Networked RFID fundamentals 4. Fine granularity supply chain management using RFID

Transcription

1 SOI Asia special lecture series Networked RFID fundamentals 4. Fine granularity supply chain management using RFID Tatsuya INABA Research Associate Graduate School of Media and Governance Keio University Auto-ID Laboratory, Japan Agenda! SCM 101! RFID for efficiency and accuracy of SCM! RFID beyond basics Page 2

2 What is SCM?! SCM stands for Supply Chain Management! Series of business practices to optimize flow of goods order order order demand Supplier Manufacturer Wholesaler Retailer goods goods goods Consumer Page 3 More than Goods! There are definitions of SCM Supply Chain Management deals with the management of materials, information, and financial flows in a network consisting of suppliers, manufacturers, distributors, and customers. Stanford Global Supply Chain Management Forum Logistics involves... managing the flow of items, information, cash and ideas through the coordination of supply chain processes and through the strategic addition of place, period and pattern values. MIT Center for Transportation & Logistics Page 4 More definitions but cover the same concept

3 Real supply chain is more complex Supplier Manufacturer Wholesaler/ Retailer Customer Page 5 Applications of RFID on SCM Page 6 Copyright 2003 Auto-ID Center, All Rights Reserved

4 Exercise! How much does a store need to place order in each order cycle? The condition of the store is below:! Condition! Lead time!time b/w order and goods arrival": 3 [days]! Inventory check cycle: 5 [days]! Check inventory level and place order to meet customer demand Page 7 Ans. with figure! Base stock inventory management policy! The store needs to have sufficient inventory to meet the demand between order placement and goods arrival. [inventory level] A B B D E E Amt of order G Base stock level C F 3days 3days 3days [day] Page 8 5days 5days

5 Ans. with equation! Base stock level (B) This part is inventory to cover lead time and review period. This part is for safety stock. Since demand is fluctuated (=value of standard deviation is not zero), the store needs to have more inventory to meet larger than average demand. z is safety factor, which decides service level of the store. To achieve 95% service level, z # 1.64 Page 9 Average inventory level [inventory level] average inv Safety stock Page 10

6 Cost of having inventory! Two types of cost incurred! Non capital inventory carrying cost! Warehouse! Obsolescence! Pilferage (Shrinkage)! Damage! Insurance! Capital inventory carrying cost! Cost of capital Page 11 There are more policies! (Q, R) policy! Inventory is checked continuously! Order quantity is fixed (Q)! Need to define re-order point (R)! (S, s) policy! Inventory is checked periodically! Order is placed when inventory level is under a certain level (s)! Order is placed up to a certain level (S)! Need to define both S and s more Page 12

7 Cost vs. Sales! Basically the more you have inventory, the more cost of having inventory incurred! But if you reduce too much inventory, you would experience sales opportunity loss! Sales opportunity loss reduces sales (of course)! Also, sales opportunity loss lowers customer satisfaction Page 13 Summary of SCM101! Lead time affects performance of the inventory management! If lead time is long, a store needs to order more and carry more.! SCM is a set of practices to maximize sales with less cost (inventory, labor, etc.) Page 14! To know accurate inventory level is fundamental of SCM! Inventory management practices premises accurate number of the inventory How can RFID contribute to improve efficiency and accuracy of SCM?

8 RFID for efficiency and accuracy! Many countings in supply chain operation!! Countings in manufacturer! In response to the order from customers! Counting in picking! Counting before shipment Countings in retailer! Counting in retailer s distribution center! Counting in receiving to check the number of items meets original order! Counting in picking! Counting before shipment! Counting in retailer s stores! Counting in receiving to check the number of items meet original order Page 15 There are many counting to operate supply chain. Why counting is important? Why counting is important?! Counting is the basis of inter-organization trade! Inter-company trade! Official point of change of custody! Buyer pays money according to the number of items received! Too many item shipment will decrease profit of supplier! Too few item shipment will cause problem in buyer (and usually penalty is imposed to suppliers for short shipment)! Intra-company trade! Official point of change of custody! May need to keep records for corporate finance Page 16

9 Labor intensive! Page 17 Impact of RFID on SCM! Capabilities of RFID! Remote read, non line of sight read, multiple read! Impact of using RFID on SCM! Make the labor intensive work less labor intensive! Make error prone manual process more accurate! Enable more frequent counting in SCM process With Barcode With RFID Page 18 Photo courtesy of CNET Networks Japan K.K.

