A strategic and global manufacturing capacity management optimisation model: A Scenario-based multi-stage stochastic programming approach

Large-scale multinational manufacturing firms often require a significant investment in production capacity and extensive management efforts in strategic planning in an uncertain business environment. In this research we first discuss what decision terms and boundary conditions a holistic capacity management model for the manufacturing industry must contain. To better understand how these decision terms and constraints have been employed by the recent model developers in the area of capacity and resource management modelling for manufacturing, 69 optimisation-based (deterministic and stochastic) models have been carefully selected from 2000 to 2018 for a brief comparative analysis. The results of this comparison shows although applying uncertainty into capacity modelling (in stochastic form) has received a greater deal of attention most recently (since 2010), the existing stochastic models are yet very simplistic, and not all the strategic terms have been employed in the current model developments in the field. This lack of a holistic approach although is evident in deterministic models too, the existing stochastic counterparts proved to include much less decision terms and inclusive constraints, which limits them to a limited applications and may cause sub-optimal solutions. Employing this set of holistic decision terms and boundary conditions, this work develops a scenario-based multi-stage stochastic capacity management model, which is capable of modelling different strategic terms such as capacity level management (slight, medium and large capacity volume adjustment to increase/decrease capacity), location/relocation decisions, merge/decomposition options, and product management (R&D, new product launch, product-to-plant and product-to-market allocation, and product phase-out management). Possibility matrix, production rates, different financial terms and international taxes, inflation rates, machinery depreciation, investment lead-time and product cycle-time are also embedded in the model in order to make it more practical, realistic and sensitive to strategic decisions and scenarios. A step-by-step open-box validation has been followed while designing the model and a holistic black-box validation plan has been designed and employed to widely validate the model. The model then has been verified by deploying a real-scaled case of Toyota Motors UK (TMUK) decision of mothballing one of their production lines in the UK after the global recession in 2010.