ISSN 1854-6250 APEM journal Advances in Production Engineering & Management Volume 15 | Number 1 | March 2020 Pfl Published by CPE apem-journal.org University of Maribor Advances in Production Engineering & Management Identification Statement APEM ISSN 1854-6250 | Abbreviated key title: Adv produc engineer manag | Start year: 2006 ISSN 1855-6531 (on-line) Published quarterly by Chair of Production Engineering (CPE), University of Maribor Smetanova ulica 17, SI - 2000 Maribor, Slovenia, European Union (EU) Phone: 00386 2 2207522,Fax: 00386 2 2207990 Language of text: English APEM homepage: apem-journal.org UniversityofMaribor University homePage: WWW.um.si APEM Editorial Editor-in-Chief Miran Brezocnik editor@apem-journal.org, info@apem-journal.org University of Maribor, Faculty of Mechanical Engineering Smetanova ulica 17, SI - 2000 Maribor, Slovenia, EU Desk Editor Martina Meh desk1@apem-journal.org Janez Gotlih desk2@apem-journal.org Website Technical Editor Lucija Brezocnik desk3@apem-journal.org Editorial Board Members Eberhard Abele, Technical University of Darmstadt, Germany Bojan Acko, University of Maribor, Slovenia Joze Balic, University of Maribor, Slovenia Agostino Bruzzone, University of Genoa, Italy Borut Buchmeister, University of Maribor, Slovenia Ludwig Cardon, Ghent University, Belgium Nirupam Chakraborti, Indian Institute of Technology, Kharagpur, India Edward Chlebus, Wroclaw University of Technology, Poland Igor Drstvensek, University of Maribor, Slovenia Illes Dudas, University of Miskolc, Hungary Mirko Ficko, University of Maribor, Slovenia Vlatka Hlupic, University of Westminster, UK David Hui, University of New Orleans, USA Pramod K. Jain, Indian Institute of Technology Roorkee, India Isak Karabegovic, University of Bihac, Bosnia and Herzegovina Janez Kopac, University of Ljubljana, Slovenia Qingliang Meng, Jiangsu University of Science and Technology, China Lanndon A. Ocampo, Cebu Technological University, Philippines Iztok Palcic, University of Maribor, Slovenia Krsto Pandza, University of Leeds, UK Andrej Polajnar, University of Maribor, Slovenia Antonio Pouzada, University of Minho, Portugal R. Venkata Rao, Sardar Vallabhbhai National Inst. of Technology, India Rajiv Kumar Sharma, National Institute of Technology, India Katica Simunovic, J. J. Strossmayer University of Osijek, Croatia Daizhong Su, Nottingham Trent University, UK Soemon Takakuwa, Nagoya University, Japan Nikos Tsourveloudis, Technical University of Crete, Greece Tomo Udiljak, University of Zagreb, Croatia Ivica Veza, University of Split, Croatia Limited Permission to Photocopy: Permission is granted to photocopy portions of this publication for personal use and for the use of clients and students as allowed by national copyright laws. This permission does not extend to other types of reproduction nor to copying for incorporation into commercial advertising or any other profit-making purpose. Subscription Rate: 120 EUR for 4 issues (worldwide postage included); 30 EUR for single copies (plus 10 EUR for postage); for details about payment please contact: info@apem-journal.org Cover and interior design: Miran Brezocnik Printed: Tiskarna Kostomaj, Celje, Slovenia Subsidizer: The journal is subsidized by Slovenian Research Agency Statements and opinions expressed in the articles and communications are those of the individual contributors and not necessarily those of the editors or the publisher. No responsibility is accepted for the accuracy of information contained in the text, illustrations or advertisements. Chair of Production Engineering assumes no responsibility or liability for any damage or injury to persons or property arising from the use of any materials, instructions, methods or ideas contained herein. Copyright © 2020 CPE, University of Maribor. All rights reserved. Advances in Production Engineering & Management is indexed and abstracted in the WEB OF SCIENCE (maintained by Clarivate Analytics): Science Citation Index Expanded, Journal Citation Reports - Science Edition, Current Contents - Engineering, Computing and Technology • Scopus (maintained by Elsevier) • Inspec • EBSCO: Academic Search Alumni Edition, Academic Search Complete, Academic Search Elite, Academic Search Premier, Engineering Source, Sales & Marketing Source, TOC Premier • ProQuest: CSA Engineering Research Database -Cambridge Scientific Abstracts, Materials Business File, Materials Research Database, Mechanical & Transportation Engineering Abstracts, ProQuest SciTech Collection • TEMA (DOMA) • The journal is listed in Ulrich's Periodicals Directory and Cabell's Directory journal University of Maribor Chair of Production Engineering (CPE) Advances in Production Engineering & Management Volume 15 | Number 1 | March 2020 | pp 1-120 Contents Scope and topics 4 Neuro-mechanistic model for cutting force prediction in helical end milling of 5 metal materials layered in multiple directions Zuperl, U.; Cus, F.; Zawada-Tomkiewicz, A.; Stçpieri, K. Improvement of logistics in manufacturing system by the use of simulation modelling: 18 A real industrial case study Straka, M.; Khouri, S.; Lenort, R.; Besta, P. Estimating the position and orientation of a mobile robot using neural network framework 31 based on combined square-root cubature Kalman filter and simultaneous localization and mapping Wang, D.; Tan, K.; Dong, Y.; Yuan, G.; Du, X. A comparison of the tolerance analysis methods in the open-loop assembly 44 Kosec, P.; Skec, S.; Miler, D. Awareness and readiness of Industry 4.0: The case of Turkish manufacturing industry 57 Sari, T.; Gule§, H.K.; Yigitol, B. Assembly transport optimization for a reconfigurable flow shop based on a 69 discrete event simulation Yang, S.L.; Xu, Z.G.; Li, G.Z.; Wang, J.Y. The impact of using different lean manufacturing tools on waste reduction 81 Leksic, I.; Stefanic, N.; Veza, I. Integrated management systems based on risk assessment: Methodology development and 93 case studies Vulanovic, S.; Delic, M.; Kamberovic, B.; Beker, I.; Lalic, B. Communication and validation of metrological smart data in IoT-networks 107 Acko, B.; Weber, H.; Hutzschenreuter, D.; Smith, I. Calendar of events 118 Notes for contributors 119 Journal homepage: apem-journal.org ISSN 1854-6250 (print) ISSN 1855-6531 (on-line) ©2020 CPE, University of Maribor. All rights reserved. Scope and topics Advances in Production Engineering & Management (APEM journal) is an interdisciplinary refer-eed international academic journal published quarterly by the Chair of Production Engineering at the University of Maribor. The main goal of the APEM journal is to present original, high quality, theoretical and application-oriented research developments in all areas of production engineering and production management to a broad audience of academics and practitioners. In order to bridge the gap between theory and practice, applications based on advanced theory and case studies are particularly welcome. For theoretical papers, their originality and research contributions are the main factors in the evaluation process. General approaches, formalisms, algorithms or techniques should be illustrated with significant applications that demonstrate their applicability to real-world problems. Although the APEM journal main goal is to publish original research papers, review articles and professional papers are occasionally published. Fields of interest include, but are not limited to: Additive Manufacturing Processes Advanced Production Technologies Artificial Intelligence in Production Assembly Systems Automation Big Data in Production Computer-Integrate d M anufacturing Cutting and Forming Processes Decision Support Systems Deep Learning in Manufacturing Discrete Systems and Methodology e-Manufacturing Evolutionary Computation in Production Fuzzy Systems Human Factor Engineering, Ergonomics Industrial Engineering Industrial Processes Industrial Robotics Intelligent Manufacturing Systems Joining Processes Knowledge Management Logistics in Production Machine Learning in Production Machine Tools Machining Systems Manufacturing Systems Materials Science, Multidisciplinary Mechanical Engineering Mechatronics Metrology in Production Modelling and Simulation Numerical Techniques Operations Research Operations Planning, Scheduling and Control Optimisation Techniques Project Management Quality Management Risk and Uncertainty Self-Organizing Systems Statistical Methods Supply Chain Management Virtual Reality in Production 4 APEM jowatal Advances in Production Engineering & Management Volume 15 | Number 1 | March 2020 | pp 5-17 https://doi.Org/10.14743/apem2020.1.345 ISSN 1854-6250 Journal home: apem-journal.org Original scientific paper Neuro-mechanistic model for cutting force prediction in helical end milling of metal materials layered in multiple directions Zuperl, U.a*, Cus, F.a, Zawada-Tomkiewicz, A.b, Stçpien, K.c aUniversity of Maribor, Faculty of Mechanical Engineering, Slovenia bKoszalin University of Technology, Department of Mechanical Engineering, Poland cKielce University of Technology, Faculty of Mechatronics and Mechanical Engineering, Poland A B S T R A C T A R T I C L E I N F O In machining of multi-layer metal materials used frequently for the manufacture of transfer sheet-metal forming tools, the cutting edge is often damaged because of cutting force peaks. Therefore, a neuro-mechanistic model, presented in this paper, has been created for accurate prediction of cutting forces in helical end milling of multidirectional layered materials. The generalized model created takes into account the complex geometry of the helical end milling cutter, the instantaneous chip thickness and the direction of depositing of the individual layer of the multidirectional layered material considered in the calculation through predicted specific cutting forces. For the prediction of specific cutting forces for individual layers a neural network is incorporated in the model. The comparison with experimental data shows that the model predicts accurately the flow of cutting force in milling of multidirectional layered metal materials for any combination of cutting parameters, tool engagement angle and directions of depositing three layers of material. The predicted cutting force values agree well with the values obtained, the maximum error of predicted cutting forces is 16.1 % for all comparison tests performed. © 2020 CPE, University of Maribor. All rights reserved. Keywords: Helical end milling; Multidirectional layered metal material; Cutting forces; Specific cutting forces; Neuro-mechanistic model; Modelling; Prediction; Artificial neural networks *Corresponding author: uros.zuperl@um.si (Zuperl, U.) Article history: Received 18 April 2019 Revised 16 March 2020 Accepted 23 March 2020 1. Introduction Tool shops making resistant transfer sheet-metal forming tools use the most up-to-date multilayer metal materials. The materials are made layer after layer by the LENS (laser engineered net shaping) process [1]. In the multi-layer material, the individual layers are deposited in different directions, therefore the machinability of the layered material is changing from layer to layer. Machinability also depends on the angle between the feeding axis of the tool and direction of layer depositing. Machining of such materials is a demanding operation requiring continuous adaption of machining parameters to momentary cutting conditions. A review of literature identifies no researches in the field of machinability of multidirectional layered metal materials. There are a few researches on machinability of difficult-to-machine nickel-based alloys [2], titanium alloys [3] and composite materials [4]. M'Saoubi et al. [5] presented an overview of the recent advances in high performance cutting of aerospace alloys and composite used in aeroengine and aero structure applications. 5 Zuperl, Cus, Zawada-Tomkiewicz, Stçpien Hojati et al. [6] and Bonati et al. [7] examined the machinability of additively manufactured Ti6Al4V alloy parts in the micro-milling with a particular emphasis on cutting forces. They found out that the microstructure of the material, in addition to the hardness of the material, has a significant impact on the cutting forces. Furthermore, the cutting forces for additively manufactured part are lower than those of standard manufactured part despite of their higher hardness. Montevecchi et al. [8] analysed cutting forces in order to examine the machinability of AISI H13 alloy. Mechanistic approach was employed to identify cutting force coefficients and to investigate the behaviour of the cutting force for laser deposition (LENS). Results outlined that the additively manufactured AISI H13 material had reduced machinability compared to the same standard material at wrought state. In machining multi-layer metal materials high cutting force peaks and excessive wear and extensive damages of the cutting tool occur because of the non-uniform shape of chips [5]. The cutting force frequencies measured in milling can be directly associated with the manner of chip formation [9]. Also the roughness of the machined surface depends strongly on the manner of chip formation and is in correlation with cutting force. The tool wear and damages are directly associated with cutting forces, therefore accurate prediction and monitoring of cutting forces is a key factor assuring quality of machining. By the cutting forces, accurately predicted, the quality of machining can be evaluated and undesirable effects on the cutting tool reduced [10]. Literature comprises a lot of researches on modeling of cutting forces in oblique cutting. The majority of cutting force models for oblique cutting are created by the mechanistic modelling technique [9]. The developed mechanistic models for oblique cutting in predicting of cutting forces take into account: radial cutter runout [11]; tool deflection [12]; system dynamics and flank wear [9]; indentation of the cutting edge into the work material [13]; dynamic chip thickness, chip forming and friction forces [14]; radius of curvature for sculptured surface machining [15]; Johnson-Cook constitutive equation [16]. In the mechanistic models, the cutting forces are associated with instantaneous uncut chip thickness through experimentally defined specific cutting forces [17]. The principal challenge in creating of those models is the work-intensive acquisition of specific cutting forces for oblique cutting. Further, the acquisition of specific cutting forces for different tool and workpiece combinations requires a great number of cutting experiments and much analytical work. In modeling of multidirectional layered materials also the non-homogeneities in materials must be considered making the determination of specific cutting forces even more exacting. Models for simulating of cutting forces in orthogonal cutting of layered laminates with different directions of fibres [4, 18, 19] and quite a few models for prediction of cutting forces in helical end milling of metals are available. For example, Kline [20] created a cutting force model by dividing the helical end mill in axial direction into differentially thin elements. For each element, he calculated the differential cutting force as the product of specific cutting force and uncut chip cross-section area. With the sum of differential cutting forces for the entire cutting edge he then determined the total cutting force on the tool. Zhang et al. [21] developed a stochastic model of cutting forces in milling of fibre-reinforced ceramic matrix composites. In this model, the cutting forces are modeled by combining the simultaneous influences of randomly distributed carbon fibres and stochastic deterioration of tool wear. Grdisek et al. [22] created a generalized prediction model considering the chip and tool geometry, cutting parameters and specific cutting forces determined by a special method from orthogonal cutting data. Wang et al. [23] developed a novel analytic cutting force model of helical milling of titanium alloy. Liu et al. [24] developed a cutting force model to predict the cutting forces and torque during helical milling of AISI D2 steel. In the recent years, the artificial neural networks (ANN) for modelling of the milling process have become popular. Zuperl et al. [25] used backpropagation ANN for modelling of cutting forces in ball-end milling based on a set of input cutting conditions. Aykut et al. [26] applied ANN for predicting cutting forces in face milling of satellite Co-based alloy under dry conditions. El-Mounayri et al. [27] introduced a radial basis model for predicting cutting forces in a ball-end milling process. Al-Zubaidi et al. in his paper [28] reviewed the previous studies and investigations on the application of artificial neural network in modeling of milling processes. 6 Advances in Production Engineering & Management 15(1) 2020 Neuro-mechanistic model for cutting force prediction in helical end milling of metal materials layered in multiple ... Balasubramanian et al. [29] performed analysis of cutting forces in helical ball end milling of Ti-6Al-4V alloys using deep neural network. On the basis of the results obtained, he concluded that the accuracy of the predicted forces was more than sufficient for any practical purpose. The models based on neural networks prove to be very accurate in predicting by searching for correlations between cutting parameters and cutting forces, whereas they are not successful in predicting cutting forces on the basis of different properties of workpiece material. Our paper discusses the generalized neuro-mechanistic model for accurate prediction of cutting forces in helical end-milling of multidirectional layered materials. The generalized model consists of a mechanistic prediction model of cutting forces for complex cutting tool geometry and an artificial neural network for prediction of specific cutting forces of a particular layer. The relevant scientific contribution of this paper is the presented methodology to build a predictive cutting force model for edge milling of multidirectional layered materials with complex cutting tool geometries. The methodology is not limited to one cutting tool; it can be extended to all other complex cutting tools where the instantaneous uncut chip thickness and direction of the layer deposition can be determined. The paper is organized as follows: Section 2 presents the methodology of creating of combined neuro-mechanistic model for prediction of cutting forces in helical end-milling of multidirectional layered material. Subection 2.1 outlines an overview of the proposed cutting force model. Subection 2.2 presents the developed mechanistic model for prediction of cutting forces by considering the geometry of the helical end milling cutter, instantaneous chip thickness and direction of depositing of individual layers of multi-layer material. Subsection 2.3 describes the artificial neural network based modeling and predicting of specific cutting forces for each direction of deposition of individual layer of layered material. Subsection 2.4 outlines the experimental set-up. In Section 3 the results of comparison of the model prediction with experimental data are analyzed. Section 4 gives the concluding remarks. 2. Materials and methods A quantitative study design was undertaken in three phases. In the first phase, edge milling of metal materials layered in different directions was executed in order to obtain orthogonal data base (milling force data) for determining specific cutting forces. Mechanistic technique was employed to predict specific cutting forces. In the second phase, an artificial neural network was built to estimate specific cutting forces. In the third phase, a generalized cutting force model was developed based on mechanistic modeling technique [17, 24]. The model was verified by comparing predicted and experimentally acquired values. The average percentage error (APE) was employed to evaluate the accuracy of the neural and cutting force model. 2.1 Overview of proposed cutting force model for multidirectional layered material Hereinafter, a combined neural-mechanistic model for prediction of cutting forces in helical end-milling of multidirectional layered metal material is presented. The model is capable of predicting the cutting forces for selected combination of cutting parameters, immersion angle and direction of deposition of individual workpiece layers. For creation of the model, the mechanistic modeling technique of complex tool geometry, dividing the helical tool cutting edge in axial direction into a great number of equal differential elements with angle offset, is used. By the use of trained artificial neural network the radial and tangential cutting force coefficients (Kr, Kt) are evaluated for each element. Based on evaluated instantaneous coefficients Kr and Kt the radial dFr and tangential dFt differential forces for each individual differential element are calculated. In the end, the total force on the cutting tooth is calculated by integrating differential forces on all elements of the cutting edge. The force on tool is determined by the sum of cutting forces on all tool cutting edges. 2.2 Creation of cutting force model for helical end mill Cutting forces are calculated for the end mill with helix angle ft, diameter D and number of cutting edges N. The milling cutter with designated cutting force direction and differential segments Advances in Production Engineering & Management 15(1) 2020 7 Zuperl, Cus, Zawada-Tomkiewicz, Stçpien is shown in Fig. 1. The two elementary forces dFt and dFr on the differential element L are determined on the basic of instantaneous values Kr and Kt, instantaneous uncut chip thickness hL and milling width. Tangential force dFt and radial force dFr acting on the differential element L of dz height are calculated according to Eqs. 1 and 2: dFr(0L,z) = [Kre + Kr -hL(0L,z)] • dz dFt{$L,z) = [Kte + Kt -hL(0L,z)] • dz (1) (2) Fig. 1 Geometry of helical end milling of multidirectional layered material with relevant cutting forces and tool dissection into differential elements The immersion (engagement) angle for differential element L on tooth j at an axial depth of cut z is calculated according to: (z) = 0 + (3) In the calculation, it is assumed that the bottom part of the cutting edge j = 0 has reference immersion angle <. The bottom end points of the other cutting edges are offset from reference cutting edge by angle < = 2n/N. This relation is written with Eq. 4: tj(z = 0) = f + jfy y = 0,1.....(iV-1) (4) where < is the angular pitch of cutting edge. Lag angle for differential element L for axial cut depth z is calculated Eq. 5: 2tan ß hL for the differential element is calculated according to Eq. 6. hL(0L,z) = fz -sinL; fz=-rz (5) (6) where f is the feed rate and n is the rotational tool speed. The cutting force components with zero chip thickness are designated as Kte and Kre. The effect of cutting tool wear is not considered because sharpened tool is used for each pass. 8 Advances in Production Engineering & Management 15(1) 2020 Neuro-mechanistic model for cutting force prediction in helical end milling of metal materials layered in multiple ... By considering the cutting trigonometry the differential axial cutting force dFa for L-th differential element is determined on the basis of dFt and dFr according to Eq. 7: dFr (sin ft — cos ft • sin an - tan ri) — dFt • cos an - tan r) h, (w, ,z) = fy - sin < dF„ (w, ,z) =------(7) ml. j iz rL aWL. j sin p • sin an • tan ^ + cos ft 1 J where p is the inclination angle of cutting edge, an the relief angle and rj is the rake angle. The radial and tangential coefficient of cutting force depend on the direction of layer deposition of layered material and uncut chip thickness determined for the instantaneous engagement angle < according to Eq. 6. The angle of instantaneous direction of deposition of layers of layered material for differential element L is determined according to Eq. 8: 0< (f)L 180 — ^ ^ = 0 The angle of deposition of a gradient layer of material is designated (8) Fig. 2 Algorithm for prediction of cutting forces in helical end milling Advances in Production Engineering & Management 15(1) 2020 9 Zuperl, Cus, Zawada-Tomkiewicz, Stçpien Transformation of elementary force dFr and dFt into differential feeding force dFy and normal force dFx by taking into account the cutting trigonometry is determined according to following equations: dFx«L,z) = dFt - cos <(z) — dFr - sin<(z) (9) dFy«L,z) = dFt - sin<(z) + dFr - cos <(z) (10) dFz«L,z) = dFa (11) By integrating the differential forces on the active part of cutting edge (from 0 to AD) the total cutting force of one cutting edge j is determined. Total cutting forces generated by the tool are determined by the addition of cutting forces on all differential elements of cutting edges. The following expressions are used: ZZN-l •S-^Ao y-^N-l a ¿iz odFy; ^0 = 2, dFz; (12) 0< <<< Fig. 2 shows the algorithm for prediction of cutting forces in milling with helical end milling cutter. 2.3 Neural model of specific cutting forces for unidirectional layered material This chapter presents the methodology of modelling and predicting of specific cutting forces for unidirectional layered metal material by the use of ANN. Specific cutting force in milling of multilayer materials depends on the direction of deposition of individual layers, instantaneous chip thickness, cutting speed and tool wear. For a certain pair tool-workpiece the specific cutting force links the cutting parameters to relevant radial and tangential cutting force. For the individual deposited metal layer of material with defined instantaneous direction of deposition 9 its amplitude is calculated according to the following expressions: (13) In the equation, Fr and Ft are measured cutting forces, the expression in the denominator is the calculated uncut chip area. The experimental data set for calculation of specific cutting forces is obtained by measurement of cutting forces in milling of multi-layer metal workpieces. All layers of the individual workpieces were made with equal direction of deposition. To train, validate and test the neural model, a total of 54 cutting experiments with three spindle speed levels, three feed rate levels, two axial depth of cutting levels and three different workpiece configurations were conducted. The data set used for training, validating and testing the neural model consists of 1890 data points. The total data set was split into input and output subsets. The input subset consists of spindle speed n, feed rate f, axial depth of cutting AD, radial depth of cutting RD, the direction of the material layer deposition uncut chip thickness h and hardness of the machined material HV. The output set consists of radial and tangential specific cutting forces. Further, the total data was randomly divided into training/validation (1260 data points) and testing data subsets (630 data points). A ten-fold cross-validation is used to validate the ANN model. Therefore, 1260 data points were randomly partitioned into 10 equal blocks (folds). Then, ANN models were systematically trained on 9 folds and validated on the remaining fold; the cross-validation process was repeated 10 times. The prediction errors of the trained ANN models were averaged and 95 % confidence interval was determined. After the validation process, a subset of 1260 data points was used to train a new ANN model. Finally, a testing data subset (630 data points) was used to test the developed ANN model. To evaluate the accuracy of the created models, the average percentage error (APE) calculated according to Eq. 14 is used: 10 Advances in Production Engineering & Management 15(1) 2020 Neuro-mechanistic model for cutting force prediction in helical end milling of metal materials layered in multiple ... |^target,i ^predicted!| _ ^qq % K (14) ^target/ I ¿=1 / where Ktargeti is the experimentally determined specific cutting force component, Kpredictedi is predicted specific cutting force component in radial and tangential direction generated by ANN and n is the number of testing data points. For modeling, a four-layer feed forward ANN with backpropagation learning algorithm is used. The rate of training 0.12 and the momentum rate 0.08 are selected for training, while for the transfer of signals between neurons the arctan transfer function is chosen. The input vector of neural network consists of spindle speed n, feed rate f, axial depth of cutting AD, radial depth of cutting Rd, the direction of the material layer deposition uncut chip thickness h and hardness of the machined material HV. The value HV did not change in the machining experiment. The output vector ANN consists of radial and tangential specific cutting forces. The ANN architecture shown in Fig. 3 and the optimum training parameters were determined through simulations. The ANN training process was completed, when 8500 iterations of training had been performed or when the prediction error has fallen below the pre-defined limit value (0.01). ■E Input vector n f h AD v Output vector Kt Kr Fig. 3 Detailed structure of ANN used for prediction of components of specific cutting forces depending on material layer deposition direction 2.4 Experimental set-up To build the cutting force model, a set of machining experiments was performed on the Heller BEA01 milling machine according to the experimental plan. Cutting conditions of performed expriments are described in Table 1. The orthogonal milling experiments with cutting conditions described in Table 1 have been carried out in order to train and test the ANN model. For determinig the specific cutting forces the stright one tooth cutting tool with a diameter of 16 mm was used. The carbide tool had 3.8° rake angle. Cooling-lubricating agents were not used. Four complex milling experiments with the same cutting conditions have been carried out in order to obtain the data for verifying the developed generalized neuro-mechanistic model. The model was verified for a segmented helical cutting tool in edge milling of unidirectional and multidirectional workpieces. In these experiments the carbide helical cutting tools with a diameter of 8.5 mm and two flutes were used. The tool from sintered tungsten carbide had 27.3° helix angle and 4.28° rake angle. Cutting edges had PVD-TiAIN coating and hardness of 1820 HV. Cooling-lubricating agents were not used. In all helical milling experiments the same tool rotational frequency 3800 min1 and the same feed rate 200 m/min were selected. The cut depth AD was adjusted to 1.8 mm and the radial cut depth RD to 0.5 mm. Advances in Production Engineering & Management 15(1) 2020 11 Zuperl, Cus, Zawada-Tomkiewicz, Stçpien Table 1 Cutting conditions of performed machining experiments and material layer deposition directions Parameter Orthogonal milling Helical milling n [min-1] 3500, 3800, 4100 3800 f [m/min] 200, 250, 300 200 Rd [mm] 1 0.5 Ad [mm] 1.6, 1.8 1.8 D [mm] 16 5.5 Layer depositing direction ^ [o] 90, 180, 135 90, 180, 135, 90/135/180 Number of teeth 1 2 Helix angle [o] 0 27.3 Radial angle [o] 3.8 4.28 Three cutting force components were measured with Kistler piezo-electric dynamometer. The signals of the measured forces were processed with dual mode charge amplifier and low pass filter of 2 kHz cut-off frequency to eliminate noises induced by vibrations of adjacent systems. Adequacy of frequency Bandwidth of dynamometer for all cutting force frequencies in the experiment was confirmed by calculations. Dynamic compensation of measured cutting forces was not employed at low tooth frequencies. Measured signals were transferred to data acquisition with Labview software. Workpiece of 50 mm length, 15 mm width and 21.8 mm height was clamped to the dynamometer. Two types of workpieces shown in Fig. 4 were used in machining experiments. The first, i.e. unidirectional workpiece type (Fig. 4a) was made from 16MnCr5 basic substrate and several stainless steel (316L) layers with a singular 0.6 mm thickness. All gradient layers were deposited in the same direction. Identical workpieces, properly machined and rotated on Z axis in respect to feed direction for 90°, 135° and 180°, were used in orthogonal milling experiments. The other, i.e. multidirectional workpiece type (Fig. 4b) was used only in complex helical milling experiment. It was made from a 16MnCr5 basic substrate and three stainless steel (316L) layers deposited in different directions. Each new layer was deposited at 45° angle with respect to the direction of deposition of previous layer. Thus, individual layers were deposited at 90°, 135° and 180° angles. Thickness of individual layers made was 0.6 mm with measured hardness 288 HV. Overlapping of laser trajectories was adjusted to 40 %. Diameter of the effective laser beam was 0.9 mm. Fig. 4 shows the structure and orientation of used test workpieces in machining experiments. a) Unidirectional workpieces b) Multidirectional workpiece Fig. 4 Structure and position of used unidirectional and/or multidirectional multi-layer workpieces in machining experiments 12 Advances in Production Engineering & Management 15(1) 2020 Neuro-mechanistic model for cutting force prediction in helical end milling of metal materials layered in multiple ... 