10 RFID pilot experiments! Many RFID experiments are conducted world wide!! Both Government and industries lead the initiative! First case: International trade! A pilot project sponsored by Ministry of Economy, Trade and Industry Japan! Applied to international supply chain (from manufacturer in China to stores in Japan)! Apparel industry! Led by Sumikin Bussan (a Japanese trading company) Page 19 Abstract of Experiment China Japan Manufacturer Distribution center Distribution center store Counting Paper less Paper less Inspection China Picking Sorting in Chinese DC Japan Distribution center Distribution center Customer Center Inspection Picking Store Customer Center Store Cross dock Store Page 20 Source: Report on METI RFID Experiment, Sumikin-bussan, p.18 (2006)

11 Impact of RFID at the Experiment Domestic DC (Japan) Store As-Is Shipment From Shanghai receipt inspection sorting shipment receipt RFID Shipment From Shanghai receipt inspection sorting shipment receipt As-Is process receiving checking (*) sorting (*) shipping , , receiving RFID process , , (*) In RFID process, these tasks are moved to main land China. Source: Report on METI RFID Experiment, Sumikin-bussan, p.25, p.109 (2006) Page [second] RFID checking gate tests Antenna RF tags Source: Report on METI RFID Experiment, Sumikin-bussan, p.62 (2006) Page 22

12 2 nd case! Second case: Store operation! A pilot project sponsored by Ministry of Economy, Trade and Industry Japan (But continue as a commercial system)! Applied to a department store operation! Apparel (shoe) industry! Led by Japanese department stores Page 23 Issus in shoe stores! Too many different kinds of shoes! Designs! Sizes! Short product lifecycle! New products are released in a season basis! Space constraint! Shoe stores in department store do not have sufficient space to display so that shoes are stocked in backrooms! Takes time to find shoes! Loss of sales and customer satisfaction Page 24

13 As-Is operation Display in sales floor How shoes are received Source: Report on METI RFID Experiment, Assc. Of Department and Apparel RFID, p.10 (2005) Page 25 Abstract of experiment Warehouse dock Ship order check Warehouse stock Warehouse Shipping area Direct shipment Sorting RF Tag application Inspection Linking RF and SKU Price tag application Stocking Picking list check Stocking Picking Order proposal Putting away Warehouse Shipping dock Before Order shipment confirmation check Shipment Shipment Shipment label Packing Replenishment Shipment process Cycle counting Stock check Shoe store Warehouse Process with RFID (newly added) Sales Payment RF tag removal RF tag recycle Process with RFID Tag recycle Source: Report on METI RFID Experiment, Assc. Of Department and Apparel RFID, p.88 (2005) Page 26

14 Store operation Receipt counting with RFID KIOSK for shoe information! Improvement of store operation! Improvement of customer satisfaction Source: Report on METI RFID Experiment, Assc. Of Department and Apparel RFID, p.60 (2005) Page 27 Consequence of efficient SCM! Efficiency brought by RFID may cut lead time! Shortened lead time enables companies to achieve the same service level with less average inventory! Reduced inventory reduces cost Page 28

15 Consequence of inaccurate SCM! Gap between physical inventory and logical inventory! Order less based on logical inventory Logical Inventory = (# of ordered items) (# of sold items) Physical Inventory = (# of ordered items) (# of sold items) (# of shrinkage) (# of human error) etc. Order less also means higher possibility of opportunity loss! Page 29 A hypothetical scenario! Think about the impact of inaccurate inventory from a hypothetical scenario! A hypothetical scenario! Demand: normal distribution (mean:10[items/day], standard deviation: 2[item])! Item lost rate: Poisson distribution (average: 0.2 [items/day])! Lead time: 3 days! Review period: 5 days! Sales period: 1 year Source: KANG, Yun, Information Inaccuracy in Inventory system, Ph.D Thesis M.I.T(2006) Page 30

