AN INTEGRATED SYSTEMS APPROACH TO RISK MANAGEMENT WITHIN A TECHNOLOGY-DRIVEN INDUSTRY, USING THE DESIGN STRUCTURE MATRIX AND FUZZY LOGIC #

http://sjie.journls.c.z AN INTEGRATED SYSTEMS APPROACH TO RISK MANAGEMENT WITHIN A TECHNOLOGY-DRIVEN INDUSTRY, USING THE DESIGN STRUCTURE MATRIX AND FUZZY LOGIC # W.F. Brkhuizen 1, J.H.C. Pretorius 2 & L. Pretorius 3* 1 1, 2 University of Johnnesburg, South Afric 3 Grdute School of Technology Mngement University of Pretori, South Afric leon.pretorius@up.c.z ABSTRACT Risk interctions exist within system nd its sub-systems, between functionl nd physicl elements in vrious dimensions such s sptil interction, informtion exchnge, mteril trnsfer, nd energy exchnge. These interctions re of multi-dimensionl complexity, nd thus re not sufficiently interpreted using conventionl mngement tools. Alterntive system representtion nd nlysis techniques re proposed in prticulr the design structure mtrix (DSM) nd fuzzy logic thinking to quntify the risk mngement effort necessry to del with uncertin nd imprecise interctions. A cement grinding plnt cse study is used to elborte on the risk mngement methodology OPSOMMING Risiko-interksies bestn binne n stelsel en sy sub-stelsels, tussen funksionele en fisiese elemente. Hierdie interksies kn gekwntifiseer word in n ruimtelike, inligting-uitruiling, mteril-oordrg of energie-uitruiling rmwerk. Die interksies is vn n multidimensionele kompleksiteit, en word nie effektief geїnterpreteer deur middel vn konvensionele beheermetodes nie. Alterntiewe stelselvoorstelling- en nliseringstegnieke kn gebruik word om die sisteeminterksies te visuliseer. Die Ontwerp Struktuur Mtriks ( design structure mtrix ), en Wsige Logik ( Fuzzy Logic ) word ingespn om hierdie interksies voor te stel en eenvoudig te kwntifiseer. n Sementnleggevllestudie word gebruik om die risikobestuurmetodologie op die proef te stel. 1 * Corresponding uthor # This rticle is n extended version of pper presented t the 2011 ISEM conference. South Africn Journl of Industril Engineering, July 2012, Vol 23 (2): pp 202-214

http://sjie.journls.c.z 1. INTRODUCTION AND RESEARCH AIM In the old dys, the tools of frming, mnufcture, business mngement, nd communiction were simple. Brekdowns were frequent, but repirs could be mde without clling the plumber, the electricin, the computer scientist or the ccountnt nd the investment dvisors. Tody, however, the tools we use re complex, nd brekdowns cn be ctstrophic, with fr reching consequences. We must be constntly wre of the likelihood of mlfunctions nd errors. [1] For the technology-driven industry, innovtion occurs where mrket needs nd technology knowhow overlp [2]. Innovtion is the ct of introducing something new. [3] When compnies re competing on the technology plyground they need to be innovtive. According to Byrd & Brown [3], the ct of introducing reltes to risk tking, nd the new reltes to cretivity, nd therefore these concepts, cretivity nd risk tking in combintion, re wht innovtion is ll bout. Risk mngement hs become one of the gretest chllenges of the 21 st century [1, 4], nd one of the min components in innovtion nd the technology-driven industry, intensifying the need for systemtic pproch to mnging uncertinties. Innovtion ƒ (Cretivity x Risk Tking) [5] To think cretively we hve to relise tht we cn t solve problems by using the sme kind of thinking we used when we creted them (Einstein [6]). Being cretive nd tking risks is one of the wys tht innovtion cn be relised [5, 7]. Loclised nd rective risk mngement techniques will not be effective in tody s globlised high technology industry [8, 9]. During the development nd design of complex engineering products, the input nd temwork of multiple prticipnts from vrious bckgrounds re required, resulting in complex interctions [10]. Risk interctions exist between the functionl nd physicl elements within such system nd its sub-systems in vrious dimensions