni in končni točki tega segmenta je tangenten na časovno os, s čimer so odpravljeni morebitni neželeni sunki v teh točkah. Orodje se premika po prvem premočrtnem segmentu zanke s konstantno hitrostjo, enako pa velja tudi za lok in zadnji premočrtni segment zanke. Profil pospeška na zadnjem premočrtnem segmentu poti je opisan s polinomom 7. reda, enako kot na prvem segmentu. To je druga faza profila oz. faza zaviranja. Največja linearna hitrost v prvem in zadnjem segmentu je obenem tudi hitrost orodja v zanki. Predstavljen je časovni potek pomikov, hitrosti in pospeškov po načrtovani poti gibanja orodja. Rešen je problem inverzne kinematike za manipulatorje antropomorfne, SCARA in kartezične zasnove, na podlagi tega pa so določeni pomiki, hitrosti in pospeški vsakega člena v kinematični verigi. Opravljene so bile simulacije z manipulatorji antropomorfne, SCARA in kartezične zasnove. Rezultati potrjujejo veljavnost predlagane tehnike, ki ima veliko pozitivnih lastnosti. Orodje gre skozi vogal dvakrat, zato je ustvarjeno trajektorijo gibanja mogoče uporabiti v celoti ali po delih. V slednjem primeru je gibanje v zanki mogoče obravnavati kot izhod orodja. Profil pospeška v karakterističnih točkah na trajektoriji je tangenten na časovno os, s čimer so odpravljeni sunki v teh točkah. Algoritem je učinkovit za opravljanje izračunov. Najbolj zamudni so izračuni geometrije trajektorije, medtem ko za določitev profila položaja, hitrosti in pospeškov zadostuje izračun koeficientov polinoma in časa gibanja. Predlagani novi pristop je primeren za izračune in je uporaben za ustvarjanje trajektorij orodij na manipulatorjih in obdelovalnih strojih. Ključne besede: načrtovanje trajektorije, oster ovinek, pot z zanko, polinomski profil pospeška, antropomorfni manipulator, manipulator SCARA, kartezični manipulator

*Naslov avtorja za dopisovanje: Naravoslovna univerza v Lublinu, Fakulteta za proizvodno strojništvo ul. Głęboka 28, 20-612 Lublin, Poljska, marek.boryga@up.lublin.pl

SI 149

Prejeto v recenzijo: 2013-11-20 Prejeto popravljeno: 2014-03-25 Odobreno za objavo: 2014-04-25

O načelih vzdrževanja kiberfizičnih sistemov Santiago Ruiz-Arenas1,2 – Imre Horváth1 – Ricardo Mejía-Gutiérrez2 – Eliab Z. Opiyo1 1 Tehnična

univerza v Delftu, Fakulteta za industrijski dizajn inženiring, Nizozemska EAFIT, Raziskovalna skupina za dizajn inženiring, Kolumbija

2 Univerza

Kiberfizični sistemi (KFS) se hitro širijo na različnih področjih. Lastnosti teh sistemov se v veliki meri razlikujejo od lastnosti linearnih kompleksnih sistemov (LKS). Zagotavljanje brezhibnega in zanesljivega delovanja je zato velik izziv, ki zahteva razvoj novih načel snovanja, implementacije in vzdrževanja. Doseganje visoke stopnje zanesljivosti je ena novejših teoretičnih in praktičnih nalog v zvezi s KFS, še posebej pri kiberfizičnih sistemih, ki so ključnega pomena za poslovanje. Znano je, da je tudi za vzdrževanje takšnih sistemov potrebno razviti in uveljaviti nova načela, ki se lahko uporabljajo v različnih kontekstih. V naši raziskavi je bil uporabljen model sklepanja, ki je prikazan na Sliki 1 (stran 816). Cilj je opredelitev razmerij med principi vzdrževanja, uporabljenimi za LKS, in sistemskimi lastnostmi KFS. Zato smo najprej raziskali, po čem se razlikujejo lastnosti sistemov LKS in KFS. V naslednjem koraku smo opredelili znana načela vzdrževanja LKS ter ugotavljali, katera med njimi so uporabna tudi za KFS, ob upoštevanju drugačnih sistemskih lastnosti. Po analizi in primerjavi lastnosti sistemov LKS in KFS smo ugotovili, da so tako pri enih kot pri drugih v fizične naprave integrirane informacijske tehnologije, da so geografsko porazdeljeni, imajo več virov energije, združujejo večje število funkcijskih enot ter, da se izvaja intenzivna interakcija s človeškimi deležniki in okoljem, v katerega so vgrajeni. KFS so običajno odprti in nelinearni sistemi, ki imajo več funkcijskih povezav med komponentami, pridobivajo informacije za krmiljenje v realnem času iz realnih procesov, imajo visoko raven avtomatizacije in inteligence ter so pogosto razviti za delo v dinamičnih ali celo agresivnih okoljih. Izhajajoč iz teh razlik smo postavili hipotezo, da je pri vzdrževanju KFS potreben drugačen pristop kot pri LKS. Zaradi visoke ravni sinergije sestavnih delov, ki je lastna KFS, ni mogoče ločevati vzdrževanja fizičnih, programskih in kiberkomponent. To je velika razlika v primerjavi z LKS, kjer se sodelovanje komponent strojne in programske opreme dogaja na nižji ravni sinergije. Na osnovi teh teoretičnih razglabljanj predlagamo strategijo vzdrževanja, ki istočasno vključuje heterogene komponente KFS. Raziskava je pokazala naslednje: (i) zaradi določene podobnosti lastnosti sistemov obstaja priložnost za neposreden prenos nekaterih načel vzdrževanja iz LKS na KFS, (ii) obstajajo načela vzdrževanja LKS, ki jih je mogoče izkoristiti z ustreznimi prilagoditvami, (iii) nekatera načela vzdrževanja pa zaradi razlik v sistemskih lastnosti ne pridejo v poštev. Nove lastnosti sistemov KFS zahtevajo razvoj novih principov vzdrževanja, ki veljajo izključno za KFS. Glavni prispevek tega članka je opredelitev novih načel vzdrževanja KFS. Kot aplikativni kontekst je bila obravnavana domena kiberfizičnih rastlinjakov. Predstavljenih je več primerov omenjenih vrst načel vzdrževanja. To delo je le prvi korak na novo raziskovalno področje. Za poglobljeno razumevanje bodo potrebne dodatne intenzivne raziskave, razvoj cele vrste načel vzdrževanja za različne KFS in njihova validacija na večjem številu praktičnih aplikacij. Prihodnje študije bodo usmerjene v identifikacijo in preučitev glavnih dejavnikov vpliva ter vzrokov odpovedi KFS. Ključne besede: kompleksni tehnični sistemi, kiberfizični sistemi, načela vzdrževanja, odpovedi, vzdrževalni svetovalni sistem

