A Suppier Evauation System for Automotive Industry According To Iso/Ts 16949 Requirements DILEK PINAR ÖZTOP 1, ASLI AKSOY 2,*, NURSEL ÖZTÜRK 2 1 HONDA TR Purchasing Department, 41480, Çayırova - Gebze, Turkey 2 Uudag University, Facuty of Engineering, Industria Engineering Department, Goruke Campus, 16059 Bursa, Turkey * Corresponding author ABSTRACT ISO/TS 16949 technica specification combines demands of a automotive manufacturers and indicates a we-defined, unique quaity assurance system target for a companies in automotive industry. This study improves a specia suppier seection and evauation system for the companies in automotive industry due to the necessities of ISO/TS 16949.A hierarchica frame is first created by Voting Anaytic Hierarchy Process (VAHP) to define the criteria and sub criteria of a quaity assurance system. VAHP is improved by integrating Quaity Function Depoyment (QFD) when assigning weights of each criterion and sub criteria of the system. Linear programming mode is used to convert the voting resuts to weights. Keywords-Anaytic hierarchy process, ISO/TS 16949, Quaity function depoyment, Suppier seection, Voting anaytic hierarchy process I. INTRODUCTION ISO/TS 16949 is a technica specification prepared by the Internationa Automotive Task Force (IATF) and Japan Automobie Manufacturers Association, Inc. (JAMA) with support from ISO/TC 176, quaity management and quaity assurance. ISO/TS 16949 technica specification is one of the most eading topics of automotive industry quaity assurance system. This technica specification combines demands of a automotive manufacturers and indicates a wedefined, unique quaity assurancesystem target for a companies in automotive industry. ISO/TS 16949 technica specification incudes terms about necessity and importance evauating suppiers reguary due to specification s needs. This is more critica for companies whichcoaborate with many suppiers. Suppier seection and evauation system shoud aso incude technica specification s musts. There are many studies which found pace in iterature about suppier seection and evauation systems. However, there is ack of suppier evauation systems for automotive industry incuding ISO/TS 16949 requirements. This study improves a specia suppier seection and evauation system for the companies in automotive industry due to the necessities of ISO/TS 16949. In the study quaity assurance system necessities are expressed in a hierarchica structure as criteria and sub-criteria. A hierarchica frame is first created by Voting Anaytic Hierarchy Process (VAHP) to define the criteria and sub-criteria. However, in a quaity assurance system there are more compex reationships. Voting Anaytic Hierarchy Process (VAHP) is improved by integrating Quaity Function Depoyment (QFD) approach when assigning weights of each criterion and sub-criteria of the system. Two QFD structures are created to represent matrice reationships between hierarchy ayers. At the next step inear programming is used to convert the voting resuts to weights. The formuization is deveoped for the situation of having inear partia information. On the next step, minimum and maximum effect bounds of sub- criteria is cacuated. The output of the mathematica programming of QFD matrice is used as an input of the mathematica programming. As ast step, the average effects of criteria and sub-criteria are normaised. The ast weights wi be used in suppier seection and evauation system directy. The point scae for the system is seected as [1-10] to et a fexibe evauation of each sub criteria. Then, suppiers are cassified in four groups in this study. II. LITERATURE REVIEW There are many studies which found pace in iterature about suppier seection and evauation systems. Some indicated methods in those studies are suitabe to be used for both suppier seection and evauation and some of them are suitabe to be used for ony seection or evauation. Akinc [1] improved a suppier seection method which minimizes annua materia cost and quantity of suppiers whie maximizing the number of suppiers with a high performance of deivery and quaity. Roodhooft and Konings [2] improved a suppier seection and evauation system by an activity based costing approach. System aows cacuating the 2356 P age
