Title: Assembly Line Balancing and Scheduling for A Hyperconnected Assembly Factory Design

Date: July 3rd, 2025 (Thursday)

Time: 9:00 am – 11:00 am EST

Meeting link: link

Join conversation

teams.microsoft.com

Wencang Bao

Ph.D. Candidate in Industrial Engineering

School of Industrial and Systems Engineering

Georgia Institute of Technology

Committee

Dr. Benoit Montreuil (Advisor), H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology.

Dr. Leon F. McGinnis, H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology.

Dr. Mathieu Dahan, H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology.

Dr. Uday Venkatadri, Department of Industrial Engineering, Dalhousie University.

Dr. Nico A. Schmid, School of Management, Department of Operations Management, IÉSEG School of Management.

Abstract

The hyperconnected assembly facility network is an innovative response to the challenges of scalability, product variety, reconfigurability, and resource shareability encountered by modern assembly factories. It builds upon the principles of hyperconnected production, flexible manufacturing technologies, and fractal layout design. However, prescriptive models for the design, implementation, and planning of such facilities remain underexplored. This dissertation focuses on two critical aspects of hyperconnected assembly factory design and planning: assembly line balancing and mixed-model sequencing. Given a target throughput, assembly line balancing involves assigning tasks to workstations, which in turn determines the required number of stations and associated equipment. Mixed-model sequencing, on the other hand, establishes the order in which products enter the assembly line, directly influencing workforce requirements. Together, these two problems form the foundation of effective assembly system planning, supporting decision-makers to bridge the gap between conceptual design and practical implementation.

In chapter 2, we investigate a new variant of the assembly line balancing problem which integrates parallel stations (PS) and variable task-worker options (VTWO). In our context, VTWO includes the selection of worker quantities for some tasks. We integrate PS and VTWO for mixed-model and multi-manned assembly lines where multiple workers are allowed to work concurrently in one station. We formulate a mixed-integer optimization model to jointly solve PS and VTWO (ALBP-PSVTWO), and we also develop two heuristic approaches which solve the problem in two phases, ALBP with PS-then-VTWO, and ALBP with VTWO-then-PS . Additionally, a logic-based Benders decomposition method is proposed to reduce the solution time for large scale problems. We carry out an experiment motivated by a real-world case study and compare the performance of the proposed models with PS-only and VTWO-only models. Depending on the inputs, the integrated model could yield 7%-20% lower costs than the traditional models.

In chapter 3, we focus on an extension of the ALBP-PSVTWO where workers possess heterogeneous skill sets. Considering the long takt time and flexible production technologies available in the hyperconnected assembly factory design, we propose a new ALBP with multi-skilled and traveling workers (ALBP-MSTW). Traveling workers are those workers who are allowed to work at several distinct stations accounting for traveling time within one takt time. Our approach introduces a novel mixed-integer programming (MIP) optimization model minimizing both worker and station costs. To handle large-scale problems efficiently, we employ a simulated annealing-based hyper-heuristic algorithm to accelerate the solving process. Finally, insights gleaned from experiments inspired by a real-world case study highlight the practical application and benefits of using traveling workers in such assembly scenarios. In the real-business case studies, the ALBP-MSTW brings 7%-15% cost savings compared with other benchmark models with the given parameters.

In chapter 4, we address the optimization of product sequencing for a mixed-model assembly line. Given the long takt time and the flexibility of worker deployment within the hyperconnected assembly facility network, we incorporate an adjustable workforce into the sequencing process, allowing for the periodic reallocation of labor. To solve the Mixed-Model Sequencing Problem with Multi-Skilled Traveling Workers and Adjustable Workforce (MMSP-MSTWAW), we develop an interactive two-phase model aiming at minimizing labor costs. The first phase applies a hyper-heuristic algorithm to explore a promising product sequence. The second phase uses a rolling-horizon optimization approach to evaluate the labor cost of the given sequence and determine the corresponding workforce requirements. These results provide feedback to the hyper-heuristic in the first phase in subsequent iterations. In the experiments extended from the industrial case study, our proposed model achieved labor cost reductions of up to 8%–13% under the given system parameters. Finally, we integrate the ALBP-MSTW and MMSP-MSTWAW models to jointly analyze the effects of task definitions and takt time on assembly line performance, offering decision-makers valuable insights for optimizing configuration and improving operational efficiency.