such s sptil interction, informtion interction, etc. The reltionships re of multi-dimensionl complexity tht cnnot be simplified using the stndrd mngement tools [11, 12]. Illustrting these risk interctions in complex system embodies the essence of the reserch im ddressed in this pper. The reserch method followed is explortory in nture, nd bcked up by cse study of complex system. To find meningful strting point for the seemingly boundless subject of risk mngement, the logicl pproch is to tke step bck into the bsic definition of risk mngement. Ech of the risk mngement processes (risk ssessment, risk identifiction, risk nlysis, risk evlution, risk tretment, nd risk monitoring nd review), nd how these processes cn be enhnced using the design structure mtrix (DSM) nd fuzzy logic thinking to ddress the uncertin, imprecise, nd multi-dimensionl nture of the interctions between system elements, were explored [13]. The pproch to enterprise risk mngement should be seen s holistic, similr to the totl qulity mngement process [1, 4, 14, 15, 17, 18], providing the opportunity to incorporte risk mngement during the design process s concurrent tsk [19]. The risk mngement model is developed concurrently (during the design phse) using product development methodologies such s conceptul modelling nd prototyping, where ultimtely the prototype is tested by mens of cse study of cement grinding section in cement production process. The result of the risk mngement model nd in essence the reserch presented in this pper is clustered DSM providing visul representtion of the system risk res similr to the methodology used in Finite Element Anlysis (FEA) [20, 12]. 203

http://sjie.journls.c.z 2. APPLICATION OF SYSTEMS ENGINEERING IN RISK MANAGEMENT Technicl components or systems do not operte in isoltion: they re, in generl, prt of lrger system [21]. For the system to function, it often involves humn interction through input effects (operting, controlling, etc.). The system responds to this interference nd provides feedbck in the form of effects or signls tht result in further ctions. A system cn therefore be influenced from outside by humn intervention, or by intervention from the environment in which it exists [17]. The intervention cn hve positive or negtive influence on the functionlity of the system, nd it cn trigger undesired side effects, from individul components within the sub-system or from the overll system itself [22]. The combined interreltionship of ll these effects hs to be crefully considered during the development of technicl systems. One of the wys to pproch complex problems is to study the underlying structure of the complex system. Systems thinking embrces holism nd cretivity to hndle complexity, chnge, nd diversity [23]. In the technology-driven industry, people re predominntly techniclly trined nd exhibit logicl, relistic, nd rtionl pproch. It still seems, however, tht people mnge risk intuitively, lrgely bsed on their pst experience [24]; but risk mngement is becoming too complex due to its multi-dimensionl nture. According to Vn Asset [5], decision-mking becomes complex when there is not single problem but n intertwined web of relted problems, when the decision or issue lies cross or t the intersection of mny disciplines (multi-dimensionl), or when the underlying process intercts on different scle levels. The fct tht consensus must be reched by group of experts to lower the uncertinty in risk sitution mens tht uncertinty contributes to mking complicted issues more complex s more nd more interfces (e.g. multiple experts) re creted [5]. Mking complex decision therefore lso involves mking n uncertin decision. The current tools nd methods for risk mngement seem to be indequte, given the complexity of the technology-driven industry [25]. Furthermore, the perceived need for systemtic pproch to risk mngement is highlighted in these seemingly unstructured existing