tota cost caused by a defined suppier in the production process of an enterprise. Li et a. [3], improved a fuzzy method to compensate bindness of peope when decision making for suppiers. Boer et a. [4] advised ELECTRE I to be used for eiminating best candidates from a big suppier candidate group and ELECTRE III for ranking those seected high performance candidate suppiers. Liu et a. [5], uses the data enveopment anaysis (DEA) to seect best suppiers for a defined product which is a cassica muti-criteria decision making probem. Tauri and Sarkis [6] suggest a new muti criteria evauation mode which uses many performance criteria for suppier performance evauation. Chen and Yang[7] indicated a method to integrate suppier and manufacturer capabiities under scope of increasing profitabiity which is improved from decreasing cost of purchased materias. Dumin and Mininno [8], appied a new muti criteria decision making method to a midde size transportation enterprise in Itay by taking continuous changing of performance criteria, financia importance and mutipurpose structure of suppier seection decision into account. Chan [9] defined five new performance indicators as addition to cassica cost and quaity performance criteria which are commony used in iterature: source usage, fexibiity, transparency, trust and innovation. Anaytic Hierarchy Process (AHP) is used as muti variabe decision making method. Wang et a. [10], improved a preventive goa programming based muti criteria decision making system which integrates AHP by considering both quaitative and quantitative factors. Pi and Low [11] cacuated the quaity oss which is caused by a unique suppier by using Taguchi ost function by using quaity, deivery on time, cost and service criteria. Lasch and Janker [12] made a research in 193 enterprises about current suppier scoring systems and proved that current systems cannot fuy meet practica needs. They proposed an aternative muti criteria decision making system which compares the idea suppier with a suppiers and introduces a ranking and cassifying resut. Hong et. a. [13] improved mathematica programming mode which can considerate changes of suppy capabiities, suppiers and customer expectations in a defined time interva. Chen et a. [14] suggested a fuzzy decision making approach for compex decision of suppier seection. The method is suitabe to use when performance variabes cannot be defined numericay and foreknowedge and subjective forecasts exist. Kumar et a. [15] improved a fuzzy muti-purpose integer programming system for suppier seection probem with three main goas: cost minimization, quaity maximization and just in time deivery. Ordoobadi [16] described a decision mode that appies fuzzy arithmetic operators to manipuate and quantify decision maker s subjective assessments. Aksoy and Öztürk [17] presented, neura network (NN) based suppier seection and performance evauation system in JIT production environment. III. PROPOSED APPROACH In this study, quaity assurance system necessities are expressed in a hierarchica structure as criteria and sub-criteria. A hierarchica frame is first created by VAHP to define the criteria and sub-criteria and QFD is used to anayse matrice reations between criteria and sub-criteria, then, to convert the voting resuts to weights, inear programming mode is used. The detais of those steps expained in foowing sections. 3.1 Criteria and sub criteria of suppier seection and evauation system There are severa criteria reated to suppier seection process described in the iterature. In this research, six criteria are determined by considering the effects to the genera quaity assurance system performance of a company. The criteria are shown beow: - Quaity performance, - Cost/cost management performance, - Deivery performance, - Management performance, - Continuous deveopment performance - Suppier s reationship performance (indicates reations between suppier and it s sub suppier) These criteria don t incude enough detais to assess a quaity assurance system and sub criteria are created for each criterion as shown beow. Sub-criteria of quaity performance are: - SC 1.1. Advanced product quaity panning shoud be appied in case of new product - SC 1.2. Quaity tracking of products and processes shoud be effective. - SC 1.3. Statistica Process Contro (SPC) shoud be appied and incuded in contro pan. - SC 1.4. Product inspection and contro activities shoud be done effectivey. - SC 1.5. Inspection and production toos shoud be caibrated. - SC 1.6. Defect prevention and soution techniques shoud be appied effectivey. - SC 1.7. Quaity targets shoud be defined. Quaity performance data shoud be recorded and controed. - SC 1.8. Interna audits shoud be done in the company in order to contro conformance to ISO/TS 16949 needs. 2357 P age
