Page V ■ 18 
Ventilation Design Handbook on Animal Research Facilities Using Static Mieroisolators 
validation work. A more realistic range of softer type flows (i.e., less extreme strain) has not been 
studied with the RNG model. The infancy of this approach prevents it from being incorporated at 
this stage. When the model becomes as tried and trusted as the present standard k-e model, it will 
be given greater attention. 
Instead of employing the eddy viscosity assumption, which assumes an equal eddy viscosity in all 
three spatial directions, a Reynolds stress model has an equation for each of the six Reynolds 
stresses themselves. This allows the modeling of the transport of each of these individual 
stresses. This is the most complex of all models and suffers accordingly. Instead of two extra 
equations we now have an extra seven. An equation for e is still required because it pops up in 
the stress transport equations. Convergence stability now becomes a serious problem. Even if 
convergence is achieved, it normally takes considerably longer than with a two-equation model. 
Prescription of boundary conditions is also tricky. Instead of setting just k and e, we now have to 
set values at supply boundaries of all stresses, not the easiest of parameters to obtain from 
experimental measurement. The question has to be asked as to whether the added theoretical 
capabilities of an RSM are worth the increased solution time and decrease in stability. 
5.2.2. 4 Modeling of Reynolds fluxes: 
The velocity-enthalpy correlations known as the Reynolds fluxes use much the same 
methodology as the Reynolds stresses. An eddy diffusivity is therefore defined as: 
5. 2. 2. 3 Reynolds stress models ( second order closure models) 
(5.19) 
where this eddy diffusivity is related to the eddy viscosity by: 
(5.20) 
where aj is the turbulent Prandtl number having a fixed value of 0.9. The next step up, as with a 
second order closure model, is to calculate each of the three fluxes from their own transport 
equations. 
