Job Shop Scheduling with Material Handling has attracted increasing attention in both industry and academia, especially with the inception of Industry 4.0 and smart manufacturing. A smart manufacturing system calls for efficient and effective production planning. On a typical modern shop floor, jobs of various types follow certain processing routes through machines or work centers, and automated guided vehicles (AGVs) are utilized to handle the jobs. In this research, the optimization of a shop floor with AGV is carried out, and we also consider the planning scenario under variable processing time of jobs. The goal is to minimize the shop floor production makespan or other specific criteria correlated with makespan, by scheduling the operations of job processing and routing the AGVs. This dissertation includes three research studies that will constitute my doctoral work.

In the first study, we discuss a simplified case in which the scheduling problem is reformulated into a vehicle dispatching (assignment) problem. A few AGV dispatching strategies are proposed based on the deterministic optimization of network assignment problems. The AGV dispatching strategies take future transportation requests into consideration and optimally configure transportation resources such that material handling can be more efficient than those adopting classic AGV assignment rules in which only the current request is considered. The strategies are demonstrated and validated with a case study based on a shop floor in literature and compared to classic AGV assignment rules. The results show that AGV dispatching with adoption of the proposed strategy has better performance on some specific criterions like minimizing job waiting time.

In the second study, an efficient heuristic algorithm for classic Job Shop Scheduling with Material Handling is proposed. Typically, the job shop scheduling problem and material handling problem are studied separately due to the complexity of both problems. However, considering these two types of decisions in the same model offers benefits since the decisions are related to each other. In this research, we aim to study the scheduling of job operations together with the AGV routing/scheduling, and a formulation as well as solution techniques are proposed. The proposed heuristic algorithm starts from an optimal job shop scheduling solution without limiting the size of AGV fleet, and iteratively reduces the number of available vehicles until the fleet size is equal to the original requirements. The computational experiments suggest that compared to existing solution techniques in literature, the proposed algorithm can achieve comparable solution quality on makespan with much higher computational efficiency.

In the third study, we take the variability of processing time into consideration in optimizing job shop scheduling with material handling. Variability caused by random effects and deterioration is discussed, and a series of models are developed to accommodate random and deteriorating processing time respectively. With random processing time, the model is formulated as a Stochastic Programming Job Shop Scheduling with Material Handling model, and with deteriorating processing time the model can be nonlinear under specific deteriorating functions. Based on a widely adopted dataset in existing literature, the stochastic programming model were solved with Pyomo, and models with deterioration were linearized and solved with CPLEX. By considering variable processing time, the JSSMH models can better adapt to real production scenarios.