- SC 1.9. Quaity poicy shoud be defined by top management and shoud be suitabe to continuousy improvement principe. Quaity management system shoud be documented. Sub criteria of cost/cost management performance are: - SC 2.1. Financia management system shoud be effective. Short and ong term financia panning shoud be done. - SC 2.2. Cost management system shoud be effective. Cost of ow quaity shoud be measured. - SC 2.3. Product panning system shoud be based on demand; and it shoud be suitabe to reach production information at basic steps of process. - SC 2.4. Inventory panning and contro system shoud be effective and based on MRP. Sub criteria of deivery performance are: - SC 3.1. Deivery performance shoud be measured by an indicator and resuts shoud satisfy customer requirements. - SC 3.2. Capacity shoud be managed to meet customer demand. - SC 3.3. Batch quantities, packaging and abeing activities shoud be suitabe to customer expectations. Products shoud be protected from deivery to customer. - SC 3.4. Storage activities shoud be effective. Sub criteria of management performance are: - SC 4.1. Reationship in the company shoud be efficient. - SC 4.2. Organization chart of the peope effecting product quaity shoud be defined. - SC 4.3. A business pan suitabe to company targets shoud be done and documented for short and ong term. - SC 4.4.Top management review meetings shoud be done on panned periodic intervas. - SC 4.5. A empoyees shoud be trained about their job, safety, quaity system. Empoyees who affect quaity directy shoud have training on the job. - SC 4.6. Company shoud meet the requirements of environment, heath and safety. - SC 4.7. Infrastructure and a inputs shoud be enough to meet product conditions. A proactive maintenance poicy shoud be panned for toos. - SC 4.8. Layout shoud be organized to hep meeting quaity needs, to have an effective usage of areas and easy materia transportation. - SC 4.9. Company shoud obey confidence principes in case of any need. Sub criteria of continuous deveopment performance are: - SC 5.1. A activities supporting continuous deveopment shoud be effective and positive resuts shoud be obtained. - SC 5.2. Customer satisfaction shoud be the first target of the company. An efficient reationship is needed to use the voice of the customer as a guide in the company. - SC 5.3. A process based quaity management system shoud be impemented. Sub criteria of suppier reationship performance are: - SC 6.1. Suppier seection system shoud be defined and appied. - SC 6.2. Suppier evauation system shoud be defined and appied; reevant data shoud be coected. The fina hierarchica structure of criteria and sub criteria is shown in Fig. 1. Figure 1. Hierarchica structure of criteria and sub criteria 3.2 Methodoogy of assigning weights of criteria and sub criteria AHP is deveoped by Saaty to assist in muticriteria decision-making probems. This method basicay uses pairwise comparisons in decision making. VAHP is seected as the base of seection and evauation system appication. VAHP is an improved version of AHP. The reasons to prefer VAHP to AHP is summarised beow: 1. VAHP is a more efficient system for group works. AHP needs a team members of the study to be at the same pace on work time. It is not possibe for a participants to be on the highest concentration eve during the study. Moreover some participants can remain passive to decare their ideas. VAHP ets each participant to decare their ideas separatey on the time that they prefer. 2. AHP uses a scae during pairwise comparison. The scae directy affects the efficiency of comparison resuts and causes a risk. 3. Appying VAHP is easier than AHP and takes ess time. 4. AHP uses pairwise comparison for assessment criteria and aso for the candidates. Comparing each pair of candidate suppiers causes a big 2358 P age
workoad. Moreover, when a company uses AHP to seect or rank suppiers, it has to refresh the same study when a new suppier is added to vendor ist. This proves that using AHP to seect suppier doesn t aow creating a suppier database. However having a suppier database is very critica especiay for the companies which have arge number of suppiers. By using VAHP method, a new suppier can easiy be assessed by the system and the resuts can be added to database. 