Rolling-horizon reliable supply chain planning for IC design house with forecast evolution and customer order behavior

In today’s highly competitive electronics market, demand for integrated circuit (IC) is increasingly uncertain due to fluctuating customer orders and continuous forecast revisions. This uncertainty challenges reliable supply chain planning (SCP), including coordinating production and resource-allocation decisions, efficiently utilizing available manufacturing capacity, and ensuring on-time delivery. In the product setting examined in this study, IC design houses manage a complex multi-stage and multi-site production network in which forecast-driven wafer production operates in a make-to-stock (MTS) mode, whereas order-driven back-end production operates in a make-to-order (MTO) mode. To improve SCP reliability under demand uncertainty, this study develops a data-driven rolling-horizon decision-support framework that integrates demand forecast evolution modeling, customer order behavior (COB), and optimization-based SCP. Two martingale model of forecast evolution (MMFE) structures, including additive and multiplicative forms, are combined with COB to construct behavior-based MMFE models that better capture realistic demand uncertainty and order dynamics. These data-driven demand evolution models are then incorporated into a rolling-horizon SCP (RHSCP) model formulated as a mixed-integer linear programming (MILP). An extended RHSCP model with stochastic lot sizing (RHSCP-SLS) is developed and solved using MILP with a piecewise-linear approximation. To evaluate the reliability and robustness of the proposed planning framework, this research establishes a rolling-horizon simulation environment and conducts experiments under different MMFE types, covariance levels, customer order behavior patterns, SCP models, and capacity utilization levels. The results provide decision-support insights into how demand uncertainty, customer ordering behavior, and capacity conditions affect supply chain performance and planning reliability.