SI 150

*Naslov avtorja za dopisovanje: Tehnična univerza v Delftu, Fakulteta za industrijski dizajn inženiring, Landbergstraat 15, Delft, Nizozemska, s.ruizarenas@tudelft.nl

Prejeto v recenzijo: 2014-04-04 Prejeto popravljeno: 2014-09-08 Odobreno za objavo: 2014-09-25

Vrednotenje vpliva faze povečanja osnega pritiska pri varjenju s trenjem na lastnosti spoja ulitkov iz različnih aluminijevo-bakrovih zlitin Barlas, Z. – Çolak, M. Zafer Barlas* – Murat Çolak

Univerza v Sakaryi, Tehniška fakulteta, Oddelek za metalurgijo in materiale, Turčija

Konvencionalne tehnologije varjenja zaradi napetostnega korozijskega pokanja in toplega pokanja v splošnem niso primerne za varjenje ulitkov iz zlitin aluminija in bakra (Al-Cu, serija 2XX.X). Zato obstaja potreba po eksperimentalnih študijah varjenja teh zlitin. Članek obravnava varivost ulitih palic iz različnih aluminijevih zlitin – z 2 % bakra (Al-2%Cu) in z 8 % bakra (Al-8%Cu) – po postopku varjenja s trenjem. Podana je analiza mehanskih, makrostrukturnih in mikrostrukturnih lastnosti testnih zvarov. Obe zlitini Al-Cu sta bili uliti v peščene forme, ulite palice pa so bile nato za varjenje obdelane na premer 20 mm in dolžino 90 mm. Torni zvari različnih zlitin Al-Cu so bili narejeni na vertikalnem rezkalnem centru. V študiji je bil preučen vpliv faze povečanja osnega pritiska na lastnosti spoja. Trdni zvarni spoji so bili ustvarjeni pri različnih vrednostih tlaka (24 in 32 MPa) in časa (20, 50 in 80 s) faze povečanja osnega pritiska. Konstantni parametri tornega varjenja so bili torni tlak 8 Pa, torni čas 60 s in torna hitrost 1500 vrt/min. Povečanje temperature med postopkom varjenja je bilo merjeno 3 mm pod stično površino in na globini 2,5 mm od zunanje površine palic Al-2%Cu, uporabljen pa je bil termopar tipa K. Za vrednotenje trdnosti torno zvarjenih preskušancev in osnovne ulite aluminijeve zlitine je bil uporabljen natezni preskus. Trdota spoja z najvišjo porušitveno natezno trdnostjo (UTS) je bila izmerjena po Brinellu in po Vickersu. Mikrostruktura na poliranem in jedkanem prerezu spoja je bila preučena z optičnim mikroskopom in z vrstičnim elektronskim mikroskopom (SEM). Za opredelitev povezav med mikrostrukturo in mehanskimi lastnostmi sta bila uporabljena mikroskop SEM, opremljen z aparatom za energijsko disperzivno rentgensko spektroskopijo (EDS), in rentgenska difrakcijska analiza (XRD). Ugotovljeno je bilo, da je postopek varjenja s trenjem primeren za spajanje ulitkov iz različnih zlitin Al-Cu. Na lastnosti spoja v nateznem preizkusu bistveno bolj vpliva tlak v fazi povečanja osnega pritiska kot čas povečanja osnega pritiska. Višji tlak v fazi povečanja osnega pritiska namreč povzroči boljšo zvezo na stiku zvara. Vrednost UTS = 142 MPa je mogoče doseči pri tlaku 32 MPa in času 80 s v fazi povečanja osnega pritiska. Med nateznimi preskusi je prišlo do krhkega zloma, ki je po analizah SEM – EDS in XRD posledica prisotnosti intermetalnega Al2Cu. Varilno območje z najvišjo vrednostjo UTS je bilo sestavljeno iz tipičnih mikrostrukturnih območij: kovina zvara (WM), toplotno vplivani pasovi (HAZ), termomehansko vplivani pasovi (TMAZ) in osnovna kovina (BM). Meritve makrotrdote po Brinellu so pokazale, da ima največjo trdoto kovina zvara, vzrok za to pa so predvsem finejša zrna v kovini zvara. Preskus mikrotrdote po Vickersu je razkril, da je na trdoto vplivala predvsem intermetalna faza Al2Cu – večja prisotnost Al2Cu v okolici piramide je namreč povečala vrednost mikrotrdote. Vršna temperatura 436 °C je bila izmerjena 3 mm pod stično površino in na globini 2,5 mm od zunanje površine palice Al-2%Cu. V nadaljnjih študijah bo mogoče opraviti še ostale mehanske preskuse, npr. upogibni preskus oz. Charpyjev udarni preskus. Za izboljšanje učinkovitosti spoja bi bilo mogoče uporabiti tudi drugačne parametre varjenja, še posebej pri osnovni zlitini Al-8%Cu. Premer varjenih palic bi bilo mogoče zmanjšati za skrajšanje časa in zmanjšanje sile povečanega osnega pritiska pri varjenju s trenjem. Vpliv zmanjšanja prereza na lastnosti zvara pa bi bilo mogoče analizirati tudi pri enakih parametrih povečanega osnega pritiska, kot so bili uporabljeni v tej študiji. Obstaja pomanjkanje eksperimentalnih raziskav varivosti ulitkov iz zlitin Al-Cu po postopkih varjenja v trdnem stanju. Palice iz različnih zlitin Al-2%Cu in Al-8%Cu so bile zvarjene s trenjem na vertikalnem rezkalnem centru, ustvarjeni spoj pa ima ugodne lastnosti. Avtorji zato verjamejo, da bodo predstavljene metode in rezultati tega članka dober prispevek k razpoložljivi literaturi. Ključne besede: varjenje s trenjem, ulitki iz zlitine Al-Cu, faza povečanja osnega pritiska, natezna trdnost, mikrostruktura, temperatura zvara

