can not converge if use variable density

alfred

Hi, there, in my model, if I use constant water density, my computation converges after around 100 iteration; while if I use piecewise linear density of water (density changes from 1000 to 972, for 278 k to 353k, no matter how many intermediate points used in between), it does not converge even after 10k iteration. The energy error still at order of 1e-4? what can I do to make it converge? thanks!