Production Planning Model and Algorithm of Dynamic Mass Customization System

Transcription

1 Producton Plannng Model and Algorthm of ynamc Mass Customzaton ystem Yu Wang *, Fang L, Jjun Yuan. Internatonal Busness chool, Jnan Unversty, Guangzhou,063 Chna.;. chool of Informaton cence, Guangdong Unversty of Busness tudes, Guangzhou, 030,Chna Abstract Based on the poston of a sngle COP, the producton plannng of mass customzaton s dvded nto two stages, mae-to-stoc producton before COP and mae-to-order producton after COP. However, n the case of multple COP, producton plannng of mass customzaton can be much more complex. hs paper presents a dynamc mass customzaton system, that s, a three-stage producton plannng model h multple COP. he three stages nclude push plannng, pushpull plannng and pull plannng. An optmzaton model for producton plannng s provded n the aspects of producton capacty, nventory constrants and stochastc demand for smulaton calculaton. he optmzaton model emphaszes on the analyss of the producton plannng durng the perod when push and pull producton coexsts. Its producton volume s consdered as the combnaton of pull producton part (customer demands) and push producton part (predcted producton volume). Moreover, ts producton plannng scheme s the combnaton of two producton schedules. Lastly, the feasblty of the proposed model s tested usng the producton plannng of turbnes. Keywords - dynamc mass customzaton; customer order decouplng pont; producton plannng; smulaton I. INROUCION Mass Customzaton s a man producton pattern of manufacturng enterprses n the st century. Mass customzaton (MC) enterprses should satsfy varous needs of ther customers. One of the ey problems s how to mae approprate producton plans accordng to customzaton requrements and respond to customers needs hout delay as well. Based on the poston of a sngle COP, the producton plannng of mass customzaton s dvded nto two stages, that s, push stage before COP and pull stage after COP. tudes show that the push stage can be mplemented steadly accordng to the reasonable plans made prevously under relatvely stable envronment. he mplementaton of pull stage depends on maret requrements or order quantty. However, n real producton systems of mass customzaton, there are always more than one COP. For nstance, Boeng Commercal Arplane Company has three COP. Multple COP can mprove the aglty of supply chan by segmentng product value chan nto a group of refned agle system. Wth the development of enterprses, sngle COP supply chan can no longer meet the demand of MC enterprses. he correlaton studes on COP are generally summarzed as follows. A sngle COP n the producton process s wdely and thoroughly dscussed. Bucln (96) connected a sngle COP structure to the postponement strategy. he poston of COP corresponds to dfferent manufacturng strateges: engneer-to-order (EO); mae-to-order (MO); and assemble-to-order (AO) (Hoestra and Romme 99, Browne et al. 996, Hggns et al. 996, acett et al. 997, Mather 999, Wortmann et al. 0). Hll (99) extended ths by addng desgn-to-order and mae-to-prnt. Upstream the COP, the raw materals are processed accordng to forecast, whle n the downstream, the raw materals are processed accordng to customer order. herefore, the COP concdes h the most mportant stoc pont, from where the customer order process starts (Jan Olhager 00). Verdouw et al (006) addressed four man underlyng factors of COP and proposed the demand-drven mult-cop chan networ model. He successfully llustrated the exstence of mult-cop n dary ndustres chan networ. Garca-astugue and Lambert (007) mentoned mult-cop n the supply chan where tme-based postponement was used. un et al. (008) proposed that mult-cop exst n sophstcate producton. Wang (008) establshed supply chan model based on mult-cop n mass dynamc customzaton. Wang (00) accomplshed a seres of correspondng studes on multple COP. Phlpet et al. (0) proposed the COP exstence n MC supply chan and elaborated the dynamc feature of COP. Arnab Banerjee et al. (0) proposed the mult-cop paradgm n global supply chan. All these relatve researches show that multple COP n supply chan s not only acceptable but also necessary. However, many authors came up h * Correspondng author. Emal: twygs@jnu.edu.cn Address:No.60 Huangpu Avenue West Road, Guangzhou,Chna.P.C.063 OI 0.03/IJ.a IN: 73-80x onlne, prnt