16 Simulation result If you operate this inventory for one year, you are expected to have appox. 100 opportunity losses! RFID system is expected to achieve accurate inventory management! Stock outs! Page 31 Summary of basic merits! Capabilities of RFID reduces manual intervention in the supply chain! Reduced manual intervention reduces lead time and inaccuracy in item handling! Reduced lead time and manual handling reduce supply chain operation cost! Accurate supply chain improves customer satisfaction and profits Page 32

17 RFID beyond Basics! Efficiency and accuracy are important BUT benefits are not limited to these benefits! Recap the functionality of networked RFID Page 33 Problem of pursuing the Basics! Costs and benefits are not balanced among supply chain partners in supply chains! Costs! Fixed cost! Cost for information system (e.g., hardware, software, system integration etc.)! Both upstream and down stream companies bear the cost! Variable cost! Cost for RF tags! Primary upstream company bears the cost! Benefits! Larger benefits in down stream companies because they deal with heterogeneous items in nature. Page 34 This is one important reason why beyond the basics is required

18 Networked RFID Internet RFID registry Service (e.g., ONS,DS) Company $! Company B RFID Info. Sys. RFID Info. Sys. RFID event repository RFID event repository Enterprise sys.!e.g. WMS, ERP"! middleware middleware Enterprise sys.!e.g. WMS, ERP"! Reader/writer Reader/writer RF tag Page 35 Fine granularity mgmt by RFID! Definition of Fine granularity management! Designers of SCM can count on availability of fine granular info! Operators of SCM can utilize fine granular info! Definition of Fine granular info in our study 1. Item level object identification (e.g., individual item, case/pallet) 2. Location info of a unique object in a real time basis and its archive 3. Location info of a unique object in an arbitrary location and its archive 4. Status info of a unique object and its archive Page 36

19 Illustration of fine granular mgmt! Fine granular info enabled by lowered data acquisition cost! 4 dimensions: what, when, where, status What When Where Status input Somewhere in the build. Who moves? Time Stock keeping level Batch mgmt Rough locating No desc. about object In the drawer John moved Item level Page 37 Time Real time mgmt Precise locating Desc. about object Page 38 Characteristics of processes enabled by fine granularity mgmt! High visibility! Processes are designed and operated under stable and accurate information about objects in supply chains! High controllability! Processes are designed and operated with an assumption of easy object finding and manipulation! High manageability! Processes are designed and operated based on number not rule of thumb regarding objects! High adaptiveness! Processes are designed and operated to dynamically adapt the changing situations

20 Introduction of FGM SCM process! Virtual inventory! Inaba, Impact of Real-time Granular Product Information on Transshipment between Stores, J. of Japan Industrial Mgmt Assc. 58(2): (2007)! Realization of SCM and CRM with RFID! Inaba, Realization of SCM and CRM by Using RFID-captured Consumer Behavior Information, J. of Networks 4(2):92-99 (2009) Page 39 Virtual inventory! Target supply chain! Short lifecycle item (e.g., seasonal item)! Stochastic demand! One supplier one buyer! Buyer has one time to decide how much to buy for sales before sales season and can not change Page 40 These are common characteristics of apparel supply chains

21 Issues of Short Lifecycle Products!! Lifecycle of products becomes shorter!! Buyer has many difficulties:! Difficult to forecast! Only one time to decide the order quantity for the season! Need sufficient inventory to get customer loyalty but need to discount if they have excess inventory Risk pooling is necessary!! Build distribution center and/or increase shipment frequency! May achieve higher service level! May increase transportation cost! Build up inventory! May achieve higher service level! May increase inventory cost (holding, discount)! Develop Virtual Inventory (Sharing inventory info among stores)! May get maximum benefit with low inventory carrying cost Page 41 Concept of Virtual Inventory! Share inventory information among retail stores! Inventory is distributed (do not miss sales opportunities)! Inventory is shared (high product availability) Inventory information is shared Distributed Inventory Virtual Inventory How much is the benefit of having virtual inventory? Page 42

22 Model Situation!!! Supplier ships goods to Buyer (two stores) Single period stochastic demand problem situation! Demand is stochastic! Only one time to decide order quantity before the season starts Inventory info! Case A: Not shared! Case B: Shared (= virtual inventory)! Store #1 and #2 are identical Buyer Store #1 Buyer Store #1 Supplier Supplier Store #2 Store #2 Case A (w/o inventory info sharing) Case B (w/ inventory info sharing) Page 43 Parameters for analytical solution Page 44

