Information and Responsiveness in Spare Parts Supply Chains

Information and Responsiveness in Spare Parts Supply Chains

Table of Contents 1.0 Motivation... 3 2.0 What is Supply Chain?... 3 2.1 Spare Parts Supply Chain... 4 2.2 Spare Part Supply Chain Characteristics... 4 2.3 Service Network Design... 5 2.4 Performance Measures in Spare Parts Supply Chains... 5 2.5 Role of Information... 7 3.0 Case: IBM - Service Logistics Organization (SPO)... 7 3.1 Service Operation at IBM... 8 3.2 Spare Parts Supply Chain Management at IBM... 8 3.3 Role of Installed Base Information... 9 Additional Reading:... 9 Question 1:... 10 Question 2:... 10 Question 3.... 11 Question 4:... 12 Question 5: (See additional reading)... 12

1.0 Motivation This case study is based on the research collaboration between RSM-Erasmus University and IBM Service Logistics Organization (IBM-SPO). The objective of this research is to analyze and develop the tools and techniques that could provide information-enriched solutions to the spare parts supply chain problems. In this case study, we would first outline some of the unique characteristics of the spare parts supply chain. The spare parts supply chains are typically different from retails supply chains, since they face slow moving demand, geographically dispersed customers and network structures. These circumstances support the role of appropriate network design and customer information for the efficient management of the spare parts supply chain. In this case study, we would describe the IBM-SPO case with specific focus to information-enriched solutions for the management of IBM s geographical layout of the spare parts distribution network. 2.0 What is Supply Chain? The supply chain consists of all parties involved either directly or indirectly in fulfilling a customer request. The supply chain includes not only the manufacturers and suppliers, but also transportation, warehouses, retailers and even customers themselves. Within each organization such as manufacturer, it includes all the functions involved in receiving, filing and fulfilling customer requests, such as product development, finance, marketing, operations and customer service. Consider the case when a customer goes to a supermarket to buy detergent. The supply chain begins with the customer itself and the next stage in this supply chain is the retail supermarket. The supermarket stocks the detergent in its shelves which is delivered to it from a warehouse or distributor by using third party logistics provider. The distributor is stocked by a detergent manufacturer who has produced the detergent by using the rawmaterial from various suppliers. These suppliers in-turn may have been supplied by the lower-tier suppliers. Second Tier Supplier First Tier Supplier Manufacturer Warehouse Retailer Customer Product Flow Funds Flow Information Flow Figure 1. A Typical Forward Supply Chain

Figure 1. graphically depicts the above forward supply chain structure. A supply chain is a dynamic chain and involves the consistent flow of products, information and funds through the supply chain. In a forward supply chain, the product flow is downstream, the funds flow upstream and information usually flows both ways. Note that the above supply chain is a linear structure, however, in many practical situations the supply chain consists of a network structure with multiple suppliers serving to many manufacturing sites as well as multiple manufacturing sites associated with many retail sites of the same company. In the current case, we would focus on typical supply chain network that exists in spare part supply chains. The details are discussed in section 2.1. 2.1 Spare Parts Supply Chain Consider buying a computer from IBM. Along with the product, IBM offers a warranty contract to provide the maintenance service in case of machine failure. To provide an efficient service IBM needs to maintain the appropriate maintenance resources such as service personnel, service tools and spare parts. Since the IBM machines are sold globally, it needs to maintain the spare parts close to customer location; however, maintaining too much spare part inventories at many locations has direct impact of supply chain costs. At the same time, the allocations of inventories at far-off locations decreases supply chain responsiveness and increases transportation costs. Therefore, IBM needs to design the service network that provides the appropriate strategic fit between service responsiveness and costs. 2.2 Spare Part Supply Chain Characteristics There are many aspects in which the spare parts supply chain differs from typical retail supply chains. Firstly, each demand realization in spare parts supply chain is an independent event and it is the result of possible machine failure. Therefore the machine reliability aspects play an important role in demand realization. The spare parts supply chain typically encounters slow moving demand which provides additional complexities for demand forecasting and planning methods as most of the planning and forecasting methods have been designed for the management of large demand volumes. Typically, the machines sales vary over time and are dependent on the product s life cycle. Therefore, the past and current installed base size and distribution provides important information for demand forecasting and supply chain planning. The customers having the same machine could be dispersed over a geographical region; therefore it necessitates the design of supply chain in a suitable network configuration. Further discussion on the impacts of various network configurations is provided on section 2.3. Last, it makes a considerable difference if the maintenance activity is designed as a proactive or reactive maintenance for spare parts supply chain. In the former case, the demand and service resource requirements are more predictable as they originate from planned maintenance cycle; whereas reactive maintenance poses additional challenges for the spare parts supply chain management.