2 producton questons for multple COP from dfferent perspectves such as defnton, poston, relaton etc. But there are few studes explored producton plans deeply for multple COP of mass customzaton. In more advanced MC model, how to mae approprate producton plans whch can meet customzaton requrements h hgh supply rate s sgnfcant. After analyzng the exstng problems of MC s producton plannng based on sngle COP, ths paper descrbes a mult-stage producton plannng model based on multple COP. In addton, ths paper bulds and solves the optmzaton model of economc producton lot n push-stage and push-pull ntegrated stage respectvely. Fnally, ts feasblty s tested usng the producton plannng of turbnes. II. PROUCION PLANNING MOEL WIH MULIPLE COP OF MA CUOMIZAION he term of producton plannng mode of MC based on multple COP s proposed n order to mae up for the shortage of tradtonal producton plannng mode of MC (producton plannng mode of a sngle COP). ome scholars analyze the customzaton demands deeply from the vew of optmzng producton operaton, and they suggest settng multple COP n MC supply chan. In the same producton system, dfferent customzaton attrbute combnatons correspond to dfferent COP so as to set dfferent delvery date for customzaton clents. Unle postponement strategy, the proposed dea maes customers choose producton customzaton as early as possble so as to meet the depth of customzaton. hen try to move the producton COP of a large number of components to clents so that these customzed products can be produced n mass producton mode as much as possble. Lastly, the personalzaton s realzed. herefore, customers h specal customzaton requrements have to pay more and wat longer, whle others pay less and wat shorter. Because busness actvtes are drven by customers needs, producton plannng, and control actvtes, the greatest challenge of multple COP producton model s producton plannng and control model. herefore, we have to put forward a new producton plannng and control model whch adapts to the multple COP producton model. he COP postonng plays an mportant role n maret-orented strategy and producton plannng of supply chan. Under the sngle COP condton, the upstream plannng of COP separaton s based on forecast, whle the downstream plannng s based on mae-to-order. It s a two-stage producton. As shown n Fg., the producton chan of mass customzaton s lnear chan structure and the producton plannng model of a sngle COP can be descrbed. he,,...j...n n Fg. and represent nodes and manufacturng ponts n supply chan. he dashed lne represents the COP locaton. COP... j n MAKE-O-OCK PROUCION FLOW IRECION MAKE-O-ORER AGE AGE Fg. wo-tage Producton Model Of a ngle COP he producton model of a sngle COP has two stages: push-producton before COP and pull-producton after COP. here s a buffer stoc on COP. But n the producton model of multple COP, there are more than one COP. At some nodes of supply chan, the producton mode can be push-producton sometmes and pullproducton at other tmes. In practce, the sch and mplementaton of producton modes are realzed by producton plannng and nstructon. In fact the producers who are guded by plannng and nstructon can not exactly understand the current producton mode of he optmzaton of the sch among COPs embodes the formulaton and control of plannng or nstructon, that s, the desgn and control of producton plannng and the remedal measure when plan fals. Pullproducton mode can formulate producton plannng based on the order requests (tme and amount), producton capacty and BOM charts. o once a clent maes an order, a precse producton plan can be formed h strct tme requrement. Yet push-producton mode maes producton plannng accordng to the forecast, so dfferent customzaton combnatons need to be forecasted and the predctve value of standardzaton modules n pushproducton mode can be acqured by the comparson and decomposton of BOM chart. he producton plannng of every node s formulated basng on nventory management polcy. In the supply chan of multple COP, the operaton plannng of certan nodes may be the combnaton of these two nds. hus, the producton plannng of MC s much more complex when there are multple COP. ang all these features nto consderaton, the mult-stage producton plannng model h multple COP s llustrated n Fg.. COP COP COP n MAKE-O-OCK Mae-to-stoc Push producton PROUCION FLOW IRECION Mxed producton Push-pull producton MAKE-O-ORER Mae-to-order Pull producton Fg. Mult-tage Producton Plannng Model of Multple COPs As shown n Fg., the producton plannng of multple COP can be dvded nto three stages accordng to the combnaton of customzaton attrbutes. he delvery tme of COP whch s nearest to customers s the shortest. he delvery tme of COP whch s farthest from customers s the longest. herefore, the schedule of producton C U O M E R C U O M E R OI 0.03/IJ.a IN: 73-80x onlne, prnt