*Naslov avtorja za dopisovanje: Univerza v Sakaryi, Tehniška fakulteta, Oddelek za metalurgijo in materiale, Esentepe kampus, Sakarya,Turčija, barlas@sakarya.edu.tr

SI 151

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, SI 152-156 Osebne objave

Doktorske disertacije, diplomske naloge

DOKTORSKE DISERTACIJE Na Fakulteti za strojništvo Univerze v Ljubljani je obranil svojo doktorsko disertacijo: ● dne 4. novembra 2014 Matej HUDOVERNIK z naslovom: »Karakterizacija procesa 3D krivljenja profilov« (mentor: prof. dr. Karl Kuzman, somentor: prof. dr. Erman Tekkaya); Upogibanje votlih tankostenskih profilov iz različnih vrst jekla pri poljubnih pogojih 2D ali 3D obremenjevanja velja za precej neraziskano področje, ki še ni v celoti okarakterizirano. Raziskovalno delo, ki je predstavljeno v domeni doktorske naloge vodi k boljšemu razumevanju odziva omenjenih profilov z ozirom na širok spekter vhodnih procesnih parametrov. Raziskave se navezujejo bodisi na konstantne ali spremenljive pogoje krivljenja. Metodološki pristop je zasnovan na osnovi analitične formulacije poenostavljenega konzolnega nosilca in realnih eksperimentalnih analiz z uporabo namenske kinematične tehnologije za krivljenje profilov, ki deluje po principu prostorske superpozicije navora. Osnovna hipoteza je testirana z uporabo kvadratnih votlih profilov iz treh različnih vrst jekla, z različnimi geometrijskimi lastnostmi prerezov. Rezultati, dobljeni z uporabo analitičnih in eksperimentalnih metod pa služijo kot kriterij za vrednotenje numeričnih MKE simulacijskih modelov s katerimi se nato izvedejo še podrobnejše analize odziva profilov pri različnih pogojih krivljenja. Sistemsko načrtovanje numeričnih eksperimentov se zagotovi z uporabo centralno-kompozitnih matrik (angl. central composite designs) in metodologije odziva površin (angl. response surface methodology). S tem se pridobijo zanesljivi statistični modeli s katerimi je moč v naprej predvideti procesne parametre za vsak posamezen vhodni material. Numerični eksperimenti se najprej izvedejo za pogoje ravninskega krivljenja pri različnih naklonih upogibne ravnine. Ob jasnem razumevanju 2D pogojev krivljenja, pa se izvedejo še numerični eksperimenti za aplikacijo prostorskega krivljenja. * Na Fakulteti za strojništvo Univerze v Mariboru so obranili svojo doktorsko disertacijo: ● dne 18. novembra 2014 Primož KOCUTAR z naslovom: »Simulacija turbulentnega toka s hibridnim LES/URANS turbulentnim modelom z uporabo SI 152

