Robust Warehouse Storage Policy Selection under Cost Uncertainty: An Integrated Pareto, Minimax-Regret, and CRITIC-TOPSIS Approach

Abstract

Warehouse storage-policy selection requires a balance between material-handling distance and the amount of space reserved for inventory. Comparisons based on one pair of cost coefficients can conceal how the preferred policy changes when the relative price of space and handling shifts. This study develops an integrated decision framework combining Pareto dominance, parametric break-even analysis, minimax regret, objective CRITIC weighting, and TOPSIS ranking. Four policies are evaluated for a three-product, four-period warehouse: dedicated three-class storage, two-class storage with fixed area, two-class storage with space consolidation, and one-class shared storage. The procedure first removes dominated policies. When a credible interval for the space-to-handling cost ratio is available, minimax regret is the primary selection rule; CRITIC-TOPSIS is retained as a complementary physical-performance ranking and as a tie-breaker when the regret result is non-unique or the cost interval cannot be specified credibly. At the baseline parameters, the respective total costs are 650, 690, 615, and 630 cost units. The space-consolidating two-class policy reduces total cost by 5.38%, storage area by 9.09%, and one-way travel distance by 2.67% relative to dedicated storage. Pareto screening removes the two dominated alternatives, and the efficient policies have a break-even space-to-handling cost ratio of 8 in the stated unit system. The minimax-regret analysis over the interval 2-12 selects the two-class space-saving policy. CRITIC assigns weights of 0.5014 to storage area and 0.4986 to travel distance; with only two criteria, these weights are equivalent to normalized criterion standard deviations because the common correlation term cancels. TOPSIS ranks one-class shared storage first with a closeness coefficient of 0.6582, followed by the two-class space-saving policy at 0.6223. Across larger benchmark problem sizes, class-based storage achieves average reductions of 4.66-10.63%. The framework provides a reproducible decision rule and a transparent comparison of monetary robustness and physical-performance preferences.