23 Optimal order quantity Optimal order quantity of case A p, w, s are fixed variables Optimal order quantity of case B h: handling cost A, B are fixed variable from p, w, s Next step:! Find range of handling cost (h) where cost of case B becomes smaller! Fine when the order quantity become largest Page 45 Upper bound of handling cost Compare order quantities of (A) and (B) AND F(Q *)%F(Q*) 0 h 0 h If handling cost is lower than h 0, case (B) s quantity is larger than case (A) s Page 46

24 Numerical Study! Simulation algorithm Start Assume Order quantities of Store #1 and #2 Generate Demand based on demand assumption Calculate profit of each store Calculate expected profit (average) Pick up the order quantities of Store #1 and #2 which maximize the expected profit of the buyer Page 47 End Parameters! Monetary values: p: 1,000 w: 650 s: 300 c: 400! Demand:! Assume normal distribution! Mean 250! Standard deviation 100 Page 48

25 Base case: sensitivity to handling cost! Deal with handling cost (h) as a variable (i) without information sharing (ii) with information sharing, h = 0 (iii) with information sharing, h = 300 (iv) with information sharing, h = 500 (v) with information sharing, h = 700 Page 49 Base case result Order QTY Expected Profit at Buyer Expected Profit at Supplier Channel Expected Profit i ,165 (100%) 122,500 (100%) 242,665 (100%) ii ,794 (114%) 124,000 (101%) 260,794 (107%) iii ,172 (107%) 124,390 (102%) 253,562 (104%) iv ,706 (103%) 123,343 (101%) 247,049 (102%) v ,554 ( 98%) 124,963 (102%) 242,517 (100%) (*100% at case (i)) Page 50

26 Observation of base case! All Buyer, Supplier and Channel increase expected profit through sharing inventory info! Even if handling cost is taken into account, expected profit of Buyer increases at around h=500! Expected profit of Supplier increases anyway If Buyer reduces handling cost, Buyer can take advantage of virtual inventory Page 51 New Business Process! Problem:! In h=700, Buyer does not have incentive to share inventory info! New Business Process Proposal:! Supplier supports part of handling cost incurred transshipping goods between stores! Incentive of Supplier:! Supplier wants to increase order quantity from Buyer! Supplier wants to have demand info at Buyer s stores [Note] This is feasible in RFID-enabled inventory control:! Supplier can get accurate number of goods transshipped between stores Page 52

27 Applied case: sensitivity to handling cost support! Deal with % of handling cost coverage by Supplier as a variable (v) with info sharing, h = 700, 0% coverage (vi) with info sharing, h = 700, 30% coverage (vii) with info sharing, h = 700, 50% coverage (viii) with info sharing, h = 700, 70% coverage (ix) with info sharing, h = 700, 100% coverage Page 53 Applied case result Order QTY Expected Profit at Buyer Expected Profit at Supplier Channel Expected Profit v ,554 ( 98%) 124,963 (102%) 242,517 (100%) vi ,065 (103%) 123,040 (100%) 247,105 (102%) vii ,900 (106%) 123,717 (101%) 251,617 (104%) viii ,595 (110%) 124,503 (102%) 256,098 (106%) ix ,794 (114%) 124,000 (101%) 260,794 (107%) (*100% at case (i)) Page 54

28 Observation of applied case! In case handling cost is expensive for Buyer to share info among stores, with the support of part of handling cost by Supplier, all Buyer, Supplier, and Channel increase expected profit from sharing inventory info among stores! Even if Supplier covers more, they do not increase expected profit. However, the demand info at Buyer will be beneficial to Supplier Page 55 Conclusion! All Buyer, Supplier, and Channel increase expected profit! If Buyer with multiple stores share inventory info among its stores (develop a virtual inventory )! Supplier does not only increases profit but gets real demand info, which can be used to improve forecast, production/logistics planning etc.! If Supplier covers transshipment cost incurred when stores exchange goods by sharing inventory info Taken advantage of the fine granular info among stores, supply chain channel has a possibility to increase channel profits Page 56