2.3 Service Network Design Typically in spare parts supply chains, the service networks are designed in hierarchical or non- hierarchical manner. Figure 2 depicts the example of such networks. The nodes depict the storage facilities (stock location) and customers and arcs denote the connections among stock locations and customers. Storage Facilities Customers Hierarchical Network Non - Hierarchical Network Figure 2. Various Network Configurations In hierarchical networks each customer is usually associated with one stock location, whereas in non-hierarchical case a customer can be serviced from many stock locations. The network configuration has direct impact on the supply chain responsiveness / costs as the hierarchical networks typically yield less responsiveness vs. non-hierarchical case. The choice of appropriate network is dependent on the strategic fit of cost vs. High responsiveness required by the supply chain. Figure 3 depicts the relationship between supply chain costs and supply chain responsiveness. For example, the need of response in automotive spareparts supply chain may not be as quick as a computer OEM who sells its product to business customer. In the Low coming section we would discuss High Low Cost various performance measures used in spare parts supply chains to determine the responsiveness of various network Figure 3. Cost Responsiveness Efficient Frontier configurations. 2.4 Performance Measures in Spare Parts Supply Chains Since the spare parts supply chains are designed to provide the maintenance service to the customers, therefore, beside costs, the primary performance measure is the ability of supply chain to provide the customer service in responsive manner. However, as depicted in figure 3, both of these performance measures are typically conflicting in nature and higher responsiveness usually incurs higher costs and vice versa. In literature, an important term to measure the supply chain responsive is the customer service level which is measured by various performance indices such as: Responsiveness

Order Fill Rate: The first one is the order fill rate, which is the fraction of customer demands that are met from stock. The order fill rate could be with respect to a central warehouse or a field warehouse or a stock location at any level in the system. Fill Rate of individual location i (fr i ) = 1 Stock Out Rate at location i The network fill rate from the customer perspective would simply be the average of all field stock locations of the network which are able to supply the items to the customers. Example: Part I: Assume there are two independent Stock Locations which are serving to their own customers individually. The individual fill rates of the stock locations are 0.9 and 0.85 respectively. Determine the Network Fill rate Solution I: Network Fill rate = n i=1 fr i n = (0.90 + 0.85)/2 = 0.875. Part II: Now let s assume that in case of a stock out at the primary location i (closest location to the customer), the customer can be served from the other location j. Now update the Network Fill rate: Solution II: Fill rate for location i customers = fr + ((1 fr ) fr ) i i j Fill rate for location 1 customers = 0.90 + ((1-0.90) 0.85) = 0.985 Similarly for location 2 customers = 0.985 Network Fill rate = 0.985 Stock-Out Rate: Stock out rate is the complement of fill rate and represents the fraction of orders lost due to a stock out. If a customer is served from only one location. Stock Out Rate of location i = ESC / Q ESC = Expected Shortage per Replenishment Cycle Q = Economic Order Quantity Back Order Level: Another measure is the backorder level, which is the number of orders waiting to be filled. To maximize customer service level, one needs to maximize order fill rate, minimize stock out rate, and minimize backorder levels. Probability of on-time delivery: probability of on-time delivery is the fraction of customer orders that are fulfilled on-time, i.e. within the agreed-upon due date.

