The reduced model aims to create a low-cost, accuratThis work proposes modeling of local problem based on the Physics-Guided Architecture of Neural Networks (PGANN), which incorporates physical variables other than input/output variables into the NN architecture.