methods of risk mngement. The unique pproch to risk mngement in technology-driven industry is similr to the nture of technology: it should incorporte structured but innovtive pproch to obtining stkeholder cceptnce. A common wy to gin understnding of complex system is to nlyse it (to mke sense of system by breking the system prt into its sub-systems) [26, 27, 23, 28]. Anlyses, however, focus on the elements of system in isoltion, nd therefore lose the reltions between prts. Systemic thinking combines nlysis (simplifying systems by tking them prt into less complicted sub-systems) nd synthesis (mking sense of system components by seeing how they fit together nd wht their reltions nd interctions re with other system components) [27]. Anlyticl thinking is used to identify the elements; syntheticl thinking is used to find the repeting pttern [27]. Figure 1 is visulistion of systemic thinking in the process plnt environment such s tht found in the cement production process cse study considered in the next section by breking plnt into its vrious sub-systems up to the component level. It lso indictes how components, res or units, nd equipment fit together in the plnt from system point of view. Adjustments to component or the functionlity of sub-system cnnot be restricted to the sub-system or component. An exmple provided by Phl & Beitz [22] is combined coupling (comprising flexible coupling nd clutch). As combined coupling it cn be regrded s system tht within mchine, or by joining two mchines, cn be regrded s n ssembly. This ssembly cn be considered s two sub-systems: flexible coupling nd clutch, which in turn cn be divided into system elements, in this cse components. It is lso possible to consider the functionl reltionship where the system coupling cn be divided into the sub-systems dmping nd clutching. 204

http://sjie.journls.c.z Figure 1: System visulistion (prtilly dopted from Tkshi et l. [30]) To nlyse the risk within such system, one would hve to nlyse the interctions nd reltionships between components, mchines, nd plnts within system, for both the subsystem functionlity nd the sub-system requirements. Eppinger & Pimmler [30] consider system reltionships long four dimensions, nd evlute the interction between two system elements bsed on the conversion of energy, mteril, nd informtion, nd lso their sptil orienttion. Strting with the fundmentl concepts of mtter nd force, one comes cross mtter in mny shpes nd forms; while force pplied to mtter ultimtely results in energy being trnsferred in different forms [22]. For the reserch presented in this pper, the interctions between components re considered long the following dimensions: Energy mechnicl, electricl, etc. Mteril with chrcteristics such s mss, structure, composition, etc. Informtion or signl exchnge. 3. DESIGN STRUCTURE MATRIX FOR RISK MANAGEMENT Throughout the engineering disciplines it is common prctice for engineers to solve complex problem by first breking it into set of smller problems tht re more esily hndled. However, the decomposition of complicted systems cn crete chllenges [30]: It might be difficult to brek up the system into suitble set of sub-systems. After decomposition it might be difficult to unite the vrious sub-systems into n overll system. To overcome these chllenges, n overll system functionlity or system requirement cn be decomposed into sub-system functionlities nd sub-system requirements. Similr to the wy in which designers estblish prticulr systems nd prticulr purposes by decomposing the system, the risks in system cn be decomposed into sub-risks of the sub-systems. This interction suggests tht risk in sub-system will interct with other sub-systems nd lso contribute to the risk in the totl system. With reference to Figure 1, this risk-reltionship cn be visulised by relising tht filure of one of the two sub-systems in either mechnicl construction ( flexible coupling or clutch ) or functionlity ( dmping or clutching ) will hve n impct on the mechnicl construction or functionlity of the system ( combined coupling ), nd in this cse the ssembly. By quntifiction of the reltionship the impct cn be estblished. A system cn, therefore, be decomposed into its vrious sub-systems, nd by identifiction nd quntifiction of risks within sub-systems (sub-risk), the system risk cn be quntified. 205

