Risk Prediction in Design of Distributed Supply Chains Using a Bi-Objective Optimization and Machine-Learning Approach