A Computationally Efficient Heuristic for a Two-Echelon, Two-Indenture Inventory Model.

Abstract

This project addresses the computational difficulties of MOD-METRIC, a two-echelon, two-indenture recoverable item inventory model which is used by the Air Force Logistics Command (AFLC). As background, the theoretical development and current solution technique of MOD-METRIC are presented. After identifying the major causes of the computational difficulties, a different approach to solving the MOD-METRIC problem, based on recent work by Kotkin and Muckstadt, is presented. The final form of the efficient heuristic proposed in this project incorporates a new estimation procedure for the important parameters of this heuristic, and also includes an expanded search over those parameters. The results of comparing heuristic solutions with solutions obtained using the current MOD-METRIC system indicate that the heuristic provides solutions that are very close to optimal at a greatly reduced computational cost for a variety of MOD-METRIC problems. An additional benefit of the heuristic is that it generates many more solutions in a single computer run than does the current system. It is recommended that users of the current MOD-METRIC system use the implemented heuristic, after gaining confidence in its accuracy. Additionally, it is hoped that this efficient heuristic will provide an incentive for AFLC to incorporate a two-indenture model into the AFLC recoverable item inventory management system. (Author)

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 1980
Accession Number
ADA095113

Entities

People

  • William H. Hughes Jr

Organizations

  • Air Force Institute of Technology

Tags

DTIC Thesaurus Topics

  • Accuracy
  • Air Force
  • Computers
  • Inventory
  • Logistics
  • Measurement
  • Metric System
  • Motivation
  • Supply Chain
  • Supply Chain Integrity
  • Supply Chain Management

Fields of Study

  • Computer science

Readers

  • Logistics and Supply Chain Management.
  • Operations Research