In the past couple of decades, computational fluid dynamics of multiphase flows has evolved tremendously. With the fast growth in computational power, researchers who once had been bounded to study the problems in small domains or for very limited number of bubbles or droplets, now are able to do the simulations of more realistic problems or to look at additional physics. However, computational simulation of multiphase flows can become exponentially expensive when thin films appears in the physical domain, either as a results of interactions between interfaces or the existence of different physics which usually leads to diverse range of length and time scales in the systems. Resolving thin films using conventional methods like adaptive mesh refinement can adversely increase the computational cost and also make the parallel scaling on massive clusters hard to achieve. Fortunately, it is possible for many cases to develop analytical subscale models which bridge the gap between length and time scales in the problem. We developed a subscale model for mass transfer in bubbly flows and demonstrated its accuracy and efficiency. This model can be essential in simulations of bubble columns, one of the most important processing units in chemical and petrochemical industries. Additionally, we developed two other subscale models, the first is for the correction of viscous forces for simulation of colliding non-coalescing droplets and the second is a thermal model for simulations of a cavitation driven heat transfer problem, critical to naval engineering applications.