Design and Analysis of Unit-load Warehouse Operations using Autonomous Vehicles.

摘要

Warehouse automation plays an important role in improving the efficiency of distribution operations in supply chains. This dissertation addresses design issues relevant to the conceptualization phase of warehouse design. Queuing models are developed to investigate trade-offs related to the multi-tier warehouses that use autonomous vehicle-based technology for unit-load operations. To store and retrieve unit-load pallets in these systems, vehicles use rails to move along x- and y-directions within a tier and use lifts to move in the z-direction between tiers.;Since the operational dynamics are complex, this research first models the operations in a single tier. A semi-open queuing network model is used to evaluate the impact of warehouse topology, vehicle assignment rules, dwell point policies, and cross-aisle location on performance measures such as throughput, utilization and cycle times. Subsequently, the model is enhanced to incorporate vehicle blocking effects. Although, incorporating blocking effects significantly increases the complexity of the queuing network models, with use of smart protocols, and a decomposition approach, solutions to the queuing models are obtained.;Next the research investigates the trade-offs in multi-tier systems that use conveyors and lifts for vertical transfer of loads. Two system variations are analyzed: the tier-captive vehicle configuration where a set of vehicles are dedicated to each tier and the pooled vehicle configuration where vehicles can process transactions in any tier. A tandem queue model of the conveyor system is combined with the queuing models of individual tiers to yield an integrated queuing network model of the entire system. Using results from analysis of a single tier that permit replacing the tier-subnetwork by an equivalent load-dependent station, the complexity of analysis is significantly reduced. A combination of parametric decomposition, embedded Markov chain analysis, and approximate mean value analysis is then used to evaluate system performance and evaluate the impact of various vertical transfer mechanisms on throughput capacity and transaction cycle times.;The suite of analytical models developed in this research will allow designers to rapidly evaluate alternative design configurations in seconds thereby helping the designer choose warehouse configurations and operating parameters that result in best performance and lowest cost.