metode robnih elementov« (mentor: prof. dr. Leopold Škerget); V doktorski disertaciji se posvečamo razvoju hibridnega LES/URANS turbulentnega modela na osnovi metode robnih elementov (MRE) za simulacijo turbulentnega toka tekočine. Za izračun toka tekočine rešujemo sistem Navier-Stokesovih enačb zapisan v hitrostno-vrtinčni formulaciji. Sistem enačb je sestavljen iz enačb kinematike vrtinčnega polja na robu in hitrostnega polja v območju, enačbe kinetike vrtinčnosti polja, energijske enačbe, ter prenosne enačbe turbulentne kinetične energije za izračun turbulentnih modelov. Za enačbo kinematike smo uporabili nefizikalno časovno shemo. Uporabili smo spojen LES/URANS hibridni turbulentni model, kjer je vmesna površina med LES in URANS območje določena s fizikalno veličino, ter je dinamično določena tekom simulacije. Za preklopni kriterij med LES in URANS območjem smo uporabili Reynoldsovo število določeno s turbulentno kinetično energijo, ter Reynoldsovo število določeno s skupno turbulentno kinetično energijo. Glavna značilnost spojenih hibridnih modelov je, da za izračun toka uporabljajo en set vodilnih enačb. LES in URANS model sta v povezavi s preklopnim kriterijem spojena v prenosni enačbi turbulentne kinetične energije. V odvisnosti od karakteristike toka, ter preklopnega kriterija, za določeno območje uporabimo podmrežno ali URANS efektivno viskoznost. Pod-mrežna ali URANS viskoznost je nadaljnje uporabljena v prenosni enačbi turbulentne kinetične energije, ter v vodilnih enačbah za izračun toka tekočine. V hibridnem LES/URANS turbulentnem modelu je LES model uporabljen za vrtince z največ energije, torej velike vrtince, ter URANS model za obstensko območje. Za LES in URANS model smo uporabili modela, ki temeljita na turbulentni kinetični energiji, pri čemer smo za URANS model izbrali eno-enačbeni model razvit za obstenska območja. Numerični algoritem smo najprej validirali na testnih primerih. Za validacijo vodilnih enačb toka tekočine smo uporabili testni primer direktne numerične simulacije toka v gnani kotanji, ter za validacijo energijske enačbe direktno numerično simulacijo toka naravne konvekcije pri nižjih Rayleighevih številih. Po uspešni validaciji z uporabo direktne numerične simulacije, smo razvit LES/URANS hibridni turbulentni model testirali na turbulentnem toku naravne konvekcije v kvadratni kotanji, s čimer smo potrdili pravilnost delovanja hibridnega modela;

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, SI 152-156

● dne 18. novembra 2014 Branka VILTUŽNIK z naslovom: »Adsorpcija Pb2+ in Hg2+ ionov z uporabo površinsko modificiranih supermagnetnih nanodelcev železovega oksida« (mentor: prof. dr. Aleksandra Lobnik); Razvoj novih nanomaterialov za odstranjevanje težkih kovin je v zadnjem času pritegnil veliko pozornost, saj postaja onesnaževanja s strupenimi kovinskimi ioni, kot so Pb2+, Hg2+, Ni2+, Cd2+, Ag+, Co2+ in tem podobni, resen okoljski in zdravstveni problem. Velik problem s težkimi kovinami predstavlja njihova sposobnost, da se kopičijo v okolju in povzročajo zastrupitve. Za razliko od nekaterih organskih onesnaževal, težke kovine niso biorazgradljive. Znano je, da ioni težkih kovin kot so Pb2+, Hg2+, Cd2+, Ni2+ in Cu2+ lahko povzročajo resne zdravstvene težave za ljudi in živali, saj se specifično vežejo na proteine, nukleinske kisline in majhne metabolite v živih organizmih, s čimer zavirajo njihove funkcije. Za odstranjevanje ionov težkih kovin iz odpadnih vod, obstaja več metod, kot so kemijsko obarjanje, ionska izmenjava, tekoče-tekoča ekstrakcija, elektroliza, koagulacija, flotacija, flokulacija, filtracija, oksidacija, reverzna osmoza, biosorpcija in adsorpcija. (Singh in sodelavci, 2011) Vsaka od naštetih metod ima svoje pomanjkljivosti v smislu učinkovitosti, stroškov in kompleksnosti. Proces elektrolize je povezan z visokimi operativnimi stroški, kemijsko obarjanje proizvede sekundarni odpadek, pri filtraciji je problematično mašenje membran in cena le-teh. Uporaba večine naštetih metod je ekonomsko neprimernih, kadar gre za nizke koncentracije težkih kovin. Zaradi tega se danes vedno bolj uporabljajo nizkocenovni biosorbenti, kot so glineni materiali, biomasa, zeoliti, aktivni ogljik in smole za ionsko izmenjavo, vendar pa se morajo izboljšati njihove adsorpcijske kapacitete, separacijska stopnja in oblika. Tudi ti adsorbenti imajo nekatere slabosti pri odstranjevanju ionov težkih kovin iz vodnih medijev, in sicer niso odzivni na spremembe parametrov kot so pH, koncentracija elektrolitov, soli in prisotnost drugih topljencev, temperature in ionske moči medija. (Ozay in sodelavci, 2010) Idealen adsorbent mora imeti močno afiniteto do ionov težkih kovin, kar je povezano z visoko specifično površino, poroznostjo, številom veznih mest, ipd. Znano je, da večja kot je specifična površina, večja je adsorpcijska kapaciteta adsorbenta. V primeru magnetnih nanodelcev bi moral idealen adsorbent zagotoviti primerno adsorpcijo pri različnih reakcijskih pogojih in bi ga kot takega lahko uporabljali za vse odpadne vode, onesnažene s težkimi kovinami, saj ga bi po končani uporabi enostavno

