RESOURCE ALLOCATION IN AN ASSEMBLY SYSTEM: FROM MANUAL PLANNING TO AUTOMATED DECISION SUPPORT

Sammanfattning

Title: Resource Allocation in an Assembly System: From Manual Planning to Automated Decision Support. Authors: Malik Jusopov & Kasper Palm. Supervisors: Dr. Sandeep Jagtap, Division of Engineering Logistics, Faculty of Engineering, Lund University Johannes de la Cour, Internal Logistics Manager, Haldex Contribution: This thesis has been a complete collaboration between the two authors. Each author has been involved in every part of the process and contributed equally. Problem Statement: Haldex currently relies on manual planning processes when allocating resources to their assembly lines, with no unified approach between planners despite having access to relevant data and tools. This results in time consuming planning that is subject to inefficiencies. Purpose: The purpose of this thesis is to evaluate and develop a decision support system for resource allocation in Haldex's assembly system that reduces manual planning effort and improves planning performance through automated decision making. Research Questions: RQ1: Which logistical parameters and design choices should be considered when developing a decision support system for short-term production planning? RQ2: What are the challenges that must be addressed when translating a planning problem into a decision support system? RQ3: To what extent can an automated decision support model improve the allocation in an assembly system, compared to the current manual planning approach? Methodology: This thesis primarily follows an abductive research approach, with a research strategy aligned with the DSR framework. Qualitative data was collected through semi-structured interviews with production planners and continuous dialogue with senior-level management. Quantitative data was also collected, mainly from company systems and employees at Haldex. Conclusion: With parameters such as order quantities, batch sizes, inventory levels, backlog, and preferred line adherence paired with design choices such as an MILP with an Excel interface, the model consistently provided a production plan that reduces both inventory levels and backlog while adhering to line preference when possible. Thus, the model showed it can improve the current planning process and reduce manual planning by providing an optimal solution in seconds. This was achieved despite challenges related to standardizing tacit knowledge, defining necessary assumptions, and adhering to the current ways of working at Haldex.