29 Realization of SCM and CRM with RFID! Physical store retailers situation! Space constraint (of course this is a strong point as well)! Difficult forecasting because of shorter product lifecycle! Competitions with both other physical stores and e-business stores! Internet-armed savvy consumers! Retailers strategies! Improve inventory turnover rate to make the most use of the store floor space (SCM: supply chain management)! Build good relationship with loyal customers (CRM: customer relationship management) Page 57 Challenges & role of RFID! Combining SCM with CRM! Challenge for retailers for long time! Differentiate members through FSP, but not loyalty basis! Critical issue: Customers do not identify themselves; too late to affect purchase decisions!! RFID fills the gap FSP: frequent shoppers program! FSP members card with RF tag and readers with a display at the store floor connects SCM and CRM Supplier MFG WS Retailer SCM RFID Customer CRM Page 58

30 New RFID equipment! RFID reader with display! Customers scan loyalty card with RF tags! Reader displays information only to the customers Photo courtesy of Dai Nippon Printing and Sears Page 59 Supporting practices! Frequent Shoppers Program (FSP)! Practice to build strong customer relationship! Rewarding customers based on customers loyalty level! Popular in airline industry and retail industry! Dynamic Pricing (DP)! Optimize revenue out of given resources! Utilize mechanism that demand of cheaper items is higher! Manipulating price to change customer demand Page 60 r(t,d): revenue, d(t): demand, C: constraint

31 System overview and flow Dynamic Pricing Engine (DPE) ) Customer DB * ' & intranet ( Inventory Management System (IMS) RFID reader with display + Loyalty Card w/rfid tag Page 61 Proposed algorithms (1)! Base Target Discount Price (BTDP) Calculation! Compute a price to achieve inventory turnover rate goal! Difference in loyalty status is not taken into account Start Get data for sales forecasting and price demand relation Get current inventory level, price, and date in sales period Forecast sales amount of the rest of the sales period Compute inventory level at the end of sales period Compute base target discount price to make the inventory level of end of sales period zero End Not Excess Excess Page 62

32 Proposed algorithms (2)! Member type Target Discount Price (MTDP) Calculation! Compute prices appropriate for each member status! Use a price computed through BTDP algorithm P BTDP : Price from BTDP P MTDP (i): Price from MTDP P List : List Price cust_class(): % of each group disc: Discount rate difference Page 63 Evaluation! Numerical study! Increase retailer s profit or not! Reward customers based on loyalty status! Control inventory turn over rate Page 64

33 Numerical study!!! Three points for evaluation A)! Increase retailer s profit or not B)! Reward customers based on loyalty status or not C)! Control inventory turn over rate successfully or not Show effectiveness of the algorithms through comparison i.! Without Dynamic Pricing ii.! With Dynamic Pricing. Without loyalty status differentiation (only difference in FSP member and non-member) iii.! With Dynamic Pricing. With loyalty status differentiation (three statuses: Platinum, Gold, and Silver; and non-member) Common settings! Poisson distribution, 3 months, 5 days/week, P List : \1,200, Disc: 4%! Demand-price curve: demand = *price! Actual demand: 120 [items/wk], Forecast range: [items/wk] Page 65 Result of numerical study ($)! Algorithms increase the profit Profit differences w/o Dynamic pricing w Dynamic pricing w/o FSP difference w Dynamic pricing w FSP difference Page 66

34 Result of numerical study (B)! Algorithms appropriately reward customers Offer low price to loyal customers Page 67 Result of numerical study (C)! Algorithms successfully control inventory level w/o Dynamic pricing w Dynamic pricing w/o FSP difference w Dynamic pricing w FSP difference Algorithms reduce the inventory level at the end Page 68

35 Page 69 Conclusion, limitations & future possibilities! Conclusion! Propose an application and two algorithms to achieve two goals together:! Increasing profits of physical store retailers by enabling them to achieve target inventory turnover rate! Rewarding FSP member customers based on their loyalty status! Connect the gap between SCM and CRM! Limitations! Privacy! Legality of the practice! Future Possibilities! More information for consumers! Not only discount but also recommendation! More information for retailers and manufacturers! Behavior information (Promotion, List price decision) Summary of this lecture! SCM 101! SCM is a promising RFID application area, since handling goods is a labor intensive and error prone process! Initial benefit of RFID in SCM is improvement of efficiency and accuracy! RFID has a potential to change SCM practices, fine granularity management Page 70