odstranili z zunanjim magnetnim poljem, predvsem na odlagališčih nevarnih odpadkov, kjer se vsakodnevno pred izpustom vode v kanalizacijo in okolje soočajo s prekomerno vsebnostjo težkih kovin. Magnetni adsorbenti bi lahko povečali učinkovitost pridobivanja oz. recikliranja težkih kovin iz odpadnih vod pri procesih kovinsko-predelovalne in oplemenitile industrije. Da bi zadostili vsem okoljevarstvenim predpisom so se v industriji primorani posluževati najnaprednejših tehnik. Monodisperzni, enodomenski nanodelci na osnovi spinelnih feritov so znanstveno in tehnološko izkazali kot zanimivi materiali za adsorpcijo ionov težkih kovin, zaradi svojih posebnih lastnosti, ki so določene s superparamagnetizmom (El-Okr in sodelavci, 2011). Superparamagnetni nanodelci s prilagojenimi površinskimi lastnostmi in visoko specifično površino izkazujejo ustrezne adsorpcijske karakteristike in predstavljajo novost na področju razvoja novih adsorpcijskih materialov. Njihova prednost v primerjavi z obstoječimi adsorbenti je tudi v tem, da jih tudi po adsorpciji lahko enostavno in hitro odstranimo iz raztopine z uporabo zunanjega magnetnega polja. V zadnjih letih se je veliko raziskovalcev ukvarjalo s sintezo magnetnih materialov, kot so magnetit (Fe3O4) za odstranjevanje ionov težkih kovin iz vodnih suspenzij, kot so Pb2+ (Yantasee in sodelavci, 2007; Li in sodelavci, 2011; Liu in sodelavci, 2009) in Hg2+ (Hakami in sodelavci, 2012; Dong in sodelavci, 2008; Parham in sodelavci, 2012; Girginova in sodelavci, 2010). Nekaj raziskovalcev se je v zadnjih letih tudi ukvarjalo s sintezo kobalt feritnih; ● dne 20. novembra 2014 Afrim GJELAJ z naslovom: »Samodejno in inteligentno programiranje CNC strojev« (mentor: prof. dr. Jože Balič); Razvoj obdelovalnih sistemov poteka danes v smeri vse večje avtomatizacije in njihove prilagodljivosti. To je povzročilo tudi pospešen razvoj sistemov za avtomatsko programiranje CNC strojev, ki so danes v veliki meri avtomatizirani. Popolno avtomatizacijo je nemogoče doseči. Zato je poudarek te doktorske naloge na razvoju sistema avtomatskega programiranja CNC strojev, ki je podprt z metodami umetne inteligence. Tak sistem mora omogočiti avtomatizacijo CNC programiranja na inteligenten način, da se tako dvigne kvaliteta programiranja, zanesljivost in določi optimalno zaporedje obdelave ter izbere optimalne rezalne pogoje. Doktorska disertacija obravnava pregled različnih pristopov k samodejnemu programiranju CNC obdelovalnih strojev in predstavlja prispevek k razvoju takšnega samodejnega inteligentnega programiranja CNCSI 153

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, SI 152-156

obdelovalnih strojev, ki bo uporabno v industrijski praksi. Raziskana in analizirana sta dva trenda razvoja in raziskav, ki med seboj hkrati konkurirata in se dopolnjujeta; to je samodejno programiranje, ki vključuje deterministične pristope in inteligentno samodejno programiranje. Poleg tega so raziskani in opisani postopki CNC-programiranja ter stanje med komercialno dobavljivimi programi in raziskavami na tem področju. Izpostavljeni so cilji samodejnega in/ ali inteligentnega programiranja in izdelan je ustrezni model, ki uporablja večkriterijsko optimizacijo, kjer so cilji podvrženi dinamičnemu tehniško-ekonomskemu okolju, ki pogosto menja pomembnost ciljev. Zaradi teh omejitev, je uporabljena večkriterijska optimizacija z genetskimi algoritmi (Ang.: MultiObjective Genetic Algorithm - MOGA) in diskretna predstavitev problema. Večkriterijska optimizacija je logično nadaljevanje dosedanjih raziskav na področju določanja optimalnega programa, ker človek-strokovnjak izvaja prav tako večkriterijsko optimizacijo pri svojem delu. Obdelava na CNCrezkalnih strojih vključuje tehniške, ekonomske in organizacijske kriterije, ki jih človek mimogrede upošteva, zato umetni sistem, ki uporablja enokriterijsko optimizacijo ne zagotavlja dovolj dobrih rezultatov za uporabo v praksi. Diskretna predstavitev problema poenostavi problem, zagotavlja omejitev iskalnega prostora in tako izboljša možnost uporabe metode v realnem času; ● dne 21. novembra 2014 Tadej TAŠNER z naslovom: »Napredni koncepti vodenja in nadzora energijsko učinkovitih hidravličnih sistemov« (mentor: izr. prof. dr. Darko Lovrec); Na področju energijske učinkovitosti industrijskih pogonskih sistemov se pojavljajo vedno strožji predpisi in zahteve za zmanjšanje porabe energije in izpustov toplogrednih plinov; pri čemer hidravlični sistemi niso izvzeti. Energijsko učinkovitost hidravličnih sistemov lahko izboljšamo tako, da izboljšamo izkoristek posameznih sestavnih delov hidravličnega pogona, in tudi tako, da izboljšamo koncept vodenja oziroma regulacije takšnega sistema. V smislu slednjega so v doktorski disertaciji predstavljeni trije možni pogonski sklopi, ki temeljijo na volumetričnem principu regulacije hidravlične energije. Zasnovani so matematični modeli vseh komponent pogonskih sklopov, tako da je omogočena statična in dinamična simulacija izkoristka vseh pogonskih sklopov. Rezultati, pridobljeni s simulacijo, so nato preverjeni še na preizkuševališču pogonskih sklopov. Na podlagi rezultatov so zasnovani različni regulacijski koncepti, med drugim tudi nov koncept, ki omogoča sledenje najvišjemu izkoristku hidravličnega sistema. Regulacijski koncepti so med SI 154

