As an update to this, and to those who are interested, this was clarified by the WASY support team.
In cases of a changing concentration BC, there is a difference between the global density ratio parameter and the density ratio applied for that BC. The global parameter should remain fixed and relates to the reference and maximum concentrations which are set by the user in the global transport settings (C0 and Cs).
When density dependent modelling is active, the Darcy flow equation includes a buoyancy term K* ([(Rho - Rho0)/Rho0] * grad(z))
In FEFLOW, the density term is determined by concentration by:
[(Rho - Rho0)/Rho0] = density ratio * ([C - C0]/[Cs-C0])
So the buoyancy term in the Darcy eqn becomes K* (density ratio * ([C - C0]/[Cs-C0]) * grad(z))
When you define a saltwater head BC is FEFLOW, it uses the global density ratio by default to set the equivalent freshwater head, using:
h_fw = h_sw + density ratio * (h_sw - z) where z is the node elevation.
This assumes C at this boundary = Cs. However, when C is not Cs, h_fw needs to be calculated using a different density ratio according to [(Rho - Rho0)/Rho0] ... where Rho is the BC density and Rho0 is the reference density, usually the minimum density in the model.
So this pretty much requires an IFM...
And it is not advisable to use free and movable surfaces for 3D density problems... So it's confined mode only.
Lots of horizontal and vertical discretisation is a given...