《系统工程》课程教学资源（英文文献）An Effective Lean Supply Inventory Management Model using VMI Hub

An Effective Lean Supply Inventory ManagementModel using VMI HubGuide Words: Supply chain, lean, supply inventoryAbstract: This paper describes an effective lean supply inventory management model using VMIhub for electronics industry. For a complex supply chain system, it is desired a lean supply inventorymanagement approach with simplicity and efficiency. The key elements of such a system areidentified as suppliers, VMI hubs, and factories. This paper firstly compares the two supply inventorymodels and explains the benefits of the model with VMI hub. Then the two key processes of the leansupply inventory model with VMI hub are further discussed, i.e. the kit-to-line process between aVMI hub and factories and the JIT suppliers pull process between suppliers and the VMI hub. Finallythe pilot run results are shown to suggest significant performance improvements in terms inventorysavingsand cycletimereductions.L.INTRODUCTIONWhile every supply chain struggles with demand and supply uncertainties, some electronicssectors like computers and mobile devices are particularly vulnerable. Those with very short productlifecycles have the greatest risk of losing the market share and obsolescence write offs. The recurringeconomic cycles further challenge these sectors which have to deal with components shortages in peakdemand periods and inventory write offs when market downturns coincide with the product's end oflife.After years of chasing surging demand and maneuvering to secure scarce supply, OEMs foundthemselves left with over excess inventory as the market suddenly turn down. Because of the longlead times for certain components on allocation and supply chains did not scale down as readily ashoped,manufacturers werenot ableto reactquicklyenough to plummeted levels of demand.In recent years, both academicians and practitioners have shown an increasing level of interest infinding ways of matching supply with demand while maintaining minimum levels of inventoriesthroughout the entire supply chain. Several researchers have examined many theoretical, as well aspractical issues involving buyer-supplier coordination, as a means of attaining successfulimplementationof Just-In-Time(JIT)baseddecisionsystems,focusingonmaterialflows,inaneffortto minimize the supply chain costs or maximize the entire chain's profitability.A JIT supply chain should produce and deliver goods just in time to be sold, subassembly just intime to be assembled, fabricate parts just in time to go into subassemblies, and purchase materials just

An Effective Lean Supply Inventory Management Model using VMI Hub Guide Words：Supply chain, lean, supply inventory Abstract：This paper describes an effective lean supply inventory management model using VMI hub for electronics industry. For a complex supply chain system, it is desired a lean supply inventory management approach with simplicity and efficiency. The key elements of such a system are identified as suppliers, VMI hubs, and factories. This paper firstly compares the two supply inventory models and explains the benefits of the model with VMI hub. Then the two key processes of the lean supply inventory model with VMI hub are further discussed, i.e. the kit-to-line process between a VMI hub and factories and the JIT suppliers pull process between suppliers and the VMI hub. Finally the pilot run results are shown to suggest significant performance improvements in terms inventory savings and cycle time reductions. I. INTRODUCTION While every supply chain struggles with demand and supply uncertainties, some electronics sectors like computers and mobile devices are particularly vulnerable. Those with very short product lifecycles have the greatest risk of losing the market share and obsolescence write offs. The recurring economic cycles further challenge these sectors which have to deal with components shortages in peak demand periods and inventory write offs when market downturns coincide with the product’s end of life. After years of chasing surging demand and maneuvering to secure scarce supply, OEMs found themselves left with over excess inventory as the market suddenly turn down. Because of the long lead times for certain components on allocation and supply chains did not scale down as readily as hoped, manufacturers were not able to react quickly enough to plummeted levels of demand. In recent years, both academicians and practitioners have shown an increasing level of interest in finding ways of matching supply with demand while maintaining minimum levels of inventories throughout the entire supply chain. Several researchers have examined many theoretical, as well as practical issues involving buyer–supplier coordination, as a means of attaining successful implementation of Just-In-Time (JIT) based decision systems, focusing on material flows, in an effort to minimize the supply chain costs or maximize the entire chain’s profitability. A JIT supply chain should produce and deliver goods just in time to be sold, subassembly just in time to be assembled, fabricate parts just in time to go into subassemblies, and purchase materials just

