An automatic method for optimizing Venturi shape in cavitation flows

Abstract

In order to lower the energy consumption of the fibrillation stage for the pulp and paper industry, a new technology need to be innovated and developed. The current research work deals with a new innovative concept based on creating cavitation in the pulp flow. A venturi nozzle is designed and optimized, where hydrodynamic cavitation is achieved by the so called Venturi effect. This paper focuses on the development of an automatic method for venturi shape optimization. The process of cavitation is hard to control and can cause high mechanical wear, therefore an optimization study of the venturi shape is performed with two main objectives. Firstly, to achieve cavitation that is sustained for as long as possible downstream and secondly to avoid cavitation at the walls. The developed method is a type of two-level optimization based on neural networks and evolutionary optimization. A number of simulations are executed and the optimization is then performed on a solver approximation instead of the real solver, which considerably reduces computation time. The obtained results show the optimal venturi configuration and the relative importance of each shape parameter. The optimal configuration is a clear improvement of the baseline configuration and an improvement also compared to all of the tested samples, thereby validating the optimization method.