Optimization of Job Scheduling and Inventory Management for a Multi-Product Make-to-Order Manufacturing System using Hybrid Metaheuristics (PSO-GA)

Project Abstract

This research presents a comprehensive optimization framework for a multi-product make-to-order (MTO) manufacturing system by integrating hybrid metaheuristics, specifically Particle Swarm Optimization (PSO) and Genetic Algorithm (GA), to jointly optimize job scheduling and inventory management. The study addresses the critical combinatorial complexity arising from lot sizing, sequencing, machine assignment, and dynamic demand in a production environment where customization and lead-time commitments are paramount. The proposed model formulates a mixed-integer nonlinear programming (MINLP) problem that captures production capacity constraints, setup costs, order due dates, backorder penalties, inventory carrying costs, and service level requirements, while accommodating product-specific processing times, routing variations, and resource contention across multiple stages. To overcome computational intractability, a two-layer hybrid optimization approach is introduced. The inner layer uses a GA to generate high-quality discrete schedules and routing decisions, leveraging crossover and mutation operators tailored to sequencing and machine assignment. The outer layer employs PSO to fine-tune continuous decision variables, such as job start times, release dates, and inventory levels, enabling smooth coordination between production and inventory policies. A cooperative exchange mechanism between layers ensures convergence toward globally feasible and economically superior solutions. The framework also incorporates constraint-handling techniques and adaptive parameter control to maintain robustness under stochastic demand and random machine failures. Data from a simulated multi-product assembly and machining environment with varying demand profiles, lead times, and return processes is used to evaluate performance. Benchmark scenarios include single- and multi-line configurations, different product mixes, and varying penalty weights for tardiness and stockouts. The performance metrics examined are total production cost, on-time delivery rate, average cycle time, work-in-process (WIP) levels, and service level attainment. Comparative analyses against traditional heuristics (e.g., pure GA, pure PSO, and constructive heuristics) demonstrate that the PSO-GA hybrid significantly reduces total cost and tardiness while improving service levels and reducing WIP, particularly in high-m variability and tight lead-time settings. Sensitivity analyses reveal the model’s resilience to parameter changes in processing times, setup costs, demand volatility, and capacity constraints. The research also investigates the impact of different inventory policies (EOQ-based, base-stock, and dynamic safety stock) on overall performance, identifying conditions under which each policy yields optimal outcomes. The results indicate that the hybrid method consistently outperforms single-method approaches by exploiting the explorative search capacity of PSO and the exploitation strength of GA, thereby achieving balanced solutions that harmonize scheduling precision with lean inventory management. The study contributes to the literature on integrated production-inventory optimization for MTO systems, offers a scalable methodological blueprint for complex manufacturing environments, and provides practical decision-support insights for operations managers aiming to enhance responsiveness, reduce costs, and sustain competitive advantage in customer-centric production networks.

Project Overview

What This Project Is About

---

A plain-language overview of how a factory can plan production and keep enough materials on hand when making several products to customer orders. The project looks at ways to schedule jobs efficiently while controlling inventory, using simple ideas and smart computer-based methods to find good, practical solutions.

---

---

The Problem It Addresses

---

Manufacturers often juggle multiple products with varying sizes, demand deadlines, and component needs. Poor scheduling or inventory decisions can lead to longer lead times, higher costs, or wasted materials. This project targets balancing timely order fulfillment with lean inventory to reduce costs and delays.

---

---

Objectives of the Project

---

1. Understand why scheduling and inventory matter in a make-to-order setting.
2. Explore a hybrid method that combines two simple search ideas to improve decisions.
3. Develop a model that captures production steps, timings, and material usage.
4. Test the model with realistic data and compare to basic planning methods.
5. Provide practical guidelines for managers to apply the method in real plants.

---

---

What You Will Do Step by Step

---

1. Study basic concepts of scheduling and inventory in manufacturing.
2. Build a simple mathematical representation of a multi-product, make-to-order system.
3. Implement the hybrid PSO-GA approach (two straightforward optimization ideas) in code.
4. Run simulations with sample data to see how orders, setup times, and inventory interact.
5. Compare results to traditional planning methods and analyze improvements.

---

---

Expected Outcome

A practical scheduling and inventory plan that reduces order lead times and lowers material costs, with clear steps for applying the method in real factories and guidance on the expected benefits and limitations.