1 Refer to PowerPoint slides on Steel Works case ---------------------------------------------------2.

. How can firms cope with huge variability in customer demand? Companies can cope with uncertainty by 1. keeping safety stock, 2. shortening production and order lead times, 3. using risk pooling strategies, 4. delaying product differentiation in the supply chain as much as possible, i.e., aggregating demand for parties upstream of the supply chain, and 5. by installing systems to achieve information sharing between suppliers and buyers, thus enabling collaborative demand forecasting. --------------------------------------------------3. What is the relationship between service and inventory levels? In general, higher inventory levels make it easier to maintain higher service levels. However, modern inventory management techniques may make it possible to increase service levels without increasing inventory levels as much as in the past. -----------------------------------------------4. What is the impact of lead time, and lead time variability, on inventory levels? The variability in demand increases as the average and the variance of lead time increase. Therefore, for a given service level, inventory le. Answer the following questions about the case at the start of the chapter. a. Based on the spreadsheet data, how would you characterize Steel Works products? What about Steel Works customers? Given your answer and the information in the case what does this suggest? b. What does the coefficient of variation tell us? Can you determine the coefficient of variation for the DuraBend and DuraFlex product lines? c. How much inventory has Steel Works been holding? How much should they have been holding? d. Although no data is given for the Custom Products division, are there any obvious opportunities that are suggested by the information in the case? a. Steel Works Steel Works

b. DuraBendTM DuraFlexTM c. Steel Works d. vels increase with longer lead times and higher lead time variance. ------------------------------------------------5. What factors should management consider when determining a target service level? The target service level depends on the mission-criticality of the product. For instance, consider a service parts vendor for equipment for which every hour of down time is very expensive. In this case, we would expect the management of the vendor company to specify a service level close to 100%. Market conditions also play an important role in determining target service levels. For commodities, we would expect relatively high service levels since customers can switch products easily if they do not find the particular product they look for. However, a lower service level may be acceptable if the product has a clear value differentiation compared to its competitors. while the expected level of inventory immediately after receiving the order is 2.2.6 (Q, R)

The reorder level s = L * AVG + z * STD * L has two components. The first component L * AVG covers the expected demand during lead time, and the second component z * STD * L is safety stock that protects against deviations from the expected demand during lead time. Therefore, immediately before th For instance, customers of a high-end server that is clearly deemed superior to the rest of the market may be willing to wait for 1-2 weeks if the manufacturer is out-of-stock. --------------------------------------------------6. Consider the (Q, R) policy analyzed in Section 2.2.6. Explain why the expected level of inventory before receiving the order is e order arrives, we expect that the first component is depleted completely and the inventory level is z * STD * L . Then, when an order of Q units arrives, the expected level of inventory is Q + z * STD * L . -----------------------------------------7. Consider the base-stock policy analyzed in Section 2.2.8. Explain why the expected

level of inventory after receiving an order is equal to while the expected level of inventory before an order arrives is 2.2.8

In the base-stock policy, at the time the warehouse places an order, this order raises the inventory position to the base-stock level (r + L) * AVG + z * STD * r L . Similar to the reorder level s in the continuous review policy discussed in Question 5, this base-stock includes two components: the average demand (r + L) * AVG until the order arrives after r + L periods, and the safety stock z * STD * r L that protects against demand uncertainty during lead time. Thus, just before an order arrives, the expected inventory on have is equal to the e a decision depending on the relative effects of these two factors. In addition, your decision would ultimately depend on the requirements of the retailers customers. ---------------------------------------11. Although we typically model inventory-related costs as either fixed or variable, in the real world the situation is more complex. Discuss some inventory-related costs that are fixed in the short term but may be considered variable if a longer time horizon is considered. For a mature product, it is reasonable to expect that the price and demand are stable in the short term. However, as the time horizon gets longer, and new products are introduced into the market, the demand and price for this particular product decrease and excessive inventories may have to be written off. Thus, inventory holding costs related to obsolescence may be regarded as fixed in the short term, but not in the long term. Some storage costs are another example of inventory related costs fixed in the short term, but variable in the long term. For instance, due to large inventories as company may have to rent multiple warehouses for a fixed lease term. However, if inventory policies are improved and turnover rates are increased in this period of time, then it may be possible to rent fewer warehouses when renewing the lease contracts. Clearly, similar arguments can be made for material handling equipment, stsafety stock z * STD * r L . In order to determine the expected inventory level right after an order arrives at time t + L, note that when inventory is reviewed at time t, the inventory position is raised to the base-stock level, and an order that was placed at time t - r arrives at time t + L. (See Figure 3-12.) Therefore, when an order arrives, the expected inventory level is L * AVG units less

than the base-stock level, i.e., is equal to r * AVG + z * STD *

r L.

