INTEGRATION PLANNING MODEL FOR PICKING AND PACKING CUSTOMER ORDERS TO MINIMIZE MAKESPAN IN FULFILLMENT CENTER FOURTH PARTY LOGISTIC

• Main Author: Yusarita, Alvi
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/76313
• Institution: Institut Teknologi Bandung

Fulfilling customer order is increasingly complex with the characteristics of e- commerce customer orders. Orders with small lot sizes and various item that arrive dynamically within a certain period. In fulfilling customer orders, each order will be grouped into a batch. Fulfillment centers that have a manual picking system (pickers-to-parts) with a human operator picking goods for each batch on a tour of the collection location. The picking process for each batch ends when the picking operator returns to the depot. Each customer order has a different delivery time because most fulfillment centers outsource delivery to third-party logistics service providers (3PL). Customer orders must be packaged before their delivery time arrives. Warehouse management a fulfillment center is critical in the customer order fulfillment process. In the current research, integration in warehouse management has the potential to produce better warehouse performance. This research integrates the process of picking and packing to fulfill an order. The difference in this research compared to previous studies is in the planning of picking and packaging which determines batches, the assignment of batches to pickers and packers, and the selection of picker routes simultaneously, taking into account the arrival time of 3PL and the minimum distance traveled by pickers. This research develops an integration model of picking and packing using the Mixed Integer Linear Programming (MILP) model with the Branch-and-Bound method. The performance criterion to be achieved in this study is minimizing the makespan. The model is tested with empirical data originating from the fulfillment center of a fourth-party logistics (4PL) company. The Mathematical model shows the global optimal value for small data set with the number of 10– 20 customer orders. To provide solutions with large data sets, the Variable Neighborhood Descent (VND) algorithm was developed according to system problems. Generating the initial VND solution using the Earliest Start Due Date Rules (ESDR) algorithm and determining the route of the pickers using the heuristic s- shape method. From the test results with the same empirical data as the mathematical model, the VND algorithm provides a solution with good quality. The gap between MILP and VND solutions on small data set is 2.59%. Based on the results of the search for solutions, the development of the algorithm in this v study was able to save around 35-50% makespan time on the review system from the current policy.