3. Results and discussion Many experimental tests with different cutting conditions have been carried out in order to validate the developed generalized neuro-mechanistic model. A sample of predicted and measured cutting forces in helical milling of unidirectional and multidirectional layered metal material is shown in Figs. 5 and 6. Samples of force values in normal, feed and axial direction are shown diagrammatically depending on the engagement angle. The measured signals of cutting forces, representing the average value of 3 measurements, are represented with full line and predicted values with dashed line. From the flow of cutting force signals it is possible to discern an obvious peak of the cutting force with the tip corresponding to one cutting-off cycle of one cutting tool edge. The cutting force peak results from the increase of chip cross-sectional area on the cutting edge from 0 upon the entry of the cutting edge into material up to maximum value upon the exit of the cutting edge from workpiece. The value and form of the cutting force signal are affected by the layer deposition on the multi-layer material. The instantaneous deposition direction relative to the tool cutting edge changes with the engagement angle. From the flow of cutting force signals in Fig. 5 it can be discerned that the cutting force values in milling of multi-layer stainless steel with 90° deposition direction are greater than when milling the material with 180° deposition direction. Also a smoother flow of cutting force signals with fewer fluctuations can be discerned. Contacts between deposited layers contribute to formation of very small broken chips. The increase of the deposition direction angle from 90° toward 135° results in strong increase of cutting force fluctuations and magnitude. Most unfavourable cutting forces occur in machining the material with deposition direction 135° (Fig. 6a), where a slightly greater chip cross sectional area of incorrect shapes and from time to time partly continuous appear. Bad cutting conditions might be ascribed to sticking of the cutting edge tip between individual deposited layers. The cutting force magnitudes in milling of unidirectional multi-layer material with deposition direction from 135° towards 180° and partly also fluctuations are decreasing. Greater fluctuations of axial force Fz can be observed in machining the material with 180° deposition direction (Fig. 5a). More favourable, somewhat larger, broken chips appear. When milling a multidirectional three-layer workpiece (90°/180°/135°) two distinct cutting force tops (Fig. 6b) can be observed. They result from great changes of instantaneous directions of deposited layers on the tool cutting edge, when it passes between individual stainless steel layers. Because of the low RD, only one cutting edge cuts at a time. At the beginning of tool rotation, the tool cutting edge moves vertically on the entire thickness of layer 3. Afterwards, it passes over the layer 2 and, in the end, it gets out of the layer 1 at engagement angle 25°. During gradual cutting of the lowest layer 3 with 90° deposition direction the first peak of all three cutting force components (Fig. 6b) occurs. A decrease of cutting forces, coinciding with cutting of the middle layer with 180° deposition direction, follows. During cutting of the highest layer with 135° deposition direction the second, somewhat smaller, peak of the cutting force component occurs. The greatest magnitudes and fluctuations of cutting force values occur when machining this layer particularly because of greater cross and unfavourable form of the chip and rough tearing of the chip from workpiece. Machining of the highest layer with 130° deposition direction contributes most to the total cutting force shown in Fig. 6b and machining of the middle layer with 180° deposition direction contributes least. The sizes, sequences and distances of peaks of measures cutting forces depend on the sequence and thicknesses of deposited layers constituting the multidirectional multilayer metal material. Comparison between measured and predicted forces for helical end milling of unidirectional multi-layer material is shown in Figs. 5a, 5b and 6a. Fig. 6b shows the comparison between measured and predicted forces for milling of multi-layer material with layer deposition directions (90°/180°/135°). Results in Figs. 5 and 6 witness that the predicted force values agree well with experimentally obtained values in milling of 180° and 135° unidirectional multi-layer workpiece. A slightly smaller agreement of predicted and measured forces is discerned in milling of unidirectional workpiece with 90° direction of laser-deposited layers. Advances in Production Engineering & Management 15(1) 2020 13 Zuperl, Cus, Zawada-Tomkiewicz, Stçpien a) 250 200 150 - 100 50 - 10 20 30 Engagement angle <|> [°] 0 10 20 Engagement angle ty [°] Fig. 5 Comparison between experimental and predicted cutting forces for unidirectional three-layer metal material with: a) with 1800 direction of layer deposition b) with 900 direction of layer deposition 250 200 -h 250 b) ------------ — J /* f 7/ a // V 7 1J igÜ^ -pr VI. / V S m V 'J V >/ r / // j // / // J /7 / / A ''Fv ' M L \%/_________ // // // // '7 7 / f/ / ' y / / a 7/ f // Y v / j // 7/ ft // // /VS s / /y? «" / /£r ' 7 y 1-'-\ 10 20 Engagement angle <|) [°] 0 10 20 30 Engagement angle fy [°] Fig. 6 Comparison between experimental and predicted cutting forces for unidirectional and multidirectional three-layer metal material: a) with 1350 direction of layer deposition b) with 90°/180°/135° directions of layers deposition The greatest deviation between the model predictions and experimental values of forces occurs in milling of multidirectional workpiece with layer deposition directions (90°/180°/135°). The results differ as follows: from 2.9 % to 8.6 % for Fx, from 3.6 % to 11.7 % for Fy and from 3.6 % to 16.1 % for Fz. The results in Figs. 5 and 6 show that the developed model appropriately predicts the general course of cutting forces in milling of multi-layer materials. 0 0 14 Advances in Production Engineering & Management 15(1) 2020 Neuro-mechanistic model for cutting force prediction in helical end milling of metal materials layered in multiple ... Table 2 Accuracy of ANN prediction model ANN Kr Kt APE [%]; Training set 4.29 4.25 APE [%]; Testing set 5.21 3.55 a) b) h [mm] h [mm] Fig. 7 Comparison of Kr and/or Kt predicted by neural model and obtained experimentally with four different directions of layer deposition; f = 200 m/min, n = 3800 min-1 The accuracy of the ANN model for predicting specific cutting forces was analysed using the APE method. The results of analysis of the model capacity are given in Table 2. The uncertainty of the predicted results, obtained from the ten-fold cross-validation, can be characterized as small, consequently it was concluded that the ANN adequately predicted Kr and Kt. The maximal prediction error of the ANN model is 5.21 % for Kr, and 3.55 % for Kj- with a 95 % confidence interval. A part of predicted values Kt and Kr depending on the layer deposition angle and uncut chip thickness is presented in two graphs in Fig. 7. Both graphs show that the predicted values of specific components of cutting forces in radial and tangential direction strongly agree with experimentally obtained values. The two specific cutting forces decrease with the increase of the engagement angle and/or instantaneous uncut chip thickness. The greatest values are reached with layer deposition angle 135°. 4. Conclusion The paper presents a combined neural-mechanistic model of the process of helical milling of multidirectional layered metal materials. For creation of model the mechanistic technique of cutting force prediction was used by dividing the helical end mill into a series of differentially thin axial elements with oblique cutting edges and calculating the differential cutting force for each element. To that end, it uses specific cutting forces predicted by the artificial neural network and incorporates in the calculation the deposition direction of individual layer of multidirectional layered metal material. The total cutting force for any immersion tool angle is calculated by integrating the differential cutting forces on the height of the tool cutting edge in contact with the multi-layer workpiece. The model is capable of predicting the cutting forces in milling of multidirectional and unidirectional metal materials with complex geometry tools. Many cutting experiments have been carried out to validate the developed model. The following conclusions can be drawn from these experiments: Advances in Production Engineering & Management 15(1) 2020 15 Zuperl, Cus, Zawada-Tomkiewicz, Stçpien • The magnitude of cutting forces in milling of multidirectional layered metal material is affected by the layer deposition direction and immersion angle of cutting tool determining the instantaneous cross-section of the chip. • The developed model appropriately predicts the general course of cutting forces during milling of multidirectional metal materials. It can be recapitulated that the model created is generally efficient (valid) for any combination of cutting parameters and direction of workpiece layer deposition. • Accuracy of the model prepared is high in milling of unidirectional metal materials and slightly worse in milling of the workpiece with multidirectional laser deposited layers. • The neural network incorporated in the model predicts the specific cutting forces for any instantaneous direction of deposition of material layers, therefore the proposed model is widely applicable and capable of simulating the cutting forces on the helical milling tool for multidirectional layered metal material. • The predicted cutting force values agree well with experimentally obtained values in milling 180° and 135° unidirectional multi-layer workpiece. A slightly smaller agreement between predicted and measured forces is found in milling unidirectional workpiece with 90° direction of layer deposition. These minor errors are not random. They might appear due to the failure to take into account the entire cutting tool geometry when calculating the specific cutting forces. • The results differ as follows: from 2.9 % to 8.6 % for Fx, from 3.6 % to 11.7 % for Fy and from 3.6 % to 16.1 % for Fz. • The maximum percentage prediction cutting force is found to be less than 16.1 % for all the cases tested. References [1] Mahmoud, E.R.I. (2015). Characterizations of 304 stainless steel laser cladded with titanium carbide particles, Advances in Production Engineering & Management, Vol. 10, No. 3, 115-124, doi: 10.14743/apem2015.3.196. [2] Tabernero, I., Lamikiz, A., Martinez, S., Ukar, E., Figueras, J. (2011). Evaluation of the mechanical properties of Inconel 718 components built by laser cladding, International Journal of Machine Tools and Manufacture, Vol. 51, No. 6, 465-470, doi: 10.1016/uimachtools.2011.02.003. [3] Jia, Z.-Y., Ge, J., Ma, J.-W., Gao, Y.-Y., Liu, Z. (2016). A new cutting force prediction method in ball-end milling based on material properties for difficult-to-machine materials, The International Journal of Advanced Manufacturing Technology, Vol. 86, No. 9-12, 2807-2822, doi: 10.1007/s00170-016-8351-8. [4] Sheikh-Ahmad, J., He, Y., Qin, L. (2019). Cutting force prediction in milling CFRPs with complex cutter geometries, Journal of Manufacturing Processes, Vol. 45, 720-731, doi: 10.1016/i.imapro.2019.08.009. [5] M'Saoubi, R., Axinte, D., Soo, S.L., Nobel, C., Attia, H., Kappmeyer, G., Engin, S., Sim, W.-M. (2015). High performance cutting of advanced aerospace alloys and composite materials, CIRP Annals, Vol. 64, No. 2, 557-580, doi: 10.1016/i.cirp.2015.05.002. [6] Hoiati, F., Daneshi, A., Soltani, B., Azarhoushang, B., Biermann, D. (2020). Study on machinability of additively manufactured and conventional titanium alloys in micro-milling process, Precision Engineering, Vol. 62, No. 1-9, doi: 10.1016/i.precisioneng.2019.11.002. [7] Bonaiti, G., Parenti, P., Annoni, M., Kapoor, S. (2017). Micro-milling machinability of DED additive titanium Ti-6Al-4V, Procedia Manufacturing, Vol. 10, 497-509, doi: 10.1016/i.promfg.2017.07.104. [8] Montevecchi, F., Grossi, N., Takagi, H., Scippa, A., Sasahara, H., Campatelli, G. (2016). Cutting forces analysis in additive manufactured AISI H13 alloy, Procedia CIRP, Vol. 46, 476-479, doi: 10.1016/i.procir.2016.04.034. [9] Song, G., Sui, S., Tang, L. (2015). Precision prediction of cutting force in oblique cutting operation, The International Journal of Advanced Manufacturing Technology, Vol. 81, No. 1-4, 553-562, doi: 10.1007/s00170-015-7206-z. [10] Yang, L., Zheng, M.L. (2017). Simulation and analysis of ball-end milling of panel moulds based on Deform 3D, International Journal of Simulation Modelling, Vol. 16, No. 2, 343-356, doi: 10.2507/IISIMM16(2)CQ9. [11] Sun, Y., Guo, Q. (2011). Numerical simulation and prediction of cutting forces in five-axis milling processes with cutter run-out, International Journal of Machine Tools and Manufacture, Vol. 51, No. 10-11, 806-815, doi: 10.1016/i.iimachtools.2011.07.003. [12] Qmar, Q.E.E.K., El-Wardany, T., Ng, E., Elbestawi, M.A. (2007). An improved cutting force and surface topography prediction model in end milling, International Journal of Machine Tools and Manufacture, Vol. 47, No. 7-8, 12631275, doi: 10.1016/i.iimachtools.2006.08.021. [13] Tuysuz, O., Altintas, Y., Feng, H.-Y. (2013). Prediction of cutting forces in three and five-axis ball-end milling with tool indentation effect, International Journal of Machine Tools and Manufacture, Vol. 66, 66-81, doi: 10.1016/i.iimachtools.2012.12.002. 16 Advances in Production Engineering & Management 15(1) 2020 Neuro-mechanistic model for cutting force prediction in helical end milling of metal materials layered in multiple ... [14] Qu, S., Zhao, J., Wang, T., Tian, F. (2015). Improved method to predict cutting force in end milling considering cutting process dynamics, The International Journal of Advanced Manufacturing Technology, Vol. 78, No. 9-12, 1501-1510, doi: 10.1007/s00170-014-6731-5. [15] Cao, Q., Zhao, J., Li, Y., Zhu, L. (2013). The effects of cutter eccentricity on the cutting force in the ball-end finish milling, The International Journal of Advanced Manufacturing Technology, Vol. 69, No. 9-12, 2843-2849, doi: 10.1007/s00170-013-5205-5. [16] Daoud, M., Chatelain, J.F., Bouzid, A. (2017). Effect of rake angle-based Johnson-Cook material constants on the prediction of residual stresses and temperatures induced in Al2024-T3 machining, International Journal of Mechanical Sciences, Vol. 122, 392-404, doi: 10.1016/i.iimecsci.2017.01.020. [17] Li, Y., Yang, Z.J., Chen, C., Song, Y.X., Zhang, J.J., Du, D.W. (2018). An integral algorithm for instantaneous uncut chip thickness measuring in the milling process, Advances in Production Engineering & Management, Vol. 13, No. 3, 297-306, doi: 10.14743/apem2018.3.291. [18] Karpat, Y., Polat, N. (2013). Mechanistic force modeling for milling of carbon fiber reinforced polymers with double helix tools, CIRP Annals, Vol. 62, No. 1, 95-98, doi: 10.1016/i.cirp.2013.03.105. [19] Kalla, D., Sheikh-Ahmad, J., Twomey, J. (2010). Prediction of cutting forces in helical end milling fiber reinforced polymers, International Journal of Machine Tools and Manufacture, Vol. 50, No. 10, 882-891, doi: 10.1016/i.iimachtools.2010.06.005. [20] Kline, W.A., DeVor, R.E., Lindberg, J.R. (1982). The prediction of cutting forces in end milling with application to cornering cuts, International Journal of Machine Tool Design and Research, Vol. 22, No. 1, 7-22, doi: 10.1016/0020-7357(82)90016-6. [21] Zhang, X., Yu, T., Zhao, J. (2020). An analytical approach on stochastic model for cutting force prediction in milling ceramic matrix composites, International Journal of Mechanical Sciences, Vol. 168, Article No. 105314, doi: 10.1016/i.iimecsci.2019.105314. [22] Gradisek, J., Kalveram, M., Weinert, K. (2004). Mechanistic identification of specific force coefficients for a general end mill, International Journal of Machine Tools and Manufacture, Vol. 44, No. 4, 401-414, doi: 10.1016/i.iimachtools.2003.10.001. [23] Wang, H., Qin, X., Ren, C., Wang, Q. (2012). Prediction of cutting forces in helical milling process, The International Journal of Advanced Manufacturing Technology, Vol. 58, No. 9-12, 849-859, doi: 10.1007/s00170-011-3435-y. [24] Liu, C., Wang, G., Dargusch, M.S. (2012). Modelling, simulation and experimental investigation of cutting forces during helical milling operations, The International Journal of Advanced Manufacturing Technology, Vol. 63, No. 9-12, 839-850, doi: 10.1007/s00170-012-3951-4. [25] Zuperl, U., Cus, F., Mursec, B., Ploi T. (2006). A generalized neural network model of ball-end milling force system, Journal of Materials Processing Technology, Vol. 175, No. 1-3, 98-108, doi: 10.1016/i.imatprotec.2005. 04.036. [26] Aykut, §., Golcu, M., Semiz, S., Ergur, H.S. (2007). Modeling of cutting forces as function of cutting parameters for face milling of satellite 6 using an artificial neural network, Journal of Materials Processing Technology, Vol. 190, No. 1-3, 199-203, doi: 10.1016/i.imatprotec.2007.02.045. [27] El-Mounayri, H., Briceno, J.F., Gadallah, M. (2010). A new artificial neural network approach to modeling ball-end milling, The International Journal of Advanced Manufacturing Technology, Vol. 47, No. 5-8, 527-534, doi: 10.1007/s00170-009-2217-2. [28] Al-Zubaidi, S., Ghani, J.A., Haron, C.H.C. (2011). Application of ANN in milling process: A review, Modelling and Simulation in Engineering, Vol. 2011, No. 9, Article ID 696275, doi: 10.1155/2011/696275. [29] Balasubramanian, A.N., Yadav, N., Tiwari, A. (2020). Analysis of cutting forces in helical ball end milling process using machine learning, Materials Today: Proceedings, In press, doi: 10.1016/i.matpr.2020.02.098. Advances in Production Engineering & Management 15(1) 2020 17 Advances in Production Engineering & Management Volume 15 | Number 1 | March 2020 | pp 18-30 https://doi.Org/10.14743/apem2020.1.346 ISSN 1854-625G Journal home: apem-journal.org Original scientific paper Improvement of logistics in manufacturing system by the use of simulation modelling: A real industrial case study Straka, M.a*, Khouri, S.b, Lenort, R.c, Besta, P.d aInstitute of Logistics and Transport, Technical university of Kosice, Kosice, Slovak Republic bInstitute of Earth Resources, Technical university of Kosice, Kosice, Slovak Republic cDepartment of Logistics, Quality and Automotive Technology, ŠKODA AUTO University, Mlada Boleslav, Czech Republic dDepartment of Economics and Management in Metallurgy, VŠB - Technical University of Ostrava, Ostrava, Czech Republic A B S T R A C T A R T I C L E I N F O The current practice and the requirements of industrial enterprises in all industrial areas require a detailed display of manufacturing systems course of events. In this paper, we studied the effects and impacts of computer simulation to improve the actual industrial production. We also verified whether the proposed simulation model and its intervention in the logistics of concrete production in a concrete manufacturing enterprise will correspond to reality. The EXTENDSIM simulation software was used. The simulation results utilization in practice has increased the actual production several times. The simulation results indicated that it is necessary to double the intensity of company supply, i.e. a frequency of entry set to 0.15 days for each timber type. This adjustment increased the performance of unutilized devices and the whole manufacturing system several times, up to 54,475 produced building timber elements, which represents an increase of production by about 199.6 % while maintaining company flexibility. © 2020 CPE, University of Maribor. All rights reserved. Keywords: Manufacturing; Logistics; Simulation; Modelling; Optimization; EXTENDSIM *Corresponding author: martin.straka@tuke.sk (Straka, M.) Article history: Received 29 September 2017 Revised 24 February 2020 Accepted 16 March 2020 1. Introduction The solution of manufacturing logistics problems in specific market area requires the utilization of the means of algorithmization, heuristics, mathematical statistics, modelling and computer simulation. A combination of the defined approaches is possible for the designing, and it solves the problems of any manufacturing system. The objective of the logistics solutions is to influence and manage material flows in the right way. The execution of a concrete manufacturing system from a complex point of view of all types of flows is possible only by taking advantage of computer simulation. The application of mathematical modelling for the designing of material, information and financial flows together is practically impossible (and what about the effect of the human factor, in determination, haphazard and so on). All these facts make computer simulation the perfect facility (in many tasks, the only possible one) for the solution of specific problems of manufacturing systems. The foregoing results clearly show the importance of computer simulation for improving the efficiency of logistics in manufacturing. The problem relates to an effective utilization of the computer simulation with EXTENDSIM simulation system to improve the efficiency of logistics in manufacturing of concrete production company, engaged in the processing of raw timber and the production of building timber in the form of beams, planks, and boards from different types of wood, such as oak, beech, spruce and pine. 18 Improvement of logistics in manufacturing system by the use of simulation modelling: A real industrial case study Timber production, in general, belongs to the most common types of industrial productions. Timber as a building material is equally popular in all the continents of the world. The methods of processing raw timber are identical, and they work on the same, respectively similar devices under the same, respectively similar conditions. This implies the importance of tackling this issue for the industrial sector in question and the universal character of the usage of this approach to increase the efficiency of logistics in manufacturing by means of computer simulation and the EXTENDSIM simulation system. The aim of this paper is to clearly define a procedure which must be followed to solve equal or similar types of tasks through the utilization of concrete EXTENDSIM simulation system serving the needs of industrial practice. The tasks that must be executed are related to some given resolution procedures that are generally applicable to streamline the logistics in manufacturing using computer simulation. The quality of the solutions and their impacts on the manufacturing system depend on the quality of the performed analysis, quality and accuracy of data acquired for the purpose of the simulation model, and the knowledge and experience of the creator of the simulation model with the work with a concrete simulation system. The subsequent correct evaluation of the results and the definition of conclusions for the implementation of measures in practice are equally important as well. 2. Literature review In order to be able to understand the essence of the functioning of the means of simulation, it is necessary to define and understand the basic principles and relations of simulation and modelling and the parts concerned. The goal of a simulation is to obtain information about the system and its elements according to presumed or defined change of characteristics of the elements and their connection. A simulation of the systems makes it possible to experiment besides real objects, resp. these objects do not have to exist Simulation as a term means an imitation of some real situation, thing, condition, action, event of process [1, 2]. Simulation is a research method, the essence of which consists of the replacement of an essential dynamic system by its model, simulator, and we perform experiments with the purpose of getting the information about the essential system. Using a simulation, it is possible to obtain answers to some very important questions, such as: "How is the system going to behave when the following (any) changes happen? Which part of the system contains critical points? How long is the distance the conveyor must overcome if two more stations are added in the system? How long will it take to run the production? How many people are needed to meet the deadline of the project?". In a simplified way, it is possible to characterize simulation by three steps: • design of the simulation model of a real system, • carrying out the experiments with the simulation model, • reversed use of the obtained results to improve the system. In our view, the objects of the simulation will be production processes representing the processes in the group of aggregates, machines and devices, on which the production processes are progressing. The amount of the production operations depends on the type of product and on the used devices, on which the production operations are taking place. Continuous production processes are the processes leading to a smooth, continuous change of the state of the production process (work of the breaker, mills, and conveyor belts). Discrete (discontinuous) production processes are the processes leading to a change of the state of the production process at certain time moments (transport of material by cars, wagons, loading of materials by an excavator). Combined production processes are characterized by combining the discrete and continuous production processes (transport of material by cars from quarry into a ball crusher). The origins of the creation of simulation models goes back to the 1950's, when universal, general programming systems were used to create simulation models. A gradual development and progress is every area of production, science and practice required increasingly difficult Advances in Production Engineering & Management 15(1) 2020 19 Straka, Khouri, Lenort, Besta simulation models and applications. General programming systems proved to be maladroit and not dynamic enough to create fast changes in the simulation models. A special category of programming systems has gradually been invented. These systems are called simulation systems. Simulation systems are adapted for the purposes of simulation. They allow us to create program simulation models in such form to create models representing the simulated systems [3-5]. The simulation models allow us to perform quick changes in the created models, in case they are needed, in such a way that the created models correspond to the changes that can occur in a real system. The evolutional phases of the simulation systems change in time and naturally evolve [6]. The recent state is characterized by object and realistically oriented simulation with the support of 3D animations and video on a professional level. According to the functional aspects of the simulation, simulation systems are oriented to activities, events and processes, wherein a change of state can occur continuously, discretely or in a combined form [6-8]. Each area of problem solving relating to logistics combines and uses the required methodology, methods, procedures and tools specific to the given area, but also the methods of other areas. The variability of some solutions within the logistics clearly predetermines the usage of methods based on the principle of multi-criteria decision-making, statistics, modelling, the principles of heuristics and computer simulation [9-11]. To achieve the highest performance with maximum production efficiency, logistics, from the strategic, tactical and operational levels, defines respectively proposes actions that lead to achievement of the required results by using all the available means of science and technology, economics and computer science [12, 13]. The aim of logistics is to create a united, integrated, optimized material flow, which is arisen from different parts of the system in the way to ensure a continuous exchange of goods and services [14]. Logistics has gradually been developing and many definitions have also been developed together with it, while new perspectives on its scope and its level of activity are still being formed. According to Hesket, Glaskowsky and Ivie [15], logistics is the management of all activities that facilitate the movement and coordination of offer and demand in creating of time and place benefits. According to Schulte [16], logistics is an integrated, market-oriented planning, creation, implementation and control of flows of material, goods, information from suppliers to enterprises, in enterprises and from enterprises to clients at optimal costs. According to Malin-dzak [17], logistics is the way, philosophy of flows management (material, information and financial), at which there are applied systematic approach, methods of planning, algorithmic thinking and coordination in order to achieve the global optimization. According to Straka [18], logistics is a system in which there is an affect to elements in order to set coordinated material, information and finance flow, resulting in, respectively, aiming at satisfying customer requirements and respective economic effect. According to Merkuyev, Merjuyeva, Piera and Guasch Petit [19], simulation models have proved to be useful for examining the performance of different system configurations and/or alternative operating procedures for complex logistic and manufacturing systems. It is widely acknowledged that simulation is a powerful computer-based tool enabling decision-makers in business and industry to improve their organisational and operational efficiency. Combining simulation with experimental design or intelligent search has been successfully adopted for simulation optimization [20, 21]. Logistics as a scientific discipline was first defined back in the 50s. Subsequently, there was the development of the MRP I - Material Requirements Planning system. Later in the 60's, the MRP II - Manufacturing Resources Planning system was created, where the original MRP was supplemented by the algorithms for capacity calculations [22]. In the 70s, the method called Just-in-time was discovered for the first time. During the 80's, the advancements in computer technology helped to accelerate the information flows. Later, fully integrated logistics systems gradually started to emerge, which resulted in cost savings and gradual replacement of manual work by mechanisation [23, 24]. The area of logistics covers a variety of technical means, such as the elements of conventional transport, the elements of production facilities, robots [25-27] and, recently, also modern drones [28]. 20 Advances in Production Engineering & Management 15(1) 2020 Improvement of logistics in manufacturing system by the use of simulation modelling: A real industrial case study 3. Presentation of the problem The investigated company is focused on timber processing of production of building timber products, such as beams, planks and boards. They are currently very popular for the construction of low cost houses the core of which is formed from beams, planks and boards. That is why there is a high demand for the products of the company for the needs of the building industry. The basic element of the products of the company is timber. The company produces beams, planks and boards of the required dimensions and types by means of timber cutting, edging, planning, milling and trimming. The raw material is supplied from various sources and the main types of woods are oak, beech, spruce and pine. All the types of raw timber for the production companies are represented in the same volume. The supply the production company is carried out by the contracted transport companies supplying one truck of timber of each timber type once a week. The described production company has its headquarters in the south-east of Slovakia in the vicinity of Kosice city. The company provides a complex timber processing, i.e. from the inputs of raw timber through its storage, cutting and drying, to manufacturing of beams, planks and boards. The waste generated during the production is processed in a contracting company to produce pellets and wood chips used for heating. As there is an increasing demand for the products of the company, which is demonstrated by the increase in the building activities within the region, there are situations during which the company is not able to meet all the requirements of the market in the short term. This is also the reason why the company is interested in identifying the bottlenecks [29] in production and design of the logistics in manufacturing in order to produce and purchase new devices that could increase their production with a minimum investment [30-32]. The entire production process of raw timber processing was developed to its present form only based on the experience of the company operators and workers. Since wood drying is the longest stage in the processing of raw timber in terms of the technological progress, the company bought a drying kiln for the batch drying of wood material. The capacity of the kiln drying device is 40 m3. Drying in this device takes 50 days on average, according to the type and thickness of the wooden elements. Within the technological process, the raw timber material is gradually shifting through the workplaces: receiving raw timber material and its storage - cutting head saw - edging saw - wood drying -crosscut saw - planning, milling, trimming - products despatch (Fig. 1). Cuttings waste is packed into sacks at special workplace for the disposal of waste. Capacity utilization of the devices is performed by means of the preparation of raw timber, its drying and the crowdedness of the system in advance. The passage time through the timber manufacturing system depends on the capacity of the individual devices, their cadence and the time necessary for the execution of the manufacturing operation and overloading of the manufacturing system as a whole unit (Table 1). Statistically, most of the time in real operation is required for drying of the raw timber material. The company provides transport of the products to customers by means of outsourcing companies. Each workstation is equipped with its own buffer necessary for the regulation of different production capacities and to ensure the fluency of production. In each part of the technological process, it is necessary to have an enough semi-finished timber element. That is why a significant amount of semi-finished products and unfinished timber elements is bound in each part of the system and a significant amount of capital is bound in warehousing, resources and in semifinished products. Advances in Production Engineering & Management 15(1) 2020 21 Straka, Khouri, Lenort, Besta Table 1 Parameters of workstations delay, and devices delay in logistics in manufacturing Timber Timber Cutting Edging Kiln Air Cross Wood Wood Wood Packing entry head saw drying seasoning cut planning milling trimming cuttings resaw saw waste Oak Beech Spruce Pine - 3 pieces -/ day / type 60 min 60 min 40 min 40 min 3.5 min 3.5 min 3.0 min 3.0 min 75 days 75 ,y— 6 months 25 days 25 days 2 min 5 min 15 min 5 min 5-8 min 4. Materials and methods The next step is a prepared comprehensive simulation model of all the activities within the timber material processing, which required a thorough analysis and compliance with the steps formalized scheme: description of production units, block diagram, simulation model, simulation runs, improvement of efficiency of logistics in manufacturing. 