Design an Optimal Blending Recipe via Scenario-based Chance-constrained Optimization

A scenario-based optimization approach is proposed to design gasoline blending recipes under uncertainty. The proposed scheme considers the nonlinear mixing law of octane, uses Monte Carlo sampling-based method to simulate parametric uncertainties, and employs a sequential algorithm with bound tightening to obtain a γ-global optimal solution. This framework offers three advantages: First, incorporating the nonlinear functions into the chance-constrained optimization will improve predictions on the key properties of the blend. Second, by accounting for uncertainty in the blending process, a solution with best expected quality under several possible conditions can be achieved. Third, the sequential algorithm with bound tightening determines the γ-global optimal solution faster than directly using state-of-the-art optimization software. A case study with nine feedstocks and two products is presented to demonstrate the effectiveness of the proposed method.