3 plannng splts nto three stages as well. he stage before COP whch s farthest from customers s n pushproducton mode. In ths stage producton forecast s done by MP. he most commonly used predcton method of enterprses, such as statstcal analyss, consders the demands as contnuous unform dstrbuted n order to formulate the sem-fnshed products processng plan, or sometmes consders the demands as normal dstrbuted n order to solve reproducton ponts problem. he stage after COP whch s nearest to customers s n pull-producton mode. In ths stage the producton actvtes are arranged accordng to orders. he producton mode between above two stages s push-pull. Its producton plannng s the sum of push and pull. he major dfference between multple COPs and a sngle COP s the push-pull producton plannng n the mddle stage of supply chan. Based on Fgure, the followng text llustrates how to calculate the producton volume of the worng pont at each stage. In Fgure 3, the dashed lne represents the processng of push mode, whle the sold lne represents the processng of pull mode. For each worng pont, producton mode determnes how t arranges the producton plan. In tage I and tage III, the worng ponts only conduct a sngle mode of producton: push producton n tage I and pull producton n tage III. In tage II, the push producton and pull producton mode coexsts n the worng ponts j+ to. here are n COPs n total n ths processng flow, each of them corresponds to a product mx and s dstrbuted among worng ponts where the correspondng sem-fnshed products are ept. For each worng pont n tage II, the combnaton of push- and pull-mode producton s adopted. hus we can decompose the producton nto pull producton part and push producton part accordng to the COPs. ae the worng pont m for example, n Fgure 3, we can see that the vertcal lne L through m ntersects h the horzontal lnes through COPs. For COP, lne L ntersects the sold lne across COP, representng that the processng correspondng to COP at worng pont m s pull producton and the producton volume s determned by actual order amount for PM. It s smlar for COP to M-. As for COP M, Lne L ntersects h the dashed part of the lne across COP M. herefore the processng of the correspondng PM M at worng pont m s push producton and the producton volume depends on the predcton of demand of PM M. hs case also suts COP M to n. Fg. he Producton Plannng Model of ecomposng COPs he followng text analyzes the producton plannng of dfferent worng ponts n tage I and tage II. For the worng ponts n tage I, the producton volume s the total predcted amount of all product mxes. For the worng ponts n tage II, the volume wll be fully dscussed. Whle for tage III of pull producton, the volume s the overall volume requred by customer for all product mxes, whch won t be dscussed n ths paper. III. PUH PROUCION PLANNING OPIMIZAION ALGORIHM AN MOEL Based on the precedng analyss, n the whole producton process, there wll be a certan amount of sem-fnshed products n the exact place where COP s. he followng text wll dscuss how to arrange push mode producton and nventory strategy of sem-fnshed products, n order to gan economc beneft of large-scale producton. Inventory cost, producton start-up cost, producton cost and shortage cost wll be taen nto consderaton. For stochastc demand, set up objectve functon on the target of mnmum total cost, then establsh constrant functons for producton and nventory capacty accordng to customzed products demand, at last fgure out the optmal producton quanttes of sem-fnshed A. Assumptons Assumpton: assume that customzed products n the supply chan go through all n processng (nodes), and the producton capacty of each node s lmted. Assumpton: assume that the amount of customzed products s proportonal to the amount of requred semfnshed products, and the producton of sem-fnshed products s drven by the demand of a certan class of customzed products; assume the ntervals of order arrval tmes of sem- fnshed products are mutually ndependent and have the same dstrbuton, subjectng to negatve exponental dstrbuton of parameterλ, and the dstrbuton functon F(Δt) s: t F( t) e λ s the average arrval rate of orders. Assume that each order demand of the correspondng sem-fnshed products s mutually ndependent and has the same dstrbuton, approxmately subjects to N (, ), and the dstrbuton functon s: G( ) = e y ( y ) dy In real producton system, demand s not an arbtrary value between(-,+ ), generally [mn,max ]. hs lmt wll be handled n the smulaton process. Assumpton3: the producton lead tme L s ndependent and has the same dstrbuton. he tme of processng a unt of sem-fnshed product s a, so the mean tme of processng unts sem-fnshed products s a. Because the length of lead tme s nfluenced by OI 0.03/IJ.a IN: 73-80x onlne, prnt

4 varous factors and there wll be a certan degree of fluctuaton, assume t subject to N(β, φ ), and L [L, L ], β=a. Its dstrbuton functon s: mn max G( ) = e y ( y ) dy Assumpton: Inventory shortage s allowed. he delay cost caused by shortage s part of shortage cost. Assumpton : the (s, ) nventory strategy s appled, n whch the safety stoc s can be set accordng to the actual stuaton. he replenshment at reproducton pont s carred out at the begnnng of each cycle. Inventory strategy parameters s and are nteger tmes of a specfed nventory resoluton. Assumpton6: there s a producton start-up cost n each producton cycle of sem-fnshed em products are used to produce customzed products, so producton set up cost of sem-fnshed products s counted as producton cost of customzed Assumpton7: he nventory cost of sem products s ncluded n the total cost. he total cost C of customzed products ncludes producton cost C, nventory cost C, shortage cost HC, and producton start-up cost K. he producton cost C ncludes raw materal cost and processng cost. ransportaton cost s neglgble. B. Modelng In a sngle COP push producton plan, assume there are nds of customzed products, and such products s plannng cycle,and t s plannng perod, have tme penalty of deferred producton. he total cost model of the nds of sem-fnshed product s mn{ ( X) t I t t s t h t max{ 0, * t X F t t ( ) {( )*( t w w w ( t X t I t t l t )} K} t fnsh delver w w w () ) s the producton cost of nds of sem-fnshed products; h t s the nventory cost of nds of semfnshed t products; max{0, * w {( w w t )*( t fnsh l t delver s the shortage cost of nds of sem-fnshed products; K s the order processng and producton preparaton cost of nds of customzed t X t F s producton capacty constrant; s t ) ABLE. HE EXPLANAION OF HE VARIABLE t [ 0, ] : the unt producton cost of nds of sem-fnshed t :, the producton amount of nds of sem-fnshed products n t perod. X t :, whether the nds of sem-fnshed products can be put nto producton n t perod, X t when postve, X t 0 otherwse. : other constant expenses of nds of sem-fnshed t : nventory amount of nds of sem-fnshed products n t perod. h : unt nventory cost of nds of sem-fnshed : unt tme penalty of deferred producton of nds of sem-fnshed w : In cycle, for nds of sem-fnshed products, the coeffcent of half amount of w customzed products on the demand. ( s nventory t w constrant. he varables n the model are shown n table:. w : In cycle, for the nds of sem-fnshed products, the amount of w customzed products on the demand. w w :- perod, nventory surplus of nds of semfnshed ( ) t : the producton completon perod of nds of semfnshed fnsh l : producton lead tme of nds of sem-fnshed t delver : delvery tme of nds of sem-fnshed K : order processng and producton preparaton cost of nds of sem-fnshed : unt capacty (tme) of the producton of nds of semfnshed : constant expense capacty of nds of sem-fnshed F : a total producton capacty on cycle )} OI 0.03/IJ.a IN: 73-80x onlne, prnt

5 Optmal batch producton t wll be dscussed on two cases: Case : no shortage case., t w w namely, () expresson does not nclude max{0, * w {( w w t )*( t fnsh l t w delver ( ), and )} second constrant t ( ) t w on w w expresson () s substtuted nto () to form expresson (3) mn{( h )* (3) t X t * h h * K } ( ) w w w (3) shows that the smaller t s, the better the result s, the bgger s, the better the result s. w w w herefore, n no shortage case, t w t t w, the mnmum total cost s w w t ( ) mn C * X K. Case : shortage case. 0, then mn{ X *[ t mn{[ *( t t t t ( )*( t w w t w l t fnsh delver t )]* X *[ t w t t l t fnsh delver )] K} ( )]*( t w w w l t ) K} fnsh delver () Because of shortage, ww t. ( ) w hen t can be splt nto two stuatons: unt producton cost s not less than, or less than punshment cost coeffcent. he optmal batch producton n these two stuatons s: *( t fnsh l t ) F X delver t t l () *( t fnsh l tdelver ) t mn In ths model, the demands are contnuous. emands arrval tmes, demands amount and lead tmes are consdered as random varables. For such a complex model, the research operaton of achvng the optmal producton batch * s mpractcal. o we use computer smulaton to solve the optmzaton model. C. mulaton oluton escrpton tep : generate smulated surplus nventory ( ) on former cycle -. a) Randomly generate producton amount ( ) [, mn max ] on cycle -. b) If t satsfes nequaton X F, contnue, otherwse return a). c) Call demand module of customzed products, randomly generate ' * w random varables w (when demand s greater