seboj primerjani tako v simulaciji kot tudi praktično na preizkuševališču. Doktorska disertacija se dotakne tudi ionskih tekočin, ki imajo velik potencial, da postanejo hidravlične tekočine prihodnosti. Ker se stisljivost ionskih tekočin razlikuje od stisljivosti hidravličnega olja ter ima velik vpliv na dinamiko in energijsko učinkovitost hidravličnih sistemov, je poudarek predvsem na stisljivosti in merjenju stisljivosti ionskih tekočin. Na koncu disertacije je predstavljen še modul za brezžični zajem podatkov, ki ga je možno preprogramirati tudi tako, da lahko izvaja vse predstavljene regulacijske algoritme. DIPLOMSKE NALOGE Na Fakulteti za strojništvo Univerze v Ljubljani so pridobili naziv univerzitetni diplomirani inženir strojništva: dne 26. novembra 2014: Alen BUBANJA z naslovom: »Izdelava 3D-tiskalnika za hitro prototipiranje« (mentor: prof. dr. Janez Kopač); Matic GOROPEVŠEK z naslovom: »Povezovanje kibernetsko-fizikalnih sistemov v pametnem laboratoriju« (mentor: prof. dr. Peter Butala); dne 27. novembra 2014: Marko HEGLER z naslovom: »Računalniško modeliranje in simulacija delta robota s pnevmatično mišico in enosmernim električnim motorjem« (mentor: izr. prof. dr. Niko Herakovič). * Na Fakulteti za strojništvo Univerze v Mariboru je pridobil naziv univerzitetni diplomirani inženir strojništva: dne 27. novembra 2014: Boštjan BRUNEC z naslovom: »Določitev lastnosti utrujanja poroznega lotus gradiva z numeričnimi simulacijami« (mentor: izr. prof. dr. Miran Ulbin, somentor: doc. dr. Janez Kramberger). * Na Fakulteti za strojništvo Univerze v Ljubljani so pridobili naziv magister inženir strojništva: dne 26. novembra 2014: Matic AMBROŽIČ z naslovom: »Kotalno frezanje jermenice za zobati jermen« (mentor: prof. dr. Janez Kopač); Rok KAPLER z naslovom: »Nadgraditev laserskega vira Er:YAG z mehanskim preklopnikom kvalitete« (mentor: prof. dr. Janez Možina, somentor: doc. dr. Matija Jezeršek);

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, SI 152-156

dne 27. novembra 2014: Klemen COTIČ z naslovom: »Geometrijsko in napetostno stanje ravninskih geometrijsko nelinearno prednapetostnih nosilcev« (mentor: prof. dr. Franc Kosel); Matjaž KOS z naslovom: »Triangulacijsko merjenje pozicije zvarnega robu pri laserskem daljinskem varjenju« (mentor: doc. dr. Matija Jezeršek); Katja RUDNIK z naslovom »Konstrukcija ohišja hladilnika z večjo togostjo« (mentor: izr. prof. dr. Jože Tavčar, somentor: prof. dr. Jožef Duhovnik). * Na Fakulteti za strojništvo Univerze v Mariboru je pridobil naziv univerzitetni diplomirani gospodarski inženir: dne 27. novembra 2014: Matjaž KOROŠEC z naslovom: »Ocena uspešnosti ponudbe storitev« (mentor: doc. dr. Marjan Leber, doc. dr. Aleksandra Pisnik). * Na Fakulteti za strojništvo Univerze v Mariboru so pridobili naziv magister inženir strojništva: dne 26. novembra 2014: David BELOVIČ z naslovom: »Koncept modularne testne celice namenjene večkratnemu transportu« (mentor: doc. dr. Aleš Belšak); Matija GLINŠEK z naslovom: »Analiza inovativnosti slovenskih proizvodnih podjetij« (mentor: doc. dr. Iztok Palčič); Matjaž HRŽIČ z naslovom: »Konstruiranje vodenja avtomatskih drsnih vrat na polimernem vodilu« (mentor: doc. dr. Janez Kramberger, somentor: prof. dr. Srečko Glodež); Sandi KEŠPRET z naslovom: »Izdelava kinematičnega modela in poprocesorja dvoosne stružnice v okolju Siemens NX« (mentor: doc. dr. Mirko Ficko); * Na Fakulteti za strojništvo Univerze v Mariboru sta pridobila naziv magister gospodarki inženir: dne 26. novembra 2014: Boštjan HRAŠAR z naslovom: »Optimizacija sistema skladiščenja in oskrbe z materiali za vzdrževanje v Gorenju d.d.« (mentorja: doc. dr. Iztok Palčič, doc. dr. Igor Vrečko); Sebastijan JURKOŠEK z naslovom: »Oblikovanje celovite storitve in zagon podjetja za osebno trenerstvo« (mentorja: doc. dr. Marjan Leber, prof. dr. Duško Uršič).

