Optimization of a tyre manufacturing plant’s lot-sizing and scheduling system

Thesis Abstract

This study addresses the persistent inefficiencies in tyre production by examining lot-sizing and scheduling practices within a contemporary tyre manufacturing plant facing volatile demand, high changeover costs, and tight delivery windows. The core problem is the misalignment between production lot sizes and dynamic demand, which amplifies setup times, work-in-progress inventory, and on-time delivery deficiencies. The aim is to develop an integrated optimization framework that simultaneously determines optimal lot-sizing and finite-capacity production schedules to minimize total production and changeover costs while meeting service level commitments. Specific objectives are (i) to quantify the impact of current lot-sizing policies on operational performance; (ii) to formulate a mixed-integer linear programming (MILP) model that integrates demand forecasts, capacity constraints, setup times, and partial backlog allowances; (iii) to incorporate a robust optimization layer to address demand and parameter uncertainty; (iv) to propose a hybrid solution approach combining exact MILP solving with metaheuristics for large-scale instances; (v) to validate the model on real-world data from the plant and compare performance against the incumbent planning system. The methodology adopts a multi-phase research design, combining quantitative optimization modeling with empirical validation. The population comprises the plant’s production lines, including tire tread and sidewall assembly, with historical monthly demand data for five product families over a two-year horizon. A stratified sampling approach yields a representative dataset of 360 daily records, selected to capture seasonality and demand shifts. Data collection instruments include production records, changeover time measurements, machine reliability logs, and demand forecasts from the plant’s enterprise resource planning (ERP) system. Instrument validity is ensured through triangulation of production logs with supervisor interviews and cross-verification against ERP extracts; reliability is established via test-retest procedures and inter-rater consistency checks for qualitative inputs. The analytical framework integrates (a) descriptive statistics to characterize current performance, (b) regression analysis to identify drivers of setup time and throughput, (c) the MILP model for lot sizing and scheduling, (d) stochastic programming or robust optimization to handle demand and yield uncertainty, and (e) a hybrid solution method employing a branch-and-price–based MILP with a genetic algorithm refinement for large-scale instances. Model specification encodes capacity limits, run lengths, setup sequences, batch sizes, inventory holding costs, and service level constraints. Ethical considerations address data confidentiality with anonymized identifiers and compliance with plant data governance. Expected findings indicate that the proposed integrated framework reduces total annualized costs by at least 12–18% relative to the current planning approach, with improvements in on-time delivery from 92% to above 98%, and a 10–15% reduction in changeover-related downtime. The robust version is anticipated to show resilience to demand fluctuations with a marginal downside of 2–3% in worst-case scenarios. Sensitivity analyses will reveal critical parameters driving performance, notably changeover times, forecast accuracy, and batch-production trade-offs. The study also anticipates quantifying the trade-off between inventory carrying costs and outsourcing risk under capacity constraints, providing managerial insights on policy levers such as policy-based ordering, demand buffering, and strategic safety stock placement. Contribution to knowledge includes (i) a novel integrative MILP-robust optimization framework for tyre manufacturing that unifies lot sizing and scheduling under finite capacity with changeover considerations; (ii) empirical validation in a real production environment, extending the literature on production planning under high-mix, low-volume automotive components; (iii) methodological advancement through a hybrid exact–heuristic solution approach validated on medium- to large-scale industrial instances; and (iv) actionable guidelines for practitioners on implementing short-cycle planning improvements, changeover reduction, and data governance for forecasting and scheduling. The main conclusion is that coordinated optimization of lot sizing and scheduling, augmented with robust optimization to absorb demand uncertainty, yields significant operational and service-level gains in tyre manufacturing. Practical recommendations include investing in data quality for demand forecasting, reducing variability in changeover times through setup reduction programs, adopting the proposed MILP framework as a decision-support tool integrated with ERP, and conducting periodic policy reviews aligned with market dynamics and plant capabilities.