4.1 The creation of a formalized scheme To make the computer simulation model, the first right step is to create a formalized scheme that represents the exact sequence of production operations, interconnecting facilities with the technical resources of the company. This step also involves retrieving the information about the parameters of the material flows in terms of volume distributions. Such a formalized scheme then represents the overall system with its features and links. In this case, the system consists of the individual elements represented by the operations with timber material. Between the operations, there are links formed by the elements of flow management of semi-finished products and of timber waste. A formalized structure (Fig. 2) is a very important basis for the creation of an actual simulation model. The limitations and capacity of the production facilities are important for correct understanding of whole system activity and the elements of logistics in manufacturing. During the subsequent simulation modelling, the individual parts of a formalized scheme will be replaced by a respective block of specific simulation software. Delay parameters (Table 1) are crucial for the setting of the simulation model blocks and for the control and solution of the critical production bottlenecks of the whole manufacturing system. MATERIAL ENTRY, OAR Hlh BUFFER MATERIAL 1 ENTRY, BEECH Hlh BUFFER ERIAL 2 SPRUCE CHIH BUFFER ERIAL 3 Y, PINE CHIH MATERIAL ENTRY, SPRI CUTTING HEAD RESAW h rFER HÍH ■FER 6 IH rFER ill- EDGING SAW 1h BUFFER HÍH I4 0UF1 « HI BUFF IH 80j)% A|R SEASONING not dried BEAMS not dried PLANKS not dried BOARDS dried BEAMS dried PLANKS dried BOARDS dried, planed BEAMS dried, planed ilaned BOARDS dried, milled BEAMS dried, milled PLANKS dried, milled BOARDS dried, trimmed BEAMS dried, trimmed PLANKS dried, trimmed BOARDS Fig. 2 The formalized scheme of building timber elements production 4.2 The logistics in manufacturing units The entire manufacturing system and logistics in manufacturing is composed of parts that are arranged in sequence according to the technological procedure. The system is mainly series oriented with the final processing of products according to the required processes. The sequence of production technology is determined by the arrangement of specialized workplaces, which are created by workplace for income and storage of raw timber material, workplace for cutting head resaw, workplace for edging saw, workplace for drying timber, workplace for cross cut saw, 22 Advances in Production Engineering & Management 15(1) 2020 Improvement of logistics in manufacturing system by the use of simulation modelling: A real industrial case study workplaces of products finalization as wood planning, wood milling, wood trimming and with workplace for packing of cuttings waste. The workplace of income and storage of raw timber material ensures the supply of the company with raw timber material, and its primary storage by the type of timber before processing. The company supply is carried out on a weekly basis with a frequency of one truck for each timber type, which represents about 20-24 units of one timber type per week, i.e. approximately 1 piece of timber material per 0.33 day. The workplace of cutting head resaw involves raw timber material cutting (depending on the thickness and length of timber) to beams, planks and boards. The output of the cutting is represented by unedged timber elements. The processing time of the elements depends on the type of timber. The cutting process of one timber piece of oak and beech requires about 60 minutes and the cutting process of one timber piece of spruce and pine requires about 40 minutes. The average output after cutting is 20 pieces of various wooden elements in the form of beams, planks and boards. The cutting generates 10 % of cuttings waste and 90 % of wooden elements continuing to the next workplace. The workplace of edging saw is used for adapting and edging of unedged wooden elements. The time for edging of wooden elements depends on the type of timber, but an average time for oaks and beeches edging is about 3.5 minutes, and for spruce and pine edging, the time is about 3 minutes for one piece of wood element. The edging of wooden elements produces 25 % of cuttings waste. After edging of the wooden elements, 10 % is sold in the raw state and 90 % of the elements will be moved to the workplace of wood drying. 10 % goes to the kiln drying, and 80 % is dried in the open stock in natural conditions. The kiln drying device has a capacity of 40 m3 for a drying batch. The cut raw timber is dried according to the requirements for the percentage of moisture and for the type of material. In general, the raw timber materials of the oak and beech types are drying for about 75 days and the spruce and pine types are drying form about 25 days. Drying of the raw timber material at the open stock in natural conditions takes about six months for all the types of timber. After drying, the timber elements are cut to the required lengths at the workplace of crosscut saw. The time for timber elements cutting to the required lengths is about 2 minutes per one piece of timber element. The cutting process generates cuttings waste, which represents 10 % of the timber material capacity. After cutting, 65 % of the timber elements are ready for sale and export, and 35 % of the timber elements are treated according to special requirements of customers at the workplaces of wood planning, wood milling and wood trimming. The workplaces of wood planning, wood milling and wood trimming provide a service for customers and their requirements for the final machining. The processing time of the wood planning and wood trimming is about 5 minutes for a one piece of any type of timber element The processing time of wood milling is about 15 minutes for a one piece of any type of timber element. All the final activities produce 5 % of cuttings waste. All the timber waste is situated in the stock of the packing of cuttings waste workplace. The packing of one waste unit requires about 5-8 minutes. Waste taking is carried out by a contracting firm and its own means of transport. Cuttings waste is used for the secondary recovery and for heating. 4.3 The model of logistics in manufacturing represented by a block diagram To create a block diagram (Fig. 3) as a basis for the actual simulation model, it is important to prepare the data and the information that are essential for the setting of the individual blocks within the simulation model [33]. From the statistical data obtained from observation and measurement in the field as well as from the technical documentation data [34], it appears that the input of the timber material into system occurs each 0.33 day, when another piece of timber material enters the system. In the simulation model, this is modelled by the four entry blocks "create" for each type of timber with a constant distribution setting of 0.33 day. Just before the first production facility, which is the cutting head resaw, there is an unprocessed timber storage which balances the cadence delays within production and the capacity of the control unit. In the Advances in Production Engineering & Management 15(1) 2020 23 Straka, Khouri, Lenort, Besta simulation model, this is represented by four blocks "queue" with the setting of "last in first out (LIFO)", material is put on itself. After that, there is a cutting head resaw type of device, where the processing of a single piece of timber ranges from 40 to 60 minutes. In the simulation model, this delay is modelled by a "lookup table" with constant distribution with delay parameters set to 40 minutes or 60 minutes for the processing of one piece and type of timber. 4.4 The logistics in manufacturing of the concrete company designed by the EXTENDSIM Each real activity, and all the processes and operations of the manufacturing system are necessary to transform and design by the blocks of concrete simulation system EXTENDSIM. Parts of the formal and block diagrams will be translated step by step, block by block to EXTENDSIM simulation model. The entry of timber into the production company, the identification of timber types and the storage of logs are represented by the blocks of "create", "set" and "queue" (Fig. 4). el blocks and for the control and solution of the critical production bottlenecks of the whole manufacturing system. The place of work of "Cutting head resaw" is modelled in the simulation model using the first "hierarchy block". The hierarchy block for this operation is created using the blocks of "activity" - "batch" - "select item out" (Fig. 5). gv ^ ^ Rn Ln oak type t buffer ^ Fig. 4 Blocks "create", "set" and "queue" which represent the entry of timber material into the company, the identification of timber type and the storage of logs 24 Advances in Production Engineering & Management 15(1) 2020 Improvement of logistics in manufacturing system by the use of simulation modelling: A real industrial case study - 1 [3383 CUTTING HEAD SAW <„. a d waste separate cutting products i |Con30ut CUTTING HEAD RESjSWV Help j cutting head f| Fig. 5 Blocks of "hierarchy block", "activity", "batch" and "select item out" which represent the cutting of the timber into semi products such as beams, planks and boards The place of work of "Edging saw" is modelled in the simulation model using the second "hierarchy block". The hierarchy block for this operation is created using the blocks of "select item out" - "queue" - "select item in" - "activity" - "select item out" (Fig. 6). iLHJiaJ ® - Du Sf sdgingsau «aaesïpaij & le; =¡1 _bu««ri EDGING SAW _h_IfeDOINOSAW~~I Iconaout \ 3 r Fig. 6 Blocks of "hierarchy block", "select item out", "queue", "select item in", "activity" and "select item out" which represent the operation edging of the timber The place of work of "Wood drying" is modelled in the simulation model using the third "hierarchy block". The hierarchy block for this operation is created using the blocks of "select item out" - "queue" - "select item in" - "select item out" - "batch" - "activity" - "select item in" (Fig. 7). mm¡ =a 1 E= WOOD DRÏING Fig. 7 Blocks of "hierarchy block", "select item out", "queue", "select item in", "select item out", "batch", "activity" and "select item in" which represent the operation drying of the timber The place of work of "Cross cut saw" is modelled in the simulation model using the fourth "hierarchy block". The hierarchy block for this operation is created using blocks "select item out" - "queue" - "select item in" - "activity" - "select item out" (Fig. 8). By combining the above described simulation model elements, several selected production processes were modelled, and they represent the actual real-life manufacturing system within the investigated company (Fig. 9). Advances in Production Engineering & Management 15(1) 2020 25 Straka, Khouri, Lenort, Besta ¿|[3?B] CROSS CUT SÍW | a || El || S3 | ^J [328] CROSS CUT SAW ill [B3][2 5Lie Comments Comments ' I84][3] Qutut I Comments | [ Queue ] Options Results j Con | Queue | Options | Results j con [ Queue | Options | Results j Cor [ Queue | Options | Results | Con Items wart here for downstream eapai Items wart here for downstream capai Items wait here for downstream capa Items war! here for downstream capai Queue statistics Current Average Length: 4054 2356,6479 Wait: 0,1823158 0,1683521 Arrivals: 14810 Departures: 10756 Queue statistics Length Wait: Arrivals: Current Average 5444 (570.2095 0.0005816 0,0621024 14695 Departures: 9251 Queue statistics Length: Wait: Arrivals: Current Average 8926 3265,7764 0,2434905 0,0928611 14702 Departures: 5776 Queue statistics Length: Watt: Arrivals: Average 17t 2,7651: Current 4032 0,0454676 0.1081364 14723 Departures: 10691 Fig. 10 Length of timber material in blocks of "queue" before processing at kiln drying facility The researched manufacturing system generates approximately 15,000 pieces of packed cuttings waste units per year. The production of cuttings waste depends on the numbers of processed timber materials and the utilization of production devices. The ratio between the quantity of the incoming material to the manufacturing system and the cuttings waste is approximately 5:1. The ratio between the quantity of the finished products and the cuttings waste is approximately 1.2:1. Other building timber materials are scattered throughout the manufacturing system. This fact makes it possible to state that the store and buffers of the whole manufacturing system contain many unfinished products. The number of elements of unfinished products can be reduced by means of thorough planning of production and by increasing the utilization of the production devices. The simulation output data show that part of the manufacturing system is overloaded, while other parts of the system have substantial reserves. We have the following possibilities in order to streamline the production: The streamlining of the manufacturing system activity does not depend on the purchase of new and expensive devices, but only on the change of the logistics in manufacturing and the management of flows. All the workstations of the company have reserves for further development. By changing the management of the company logistics in manufacturing, we will increase the performance of timber drying in the drying device, thereby also increasing the performance of the entire manufacturing system. The company has enough storage capacity and it is therefore possible to increase the intensity of company supply. The simulation results indicate that it is necessary to double the intensity of company supply, i.e. a frequency of entry set to 0.15 days for each timber type. This adjustment increases the performance of unutilized devices and the whole manufacturing system several times, up to 54,475 produced building timber elements, which represents an increase of production by about 199.6 % while maintaining company flexibility (Fig. 11). This adjustment increases the resources at the input to the system, but the performance of the devices will increase several times and thus the whole manufacturing system. This implies that the intensity of the input material to the system also represents a bottleneck for the produc- 1691 (at Report j Animation j Comments^ Passes items out of the simulation ■ Reports results - Number E"ted Ig rtore R«seis 1 5964 a 2 3516 a 3 2423 a 4 1417e a 5 9340 a 6 4515 0 7 3914 0 6 2435 0 0 1550 0 ID 107? H 11 726 0 12 444 0 13 2219 0 14 1327 0 IS 847 0 xfc-l,u>Mfc-l,l>Mfc-l,2>" "> Mfc-lj','", Mfc-l,m] , j = 1,2,",m (11) Where, xk_lv and xk_lm are robot and characteristic position information, respectively; xk-i,u is control input amount to robot. © Cubature points are propagated through equation of state, and the prior estimates (weights) are calculated as: xfc|fc-l =f(xk-l,v,xk,u) (12) © According to cubature seeking rule, square-root covariance of robot attitude estimation can be calculated: ^fc|fc-l = ^r(^fc!k-l^Q,k-l), Qk-1 (13) @ Apply cubature transformation to approximate the third-order state of robot. X^-lv = 2 (ns+nm)^,i = 1 Äfc|fc-l,v (14) Where, qr means the matrix is decomposed into lower triangle and upper triangle forms; and covariance prediction value is: Pfc|fc-1 = 2(nr+n^/ ^ 1 *fc|fc-l(*fc|fc-l) _xfc|fc-lxfc|fc-l + Qfc (15) © Estimation (weight) and state error innovation are represented by Skik_t. c* - 1 ■ r^*,1 -v1 r*,2 -v2 " ~*,2(ns+nu) _2(ns+nu)] (16) Dfc|fc-1 ~ J2(ns+nm) [ Afc|fc-1, xfc|fc-l /tfc|fc-l,,xfc|fc-l Afc|fc-1 J (±D) (2) Measurement update phase The z'-th feature point observation generalized from the nonlinear observation model is expressed as the posterior estimation formula: zfc|fc-l = ^(xfc|fc-l,v,xfc|fc-l,m)+^z = ^(xfc|fc-l,v,Mfc|fc-l) + (17) Where, xk|k_1|V is attitude information of robot in the prior estimation, which does not change with observation of a certain feature point; is estimated value of z'-th feature point at time k — 1; Observation noise is Sz~N(0, R). ® First calculate the z'-th cubature point. xfc-l = +xfc|fc-l,i = 1,2,",2(.ns +nu) (18) xfc-l = [xfc|fc-l,v,ufc|fc-l,l,ufc|fc-l,2,"",ufc|fc-l,rn] (19) is cubature point set of z'-th cubature point xk|k_1|V of robot's own state and z'-th cubature point ufc|fc-ij of y-th characteristic state quantity, j = 1,2, — , m. © Propagation cubature point 34 Advances in Production Engineering & Management 15(1) 2020 Estimating the position and orientation of a mobile robot using neural network framework based on combined square- ... -^(^fclfc-i'^fclfc-ij) (20) ® According to the rule for finding the cubature, calculate the observation and prediction values: *W-I=^s^j=izlk\k-1 (21) ® Calculate observation and forecast estimate (weight) covariance error innovation square-root formula: Szz,k\k-1 = qr {[slkik-1VR]7} (22) Where, Rk = SR>kSR)k. Innovation error is: = * = 1,2, '" ,2n5 (23) ® Square-root of estimated (weight) cross-covariance is: PxzMk-l =^k|k-l{£lJc|k-l) (25) T Kalman gain is: Wk (27) According to Kalman gain expression in CKF algorithm, square-root of correlation error covar-iance can be obtained: Sw = qr[^k|k_1 -W^^W^^] (28) Under the given initial conditions, after the state prediction update iteration of SRCKF-SLAM algorithm, system's state estimation is obtained, and robot's navigation in the environment and positioning and observation of target are achieved. However, with the creation of SLAM map database, the cubature point set and its weights have been continuously updated and expanded, which has led to a reduction in the algorithm's filtering estimation accuracy. A neural network framework based on SRCKF-SLAM algorithm is proposed to reduce square-root cubature points. The weight matrix of set and the output value of network are regarded as state quantities and measured values, respectively, and state space is used to describe training. 2.2 Neural network structure model Let the number of nodes in input layer of neural network be n, the number of nodes in hidden layer be m, and output layer be a non-time-varying neuron node. Then network model is shown in the Fig. 1 [12]: Fig. 1 The model of neural network Advances in Production Engineering & Management 15(1) 2020 35 Wang, Tan, Dong, Yuan, Du Let dimension of input signal be n, so the number of samples is the same as the dimension of input signal. In Fig. 1, x1,x2,---,xn is input signal to be trained, and its corresponding weight is: w1,w2,---,wn. The function at each input node of hidden layer is zt = X(W(. Input signal of output layer is: z = YS=1xiwi = xTw (29) The general formula for nonlinear relationship between output of neural network and input is: y = f I vM1'^ a-]] (30) Where, x¿ and y are network input and output. is hidden layer excitation function. Ojis the scaling function factor. bj is translation function factor. /(•) is activation function of output layer. Wij(t) is the weight between j-th neuron in hidden layer and the z-th neuron in input layer, and Vj is the weight between output layer and the j-th neuron in hidden layer. It is generally believed that increasing the number of hidden layers can reduce network errors, but also complicate network. Generally, the design of neural networks should give priority to 3-layer networks. When accuracy allows, reduce the number of nodes in hidden layer as much as possible [13-15]. State space of network consists of a weight matrix, network input and output, weight update functions, and parameterized non-linear functions. Assuming that there are / and J neurons in each of the two adjacent network layers, the weight matrix between the two layers is W¡j and the weight space of neural network is W = {YWij}. Let W¡j be the weight matrix between layer / and layer J consisting of w¿J (i = 1. ../;_/ = 1..._/). Unitized weight matrix is W¡j: WIJ=F1(wij) = w„ (31) x,x w?--1 lJ The optimal estimate of the cubature point weights (state quantities) of SRCKF is obtained through neural network framework. State space model of neural network with weight as state quantity can be expressed as: ( wk+1 =Fkwk + qk {zk = ^Jk(wk,xk) + rk ( ) Where, two kinds of Gaussian noise obey qk~N(0,Q) and rk~N(0,R), respectively. Output of neural network at time k + 1 is the iterative training result of the observation equation of network, which is determined by network weights and network observation noise at time k; Fk and ipk represent the linear state mapping and observation nonlinear mapping of network. Article selection is Fk = lk. 3.Results and discussion 3.1 Neural network framework based on proposed SRCKF-SLAM algorithm Establish a neural network framework and tracking algorithm based on SRCKF-SLAM. The process includes: (1) Update iterative formula according to the state of SRCKF-SLAM, generate and propagate cubature points, solve the prior estimates of the attitude of robot and calculate robot's attitude information and square-root covariance. (2) Set the initial weight for the innovation error (weight) between the observation and estimation of the cubature point state with the new square-root sub-form in SRCKF in step (1) as the input of neural network. Hermit function is selected as excitation function, a network state space 36 Advances in Production Engineering & Management 15(1) 2020 Estimating the position and orientation of a mobile robot using neural network framework based on combined square- model is established to train the SRCKF state quantity, and the filtered new optimal estimate and gain are obtained as output of neural network, which together serve as the next state parameter update. The result of the expected output vector is that the position error of the feature point is required to be minimal, so network output layer node is set to 3 (Observe coordinates in the x-direction and y-direction, and observation angle of robot heading angle so network input layer vector is constrained, and the number of nodes is set to 5 (Coordinates in the x-direction, coordinates in the y-direction, the size of robot's heading angle and controls input noise and observation noise), hidden layer is first set to 10 nodes and is automatically adjusted by Matlab software after network training programming. (3) Bring the trained estimates (weights) into the SRCKF algorithm to iterate, propagate cuba-ture points and calculate state update parameters. Square-root of the estimated cross-covariance in step (1) is improved to a neural network framework state model as: Kalman gain is: wí(fc|fc-i(£fc|fc-i) ¿=1 Wk = (Pxz,fc|fc-l/>S'Jz,fc|fc-l)>S'zz,fc|fc-l (33) (34) According to Kalman gain expression in CKF algorithm, square-root of correlation error covar-iance can be obtained: Sfc|fc = qr[(kik-1 -W£fc|fc-i,WSfc|fc] 3.2 Verification and analysis of feature map simulation experiments based on SLAM Let initial state of robot be zero. Initial values of robot and feature covariance are P, diag[330]T and Pmo —diag[0.20.20]T , so sensor sampling frequency is T — 0.02s. System noise (35) vO — 0.2' 0 0.1^0 of-)' V180/ Robot motion parameters are as and observation noise are Q = r m\2 r zn\2 , R = follows: speed v = 2 m/s, speed error av = 0.2 m/s, angular velocity a = 15 °/s, ranging error 0.25 m, angle error oa = 1°/s. In the experimental environment (Fig. 2), SRCKF-SLAM, SRUKF-SLAM, SCDKF-SLAM were subjected to 50 independent repeated simulation experiments, so the estimation errors in the x-direction and y-direction of robot path were analyzed and compared. Fig. 2 Simulation area Advances in Production Engineering & Management 15(1) 2020 37 Wang, Tan, Dong, Yuan, Du 4000 5000 6000 7000 8000 9000 t/S Fig. 3 The estimation error in the x-direction of robot path under the three SLAM algorithms 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 t/s Fig. 4 The estimation error in the y-direction of robot path under the three SLAM algorithms It can be seen from Figs. 3 and 4 that the position estimation error in the x-direction of robot path obtained using SRCKF-SLAM algorithm is [0, 0.8] m, and position estimation error in the y-direction is [0, 1.2] m. Compared with SRUCF-SLAM and SRCDKF-SLAM filter estimation errors, the filter estimation error is the smallest, and the filter estimation accuracy is superior to them. The experimental data fully proves that using SRCKF-SLAM algorithm for mobile robot navigation and positioning estimation can achieve high accuracy and good stability. Then compare the superiority of the target feature position of cubature point estimation (weight) under neural network framework based on the new and traditional SRCKF-SLAM algorithm. 3.3 Experimental verification and analysis of neural network frameworks based on SRCKF-SLAM • Simulation experiment and analysis of feature position estimation Hermit function is selected as excitation function, and the expression is: f(x) = "('-"^'H-4) (36) Initialization algorithm can effectively reduce overshoot and shorten the convergence time of algorithm. Network itself has a certain good weight, so this difference can effectively reduce the possibility of divergence during network learning process. In this paper, initial network weight w is a random number between [—1,1], and the noise follows normal distribution[0,0.1], the learning rate is 0.4, the momentum term coefficient is 0.6, and the number of learning times is 800. According to the function expression, P0 is 0.55, a = 0.2, b = 0.01. 38 Advances in Production Engineering & Management 15(1) 2020 Estimating the position and orientation of a mobile robot using neural network framework based on combined square- ... In order to better compare the filtering estimation accuracy of SRCKF-SLAM filtering estimation algorithm of the cubature point weights and the traditional SRCKF-SLAM algorithm after neural network, increase the system noise and observation noise, set Q = 0.22 0 (2n\2 (3n\2 U80/ U80/ , R = 0.12 (-2-Y . 180 0 (-) 180 Other initial conditions remain unchanged, and 50 independent repeated simulation experiments are performed, and the estimated errors in the x-direction and y-direction of the target feature position are analyzed and compared. From Figs. 5 and 6, it can be seen that under the conditions of increasing system noise and observation noise, as the simulation progresses, due to the setting of the reference path, the walking direction of robot is constantly changing, so the position estimation error has fluctuations during the simulation. The traditional SRCKF-SLAM algorithm's position estimation errors in the x-direction and y-direction are respectively [0, 1.8] m and [0, 2.6] m. The filtering algorithm based on SRCKF-SLAM's cubature point estimation (weight) after neural network framework, the position estimation errors in the feature x-direction and y-direction are respectively [0, 1.2] m, [0, 1.7] m. The filtering estimation accuracy is higher than the former, which is consistent with the theoretical analysis. SRCKF-SLAM Neural network training algorithm based SRCKF-SLAM 4000 5000 6000 7000 8000 9000 t/s Fig. 5 The estimation error in the x-direction of the characteristics under the two SLAM algorithms SRCKF-SLAM Neural network training algorithm based SRCKF-SLAM 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 t/s Fig. 6 The estimation error in the y-direction of the characteristics under the two SLAM algorithms Advances in Production Engineering & Management 15(1) 2020 39 Wang, Tan, Dong, Yuan, Du • Observation of features under different SLAM algorithms Based on SLAM filtering algorithm, the position error observations in the x-direction and y-direction of features are collected. Observe the characteristics in Fig. 2 of the simulation area. The three filtering algorithms using the SRUCF-SLAM, SRCKF-SLAM and SRCKF-SLAM cubature point weight training algorithms all observe 82 feature points, indicating that the new algorithm is feasible for SLAM feature target observation. It can be seen from Figs. 7 and 8 that the above-mentioned three algorithms reduce the position error in the x-direction and the y-direction of the features in turn, indicating that the SRCKF-SLAM algorithm after the cubature point weights are trained by network has the highest accuracy. And the algorithm error is convergent and stable, which is consistent with the theoretical analysis. + Observation characteristics of SRUKF-SLAM algorithm | + + + + ■ o o CD 0 + ++t ++ + v + _+ ++t_+ v+ + + —++I+ —I—I— c o O 0 O O o ° OO ?O oo00o0o0c o o o o o ) ^o O o Observation characteristics of SRCKF-SLAM algorithm o o o oo o° ° OO o° o ?oo o_QÎO_pO O ? 