http://sjie.journls.c.z The methodology tht is used in this reserch to represent nd nlyse dependencies nd reltions between items is known s the design structure mtrix (DSM), nd ws introduced by Stewrd in 1967 [31] nd in 1981 [32]. These ppers re considered the origins of the DSM field. The mjor ide of Stewrd s pproch ws to hndle uncertinty in complex systems by exploring the structure of problem [25, 12]. Figure 2 illustrtes, from the current reserch, where the DSM field cn be used within the risk mngement process for risk identifiction, by identifying the reltionship between system elements. Figure 2: Design structure mtrix within the risk mngement process (prtilly dopted from Longfellows [33]) The introduction of the design structure mtrix into the risk identifiction process (s illustrted in Figure 2), nd mking use of the quntifiction suggested by Eppinger & Pimmler [30] to identify the interctions tht my occur between the functionl nd physicl elements, considers 1) ssocitions of physicl spce nd lignment, 2) ssocitions of energy exchnge, 3) ssocitions of informtion (signl & mesurement) exchnge, nd 4) ssocitions of mterils (process) exchnge. Figure 3: Interction between elements, represented s vector with four scores For the purposes of this reserch, the interction must be systemticlly identified to quntify nd mnge the risk of the complete system. Therefore, to identify, quntify, nd mnge the risks res of complete system, the risk res of the vrious components nd 206

http://sjie.journls.c.z elements within the system must be systemticlly identified nd quntified. Referring to Figure 3, the four generic interctions re defined s follows: Sptil: A sptil-type interction identifies the needs for djcency or orienttion between elements. Energy: An energy-type interction identifies the needs for energy trnsfer between two elements. Informtion: An informtion-type interction defines the needs for informtion, signl, or mesurement exchnge between two elements. Mteril: A mteril-type interction identifies needs for mterils exchnge between two elements. [30] The rnges of interctions used in this reserch re defined in Tble 1. Tble 1: Generl interction quntifiction scheme (prtilly dopted from Eppinger & Pimmler [30]) 4. FUZZY LOGIC FOR RISK MANAGEMENT Figure 4: Fuzzy logic for risk mngement (prtilly dopted from Longfellows [33]) 207

http://sjie.journls.c.z The interctions of risk between elements re obtined by interviewing vrious system experts. In the cse study introduced in the next section, the system experts were plnt nd process specilists. However, in evluting this interction it is difficult to provide cler-cut definition of wht interctions re HIGH, MEDIUM, or LOW ; such vgueness cn be ddressed in the fuzzy set theory [34, 35, 36]. All fuzzy rules pply t ll times, nd they pply in prllel. The fuzzy sets re converted to crisp output vlues by mens of process clled defuzzifiction [37] process to get non-fuzzy vlue tht best represents the possibility distribution of n inferred fuzzy control ction [38]. The selection of defuzzifiction procedure depends on the properties of the ppliction [39, 40, 41]. The weighted verge defuzzifiction provides n cceptble ccurcy with reltively simple mthemtics [42]. 