A procedure for approximating a complex hydrodynamic model by the adaptive time delay method

Model reduction in water systems provides an efficient technique for simplifying complex hydrodynamic models without eliminating the key dynamic attributes, and is inevitable for model-based control. In particular, the first order model with adaptive time delay (ATD) has proven to be suitable for approximating hydrodynamic behavior of river reaches. Experimental results of identifying parameters of ATD models from a complex hydraulic model (e.g., HECRAS) show that the state-of-art of existing methods is not enough, because time constant and time delay mutually interact, are variably and significantly depend on flood wave characteristics. Hence, the present work proposes a procedure to identify these parameters. At first, the hydrograph of extreme flood event is generated and analyzed to derive a characteristic hydrograph. Secondly, the typical outflow is simulated by HECRAS model and the inflow and outflow generated from HECRAS are used to estimate parameters of ATD model by nonlinear programming technique. A set of time constant and time delay corresponding to characteristic flood wave will be validated by a specific flood event. This procedure is an advancement of previous methods which model flow by a simplified channel geometry. In addition, it compensates deficiencies of existing methods attempting to fix a time delay as an integer value or a constant value which introduces truncation error during routing process leading to model instability. Therefore, the approach is capable to different types of flow, and returns very promising results in real time flood forecasting and control.