0°°00 o 90000 000060° ** . **** *** 10 20 30 40 50 60 70 80 landmarks Fig. 7 The position error in the x-direction of landmarks * Observation characteristics of Neural network training algorithm based on SRCKF-SLAM algorithm 0 10 20 30 40 50 60 70 80 90 landmarks Fig. 8 The position error in the y-direction of landmarks 3.4 Possible practical implementation of the proposed SRCKF-SLAM algorithm At present, single non-linear filtering algorithms used to deal with SLAM problems mainly includes EKF, UKF, CDKF, CKF, etc. Later, scholars further improved above several algorithms and proposed their derivative algorithms such as SRCDKF, SRUKF and SRCKF, then their accuracy are mainly discussed below. In reference [16], accuracy of EKF, UKF and CDKF are compared, and it is concluded that when EKF algorithm was used to deal with nonlinear model, Jacobian matrix was needed to be solved, 40 Advances in Production Engineering & Management 15(1) 2020 Estimating the position and orientation of a mobile robot using neural network framework based on combined square- ... and Taylor expansion and high-order terms were needed to be eliminated, and larger truncation error was introduced, which led to larger algorithm error. UKF is a method that uses weighted statistical linear regression to achieve random linearization, which can approximate the Gaussian distribution. In theory, UT transformation could approach the posterior mean and covariance of the nonlinear Gaussian system with three times of Taylor precision. However, when UKF algorithm generated a set of random Sigma points with a large number, it would lead to more complicated calculation of state quantity parameters. The Central Difference Kalman filter (CDKF) belongs to a suboptimal Gaussian filter. This algorithm uses polynomial interpolation to calculate multidimensional integrals. It uses a function sequence to approximate the integrand, which is similar to UT transform in UKF. It is just that the sampling point weights and calculation of prediction covariance expressions are different from the UKF algorithm in the form. The calculation of this algorithm is simple and easy to implement, but it is different from the selection of sigma points of UKF. In theory, the accuracy of CDKF and UKF algorithms is equivalent or slightly higher. According to references [17-18], it can be known that when using the CDKF algorithm to process non-linear models, the solution of the complex Jacobian matrix is avoided, the shortcomings of EKF algorithm are overcome, the linearization error is smaller than of EKF. The experiments have shown that it is less sensitive to state covariance and has a faster approximation speed than UKF. Later scholars collectively referred to the UKF / CDKF as the sigma point Kalman filter (SPKF). Then we discuss the accuracy problems of UKF, CDKF and CKF algorithms: the basic idea of the three filtering algorithms is to generate several groups of weight points through different methods, then calculate the parameters such as the mean value and covariance of propagation, and update state and measurement of algorithm. A large number of references, such as [19-20], have proved that the accuracy of filter estimation of the three algorithms is equivalent, and can reach the second order. For similar algorithms, another performance indicator that needs to be considered is computational complexity. According to the literature [20], it is verified that the computational complexity of the three algorithms is reduced in order, so that their filtering performance is sequentially enhanced. However, for the SPKF filter and ordinary CKF filter, there is still a problem of computational divergence. Later, students introduced SR ideas into these methods. According to reference [9], Square-root can not only propagate square-root of state covariance, but also ensure the symmetry and semi positive setting of covariance matrix, and improve the accuracy, robustness and stability of filtering algorithm. So combining with SR idea, this paper applied three groups of filtering algorithms SRUKF, SRCDKF and SRCKF to SLAM nonlinear model, compared the filtering performance, and got the conclusion in this section. That is, SRCKF-SLAM algorithm had the highest filtering accuracy and the best stability. However, in order to get higher accuracy, this paper established a more suitable state model for engineering application after training the state quantity of SRCKF algorithm with neural network, and compared it with SRUKF and SRCKF algorithms without neural network framework, and obtained another group of experimental conclusions in this section. It can be seen from the introduction and the theory of filtering in this paper, square-root cuba-ture Kalman filter algorithm uses numerical integration to calculate the mean and covariance of the nonlinear random model, which avoids the derivative operation and reduces the computational complexity. Moreover, the algorithm propagates square-root of the state covariance, ensures the symmetry and semi positive finalization of the covariance matrix, and improves the accuracy, robustness and stability of the filtering. If the new algorithm is applied to the mobile robot, for example, in the practical application system of the aviation strategic missile in military field, when the tracked target encounters air resistance in the course of navigation, its state equation and observation equation become highly nonlinear. While in the estimation of the tracking state of the reentry ballistic target with unknown ballistic coefficient, the new algorithm can greatly reduce the tracked target state estimation error and improve the estimation accuracy. Therefore, the time to solve the its position and speed information is short, and the navigation operation speed is fast. Advances in Production Engineering & Management 15(1) 2020 41 Wang, Tan, Dong, Yuan, Du 4. Conclusion Aiming at the problems of weak observability of system under non-Gaussian, non-linear, and large noise, leading to unstable filtering algorithms and slow convergence, a SRCKF-SLAM filtering algorithm was used to estimate the position of robot and target. Since the learning and training of neural network could be regarded as seeking the best weight parameter through optimal estimation, and weight matrix and network output value could be regarded as state quantities and measured values, respectively, the training was described by state space. Moreover, neural network could have good convergence and robustness without an accurate mathematical model. An SRCKF-SLAM filtering algorithm based on neural network framework with cubature point estimates (weights) was proposed. Simulation experiments showed that the new filter algorithm was feasible and effective for feature observation after training with neural network. Under the conditions of increasing system noise and observation noise, position estimation errors of feature in the x-direction and in the y-direction were respectively [0, 1.2] m and [0, 1.7] m, which improved estimation accuracy compared to other traditional filtering algorithms such as SRCCK-SLAM, SRUCF-SLAM and SRCDKF-SLAM. If the new algorithm is applied to mobile robot, for example, in the practical application system of aviation strategic missile model in military field, it can greatly reduce tracked target state estimation error and improve estimation accuracy. The time to solve its position information is short, so the navigation operation speed is fast. Acknowledgement Funds support: The Key Science and Technology Program of Henan province, China(182102110295), and the project of Henan Science and Technology Think Tank (HNKJZK-2019-30B), and the Key Science and Technology Program of Anyang(121), and the Education and teaching reform of Anyang Institute Of Technology(AGJ2019053). References [1] Smith, R., Self, M., Cheeseman, P. (1988). Estimating uncertain spatial relationships in robotics, In: Lemmer, J.F., Kanal, L.N. (eds.), Machine Intelligence and Pattern Recognition, Amsterdam North-Holland, Vol. 5, 435-461, doi: 10.1016/B978-0-444-70396-5.50042-X. [2] Yuan, G.-N., Wang, D.-D., Wei, Y.-H., Hong, W. (2013). Particle filter SLAM algorithm for underwater oil pipeline leakage detection and positioning, Journal of Chinese Inertial Technology, Vol. 21, No. 2, 204-208. [3] Sun, Y., Lu, D., Chen, Q. (2013). An improved cubature Kalman filters based on strong tracking, Journal of Huazhong University of Science and Technology (Natural Science Edition), Vol. 41, 451-454. [4] Wang, H., Fu, G., Li, J., Yan, Z., Bian, X. (2013). An adaptive UKF based SLAM method for unmanned underwater vehicle, Mathematical Problems in Engineering, Vol. 2013, Article ID 605981, doi: 10.1155/2013/605981. [5] Yuan, G., Wang, D., Tan, K. (2013). SLAM algorithm based on the grid map fuzzy logic, Journal of Huazhong University of Science and Technology (Natural Science Edition), Vol. 41, No. 9, 32-36. [6] Yuan, G., Hu, Z., Zhang, J., Zhao, X., Fu, C. (2016) Novel neural network framework based on iterated cubature Kalman filter, Computer Science, Vol. 43, No. 10, 256-261. [7] Feng, K., Li, J., Zhang, X., Zhang, X., Shen, C., Cao, H., Yang, Y., Liu, J. (2018). An improved strong tracking cubature Kalman filter for GPS/INS integrated navigation systems, Sensors, Vol. 18, No. 6, 1919-1940, doi: 10.3390/s18061919. [8] Yue, C., Bo, Y.-M., Wu P.-L., Tian, M.-C., Chen, Z.-M. (2017). Non-cooperative space target tracking based on fuzzy iterative square-root cubature Kalman filter, Journal of Chinese Inertial Technology, Vol. 25, No. 3, 395-398. [9] Gao, W., Zhang, Y., Sun, Q., Guan, J. (2014). Simultaneous localization and mapping based in iterated square root cubature Kalman filter, Journal of Harbin Institute of Technology, Vol. 46, No. 12, 120-124. [10] Adam, S.P., Karras, D.A., Magoulas, G.D., Vrahatis, M.N. (2014). Solving the linear interval tolerance problem for weight initialization of neural networks, Neural Networks, Vol. 54, 17-37, doi: 10.1016/j.neunet.2014.02.006. [11] Kim, K.J., Park, J.B., Choi, Y.H. (2006). The adaptive learning rates of extended Kalman filter based training algorithm for wavelet neural networks, In: Gelbukh A., Reyes-Garcia C.A. (eds.), MICAI2006: Advances in Artificial Intelligence, MICAI2006, Lecture Notes in Computer Science, Vol. 4293, 327-337, Springer, Berlin, Germany, doi: 10.1007/11925231 31. [12] Li, W., Fang, J. (2014). BP neural network for mobile robot self-tuning PID controller PID controller design, Advanced Materials Research, Vol. 898, 755-758, doi: 10.4028/www.scientific.net/AMR.898.755. [13] Ahmad, M.A., Azuma, S.-I., Sugie, T. (2014). Performance analysis of model-free PID tuning of MIMO systems based on simultaneous perturbation stochastic approximation, Expert Systems with Applications, Vol. 41, No. 14, 6361-6370, doi: 10.1016/j.eswa.2014.03.055. 42 Advances in Production Engineering & Management 15(1) 2020 Estimating the position and orientation of a mobile robot using neural network framework based on combined square- ... [14] Du, X., Zhao, Y., Yuan, G., Xia, G., Chang, S. (2014). A novel initial method of fuzzy wavelet network in once-through steam generator control, In: Proceedings of the 33rd Chinese Control Conference, Nanjing, China, 86938698, doi: 10.1109/ChiCC.2014.6896461. [15] Sadati, N., Ghaffarkhah, A., Ostadabbas, S. (2008). A new neural network based FOPID controller, In: Proceedings of2008 IEEE International Conference on Networking, Sensing and Control, Sanya, China, 762-767, doi: 10.1109/ ICNSC.2008.4525318. [16] Chatterjee, S., Sadhu, S., Ghoshal, T.K. (2017). Improved estimation and fault detection scheme for a class of nonlinear hybrid systems using time delayed adaptive CD state estimator, IETSignal Processing, Vol. 11, No. 7, 771779, doi: 10.1049/iet-spr.2016.0380. [17] Al-Shabi, M. (2017). Sigma-point smooth variable structure filters applications into robotic arm, In: 7th International Conference on Modeling, Simulation, and Applied Optimization (ICMSAO), Sharjah, United Arab Emirates, doi: 10.1109/ICMSA0.2017.7934865. [18] Li, J., Zhang, J.-X., Zhang, Y.-H., Chen, L.-J. (2017). Estimation of vehicle state and parameter based on strong tracking CDKF, Journal of Jilin University Engineering and Technology Edition, Vol, 47, No. 5, 1329-1335, doi: 10.13229/j.cnki.jdxbgxb201705001. [19] Wang, G., Li, N., Zhang, Y. (2016). Diffusion distributed Kalman filter over sensor networks without exchanging raw measurements, Signal Processing, Vol. 132, 1-7, doi: 10.1016/j.sigpro.2016.07.033. [20] Chen, M. (2017). The sequential WSNs target tracking algorithm based on adaptive SR-CKF, Chinese Journal of Sensors and Actuators, Vol. 30, No. 8, 1220-1225, doi: 10.3969/j.issn.1004-1699.2017.08.016. Advances in Production Engineering & Management 15(1) 2020 43 Advances in Production Engineering & Management Volume 15 | Number 1 | March 2020 | pp 44-56 https://doi.Org/10.14743/apem2020.1.348 ISSN 1854-625G Journal home: apem-journal.org Original scientific paper A comparison of the tolerance analysis methods in the open-loop assembly Kosec, P.a, Skec, S.a,b*, Miler, D.a aUniversity of Zagreb, Faculty of Mechanical Engineering and Naval Architecture, Zagreb, Croatia bTechnical University of Denmark, Kongens Lyngby, Denmark A B S T R A C T A R T I C L E I N F O Dimensional and geometric tolerances affect both the cost and the functionality of a given product. Finding the acceptable trade-off between the two is among the common engineering tasks. Thus, many tolerance analysis methods are developed to help engineers and assist in the decision-making process. In this article, the authors have assessed four tolerance analysis methods by applying them to the open-loop assembly. The results obtained by the tolerance chart (worst-case) method, Monte-Carlo simulation, vector-loop analysis, and the Unified Jacobian-torsor model were analysed and compared. Additionally, the overview and application guidelines are included for each of the methods, aiming to help both researchers and practitioners. The results have confirmed that there are significant variations in the outputs across the observed methods, implying the need for informed method selection. © 2020 CPE, University of Maribor. All rights reserved. Keywords: Assembly; Open-loop assembly; Tolerance analysis; Computer aided tolerancing; Tolerance chart analysis; Unified Jacobian-torsor model; Monte Carlo method; Vector-loop analysis *Corresponding author: stanko.skec@fsb.hr (Škec, S.) Article history: Received 6 September 2019 Revised 11 March 2020 Accepted 17 March 2020 1. Introduction During the design phase, tolerances are assigned to nominal dimensions, ensuring successful assembly while retaining the manufacturing costs at an acceptable level. As the complexity of mechanical design increases, keeping track of the tolerances becomes harder. To mitigate the problem, tolerance analysis methods of various complexity are available. The methods range from simple, 1D tolerance chart analysis, to advanced procedures requiring the use of advanced mathematical models. Examples include vector loop, Unified Jacobian-torsor, T-maps, and Skin model Shapes. Furthermore, tolerance analysis methods can be divided by several criteria: an approach to the analysis, identification process, and the calculation procedure of dependent dimensions [1]. Before analysing, tolerances are assigned to assembly features and are organised into stacks, easing the variation analysis. Stacks are then used to analyse the assembly by reading the drawings or by assigning tolerances on computer-aided drawing (CAD). The tolerances are then stacked into loops using points, surfaces, vectors, or joints, among others - depending on the method [1]. Manual charting is frequently used when solving simple problems consisting of few dimensions. As the number of dimensions increases, its reliability decreases - it is error-prone and tiresome. Additionally, manual analysis is hard to perform in 2D and 3D tolerancing problems. 44 A comparison of the tolerance analysis methods in the open-loop assembly The tolerancing problem complexity further increases when the geometric tolerances are necessary [2]. Geometric tolerances are defined by 3D tolerance zones, rendering most of the simpler methods unusable. Thus, computer-aided tools (CAT) were developed, increasing the capabilities in terms of the number of available approaches and mathematical models. Many such tools are developed and successfully applied (VisVSA, 3DCS, CETOL, OpTol) [3] in the industrial environment. Unfortunately, various proprietary CAT tools use different mathematical models to define and analyse tolerances, meaning that the obtained results may differ [4]. State-of-the-art CAT tools allow users to model assembly stacks with point-to-point features. The contributing tolerances are identified and arranged into suitable stacks or loops [5] as each method is compatible with a specific stacking procedure to build the stacking equation. In recent papers, many researchers have studied differences and similarities of tolerance analysis methods. Studies considered the contributing tolerances from multiple directions [6], the angular deviation of the adjustable element, or a critical assembly feature (functional requirement) [6]. Also, the form [1] and interaction of the multiple tolerances in the 3D context is defined by the geometric drawing and tolerancing (GD&T) standards [1, 10]. Due to frequent changes in GD&T standards [7] such as ISO 8015 [2] and ASME Y14.5 [8, 9], continuous support of the tolerance analysis methods is needed. Various assembly applications are described as a system of open-loop or closed-loop that must be solved together. The open-loop describes a dimension stack terminated with a gap or a critical assembly feature. The closed-loop defines a closure constraint for the assembly, implying that adjustable elements are in the assembly. Thus, the critical difference between the open-loop and closed-loop assemblies is the existence of gap; in the open-loop assemblies, we anticipate that gap dimension must be properly toleranced to allow us to form an engineering fit with another part (for the schema of the open-loop assembly, (please see Fig. 3). Those elements, gap or functional requirement, are the result of part tolerance accumulation. If there are no adjustable components, there is no need for closed-loops - the assembly model is composed only of open-loops [2]. In recent studies, methods for tolerance analysis were compared using the closed-loop examples. The aim was to determine the advantages and shortcomings of each method, along with the differences in output (e.g. [10-15]). To the best of our knowledge, mentioned research studies have not considered the open-loop assemblies. Hence, the contribution of the article at hand is the evaluation of the tolerance analysis methods on open-loop problems. Furthermore, besides the scientific contribution, this article aims to provide the practitioners with a simple review and guidelines for the application of each method. To achieve this, we have compared four different methods: tolerance chart method, Monte Carlo method, vector loop model, and Unified Jacobian-torsor model. Each method was applied to an open-loop assembly, allowing for comparison in performances and outcomes. 2. Methods and materials In this research study, four tolerancing methods were compared: tolerance chart method [16], Monte Carlo method, vector-loop model [15], and Unified Jacobian-torsor (see Section 2.1). Each method is described, along with the steps necessary to apply it. Those include tolerancing problem identification, mathematical modelling, and calculation procedures. 2.1 Used tolerance analysis methods Tolerance chart method is the most frequently used tolerance analysis method in the industry [16], mostly due to its simplicity. It is widely used for solving problems concerned with dimensional tolerances, although the recent improvements enabled its application to geometric tolerances [15]. The method is one-dimensional; in order to apply it to the multi-dimensional geometric tolerances, they must be converted to 1D space [15]. Tolerance chart method can be performed on both the part and assembly level. For assembly level, parts included in the tolerance chain represent one of the tolerance end-points (maximum or minimum). Each part is placed against its mating part in one of its tolerance end-points. As a Advances in Production Engineering & Management 15(1) 2020 45 Kosec, Škec, Miler result, the worst-case tolerance chart method illustrates the minimal and maximal variation of a functional requirement based on the values in the tolerance chain [9]. When performing the tolerance chart method analysis, the first step is to set a goal by labelling the chain starting and ending points [16]. The starting point is selected on one edge and the ending point on the opposite edge of the analysed feature (see Fig. 1). The chain indicator is placed to determine the direction of the dimension vector and is either positive or negative [17]. The vector pointing toward the chain end-point is marked "©", and the vector pointing opposite of the end-point is marked "©". The indicator shows whether to add or subtract dimensions and tolerances during the stack calculation. Additionally, it simplifies the interpretation of tolerance chart results [16]. The resulting dimension chain is the shortest possible and consists only of known dimensions - dimensions set by designers. Tolerance chart method was used in recent studies [5, 6, 9, 15-17], mostly as a reference for the comparison of advanced tolerance analysis methods. Its most important advantage is simplicity; no computational tools are needed as it can be carried out by hand. The downside is that the user has to keep in mind all the standard rules [2, 8] for creating the stacks, making the process error-prone. Besides, the tolerance chart method creates stacks in one direction and ignores the contributions of others, possibly providing unsatisfactory results. 9+0,2 Gj Z_i¡ = l ¿—¡j=l = SO - 3S = 1S TOL = V" TOLi = ±2.9 Final dimension: 1S ±2.9 Fig. 1 Tolerance chart method A plethora of statistical approaches was introduced to conduct non-linear statistical tolerance analysis. A typical example is the Monte Carlo simulation (MCS) based on the algorithm of the same name. It utilises random sampling input values to calculate the output results. For a given input vector x, the number of sampling values n is determined {x1,x2,...,xn}. By using the mathematical model (transfer function) y = f{x) new output vector of same length is found {yi,y2,-,yn}. Finally, the output results y are analysed by calculating statistical data such as mean, standard deviation, or range. Monte Carlo simulation (MCS) is a beneficial tool for tolerance analysis of mechanical assemblies. Its main advantage is flexibility and ability to use various non-normal input or output distributions [18]. A large set of sample parts is created by randomly assigning a tolerance value to each nominal dimension. Values are selected within the tolerance interval to simulate the manufacturing variation [18]. The process is repeated until enough output data is acquired to enable the use of statistical techniques. It allows the calculation of the mean value, standard deviation, range, upper and lower specification limit, and share of rejected samples [19]. Define the problem Analyse the data Assign the expected tolerance distribution to each dimension Apply transfer function Estimate the required number of runs Randomly generate tolerances (at input) Fig. 2 Monte Carlo simulation for implicit assembly constraints 46 Advances in Production Engineering & Management 15(1) 2020 A comparison of the tolerance analysis methods in the open-loop assembly In this article, MCS is applied as an extension to the Tolerance chart method. A modified form of MCS (McCATS) accounting for the implicit assembly variations was used, as suggested in [20]. In the modified simulation, the random parts are sent to the assembly function, which iteratively solves the tolerance chart equations for the dependent assembly variations [20]. The process is repeated until a sample of a suitable size to produce the assembly histogram is created. The steps necessary to carry out the tolerance analysis using the MCS are shown in Fig. 2. The vector loop model is a stack-up technique used to extend the stack analysis to two and three-dimensional assemblies [1]. The idea of the vector loop method is to use vectors to describe the dimensions and associated tolerances. Vectors are arranged in loops to determine the assembly deviations. Tolerance analysis problems are solved using the kinematic concept; contact points are set as kinematic joints. A number of possible motions is defined for each joint (i.e. degrees of freedom), along with the local datum reference plane. Three types of variations are described in vector loop model: dimensional variations (lengths and angles), kinematic variations (small adjustments between mating points, joints) and geometric/feature variations (position, roundness, angularity) [1]. Dimensional and geometric tolerances are described as additional degrees of freedom on the kinematic joints [1]. Kinematic simplification is required to represent geometric tolerances in such way. Thus, in the vector loop model, geometric tolerances are included only at mating points, in the direction defined by the type of kinematic joint [1]. They are described as additional translational and rotational transformations (displacement vectors, rotation matrices) -as gaps with zero-length nominal dimension vectors. The assembly graph is a diagram that represents the analysed assembly, including its parts, dimensions, mating conditions, functional elements, and functional requirements. The graph is used to represent any linear dimension in the assembly as a vector (see Fig. 3). Vectors are connected and form chains or loops, reflecting how assembly parts stack-up together. The associated tolerance is included as a small kinematic adjustment of such a vector (gap) [1, 12]. Such representation allows us to determine the functional requirements of an assembly. Stack-up functions are built by including the vector variations involved in each chain into implicit kinematic equations. As such, they can then be solved using various mathematical approaches [1, 6]. For each part in the tolerance chain, a local datum reference frame (DRF) is added to identify the relevant features of a part for tolerance analysis. DRFs are then connected using datum paths representing geometric layouts, which define the direction and orientation of vectors forming the loop [1]. They are created by stacking and chaining the dimensions that locate the contact point between two parts. After creating datum paths, the vector loops can be created by connecting datums. Loops can be open or closed, depending on the functional requirement of the tolerance analysis. The number of closed loops is calculated as L = J — P + 1, where J is the number of the mating points, and P the number of parts. g A Fig. 3 Assembly graph and the example of vector loops Advances in Production Engineering & Management 15(1) 2020 47 Kosec, Škec, Miler After defining the vector loops, the calculation is carried out [1, 11]. When considering the closed-loop problem, the equations are often non-linear; they must be linearized using direct linearization method [1, 11], producing approximate results. Thus, vector loop, when using direct linearization, is unable to generate true worst-case results [4, 11]. When the open-loop problem is considered, deviations are calculated directly using explicit equations [11]. Unified Jacobian-torsor (JT) method [21] is a 3D tolerance analysis method. It uses the Jaco-bian matrix to relate the functional requirement (FR) and virtual joints displacements. JT advances the punctual small-displacement variables of the Jacobian formulation to represent tolerance zones using the torsor model and interval arithmetic. It offers more output information on the FR, reducing the size of the analysed model since it is no longer point-based [21]. Torsor model uses small displacement screws to establish tolerance zones of points, curves, and surfaces [21, 22]. Each real surface is modelled by a substitution surface defined by a set of screw parameters that are modelling the deviations from nominal geometry [23]. Screw parameters are arranged in torsors containing translational components of a point (u, v, w) and a, fi, S as rotational components with respect to the nominal geometry: T = \fi vL (1) wJ R where R is DRF used to evaluate the screw components. Torsor model can fully define the tolerance zones due to its ability to shape spatial volumes within which the surfaces are deviating [10]. The procedure of Unified Jacobian-torsor method consists of 4 steps [24]. The first step is to identify all functional elements (FEs) affecting the FR by distinguishing kinematic chains involving the functional condition or part under study. Functional element can be any point, curve or surface of a part and creates internal or kinematic pairs [21]. The second step is to associate a torsor or screw parameter to each element (surface, axis) of the kinematic chain. Torsors express the degrees of freedom and the allowable element displacements and their bounds. Small displacements are applied to parts' geometrical features affecting the FR [21], after which the Jacobian matrix is used to determine relative positions and orientations of torsors within the chosen kinematic chain (step three) [22]. The final step is to combine torsor and Jacobian to provide a matrix equation. Solving a resulting matrix using interval algebra provides the functional condition bounds. 2.2 Assembly model for case study The above-described methods were compared by analysing a 3D tolerancing problem. The assembly consisting of the cantilever and the rotating handle (open loop) was used as an example. Thus, both the dimensional and geometric tolerances were considered. The functional requirement deviations are assessed using each of the methods, while the results are compared in Section 3.5. The nominal dimension (DIM), upper deviation limit (UDL), and lower deviation limit (LDL) were calculated. The comparison is focused on the similarities and differences between results obtained by each method. Differences in procedures and calculation approaches are also observed. A simple rotating handle assembly consisting of four parts was used to carry out the comparison between the methods (see Fig. 4). The pole (2) is fixed to the bottom plate (1), while the lever (3) is mounted onto the journal located on the pole. The handle (4) is installed into the bore located on the lever. Tolerances were assigned to all the dimensions apart from a distance between the handle (4) and the base plate (1), which is selected as a functional requirement). 48 Advances in Production Engineering & Management 15(1) 2020 A comparison of the tolerance analysis methods in the open-loop assembly The positional tolerance between the top surface of the base and its cut-out was included. The contact between the base and the cylindrical base of the pole is considered ideal. The parallelism tolerance between the axis of the cylindrical pin located on the pole and the bottom surface of the pole was also included. The lever is mounted onto the pole pin (see Fig 4) by clearance fit 045 H8/g7. On the opposite side of the lever, the handle is mounted into the bore with a clearance fit 045 G6/h7. Regarding the geometric tolerances, the parallelism between two lever bores and perpendicularity between the handle and the mounting sleeve wall were required. Each method was then applied to the above-described open-loop assembly. The results were compared according to three criteria: • identification of the contributing tolerances, • calculation of the dependent dimension (functional requirement), • analysis of calculation differences compared to the assemblies with closed loops. An assembly graph was created for each of the methods except for the Monte Carlo simulation, as it is based on Tolerance chart method. 3. Results and discussion 3.1 Tolerance chart results Tolerance chart method is mostly used for dimensional tolerances, even though the recent modifications have enabled the analysis of geometric tolerances as well [15]. The geometric tolerances are to be transformed into their dimensional counterparts. Yet, such transformation does not account for the angular surface deviation. In this article, the tolerance chart method is applied only to dimensional tolerances. Tolerance chain consisting of base plate bore depth (a), length between the pole basis and pole journal axis (b), tolerance fit between pole journal and lever bore (c and d), distance between lever bore axes (e), and tolerance fit between lower lever bore and handle (f and g). As mentioned, the distance between the top surface of the base plate (part 1) and handle (4) is a functional requirement. Indices U and L were included to denote the upper and lower deviation limit, respectively. Advances in Production Engineering & Management 15(1) 2020 49 Kosec, Škec, Miler Dimension + - TOL. a 15 0 b 317 0 c 22.5 cut = -0.009/2; clt = -0.034/2 d 22.5 dut = 0.039/2; dlt =0 e 150 0 f 22.5 fut = 0.025/2; dlt = 0.006/2 g 22.5 gut = 0/2; glt = -0.025/2 Sum 362 210 Nominal functional requirement dimension: DIMtc = -a + b + c- d- e- / + g = 152 mm Stack equation for the upper deviation limit: UDMtc = -a + b + cut -dlt -e - flt +gut = -0.008 mm Stack equation for the lower deviation limit: UDMtc = -a + b + clt -dut — e — fut +glt = -0.062 mm Fig. 5 Application of tolerance chart method The tolerance stack coordinate system is defined next; the starting point is set at the base plate surface (1). The upward dimension is shown in Fig. 5 is selected as positive and marked with the indicator "©", while the downward is negative and marked with "©". The direction of the tolerance chain is chosen arbitrarily, but it is important to respect the specified direction along the chain. Finally, the results are calculated by adding and subtracting values along the tolerance chain and shown in Fig. 5. 3.2 Monte Carlo simulation results Monte Carlo simulation was applied following the procedure explained in Section 2.1. Determining the appropriate distribution to each of the tolerances was the crucial step, as it affects the results. The distribution of geometric tolerances along with the interval between the upper and lower deviation limit most frequently follows the normal distribution. Tolerance fits are asymmetrical, requiring the use of the skewed distribution according to [19]. Distribution of input values is described using ± 3o process range (6ct). Since data about the manufacturing process was not available, 3DCS CATS software was used to determine the distribution models of tolerance fits. According to 3DCS, tolerance fits have unimodal continuous probability distribution called Pearson 1. Since tolerance limits assigned to dimension c are negative, the distribution model is skewed left from the nominal dimension. The same can be concluded for the tolerance assigned to dimension g. On the other hand, tolerance limits assigned to dimensions d and f are positive, and distribution is right-skewed. Runs 2000 1SÛ Nominal 152.000 mm 135 Mean 151.999 mm STD 0.018 mm 100 6STD 0.110 mm 75' LSL 151.000 mm USL 153.000 mm 50 EST. TYPE Pearson I 25' EST. LOW 151.958 mm EST. HIGH 151.995 mm 0 - 151,90 EST. RANGE 0.085 mm + 3STD 152,0S4fmnn| Fig. 6 Monte Carlo assembly results and histogram 50 Advances in Production Engineering & Management 15(1) 2020 A comparison of the tolerance analysis methods in the open-loop assembly Next step is to define the variation model function. Since Monte Carlo is applied to tolerance chart method, tolerance stack equations are used to define it. After running the simulation for n = 2000 times with randomized input variables, an output model for FR was created. A variation analysis provides descriptive statistics, inferential statistics data, and a histogram (shown in Fig. 6). By adding and subtracting the input variables using the variation model functions, distribution of FR tolerances was found. The resulting functional requirement distribution is also Pearson 1. However, it is not as profoundly left- or right-skewed as are the input variables. 