Basic explanation of turbulence models
 

I've been searching for a while for a low-level description of how the turbulence models work and am not finding very much. So far I have:

K-e:
Uses turbulence energy (K) and it's rate of dissipation (e). Good standard model that works in most industrially relevant cases.
Not good with:
- flow reattachment
- flow seperation
- flow recovery


SST:
Uses K-e in free flow and K-w (w is the frequency of eddie currents) which is better for wall boundary layers.
- bad with flow recovery after flow reattachment


BSL:
Baseline Reynolds Model? Can be used with a rougher mesh?

SSG:
Speziale-Sarkar-Gatski. Better than BSL for high Reynolds numbers?

As you can see I have pretty much no information on BSL and SSG. Any information will be greatly appreciated!

Cheers
Ash :)

Dear Ash,
I suggest that the help documentation of any CFD code may help you. For example, the section of "Turbulence Models" in CFX Modeling Guide offers a basic capability description of various turbulence models. And the Theory Guide tells how these turbulence models work.
Sorry for no detailed information to help you.

The turbulence modelling textbook "turbulence Modelling for CFD" by Wilcox is a excellent textbook in this area.

To correct some of your points:
- k-e, has known problems at stagnation points, separations (both detactment and reattachment) but pretty good for bulk flows. Very poor at rotating flows and anisotropic turbulence.
- SST - better at all these conditions, has models available which might help for rotating flows and turbulence transition and a few other special cases. This should be your default selection for a turbulence model unless you have a very good reason to choose something else.
- All the Reynolds stress models - can capture flow rotation and anisotropic turbulence, but rarely end up being much more accurate than a reasonable 2-eqn model anyway due to the large number of assumptions required to get them to work. Also devilishly hard to converge.
- And then there is DES, SAS, LES, DNS - these are advanced models which you should only consider when you have a good knowledge of turbulence models. Using them without knowing what they do is a receipe for disaster.