Click for next page ( 42


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 41
4 OPTIMIZATION OF MATERIEL MANAGEMENT LIMITATIONS OF THE CURRENT SYSTEM The Defense Logistics Agency's (DLA) Standard Automated Materiel Management System (SAMMS) was designed in the 1960s and implemented mainly in the early 1970s. Although SAMMS generally permits the DLA to provide good service to its customers, SAMMS is not very efficient. Indeed, SAMMS makes very little attempt to achieve the integrated optimization of procurement and distribution decisions that is envisioned in DLA's Strategic Plan. For example, the size of order quantities for about one-half of DLA's two million items are made by the Harris square-root formula which dates to 1915. For each item, this single-variable optimization model trades off the fixed cost incurred with every purchase against its unit storage cost, assuming that the demand rate for the item is independent of time. This model has many limitations, including the following: It does not exploit the scale economies that are present in procurement because of lower prices that are often available with larger buys. It does not consider the opportunities to buy the item at times when there are favorable prices. It does not take account of the fact that item demands are both uncertain and time-dependent. It does not take account of the shelf life of the item. It does not consider existing item inventories nor the possibility of disposal of the item. It does not consider the location and amounts of the item stocked throughout the system. It does not provide for coordinating the ordering decisions of the item with other items to provide for support of weapon systems, annual procurement budget limitations, warehouse space limitations, or substitution of one item for another. 41

OCR for page 41
42 Although SAMMS does consider some of the above issues in some way, it often does so in a piecemeal, rather than in an integrated, fashion. The result of the piecemeal approach is suboptimization of the system as a whole. Here are two examples: . . A half-hearted attempt is made in SAMMS to take account of scale economies in procurement by giving buyers of some items the opportunity to ask vendors to report price breaks. But even if this information is provided by the vendor (which is not required) and reported back to the item manager, the latter is provided with no tools to calculate the optimal order quantity in light of the price breaks. The Harris square-root formula will not do this. SAMMS focuses on item fill rates to measure service. But this ignores the fact that if a system needing the item is down, it really does not make much difference whether the system is down because one or several parts are inoperative. Yet focusing only on item fill rates as the measure of service obscures this fact. Many other examples could be given. INTEGRATION OF MATERIEL MANAGEMENT WITH LARGE-SCALE OPTIMIZATION Research at universities during the last three decades has made major advances in developing models and optimization techniques for large-scale inventory systems. For example, a multi-echelon generalization of Harris' square-root formula has been found that coordinates the times and sizes of orders at each stocking point, that guarantees an average cost within 2 percent of the minimum possible, and for which computations can be done very rapidly for very large numbers of stocking points. For example, for the case of one warehouse supplying many retailers, the most difficult part of the computations requires only sorting a list of numbers -- two for each stocking point. Some of this research has been applied in recent years to achieve integrated optimization of materiel management at several large firms that simultaneously reduced investment in inventories and improved customer service. Here are some examples. . The Chicago Pneumatic Tool Corporation used large-scale optimization to reduce inventories by 43 percent while increasing the average fill rate from under 80 percent to over 90 percent in a two-echelon inventory system with about 40,000 stocking units and uncertain demands.

OCR for page 41
43 One of the world's largest manufacturers of reprographics equipment used large-scale optimization to reduce inventories by 80 percent while increasing average fill rate from under 65 percent to 89 percent in a four-echelon inventory system with tens of thousands of part types and uncertain demands. One of the world's largest computer manufacturers used large-scale optimization to minimize inventories in a multi-echelon inventory system subject to lower bounds on computer and part fill rates with uncertain demands. We believe that it is now possible and timely to develop and implement a much more efficient materiel management system than DLA now has in which large-scale optimization is used to effectively integrate supply decisions. The DLA has effectively used large-scale optimization in a limited way for a number of years, for example, to evaluate bids for fuels (which account for 42 percent of the DLA's total dollar investment). But experience in industry suggests that a DLA materiel management system that makes much broader use of large-scale optimization to integrate supply decisions is likely to increase system availability and permit significant reduction in inventories and procurement costs, particularly if visibility of military service inventories (i.e., information on the amounts and locations of such inventories) is provided to the DLA. Such a system should have the following features among others: Exploits scale economies in procurement and opportunities to buy at favorable prices to reduce average procurement costs while rationing procurement dollars among items to keep within budget limitations. Coordinates decisions on inventory levels of items at the various locations, including vendors, and redistribution of stocks to improve service and reduce inventories, while respecting space limitations. Consolidates shipments and selects economical carriers to reduce transportation costs. Provides effective methods to supply rapid mobilization. ADVANCES IN LARGE-SCALE OPTIMIZATION One of the developments making possible the use of large-scale optimization to achieve integrated materiel management is the rapid improvement in methods for solving large-scale linear, integer, nonlinear, dynamic and network optimization problems in recent years. These developments, as well as the continuing increase in the power of computers, mean that the possibility of coordinating even as large a system as DLA's may be within reach.

