This study investigates the effect of water injection in the high pressure chamber of a cold spray nozzle. A de Laval nozzle geometry with constant back pressure and temperature is modeled numerically using Reynolds Stress Model coupled equations. Water spray with a droplet size of 10 – 100 μm is modeled using both uniform and Rosin-Rammler size distributions. The two-phase flow of gas-liquid is modeled using an unsteady discrete phase mass source with two-way coupling with the main gas flow. Upon injection, the droplets in the water spray evaporate while travelling through the nozzle due to momentum and energy exchange with the gas flow. The evaporation behavior in presence of water content is modeled and a correlation between the initial diameter and the diameter just before the throat is obtained. As a result, the proper droplet size distribution with a fully evaporative spray can be used as a carrier of nano-particles in cold spray nozzles. Having the results, guide us to substitute the un-evaporated part of the droplet with an equal diameter agglomerate of nano particles and find a minimum fraction of nano particles suspended in the liquid which guarantees fully evaporative liquid spray injection.

The two-phase flow properties of copper particle laden nitrogen are measured and compared to computational fluid dynamic calculations, determining the achievable degree of consistency between model and reality. Two common, commercial nozzles are studied. A two-way coupled Lagrangian scheme along with the RSM turbulence model is used to track the particles and to model the interactions between the gas and the particulate phase. Significant agreement is found for the geometrical gas flow structure, the resulting particle velocities, and the dependence of the two-phase flow on the particulate phase mass loading. The particle velocities decrease with increasing mass loading, even for modest powder feed rates of less than 3 g/s. The velocity drop occurs even when the gas flow rate is kept constant. Adiabatic gas flow models neglecting the energy consumption by the particles are thus inaccurate, except for very dilute suspensions with low technical relevance. For the cases modeled, the experiments evidence the high predictive power of the chosen approach.