2.5 Role of Information Potentially the biggest driver in supply chain performance is the information about products, processes, customers, and suppliers. Some of the aspects of information that trigger supply chain performance are its accuracy, accessibility and relevance. 1. Information must be accurate: Without information that gives the true picture of the state of the supply chain, it is very difficult to make good decisions. That is not to say that information should be 100% correct, rather the data available should paint a picture that is at least directionally correct. 2. Information must be accessible in timely manner: Often accurate information exists but by the time it is available, it is either out of date or, if it is current then it is not in an accessible form. To make good decisions, a manager/planner needs good information that is easily accessible. 3. Information must be of a right kind: Decision makers need information that they can use. Often companies have large amounts of data that is not helpful in making a decision. Companies must think about what information should be collected so that the valuable resources should not be wasted in gathering meaningless information. The information enables the supply chain managers to make decisions that directly impact its responsiveness and efficiency. It provides the customer proximity and demand information to the supply chain managers which is used to design and plan the supply chain in an efficient way. The various sources of information such as installed-base information and request handling database provide the product distribution and its usage history information to the supply chain managers which could be used in designing and managing a spare parts supply chain in an optimal manner. 3.0 Case: IBM - Service Logistics Organization (SPO) IBM is a computer manufacturer with various products for business and individual customers. These products range from high end server machines to low end products. In this case, we would focus on the service logistics operations for IBM s business customer. These machines are sold to business organizations in all sorts of industries and due to their critical position in IT infrastructure; it is a requirement to have high availability of these machines. IBM sells these machines with expensive and time critical service / warranty contracts, which could bind IBM to provide service (in case of machine failure) in less than 2 hours. The service range of service deadlines in these contracts is 2 hrs, 4 hrs, 8 hrs, or next day service. To provide service at such a short notice, IBM needs to maintain efficient service operations. The IBM s Service Logistics Organization is responsible for maintaining spares for the spare parts supply chain in this regard.

3.1 Service Operation at IBM Once the machine fails at the customer location, the service is performed by a closely coordinated operation between service organization and spare parts organization (IBM- SPO). Figure 4 depicts the sequence of a field service operation. Based on remote and onsite diagnostics, the part requests are initiated by the CE (Customer Engineer). This part is delivered form the nearest stock location where it is being maintained according to the stock levels determined during the planning cycle. Service Response Time Inventory Response Time Logistics Delay Time Remote Diagnostics On-site Diagnostics On-Site Repair Machine Fails CE Arrives Parts Arrive Customer Calls CE Order Parts if Necessary Repair Job Complete Fig 4. Sequence of Maintenance Operation at IBM 3.2 Spare Parts Supply Chain Management at IBM The IBM s service logistics network spreads across the globe. For efficient functioning, it has been subdivided into various GEOs (i.e. geographical area). Each of the GEOs is responsible for the availability and delivery of spare parts within that geographical region. The EMEA (Europe, Middle-East and Africa) GEO delivers spare parts to its local customers by means of an inventory network throughout the EMEA GEO. This network is managed by the IBM-SPO organization and it has about 150 stock locations (field location) within EMEA region. Fig 5 depicts the geographical spread of IBM s customers in Germany, Austria and Switzerland for a specific machine model. The 2 hrs CRS (Committed Recovery Service) denotes the customer having a 2 hrs committed service contract. Similarly, there are 4 hrs and 8 hrs committed service contract customers in the region. The stock locations are dispersed throughout the region. The supply chain management operations at IBM-SPO include demand forecasting cycle followed by an inventory planning cycle and subsequently the supply chain execution. As the customers are spread throughout the geographical region, therefore IBM-SPO uses the historical usage as well as customer location information (installed base) to predict the regional demand forecasts. The inventory planning for central and field location is performed by using specific optimization algorithms that determine the appropriate stocking levels ate all locations. The execution stage also attempts to minimize the transportation costs by choosing the appropriate stock location for service request fulfillment.