3.3 Vector-loop results An assembly graph describing the open-loop of the assembly and its vector loop tolerance chain (or datum path) was created. It was used to identify of the number of vector chains and loops involved in the assembly (Fig. 4). Since each part is in contact with its two neighbouring parts only once, this assembly contains one open loop. Same can be seen on the assembly graph (Fig. 7) where each arrow is representing the contact between parts. Vector loop is open at the gap (noted g) between the Handle (Part 4) and Base (Part 1). The datum path (Fig. 7, right) connects the point, surface, axis or DRF of a part with next part's point, surface axis or DRF. DRFs have been assigned to each part with respect to the origin coordinate system at the top of the Base (Part 1). All the DRFs have a horizontal x-axis and vertical z-axis. Origin coordinate system is set in such a way that positive direction of Z0 axis corresponds with the positive direction of a tolerance chain in Tolerance chart method. This eases the tracking and method comparison. The geometric tolerances were also accounted for. Each tolerance was represented as an additional vector of magnitude equal to ± i/2, where t is the width of corresponding tolerance field (see Fig. 4; 0.2 for the positional, and 0.1 for parallelism and perpendicularity tolerances). The additional vectors represent gaps between parts contacting points and were denoted based on the corresponding nominal dimension. The position tolerance on the Base cut-out with respect to the datum A (apos) is represented as a translation vector of the surface in the z-direction (Fig. 7). The parallelisms applied to the Pole's pin (¿par), and Lever holes (epar) with respect to the datum B were also represented as translation vector along the z-axis [1]. Perpendicularity applied to the horizontal axis of the Handle (^per) with respect to the datum D can be described as a translation along x-axis [1]. According to the assembly graph there are / = 3 contacting points and P = 4 parts, resulting in 0 closed loops (Eq. 1 was used). There is also one open-loop functional requirement. c+c Var. Tol. description Gap vector length ßpos Position (Base top to cutout) ±0.1 ¿par Parallelism (Pole's to bottom) ±0.05 Cd Dimensional tol. (Pole's pin) Cdu = -0.009; Cdl = -0.034 dd Dimensional tol. (Lever's upper hole) ddu = 0.039; ddl = 0 epar Parallelism (Lever hole axes) ±0.05 fd Dimensional tol. (Lever's lower hole) fdu = 0.025; fdl = 0.009 f)d Dimensional tol. (Handle) fdu = 0.0; fdl = -0.025 hpar Perpendicularity (Handle axis) ±0.05 Uppervalue of functional requirement: FRU = -(a- apos) + (b + bpar) + c + cdu d + dd] . _ - 2 2 ^ 6parJ f~ fd\ , 9 + S'du , . -, CI Hi —2— + —2- Par = 152.241 mm Lower value of functional requirement: Ffi] = -(a + apos) + (b-bpar) + c + cdl d + ddu par J f + fdu | a 2 9di -(e + ePar) -hpar = 151.680 mm Fig. 7 Vector loop assembly graph and results Advances in Production Engineering & Management 15(1) 2020 51 Kosec, Škec, Miler 3.4 Unified Jacobian-torsor results Before creating the assembly graph (Fig. 8), it was necessary to identify the functional elements (FE) and functional requirements (FR). Also, it was required to differentiate between the internal and kinematic pairs. For the assembly at hand, there are four internal and one kinematic pair. First internal pair (FE0-1) is located on the Base (Part 1), as the positional tolerance defined between its top and cut-out surface corresponds to functional surfaces 0 and 1 on the assembly. The parallelism tolerances define FE2-3 and FE4-5. Internal pair FE6-7 is defined by the perpendicularity tolerance set on the Handle (Part 4). Only kinematic pair (FE1-2) is set between the Base cut-out and Pole's bottom. However, the contact is assumed to be ideal so that it will not impact the analysis. Two more contacts defined by tolerance fits (between Pole and Lever, and between Lever and Handle) were not set as kinematic pairs even though they are in physical contact. This means they are defined as important conditions to be satisfied between two FEs. So, according to [13], they are then defined as functional requirements that will be taken into account in the analysis as kinematic pairs. ¿3 *4 ÏS -t /S t 1/7 > Lever FR3.4 % /FE« A 5 ~ Functional surface- { i ! Parts: Imcrnalpai:: ------ Kinematic pair: I fr a Handle ■Ov FE,, 7 I 6 ............ -- 1 X, i'1 X0 Upper value of functional requirement: FRU = FRnomina]+FR = 152.28 mm Lower value of functional requirement: FR] = Fflnominai +FR = 151.143 mm Fig. 8 Jacobian-torsor method and results Jacobian matrix and small-displacement torsor vector were calculated for each internal and kinematic pair. A small-displacement torsor vector was also calculated for each FE (based on torsor representing the tolerance zone [21]): [Fß]T = u v w £ ß ^ ü v w a ß S_ , [FEi] = FR U V w U V w a ß S_ a ß 5. (2) FEi For each tolerance, translational and rotational components inside the tolerance zone were determined [21]. Since the direction of a functional requirement is along the z-axis and rotations that would influence functional requirement are around x and y-axis, w, a and p component must be calculated. Contact between functional surfaces 0 and 1 in the FE0-1 is a planar contact with normal containing one translation component (w) and two rotational components (a, ft) [12, 21]. Components are calculated using the equations for planar surface according to [21]. Internal pairs FE2-3, FE4.5, and FE6-7, and functional requirements FR3.4 and FR5-6 are defined by the tolerance fits. They use translational components v and w and rotational components p and S of a slipping pivot with the axis [21]. Below is a table containing displacements torsors for each internal and kinematic pair. 52 Advances in Production Engineering & Management 15(1) 2020 A comparison of the tolerance analysis methods in the open-loop assembly FEo-i 0 0 01 0.0013 -0.0013 0 0 0 =0¡1 0.0013 -0.0013 0 Table 1 Displacement torsors for each internal and kinematic pair FE2-3 0 0.05 0.05 0 0 -0.05 -0.05 0 0.004 -0.004 0.004 -0.004. FR3-4 0 0.00365 0.0045 0 0.00365 0 -0.0003 0.0045 0 0.0004 -0.0003 0.0004. FE45 0 0.05 0.05 0 0 -0.05 -0.05 0 0.004 -0.004 0.004 -0.004. FRs-6 0 0.025 0.025 0 -0.002 0 0.045 0.045 0 -0.0004 -0.002 -0.0004. FE6-7 0 0 0.05 0.05 0 0.001 -0.001 -0.05 -0.05 0 0.001 -0.001 Jacobian matrix is calculated according to the procedure presented in [21]. Its purpose is to calculate the effect of the traditional torsor set for each functional element (FEE) on the functional requirement (FR) of the assembly [21]. Finally, after calculating small-displacement torsor vectors and Jacobian matrices for each FE, the same can be done for FR: FR=J^FE = [ -0.058 -0.434 0.239 0.184 0. 665 -0. 947 0.001 -0.001 0.005 -0.011 0.004 ]' -0.010J (3) 3.5 Comparison of the results obtained by different methods After analysing the assembly presented in Section 3.1 using the four methods, results are presented in Table 2. Abbreviations are used to ease the result disambiguation; TC was used for tolerance chain method, MC for Monte Carlo simulation, VL for the vector loop method, and JT for the Unified Jacobian-torsor method. Since it is not possible to include geometric tolerances in TC and MC analysis, two results were provided for VL and JT methods. The first batch of results included dimensional tolerances, while the second includes both. Besides the quantitative analysis results, qualitative properties such as the scope and perceived complexity of each method were assessed. Proprietary CAT tools that are used in day-to-day work are often perceived as black boxes. That means that the users are frequently not familiar with the underlying processes and mathematical models. Besides, the tolerance analysis methods used in CAT tools are often not complying to the technical standards. For this reason, we have analysed underlying tolerance analysis methods, aiming to determine their advantages and shortcomings. When considering the dimensional tolerances (DT) exclusively, TC, MC, and VL provide similar results, with FRu being practically equal. Contrary to the upper value of the functional requirement, the deviations in lower (FRl ) are greater - TC and VL provide more conservative results when compared to MC. A significant deviation in FRu was found when calculating it using JT. Unlike other analysed methods, in Jacobian-torsor method tolerance analysis is carried out using a tolerance zone as a basis (instead of points), causing the afore-mentioned variations in results. By using zones and surfaces instead of points, it is possible to create a more credible representation of a realistic case. Table 2 Method comparison Scope FRu FRl A Applicability Complexity Tolerance chain method DT 151.992 151.938 0.054 For simple 1D tasks Simple, carried out by hand Monte Carlo simulation method DT 151.995 151.958 0.037 Simple tasks, statistical analysis Statistical tools required Vector Loop method DT 151.991 151.930 0.061 Multi-dimensional Carried by D> 152.243 151.143 1.100 hand/more complex Unified Jacobian-torsor method DT 152.369 151.946 0.423 Multi-dimensional, automation Requires mathematical tools, enables D> 152.665 151.053 1.612 automation Advances in Production Engineering & Management 15(1) 2020 53 Kosec, Škec, Miler The MC method is the least conservative due to its statistical approach - when carrying out the tolerance analysis using the MC method, most extreme cases are excluded from the analysis and counted as write-offs. The advantage of such an approach is that it reduces the cost of manufacturing equipment; it is less expensive to write-off a portion of parts, then to purchase more accurate manufacturing tools. Thus, the analysis method should be selected in accordance with the manufacturing process. Statistical approaches are suitable when analysed products are manufactured in large series, while prototypes and one-of-a-kind products warrant the use of more complex and conservative methods, such as JT. The applicability of methods regarding the tolerancing problem dimensionality should be addressed next. As applied in this study, by using TC and MC only 1D problem containing dimensional tolerances can be solved. This drawback can be partially mitigated by converting the geometric tolerances into their dimensional counterparts; however, methods remain limited to 1D problems. In comparison, VL and JT were developed with having the 2D and 3D tolerancing problems in mind. Besides the dimensional tolerances, both methods can be used to analyse the geometric tolerances as well. However, there is a significant difference between VL and JT in terms of tolerance representation. The former, vector loop, observes a set of tolerances simultaneously, forgoing the possible interactions among them. The latter, Unified Jacobian-torsor, includes both the translational and rotational components, thus including different tolerances as complementary. The procedure complexity of each method should also be considered. TC is by far the most straightforward method and can be carried by hand. It is suitable for simple tolerancing problems that engineers solve daily. Second is the VL, which requires an additional schema of the vector loop. By procedure complexity, MC comes next It is a statistical method, meaning that it requires a large sample in order to provide significant results - experience with similar parts and their tolerances is necessary. The last method is JT, which is found to be the most complex. To carry it out, it requires the detection of functional elements, functional requirements, and kinematic pairs. When comparing the method performance on open and closed-loop tolerancing, changes were detected only in VL. The vector-loop method is affected by the procedure system of open and closed loops. In cases where only open-loop tolerances are used, the vector loop method is reduced to explicit equations. This allows for a direct calculation of the functional requirement values. In other words, the VL method loses its advantage to TC. The limitations of the study should also be considered. Each of the methods is carried out strictly according to the literature, without additional data manipulation (for example, geometric tolerances were not converted to dimensional ones). Additionally, when carrying out MC, it should be stressed that previous knowledge about the manufacturing process and manufacturing tool properties is necessary to enable satisfying approximation of tolerance distribution. Lastly, during the planning of the product design process, in addition to tolerance analysis methods, engineers should also consider applying the tolerance optimisation methods. Several studies have been carried out on the subject, such as [25, 26]. Using optimisation algorithms to tolerancing problems allows us to find the optimal trade-off between the tolerances, manufacturing costs, and quality loss [25]. Such an approach would surely increase the design effectiveness, increasing its market success. 4. Conclusion Simple tolerancing problem was used to assess the similarities and differences between four tolerance analysis methods: Tolerance chart ("Worst-case analysis"), Monte Carlo Simulation method, Vector-loop method, and Unified Jacobian-torsor. Open-loop assembly was used to illustrate the problem-solving process using each of the methods. Based on the results, the authors have concluded the following: 54 Advances in Production Engineering & Management 15(1) 2020 A comparison of the tolerance analysis methods in the open-loop assembly • Tolerance chart and Monte Carlo Simulation methods do not account for the geometrical tolerances. This results in overly optimistic results; however, both methods are only suitable for solving simple, 1D tolerancing problems. • The unified Jacobian-torsor method was found to be most conservative (i.e. provided the most substantial deviations in functional requirement), followed by Vector-loop, Tolerance chart, and Monte Carlo Simulation, respectively. • Tolerance chart is the simplest and thus suitable for solving many day-to-day tolerancing problems. Monte Carlo Simulation and Unified Jacobian-torsor require more detailed analysis and know-how and are suitable for more pressing problems. Vector loop can be considered the middle ground - it offers good results at the moderate complexity. • When comparing the method performance in open-loop assemblies to closed-loop ones, differences are detected only in Vector-loop method. The field of tolerance analysis is fruitful, and there is more work to be done. Following this study, the authors aim to analyse the performance of tolerancing methods by carrying out an industrial case study. The part deviations measured during the quality assurance are to be compared to the values provided by analysis methods, providing additional insight. Acknowledgement This paper reports on work funded by the Croatian Science Foundation project IP-2018-01-7269: Team Adaptability for Innovation-Oriented Product Development - TAIDE. References [1] Polini, W. (2011). Geometric tolerance analysis, In: Colosimo, B., Senin, N. (eds.), Geometric tolerances: Impact on product design, quality inspection and statistical process monitoring, Vol. 2, Springer, London, United Kingdom, 39-68, doi: 10.1007/978-1-84996-311-4 2. [2] International organisation for standardisation (2011). ISO 8015-2011 - Geometrical product specifications (GPS) - Fundamentals - Concepts, principles and rules, ISO, Geneva, Switzerland. [3] Sigurdarson, N., Eifler, T., Ebro, M. (2018). The applicability of CAT tools in industry - Boundaries and challenges in tolerance engineering practice observed in a medical device company, Procedia CIRP, Vol. 75, 261-266, doi: 10.1016/j.procir.2018.04.066. [4] Corrado, A., Polini, W. (2017). Manufacturing signature in jacobian and torsor models for tolerance analysis of rigid parts, Robotics and Computer-Integrated Manufacturing, Vol. 46, 15-24, doi: 10.1016/j.rcim.2016.11.004. [5] Ramnath, S., Haghighi, P., Chitale, A., Davidson, J.K., Shah, J.J. (2018). Comparative study of tolerance analysis methods applied to a complex assembly, Procedia CIRP, Vol. 75, 208-213, doi: 10.1016/j.procir.2018.04.073. [6] Chen, H., Jin, S., Li, Z., Lai, X. (2014). A comprehensive study of three dimensional tolerance analysis methods, Computer-Aided Design, Vol. 53, 1-13, doi: 10.1016/j.cad.2014.02.014. [7] Morse, E.P., Shakarji, C.M., Srinivasan, V. (2018). A brief analysis of recent ISO tolerancing standards and their potential impact on digitalization of manufacturing, Procedia CIRP, Vol. 75, 11-18, doi: 10.1016/j.procir.2018. 04.080. [8] American Society of Mechanical Engineers (2004). ASME Y14.5 - Mathematical definition of dimensioning and tolerancing principles, ASME, New York, USA, 1-15. [9] Shen, Z., Ameta, G., Shah, J.J., Davidson, J.K. (2005). A comparative study of tolerance analysis methods, Journal of Computing and Information Science in Engineering, Vol. 5, No. 3, 247-256, doi: 10.1115/1.1979509. [10] Marziale, M., Polini, W. (2011). A review of two models for tolerance analysis of an assembly: Jacobian and torsor, International Journal of Computing Integrated Manufacturing, Vol. 24, No. 1, 74-86, doi: 10.1080/0951192X. 2010.531286. [11] Chase, K.W., Magleby, S.P., Gao, J. (2004). Tolerance analysis of 2-D and 3-D mechanical assemblies with small kinematic adjustment, Advanced Tolerancing Techniques, Vol. 218, 1869-1873. [12] Ghie, W. (2009). Statistical analysis tolerance using jacobian torsor model based on uncertainty propagation method, The International Journal of Multyphysics, Vol. 3, No. 1, 11-30, doi: 10.1260/175095409787924472. [13] Ghie, W., Laperriere, L., Desrochers, A. (2010). Statistical tolerance analysis using the unified Jacobian-Torsor model, International Journal of Production Research, Vol. 48, No. 15, 4609-4630, doi: 10.1080/002075409028 24982. [14] Wang, Y. (2008). Closed-loop analysis in semantic tolerance modeling, Journal of Mechanical Design, Vol. 130, No. 6, Article No. 061701, doi: 10.1115/1.2900715. [15] Schleich, B., Wartzack, S. (2016). A quantitative comparison of tolerance analysis approaches for rigid mechanical assemblies, Procedia CIRP, Vol. 43, 172-177, doi: 10.1016/j.procir.2016.02.013. Advances in Production Engineering & Management 15(1) 2020 55 Kosec, Škec, Miler [16] Magadum, R., Allurkar, B.S. (2015). A comparative study of fasteners tolerance analysis methods, International Journal of Scientific & Engineering Research, Vol. 6, No. 6, 700-705. [17] Fischer, B.R. (2015). Mechanical Tolerance Stackup and Analysis, CRC Press, Boca Raton, USA. [18] Yan, H., Wu, X., Yang, J. (2015). Application of Monte Carlo method in tolerance analysis, Procedia CIRP, Vol. 27, 281-285, doi: 10.1016/j.procir.2015.04.079. [19] Schenkelberg, F. (2016). Statistical tolerance analysis - Basic introduction, FMS Reliability Publishing, Los Gatos, USA. [20] Chase, K.W., Gao, J., Magleby, S.P. (1995). General 2-D tolerance analysis of mechanical assemblies with small kinematic adjustments, Journal of Design and Manufacturing Automation, Vol. 5, 263-274. [21] Desrochers, A., Ghie, W., Laperriere, L. (2003). Application of unified jacobian-torsor model for tolerance analysis, Journal of Computing and Information Science in Engineering, Vol. 3, No. 1, 2-14, doi: 10.1115/1.1573235. [22] Ghie, W. (2012). Tolerance analysis using Jacobian-Torsor model: Statistical and deterministic applications, In: Cakaj, S. (ed.), Modelling simulation and optimization - Tolerance and optimal control, InTech, Rijeka, Croatia, 147-160, doi: 10.5772/9043. [23] Desrochers, A., Delbart, O. (1998). Determination of part position uncertainty within mechanical assembly using screw parameters, In: ElMaraghy H.A. (ed.), Geometric design tolerancing: Theories, standards and applications, Springer, Boston, USA, 185-196, doi: 10.1007/978-1-4615-5797-5 14. [24] Peng, H., Lu, W. (2017). Three-dimensional assembly tolerance analysis based on the Jacobian-Torsor statistical model, In: Proceedings of 3rd International Conference on Mechatronics and Mechanical Engineering (ICMME 2016), Les Ulis, France, doi: 10.1051/matecconf/20179507007. [25] Siva Kumar, M., Kannan, S.M., Jayabalan, V. (2009). A new algorithm for optimum tolerance allocation of complex assemblies with alternative processes selection, The International Journal of Advanced Manufacturing Technology, Vol. 40, No. 7-8, 819-836, doi: 10.1007/s00170-008-1389-5. [26] Ramesh Kumar, L., Padmanaban, K.P., Balamurugan, C. (2016). Optimal tolerance allocation in a complex assembly using evolutionary algorithms, International Journal of Simulation Modelling, Vol. 15, No. 1, 121-132, doi: 10.2507/IISIMM15(1)10.331. 56 Advances in Production Engineering & Management 15(1) 2020 Advances in Production Engineering & Management Volume 15 | Number 1 | March 2020 | pp 57-68 https://doi.Org/10.14743/apem2020.1.349 ISSN 1854-6250 Journal home: apem-journal.org Original scientific paper Awareness and readiness of Industry 4.0: The case of Turkish manufacturing industry Sari, T.a*, Güle?, H.K.b, Yigitol, B.a Department of Management Information Systems, Faculty of Social Sciences and Humanities, Konya Food and Agriculture University, Konya, Turkey bDepartment of International Trade and Business, Faculty of Social Sciences and Humanities, Konya Food and Agriculture University, Konya, Turkey A B S T R A C T A R T I C L E I N F O The concept Industry 4.0 (I4.0) represents intelligent production processes combining cyber and physical systems through a set of technologies such as internet of things, big data and cloud computing. Transition to Industry 4.0 is expected to cause formidable structural changes, productivity increments and competitiveness in manufacturing industry in all over the world. This study aimed to investigate the general approach to the concept of Industry 4.0 and levels of adoption of the basic Industry 4.0 technologies in manufacturing firms across Turkey. For this purpose, a survey was conducted with 427 firms with various sizes (micro, small, medium and large) operating in six sub-sectors (automotive; electronic; machinery; chemical; food; and textile) of Turkish manufacturing. The paper examined nine I4.0 technologies: autonomous robots, big data applications, cloud computing, cyber security, simulation approaches, additive manufacturing, system integration, internet of things, and augmented reality. The results revealed that, there is a significant correlation between the degrees of importance and implementation of the basic Industry 4.0 technologies. Moreover, I4.0 implementation degree increases as the firm size increases. The top three industries in Turkish manufacturing that use the most basic Industry 4.0 technologies are automotive industry, electrical and electronics, and machinery, respectively. The analyses are aimed to achieve a better understanding of the concept Industry 4.0 by comparing different groups of manufacturers. © 2020 CPE, University of Maribor. All rights reserved. Keywords: Industry 4.0; Additive manufacturing; Autonomous robots; Cloud technologies; Cyber security; Internet of things (IoT); Big data; Augmented reality *Corresponding author: tugba.sari@gidatarim.edu.tr (Sari, T.) Article history: Received 3 December 2019 Revised 9 March 2020 Accepted 17 March 2020 1. Introduction Industry 4.0 (I4.0) defines a new technological era (i.e. the intelligent factory system that combines embedded system manufacturing technologies and intelligent production processes) that will change the industrial and production processes [1], while providing production ecosystems produced by intelligent systems with autonomous features such as self-structuring, self-monitoring and self-healing [2]. Integration of these systems creates an intelligent manufacturing environment which yields higher productivity and faster production of customized and high-quality products. The main idea in here is to capture a competitive industrial advantage with the help of the state of art technologies. This concept not only changes the dynamics of competition between the nations and the companies in the business world, but also reshapes the way of doing business and setting strategies for the future. Just like other industrial revolutions, The 4th Industrial Revolution is a process that requires structural changes, new business models, de- 57 Sari, Güleg, Yigitol structive innovations and paradigm shifts from centrally controlled to decentralized production processes [3, 4]. Another aspect of Industry 4.0 is its potential to contribute to the sustainability of future supply chains with optimum use of resources [5]. The concept of Industry 4.0 was first used in 2011 at the Hannover Fair in Germany [6]. Then it was developed to create a national policy by The German Government and dispersed to other countries in the world under the label of The 4th Industrial Revolution. Strategic initiatives of many countries around the world aim to catch the benefits offered by this new digital era. However, the structure of the manufacturing industry of each country and their advancement in production technologies differ. For this reason, some indices are needed to measure the compliance of countries with industry 4.0. The Roland Berger Readiness Index, defines four Industry 4.0 readiness categories for European countries: (1) Frontrunners with high manufacturing share and high readiness: Germany, Sweden and Austria; (2) Potentialists with low manufacturing share and high readiness: Belgium, Finland, UK, France, Netherlands; (3) Traditionalists with high manufacturing share and low readiness: Slovenia, Slovakia, Czech Republic, Hungary and Lithuania; (4) Hesitators with low manufacturing share and low readiness: Portugal, Greece, Poland, Bulgaria, Italy, Spain and Turkey [7]. Although Turkey seems to be one of the hesitators with low digitalization level in between European Countries, TUSIAD (Turkish Industry and Business Association's industry report reveals the potential of Turkish manufacturing industry for Industry 4.0 transformation. The purpose of this study is to investigate the approaches and readiness of Turkish manufacturing firms to Industry 4.0 by using self-assessment tools. For this purpose, a survey was conducted with firms operating in six different sub-sectors of Turkish manufacturing. The outline of the article is designed as follows: The next section gives theoretical background of the term Industry 4.0 and the key technologies it covers. The third section presents a comprehensive review of the existing literature on Industry 4.0 including articles and reports. Fourth section presents the methodology of data collection, and the descriptive statistics. Section five presents the basic findings of the survey and the results of the analysis while section six presents concluding remarks of this study and discusses future studies. 2. Theoretical background of Industry 4.0 The number of studies on industry 4.0 is increasing since it was first introduced in 2011, and yet there is no single common definition of industry 4.0. Based on the common points of the definitions, industry 4.0 can be defined as the production systems that production is made by smart and autonomous machines that can communicate and coordinate with each other [9]. In the new era of digitalization, companies are expected to face a new dynamic environment focused on machine learning and real time data processing. Firms will be able to accelerate their data collection and analysis methods through flexible production process. They would produce higher quality products with lower costs and hence increase their efficiency with the opportunities offered by Industry 4.0. In literature, Industry 4.0 encompasses nine technologies which are big data and analytics, autonomous robots, simulation, horizontal and vertical system integration, the internet of things, cybersecurity, the cloud, additive manufacturing, and augmented reality [10]. The classifications of key technologies of Industry 4.0 and the brief descriptions of each technique are as follows: Big data and analytics: Large-scale datasets are a combination of large, complex, diverse and heterogeneous data generated from a variety of sources, such as click flows, sensors, video sharing, business processes, and social networks [11]. Data and data analytics are critical in digital and smart factories emerging with Industry 4.0. Therefore, data from many different sources need to be collected, classified, evaluated and made available for decision-making in real time. Autonomous robots: The organizational structure that promotes human-robot collaboration in the production process is emphasized. Robots will interact with each other in this structure via internet. These collaborative robots (Cobots) will not only work closely with other robots, but also humans and learn from them for better practices [12]. 58 Advances in Production Engineering & Management 15(1) 2020 Awareness and readiness of Industry 4.0: The case of Turkish manufacturing industry Simulation: The physical production processes in factories can be reflected by a virtual environment with the help of simulation technologies. For example, quality improvement in functions such as testing, setup and installation can be performed in virtual environments. In addition to quality enhancement, cost savings can be achieved by decreasing setup and maintenance times with simulation technologies. Horizontal and vertical system integration: Companies, departments, functions and capabilities will become more consistent with Industry 4.0, and cross-company and universal data integration networks will evolve to autonomous value chains. Horizontal integration allows firms to create, an effective ecosystem to share information, financial resources and materials, while vertical integration enables them to obtain flexible and restructured production systems [12]. The internet of things (IoT): "The IoT refers to an inter-networking world in which various objects embedded to electronic sensors, actuators or other digital devices where they can connect to each other and exchange the data they collected" [14, 15]. IoT technologies can be used for automation of various operations such as remote controlling, machining, lighting and heating. Furthermore, decision making is decentralized by machine to machine (M2M) interactions via internet of things. Cybersecurity: Perhaps one of the most discussed negative aspects of digitalization is the issue of cyber security. Users of Industry 4.0 technologies will not only face the traditional cyber security challenges, but also the unique security and privacy challenges inherent in the digital Industry [2]. The mechanization of production processes and internet-based interaction and communication between machines increase the need for security especially in the critical industrial systems. Therefore, secure and reliable systems are vital for sustainability in Industry 4.0. The cloud: Due to data-based production systems, the amount of data produced increases and effective management and storage of this data becomes important. Unlike the traditional storage services, cloud technologies provide storage space in a smaller area for a large amount of data from various sources [16, 17]. "Machine data and functionality will be deployed to the cloud, enabling more data-driven services for production systems" [10]. Additive manufacturing: Additive manufacturing technologies are the technologies that enable a model to be produced by means of three-dimensional computer aided design systems without the need of any process planning [18]. The system receives the information needed from a computer-aided design (CAD) program [19]. This technology facilitates the production of products with complex production processes. Augmented reality: "Augmented reality (AR) can be defined as a computer graphics technique where virtual symbols are superimposed to a real image of the external world" [20]. Augmented reality technologies ease daily lives of the users by providing information about objects [21]. 