Paint Shop Scheduling Optimization
As the project owner, I designed and implemented a simulation-based optimization model to solve a real-world car paint shop scheduling problem, with the primary objective of minimizing color changeover costs and idle times within a highly constrained industrial environment.
This project was conducted in the context of a course on simulation and metaheuristics, and it allowed me to apply a wide range of techniques covered in the curriculum.
The System
The paint shop environment was modeled using a discrete-event simulation framework, capturing operational constraints such as buffer capacity between workstations, priority rules for certain vehicle types, cleaning time between color transitions, and the sequence-dependent nature of changeover costs (e.g., switching from white to black is more costly than from grey to silver).
Methodology
To generate initial schedules, I implemented multiple heuristic approaches including First-Come-First-Served (FCFS), color batching, and greedy lookahead algorithms. To further improve these schedules, I applied simulated annealing, a probabilistic metaheuristic that explores the solution space by accepting not only improvements but occasionally worse solutions to escape local optima. The neighborhood generation mechanisms included random swaps, reinsertions, and batch merges. Temperature decay functions and acceptance criteria were carefully tuned based on empirical testing.
Technologies & Evaluation
Throughout the simulation runs, I evaluated performance using metrics such as total changeover cost, idle machine hours, paint station utilization rates, and the number of color transitions. The final model was implemented in Python using NumPy for logic and data manipulation, and Matplotlib to visualize scheduling output in the form of Gantt charts and trend plots. Additionally, Excel was used for validating output data and comparing the performance of different approaches.
Outcome & Impact
The simulated annealing algorithm consistently outperformed static heuristics, achieving up to 40% fewer color changeovers and approximately 25% reduction in idle time under realistic workloads. This project not only deepened my understanding of simulation-based optimization and metaheuristics but also demonstrated how theoretical knowledge from operations research courses can be translated into impactful solutions in a manufacturing context.
“In manufacturing, every minute lost to inefficient scheduling is money wasted. This project proves how smart optimization can paint a different picture.”