A work lesson I keep having to relearn.
If you use fields to push and force a fluids movements you open the door for unexpected and unnatural results.
Use fields sparingly. Think hard as to how you can get a fluid to move with only internal fluid dynamics (not the internal "turbulence") and weak motion fields. Sometimes it's turning a fluid in it's side so buoyancy is left or right. Sometimes its having a strong + temp and weak - density buoyancy.
I spent time doing a crack dust sim and kept getting crappy looking sim's. I played with all the settings, even conservatively went back and eased into it. I had a line of emitters along a curve and a turbulence field in the front to disrupt the "air" before the crack emission. The results were somehow unnatural, even at its best. Not to say I won't try it again but its not how dust would react in real life and it looks almost unconsciously fake when you see it CG. What I mean is, when a crack happens in real life there's no turbulent air on the leading edge of the crack, so adding one in CG doesn't help.