Legend: Legend: IBM Stock Location IBM Stock Location 2 hour CRS 2 hour CRS 4 hour CRS 4 hour CRS 8 hour CRS 8 hour CRS Base Maintenance + Base Base Warranty Maintenance + Base Warranty Machine Installed Base Switzerland Fig 5. Geographical Spread of IBM Customers 3.3 Role of Installed Base Information The role of installed base information for service logistics planning at IBM could not be undermined. It allows IBM to be more customer-focused as it provides the IBM with the location specific information about the customer. IBM encounters slow moving demand in their geographical service network which necessitates the attainment of demand s geographical spread and density information for improved planning. The transportation requirements and associated cost information is also deduced by using the installed base information. This also highlights the criticality of the planning system on information quality. Additional Reading: M. Jalil, R. Zuidwijk, H. Krikke, Yesterday proactive, responsive today - Use of Information to enhance planning in closed loop supply chains. Environment Conscious Manufacturing, CRC Press, Taylor & Francis (to be published in winter 2007).

Question 1: Discuss the following statement Slow moving demand and strict service levels induce relatively high inventory costs. Question 2: The current IBM management is considering various options to redesign their supply chain to improve the responsiveness in the network. Let s assume that the current network is composed of 4 independent field stock locations with each one serving their customers in their own geographical area. The fill rates for all locations are given in Table 1. Stock Location Fill Rate Stock Out Rate 1 0.90 0.10 2 0.85 0.15 3 0.92 0.08 4 0.89 0.11 Table 1. Fill Rate of Various Stock Locations Part 1. Determine the network fill rate. Part 2. A planner at IBM recommends that IBM should change the field stock location 2 to the status of central stock location. In other words, it should not serve the customers directly but support the field stock locations in case they are out of stock. The network configuration (which is now hierarchical) is depicted in Figure 6. Now calculate the network fill rate. 2 1 3 4 Customers Customers Customers Figure 6. Modified Hierarchical Network Part 3. A recent MS supply chain management graduate who joined IBM-SPO planning department recommends that IBM should attempt to design a non-hierarchical network. In other words, the central location should not be introduced, but the customers should be

allowed to be served from any location as depicted in Fig 7. Now calculate the network fill rate. 1 2 3 4 Figure 7. Non-Hierarchical Network Part 4. Discuss the results of part 1, 2 & 3 of this exercise. Question 3. Customers Customers Customers Customers Consider the Supply Network in Figure 8 that consists of three stock locations and various customer locations. Part I: At this point, due to the limited information available about the customers, the company does not know the total number of its customers and their locations. The forecasting department, however, based on historical demand data predicts that in the next period, there would be demand of total 6 units. How would you allocate the 6 units intuitively to each of the stock locations? Also consider if the demand is uncertain; then how would it impact the original stocking decision? Support your judgment with arguments. (0,10) (10,10) 3 Y Coordinates 1 2 (0,0) (10,0) X Coordinates Figure 8. Supply Network without customer information (note one small square = 10 km * 10 km).

Part 2: suppose from Installed base information you are able to acquire the customer location information and by contacting your logistics provider, you are able to acquire the possible delivery routes (as shown in Figure 9) and per unit transportation cost i.e. 10 / km. Now, assign the total 6 units to each stock location (assume each customer location has equal contribution to demand forecast). Interpret your results. (0,10) (10,10) 3 Y Coordinates 1 2 (0,0) (10,0) X Coordinates Figure 9. Supply Network with customer information (note one small square = 10 km * 10 km). Question 4: In section 2.5, it has been discussed that information quality could have impact on supply chain performance. Which network topology (i.e. hierarchical or non-hierarchical topology) would be more susceptible to degradation in information quality? Why? Question 5: (See additional reading) Discuss the situation if IBM introduces the proactive maintenance policy instead of its current reactive maintenance policy. What would be the implication of such policy on the supply chain operations (such as forecasting and planning)? What type of information does IBM need to introduce such policy? What could be the different sources/methods to acquire such information? Discuss the pros and cons of each of these sources/methods?