Jan 1, 1995
Evaluations of Alternative Approaches
Over the past twenty-five years, the Air Force has rationalized its means of managing the repair and allocation of recoverable spare parts used for aircraft maintenance. In the late 1970s, the Air Force implemented D028, a "central stock leveling" system that allowed for global control of stockage objectives. Although this approach reduced inefficiencies for over a decade, it was not easily adaptable to Operation Desert Storm and was consequently dropped in favor of the traditional Standard Base Supply System (SBSS). As a result, stockage objectives at individual bases were once again set locally, without any reference to systemwide spares requirements and aircraft availability goals.
In a recent evaluation of alternative approaches to setting stockage objectives, RAND researchers Louis W. Miller and John B. Abell demonstrate that most well-known central stock leveling methods are more effective than the SBSS. DRIVE (Distribution and Repair in Variable Environments) is a computer-based system used to determine repair priorities and base allocations for many aircraft parts. The authors recommend that DRIVE's role be extended to determine stockage objectives for these same parts. In cases where DRIVE is not already in use, an upgraded version of D028 would probably be the most effective stock leveling choice.
The authors also discuss several ways of ensuring that DRIVE is used to maximum effectiveness, focusing primarily on an approach that minimizes certain inconsistencies and on the need to adjust for the fact that DRIVE and similar models tend to favor the requirements of large bases and simpler weapon systems.
Central stock leveling is the determination of stockage objectives by a central authority, such as an air logistics center, that provides spare parts for aircraft located at various bases. Determining stock levels at such a center offers two important advantages over local computation of stockage needs at each base. First, it gives decisionmakers an overview of assets and requirements that allows for effective management of spare parts according to systemwide priorities. Second, it sets up an automatic process that ensures that unfilled requisitions are kept to a minimum.
The authors evaluated the performance of several models for determining parts requirements and allocating stock levels. These included METRIC (and VARI-METRIC), the Aircraft Availability Model, and DRIVE—models that share many assumptions and the same abstract view of the logistics system. All of these approaches performed at nearly equal levels in the evaluation, but all yielded stockage objectives more effective than those of the SBSS.
The authors recommend DRIVE, which is already performing logistics functions for various spare parts, as the logical choice for central stock leveling. The recommended version is Desktop DRIVE, which limits the scope of operations enough to allow for human intervention and manageable coordination. Periodically, whenever there is a change in force beddown, flying-hour program, mix of aircraft among bases, or other important factor, DRIVE recomputes the stockage objectives for both LRUs (line-replaceable units) and SRUs (shop-replaceable units). Desktop DRIVE is now being demonstrated for stock leveling at the Ogden Air Logistics Center for F-16 avionics components.
In analyzing DRIVE's performance, the authors observed that DRIVE occasionally recommends allocations of assets to bases that have no unfilled requisitions. DRIVE also sometimes ignores existing requisitions for some time. The authors claim that these inconsistencies can be minimized by computing stockage objectives centrally with a simple adaptation of DRIVE similar to the one they constructed for the purpose of comparing DRIVE with other models and approaches.
Unless DRIVE is already being used for prioritizing repairs and allocating assets, it should not be employed simply for determining stock levels. In such cases, the Air Force Materiel Command's Central Stock Leveling System, D028, seems to be the best approach. While D028 may have had data or data-processing problems that inhibited its effectiveness, its underlying logic is sound. D028 is clearly superior to the SBSS and other approaches that do not have an overview of the total system, and should again be made a viable system for central stock leveling.
Optimization models such as METRIC, the Aircraft Availability Model, and DRIVE determine stock levels according to a systemwide objective function (such as the expected availability of aircraft worldwide). The authors found that these models typically favored the needs of large bases over smaller ones and simpler weapon systems over more complex ones (where complexity is measured in terms of the number of LRUs making up the weapon system). In analyzing this "heterogeneity effect," the authors found that special attention is required from decisionmakers.
The figure illustrates the problem. The three bases shown by the black bars have simpler aircraft, comprising only 54 LRUs. The gray bars represent more complex aircraft with 108 LRUs. The bases also differ in the number of aircraft they possess, varying from 18 to 36 to 72 primary authorized aircraft (PAA). The differences in the percentage of aircraft that are unavailable for operational use underscores the substantial effects of PAA and weapon system complexity on the determination of stockage objectives by models such as METRIC and DRIVE.
In the process of setting performance goals for DRIVE and similar models, decisionmakers need to understand that such distortions exist and should consider setting stockage goals by individual base and weapon system rather than applying them globally. Without such adjustments, some commanders may feel that the system does not properly account for their needs.
Relying on the SBSS and negotiated stock levels leads to inadequate performance levels. These approaches, therefore, should be replaced by DRIVE (where appropriate) or by an improved version of D028. DRIVE offers great promise as a decision support system and its potential should be further explored. Evaluations of DRIVE with SRUs, not included in this study, could quantify the advantages of DRIVE's surgical treatment of SRUs and show the system to full advantage. Additionally, the authors recommend a review of the ways in which the Air Force sets aircraft availability goals for models like DRIVE and the development of an expected availability objective function to ease DRIVE's implementation. Taking full advantage of DRIVE for central stock leveling would contribute significantly to the Air Force's goal of providing adequate levels of aircraft spare parts at least cost.