OCR for page 41
44 OPERATIONS RESEARCH AND ECONOMIC ANALYSIS OFFICE Presentations by senior staff of the DLA to the committee in August 1988 revealed that an important goal of the DLA was to optimize its use of resources in providing the best possible service to its customers. However, it did not appear that the DLA senior staff at that meeting clearly understood the role that methods of mathematical optimization might play in helping the DLA optimize its use of resources. For that reason, a subcommittee of this committee met with the leadership of the Operations Research and Economic Analysis Office and several individuals with expertise in the use of large-scale optimization in industrial and military supply systems in mid-February of 1989. The purpose of the meeting was to discuss the extent and appropriateness of the use of optimization within the DLA. That meeting revealed that optimization was indeed used when appropriate, along with other tools, to address a number of important problems within the DLA. Indeed, large-scale optimization is being used by the DLA to evaluate bids for fuels. One of the outside experts attending the meeting had consulted with the DLA in the development of large-scale optimization models about five years earlier. But very few such efforts were currently under way to expand the use of large-scale optimization to improve integrated materiel management in the DLA. Moreover, the DLA did not appear to have any significant in-house expertise in large-scale optimization. Instead, it appeared to the subcommittee that emphasis was now being placed by the DLA on optimization by means of personal computers (PC). PCs no doubt have a use for this purpose, especially to provide local distributed capability to interact with more powerful systems. However, the magnitude of the DLA's materiel management problems far exceed the capabilities of even the fastest PCs. For this reason, mainframes are also needed to do the kinds of large-scale optimization that we think the DLA should be doing. Accordingly, we believe that the DLA should acquire some in-house capability in large-scale optimization, both personnel and software. Moreover, and especially as long as its in-house capability in large-scale optimization is less than it should be, we recommend that the DLA make greater use of consultants and advisory committees to provide help in this area. In fact we believe that the DLA could benefit from long-term advisors in large-scale optimization and logistics research more generally. Such advisors should include a wide range of talents, both practitioners and researchers. The first are needed to assure that recommendations are practical and the second to assure that opportunities to take advantage of advances in logistics and large-scale optimization research are not lost. The committee suggests that the level and location of the Operations Research and Economic Analysis Office within the DLA may be too remote from the director and in need of review. Two observations lead to this. First, while it appears to us that the sensor members of the

OCR for page 41
45 Operations Research and Economic Analysis Office have a good understanding of the proper role and use of important tools 1ike optimization, simulation and expert systems in helping the DLA perform its mission, it appears to us that the senior staff of the DLA are not as well informed in this area. This information gap might diminish if the Operations Research and Economic Analysis Office were placed at a higher level within the DLA or if any other means were chosen to increase the visibility and use of this expertise in its senior management decision making. Second, while many of the ways in which we would like to see the models of the DLA's materiel management system improved are already either under study or planned for future study by the Operations Research and Economic Analysis Office, some members of the committee felt that studies by the office were not playing as important a role within the DLA as they should. This problem might also be alleviated if the office were located closer to the director than its present site in Richmond, Virginia. However, we have not made an organizational study, and we recognize that alternative approaches to accomplish this could certainly be developed, some no doubt superior to those suggested. The important point is that the information gap between senior DLA management and the Operations Research and Economic Analysis Office be closed.

OCR for page 41