3. Literature review The literature on the readiness and preparedness to Industry 4.0 is increasing with different types of publications. Our review on the current literature includes both academic papers and industry reports is given in Table 1. Table 1 Literature review on Industry 4.0 Author Topic/Aim is Method Results Ratnasingam et al. (2019) [22] to measure Industry 4.0 readiness of the firms operating in the furniture sector in Malaysia. Survey method - Machining centres and finishing processes are the processes requiring the most technological infrastructure - Driving forces on the way to digitalization are higher production capacity, cost, product characteristics and government policy. - The sector is not yet ready for Industry 4.0 Machadoa et al. (2019) [23] to appraise the readiness of seven companies on the digitalization Survey and case study - Firms are at the beginning of the Industry 4.0 process. - One of the biggest obstacles related to Industry 4.0 is stated as lack of knowledge Schumacher et to present a model to manu- Model design - Model consists of 65 critical success factors to assess maturi- al. (2019) [24] facturing companies related to Industry 4.0 ty, and roadmaps with 10-step. - The model was applied on manufacturing companies in Hungary, Austria, Germany, China, Slovakia and India. Advances in Production Engineering & Management 15(1) 2020 59 Sari, Güleg, Yigitol Table 1 (continuation) Pacchini et al. (2019) [25] to propose a model of readiness of a manufacturing companies for Industry 4 Model design - The model covers Industry 4.0 principles and technologies. - This model was implemented in auto-parts manufacturing company in Brazil. - The model provides information about the challenges in the transition to I4.0 to managers, and contributes to both theory and practice. Castro et al. (2019) [26] to present a model for self-assessment on the i4.0 readiness Model design - This model covers six dimensions: smart factory, data-driven services, smart operations, strategy and organization, smart products and human resources. - The result of this study shows that how a company can make better Industry 4.0 readiness level by using SHIFTo4.0. Stentoft et al. (2019) [27] to measure digitalization readiness level of small and medium-sized manufacturers Survey method - Danish SMEs have a low to moderate Industry 4.0 preparation. - Incentive implementations cause an in-crease in Industry 4.0 readiness Castelo-Branco et al. (2019) [3] to measure factors and the Factor and degree of adoption of Industry cluster analyses 4.0 - There is a large distribution between countries in required conditions for that readiness. Nick et al (2019) to search Industry 4.0 ap-[28] proaches in some countries such as Germany, Austria and Hungary Model design - This results help to capture the different phases of digitization and Industry 4.0 in these countries - This study defined objectives, strategies and also solutions for Industry 4.0 related problems Medic et al. (2019) [ 29] to evaluate the usage of advanced manufacturing technologies in Serbia FAHP and PRO-METHEE model - The results show that, digital data sharing among supply chain members, production planning and production control systems are important issues in the context of I4.0 implementation. Resman et al. (2019) [30] to propose a new model based Model design on I4.0 technologies for smart factory planning - The proposed model is easy to use and offers a more reliable and simple modelling of smart factory compared to reference architectural model of I4.0. Mittal et al. (2018) [31] to analyse available systems and Industry 4.0 maturity models Model design - This study provides information to help improvement of realistic smart manufacturing. Vieira et al. (2018) [32] to introduce a R&D agenda for Literature re-discrete-even simulation view (DES) in the context of I4.0 - The significant DES characteristics are: automation of data exchange, automatically generated simulation models and visualization. Kusiak (2018) [33] to provide a theoretical framework of smart manufacturing Theoretical study - Manufacturing technology, sustainability, networking, data analysis, material science and predictive engineering are the essential issues of smart manufacturing. Basl (2017) [34] to analyse the level of knowledge of firms and employees about I4.0. Survey method - 40 % of companies deal with Industry 4.0 for more than one year. - 75 % of companies says that their main reason for not implementing Industry 4.0 is low awareness of the topic. Deloitte report (2017) [35] to evaluate Industry 4.0 readiness. Survey method - Public institutions will be more effective in shaping society, the greatest impact on companies is to deliver the best possible product / service to customers, two most talked topics in the past year is to develop / create new products and services and improve productivity, technological initiatives are mostly in operations / processes. Berger report (2017) [7] to determine the advantages and disadvantages of Turkish food and beverage Industry in terms of Industry 4.0. Survey method - Although Turkey has the positive position in rankings, it is in the low country group in the readiness of Industry 4.0. - There are advisory steps as to how Turkey can ensure and improve readiness of Industry 4.0. Europarl report (2016) [36] to introduce related issues with Industry 4.0 such as its challenges and technologies. SWOT analysis - The supporting policies of the European Union for Industry 4.0; cyber security risks; and technological, social and business changes were discussed. - New policy proposals were presented. TUSIAD report (2016) [8] to analyse the opportunities emerging from Industry 4.0 and to demonstrate the potential of Turkish manufacturing Industry. Survey method - It was observed that some concrete steps have already taken for Industry 4.0 in Turkey. - It is sated that there is high-level of awareness and interest in the Industry to benefit from Industry 4.0 technologies and create competitive advantage in Turkey. Infosys report (2015) [37] to find out what the companies' plans for the technology journey are. Survey method - Although the majority of the companies are aware of the potential power of the Industry 4.0 technologies, only a few of them use these concepts in practice. - China is in a leading position compared to other countries. - The process Industry is the leader to the other industries. 60 Advances in Production Engineering & Management 15(1) 2020 Awareness and readiness of Industry 4.0: The case of Turkish manufacturing industry 4. Research method and sample description This study tries to find out the readiness of Turkish manufacturing companies to the new digital era by examining the implementation degrees of basic Industry 4.0 principles in their businesses. The sampling manufacturers were drawn from the database of "The Union of Chambers and Commodity Exchanges of Turkey". The data was collected through a survey conducted by phone interviews. The data includes 427 observations (respondent firms) from six different manufacturing sectors. Sampling universe is determined as 600 firms at the beginning of the study, and we were able to have response from about 70 % of these firms. The sample industries are chosen in accordance with the Industry 4.0 report of TUSIAD, (2016) [8]. The sub-industries included in the study are electrical and electronic products, machinery, food and beverage, chemical, automotive, and textile manufacturing industries. Table 2 summarizes the distribution of these firms by industries. The distribution of respondent firms by size is given in the Table 3. The sample includes micro, small, medium and large firms. The basic distinction between various size segments is based on SMEs (small and medium-sized enterprises) definition of EU (European Union) and Turkey. According to its definition, SMEs are examined mainly in three categories which are micro, small and medium-sized enterprises with personnel numbers "1-9", "10-49", "50-249" respectively [38, 39]. Since this study is aimed to investigate companies in all sizes, the segment of large firms having employees more than 250 is also included. In order to be able to make accurate comparisons, the composition of the firms in the sample universe was chosen homogenously. According to Table 4 which indicates the respondent's position in the company, the great majority of the firms (252 firms) are represented by a "production manager" with the ratio of 59.3 %. The titles following "production manager" are general manager (20 %), company owner (6.1 %) and R&D manager (4.2 %), respectively. SPSS 22.0 software package is used to analyse the data collected via survey. Table 2 Distribution of firms by sector Industries Number of firms Ratio Electrical and Electronic Products Manufacturing 71 16.7 General and Special Purpose Machinery 70 16.5 Food and Beverage Manufacturing 71 16.7 Chemical Manufacturing 75 17.6 Automotive Manufacturing 68 16.0 Textile / Clothing Manufacturing 70 16.5 Total 425 100.0 Table 3 Distribution of firms by size Size Number of employees Number of firms Ratio Micro 1-9 106 24.9 Small 10-49 107 25.2 Medium 50-249 107 25.2 Large 250 and more 105 24.7 Total 425 100.0 Table 4 Respondent's title in the firm Respondent's title Number of firms Ratio Production Manager 252 59.3 General Manager / Business Manager / General Coordinator 85 20.0 Owner / Partner 26 6.1 R & D Manager 18 4.2 Production Planning Manager 17 4.0 Quality Manager 11 2.6 Technology Manager 7 1.6 Factory Manager 5 1.2 Assistant General Manager 3 0.7 Other 1 0.2 Total 425 100.0 Advances in Production Engineering & Management 15(1) 2020 61 Sari, Güleg, Yigitol 5. Results and discussion The questions covered in the survey are aimed to understand the readiness of respondent firms for Industry 4.0 and the degree of implementation of basic Industry 4.0 concepts into their businesses. The first question of the survey is aimed to find out approaches of the firms towards the Industry 4.0 and it is adapted from the study of Basl [34]. Table 5 gives the respondents' answers. According to Table 5, although the great majority of the firms heard about the term Industry 4.0 with the ratio of 80.7 %, only 15.3 % (corresponds to 65 firms) of them are dealing with it for more than 1 year. By combining the new implementers, we can conclude that the total of 26.8 % firms are implementing the concepts of Industry 4.0. While 21.2 % of the companies admitted that they have not enough knowledge related to the concept of Industry 4.0, 32.7 % of them haven't considered its implementation due to various reasons. Considering the approaches of the firms to Industry 4.0 technologies by sub-sectors, it can be concluded that the automotive manufacturing industry is the leader in implementing Industry 4.0 principles as expected. Together with the new implementers and long term users, a total of 35.3 % of the firms operating in automotive manufacturing industry have adopted Industry 4.0 principles into their business operations. Indeed, the automotive industry is expected to yield better results compared to the other industries, because it experiences the best practices of Industry 4.0. Following automotive industry, electrical and electronic and machinery manufacturing industries are the second and third industries with the highest ratio of the firms implementing Industry 4.0 practices. While 20.8 % of chemical manufacturers have adopted Industry 4.0 technologies, the ratios tend to decrease in textile and clothing (14.2 %) and food and beverage (12.8 %) industries. Data reveals that the great majority of the firms in textile and clothing industry have not heard about the term Industry 4.0. The ratio in this industry is twice as much as the average of all covered industries. The food and beverage industry has the highest ratio (34.3 %) among the companies who have heard the term Industry 4.0, but have no idea what exactly it covers. It can be said that both of these sectors need to be well informed about this new industry revolution and its content. The summary of the distribution of the approaches to Industry 4.0 technologies by the size of the firms are given in the Table 6. According to Table 6, there is a positive and strong relationship between the size of the firms and their implementation of Industry 4.0 principles into their businesses. While more than half of the large firms have been implementing Industry 4.0 concepts, this ratio decreases as the firm size decreases. On the other hand, the number of companies who have not heard the topic Industry 4.0 increases as the firm size decreases. These results are in the line with the results of European Parliament Industry 4.0 Report suggesting that a large amount of investment is required by firms to catch the benefits of Industry 4.0 [32]. It is predicted that the necessary investment is about €40 billion annually until 2020 for Germany alone. In addition to capital and managerial needs, lack of qualified employees, and knowledge are the basic limitations that prevent companies from implementing Industry 4.0 technologies. Table 5 General approach to the topic Industry 4.0 Statements Number of firms Ratio We have been dealing with Industry 4.0 for more than 1 year. We are trying to implement the concepts of Industry 4.0 right now. We know the term Industry 4.0 but we haven't considered its implementation so far. We have heard the term Industry 4.0 but we have no idea what exactly it covers. We have not yet heard about the topic Industry 4.0 Total 65 49 139 90 82 425 15.3 11.5 32.7 21.2 19.3 100.0 62 Advances in Production Engineering & Management 15(1) 2020 Awareness and readiness of Industry 4.0: The case of Turkish manufacturing industry Table 6 Summary of firms' approach to Industry 4.0 technologies by firm size Statements Firm size Micro Small Medium Large We have been dealing with Industry 9.4% 14.9 % 28% 55.2 % 4.0 for more than 1 year, or we are trying to implement the concepts of Industry 4.0 right now. We have not yet heard about the 35.8% 23.4% 13.1% 4.8 % topic Industry 4.0. In the next question, the managers are asked in what extent they apply Industry 4.0 principles and in which organizational department (unit) in their firm. According to the answers of the respondents, it is concluded that, 7.5 % of all firms implement Industry 4.0 principles in "finance and accounting departments" in advanced level. "IT department" follows "finance and accounting" with the percentage of 6.8 %. Following finance and accounting and IT departments, "production" (6.3 %), "product design" (5.9 %), "research and development" (4.4 %), "repair and maintenance" (4.2 %), "marketing and sales" (4 %), "purchasing" (3 %) and logistics (3 %) departments are using I4.0 principles in advanced level. Table 7 lists the implementation degrees of the 9 basic technologies that constitute Industry 4.0. There are 9 technologies constituting Industry 4.0, which are autonomous robots, big data applications, cloud computing, cyber-security, simulation, additive manufacturing, system integration, internet of things and augmented reality [10]. The respondents are asked about the degree of their usage of each technology. They are given 1-7 Likert scale with the ratings from "1-never" to "7-advanced level". The resulting average scores are between 1.59 and 2.96 which are quite low, since the great majority of the firms responded as "never (1)". According to Table 7, the most used technology is cyber security with the mean of 2.96. It is followed by cloud technologies (2.22) and system integration (2.19), respectively. Having the mean of 1.59 augmented reality is the least used technology among these nine technologies. Although the means of implementation degrees are quite low in general, there are a numerous firms that use these technologies in an advanced level. Fig. 1 indicates the total number of the advanced implementers of each Industry 4.0 technology and their distribution by sectors. Fig. 1 shows that cyber security is the most popular technology among all industries. While autonomous robots are widely used by electrical and electronic products manufacturers, system integration is mostly implemented by automotive manufacturers. The usage of big data and cloud technologies are relatively rare between the advanced users in all industries. We also used the ANOVA statistical method to test the differences between all types of sectors in implementation degrees of Industry 4.0 technologies. The probability value is below 0.05 in nine Industry 4.0 technologies. This result shows that there is statistically significant differences in the means of implementation degrees of Industry 4.0 technologies among the different types of industries. The respondent firms are also asked in what extend the key concepts of Industry 4.0 are important to them (on a scale of 1 to 7). The results point out that, all of the concepts are important for the managers. As one can see in Table 8, the means of importance degrees are between 6.25 and 5.23, which are very close to the highest rate 7. According to the respondent firms, cyber security is the most important technology followed by autonomous robots and big data, respectively. Among these nine I4.0 technologies, augmented reality is found to be the least common technology, since it has the lowest implementation degree with the mean of 1.59 and the lowest importance degree with the mean of 5.23. Hence, we can conclude that, augmented reality will be the most difficult I4.0 technology to introduce in the Turkish manufacturing sector. Table 9 summarizes the correlations between implementation and importance degrees of Industry 4.0 technologies. Having a correlation coefficient r = 0.822, proves a positive and significant relationship between the implementation degrees of Industry 4.0 technologies and their importance to the firms. This result indicates that the manufacturing companies try to use the Industry 4.0 concepts which they find the most important. Advances in Production Engineering & Management 15(1) 2020 63 Sari, Güleg, Yigitol Table 7 Implementation degrees of Industry 4.0 technologies I 4.0 technologies Number of Mean Standard Number of firms deviation advanced users Cyber Security 427 2.96 2.01 60 Cloud Technologies 425 2.22 1.63 19 System Integration 425 2.19 1.69 30 Internet of Things 425 2.06 1.58 19 Simulation Technologies 425 2.02 1.66 25 Big Data 425 1.94 1.55 18 Autonomous Robots 427 1.92 1.64 25 Additive Manufacturing 425 1.72 1.44 22 Augmented Reality 425 1.59 1.40 21 Cyber Security System Integration Simulation Technologies 16 14 12 10 8 6 4 2 0 I . ■ i h i ■ I---I—KM-N/i-\ \—i/v\i—m-1—i/N—fT^—v Fig. 4 Scopes of workstations for each AGV under different number of AGVs 3.3 Many-to-many The last strategy is that many vehicles fulfill the transport request of many workstations within an area, and is denoted as many-to-many. Under this strategy, the front workshop is divided into two areas. The first area contains the workstations from AS00 to AS05, and the second one contains the workstations from AS06 to AS10. The first and second area is equipped with AGV1 and AGV3, respectively. And products in the rear workshop is still transported by the AGV2. For the first area, similar to the one-to-many strategy, six skip cars are used to carry the four parts of a product, and the skip cars are transported by an AGV. For the two areas of the front factory, the many-to-many strategy is used. Under this strategy, a certain area, which contains many workstations, is equipped with many AGV vehicles. When a vehicle finished its current transportation, it becomes available and can be used by all other workstations within this area. When a workstation a finishes its operation, and its next workstation is idle, the workstation a generates a transport request and writes the request to the Distribution Table. Then, a method is accessed every 5 s to allocate the requests of the Distribution Table to a vehicle selected from the Available Vehicles. The Available Vehicles is the vehicle set of available vehicles within an area. If both the number of requests and the Available Vehicles are greater than 0, the first transport request is removed from the Distribution Table and assigned to an available vehicle. The vehicle is selected by the shortest distance principle, which first selects the vehicle closest to the workstation. Then the vehicle moves to its destination for loading or unloading. The pseudocode for allocating is shown in Fig. 5. Procedure 3 Allocate requests to vehicles m=sizeof(Distribution Tabic) n=sizeof(Available Vehicles) while m >0 and n >0 do Remove the first transport request from the Distribution Table Remove a vchiclc from the Available Vehicles by the shortest distance principle Get the Target Workstation and Target Track of the request Move the vehicle to the Target Track by the shortest path principle Record the current time t\ Update the Distribution Tabic and Available Vehicles m=sizeof(Distributioii Table) n=sizeof(Available Vehicles) end while Fig. 5 The procedure for allocating Advances in Production Engineering & Management 15(1) 2020 73 Yang, Xu, Li, Wang A sensor for loading and unloading is set beside every workstation. When a vehicle passed by the sensor, the sensor is triggered. If the current workstation is the Target Workstation, the vehicle stops. Furthermore, if the vehicle is occupied, the vehicle unloads its product to the Target Workstation. Then, the vehicle becomes available and waits for a new transport request. If an empty vehicle reaches to its Target Workstation, the vehicle first loads the product from the Target Workstation and then moves to its destination by the shortest path principle. The pseudocode for loading and unloading is shown in Fig. 6. In summary, three assembly transport strategies, which are one-to-one, one-to-many, and many-to-many, are proposed for the front workshop. Moreover, for the rear workshop, only the many-to-many strategy is used. Procedure 4 Load or unload if the current workstation is the Target Workstation then Stop the vehicle if the vehicle is occupied then Unload the product from the vehicle to the target workstation Delete the destination of the vehicle Record the current time, ¿2 Calculate the transportation time t, t — Î2 — t\ Append the vehicle to the Available Vehicles Waituntil a new request is assigned to the vehicle else Load the product from the Target Workstation to the vehicle Get the Target Workstation and Target Track of the product Move the vchicle to the Target Track by the shortest path principle end if end if Fig. 6 The procedure for loading and unloading 4. Modelling and optimization of the proposed transport strategies This section establishes a production simulation model for the RFS. The three assembly transport strategies are implemented and verified in the model. Besides, production process and vehicle configuration of each strategy are optimized. 4.1 Establishment of the simulation model The Plant Simulation software is used to establish the simulation model of the RFS. The simulation model mainly contains the ASRS, workstations, vehicles, scheduling strategies, experiment module, and statistical analysis. For more modelling information about the workshop logistics processes, please refer to [26]. Finally, the simulation model of the RFS is shown in Fig. 7. The production is operated 7 hours a day and lasts for 10 days. The throughput, facility utilization, and vehicle quantity of those three transport strategies are evaluated and optimized. 74 Advances in Production Engineering & Management 15(1) 2020 Assembly transport optimization for a reconfigurable flow shop based on a discrete event simulation 4.2 Optimization of the one-to-one strategy Recall that the front workshop contains 10 workshops and an ASRS, and the rear workshop contains four workstations. Firstly, sufficient vehicles (11 AGV1 and 5 AGV2) are provided for the two workshops to evaluate production performance. After 10 days of production, the throughput is 390, and the facility utilization is shown in Fig. 8A. Fig. 8A shows that the overall facility utilization is relatively low, and many workstations are blocked. The blockage appeared in AS10 and its front workstations. This is because the finished product in AS10 is to be transported away by the vehicles of the rear workshop. The rear workshop uses the many-to-many transport strategy, which may wait for a long time before a vehicle is available. Moreover, there is a long distance between the AS10 and the AS11. Thus, the finished product in the AS10 may wait for a long time before transported away, which results in a blockage in AS10. When AS10 is blocked, the workstations in front of it will also be blocked. To lessen the blockage, we set buffers after AS10 and before AS11, namely BF10 and BF11, respectively. Buffer sizes are optimized by the design of experiment (DOE) method. The buffer size of BF10 and BF11 are increased from 0 to 3 and from 0 to 5, respectively. A total of 24 experiments are carried out, and the output is shown in Fig. 9A. Fig. 9A shows that when buffers are set, the throughput increases significantly. The maximum output is 566, which is 45.1 % higher than the output before the buffer optimization. Among the experiments reaching the maximum output, the 10 th experiment uses the least buffer size. In this experiment, the buffer size for BF10 and BF11 are 1 and 3, and this buffer configuration is chosen as the best one. The facility utilization under this buffer configuration is shown in Fig. 8B. It shows that after the buffer optimization, facility utilization of all workstations increases obviously, and blockages reduce significantly. However, workstations AS04 and AS11 still have some blockages. This is because the next workstation of AS04 and AS11 have more operation time, and no extra space exists besides the two workstations. I Working I Setting-up ] Waiting ] Blocked I PoweringllpDown I Failed ] Stopped I Paused I Unplanned AS03 AS05 AS07 AS09 AS11 AS13 AS02 AS04 AS06 AS08 AS 10 AS 12 Station 70-; § 60 'H 504 o G 40 O) 2 30-r ID C/} Possibil-yt ity of 3 detection b a b o ID Q c (0 CL* C£ Advances in Production Engineering & Management 15(1) 2020 97 Vulanovic, Delic, Kamberovic , Beker, Lalic Table 3 Scale for the assessment of the severity of consequences "I Effect c£ Severity of consequences Consequences of quality degradation (ISO 9001) Consequences of environment degradation (ISO 14001) Consequences of OH&S degradation (ISO 18001) 10 Failures may lead to the loss of important operational functions for a long period of time. Lawsuits by cus-Extremely tomers and serious sanc-high tions from the government are inevitable. Possible cessation of the company. Permanent degradation of the environment on a large area with contamination that can be expanded by emissions into the air, water or land. The level of contamination directly threatens human health. Lawsuits and sanctions from the government are inevitable. Possible cessation of the company._ Failures may lead to emergency situations with multiple fatalities among employees and other stakeholders. Lawsuits from stakeholders and serious sanctions from the government are inevitable. Possible cessation of the company. Critically high Failures may lead to the loss of operational functions for a short period of time. Lawsuits by customers and bad publicity are inevitable. Financial losses are sometimes irrecoverable. Permanent degradation of the environment on a small area with contamination that can be expanded by emissions into the air, water or land. Human health is indirectly affected. Sanctions from the government and bad publicity. Financial losses can be irrecoverable. Failures may lead to fatalities among employees. Lawsuits from employees, sanctions from the government and bad publicity are inevitable. Financial losses are sometimes irrecoverable. Very high Failure leads to the suspension of supply / provision of contracted services. In addition to direct financial losses, permanent loss of customers and the market position is expected. Local degradation of the environment with no possibility of expanding pollution. Sanctions from the government and bad publicity of the company are expected. Significant violation of employees' health, which cannot be compensated (permanent inability to perform work-related activities). A lawsuit by employees and bad publicity are expected. 7 High Failure leads to a brief interruption of operational functions and delays in agreed deadlines. Expected financial losses to legitimate customer complaints. Constant degradation of the environment over a long period of time (over one year). The consequences are not fully recoverable. Permanent violation of health and work ability of employees due to serious injuries or chronic illnesses (30% disability) Moderately high Failure leads to a number of inconsistencies in the process of work. Nonconforming products / services are delivered to end-users, which leads to their dissatisfaction. The consequences for the environment are evident, but not fatal for wildlife. There is no possibility of spreading negative impact, but the consequences may not be fully recoverable. Injuries / ill-health that temporarily violate work ability, but full recovery is possible (sick leave up to 3 months) 5 Medium Systematic deviations lead to hidden defects in the product / services, leading to claims under warranty and dissatisfaction of customers. It is necessary to make an effort to overhaul consequences for the environment after the completion of work activities. Injuries / ill-health that temporarily violate work ability, but full recovery is possible (sick leave up to 1 month) 4 Moderate Deviations can lead to small inconsistencies in the process. The occurrence leads to dissatisfaction of customers and financial losses. The effects of pollution are present over a longer period of time after the cessation of activities. The environment is able to regenerate inone year period Injury / ill-health that requires a simple and short-term medical intervention (loss of up to 1 week). A decline in the performance of the process, which causes 3 Low the appearance of a small number of nonconformities. End users are not affected. The effects of local pollution are present in a short period after the cessation of activities. The environment is able to regener- Injury / ill-health that requires first aid by internally trained persons (loss to one business day) 2 Very low Decline in performance processes causing loss of time. Adverse effects are very limited. Environmental impact exists only as long as the current work activities. Slight degradation of health. After little intervention of the injured person, operation continues unimpeded. 1 Insignificant Deviations have no effect. Deviations have no effect. Deviations have no effect. 9 8 6 98 Advances in Production Engineering & Management 15(1) 2020 Integrated management systems based on risk assessment: Methodology development and case studies The severity of consequences Assigning the values for each assessed factor can be one of the greatest problems in implementation of any risk assessment method including FMEA. Converting qualitative into quantitative values is always fraught with the subjective attitude of risk assessors. Therefore, it is recommended to establish specific guidelines for the evaluation of factors as precisely as possible. The scale for evaluation of the severity of consequences for three groups of risks linked to the ISO 9001, ISO 14001 and OHSAS 18001 is proposed in Table 3, but every organization can define its own scale tailored to its needs. The probability of occurrence It is much easier to quantify probability of occurrence, especially if empirical data are available. The scale for the probability assessment is given in Table 4. Table 4 Scale for the probability assessment Rank Probability of occurrence Likelihood of failure occurrence, or frequency of its repetition 10 Extremely high More than 1 occurrence per day / more than 3 occurrences in 10 events. 9 Critically high 1 occurrence on every 3 to 4 days / 1 occurrence in 10 events. 8 Very high 1 occurrence per a week / 5 occurrences in 100 events. 7 High 1 occurrence per a month / 1 occurrence in 100 events. 6 Moderately high 1 occurrence per 3 months / 3 occurrences in 1.000 events. 5 Medium 1 occurrence per 6 to 12 months / 5 occurrences in 10.000 events. 4 Moderate 1 occurrence per year / 6 occurrences in 100.000 events. 3 Low 1 occurrence per 3 years / 6 occurrences in 1.000.000 events. 2 Very low 1 occurrence per 3 to 5 years / 2 occurrences in 10.000.000 events. 1 Insignificant 1 occurrence in more than 5 years / less than 2 occurrences in 100.000.000 events. The probability of detection Since the Model should be applicable in organizations of all types and sizes, it is almost impossible to define a scale from 1 to 10 that would be applicable for every possible process. The possibility of detection is often limited by the existence of measuring equipment in the company and by the nature of the process itself. Due to these limitations and the fact that the probability and consequence are prevalent in determining any risk in most risk assessment methods [21], the matrix for detectability assessment was simplified. According to Sankar & Prabhu [25], rankings of detectability can vary even for the same type of deviation (risks), so the authors of this paper created a scale in the range of 1 to 3, which is shown in Table 5. After defining the scales for all criteria required for conducting the FMEA method, it is clear that potential results of assessed risks (risk priority numbers - RPNs) are in a range from 1 to 300. Table 5 Scale for the detectability assessment Rank Probability °f Description of detectability _detection_____ 3 Low The process is difficult to control, and the effect of the failure is very difficult to detect. 