Robust optimization of hydrogen network

Abstract

Process integration is an effective way to reduce hydrogen utility consumption in refineries. A number of graphical and mathematical programming approaches have been proposed to synthesis the optimal network. However, as the operation of refineries encounters uncertainty with the rapidly changing market and deteriorating crude oil, existing approaches are inadequate to achieve robust hydrogen network distribution due to the uncertain factors. In this paper, robust optimization is introduced as a framework to optimize hydrogen network of refineries under uncertainty. In this framework, a number of scenarios representing possible future environments are considered. Both model robust and solution robust are explicitly incorporated into the objective function. A possible optimal network distribution which is less sensitive to the change of scenarios and has the minimum total annual cost is achieved by the tradeoff between the total annual cost and the expected error. Case studies indicate that this method is effective in dealing with hydrogen network design and planning under uncertainty in comparison to the deterministic approach and the stochastic programming method.