5. AHP is suitabe to seect the best suppiers through candidates, however is not suitabe to assess a unique suppier. VAHP uses the voters idea when ranking criteria or subcriteria. nvoters rank separatey the criteria on the same hierarchy eve. For exampe each voter ranks the criteria at the first hierarchy eve: quaity, cost, deivery, management, continuous improvement, suppier reationship. Then, each voter ranks the sub criteria at the second hierarchy eve which beong to the same criteria. For exampe, SC 1.1-1.9 beongs to quaity criteria and these nine sub criteria are ranked by each voter. SC 2.1-2.4 beongs to cost criteria and these four sub criteria are ranked by each voter. At the ast step, VAHP uses Noguchi s [18] tota ranking method to assign weights to each criterion and subcriteria due to the votes of n voters. However, in rea ife, quaity assurance system incudes more compex reationships. In Fig 1, each sub criteria beongs to the criteria which effects more, but, aso, sub-criteria are not competey independent from other criteria. For exampe, SC 6.1. effects suppier reationship criteria directy, and the effect is high. However, SC 6.1 may have effect on quaity, cost, deivery, management or continuous improvement criteria. SC 6.1. s effect to other criteria may be smaer but they are aso important. To assess this compex reationship quaity function depoyment (QFD) approach is used. Han et a. [19] deveoped a inear programming mode to assess matrix reationships between hierarchy eves and to assign weights to each criterion and sub criteria due to vote of n voters. Two QFD structures are created to represent matrice reationships between hierarchy eves. Tabe 1 shows the matrice structure incuding sub criteria s effects to criteria. Tabe 2 shows the matrice structure incuding criteria s effects to genera quaity assurance system performance. In QFD matrices, x, y, z, [ ] show the degree of reations. x represents very strong reationship, y represents strong reationship, z represents weak reationship and [] represents no reationship in case. In a methodica voting appication, this evauation is done by 15 ISO/TS 16949 auditors separatey and the most voted degree is seected as the resut of a ce. For Tabe 1. Reationship matrice between criteria and sub criteria Tabe 2. Reationship matrice between criteria and genera quaity assurance system performance exampe, if %60 of the 15 voters seected x for the effect of SC 1.1. to quaity, the resut of the voting for this ce is seected as x. In this study, the inear programming mode to convert the voting resuts to weights, introduced by Han et a. [19] is used. The formuization is deveoped for the situation of having inear partia information. The reason for this inear partia information acceptance is possibiity of decision making under time pressure or ack of data, difficuty in expression ideas by numbers or by some kind of degrees, compex and uncertain environment and ack of concentration. In a QFD matrice, when customer attributes (CA) are on the rows and when the engineering characteristics (EC) are on the coumns, notation of the inear programming mode is shown beow: c i customer attribute e i engineering characteristics hierarchy eves of CA s {0,1,,L} L+1 eve of the hierarchy tree number of CA at th eve m Y j the set of chid nodes ofjth attribute at the th eve 0 Y j the set of customer attributes at owest eve (Y 0 = Y 0 j ) d i importance rating of ith attribute at th eve (d 0 i = d i ) r ij effect of the jth engineering characteristic to owest eve ith customer attributes we j weight of each engineering characteristic 2359 P age
θ min (ec j ) θ max (ec j ) minimum absoute bound of ec j maximum absoute bound of ec j minimum effect of ec j is cacuated maximum vaue of r ij s are put in the mode. Rrepresents the constraints due to partia information of voters aboutr ij. Rset can be divided into R i sets for each c i (i=1,2,,m 0 ). R i set is composed of ER i partia n information sets from voters or deciders and j =1 r ij = 1. So that two sets can be defined asr=i=1m0ri and n R i = ER i { j =1 r ij = 1} (Han et a., 2004). Constraint sets can be defined to represent partia information of c i s. γ(d i ) (=1,2,..,L) is the partia information set from voters or deciders for d i γ(d 2 3 )is partia information set 2 of voters or deciders for d 3 iny 2 3. Aso for a c i s on n 1 eve (=1,2,..,L), a constraint is d i = d j. For a c i s on eve (=1,2,..,L), φ = { i Y i n 1 d i Y i i = d j }set is defined. This equation shows that sum of the vaue of chid nodes are equa to the vaue of parent node. m Then, 0 i=1 d 0 i = 1 is defined for customer attributes constraints and the sum of the vaue of customer attributes on owest eve is 1. These three constraints set for c i s are defined by set H = γ d i φ m 0 0 { i=1 d i = 1}. Linear programming mode set is defined as beow: 3.3 Assigning weights of criteria and sub criteria In suppier seection and evauation system probem, sub criteria are in the pace of e i s and criteria are in the pace of c j s for the QFD matrice in Tabe 1and, for the QFD matrice in Tabe 2. criteria are in the pace of e i s and