* Na Fakulteti za strojništvo Univerze v Mariboru sta pridobila naziv diplomirani inženir strojništva (UN): dne 14. novembra 2014: Jernej KOLAR z naslovom »Rekonstrukcija tunelske žarilne peči« (mentor: prof. dr. Niko Samec, somentor: dr. Jurij Iljaž); dne 28. novembra 2014: Kristjan KOLAR z naslovom »Kontrola zvarnih spojev z neporušitvenimi metodami« (mentor: prof. dr. Bojan Ačko, somentorja: prof. dr. Aleš Hribernik, doc. dr. Tomaž Vuherer). * Na Fakulteti za strojništvo Univerze v Ljubljani so pridobili naziv diplomirani inženir strojništva: dne 7. novembra 2014: Tomaž KOS z naslovom: »Končna kontrolna priprava za avtomobilski izpušni sistem« (mentor: doc. dr. Andrej Lebar, somentor: doc. dr. Joško Valentinčič); Aleš PADARŠIČ z naslovom: »Vpliv sobne hladilne naprave na ugodje v prostoru« (mentor: doc. dr. Matjaž Prek, somentor: prof. dr. Vincenc Butala); Gregor PAPEŽ z naslovom: »Plamensko varjenje, še vedno aktualen postopek?« (mentor: prof. dr. Janez Tušek). * Na Fakulteti za strojništvo Univerze v Ljubljani je pridobil naziv diplomirani inženir strojništva (VS): dne 7. novembra 2014: Andrej MLAKAR z naslovom: »Razvoj naprave za posluževanje injekcijske stiskalnice za brizganje gume« (mentor: prof. dr. Marko Nagode). * Na Fakulteti za strojništvo Univerze v Mariboru so pridobili naziv diplomirani inženir strojništva (VS): dne 27. novembra 2014: Benjamin BALAŽIC z naslovom: »Servo pogoni polnilne linije« (mentor: prof. dr. Iztok Potrč, somentor: izr. prof. dr. Tone Lerher); Damijan JAVORNIK z naslovom: »Konstruiranje elevatorja kot podpora transportnemu sistemu uplinjevalca« (mentor: prof. dr. Iztok Potrč, somentor: izr. prof. dr. Tone Lerher); Tadej STRAŠEK z naslovom: »Podajalni robot za varjenje gozdarskih vitlov« (mentor: izr. prof. dr. Karl Gotlih, somentor: doc. dr. Tomaž Vuherer). SI 155

