Ehsan Shirazi



Date of Graduation


Document Type

Problem/Project Report

Degree Type



Statler College of Engineering and Mineral Resources


Industrial and Managements Systems Engineering

Committee Chair

Kenneth Currie

Committee Co-Chair

Leily Farrokhvar

Committee Member

Thorsten Wuest


In a traditional storage system, when one of the mechanical or electrical parts or the software fails to operate correctly, it may lead to the failure of the entire system imposing huge costs to the system (Gue et al., 2014). However, in a grid storage network, if a grid fails to operate, the system continues to operate. Therefore, failure of a system component will not jeopardize the whole system functionality. As such, reliability is one of the key advantages of grid storage networks. Another benefit of grid storage systems is flexibility. Various systems such as GridFlow, GridPick, GridStore, are developed that consist of multiple autonomous conveyors to store and move items in different directions. A unit module is capable to work independently. To retrieve an item, empty modules open a temporary aisle to move the item to the retrieval point. This study developed a tote retrieval algorithm for high-density storage systems (HDSS). The algorithm is capable of retrieving totes from all sides of a storage puzzle. This is novel since methodologies of prior studies were confined to retrieving items from a single grid or a single side of a puzzle. The algorithm was decentralized and agent-based where each grid acts as an agent. The study undertook developing algorithms that could minimize the number of movements in the network for tote retrievals and to prevent deadlocks. Deadlock prevention algorithms are capable to resolve a diverse range of situations that cause network deadlocks. An object-oriented software program was developed that implemented the algorithms. The software tool simulated puzzles of various sizes with different number of escorts available for retrieving requested totes. Thousands of iterations of the puzzle configurations were resolved for the analysis. It was found that incremental increase in the number of escorts in the puzzle reduces the number of movements for retrieving totes. However, depending on the puzzle size, there were increments that additional escorts had minimal impact. It was also found that average retrieval movements for 3 totes increase logarithmically with the number of cells in a puzzle. To validate the methodology and its software implementation, results from the simulation were compared with the results from other studies with mathematical solutions. The program replicated the optimum number of movements for a puzzle that contained one escort and retrieved one tote.