-----------------------------------------8. Imagine that you operate a department store. List five products you sell, and order them from lowest target service level to highest target service level. Justify your ordering. 5 Possible items for discussion are as follows: 1. Magazines order quantity has to be the average demand for a magazine that the store expects to sell. The service level is low and the store would not order a large quantity as unsold magazines would be worthless 2. BestSellers order quantity would be slightly higher than average demand as the store can expect higher than average demand for such books 3. CDs order quantity would be higher than average demand for the titles where the store expects a higher demand. Otherwise the service level will be low for infrequently sold CDs. 4. Potato Chips order quantity typically would cover expected demand over a number of days. Store needs a high service level for such items and the longer shelf life of the products allows a larger order quantity. 5. Milk service level is high and hence order quantity would have to meet a high percentage of expected demand; although milk is perishable, this is managed through frequent deliveries, which allows the store to only match demand with available inventory over a few days. ------------------------------------------------------9. Consider a supply chain consisting of a single manufacturing facility, a crossdock, and two retail outlets. Items are shipped from the manufacturing facility to the cross-dock facility and from there to the retail outlets. Let L1 be the lead time from the factory to the cross-dock facility and L2 be the lead time from the crossdock facility to each retail outlet. Let L = L1 + L2. In the analysis below, we fix L and vary L1 and L2. a. Compare the amount of safety stock in two systems, one in which lead time from the cross-dock facility to a retail outlet is zero (i.e., L1 = L and L2 = 0) and a second system in which the lead time from the factory to the cross-dock facility is equal to zero (i.e., L1 = 0 and L2 = L). b. To reduce safety stock, should the cross-dock facility be closer to the factory or the retail outlets? For this purpose, analyze the impact of increasing L1, and therefore decreasing L2, on total safety stock. L1 L2 LL1 L2L L1 L2 a. (L1LL20) (L10L2L)

b. L1 L2 Observe that the longer L1, the more time the system has before allocation of inventory to the retailers need to be made by the cross-dock facility. Thus, the longer L1 the more the system can take advantage of the risk pooling concept. Hence, the total amount of inventory is smaller when the cross dock facility is closer to the retail outlet. -------------------------------------10. Suppose you are selecting a supplier. Would you prefer a supplier with a short but highly variable delivery lead time or a supplier with a longer but less variable lead time? The answer is not immediately clear because the required safety stock depends both on the average and on the variance of the lead-time. The retailer would have to makorage racks, insurance, personnel, etc. -----------------------------------------------------12. When is a model such as the economic lot sizing model, which ignores randomness, useful? Such deterministic models can be used as proxies for the more realistic stochastic models if the planning horizon is short, and the parameters of the problem are expected to be relatively stable over this time frame. However, most importantly, simple models can illustrate the basic trade-offs in a given type of problem which also translate into more realistic and complex situations. For instance, the optimal policy for the economic lot sizing model balances ordering and inventory holding costs which is a general insight for more sophisticated systems as well. -------------------------------------------------13. What are the penalties of facing highly variable demand? Are there any advantages? There are implicit and explicit penalties associated with a highly variable demand. For instance: 1. As discussed, the level of safety stock is proportional to the variability in demand, i.e., the higher the variability in demand the higher the inventory holding costs. 2. From a manufacturers perspective, highly variable demand means that utilization of equipment will greatly fluctuate, and equipment will sit idle when demand is low. 3. From a managerial perspective, high variability makes planning a very complex task that requires additional resources, sophisticated models and tools. On the other hand, if a company is successful at implementing strategies to cope with high