Strojniški vestnik - Journal of Mechanical Engineering 60(2014)12, SI 152-156

V spomin prof. dr. Antonu Kuhlju Pred nedavnim nas je hudo prizadela vest, da je umrl naš sodelavec, spoštovani prof. dr. Anton Kuhelj, dolgoletni visokošolski učitelj, prodekan in dekan na Fakulteti za strojništvo, Univerze v Ljubljani, naš vzornik in prijatelj. Prof. dr. Anton Kuhelj se je rodil leta 1934 v Ljubljani, kjer je maturiral kot najboljši maturant generacije gimnazijcev, ki se je učila tako latinščine kot tudi grščine. Diplomiral je na Fakulteti za strojništvo leta 1960. Po diplomi je dve leti delal na Inštitutu Jožef Stefan, nato pa skoraj dve leti spoznaval praktične vidike dinamike na Centrali za dinamiko v podjetju Sulzer AG v Winterthuru v Švici. Po vrnitvi v domovino se je zaposlil na Fakulteti za strojništvo, Univerze v Ljubljani, kjer je bil leta 1964 izvoljen za asistenta za področje mehanike. Leta 1969 je magistriral ter leta 1972 doktoriral pri prof. dr. Ervinu Prelogu. V naziv docenta je bil izvoljen leta 1973, v naziv izrednega profesorja leta 1978 ter v naziv rednega profesorja leta 1984. Na Fakulteti za strojništvo, Univerze v Ljubljani, je poučeval predmete s področja mehanike, na začetku na visokošolskem in višješolskem ter kasneje na univerzitetnem študiju, vse do svoje upokojitve, decembra 1997. Predmeti, pri razvoju katerih je pustil neizbrisljiv pečat, so s področja dinamike. Zasnoval je predmet Dinamika strojev, v sklopu katerega je poučeval poglobljene teoretične podlage, ki so bile tako takrat kot tudi danes nepogrešljive sodobnemu inženirju strojništva. Za potrebe dodiplomskega študija je napisal visokošolska učbenika Kinematika ter Dinamika. Na podiplomskem študiju Fakultete za strojništvo, Univerze v Ljubljani je bil nosilec oz. sonosilec predmetov Dinamika in vibracije, Lomna mehanika ter Akustična emisija in hrup. V obdobju aktivnega poučevanja je pod njegovim strokovnim mentorstvom zaključilo študij večje število študentov na diplomskem, magistrskem ter doktorskem študiju. Profesor Kuhelj je ustanovil ter do upokojitve tudi vodil Laboratorij za dinamiko strojev in konstrukcij, v katerem se je pod njegovim vodstvom začelo ter razmahnilo tako teoretično-raziskovalno kot tudi aplikativnorazvojno delo na področju dinamike mehanskih sistemov, predvsem za potrebe slovenske industrije. Raziskovalci njegovega laboratorija so združevali teoretično-analitične raziskave z eksperimentalnim delom na področju dinamike. Bil je pobudnik eksperimentalnih raziskav, za katere je v letih vodenja laboratorija pridobil eksperimentalno opremo, ki je omogočala primerjavo izračunanih ter izmerjenih spremenljivk v dinamiki. Aktiven je bil na področju dinamike rotorjev, modeliranja dinamičnega obnašanja realnih tehniških sistemov, utrujanja kovin ter zmanjševanja hrupa. Zaradi velikega ugleda, ki ga je profesor Kuhelj užival v ožji in širši družbi, so mu bile zaupane številne pomembne funkcije. Bil je član Znanstvenega sveta za elektrotehniko in strojništvo, predsednik Znanstvenega sveta za raziskave za področje proizvodne kibernetike, obdelovalnih sistemov, robotike, konstruiranja in varjenja, ter član Odbora za raziskovalno infrastrukturo Republike Slovenije. Bil je tudi vodja usmerjenega raziskovalnega programa »Konstruiranje delovnih strojev in motorjev« na ravni Republike Slovenije. Na Fakulteti za strojništvo, Univerze v Ljubljani, je profesor Kuhelj opravljal pomembna vodstvena dela kot prodekan za znanstvenoraziskovalno delo v obdobju 1985−1987 ter kot dekan fakultete v letih 1987−1988. Vodil je več usmerjenih raziskovalnih programov in s tem omogočil sredstva za aktivno delovanje večjega števila mlajših raziskovalcev na Fakulteti za strojništvo. Profesor Kuhelj ni bil le odličen pedagog in strokovnjak, bil je predvsem rahločuten in preudaren človek širokih misli. Širino mu je dala njegova družina, klasična izobrazba, znanje več tujih jezikov in ne nazadnje tudi bivanje in pridobivanje strokovnih izkušenj v tujini. Vsako odločitev je podrobno premislil in čas je pokazal, da so bile njegove odločitve vedno pravilne. Profesor Kuhelj je bil ustvarjalen in široko razgledan človek. Počaščeni smo, da smo bili njegovi študentje, sodelavci in prijatelji. Slava mu in večni spomin. Prof. dr. Miha Boltežar SI 156

Technical Editor Pika Škraba University of Ljubljana, Faculty of Mechanical Engineering, Slovenia

Founding Editor Bojan Kraut

University of Ljubljana, Faculty of Mechanical Engineering, Slovenia

Vice-President of Publishing Council Jože Balič

University of Maribor, Faculty of Mechanical Engineering, Slovenia Cover: Hydrokinetic Darrieus turbine with horizontal axis of rotation (middle) that uses kinetic energy of water current without impoundment or side channels and thus enables a quick, cost effective installation on a shallow riverbeds for electricity production with minimal environmental impact. The CFD simulation was performed with a flow driven approach, which better corresponds to the real operating conditions. Simulation results for two different turbine positions are shown above and below. Courtesy: Matjaž Fleisinger, University of Maribor, Faculty of Mechanical Engineering, Slovenia

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

http://www.sv-jme.eu

60 (2014) 12

Since 1955

Papers

769

Matjaž Fleisinger, Matej Vesenjak, Matjaž Hriberšek: Flow Driven Analysis of a Darrieus Water Turbine

777

Ivan Dunđerski: Managing Vehicle Acceleration Properties by Programming Functions for Engine Torque Control

789

Primož Potočnik, Tomaž Berlec, Alojz Sluga, Edvard Govekar: Hybrid Self-Organization Based Facility Layout Planning

797

Nataša Vujica Herzog, Stefano Tonchia: An Instrument for Measuring the Degree of Lean Implementation in Manufacturing

804

Marek Boryga: Trajectory Planning of an End-Effector for Path with Loop

815

Santiago Ruiz-Arenas, Imre Horváth, Ricardo Mejía-Gutiérrez, Eliab Z. Opiyo: Towards the Maintenance Principles of Cyber-Physical Systems

832

Zafer Barlas, Murat Çolak: Evaluation of the Influence of Upset Stage on Joint Properties of Friction Welded Dissimilar Aluminum-Copper Cast Alloys

Journal of Mechanical Engineering - Strojniški vestnik

Contents

12 year 2014 volume 60 no.

Strojniški vestnik Journal of Mechanical Engineering