There's no way to do this directly that I'm aware of. You can, however, use the results from a CFD program to inform your inputs to DOE2. For example, you can run steady-state simulations in your CFD software over a variety of ambient temperatures and windspeeds. You could then use the data from these simulations to calculate the equivalent Sherman-Grimsrud coefficients for use in a DOE2 model (assuming you can get a good fit). If you really do need the CFD coupled to the energy simulation, then you might want to look at other engines besides DOE2. You might also want to check out the airflow network model in EnergyPlus which is like a good compromise between empirical methods and full CFD.