in time to be transformed into fabricated parts. The JIT approach provides the right materials, in therightquantitiesandquality,justintimeforproduction.Fawcett and Birou predicted that “"future manufacturing strategies will place significant emphasison the control of purchased inventory, increasing the value of a JIT procurement system"Today'semphasis on lean manufacturing further supports the need for efficient JIT supply chains. Therefore, itis critical that JIT suppliers identify and address performance issues as effectively as possible.TheJITpurchasingmethod isanimportanttechniqueoftheJITphilosophy,whichisregardedasoneof the most important productivity enhancement management innovations.JIT purchasingadvocates smaller sized andmorefrequent orders.ThisapproachwasoriginallyexploredbyKanzleras a method for reducing inventory levels at the Fordson Tractor Plant in the 1920s [3].Early studies conducted through the 1980s, including Goyal's pioneering work, focused only onjointlot sizing and buyer-supplier coordination for a single buyer and a single supplier based on alot-for-lot approach [4]. Subsequent researchers who followed this path had advanced the notion ofintegrationthrougheithernovel lotsplitting techniques or by examining more complex structures involving multiple buyers and/orsuppliers.A major gap in the existing literature, much of which focuses on direct, two-echelonbuyer-supplier coordination, is the issue concerning procurement of materials from suppliers by themanufacturing stage, integrated with the coordinated decisions of production and distribution toretailers. In view of this gap, this paper addressed the issue in the context of a real case for electronicsindustry. It is assumed that the factory is building to orders, therefore the production schedule islinked to the product distribution and delivery plan.Againstthisbackground,theaimofthispaperistointroduceaneffectiveleansupplyinventorymanagement approach and its implementation for electronics industry.The rest of this paper will firstdescribe two supply inventory management models and explain the advantages of having a VMI hubin the system. Following the comparison of the two models, Section Ill will discuss theimplementation issues of the supply inventory model with VMI hub. Then, the pilot run results aredemonstrated and discussed.A summary of the paper and lessons learnt conclude this paper.II.COMPARISONOFTWOSUPPLYINVENTORYMODELSThe electronics supply chain under study is just like those of many multi-national companies.There are a number of factories to support the global market. Some factories are self-owned facilitiesand some are ODMs. This study is focusing on the factories cluster in Asia, which provide majority ofthe products to worldwide market. The two options of the supply inventory models are discussed inthis section, i.e. either allowing the factories to manage their own supply inventories, or consolidatingthe suppliers shipments through a VMI hub. The advantages and disadvantages of the two models arecompared, with more emphasize on the impacts on the inventory level and the delivery cycle time

in time to be transformed into fabricated parts. The JIT approach provides the right materials, in the right quantities and quality, just in time for production. Fawcett and Birou predicted that “future manufacturing strategies will place significant emphasis on the control of purchased inventory, increasing the value of a JIT procurement system”. Today’s emphasis on lean manufacturing further supports the need for efficient JIT supply chains. Therefore, it is critical that JIT suppliers identify and address performance issues as effectively as possible. The JIT purchasing method is an important technique of the JIT philosophy, which is regarded as one of the most important productivity enhancement management innovations. JIT purchasing advocates smaller sized and more frequent orders. This approach was originally explored by Kanzler as a method for reducing inventory levels at the Fordson Tractor Plant in the 1920s [3]. Early studies conducted through the 1980s, including Goyal’s pioneering work, focused only on jointlot sizing and buyer–supplier coordination for a single buyer and a single supplier based on a lot-for-lot approach [4]. Subsequent researchers who followed this path had advanced the notion of integration through either novel lot splitting techniques or by examining more complex structures involving multiple buyers and/or suppliers. A major gap in the existing literature, much of which focuses on direct, two-echelon, buyer–supplier coordination, is the issue concerning procurement of materials from suppliers by the manufacturing stage, integrated with the coordinated decisions of production and distribution to retailers. In view of this gap, this paper addressed the issue in the context of a real case for electronics industry. It is assumed that the factory is building to orders; therefore the production schedule is linked to the product distribution and delivery plan. Against this background, the aim of this paper is to introduce an effective lean supply inventory management approach and its implementation for electronics industry. The rest of this paper will first describe two supply inventory management models and explain the advantages of having a VMI hub in the system. Following the comparison of the two models, Section III will discuss the implementation issues of the supply inventory model with VMI hub. Then, the pilot run results are demonstrated and discussed. A summary of the paper and lessons learnt conclude this paper. II. COMPARISON OF TWO SUPPLY INVENTORY MODELS The electronics supply chain under study is just like those of many multi-national companies. There are a number of factories to support the global market. Some factories are self-owned facilities and some are ODMs. This study is focusing on the factories cluster in Asia, which provide majority of the products to worldwide market. The two options of the supply inventory models are discussed in this section, i.e. either allowing the factories to manage their own supply inventories, or consolidating the suppliers shipments through a VMI hub. The advantages and disadvantages of the two models are compared, with more emphasize on the impacts on the inventory level and the delivery cycle time

