The 2D simulation has an effect in obtaining this difference. Fluent recognises the jet depth is 1 m, which is not true ... To solve this issue, you need to have few cells in the z direction, , then apply symmetry plane. You can actually simulate 1 quarter of the room, and apply symmetry ... nMore smart solution ... make axisymmetry ... having rectangular or cylindrical room will not make a difference if it is large enough.nThe captured photo, the impinged fresh water particles, started to diffuse horizontally when the jet velocity became very small ... may be the salinity played a role.nFinally, I believe, if you left your simulation to continue, you might get something close to the photo .. nThe last density shows that the mixture density is similar to the impinged water density. I can't explain the physics of this ... it does not make sense . Probably this happened when you were running the simulation with wall at the top. !!nnmy recommendation, repeat the simulation ... asymmetric ... outlet pressure on top. And monitor the density in few points away from the jet.n