genera performance of quaity assurance system is in the pace ofc j. Mathematica mode is soved first for the QFD in Tabe 2, and then for the QFD in Tabe 1. In partia information case reationship between x, y, z voting of deciders can be defined as 1.5y x 3y, 1.5z y 3z and 3z x 6z. Tabe 3 (a, b) presents the resuts of the mathematica programming mode cacuating min/max r ij vaues of QFD matrice in Tabe 1. Tabe 3(a). The vaue interva of r ij s of matrice in Tabe 1 s.t. and s.t. θ min ec j = min m 0 d i i=1 δ (ec j )(1) d i H(2) d i 0, i = 1,2,, m, = 0,1,2,, L(3) δ ec j = minr ij (4) r ij R i (5) r ij 0 (6) Tabe 3(b). The vaue interva of r ij s of matrice in Tabe 1 In this inear programming set, first the vaue of δ (ec j )wi be derived for a c i (i Y 0 ) by soving the second inear programming mode and then θ min (ec j ) wi be minimized by putting the derived vaue of δ (ec j ) in the objective function. By soving the second inear programming probem, minimum vaue of a r ij is cacuated; by soving the same inear programming mode after putting max to the objective function maximum vaue of a r ij is cacuated. At the second stage the first inear programming mode is soved by putting minimum at the objective function to cacuate the minimum effect bound of ec j and the same programming mode is soved by putting maximum at the objective function to cacuate the maximum effect bound ofec j. When minimum effect of ec j is cacuated minimum vaue of r ij s are put in the mode and when Next step incudes the cacuation of minimum and maximum effects bounds of sub criteria. The x, y, z effects of criteria to genera quaity assurance system performance is used as d i in the mathematica mode. This means that when a criterion has a higher effect to genera performance, its importance rating d i is higher. When having partia information x, y, z reationship in Tabe 2 can be stated as: 1.5d 6 d 1 3d 6 and d 1= d 2 = d 3 = d 4 = d 5 Tabe 4(a, b) shows the minimum-maximum effect bounds and average effects of a sub criteria. 2360 P age
Tabe 4 (a). Effect bounds of sub criteria to a criteria Tabe 4 (b). Effect bounds of sub criteria to a criteria As a ast step, the average effects of criteria and subcriteria are normaised via equation (7): r j = r j and r j r j j = 1(7) j Tabe 5 (a, b) and Tabe 6 show the average effects of a criteria and subcriteria after normaisation. Tabe 5 (a). Average normaised effects of sub criteria Tabe 5 (b). Average normaised effects of sub criteria Tabe 6. Average normaised effects of criteria 3.4 Design of suppier seection and evauation system By appying AHP, the fina weight of each sub criteria is cacuated by mutipying the weight of the sub criteria and the criteria that it beongs in hierarchica eve. Fina weights of sub criteria are shown in Fig.2. Fina weights seen in Fig 2 wi be used in suppier seection and evauation system directy. The point scae for the system is seected as [1-10] to et a fexibe evauation of each sub criteria. This means that an auditor wi rate each sub criteria between 1 and 10; and mutipy each rate of the sub criteria with the weight of reated sub criteria in Fig.2. Suppiers are cassified in four groups according to their scores in this system. Figure 2. Fina weights of sub criteria - Cass_A suppier: System definitey advices to work with this cass. - Cass_B suppier: A System advice to work with this cass, but, an action pan is demanded from suppier to reach Cass_A eve after 6 months. - Cass_C suppier: System doesn t advice to work with this cass. An action pan is demanded from suppier to reach Cass_B after 3 months. If the suppier is successfu to reach Cass_B, it s evauated again due to rues of Cass_B suppiers. - Cass_D suppier: System definitey doesn t advice to work with this cass. Action pan is not demanded. In the proposed system maximum score that a suppier may have is 1.75531 and the minimum score is 0.17553. So the score intervas of the casses are defined as beow: - Cass_A suppier: [1.36037-1.75531] - Cass_B suppier: [0.96542-1.36036] - Cass_C suppier: [0.57048-0.96541] - Cass_D suppier: [0.17553-0.57047] IV. CONCLUSION ISO/TS 16949 technica specification is one of the most eading topics of automotive industry quaity assurance system. This specification combines separate demands of automotive manufacturer companies in Europe and America and presents a unique quaity assurance standard to a companies in automotive industry. ISO/TS 16949 technica specification incudes terms about necessity and importance evauating suppiers reguary due to specification s needs. Quaity assurance system s efficiency is directy inked to the quaity assurance system of suppiers. This is more critica for companies which coaborate with many suppiers. Suppier seection system shoud aso incude technica specification s musts. This study proposes a systematic approach for automotive industry to evauate current suppiers and seect new suppiers due to ISO/TS 16949 requirements. The proposed system can aso be used in a company to assess their own quaity assurance system. Proposed system is very simpe to 2361 P age