variability in demand, it may be possible to leverage on these to increase market share and/or revenue if the competitors are not as successful. ---------------------------------------------14. Give a specific example of risk pooling (a) across locations, (b) across time, and (c) across products. (a) (b) (c) a. Risk pooling across locations: combining several warehouses into a single central warehouse. b. Risk pooling across time: using quarterly demand forecasts instead of monthly forecasts to do capacity planning. c. Risk pooling across products: designing products with maximum commonality and delaying product differentiation in the supply chain as much as possible. -----------------------------------------------15. When would you expect demand for a product in two stores to be positively correlated? When would you expect it to be negatively correlated? If pricing strategies, service levels and quality of service in two stores are similar, then we would expect the demand in these two stores to be positively correlated. However, assume that while the overall market demand is relatively stable, one of the stores is running a promotion. In this case, we would expect that the promotional campaign would steal sales from the other store, so that the demand in the two stores would be negatively correlated. --------------------------------------------16. Consider the first Walkman model introduced by Sony. Discuss which forecasting approach would be most useful at the start, in the middle, and toward the end of the product life cycle. Now, consider a more recent Apple iPod model. How would your assessment of the appropriate forecasting techniques change? (Sony) WalkmanTM iPod In the absence of historical data, judgment and market research methods would be most useful at the beginning of the life cycle of the first Sony WalkmanTM. As more data becomes available during the product adoption phase, time-series methods could be employed successfully. Then, as the product matures and the manufacturer has a better understanding of the factors that affect demand, both time-series and causal methods could prove effective. When introducing a more recent Walkman model, we would expect Sony to rely much less on judgment and market research methods compared to the very first model. Years of experience, knowledge and data can be used to develop accurate quantitative time-series and causal models. ---------------------------------------------------

17. Technical question: KLF Electronics is an American manufacturer of electronic equipment. The company has a single manufacturing facility in San Jose, California. KLF Electronics distributes its products through five regional warehouses located in Atlanta, Boston, Chicago, Dallas, and Los Angeles. In the current distribution system, the United States is partitioned into five major markets, each of which is served by a single regional warehouse. Customers, typically retail outlets, receive items directly from the regional warehouse in their market. That is, in the current distribution system, each customer is assigned to a single market and receives deliveries from one regional warehouse. The warehouses receive items from the manufacturing facility. Typically, it takes about two weeks to satisfy an order placed by any of the regional warehouses. Currently, KLF provides their customers with a service level of about 90 percent. In recent years, KLF has seen a significant increase in competition and huge pressure from their customers to improve the service level and reduce costs. To improve the service level and reduce costs, KLF would like to consider an alternative distribution strategy in which the five regional warehouses are replaced with a single, central warehouse that will be in charge of all customer orders. This warehouse should be one of the existing warehouses. The company CEO insists that whatever distribution strategy is used, KLF will design the strategy so that service level is increased to about 97 percent.

Answer the following three questions: a. A detailed analysis of customer demand in the five market areas reveals that the demand in the five regions is very similar; that is, it is common that if weekly demand in one region is above average, so is the weekly demand in the other regions. How does this observation affect the attractiveness of the new system? b. To perform a rigorous analysis, you have identified a typical product, Product A. Table 2-11 provides historical data and includes weekly demand for this product for the last 12 weeks in each of the market areas. An order (placed by a warehouse to the factory) costs $5,550 (per order), and holding inventory costs $1.25 per unit per week. In the current distribution system, the cost of transporting a product from the manufacturing facility to a warehouse is given in Table 2-12 (see the column Inbound). Table 2-12 also provides information about transportation cost per unit from each warehouse to the stores in its market area (see the column Outbound). Finally, Table 2-13 provides information about transportation costs per unit product from each existing regional warehouse to all other market areas, assuming this regional warehouse becomes the central warehouse. Suppose you are to compare the two systems for Product A only; what is your recommendation? To answer this question, you should compare costs and average

inventory levels for the two strategies assuming demands occur according to the historical data. Also, you should determine which regional warehouse will be used as the centralized warehouse. c. It is proposed that in the centralized distribution strategy, that is, the one with a single warehouse, products will be distributed using UPS Ground Service, which guarantees that products will arrive at the warehouse in three days (0.5 week). Of course, in this case, transportation cost for shipping a unit product from a manufacturing facility to the warehouse increases. In fact, in this case, transportation costs increase by 50 percent. Thus, for instance, shipping one unit from the manufacturing facility to Atlanta will cost $18.Would you recommend using this strategy? Explain your answer.

KLF KLF 2 KLF 90% KLF KLF

 97% a. b. A 2-11 12 5,550 1.25 2-12 2-12 2-13 A c. UPS 3 0.5 50% 18

a. The benefits of risk pooling increase as the correlation between demands decrease. Therefore, similarity of demand across the five regions makes the proposed system less appealing. b. The total cost of the decentralized system is $9,272 per week. In the centralized system, LA is the best location with the total cost of $6,545 per week. Please refer to the attached spreadsheet Chapter_2_Question_17.xls for details. c. In this case, the minimum total cost is $8,808 per week, and the central warehouse is located in LA. In other words, the decrease in the inventory holding costs due to the decreased lead time between the manufacturing facility and the warehouse is more than offset by the increase in the transportation costs. Please refer to the attached spreadsheet Chapter_2_Question_17.xls for details.