5. CASE STUDY Figure 5: Weighted verge defuzzifiction [43] The lifeblood of technology-driven industry consists of n orgnistion s bility to develop new products nd to integrte risk mngement holisticlly, thus requiring risk mngement to be one of the concurrent tsks in the design process. By including risk mngement s concurrent tsk in the design process, it cretes the opportunity to develop the risk mngement model, following the sme pproch nd principles used for product development. By dopting this developmentl pproch, conceptul model nd prototype ws constructed s briefly described in the previous sections nd evluted using cse study. With focus on the cement grinding section of the cement process, nd for the purpose of the cse study, the following sub-systems of the cement grinding process were defined: Clinker trnsport Additives trnsport Finished product trnsport Recircultion trnsport Grits trnsport Mill drive Mill feeding Seprtor (nd cyclones) System filter Hot fuel oil supply Hot gs genertor Mill hydrulic Compressed ir / cooling wter 208

http://sjie.journls.c.z The reltionship nd interctions between elements of the model referred to in the previous sections re obtined by interviewing the vrious system experts. In the cse study, smll group of five plnt experts were sked to complete survey nd provide quntifiction of the interction between plnt components, bsed on their experience nd the plnt process. The survey used for the cse study, nd the results obtined from the survey, form prt of doctorl study presented t the ISEM conference in 2011 [44, 45]. After dt collection nd mpping of the interdependencies into the design structure mtrix on the sptil, energy, mteril, nd informtion levels, fuzzy logic is used to determine the overll interdependence between elements by pplying the fuzzy rules chosen. A simplified version of the fuzzy controller is presented below for discussion purposes. Chosen fuzzy rules: Rule 1: IF (Sptil is required) OR (Informtion is required) OR (Energy is required) OR (Mteril is required) THEN Risk Mngement Effort = HIGH Rule 2: IF (Sptil is desired) OR (Energy is desired) OR (Informtion is desired) OR (Mteril is desired) THEN Risk Mngement Effort = MEDIUM Rule 3: IF (Sptil is indifferent) OR (Energy is indifferent) OR (Informtion is indifferent) OR (Mteril is indifferent) THEN Risk Mngement Effort = LOW The following exmple vlues re chosen to illustrte the ppliction of the fuzzy rules nd the fuzzy controller output: (Sptil, Informtion, Mteril, Energy) = (5,7,2,1) Figure 6: Appliction of rules in the fuzzy controller For simplicity, the output set of the fuzzy controller is defined s risk mngement effort tht cn be LOW, MEDIUM, or HIGH. In Figure 6 the evlution of Rule 1, Rule 2, nd Rule 3 provides the following: Rule 1 Input is MAX(0,0.25,0,0,0) = 0.25 nd Output is 6 (High) Rule 2 Input is MAX(0,0.25,0,0,0,0,0.5,0,0) = 0.5 nd Output is 4 (Medium) Rule 3 Input is MAX(0,0,0,0,0.667,0.333,0) = 0.667 nd Output is 2 (Low) The output cn be defuzzified using the weighted verge formul illustrted in Figure 7: Risk Mngement Effort = ((0.25x6)+(0.5x4)+(0.667 x 2)) / (0.25+0.5+0.667) = 3.41 209

http://sjie.journls.c.z Figure 7: Fuzzy controller output (risk effort) Tble 2: Reltionship rting Numeric scle Mening 0 Zero risk mngement effort required 2 Low risk mngement effort required 4 Medium risk mngement effort required 6 High risk mngement effort required 8 Very high risk mngement effort required The reltionship rting for 3.41 clssifies the risk mngement effort requirement s lying between Low nd Medium (s defined in Tble 2). Figure 8: Fuzzy controller output vlues represented in DSM The resulting DSM from the defuzzifiction process (represented in Figure 8) will be single mtrix contining the combined weighted contributions (resulting from the fuzzy rules) from the four input DSMs (contining the components for sptil, informtion, mteril, nd energy). The DSM cn be mnipulted using DSM tools such s clustering [11] to rerrnge the DSM elements to obtin clusters of highly intercting components while minimising the inter-cluster interctions. During the clustering process, the dt is not modified: mtrix rows nd columns re only swpped pir-wise (therefore lso keeping the interreltionship between DSM elements) to obtin different mtrix lyout. The new 210