use and ready to appy in industry. It doesn t need neither a specia training about the evauation system nor ISO/TS 16949 for appication. In this study quaity assurance system necessities are expressed in a hierarchica structure as criteria and sub criteria. This is important to et an auditor to focus on detais of a quaity assurance system. VAHP methodoogy is improved to assess matrice reationships of quaity assurance system hierarchy and QFD discipine is used to assess the reationships between criteria and sub criteria. QFD et to reaize the effect of a sub criteria to a criteria, not ony to criteria that it beongs at the hierarchy. At the next step inear programming is used to convert the voting resuts to weights. Suppiers are cassified in four groups in this system. The proposed system can aso recommend to decision makers in automotive industry for each suppier cass when evauate current suppiers and seect new suppiers due to ISO/TS 16949 requirements. REFERENCES [1] U. Akinc, Seecting a set of vendors in a manufacturing environment, Journa of OperationsManagement, 11, 1993,107-122. [2] F. Roodhooft, J. Konings, Vendor seection and evauation an Activity Based Costing approach, European Journa of Operationa Research, 96(1), 1997, 97-102. [3] C.C. Li, Y.P. Fun, J.S. Hung, A new measure for suppier performance evauation, IIE Transactions on Operations Engineering, 29, 1997, 753-758. [4] L. Boer, L.Wegen, J.Tegen, Outranking methods in support of suppier seection. European Journa of Purchasing&Suppy Management, 4, 1998,109-118. [5] J. Liu, F.Y.Ding, V.La, Using data enveopment anaysis to compare suppiers for suppier seection and performance improvement. Suppy Chain Management:An Internationa Journa, 5(3), 2000, 143-150. [6] S. Tauri, J. Sarkis, A mode for strategic suppier seection, The Journa of Suppy Chain Management, 38 (1), 2002, 18-28. [7] C.C.Chen, C.C. Yang, Tota-costs based evauation system of suppier quaity performance", Tota Quaity Management, 14(3), 2003, 325-39. [8] R. Dumin, V. Mininno, Suppier seection using a muticriteria decision aid method, Journa of Purchasing and Suppy Management, 9, 2003, 177 187. [9] F.T.S. Chan, Interactive seection mode for suppier seection process: an anaytica hierarchy process approach, Internationa Journa of Production Research, 41, 2003, 3549 3579. [10] G. Wang, S. H. Huang, J. P. Dismukes, Product-driven suppy chain seection using integrated muti-criteria decision-making methodoogy, Internationa Journa of Production Economics, 91(1),2004, 1 15. [11] W.N. Pi, C. Low, Suppier evauation and seection via Taguchi oss functions and an AHP, Internationa Journa of Advanced Manufacturing Technoogy, 27, 2006, 625-630. [12] R. Lasch, C.G. Janker, Suppier seection and controing using mutivariate anaysis, Internationa Journa of Physica Distribution & Logistics Management,35(6), 2005, 409-425. [13] G.H., Hong, S. C. Park, D. S. Jang, H. M. Rho, An effective suppier seection method for constructing a competitive suppy reationship, Expert Systems with Appications, 28(4), 2005,629 639. [14] C.T. Chen, C. T. Lin, S. F. Huang, A fuzzy approach for suppier evauation and seection in suppy chain management, Internationa Journa of Production Economics, 102, 2006, 289-301. [15] M. Kumar, P. Vrat, R. Shankar, A fuzzy programming approach for vendor seection probem in a suppy chain, Internationa Journa of Production Economics, 101(2), 2006, 273-285. [16] S.M. Ordoobadi, Deveopment of a suppier seection mode using fuzzy ogic, Suppy Chain Management: An Internationa Journa, 14(4), 2009, 314 327. [17] A. Aksoy, N. Öztürk, Suppier seection and performance evauation in just-in-time production environments, Expert System with Appications, 38, 2011, 6351-6359. [18] H. Noguchi, M. Ogawa, H. Ishii, The appropriate tota ranking method using DEA for mutipe categorized purposes, Journa of Computationa and Appied Mathematics, 146, 2002, 155-166. [19] C.H. Han, J.K. Kim, S.H. Choi, Prioritizing engineering characteristics in quaity function depoyment with incompete information: a inear partia ordering approach, Internationa Journa of Production Economics, 91(3), 2004, 235-249. 2362 P age