There are many options. The choice will depend on the particle loading and on the computational resources you have. The easiest and less expensive (least accurate as well) is to deal with it a single flow with modified (effective) thermo-physical properties. There are many empirical models for that. Another approach is to deal with it as (somehow) as a single phase but in the mixture content. in this model you have to model the slip velocity between the carrier fluid and the particles and use a volume fraction equation . Another approach is to treat both of the carrier fluid and the particles in the framework of a full-Eulerian approach. Here, you will need (in addition to the volume fraction) distinctive momentum and energy equations for each phase. The discrete particle modeling DPM is also possible provided that the particle loading is sufficiently small ( to be conservative, I would say a volume fraction less than 3%). Direct simulation for the particles is prohibitively expensive and is not feasible. Commercial packages like Fluent and CFX are equipped with the previously mentioned two-phase flows. You have to keep in mind that these models are full of empiricism and modeling uncertainties.

We are wondering what will be the results considering two possibilities:

1. To treat the radiator as one black box and determine the heat transfer based on the inlets and outlets conditions; OR

2. To study the effect of single phase or multphase conditions in the internal radiator flows.

What is the best way to simulate card radiator using Fluent?

We are reviewing past CFD models for car radiator. Many of them look at only one tube with fins. Is this the best way to simulate the car radiator? Or should at least model half the radiator?

For symmetrical assumption, should we consider the full length of the fin or only half-length? Since a full length fin normally touches two two tubes?