http://sjie.journls.c.z groupings or clusters represent reorgnised frmework of the product rchitecture, visully showing the system elements tht hve the highest interction. The Cmbridge Advnced Modeller by the Engineering Design Centre of the University of Cmbridge is used in the reserch for DSM nlysis nd clustering (Refer to Figure 9 nd 10) [49, 50]. Figure 9: Output Mtrix (CAM Representtion). Figure 10: Clustered Output Mtrix When the DSM elements represent design components (i.e. component-bsed DSM) the purpose of the mtrix mnipultion becomes the finding of comprtments of DSM elements (i.e. clusters or modules) tht re mutully exclusive or mrginlly intercting subsets tht is, clusters s groups of elements tht re interconnected mong themselves to n importnt extent while being little connected to the rest of the system [46, 47]. In the risk mngement domin, clustering of the system highlights risk res in the system. Clustering of the mtrix is similr to the ppliction of Finite Element Method (FEM) or Finite Element Anlysis (FEA) where detiled visulistion of bending or twisting in structures is provided 211

http://sjie.journls.c.z [48]. The clustering provides visulistion of the high interction res within system, nd thereby risk stresses (Refer to Figure 11). 6. CONCLUSION 212 Figure 11: Risk mngement effort Technology-driven industry cn be very complex, due to its multi-dimensionl chrcteristics nd concurrently running tsks. One of the most efficient wys to pproch complex problems is to follow systemtic pproch nd study the underlying structure of the complex system. In most instnces, however, the nlysis of system (breking the system prt) focuses on the elements of the system in isoltion, nd the reltions between sub-systems re lost. This seprtion between sub-systems during nlysis requires systemic pproch tht will combine the nlysis of the system with synthesis of the system. This therefore suggests methodology where the complex system is broken into meningful sub-systems without losing reltions between the sub-systems. An evlution of the vrious risk identifiction nd nlysis techniques highlighted lck in their bility to ccount for dependencies nd reltionships between system components. The methods tht consider the reltionships between system components (e.g. FMEA) re cumbersome. Tools nd techniques re required for system decomposition nd integrtion. The DSM instrument ws used to model the system nd visulise the risk reltionship between system components in vrious dimensions (sptil, mteril, informtion, energy). The DSM representtion of the system llowed for systemic interprettion of the system by breking the system into its sub-systems, while still keeping the reltionship between the sub-systems, thereby estblishing one of the key contributions of the reserch. Risk mngement involves fctors tht defy clssifiction into crisp sets, nd the quntifiction of risk vries mong different individuls nd groups, bsed on their perception nd experience. In the cse study it ws shown successfully tht fuzzy logic cn define vlues between the conventionl digitl logic ( ON nd OFF or 1 nd 0 ) providing more humn wy of thinking. The input to the decision-mking logic (the fuzzy logic rules) is bsed on humn expert knowledge bse, nd mimics how experts solve problems. It lso mkes fuzzy logic controller fit comfortbly into the risk nlysis process. The reserch presented here successfully integrtes lterntive system representtion nd nlysis techniques [51]. In prticulr, the design structure mtrix nd fuzzy logic quntify