A.Allow a Factory to Manage its Own Supply InventoryFig.I shows a scenario of allowing an individual factory to manage its own supply inventorieswithout linkingwith otherfactories.The individualfactorywill receiveordersfromdedicated marketand react by JIT pulling of components directly from its suppliers.SupplierFactorySupplierCustomer2SuppliermPullFig.1.Afactoryto manage its own supply inventoryIn this model, the inventories of all components will be stored at the factory store room, and thefactory prefers directly pulling from suppliers based on customer orders, so that the inventory costwould be minimized. For illustration purpose, an important component is selected to study the pullingeffect between the factory and one of the suppliers.The component is one of the major components offinal products, supplied by a particular supplier.Currently,the factory places orders to suppliers in batches to offset fixed ordering costs incurredevery time an order is placed when inventory levels drop. Because suppliers and the factorybuyer-planners react to changes in conditions as they occur, they used a continuous-time approach andthe corresponding optimal policy, a (Q, R) policy, to manage the component inventories.In a (Q, R) policy, as shown in Fig.2, R denotes the reorder point and Q is the size of the orderAn order with a lot size Q is placed whenever system inventory drops to R. The reorder point R isclosely related to safety stock ss, which is catering for both the uncertainty of customer demandduring the lead time and the uncertainty of lead time itself. Assume the demand during the lead time isDL and the standard deviation of demand during the lead time is o. Specifically, safety stock is thedifferencebetweenthe reorder point and the average requirements during replenishment lead time, i.e. R = ss + DL.Further assume the demand for the component per period is normally distributed with a mean ofD and standard deviation of o,; the lead time for replenishment follows normal distribution N(L,sL)In order to evaluate the Cycle Service Level (CSL) given a lot size Q, it needs to calculate theprobability of stock-out of the component if demand during the lead time exceeds the R. A stock-outmay occur in a cycle if the demand during the lead time is more than the reorder point R. The safety

A. Allow a Factory to Manage its Own Supply Inventory Fig.1 shows a scenario of allowing an individual factory to manage its own supply inventories without linking with other factories. The individual factory will receive orders from dedicated market and react by JIT pulling of components directly from its suppliers. Fig.1. A factory to manage its own supply inventory In this model, the inventories of all components will be stored at the factory store room, and the factory prefers directly pulling from suppliers based on customer orders, so that the inventory cost would be minimized. For illustration purpose, an important component is selected to study the pulling effect between the factory and one of the suppliers. The component is one of the major components of final products, supplied by a particular supplier. Currently, the factory places orders to suppliers in batches to offset fixed ordering costs incurred every time an order is placed when inventory levels drop. Because suppliers and the factory buyer-planners react to changes in conditions as they occur, they used a continuous-time approach and the corresponding optimal policy, a (Q, R) policy, to manage the component inventories. In a (Q, R) policy, as shown in Fig.2, R denotes the reorder point and Q is the size of the order. An order with a lot size Q is placed whenever system inventory drops to R. The reorder point R is closely related to safety stock ss, which is catering for both the uncertainty of customer demand during the lead time and the uncertainty of lead time itself. Assume the demand during the lead time is DL and the standard deviation of demand during the lead time is . Specifically, safety stock is the difference between the reorder point and the average requirements during replenishment lead time, i.e. R = ss + DL. Further assume the demand for the component per period is normally distributed with a mean of D and standard deviation of ; the lead time for replenishment follows normal distribution N(L,sL). In order to evaluate the Cycle Service Level (CSL) given a lot size Q, it needs to calculate the probability of stock-out of the component if demand during the lead time exceeds the R. A stock-out may occur in a cycle if the demand during the lead time is more than the reorder point R. The safety

inventory, ss,can be calculated such that the following is true:(1)CSL=F (ss + DL,DL, o,)whereDL=LXD,0,=Lo + D?0(2)Given the distribution of the demand during the lead time L, it is able to obtain the CSL for thecomponentunderdiscussionOn the other hand, with a desired level of CSL, e.g.95%, it is possible to determine the safetyinventoryfrom thebelow equation:SS = F-1(CSL)X (3) InventoryQSafetyDInventoryRun Out PointLeadTimeReorderPointUncertaindemandUncertainleadtimeFig.2.Safety inventory cateringfor demand and supplyuncertaintyFrom above (1), (2) and (3), it can be concluded that the safety inventory is determined by thestandard deviation of the demand during the lead time given desired cycle service level. The safetyinventory can be relativelyhigh when the lead time and the standard deviation of the lead time arebothlarge.Thisisparticularlytruewhenthesuppliersarefardistanceawayfromthefactoriesandtherefore the variances of the transportation lead time are also large.B. VMI Hub to Manage Supply Inventories for FactoriesAnother model of managing supply inventories is illustrated in Fig.3, where a VMI(Vendor-Managed Inventory) hub is adopted to consolidate all the suppliers' shipments for multiplefactories. The VMI hub consists of two segregated inventories, SOI and FOL. SOI refers to SuppliersOwned Inventory, which means the materials are under suppliers' books until they reach the agreedownership transferring dates or receive the request to be transferred. SOI becomes Factory OwnedInventory (FOl)once they aretransferred.Factories could pull FOIwith a fast turnaround time

inventory, ss, can be calculated such that the following is true: CSL = F (ss + DL,DL, ) (1) where DL=L D, (2) Given the distribution of the demand during the lead time L, it is able to obtain the CSL for the component under discussion. On the other hand, with a desired level of CSL, e.g.95%, it is possible to determine the safety inventory from the below equation: (3) Fig.2. Safety inventory catering for demand and supply uncertainty. From above (1), (2) and (3), it can be concluded that the safety inventory is determined by the standard deviation of the demand during the lead time given desired cycle service level. The safety inventory can be relatively high when the lead time and the standard deviation of the lead time are both large. This is particularly true when the suppliers are far distance away from the factories and therefore the variances of the transportation lead time are also large. B. VMI Hub to Manage Supply Inventories for Factories Another model of managing supply inventories is illustrated in Fig.3, where a VMI (Vendor-Managed Inventory) hub is adopted to consolidate all the suppliers’ shipments for multiple factories. The VMI hub consists of two segregated inventories, SOI and FOI. SOI refers to Suppliers Owned Inventory, which means the materials are under suppliers’ books until they reach the agreed ownership transferring dates or receive the request to be transferred. SOI becomes Factory Owned Inventory (FOI) once they are transferred. Factories could pull FOI with a fast turnaround time

normally within one day.SupplieVMIHubSupplierCustomerFactoryCustomeraCTOSuppliePushPullFig.3.Supply inventorymanagement withVMI hubUnder VMI arrangement, suppliers decide how much inventory to ship and when to ship whilefactories just set target inventory levels and record suppliers'deviations from the targets.Factorieswithdraw inventories from FOI only when needed. In addition, factories do not own the inventory inSOI, which is owned by suppliers instead and charged to factories indirectly through componentpricing. The cost of maintaining inventory in SOI is, however, eventually included in the final pricesof thefinal products.Therefore, any reduction in inventorybenefits factories'customers directly byreducingproductprices.In Fig.3, the VMI hub is normally a decoupling point, where pull processes triggered bycustomer demands, and push processes are initiatedby the information sharing between thefactoriesand suppliers. However, this paper will introduce a Suppliers Pull (JSP) process, which is to establisha JIT pulling system between suppliers and the VMI hub.To understand the impact of VMI hub on supply inventories, it is necessary to find out thedistribution of the aggregate demand across all the product models of multiple factories for thecommon components, Assume the aggregate demand of the same component discussed previously isnormally distributed with a mean of Rc, standard deviation of o, and the demands across all theproduct models are independent, that means all correlation coefficients and thus covariance is zero.In our context, because different product models are manufactured by different factories andserve for different market segments, it is acceptable to assume that the demands of the commoncomponent across product models are independent. Therefore, it concludes that the aggregate demandhas a much lower standard deviation than that of the sum of individual demands across productmodelsofmultiplefactories.Since the required safety inventory is proportional to the standard deviation of the demand duringthe replenishment lead time, it is thus likely to conclude that aggregating demands will reduce theamountofsafetyinventoryrequiredwithouthurtingthecomponentavailability.Similarly, the impact of aggregate demand on safety inventory can be evaluated by desired cycle

