A pooling-based stochastic order-picking system considering priority queues

E-commerce is a phenomenon that shapes the shopping needs of today's world and enables every product to reach our homes with a single click, shaping the life of the consumption society of the 21st century. E-commerce is a market whose users are increasing every year. Especially in the post-covid period, people far from e-commerce applications have also started to use this system. In this thesis, the bottleneck created by online orders and the increasing workload in online order retailers on special days are focused on. Markov models are a widely used method in modeling the preparation and waiting process of such orders. Future orders can be determined with Markov chains and the waiting times in the queue can be found as exact solutions without the need for simulation. Classifying orders and prioritizing certain order classes can avoid bottlenecks. Priority orders are generally expected to include more valuable product lines, as retailers want to deliver more valuable items sooner to satisfy relevant customers sooner. In this study, an order fulfillment model using the priority model is envisaged and two models are considered with priority orders. In the first model, we assume different service rates for different priority classes. Thus, we aim to specify the conditions where the priority order model decreases expected waiting time and increase the customer satisfaction. In the second model, same service rates for different priority classes are assumed but order cancellations are also considered by introducing a threshold waiting time level for every customer's waiting sensitivity. As waiting time sensitivities, deterministic and exponential values are considered. Moreover, the proportion of serviced customers as well as the expected waiting time is regarded as another performance measure of the system because order cancellations are possible. Lastly, numerical results are obtained to show our statements.