http://sjie.journls.c.z the risk mngement effort necessry to del with uncertin nd imprecise interctions between system elements. As further nd finl contribution, the mtrix formt of the clustered DSM provides grphicl overview of the system risk, highlighting high risk stress res similr to the methodology used in finite element nlysis (FEA). The pproch outlined in this pper provides plusible pproch to mnging risk in technology-driven industry. Further opportunity for refinement of the method is recommended through reserch. Prticulr res of interest re the introduction of fuzzy logic controller feedbck to incorporte risk tretment nd monitoring, ppliction to project risk mngement, nd expnding the DSM to Domin Mpping Mtrix (DMM). REFERENCES [1] Bernstein, P.L. 1998. Aginst the Gods: The remrkble story of risk, John Wiley & Sons. [2] Bermn, S.J. & Hgn, J. 2006. How technology-driven business strtegy cn spur innovtion nd growth, Strtegy nd Ledership, Vol. 34 (2). [3] Byrd, J & Brown, P.J. 2003. The innovtion eqution: Building cretivity nd risk tking in your orgniztion, Wiley. [4] Bieri, S. 2001. Disster risk mngement nd the systems pproch. [Cited June 20, 2010.] http://www.drmonline.net/drmlibrry/systems.htm. [5] Vn Asselt, M.B.A. 2000. Perspectives on uncertinty nd risk, Kluwer. [6] Hrris, K. 1995. Collected quotes from Albert Einstein. [Cited August 27, 2010.] http://rescomp.stnford.edu/~cheshire/einsteinquotes.html. [7] Anders, G. Who knew? Some of the predictions we mde decde go were wy off, The Wll Street Journl, 9. [Cited Februry 20, 2010.] [8] Adms, J. Risk Mngement: It s not rocket science... It s much more complicted. [Online]. [Cited Februry 8, 2010.] http://www.rmmg.com/mgtemplte.cfm?section=mgarchive&nvmenuid=304 &templte=/mgzine/displymgzines.cfm&aid=3330&showarticle=1. [9] Winzker, D.H. 2005. A holistic mngement model for the trnsformtion of high technology engineering compnies for sustined vlue cretion nd globl competitiveness, Johnnesburg: University of Johnnesburg. [10] Bermn, S.J. & Hgn, J. 2006. How technology-driven business strtegy cn spur innovtion nd growth, Strtegy nd Ledership, Vol. 34 (2). [11] Yssine, A.A. 2004. An Introduction to modeling nd nlyzing complex product development processes using the design structure mtrix (DSM) method. [Online] [Cited: 07 28, 2010.] http://users.ipfw.edu/reddpv01/dsmtutoril.pdf. [12] Dnilovic, M. & Sndkull, B. 2005. The use of dependence structure mtrix nd domin mpping mtrix in mnging uncertinty in multiple project situtions, Interntionl Journl of Project Mngement, Vol. 23, pp. 193-203. [13] Cooper, D.R. & Schindler, P.S. 2003. Business reserch methods, 8 th ed., McGrw-Hill Irwin. [14] Childers, S.R. & Long, J.E. 2000. A concurrent methodology for the systems engineering design process. [Cited August 3, 2010.] http://www.vitechcorp.com/ whiteppers/files/ 200611131842060.Long_Childers_1994.pdf. [15] Conrow, E.H. 2004. Effective risk mngement, some keys to success, 2 nd ed., Americn Institute of Aeronutics nd Astronutics. [16] Gorrod, M. 2004. Risk mngement systems process, technology nd trends, Plgrven Mcmilln. [17] Jmes, M. 1995. Risk mngement in civil, mechnicl nd structurl engineering, Proceedings of the conference orgnized by Helth & Sfety Executive in co-opertion with the Institution of Civil Engineers, held in London on 22 Februry 1995. [18] Tippins, S.C. 2004. Risk mngement: Where is it nd where does it belong?, Risk Mngement, Vol. 6 (3), pp. 9-11. [19] Kusik, A. 1999. Engineering design: Products, processes nd systems, Acdemic Press. [20] Gillow, K. 2005. A finite element method tutoril, Computtionl Biology Group: Finite Element Method Tutoril. [21] Henderson, R.M. & Clrk, K.B. 1990. Architecturl innovtion : The reconfigurtion of existing product technologies nd the filure of estblished firms, FORM Globl, Vol. 35, pp. 9-15. [22] Phl, G. & Beitz, W. 1996. Engineering design: A systemtic pproch (edited by Ken Wllce nd trnslted by Ken Wllce, Lucienne Blessing, nd Frnk Buert), 2 nd ed., Springer. [23] Sge, A.P. & Rouse, W.B. 1999. Hndbook of systems engineering nd mngement, John Wiley & Sons. [24] Lindemnn, P. 2009. The design structure mtrix (DSM) http://www.dsmweb.org/. [Cited October 15, 2011.] 213

