1: \begin{abstract}

2: % The current energy transition promotes the convergence of operation between the power and natural gas systems. In that direction, it becomes paramount to improve the modeling of non-convex natural gas flow dynamics within the coordinated power and gas dispatch. In this work, we propose a neural-network-constrained optimization method which includes a regression model of the Weymouth equation, based on supervised machine learning. The Weymouth equation links gas flow to inlet and outlet pressures for each pipeline via a quadratic equality, which is captured by a neural network. The latter is encoded via a tractable mixed-integer linear program into the set of constraints. In addition, our proposed framework is capable of considering bidirectionality and linepack capability without having recourse to complex and potentially inaccurate convexification approaches. We further enhance our model by introducing \textit{i)} a reformulation of the leaky rectifier linear unit activation function, and \textit{ii)} a physics-informed sampling of the training dataset which both improve the tightness computational efficiency. An extensive numerical study based on the real-life Belgian power and gas systems shows that the proposed methodology outperforms the existing convexification approaches in terms of accuracy and tractability. 

3: % \end{abstract}

4: