Gas Flow Models and Computationally Efficient Methods for Energy Network Optimization

The equations governing gas flow dynamics are computationally challenging for energy network optimization. This paper proposes an efficient solution procedure to enable tractability for an hourly resolved yearly decision horizon. The solution procedure deploys linear and second-order cone gas flow models alternatively based on the length–diameter ratio of pipes, achieving maximum efficiency within accuracy limits. Moreover, it addresses the computational complexity of bidirectional pipe flows by fixing the associated integer variables according to a preceding optimization with a static flow approximation. The procedure also precisely aggregates parallel and serial pipes for increased efficiency. Mathematical derivations and single-pipe analyses substantiate the model selection criterion. Network optimizations validate the accuracy, success rate, and scalability of the procedure, achieving up to 3.1% cost savings compared to static models, enhancing the success rate by a minimum of 96%, and boosting computational efficiency up to 3 orders of magnitude over full dynamic models.