normally within one day. Fig.3. Supply inventory management with VMI hub Under VMI arrangement, suppliers decide how much inventory to ship and when to ship while factories just set target inventory levels and record suppliers’ deviations from the targets. Factories withdraw inventories from FOI only when needed. In addition, factories do not own the inventory in SOI, which is owned by suppliers instead and charged to factories indirectly through component pricing. The cost of maintaining inventory in SOI is, however, eventually included in the final prices of the final products. Therefore, any reduction in inventory benefits factories’ customers directly by reducing product prices. In Fig.3, the VMI hub is normally a decoupling point, where pull processes triggered by customer demands, and push processes are initiated by the information sharing between the factories and suppliers. However, this paper will introduce a Suppliers Pull (JSP) process, which is to establish a JIT pulling system between suppliers and the VMI hub. To understand the impact of VMI hub on supply inventories, it is necessary to find out the distribution of the aggregate demand across all the product models of multiple factories for the common components. Assume the aggregate demand of the same component discussed previously is normally distributed with a mean of Rc, standard deviation of , and the demands across all the product models are independent, that means all correlation coefficients and thus covariance is zero. In our context, because different product models are manufactured by different factories and serve for different market segments, it is acceptable to assume that the demands of the common component across product models are independent. Therefore, it concludes that the aggregate demand has a much lower standard deviation than that of the sum of individual demands across product models of multiple factories. Since the required safety inventory is proportional to the standard deviation of the demand during the replenishment lead time, it is thus likely to conclude that aggregating demands will reduce the amount of safety inventory required without hurting the component availability. Similarly, the impact of aggregate demand on safety inventory can be evaluated by desired cycle

service level (CSL), given a lead time of L, average demand during the lead time RL, and standarddeviation of demand during lead time or. Recall from (3), the required safety inventory isproportional to the standard deviation of the demand during the lead time. It is proved that a smallerstandard deviation of the aggregate demand will reduce the amount of safety inventory required tomeet the desired cycle service level.In addition, the VMI hub is located much closer to factories than most suppliers, therefore theaveragelead timeLand the standarddeviation of lead time SL aremuch smallerthan thatbetweenfactories and suppliers. From (2), it indicates that the standard deviation of demand during lead time, will be much smaller too.By and large, the safety inventory of using VMI hub to manage supply inventory is expected tobemuch lower than that managed by individual factories due to the twokey reasons: (a)the aggregatedemand of the component across all the product models has a much lower standard deviation than thatof the sum of individual demands; (b) the much shorter distance between factories and VMI hub alsocontributes to a much lower standard deviation of demand during the replenish lead time. Thediscussions in this section may also support the observations of that many multinational companieswith global market adopted VMI hubs to support their factories by managing their supply inventories.IIL.IMPLEMENTATIONOFVMIHUBBASEDSUPPLYINVENTORYSYSTEMFig.4 is a schematic diagram of the VMI hub based supply inventory management model, Thereare three key processes for such a supply inventory system. The first process is the JIT FinishedGoods (FG) Pull from factories to customers. The second process is the JIT Kit-to-lines (KTL) Pullprocess between the VMI hub and factories. The KTL process is to ensure the factory only pullsmaterials needed for making to orders with smallest possible lot sizes. The third process is JITSuppliersPull(JSP)process,whichistoensurejustenoughmaterialsarepulledintotheSOIoftheVMI hub.Push out Material Iinventory (Lean)Push out Material Inventory (lean)FG Demand Pull (Lean)JITSupplierJITFGJIT Kit-To-LinePullPullPullHubsSuppliersFactoriesCustomersReal-timeDemand Signal, Confirmation&Supply Visibility (information Flow)Fig.4. Schematic diagram of the VMI hub supply inventory modelA. JIT Kit-To-Line Pull ProcessFig.5 describes theKTL process in more details by showing the major information flows and