http://sjie.journls.c.z [25] Dnilovic, M. & Browning, T.R. 2007. Mnging Complex product development projects with design structure mtrices nd domin mpping mtrices, Interntionl Journl of Project Mngement, Vol. 25 (2007) pp. 300-314, Elsevier. [26] Jckson, M.C. 2004. Systems Thinking, John Wiley & Sons. [27] Wegmnn, A. 2010. Systemic thinking. [Cited September 20, 2010.] http://lmspeople.epfl.ch/blbko/professionl/systemic_thinking/index.htm. [28] Li, C., Pretorius, L. & Nel, A.L. 2006. Identifying risks by using systems engineering pproch: A technology mngement cse study, IAMOT. [29] Tkhshi, J., Node, T., Htt, N. & Fuji, A. 2003. Appliction of risk bsed mintennce to life nd finncil ssessment t cement plnts, Engineering Review, Vol. 36 pp.52-58. [30] Eppinger, S.D. & Pimmler, T.U. 1994. Integrtion nlysis of product decompositions, Minnepolis, 1994. [31] Stewrd, D. 1967. The design structure systems, Generl Electric Report No. 67APE6, Sn Jose. [32] Stewrd, D. 1981. The design structure system: A method for mnging the design of complex systems, Trnsctions on Engineering Mngement, Vol. EM-28 (3), pp. 71-74. [33] Longfellows. 2008. http://www.longfellows.com.u/imges/risk_mngement_ digrm.jpg. [Cited June 11, 2010.] [34] Verworn, B. & Hersttt, C. 2001. Mnging the fuzzy front end of innovtion, Working Pper, Deprtment of Technology nd Innovtion Mngement, Technicl University of Hmburg. [35] Cheung, W.W.L., Pitcher, T.J. & Puly, D. 2005. A fuzzy logic expert system to estimte intrinsic extinction vulnerbilities of mrine fishes to fishing, New Reserch nd Expert Systems, Vol. 124 pp. 97-111, [36] Mckone, T.E & Deshpnde, A.W. 2004. Cn fuzzy logic bring complex problems into focus? Modeling imprecise fctors in environmentl policy, Lwrence Berkeley Ntionl Librry. [37] Zdeh, Lofti A. 1965. Fuzzy sets, informtion nd control, Deprtment of Electricl Engineering nd Electronic Reserch Lbortory, University of Cliforni, Vol.8, pp. 338-353, Berkeley. [38] Nissen, V. 2007. Mngement pplictions of fuzzy control, 52 nd Interntionles Wissenschftliches Kolloquium, Technische Universität Ilmenu. [39] Simoes, M.G. 2010. Introduction to fuzzy control, Colordo School of Mines - Engineering Division. [40] Brulé, J.F. 1985. Fuzzy systems - A tutoril. http://www.ustinlinks.com/fuzzy/ tutoril.html. [41] Kosko, B. 1993. Fuzzy thinking: The new science of fuzzy logic. http://www.intelligentsystems.info/clsses/ee509.9.pdf. [Cited My 2, 2011.] [42] Zdeh, L.A. 1968. Fuzzy lgorithms, informtion nd control, Deprtment of Electricl Engineering nd Electronic Reserch Lbortory, University of Cliforni, Vol.12, pp. 94-102, Berkeley. [43] Ross, T.J. 2010. Fuzzy logic with engineering pplictions, 3 rd ed., Wiley. [44] Brkhuizen, W.F., Pretorius, L. & Pretorius, J.H.C. 2011. An integrted systems pproch to risk mngement within technology driven industry using the Design Structure Mtrix nd Fuzzy Logic, University of Johnnesburg. [45] Brkhuizen, W.F, Pretorius, L. & Pretorius, J.H.C. 2011. An pproch to innovtion in risk systems, ISEM 2011 Proceedings. [46] Browning, T.R. 2001. Applying the design structure mtrix to system decomposition nd integrtion problems: A review nd new directions, IEEE, Trnsctions on Engineering Mngement, Vol. 48. [47] Arvind, H., Rjgopl, C. & Somn, K.P. 2010. A simple pproch to clustering in excel, Interntionl Journl of Computer Applictions, Vol.11, pp. 19-25. [48] Eriksson, K., Estep, D., Hnsbo, P. & Johnson, C. 1995. Introduction to dptive methods for differentil equtions, Act Numeric, pp. 105-158. [49] University of Cmbridge. 2011. Cmbridge Advnced Modeller (CAM). [50] Biedermnn, W., Browning, T., Eppinger, S., Kreimeyer, M. & Murer, M. 2009. DSMweb. http://www.dsmweb.org/. [Cited August 1, 2011.] [51] Wrfield, D., 2005. IS/IT reserch: A reserch methodologies review, Journl of Theoreticl nd Applied Informtion Technology, pp. 28-35. 214