Improvement of three-phase flow numerical simulations by means of novel hybrid methods

Numerical simulations of three-phase flows are facing the challenge that their mathematical models include a lot of assumptions and the equation systems often deliver controversial solutions. The object of this study is the improvement of numerical simulations of a three-phase (solid, gas, liquid) flow according to the four-way coupling Eulerian-Eulerian frame. Following the strategy of incorporating a priori knowledge in a system, initial velocity information achieved by several experimental and numerical techniques is implemented in the numerical simulations. Particle image velocimetry (PIV) data are employed in a numeroexperimental hybrid and artificial neural network (ANN) data in numeroneuronal and neuroexperimental hybrids, where the ANNs are trained with numerical or PIV data, respectively. The employment of the three presented hybrid methods affords better convergence of the numerical simulations, delivers more accurate numerical results and enables saving of computational time, thus, more precise information about the behaviour of the fluid mechanical system is faster achieved.