service level (CSL), given a lead time of L, average demand during the lead time RL, and standard deviation of demand during lead time . Recall from (3), the required safety inventory is proportional to the standard deviation of the demand during the lead time. It is proved that a smaller standard deviation of the aggregate demand will reduce the amount of safety inventory required to meet the desired cycle service level. In addition, the VMI hub is located much closer to factories than most suppliers, therefore the average lead time Land the standard deviation of lead time SL are much smaller than that between factories and suppliers. From (2), it indicates that the standard deviation of demand during lead time will be much smaller too. By and large, the safety inventory of using VMI hub to manage supply inventory is expected to be much lower than that managed by individual factories due to the two key reasons: (a) the aggregate demand of the component across all the product models has a much lower standard deviation than that of the sum of individual demands; (b) the much shorter distance between factories and VMI hub also contributes to a much lower standard deviation of demand during the replenish lead time. The discussions in this section may also support the observations of that many multinational companies with global market adopted VMI hubs to support their factories by managing their supply inventories. III. IMPLEMENTATION OF VMI HUB BASED SUPPLY INVENTORY SYSTEM Fig.4 is a schematic diagram of the VMI hub based supply inventory management model. There are three key processes for such a supply inventory system. The first process is the JIT Finished Goods (FG) Pull from factories to customers. The second process is the JIT Kit-to-lines (KTL) Pull process between the VMI hub and factories. The KTL process is to ensure the factory only pulls materials needed for making to orders with smallest possible lot sizes. The third process is JIT Suppliers Pull (JSP) process, which is to ensure just enough materials are pulled into the SOI of the VMI hub. Fig.4. Schematic diagram of the VMI hub supply inventory model A. JIT Kit-To-Line Pull Process Fig.5 describes the KTL process in more details by showing the major information flows and

materials flows between the VMI hub and one of the factories.An on-line system linking the factoryand the VMI hub is built for triggering the pulling of kits based on the kanban signal. Once a kanbanis available for a line, the VMI hub is informed to deliver the kit in four hours time. Once the kit isready, the confirmation is sent back to the factory to inform the delivery of the kit. All requested partsof one model are classified into four types: A, B, C and D. Parts A and B are either expensive parts orunique parts to a model. C and D parts are either cheaper parts or none bulky parts and space taken isnot a concern. Based on such clarification, A and D parts will be pulled directly from VMI hub in kitswhen a factory line needs them.B and C parts will be stored in factory FOI by daily replenishment.The information sharing of production schedule to the VMI hub is very important to support theKTLprocess.Production scheduleis sharedwithVMIhubonweeklybasisand updated ondailybasis.Such information haring is critical for the VMI hub to prepare resources and bin-down pallets one dayahead beforetheactualkitspulling.Moreimportantly,with suchinformation sharing,VMIhub couldcheck the availability of materials at VMI hub and feedback any parts constraints after running theconstraint report.infoflow7.ReturnedPartsMtl.ffovVMI HubFactoryProduction Line5,NextDavSchBreakBulkLabeling&WIPAtSOIScanning6.ReplenishB&CitedayPlarsoCEEEEE3.Kanban Material delivered within 4hrsWebForm Puli-A,B.C.D items"byDJ bylinetriggeredbykanbanWebForm Pull"A&D items"byDJ by Linetriggered bykanbarFig.5.Kit-to-lineprocessbetweenVMIhubandfactories.B.JITSuppliers-PullProcessOnce Kit-to-line process is established and stabilized, the next step is to implement theJust-in-timeSuppliersPull(JSP)process,whichisdesignedtoestablishaJITpullingsystembetweensuppliers and the VMI hub. JSP process needs to consider the information exchanges among suppliers,VMI hubs and the factories. In addition, it has to take into consideration of the production lead timesof both the suppliers and the factories.Fig.6 briefly illustrates the processes and information flows involving the factory, VMI hubs, andsuppliers. Firstly, the demand forecast should be shared with suppliers to prepare for raw materials in

