A computationally efficient spray model is presented for the simulation of transient vaporizing engine sprays. It is applied to simulate high-pressure fuel injections in a constant volume chamber and in mixture preparation experiments in a light-duty internal combustion engine. The model is based on the Lagrangian-Particle/Eulerian-Fluid approach, and an improved blob injection model is used that removes numerical dependency on the injected number of computational parcels. Atomization is modeled with the hybrid Kelvin-Helmholtz/Rayleigh-Taylor scheme, in combination with a drop drag model that includes Mach number and Knudsen number effects. A computationally efficient drop collision scheme is presented, tailored for large numbers of parcels, using a deterministic collision impact definition and kd-tree data search structure to perform radius-of-influence based, grid-independent collision probability estimations. A near-nozzle sub-grid scale flow-field representation is introduced to reduce numerical grid dependency, which uses a turbulent transient gas-jet model with a Stokes-Strouhal analogy assumption. An implicit coupling method was developed for the Arbitrary Lagrangian-Eulerian (ALE) turbulent flow solver. A multi-objective genetic algorithm was used to study the interactions of the various model constants, and to provide an optimal calibration. The optimal set showed similar values of the primary breakup constants as values used in the literature. However, different values were seen for the gas-jet model constants for accurate simulations of the initial spray transient. The results show that there is a direct correlation between the predicted initial liquid-phase transient and the global gas-phase jet penetration. Model validation was also performed in engine simulations with the same set of constants. The model captured mixture preparation well in all cases, proving its suitability for simulations of transient spray injection in engines. (C) 2015 Elsevier Ltd. All rights reserved.