An ensemble method for sensor optimisation applied to falling liquid films

Multiphase flow problems are often extremely complex due to their strong nonlinearity. To study multiphase flow, it is important to simulate or measure key parameters accurately, such as pressure drops and flow rates. Therefore, it is essential to place the sensors at the locations with high impact, and to avoid locations with low impact, where impact is determined by a function such as one of the key variables like pressure drop or flow rate. In this paper, an ensemble method is used to optimise sensor locations for falling film problems based on an importance map. The importance map can identify the important regions according to a target function. The sensor locations are selected based on the importance map, the variation of the variables, and the costs of performing the measurements. We demonstrate the approach by applying data assimilation and show that the optimised sensor locations can significantly improve the data assimilation results. Through sensitivity analysis, sensor optimisation, and data assimilation, this study, for the first time, provides a systematic linkage between the experiments and the models for falling film problems. It also presents a new goal or target based method for sensor placement. This method can be extended to other complex multiphase flow problems.