Research-Thesis: Makespan Minimization In Warehouse Order Picking Via Integrated Batching And Routing: A Metaheuristic Approach Under 2D Placement And Pallet Feasibility Constraints

Abstract

Efficient warehouse management increasingly requires the simultaneous optimization of multiple operational layers, yet literature often treats order batching, packing, and routing as independent problems. This thesis proposes an integrated optimization framework designed to minimize the maximum makespan in a multi-picker warehouse environment while adhering to strict two-dimensional packing constraints. The methodology employs a First-Fit Decreasing Algorithm combined with a 2D Bottom-Back-Left-Fill heuristic to ensure that batches are not only volumetrically efficient but also physically feasible for palletization. To refine the initial solutions, two variants of Simulated Annealing are developed: one focusing on route-level distance reduction and the other on workload balancing (makespan). The framework was evaluated across diverse experimental scenarios involving varying order sizes (100 to 1000) and picker counts (2 to 8). Experimental results demonstrate that the Routing-Based SA consistently outperforms the makespan-based approach, yielding higher improvement percentages by exploiting structural inefficiencies within picking tours. The findings reveal that the initial greedy baseline achieves a high average pallet utilization of approximately 87%, confirming the effectiveness of the BBLF logic. Furthermore, the study shows that in high-density instances, the marginal benefits of routing optimization decrease as picking routes become more spatially constrained. By bridging the gap between geometric packing feasibility and heuristic routing optimization, this research provides a robust decision-support tool for modern distribution centers, highlighting that true operational efficiency stems from the holistic coordination of batching, packing, and routing decisions.