2 Medium Process perpetrators perform a visual process control / periodic measurements of numerical values / counting attribute values. 1 High The failure is detected and controlled automatically or semi-automatically in a way that _prevents the occurrence of deviations._ 3.5 Influence of risks on IMS documentation A common hierarchical structure of IMS documents is shown in Fig. 2. The number and scope of IMS documents that should exist in an organization and describe identified processes (according to the process approach) have not been defined by any normative references. The authors of this paper consider that the ranks of risks in the existing processes are directly aligned with the type and number of needed documentation. The exact relations between assessed risks and IMS documents are described in the further text. Advances in Production Engineering & Management 15(1) 2020 99 Vulanovic, Delic, Kamberovic , Beker, Lalic LEVELO IMS Policy and objectives / \ Documents that define level 2 / procedures \ realization of processes / \ Documents that define LEVEL 3 / INSTRUCTIONS \ realization of activities / \ Documents that provide level 4 / records \ evidence of performed activities Fig. 2 Hierarchical structure of IMS documents Level 0 - The Policy and objectives The IMS Policy and objectives, along with the mission and vision, represent the highest level of documents in a company that are certainly affected by strategic risks. Each management standard requires periodic management review of the system that should result in the adaptation of the policy and objectives of an organization in accordance with existing conditions. In this way, through the institution of management review, changes in operational risks indirectly reflect the changes in the policy and objectives of an organization, but the policy and objectives are not directly treated by the Model. Level 1 - The Manual Specification PAS 99 recommends the development of a single Manual that describes the overall management system [11]. This IMS Manual refers to the procedures and instructions of an integrated management system. Usually there is no need to describe all the processes by separate procedures because they can be very numerous and heavily burden the administration of a system. Therefore it is sufficient to describe some processes with minor risks (rank 1) just in the Manual. Level 2 - Procedures Some management systems insist on certain mandatory procedures but the system should be also documented with procedures that describe primary processes of the organization. The designed Model proposes that moderate risks (rank 2) must be described by procedures. When documenting the observed process by a procedure, particular attention should be focused on those activities where the moderate risks were identified. In this way, documentation should be adapted to the risks of the organization which minimizes the possibility of deviations and errors in processes. Level 3 - Instructions The Model suggests that all activities with high risks (rank 3) should be described by separate instructions. The appearance of risks in rank 3 certainly implies the development of a procedure for the entire process in which risky activity occurs, but a separate instruction that will accurately describe the observed activity is also needed. In that way, all high-risk activities would be separately described in detail, which would enable workers to adequately perform their tasks in safe manner. Level 4 - Records This level of documentation is actually dependent on the above-described documents of a system (especially procedures and instructions) which define the number and format of records that need to be kept. The designed Model does not foresee the changes that are directly related to the records, but their number indirectly depends on the level of estimated risks, since risks affect the quality and quantity of procedures and instructions in IMS. 100 Advances in Production Engineering & Management 15(1) 2020 Integrated management systems based on risk assessment: Methodology development and case studies Added Level 5 - Action sheets Besides customary IMS documents, the Model introduces another level of documentation that is related to the most severe risks that can lead to disastrous consequences. Emergency situations are specially treated by ISO 14001, OHSAS 18001, ISO 22000, etc. (requiring preparedness for response in case of emergencies). The Model therefore requires the existence of special instructions called "Action sheets" which define preventive and recovery measures in case of emergency situations for all severe risks (rank 4). Action sheets should be posted in a visible location near a potential emergency spot which should enable everyone to react properly in case of emergency. 4. Results and discussion 4.1 Risk based model for the design of IMS documentation Materials and methods used in previous chapter resulted with the universal risk based Model for IMS documentation design, which is a main result of this paper. To avoid arbitrariness in defining the scope and structure of IMS documents, the proposed Model introduces precise criteria that can be used to determine the number and types of documents required, depending on the rank of assessed risk. The number of repetitions of all possible RPN results obtained by modified FMEA method is shown in the histogram in Fig. 3, which shows that the distribution of RPN values is far denser at the beginning of the scale where RPN values are lower. The occurrences of high-risk deviations in practice are extremely rare, and low-risk deviations are far more common, which is in accordance with the distribution of RPN values shown in Fig. 4. The density of distribution of RPN values was taken into account when risk rankings were defined. Since there are four levels of IMS documents that are treated by the proposed Model, the authors predicted four risk rankings within a range from 1 to 300 RPN values: • Minor risks (Rank 1) - risks are in the interval 1 < RPN < 50. • Moderate risks (Rank 2) - risks are in the interval 50 < RPN < 100. • High risks (Rank 3) - risks are in the interval 100 < RPN < 150. • Severe risks (Rank 4) - risks are in the interval 150 < RPN < 300. The description of IMS documents which directly depend on the rank of estimated risks is shown in Table 6. Risk rankings in practice can be subject to change. Each organization could define its own scale for risks according to the objectives and available resources, so rankings should be periodically reviewed and adapted to the needs of the organization. 12 r 10 RPN values Fig. 3 The distribution of RPN values Advances in Production Engineering & Management 15(1) 2020 101 Vulanovic, Delic, Kamberovic , Beker, Lalic Table 6 IMS documentation depending on the rank of the estimated risks RPN of analyzed RANK ^ . c . , IllJIf, , c . , Description of required IMS documents process/activity of risk_____ RPN < 50 Minor risks Analyzed process should be described in the IMS Manual (RANK 1) 50 < RPN < 100 Moderate risks Analyzed process should be described by a Procedure and briefly described (RANK 2) in the IMS Manual 100 < RPN < 150 High risks Analyzed activity should be described by a separate instruction (RANK 3) 150 < RPN < 300 Severe risks Analyzed activity should be described by a separate instruction and by an _(RANK 4)_action sheet for the prevention and treatment of possible emergency_ It may be noted that, irrespective of the assessed risks, all processes should be described (at least roughly) in the Manual of the integrated management system, which is required by ISO/TR 10013:2001 [26]. The procedure is required only if the analyzed process has risks in rank 2. Risks in ranks 3 and 4 indicate that separate instructions for analyzed activities should be made. When it comes to risks in rank 4, at a first sight it could be stated that there is an overlap in the documentation for the same activities. However, the duplication of documentation does not actually exist because action sheets define just activities for the prevention and remediation of potential emergencies, but the way of carrying out the observed activities and corresponding responsibilities is defined by appropriate instructions. In addition, action sheets are designed for all present persons, regardless of their familiarity with the company processes. Unlike action sheets, the instructions are designed directly for the workers. 4.2 Case studies The risk based Model for the design of IMS documentation has been applied in 3 various companies in Serbia: • Company A is engaged in designing buildings and facilities and has implemented integrated management system which includes ISO 9001, ISO 14001 and OHSAS 18001. • Company B is engaged in civil engineering, transport, asphalt production and exploitation of mineral materials. It has implemented integrated management system which includes ISO 9001, ISO 14001 and OHSAS 18001. • Company C has implemented integrated management system which includes ISO 9001, ISO 14001, OHSAS 18001 and ISO 22000. It has three branches: - The first branch is engaged in the design, construction and maintenance of all types of gas, electrical and ventilation installations. - The second branch is engaged in the production and storage of fresh and frozen goods (food, fruit and vegetables). - The third branch is engaged in the production and packaging of brandies from fruits and grapes. All three companies already had their integrated management system implemented and the comparison between the old and new IMS documentation was made. In order to facilitate the comparison and analysis of the results obtained, the following restrictions were made: • Only the processes that have previously been identified and described in companies were analyzed and their risks were assessed (this was the only way for objective comparison between the old and new IMS documentation). • System processes that are required to be documented in accordance with the implemented standards (such as: document control, record control, internal audit, corrective actions, etc.) were not considered by the Model. • The categories of risk taken into account were in direct correlation to the standards applied in the organization. Although the key performance indicators are often expressed through financial results of the company [27], the Model does not predict financial indicators for several reasons: 102 Advances in Production Engineering & Management 15(1) 2020 Integrated management systems based on risk assessment: Methodology development and case studies • The financial success of the company is not the subject of any management standard. • Profit is just a consequence of consistent quality of the processes, products or services, which wins long-term customer satisfaction. Therefore, any deviations in the processes of the company influence its profit directly or indirectly. After drawing flow charts for each analyzed process, the authors created corresponding FMEA tables, which resulted with a resume of IMS documents required by the proposed Model. Comparative analysis between the documentation that existed in the companies A, B, and C in the past, and the documentation designed by new research model, showed following results: • 9 specific processes were identified in the integrated management system of Company A. The primary system in the Company A consisted of 10 documents (6 Procedures + 4 Instructions) and the documentation designed by new model consisted of 8 documents (7 Procedures + 1 Instruction). It is obvious that number of IMS documents in Company A does not vary much depending on the approach used, but there are major differences regarding the type of designed documentation. This dispersion can be explained by the inconsistent application of the process approach when the documentation was primarily designed. Generally, Company A does not face great risks, especially when it comes to endangering health and safety at work or the environment, since the company is engaged only in the design and supervision, and not in the construction works. Therefore, the number of documents designed for this company in accordance with the proposed model is very small and covers mainly the quality aspect in accordance with ISO 9001. • 12 specific processes were identified in the integrated management system of Company B. The primary system in the Company B consisted of 35 documents (10 Procedures + 13 Instructions + 12 action sheets) and the documentation designed by new model consisted of 48 documents (16 Procedures + 25 Instructions + 8 action sheets). There is a big difference in the size and type of business between Company A and Company B, so the IMS documentation in those two companies differs a lot, although it covers the same set of standards (ISO 9001, ISO 14001 and OHSAS 18001). It can be seen that total number of required documents designed by a risk-based model is higher by 13 than the documentation that was primarily designed. The analysis of the processes shows that unlike company A, Company B faces significant risks, especially when it comes to environment and occupational health and safety. That is why the designed documentation is "focused" on the most risky processes such as: construction, production of asphalt mass and machinery manipulation, to which most instructions and all action sheets are directed. Such a distribution of documentation was predictable, since the model was developed to be consistent with the risks of the company concerning the applied standards. • 24 specific processes were identified in the integrated management system of Company C. The primary system in the Company C consisted of 34 documents (23 Procedures + 11 Instructions) and the documentation designed by new model consisted of 55 documents (25 Procedures + 26 Instructions + 4 action sheets). Company C has a much more complex structure and wider scope of activities than other companies analyzed in the case study. In addition, this company has the most complex management system, which comprises four standards (ISO 9001, ISO 14001, OHSAS 18001 and ISO 22000). The processes occurring in the branch which deals with the design, construction and maintenance of gas and ther-mo-technical installations can in some ways be compared with the processes already analyzed in companies A and B. However, other two branches have specific processes that are also subject to requirements ISO 22000, which generates a new group of risks and increases the number of required documentation. According to the set goal, the designed documentation is focused on the most risky processes such as: construction, machinery manipulation, production of frozen products, service storage of frozen products and production of brandy. • Case studies generally showed that primary implementation of IMS was not carried out systematically and according to the process approach which is one of the core principles of quality management. This resulted with non-consistent number and types of IMS docu- Advances in Production Engineering & Management 15(1) 2020 103 Vulanovic, Delic, Kamberovic , Beker, Lalic ments. For instance, that is why number of documents in the Company B was even slightly larger than in the Company C which is far more complex and even has additional standard implemented. Such omissions cannot happen when applying the proposed Model. Besides mentioned, previous documents were not aligned with existing risks in processes and by that they don't meet the needs of the organization. The practical application of the Model suggests that in some cases the same processes carry different risks in different companies. Risks vary according to an impact that the analyzed process has on the performance of the company and it should be treated and documented according to it. Therefore, there are no "a priori risky processes". 5. Conclusion The paper presents all phases of the development and implementation of the Model for designing IMS documentation based on risk assessment in organizations. In order to create this Model, the authors: • Adapted flow chart, as a universal tool suitable for graphically displaying and analyzing essential process elements, • Selected and adapted the FMEA method, as a universal tool suitable for risk assessment in the processes of the company, related to the applied management standards, • Established a universal matrix for ranking different types of risk, • Defined the relationship between the assessed risks and required IMS documentation. For the purpose of practical verification of the designed Model, the following actions were performed: • Implementation of the Model in three diverse companies that already had an integrated management system with at least three management standards implemented, • Comparative analysis of the documentation obtained from a risk based Model and previous IMS documentation that existed in organizations. Case studies showed that the Model can be applied in the organizations of all sizes and types in a way that the number and scope of IMS documents directly depend on the risks in the existing processes. That documentation should always be changed along with the changes of the company's risks. The risk based model for the design of IMS documentation has given precise guidelines that each company can use in creating the optimal level of IMS documentation. However, every organization can change the projected Model by adapting the scales for risk ranks in accordance with its goals and needs. Organizations that thrive in the field of risk management often desire to bring it to a higher level by reducing risk appetite. The authors have also identified some weaknesses of the proposed approach, such as: • Implementation of the Model depends on expertise and evaluation of FMEA team, • Drawing the flow charts and FMEA matrix for every process can be time consuming, • Current Model does not include financial, reputational, compliance, or similar risks, • Current Model does not define operational actions for the reduction of assessed risks. In order to correct recognized weaknesses, the authors anticipate following directions for further research: • Developing adequate software would facilitate the implementation of the Model. Creating a database with all potential deviations, their consequences and associated risks, would help FMEA team to get the job done, • Creating new types of risks related to the financial effects should broaden the use of proposed Model, • Extending the Model with actions for the reduction of assessed risks should add operational value to the Model implementation. 104 Advances in Production Engineering & Management 15(1) 2020 Integrated management systems based on risk assessment: Methodology development and case studies Acknowledgement Special thanks are addressed to the associates of Research and Technology Center Ltd, Novi Sad, Serbia, especially for carrying out the practical part of the research concerning activities in the case studies. References [1] Beckmerhagen, I.A., Berg, H.P., Karapetrovic, S.V., Willborn, W.O. (2003). Integration of management systems: Focus on safety in the nuclear industry, International Journal of Quality & Reliability Management, Vol. 20, No. 2, 210-228, doi: 10.1108/02656710310456626. [2] Rebelo, M., Santos, G., Silva, R. (2013). Conception of a flexible integrator and lean model for integrated management systems, Total Quality Management & Business Excellence, Vol. 25, No. 5-6, 683-701, doi: 10.1080/ 14783363.2013.835616. [3] Bekcic, S., Kelecevic, N., Marinkovic, V., Tasic, L., Krajnovic, D. (2013). Approach to the integration of management systems in a pharmaceutical organization, Indian Journal of Pharmaceutical Education and Research, Vol. 47, Vol. 3, 19-25. [4] Domingues, J.P.T., Sampaio, P., Arezes, P.M. (2015). Analysis of integrated management systems from various perspectives, Total Quality Management & Business Excellence, Vol 26, No. 11-12, 1311-1334, doi: 10.1080/ 14783363.2014.931064. [5] International Organization for Standardization, ISO 31000: 2009 Risk management - Guidelines on principles and implementation of risk management, ISO Copyright Office, Geneva. [6] Pojasek, R.B. (2006). Is your integrated management system really integrated?, Environmental Quality Management, Vol. 16, No. 2, 89-97, doi: 10.1002/tqem.20124. [7] Wilkinson, G., Dale, B.G. (2006). Integrated management systems: A model based on a total quality approach, Managing Service Quality: An International Journal, Vol. 11, No. 5, 318-330, doi: 10.1108/09604520110404040. [8] Zeng, S.X., Shi, J.J., Lou, G.X. (2007). A synergetic model for implementing an integrated management system: An empirical study in China, Journal of Cleaner Production, Vol. 15, No. 18, 1760-1767, doi: 10.1016/j.jclepro. 2006.03.007. [9] Zwetsloot, G.I.J.M. (1995). Improving cleaner production by integration into the management of quality, environment and working conditions, Journal of Cleaner Production, Vol. 3, No. 1-2, 61-66, doi: 10.1016/0959-6526(95)00046-H. [10] J0rgensen, T.H., Remmen, A., Mellado, M.D. (2006). Integrated management systems - Three different levels of integration, Journal of Cleaner Production, Vol. 14, No, 8, 713-722, doi: 10.1016/j.jclepro.2005.04.005. [11] British Standard Institute (BSI), (2012). PAS 99 Specification of common management systems requirements as a framework for integration, BSI Copyright Office, London. [12] Karapetrovic, S., Willborn, W. (1998). Integration of quality and environmental management systems, The TQM Magazine, Vol. 10, No. 3, 204-213, doi: 10.1108/09544789810214800. [13] Karapetrovic, S. (2002). Strategies for the integration of management systems and standards, The TQM Magazine, Vol. 14, No, 1, 61-67, doi: 10.1108/09544780210414254. [14] Labodová, A. (2004). Implementing integrated management systems using a risk analysis based approach, Journal of Cleaner Production, Vol. 12, No. 6, 571-580, doi: 10.1016/j.jclepro.2003.08.008. [15] Salomone, R. (2008). Integrated management systems: Experiences in Italian organizations, Journal of Cleaner Production, Vol. 16, No. 16, 1786-1806, doi: 10.1016/j.jclepro.2007.12.003. [16] Asif, M., Fisscher, O.A.M., De Bruijn, E.J., Pagell, M. (2010). An examination of strategies employed for the integration of management systems, The TQM Journal, Vol. 22, No. 6, 648-669, doi: 10.1108/1754273101108 5320. [17] Bernardo, M., Casadesus, M., Karapetrovic, S., Heras, I. (2009). How integrated are environmental, quality and other standardized management systems? An empirical study, Journal of Cleaner Production, Vol. 17, No. 8, 742750, doi: 10.1016/j.jclepro.2008.11.003. [18] Bernardo, M., Casadesus, M., Karapetrovic, S., Heras, I. (2012). Do integration difficulties influence management system integration levels?, Journal of Cleaner Production, Vol. 21, No. 1, 23-33, doi: 10.1016/j.jclepro.2011. 09.008. [19] Asif, M., De Bruijn, E.J., Fisscher, O.A.M., Searcy, C. (2010). Meta-management of integration of management systems. The TQM Journal, Vol. 22, No. 6, 570-582, doi: 10.1108/17542731011085285. [20] Mezinska, I., Lapina, I., Mazais, J. (2015). Integrated management systems towards sustainable and socially responsible organisation, Vol. 26, No. 5-6, 469-481, doi: 10.1080/14783363.2013.835899. [21] International Organization for Standardization (ISO), (2009). ISO 31010:2009 - Risk management - Risk assessment techniques, ISO Copyright Office, Geneva. [22] McDermott, R.E., Mikulak, R.J., Beauregard, M.R. (2008). The basics of FMEA, Second edition, CRC Press, Taylor & Francis Group, New York, USA. [23] Vulanovic, S., Beker, I., Radlovacki, V., Delic, M. (2012). The appliance of work flow diagram as a tool for identification and grouping of failures in processes of integrated management system, International Journal Advanced Quality, Vol. 40, No. 1, 23-26. Advances in Production Engineering & Management 15(1) 2020 105 Vulanovic, Delic, Kamberovic , Beker, Lalic [24] Banduka, N., Veža, I., Bilic, B. (2016). An integrated lean approach to process failure mode and effect analysis (PFMEA): A case study from automotive industry, Advances in Production Engineering & Management, Vol. 11, No. 4, 355-365, doi: 10.14743/apem2016.4.233. [25] Sankar, N.R., Prabhu, B.S. (2001). Modified approach for prioritization of failures in a system failure mode and effects analysis, International Journal of Quality & Reliability Management, Vol. 18, No. 3, 324-336, doi: 10.1108/ 02656710110383737. [26] International Organization for Standardization (ISO), (2001). ISO/TR 10013:2001 Guidelines for quality management system documentation, ISO Copyright Office, Geneva. [27] Delic, M., Radlovački, V., Kamberovic, B., Maksimovic, R., Pečujlija, M. (2013). Examining relationships between quality management and organisational performance in transitional economies, Total Quality Management & Business Excellence, Vol. 25, No. 3-4, 367-382, doi: 10.1080/14783363.2013.799331. Appendix A The list of the abbreviations in the paper: FMEA Failure mode effect analysis IMS Integrated management system OH&S Occupational health and safety PDCA Plan-Do-Check-Act RPN Risk priority number 106 Advances in Production Engineering & Management 15(1) 2020 Advances in Production Engineering & Management Volume 15 | Number 1 | March 2020 | pp 107-117 https://doi.Org/10.14743/apem2020.1.353 ISSN 1854-6250 Journal home: apem-journal.org Original scientific paper Communication and validation of metrological smart data in IoT-networks Acko, B.a*, Weber, H.b, Hutzschenreuter, D.b, Smith, I.c aUniversity of Maribor, Faculty of Mechanical Engineering, Maribor, Slovenia bPhysikalisch-Technische Bundesanstalt, Braunschweig, Germany cNational Physical Laboratory, Teddington, United Kingdom A B S T R A C T A R T I C L E I N F O An Internet of Things-network (IoT- network) allows for the communication of data both within the network and to data hubs. However, the usefulness of the data depends on its ability to be interpreted correctly. For metrology data, effective use of the data is only possible if the numerical value, associated unit and uncertainty, expressed in a standard format, are also available. In order to develop, provide and distribute a formal framework for the transmission of metrology data on the basis of the International System of Units, European project EMPIR 17IND02 SmartCom was agreed between the European Commission and the European Association of National Metrology Institutes (Euramet). The SmartCom project aims to provide the methodological and technical foundation for the unambiguous, universal, safe and uniform communication of metrological smart data in the IoT and Industry 4.0. The project will increase the industrial capabilities and the provision of regulations for data exchange in the IoT. It will also assist countries within the European Union (EU) and those with an association agreement with the EU in developing products that are able to communicate in IoT environments worldwide. In addition to describing the general ideas and aims of the project, this article presents the research results achieved in the first midterm period. © 2020 CPE, University of Maribor. All rights reserved. Keywords: Metrology; Measurement metadata; Information and communication technology (ICT); Smart Data; Data communication; IoT-communication; IoT-networking; Digital calibration certificate *Corresponding author: bojan.acko@um.si (Acko, B) Article history: Received 27 December 2019 Revised 26 March 2020 Accepted 28 March 2020 1. Introduction The current paradigm for industrial and engineering quality assurance is based on production processes that have short- to medium-term stability along with a significant investment in quality inspection. One of the goals of Industry 4.0 and data-centric engineering [1-4] is to use multiple sensors that measure all aspects of the production process. The resulting complete set of measured data can then be used to understand, in much greater detail, the performance of the system so that the process can be kept close to its ideal state, leading to reduced downtimes, fewer rejected parts, improvements in quality, better organised maintenance, better conservation of energy and resources, and increased business success. However, data-centric engineering depends on being able to assess the quality of the measurement information through the metro-logical concepts of traceability and uncertainty [5]. Rapid growth in digital communication such as cloud applications, distributed networks, smart devices and intelligent network architectures [3] demands new concepts for decision making based on reliable information. Existing cloud storage and services provide state-of-the-art capabilities for storing data but, on their own, provide no information on the provenance of the data nor on how the data can be interpreted correctly. Smart data is used to overcome this situation. In the scope of this article, 107 Acko, Weber, Hutzschenreuter, Smith the term "metrological smart data" refers to digital information comprising metrology data formatted for further data consolidation and analysis of the data in Internet of Things-networks (IoT-networks). Such smart data provides metadata, such as measurement units and uncertainties, describing the meaning and purpose of the underlying data in a machine-understandable form. Thereby, benefit lies in using common standards for the provision of relevant metadata. More traditional data, such as raw data from measurement sensors or measurement data in calibration certificates, cannot be classified as smart. In many cases, such data has incomplete information (e.g. when a sensor outputs numerical values but with no associated unit) or the metadata can only be accessed by human-interpretation of a (paper) document. Therefore, an essential component of a digitally-enabled metrology landscape for the IoT that can address the requirements of calibration, traceability and legal metrology [5, 6] is the automatic and secure communication of all relevant elements of the data and metadata formatted as smart data. This communication allows the unambiguous and correct interpretation of the data [7]. The interoperability of metrology data is severely degraded if essential information is lost or corrupted and current protocols do not address this issue. In general, the confusion, ambiguity and incorrect interpretation caused by missing metadata, diversity of units, etc., represent significant risks for future investment in IoT-technologies. If the IoT is to bring its benefits to society, it must be founded on well-engineered principles, including those derived from the metrological concepts of traceability, uncertainty and interoperability. In addition, to avoid future loss of information and consequent impact on decision-making, and to make secure human lives and environment, the exchange of metrological information (measurement results and assigned information) must be defined for all measurement tasks. Today, measurement results are communicated using base units of the International System of Units (SI) but also using domain-specific units such as foot, radian, inch, weber, gallon, etc. A BIPM brochure [8] provides guidance on using such derived units. However, this system is insufficient for the automated data processing required in the IoT, where information must be understood unambiguously and worldwide. One major goal of the research presented here is to define a data exchange format where the expression of measurement results in SI units [8-11] is mandatory. Optional information such as domain-specific or derived units will be covered, as well as additional information. The presented research is focused on establishing the secure, unambiguous and unified exchange of data in all communication networks where metrological smart data is needed. It aims to develop, provide and distribute a formal framework for the transmission of metrological data based on the SI. The framework will be applicable to all metrology domains. Furthermore, a worldwide-applicable concept for the use of digital calibration certificates (DCCs) will be made available for the first time. The development of demonstrators in two industrial domains will also prove the benefit and innovation potential for industry of the outputs of the project. The most important scientific contributions of this research are: • Establishing minimum required metadata models for the digital exchange of measurement data from a study of various international guides in the field of "traditional" measurement data and uncertainty representation, • Establishing minimum required information to be contained in DCCs and requirements for the secure usage of DCCs from a study of standard documents and samples of "traditional" paper calibration certificates, • Creating uniform metadata models that can help to facilitate interoperability of research data and increase reusability. 