Enhancing the robustness of water resource simulation models based on network linear programming

In Australia, water resource simulation models using network linear programming (NLP) are widely used. This study identifies two potential problems when using NLP models at daily time steps, which, if not addressed, could result in poor outcomes. These problems are the failure of iteration to converge to good solutions and suboptimal water allocations due to a lack of foresight. It is shown how enhancements to the WATHNET model have addressed these issues. The outcome is a more robust framework than provided by the current generation of Australian NLP-based models.