materials flows between the VMI hub and one of the factories. An on-line system linking the factory and the VMI hub is built for triggering the pulling of kits based on the kanban signal. Once a kanban is available for a line, the VMI hub is informed to deliver the kit in four hours time. Once the kit is ready, the confirmation is sent back to the factory to inform the delivery of the kit. All requested parts of one model are classified into four types: A, B, C and D. Parts A and B are either expensive parts or unique parts to a model. C and D parts are either cheaper parts or none bulky parts and space taken is not a concern. Based on such clarification, A and D parts will be pulled directly from VMI hub in kits when a factory line needs them. B and C parts will be stored in factory FOI by daily replenishment. The information sharing of production schedule to the VMI hub is very important to support the KTL process. Production schedule is shared with VMI hub on weekly basis and updated on daily basis. Such information haring is critical for the VMI hub to prepare resources and bin-down pallets one day ahead before the actual kits pulling. More importantly, with such information sharing, VMI hub could check the availability of materials at VMI hub and feedback any parts constraints after running the constraint report. Fig.5. Kit-to-line process between VMI hub and factories. B. JIT Suppliers-Pull Process Once Kit-to-line process is established and stabilized, the next step is to implement the Just-in-time Suppliers Pull (JSP) process, which is designed to establish a JIT pulling system between suppliers and the VMI hub. JSP process needs to consider the information exchanges among suppliers, VMI hubs and the factories. In addition, it has to take into consideration of the production lead times of both the suppliers and the factories. Fig.6 briefly illustrates the processes and information flows involving the factory, VMI hubs, and suppliers. Firstly, the demand forecast should be shared with suppliers to prepare for raw materials in

suppliers'warehouse. Secondly,the short term production planning forecast needs to be sent tosuppliers to prepare their production and suppliers need to confirm the production plan with thefactory if any changes.Thirdly,safety stocks are needed at both suppliers'sites andVMIhubs cateringfor potential uncertainties and risks from production, materials quality, and transportation lead timesThe requested components based on the short term production planning forecast are shipped to VMIhubforkittingtothefactorylines.4HubshiptoFactory③Supplie0lAHProductionLinegForecas2rnlweekine2.WeeksplanFig.6. JIT suppliers pull from suppliers to the hubIV.PILOTRUNRESULTSANDANALYSISForthepurposeofproofoftheconcept,pilotrunswerecarefullyplannedandconductedwithsixselected suppliers and around twenty product models. The pilot runs continued for three months andhundreds shipments weredelivered and monitored.The inventoryand cycletimeare measured as KeyPerformance Index (KPI). The components inventory measured is a sum of both the FOI in VMI huband FOI at factory store room. The cycle time refers to the time starting from when the pulling signalis sent from the factory to when the components are received by the factory.Table I shows the results from the pilot run in terms of the two KPIs measured. The inventorylevels reduction from each of the six suppliers are measured respectively and an average of 57%inventory reduction was achieved. On the other hand, the reduction of cycle times for each supplierwas alsorecorded and anaverageof 43.4%cycle timewas reduced.Tablel The Results From Pilot RunsKPISupplierResultsReduced Inventory by $487.4K (-54%)InventoryReductionPBReduced InventorybyS2293K(-67.3%CReduced Inventoryby$345K(-45.2%DReducedInventorybyS430K(-47.8%LReducedInventorybyS43.5K(-45.3%EReduced Inventoryby71.5K.(-82.5%)Total:USD3.670K.average(-57%)ACycle Time ReductionReducedCycletimeby3days(-428%)BReducedCycletimeby3days(42.8%)ReducedCycletimeby3days(-37.5%)AReducedCycletimeby3days(-37.5%)ReducedCycletimeby4days(-57%)ReducedCycletimeby3days(-42.8%)[Average:3.16days.average(43.4%)One more factor of such a lean supply inventory management system needs to be considered is

suppliers’ warehouse. Secondly, the short term production planning forecast needs to be sent to suppliers to prepare their production and suppliers need to confirm the production plan with the factory if any changes. Thirdly, safety stocks are needed at both suppliers’ sites and VMI hubs catering for potential uncertainties and risks from production, materials quality, and transportation lead times. The requested components based on the short term production planning forecast are shipped to VMI hub for kitting to the factory lines. Fig.6. JIT suppliers pull from suppliers to the hub IV. PILOT RUN RESULTS AND ANALYSIS For the purpose of proof of the concept, pilot runs were carefully planned and conducted with six selected suppliers and around twenty product models. The pilot runs continued for three months and hundreds shipments were delivered and monitored. The inventory and cycle time are measured as Key Performance Index (KPI). The components inventory measured is a sum of both the FOI in VMI hub and FOI at factory store room. The cycle time refers to the time starting from when the pulling signal is sent from the factory to when the components are received by the factory. Table I shows the results from the pilot run in terms of the two KPIs measured. The inventory levels reduction from each of the six suppliers are measured respectively and an average of 57% inventory reduction was achieved. On the other hand, the reduction of cycle times for each supplier was also recorded and an average of 43.4% cycle time was reduced. Table1 The Results From Pilot Runs One more factor of such a lean supply inventory management system needs to be considered is

thecostincurredfortheVMIhuboperations,comparedtothescenarioof linkingthefactorydirectlywith suppliers.Because the VMI hubs are run bya 3PL, and the quotation from the 3PLis justified bycomparing with the factory's operational cost savings, which could be obtained from transferring thereceiving and kitting activities to the VMI hubs. The factories need less manpower and space forinbound logistics due to JIT Kitto-line process and much lower inventory at factory store room. Suchsavings could be more significant when multiplefactories are supported by the same VMI hub.Theresultfrom the pilot runs shows an average of approximate 10%operational cost savings.However, there is a fixed overhead cost of a VMI hub which requires a minimum productionvolume to secure the breakeven point, it could happen when the production volume drops to asignificantly lower degree than the specified minimum production volume.V.CONCLUSIONA lean supply inventory management system design requires a systems engineering approach toincorporatealltherelevantsystemelementsandsystemconstraintsintothedesign.Foracomplexsupplychainsystem,itisdesiredalean supplyinventorymanagement systemwithsimplicityandefficiency.The key elements are identified as suppliers, VMI hub, and factories.This paper focuses more on the supply inventory management system, though the distributionsystem is another challenging part of the whole supply chain. The two processes are designed for theVMI hub based lean supply inventory management system. First one is the kitto-line process betweenVMI hub and the factories, and the second one is the JIT suppliers pull process between suppliers andthe VMI hub. In addition, comprehensive information systems are needed to support such a VMI hubbased lean supply inventory management system.In a summary, the results from both the theoretical discussions and the pilot runs demonstratedthat supply chain performance could besignificantly improvedby theproposed lean supply inventorymanagement system. Even with those suppliers with long lead time, the inventory and cycle time werereduced because the VMI hub can consolidate the demands and shorten the lead times of theshipments from suppliers tofactories

the cost incurred for the VMI hub operations, compared to the scenario of linking the factory directly with suppliers. Because the VMI hubs are run by a 3PL, and the quotation from the 3PL is justified by comparing with the factory’s operational cost savings, which could be obtained from transferring the receiving and kitting activities to the VMI hubs. The factories need less manpower and space for inbound logistics due to JIT Kitto-line process and much lower inventory at factory store room. Such savings could be more significant when multiple factories are supported by the same VMI hub. The result from the pilot runs shows an average of approximate 10% operational cost savings. However, there is a fixed overhead cost of a VMI hub which requires a minimum production volume to secure the breakeven point; it could happen when the production volume drops to a significantly lower degree than the specified minimum production volume. V. CONCLUSION A lean supply inventory management system design requires a systems engineering approach to incorporate all the relevant system elements and system constraints into the design. For a complex supply chain system, it is desired a lean supply inventory management system with simplicity and efficiency. The key elements are identified as suppliers, VMI hub, and factories. This paper focuses more on the supply inventory management system, though the distribution system is another challenging part of the whole supply chain. The two processes are designed for the VMI hub based lean supply inventory management system. First one is the kitto-line process between VMI hub and the factories, and the second one is the JIT suppliers pull process between suppliers and the VMI hub. In addition, comprehensive information systems are needed to support such a VMI hub based lean supply inventory management system. In a summary, the results from both the theoretical discussions and the pilot runs demonstrated that supply chain performance could be significantly improved by the proposed lean supply inventory management system. Even with those suppliers with long lead time, the inventory and cycle time were reduced because the VMI hub can consolidate the demands and shorten the lead times of the shipments from suppliers to factories