2. Research objectives, methods and the structure of research The overall objective of the research is to provide the methodological and technical foundation for the unambiguous, universal, safe and uniform communication of metrological data in the IoT. Guidelines are being developed that can be used, for example, for the definition of pre-normative 108 Advances in Production Engineering & Management 15(1) 2020 Communication and validation of metrological smart data in IoT-networks standards for the IoT and the supplementation of existing standards in order to define and harmonise the dissemination of measurement results and associated information. The specific objectives of the presented research are: 1. To define the requirements for a uniform, unambiguous and safe exchange format for measurement data and metrological information in an IoT network. The exchange format shall be based on the definition of SI units and meet central requirements from standards, guidelines and legal metrology. 2. To develop and establish secure DCCs. This objective requires consideration of exchange formats for administrative information, data transfer, cryptographic requirements, authentication and digital signatures. 3. To develop an online validation for services system for the types of data format as addressed under objectives 1 and 2. 4. To develop a reliable, easy-to-use, validated and secure online conformity assessment procedure designed for cloud system applications for legal metrology. The online conformity assessment procedure should also be applicable to calibration services and provide compliance with current international and European standards. 5. To build and validate demonstrators involving running applications from industrial stakeholders, to facilitate the uptake of the technology and measurement infrastructure developed in the project by the measurement supply chain, standards developing organisations and end users, and to work towards a European platform for metrological calibration services. The idea of the research with the objectives is presented in Fig. 1. Technical realisation of the research is divided into 5 work packages: • WP 1: Universal format for transfer of metrological data via digital communications, • WP 2: Digital calibration certificate (DCC) considering technical and legal requirements, • WP 3: Online validation of data formats and DCCs in digital communications, • WP 4: Online conformity assessment in legal metrology, • WP 5: SmartCom demonstrators. Each work package is split into diverse activities, which are only briefly described in this section. CALIBRATION CERTIFICATE Metrological information ANALYSE REQUIREMENTS (ISO/IEC 17025, data security...) DEFINE DATA - (text, numbers, symbols, units,...) DEFINE METADATA - • Formal data • Quantities • Numbers • Units • Comments DEFINE TRANSFER FORMATS DEFINE TRANSFER SECURITY MACHINE TOOL On-line: • Decision about quality • Controlling process parameters • Records and statistics ... Nominal dimensions, tolerances,... On-line data exchange Measured values, ... Conformance information •Formal data • Conformity statement • Numbers • Units • Comments t VERIFICATION CERTIFICATE ANALYSE REQUIREMENTS (ISO/IEC 17020, data security...) DEFINE DATA (text, numbers, symbols, units,...) DEFINE METADATA DEFINE TRANSFER FORMATS DEFINE TRANSFER SECURITY MEASUREMENT NSTRUMENT DEFINE DATA (text, numbers, symbols, units,...) DEFINE METADATA DEFINE TRANSFER FORMATS DEFINE TRANSFER SECURITY Legal requirements,.. On-line data exchange Measured values,.. LEGAL USER (e.g. petrol station) On-line: • Records • Decisions about instr. adjustments and services • Invoice creation • Reverification plans... Fig. 1 Schematic presentation of the research goals and outcomes Advances in Production Engineering & Management 15(1) 2020 109 Acko, Weber, Hutzschenreuter, Smith 2.1 Universal format for transfer of metrological data via digital communications The aim of this work package is the elaboration and definition of a fundamental description applicable to all metrological data [5-7] used in digital communication. For the first time, and linked to the worldwide-established SI, project partners have defined the measures that are essential for the easy-to-use, safe, harmonised and unambiguous digital exchange of metrological data. The use of non-SI units has also been considered. The resulting universal format for the digital transfer of metrological data realises an implementation of the minimum requirements from guides in both metrology and Information and Communication Technology (ICT). This part of the research is complete and details are presented in Section 3.1. 2.2 Digital calibration certificate (DCC) considering technical and legal requirements Within this work package, a universal structure of a DCC has been established. In order to obtain global acceptance of DCCs, the requirements for physical calibration certificates (paper form) of leading countries were first analysed. Using these requirements, a flexible and universal data structure was defined. The DCC was realised using extensible markup language (XML) [12, 13] with the metrological data represented according to the specifications developed in WP 1. In contrast to physical calibration certificates, new framework conditions required for digital communication [13, 14], such as the minimum requirements for transfer of encrypted data, authentication and digital signatures, will be developed and established (in a worldwide context and for the first time). Partial results are presented in Section 3.2. 2.3 Online validation of data formats and DCCs in digital communications This work package aims to establish a worldwide accessible online service for all applications where metrological data is exchanged, and the validation of metrology information within a DCC. Best practice guides for this service will be produced in order to guide software engineers, purchasers of products used in "intelligent" communication networks, and managers of quality management systems. A classification of metrology data regarding its usability for machines has been developed. The highest classification, termed "Platinum" or "Next generation", refers to data provided using only SI base units. Other classifications include "Gold", "Silver" and "Bronze" while the lowest classification "Improvable" refers to data that does not include sufficient information, for example, omitting a measurement unit entirely. This part of the research is ongoing and is expected to be concluded in 2020. 2.4 Online conformity assessment in legal metrology This work package will study requirements for a user-oriented and easy-to-establish online conformity assessment system that fulfils the general needs of legal preconditions. The research will focus on industrial users that develop sophisticated networks within digital networks and whose previous developments could not come to market due to restriction by national laws. The online conformity assessment system will consist of three parts: • XML-based communication interface [12], • Unified user interface (UniTerm), and • Security concept for the transmission of metrological information into a "world" outside the restricted and economic environment [13, 14]. This part of the research is ongoing and is expected to be concluded in 2021. 2.5 SmartCom demonstrators Pilot applications (demonstrators) will be implemented in this work package to prove the met-rological usability of the concepts developed during the research. The demonstrators will comprise the application of the validation system implemented in the TraCIM platform [15] on data from DCCs and an application of the UniTerm. 110 Advances in Production Engineering & Management 15(1) 2020 Communication and validation of metrological smart data in IoT-networks 3. Midterm research results 3.1 Machine-readable SI format for the exchange of metrological data This part of the research specifies the basic principles for the exchange of machine-readable data for all applications that transfer or require measurement data according to the specifications of the Système International d'Unités (SI) [8]. The research results thus provide the basis for the harmonised, clear, secure and economical exchange of digital measurement values for a universe in which digital data is being transferred in accordance with the specifications introduced below. This new approach addresses the need for improvement in secure data communication in the sense of reducing the risk of incomplete and incompatible data exchange such as mixing up length measurement values with units inch and centimetre. Calculations combining such incompatible data can lead to catastrophic results. In safety-critical areas, a consequence could be loss of human life. Using the new approach will also improve the universality of communication, as the data and its metadata are based on common minimum required information from highly-authorative international guides such as VIM [5], GUM [7] and the BIPM SI brochure [8]. Other data models are in many cases very domain specific and hence only usable in their field of application. For the digital exchange of metrological data, it is essential to associate at least one value to a corresponding unit [13, 14, 16]. These two pieces of information enable a statement to be made about the value of a quantity that can be interpreted according to the SI. Because of its indivisibility and fundamental importance, this form of representation is termed "atomic" (example: 1 kg). The complete indication of a measured quantity may contain additional information such as a specification of measurement uncertainty [7] and a time stamp. For a single quantity, measurement uncertainty is usually expressed by a coverage interval corresponding to a specified coverage probability [7, 17, 18]. A time stamp [13] is required if probing is undertaken for time-variant materials (or substances) and if a measured quantity or a constant is interpreted over a longer period of time. Fundamental physical constants (e.g. the Planck constant) [19-21] have changed several times since the introduction of the SI [8]. In a digital network in which existing and new applications communicate with each other, even greater importance than before must be assigned to the SI. The ability to use the SI to describe all physical processes using only seven base units leads to an unprecedented clarity that is fundamental for the secure exchange of data. It is important to distinguish between human-to-human and machine-to-machine interfaces. The specifications presented here primarily relate to an automated communication. It is essential for communication between machines and algorithms operating in an innovative digitised value chain. The basic idea of the machine-readable format for SI units is presented in Fig. 2. Digital System of Units (D-SI) feet, Pa binary ÍLlhi GUIS '¿k [&R] Fig. 2 Schematic presentation of the Digital System of Units (D-SI) as the universal format for the digital exchange of metrological data [22-23] metre, °C, decimal Ä. Ö, Ü, Méîpa bar unambiguous efficient easy to understand exchangeable Advances in Production Engineering & Management 15(1) 2020 111 Acko, Weber, Hutzschenreuter, Smith One of the research outcomes is also an adapter to allow the integration of non-SI units into the machine-readable D-SI data model [22-24]. This adapter is termed the "hybrid data model" or in short "hybrid". The outcomes of this research segment include digital formats for: • Machine-understandable unit format for Si-base units and units provided by the BIPM SI brochure [8], • Real quantities, • Complex quantities, • Lists of real quantities (vector of real quantities), • Lists of complex quantities (vector of complex quantities), and • Coverage regions (related to multivariate measurement uncertainty). For reasons of space, examples are presented only for the cases of a real quantity and a hybrid data model (for non-SI units). The complete D-SI data model and a reference implementation in XML (extensible markup language) have been published [22-24]. Real quantity The uniform data format for real quantities is shown in Fig. 3. The model contains the measurement value with a corresponding unit, the measurement uncertainty in the same unit and a time stamp. The components of the real quantity type in Fig. 3 were defined by considering the requirements of the most important metrological normative documents such as SI brochure [8], GUM [7], VIM [5], ISO 80000 [10, 11] and CODATA [19-21]. The data types of the components consider important standards from computer science such as IEEE 754, RFC 362 (UTF-8) [14] and ISO 8601 [13]. An example of an extended real quantity XML format [24, 25] is shown in Fig. 4. Y =y [SI] U9S(k) distribution UTC real quantity type extended components A. (of the real quantitjhjype) a* ■8 4> 3 C 3 O £ • « T3 M c D TJ I 01 2 m ï> S V s Basic real with expanded measurement uncertainty Basic real with coverage interval (probabilistic-symmetric) sub type Measurement value SI unit Measurement uncertainty Expansion factor Coverage probability Coordinated Universal Time Fig. 3 Uniform data format for real quantity temperature 20.10 Basic real \degreecelsius <3i : uncertainty>0 . 50 Expanded 2 0.95 uncertainty nontial Fig. 4 Example of XML implementation of real with expanded uncertainty 112 Advances in Production Engineering & Management 15(1) 2020 Communication and validation of metrological smart data in IoT-networks constant quantity type components (of the constant quantity type) label 01 3 ro > 'c 3 dateTirne uncertainty distribution constant quantity with an exact value constant quantity with an uncertainty (S) sub type Fig. 5 Data model for machine-readable fundamental physical constants Example a : XML representation of Planck constant before 2019-05-20: CODATA 2014 value Planck constant <31:value>6.626070040e-34 <31 : unjLt>\ joule\second <31:dateTime>2015-O6-25T00: 00:00Z0.000000081e-34 Example b : XML representation of Planck constant since 2019-05-20: new SI value Planc)t constant6,62607Q15e-34\joaXe\hertz\tothe( -X} <3i :uncextainïy>0 0,304S006 \metre l ft(U.S. survey) Fig. 7 Example of XML implementation of a real quantity with a non-SI unit in the hybrid data model 3.2 Digital calibration certificate (DCC) considering technical and legal requirements In the future, calibration services will require the exchange of comprehensive digital content of all kinds between customers, applicants and calibration service providers. Digital interfaces must therefore be developed and provided in such a way that the following aspects are guaranteed: authenticity; completeness of the transmitted data; data integrity and manipulation protection as well as protection of confidentiality. The first step for the exchange of calibration certificates in a digital environment is to have a uniform and internationally recognised structure of such digital documents. The following specification sets out the basic design for the structure of a DCC that was developed in WP 2 of the SmartCom project. This basic structure is founded on agreed standards, including ISO/IEC 17025 for calibration certificates [26] and internationally-accepted guides like the SI brochure [8], CODATA [19-21], VIM [5] and GUM [7]. Fundamental DCC-Layout The DCC designed within the project is structured in four layers and is presented in Fig. 8. These layers contain both regulated data, which are mandatory, and unregulated data, which are optional and contain additional information that does not necessarily have to be machine-readable. • Administrative shell The administrative layer represents regulated (administrative) data. It contains required information of core interest (i.e. is not optional), for the unambiguous identification and collection of administrative information of the DCC. This information includes the unique DCC ID, identification of calibration laboratory, customer and items. • Calibration results This layer contains a regulated area of measurement results according to the rules for the D-SI format [22, 23]. Moreover, individual additional information can be entered here in an unregulated area, e.g. individual calibration information, considering influence conditions, calibration methods and individual results. • Individual information For general, optional, and additional comments, calculation tables and graphics of any individual data formats, typically requested by the recipient of the certificate. • Optional attachment Here, a human viewable file can be stored (e.g. PDF format), which will typically be a conventional analogue calibration certificate. This layer will not be machine readable. 114 Advances in Production Engineering & Management 15(1) 2020 Communication and validation of metrological smart data in IoT-networks The structure described above provides for the integration of all aspects of ISO/IEC 17025 [26]. Firstly, industrial applications have been considered and have proven to be suitable for industrial requirements. The metrological data included in digital certificates must consist of a numerical value and a corresponding unit as a minimum. These specifications are described in section 3.1. The structure described above is not dependent on any programming language and will work with many file formats including XML and JSON. Used identifiers and correct expression of measurement results have been taken from worldwide harmonised guides and standards to ensure international (machine) readability of the documents. 1. Administrativeshell (mainly mandatory) 2. Calibration results (partly regulated) 3. Individual information (not regulated) 4. Optional Attachment: „Human readable document"(e. g, PDF) Fig. 8 General structure of digital calibration certificates 4. Conclusion SmartCom is one of the first projects in metrology to define the universal minimum requirements for machine-readable data exchange in digital communication. The Digital System of Units (D-SI) metadata model presented in this paper can help developers of data formats to implement their data in an unambiguous, easy-to-use, safe and uniform way that is based on the International System of Units and other internationally accepted guides. It provides a data basis for representing data in future digital applications in metrology, such as: data for metrological services; data exchanged between virtual measuring instruments (DCC is virtual representation of properties of measuring artefacts and instruments). Analysis of big data is also facilitated if data is based on common terminology in metrology. Digital calibration certificates (DCCs) are a very important application of the metrological data exchange. Using the principles from the D-SI, XML as machine-readable format and fundamental requirements from ISO/IEC 17025, a general structure for DCCs was specified. In the future, DCCs will record all aspects of the calibrated items and make them available to a comprehensive quality management system. With these complete data sets, the performance of systems and processes can then be captured effectively and efficiently, allowing data analytic methods to provide information on optimised system performance. This activity leads to reduced downtime, less waste, significant improvement in quality, and ultimately greater economic success. The work in the SmartCom project will be concluded in 2021. Until then, further tools will be developed to support the application of the DCC and D-SI [15]. The elaboration of aspects of cryptography will be of great importance for the transmission and use of DCCs. Suitable methods must be used to guarantee the integrity, completeness and authenticity of the calibration data. This area has proven to be particularly complex. No international standard in metrology is yet available for secure transmission, digital stamps and signatures or the withdrawal of digital data [27]. Preliminary approaches are using or envision the use of regulations such as the European eIDAS law [28] and blockchain technology in various areas such as Legal Metrology [29]. Finally, web-based services are being developed to help users of the D-SI and DCC to validate the correct usage of the data structure. PTB and NPL are planning to make available the validation services under the auspices of the TraCIM online test system. Ei^KJ iro Ci, + Framework conditions + legal requirements Advances in Production Engineering & Management 15(1) 2020 115 Acko, Weber, Hutzschenreuter, Smith Acknowledgement The authors would like to acknowledge funding of the presented research within the European Metrology Programme for Innovation and Research (EMPIR) in the Joint Research Project 17IND02 SmartCom. Furthermore, we like to thank all partners of the SmartCom project and all involved colleagues from PTB - especially Siegfried Hackel - for their commitment to the development of the D-SI and the DCC. References [1] Gajsek, B., Marolt, J., Rupnik, B., Lerher, T., Sternad, M. (2019). Using maturity model and discrete-event simulation for Industry 4.0 implementation, International Journal of Simulation Modelling, Vol. 18, No. 3, 488-499, doi: 10.2507/HSIMM18(3)489. [2] Vieira, A.A.C., Dias, L.M.S., Santos, M.Y., Pereira, G.A.B., Oliveira, J.A. (2018). Setting an Industry 4.0 research and development agenda for simulation - A literature review, International Journal of Simulation Modelling, Vol. 17, No. 3, 377-390, doi: 10.2507/IISIMM17(3)429. [3] Resman, M., Pipan, M., Šimic, M., Herakovič, N. (2019). A new architecture model for smart manufacturing: A performance analysis and comparison with the RAMI 4.0 reference model, Advances in Production Engineering & Management, Vol. 14, No. 2, 153-165, doi: 10.14743/apem2019.2.318. [4] Vujica Herzog, N., Buchmeister, B., Beharic, A., Gajsek, B. (2018). Visual and optometric issues with smart glasses in Industry 4.0 working environment, Advances in Production Engineering & Management, Vol. 13, No. 4, 417428, doi: 10.14743/apem2018.4.300. [5] JCGM 200:2012 (2012). International vocabulary of metrology - Basic and general concepts and associated terms (VIM), 3rd edition, 2008 version with minor corrections, from https://www.bipm.org/en/publications/guides/, accessed November 8, 2019. [6] Euramet (2008). Metrology - in short, 3rd edition, from https://www.euramet.org/publications-media-centre/ documents/metrology-in-short/, accessed November 18, 2019. [7] JCGM 100:2008 (2008). Evaluation of measurement data - Guide to the expression of uncertainty in measurement, 1st edition, from https://www.bipm.org/en/publications/guides/, accessed October 21, 2019. [8] BIPM (2019). The international system of units (SI), 9th edition, from https://www.bipm.org/en/publications/si-brochure/. accessed November 8, 2019. [9] Thompson A., Taylor, B.N. (2008). Guide for the use of the international system of units (SI). NIST Special Publication 811, National Institute of Standards and Technology, Gaithersburg, USA, doi: 10.6028/NIST.SP.811e2008. [10] ISO 80000-1 (2009). Quantities and units - Part 1: General, 1st edition, International Organization for Standardization, Geneva, Switzerland. [11] IEC 80000-13:2008 (2008). Quantities and units - Part 13: Information science and technology, International Organization for Standardization, Geneva, Switzerland. [12] World Wide Web Consortium Extensible Markup Language (XML); version 1.0; 5th edition, from https://www. w3.org/TR/xml/, accessed November 6, 2019. [13] ISO 8601:2004 (2004). Data elements and interchange formats - Information interchange - Representation of dates and times, International Organization for Standardization, Geneva, Switzerland. [14] RFC 3629 UTF-8 (2003). A transformation format of ISO 10464, https://tools.ietf.org/html/rfc3629. accessed November 10, 2019. [15] PTB (2019). TraCIM service operated at PTB, from https://tracim.ptb.de, accessed November 29, 2019. [16] IEC TS 62720:2017 (2017). Identification of units of measurements for computer-based processing, International Organization for Standardization, Geneva, Switzerland. [17] JCGM 101:2008 (2008). Evaluation of measurement data - Supplement 1 to the "Guide to the expression of uncertainty in measurement" - Propagation of distributions using a Monte Carlo method, 1st edition, from https://www.bipm.org/utils/common/documents/icgm/lCGM 101 2008 E.pdf, accessed October 5, 2019. [18] JCGM 102:2011 (2011). Evaluation of measurement data - Supplement 2 to the "Guide to the expression of uncertainty in measurement" - Extension to any number of output quantities, from https://www.bipm.org/utils/common/documents/icgm/lCGM 102 2011 E.pdf, accessed October 2019. [19] Mohr, P. J., Newell, D. B., Taylor B. N. (2015). CODATA Recommended values of the fundamental physical constants: 2014, doi: 10.6028/NIST.SP.961r2015, accessed December 16, 2019. [20] NIST (2018). CODATA Internationally recommended 2018 values of the fundamental physical constants, from http://physics.nist.gov/constants, accessed September 12, 2019. [21] PTB (2019). Complete list of machine readable CODATA 2018 values, from https://ptb.de/si/codata/ m2m constants CODATA 2018 all v2018.1.xml, accessed December 09, 2019. [22] Hutzschenreuter, D., Härtig, F., Heeren, W., Wiedenhöfer, T., Forbes, A., Brown, C., Smith, I., Rhodes, S., Linkeová, I., Sykora, J., Zeleny, V., Ačko, B., Klobučar, R., Nikander, P., Elo, T., Mustapää, T., Kuosmanen, P., Maennel, O., Hovhannisyan, K., Müller, B., Heindorf, L., Paciello, V. (2019). SmartCom Digital System of Units (D-SI) Guide for the use of the metadata-format used in metrology for the easy-to-use, safe, harmonised and unambiguous digital transfer of metrological data, (Version D-SI 1.3), Zenodo, doi: 10.5281/zenodo.3522631, accessed November 21, 2019. 116 Advances in Production Engineering & Management 15(1) 2020 Communication and validation of metrological smart data in IoT-networks [23] Hutzschenreuter, D., Härtig, F., Wiedenhöfer, T., Smith, I., Brown, C. (2019). D-SI in short - Digital brochure on establishing the use of units in digitalised communication, Zenodo, doi: 10.5281/zenodo.3522074, accessed November 28, 2019. [24] Hutzschenreuter, D., Härtig, F., Wiedenhöfer, T., Hackel, S.-G., Scheibner, A., Smith, I., Brown, C., Heeren, W. (2019). SmartCom Digital-SI (D-SI) XML exchange format for metrological data version 1.3.0, doi: 10.5281/zenodo.3366902, accessed October 24, 2019. [25] Wright, J. (2019). siunitx - A comprehensive (SI) units package, Version v2.7t, from http://mirrors.ctan.org/macros/latex/contrib/siunitx/siunitx.pdf. accessed November 3, 2019. [26] SIST EN ISO/IEC 17025:2017 (2017). General requirements for the competence of testing and calibration laboratories, Slovenski inštitut za standardizacijo (Slovenian Institute for Standardization), Ljubljana, Slovenia. [27] Nikander, P., Elo, T., Mustapää, T., Kuosmanen, P., Hovhannisyan, K., Maennel, O., Brown, C., Dawkins, J., Rhodes, S., Smith, I., Hutzschenreuter, D., Weber, H., Heeren, W., Schönhals, S., Wiedenhöfer, T. (2020). Document specifying rules for the secure use of DCC covering legal aspects of metrology, doi: 10.5281/zenodo.3664211, accessed March 24, 2020. [28] Regulation (EU) No 910/2014 of the European Parliament and of the Council of 23 July 2014 on electronic identification and trust services for electronic transactions in the international marked and repealing Directive 1999/93/EC (2014). Official journal of the European union, L 257, 73-115. [29] Peters, D., Wetzlich, J., Thiel, F., Seifert, J.-P. (2018). Blockchain applications for legal metrology, In: Proseedings of 2018 IEEE International Instrumentation and Measurement Technology Conference (I2MTC], Houston, USA, doi: 10.1109/I2MTC.2018.8409668. Advances in Production Engineering & Management 15(1) 2020 117 Calendar of events • 5th International Conference on 3D Printing Technology and Innovations, March 16-17, 2020, Berlin, Germany. • 8th International Conference and Exhibition on Automobile & Mechanical Engineering, May 18-19, 2020, Berlin, Germany. • International Conference on 3D Printing, Advanced Robotics and Automation (3DPARA), May 21-22, 2020, London, UK. • 20th International Conference on Materials Science and Engineering, May 25-26, 2020, Osaka, Japan. • 6th International Conference and Expo on Ceramics and Composite Materials, June 8-9, 2020, Frankfurt, Germany. • 26th International Conference on Advanced Materials, Nanotechnology and Engineering, June 17-18, 2020, Brisbane, Australia. 118 Advances in Production Engineering & Management 15(1) 2020 Notes for contributors General Articles submitted to the APEM journal should be original and unpublished contributions and should not be under consideration for any other publication at the same time. Manuscript should be written in English. Responsibility for the contents of the paper rests upon the authors and not upon the editors or the publisher. Authors of submitted papers automatically accept a copyright transfer to Chair of Production Engineering, University of Maribor. For most up-to-date information on publishing procedure please see the APEM journal homepage apem-journal.org. Submission of papers A submission must include the corresponding author's complete name, affiliation, address, phone and fax numbers, and e-mail address. All papers for consideration by Advances in Production Engineering & Management should be submitted by e-mail to the journal Editor-in-Chief: Miran Brezocnik, Editor-in-Chief UNIVERSITY OF MARIBOR Faculty of Mechanical Engineering Chair of Production Engineering Smetanova ulica 17, SI - 2000 Maribor Slovenia, European Union E-mail: editor@apem-journal.org Manuscript preparation Manuscript should be prepared in Microsoft Word 2010 (or higher version) word processor. Word .docx format is required. Papers on A4 format, single-spaced, typed in one column, using body text font size of 11 pt, should not exceed 12 pages, including abstract, keywords, body text, figures, tables, acknowledgements (if any), references, and appendices (if any). The title of the paper, authors' names, affiliations and headings of the body text should be in Calibri font. Body text, figures and tables captions have to be written in Cambria font. Mathematical equations and expressions must be set in Microsoft Word Equation Editor and written in Cambria Math font. For detail instructions on manuscript preparation please see instruction for authors in the APEM journal homepage apem-journal.org. The review process Every manuscript submitted for possible publication in the APEM journal is first briefly reviewed by the editor for general suitability for the journal. Notification of successful submission is sent. After initial screening, and checking by a special plagiarism detection tool, the manuscript is passed on to at least two referees. A double-blind peer review process ensures the content's validity and relevance. Optionally, authors are invited to suggest up to three well-respected experts in the field discussed in the article who might act as reviewers. The review process can take up to eight weeks on average. Based on the comments of the referees, the editor will take a decision about the paper. The following decisions can be made: accepting the paper, reconsidering the paper after changes, or rejecting the paper. Accepted papers may not be offered elsewhere for publication. The editor may, in some circumstances, vary this process at his discretion. Proofs Proofs will be sent to the corresponding author and should be returned within 3 days of receipt. Corrections should be restricted to typesetting errors and minor changes. Offprints An e-offprint, i.e., a PDF version of the published article, will be sent by e-mail to the corresponding author. Additionally, one complete copy of the journal will be sent free of charge to the corresponding author of the published article. APEM journal Advances in Production Engineering & Management Chair of Production Engineering (CPE) University of Maribor APEM homepage: apem-journal.org Volume 15 | Number 1 | March 2020 | pp 1-120 Contents Scope and topics Neuro-mechanistic model for cutting force prediction in helical end milling of metal materials layered in multiple directions Zuperl, U.; Cus, F.; Zawada-Tomkiewicz, A.; Stçpierï, K. Improvement of logistics in manufacturing system by the use of simulation modelling: A real industrial case study Straka, M.; Khouri, S.; Lenort, R.; Besta, P. Estimating the position and orientation of a mobile robot using neural network framework based on combined square-root cubature Kalman filter and simultaneous localization and mapping Wang, D.; Tan, K.; Dong, Y.; Yuan, G.; Du, X. A comparison of the tolerance analysis methods in the open-loop assembly Kosec, P.; Skec, S.; Miler, D. Awareness and readiness of Industry 4.0: The case of Turkish manufacturing industry Sari, T.; Gùle§, H.K.; Yigitol, B. Assembly transport optimization for a reconfigurable flow shop based on a discrete event simulation Yang, S.L.; Xu, Z.G.; Li, G.Z.; Wang, J.Y. The impact of using different lean manufacturing tools on waste reduction Leksic, I.; Stefanic, N.; Veza, I. Integrated management systems based on risk assessment: Methodology development and case studies Vulanovic, S.; Delic, M.; Kamberovic, B.; Beker, I.; Lalic, B. Communication and validation of metrological smart data in IoT-networks Acko, B.; Weber, H.; Hutzschenreuter, D.; Smith, I. Calendar of events Notes for contributors Copyright © 2020 CPE. All rights reserved. apem-journal.org 4 5 31 44 57 69 81 93 9771854625008