than ', t wll be out of stoc). hen w t t ( ) = w ( ) w = w= tep : generate smulated state on cycle. a) mae j 0, C ( j) nfnty. b) Randomly generate producton amount [ mn,max ] on cycle. c) If t satsfes nequaton X F, contnue, otherwse return b). d) Call module of generatng random varables to randomly generate producton lead tme at the end of cycle. e) Call module of generatng random varables to randomly generate ' random varables t j whch satsfy: + ' j= t > l t l j ( ) ' j= j t t ( ) f) Call module of customzed products demand to randomly generate ' random varables w, w t t )f t w w ( ) When t = w= = w= w t t = w w ( ) the mnmum total cost s C * t X t K F X t Otherwse execute ) l ) If *( t fnsh l tdelver ) and, the total cost s mnmum. Otherwse, when t mn the total cost s mnmum.. Push-pull Producton Plannng Model h multple COP COP (=,,,n) corresponds to product mx (PM ). he worng ponts upstream COP conduct push producton, h predcted demand amount. t t OI 0.03/IJ.a IN: 73-80x onlne, prnt

6 hrough the precedng economc lot producton model, the optmal producton batch s * and the optmal producton cycle s *. herefore, assume the producton volume of sem-fnshed products for PM by the worng ponts upstream the COP s: n = P m = t he worng ponts downstream COP conduct pull producton. At COP, the common sem-fnshed products of PM are stored. When the order for any product of PM arrves, the worng ponts downstream the COP contnue processng. herefore durng cycle, the producton volume depends on the actual demand, denoted by Z t. It s obvous that the relatve poston of worng ponts n tage II and COPs s dfferent from each other. hus, the total producton volumes of worng ponts are the combnaton of the pull producton volumes and the push producton volumes, but each of them s varous. From the above analyss, durng cycle the producton volume of worng ponts to j s, and that of worng ponts + to n s n Z n. For worng ponts n tage II, tang the worng ponts j+ for example, snce t s behnd COP and ahead of COP to n, ts producton volume should be the combnaton of pull producton volume of PM and push producton volumes of PM to n, that s Z n. Based on the economc lot model of sem-fnshed products, the followngs present a general model of producton volume calculaton of the worng ponts n mult-cop MC producton system. () enote the node n tage I (ncludng node to j) as m (m =,,,j), thus n = P m = t P s the productve tas of m m. () enote the node n tage III (ncludng node + to m) as m 3 (m 3 =+,,m), thus Pm Z 3 Z... Zn Pm s the productve tas of 3 m 3. Z s the actual demand for customzed product. (3) enote the node n tage II (ncludng node j+ to ) as m (m =j+,,) and assume m s postoned just before COP M, thus P m t n Z t j P s the producton tas of m m. IV. APPLICAION EXAMPLE team turbne s a nd of technology-ntensve product h hgh relablty requrement. Its structure s very complex and ts man components are requred to have hgh precson and be resstant to hgh-temperature and mpact. Each steam turbne has nearly 000 parts. team turbnes embody the desgn prncple of combnaton products: decompose t nto a varety of standard modules, just le buldng a house h blocs. Accordng to customers ndvdual needs, lmted number of standard modules can be combned to produce dfferent customzed Currently, there are hundreds of steam turbne type models. team turbne plants manly consst of fve worshops: the daphragm worshop, the cylnder plant, the rotor assemblng shop, the tan shop and the fnal assembly shop. he supply chan of more turbne producton processng s smplfed nto Fg. n whch four COP are desgned. Fg. Mult-COP Mass Customzaton Producton ystem of urbne Machne he turbne's producton chan (Fg. ) s abstracted nto the followng lnear process. nce every node from node to node 8 contans multple processng stes, for example node s the abstracton of bendng clasp, castng, rough turn and nstalled welded. Node names are smplfed by numbers n Fg.. et COP at node 3,,6, Producton Flow recton tage producton COP COP3 tage producton COP COP tage 3 producton Fg. Abstract Lner agram of urbnes Producton OI 0.03/IJ.a IN: 73-80x onlne, prnt

