Supply chain network design is one of the key problems in Industrial Engineering and Operations Management. In developing countries such as Bangladesh, the issue is exacerbated by periodic fluctuations in demand and disruptions in the process, particularly when it comes to production and distribution of fast-moving consumer goods. This research proposes and introduces a decision support scheme that combines the operations research optimization techniques with artificial intelligence demand forecasting techniques. By means of a two staged stochastic mixed integer model, the framework aims at optimizing facility locations, capacity pattern, and transportation flows in such a way that the expected network operating cost is minimized. An Artificial Neural Network (ANN) is used to forecast dynamic demand, i.e. nonlinearity and time-varying characteristics of demand. These forecasts are used as input to the stochastic optimization model. Expected expenditure includes facility setup costs, transportation costs, amount kept in inventory, and penalties for disruption. AI-based forecasts enable an improved relationship between cost and risk in an uncertain environment. Scenario based modeling accounts for demand fluctuations and logistic interruptions at regional levels. A numerical case study in the food and beverage industry in Bangladesh shows the hybrid of AI optimization has a better service continuity and network robustness than conventional deterministic planning. The framework provides a step-by-step decision support mechanism for the design of resilient and cost effective smart supply chain networks - even in a manufacturing / logistics uncertain environment.

Keywords

Operations research, artificial neural networks, stochastic optimization, supply chain disruption, FMCG logistics.