Simulation-based Decision Support Tools for Demand Disaggregation in a Supply Chain Characterised by High Routing Flexibility

Routing flexibility is a key process feature in supply chains characterised by complex hierarchy at different manufacturing levels. When routing alternatives exist, controlling the supply chain to ensure uniformity and quality of the production outcome becomes a significant challenge, also from a demand planning perspective. In this paper, the demand planning problem at a supply chain operating in the semiconductor industry is investigated. Special attention is paid to mid-term demand planning when production orders are not yet finalised and aggregated demand forecast is considered. Within this planning frame, the demand planners face the difficult task of disaggregating aggregated demand into finer granularity products in order to generate provisional production plans that will be used to foresee potential capacity adjustment requirements. The demand disaggregation process entails routing decisions that also incorporate restrictions occasionally imparted by final customers on eligible routes for a specific product type. Historical demand patterns and routing constraints currently constitute the main decision drivers in the demand disaggregation process; likewise, the only objective accounted for is the timely satisfaction of customers’ orders. However, disregarding capacity constraints and ignoring incoming future demand characterised by more stringent routing requirements leads to uncapacitated production plans which might cause significant lateness in the orders’ production. In this study, simulation-based solution approaches able to facilitate the demand planners in the complex task of disaggregating demand forecast are developed. It is shown how either analytical simulation algorithms or discrete event simulation can be used to quantify the lateness deriving from the allocation of different demand profiles and predict the impact of future demand on the optimal disaggregation logic. For analysis purposes, the supply chain will be modelled as a serial-parallel multistage manufacturing system for which the facilities operating in parallel at each stage are similar from a process viewpoint but present different capacity.