7 Related parameters nformaton of the sem-fnshed products n one month are shown n table. ABLE.RELAIVE PARAMEER ABLE OF PROUCION AN INVENORY OF EMI-FINIHE PROUC 3 h w K F (Yuan) (Yuan) (Yuan) (Part) Accordng to the data n able, producton tas of each stage node n a certan perod s calculated ndvdually. () Use node as an example on tage. Push producton s used on ths stage, whch means the producton plannng s based on the forecast demand of sem-fnshed o the producton tas on node s the sum of optmal producton batch of each semfnshed product,that s P t Accordng to the model n 3.. and smulaton process n..3, the calculaton process of t s as follows: Frst, smulate surplus of nds of ( ) customzed products n the last cycle accordng to step one of the smulaton.hey are 00 peces, 60 peces, 0 peces and 30 peces. econd, gan random total demand of nds of customzed products on a certan lead tme accordng to step two of the smulaton, hey are w w w 0, 990, w w w 60, 3 3w 3w w w w w 80. For the capacty constrants, maxmum producton batches of customzed products of class, are F X t max 333, F X t max 86,whch are respectvely less than demand 0 and w w w w w w 80. o the optmal producton batch of customzed products of class, should be t 333, t 86. Maxmum producton batches of customzed products of class,3 are F X t 6, max F X3 3 3 t max 66,whch are respectvely 3 greater than demand. o the optmal producton batch of customzed products of class,3 are 990, 3t 60.In summary, t P = 79 t ()Use node 8 as an example n tage. Pull producton s used n ths stage, whch means the producton s based on customers demand. o the producton tas on node 8 s the sum of demand amount of customzed P 8 t (3)Use node as an example on tage 3. Push and pull producton s combned n ths stage. o the producton tas on node s the sum of customers demand amount on node 3, and forecast optmal producton batch on node 6,7. In summary, 7 P t = t Wth the example of turbnes producton plannng, the feasblty of mult-cop mult-stage producton plannng model s proved. herefore, n real world applcatons, ths model can be used to calculate the producton tass of push stage and push-pull combned stage of mult-cop to mae a better arrangement of producton. OI 0.03/IJ.a IN: 73-80x onlne, prnt

8 V. CONCLUION Frstly, ths paper analyzes two-stage producton plannng based on sngle COP, and sets up a multstage producton plannng model based on multple COP of mass customzaton. econdly, the paper dvdes producton plannng n supply chan h multple COP nto three stages: push producton plannng before COP based on forecast; push-pull producton plannng based on forecast and order; pull producton plannng after COP based on mae-toorder. Accordng to the three-stage model, optmzaton models of push and push-pull producton plannng are establshed and tested usng an example of steam turbnes. he operatng mechansm of producton plannng based on mult-cop of MC can be guaranteed. CONFLIC OF INERE he authors confrm that ths artcle content has no conflcts of nterest. ACKNOWLEGEMEN he paper s supported by the project of Natonal Natural cence Foundaton, Chna (No.7303;7708); the project of Chna Postdoctoral cence Foundaton (No );the project of Natural cences of Guangdong Provnce, Chna (No. 0A ). REFERENCE [] Arnab Banerjee, Bjon arar. Multple decouplng pont paradgms n a global supply chan syndrome: a relatonal analyss. Internatonal Journal of Producton Research., Frst,0. [] C.N. Verdouw, A.J.M. Beulens.. Bouwmeester, Modellng demand-drven chan networs usng multple COP Proceedngs APM'006 Lean Busness ystems and Beyond. Poland,Wroclaw,80:3338,006. [3] Garca-astugue, J.. and Lambert, M.., 007. Interorganzatonal tme-based postponement n the supply chan. Journal of Busness Logstcs, 8 (), 7 80.Garca- astugue and Lambert,007. [] Mason-Jones, R., Naylor, B., and owll,.r., 000. Lean, agle or leagle? Matchng your supply chan to the maret place. Internatonal Journal of Producton Research,38 (7): ,999. [] Phlp Hedensterna, Amos H.C. Ng. ynamc mplcatons of customer order decouplng pont postonng", Journal of Manufacturng echnology Management, (8):03 0,0. [6] Flavo. Foglatto, Govan J.C. da lvera, ens B-orensten (0). he mass customzaton decade: An updated revew of the lterature,int. J. Producton Economcs, (38) : [7] EevaJärvenpää, PasLuostarnen (0). ynamc operaton envronment--owards ntellgent adaptve producton systems. Assembly and Manufacturng (IAM), IEEE Internatonal ymposum:-,0. [8] Buffngton, Jac (0). Comparson of mass customzaton and generatve customzaton n mass marets. Industral Management & ata ystems,,():-6,0. [9] wamnathan, J.M. and ayur,.r.. Managng broader product lnes through delayed dfferentaton usng vanlla boxes.management cence, (), 6 7,998. [0] Yu Wang, Je Ln. upply Chan Model Based on Mult- COPs n Mass ynamc Customzaton[C].Proceedngs of the Internatonal Conference on Informaton Management, Innovaton Management and Industral Engneerng. Los Alamtos: Inst. of Elec. and Elec. Eng. Computer ocety, -,008. [] X.Y.un. Postonng multple decouplng ponts n a supply networ. Internatonal Journal of Producton Economcs,3: 9396,008 OI 0.03/IJ.a IN: 73-80x onlne, prnt