A novel feature-rich discrete phase solver is presented and its applications to high-speed liquid sprays are discussed. The solver has been implemented in the 3D CFD VECTIS suite, which is a tool for solving advanced industrial fluid flow problems. Eulerian/Lagrangian representation has been adopted for the analysis of the two-phase flows with the discrete phase statistically grouped in computational parcels. The solver features a set of detailed physical sub-models which are particularly suitable for high-speed sprays, namely the droplet evaporation, aerodynamic drag and heat transfer, primary and secondary breakup and droplet impingement models.

The integration of the discrete phase is based on an advanced face-to-face tracking algorithm with variable time step control used to maximize the efficiency of the time integration of the parcel balance equations. The time step choice is done individually for each parcel based on the estimation of the droplet response times in the mass, momentum and energy balances. This ensures that the time resolution of the solution is sufficient for an accurate prediction of the inter-phase transfer processes.

The phase coupling is an important issue for modelling of highly separated two-phase flows with large slip velocities between the phases. A kernel-function approach has been used in the code for the evaluation of continuous flow properties at droplet positions as well as for the distribution of the sources from droplets to their gaseous surroundings. This approach minimizes undesirable effects of the computational grid on the spray formation prediction.

The code adopts an up-to-date approach to the spray breakup modelling. A hybrid breakup model improves the accuracy of the breakup predictions for a wide range of conditions found in high-speed sprays by combining two breakup models. The variable time stepping approach has been extended to take into account the droplet characteristic breakup time to predict correctly the behaviour of distorting and possibly disintegrating droplets.

The implemented solver has been applied to a number of test cases with different injector geometries and with injection pressures up to the level typical for modern diesel engines. The predicted spray structures were validated with experimental data demonstrating good accuracy of the spray solver. The computational costs of the discrete phase solver have been found to be reasonable by comparison with the overall cost of the 3D flow solution.