Melting ice - Sudden phase change of entire interior
 

I'm trying to model the melting of a piece of ice next to some water using the melting/solidification model. I've attached the model ("Model") so you can see where the ice is and where the water is (of course, the whole rectangle is meshed even though it doesn't show). The goal is to get something that makes sense and looks right, and then i will change properties and layout and so on to be more realistic later.

I'm getting wierd results that i will explain at the bottom, but if you're interested here's first how i set up everything:

I am doing a transient study with gravity applied.
I defined a "ice-water" material with the properties of liquid water using boussinesq density. I also defined the latent heat, and i set T_liq = 270K and T_sol = 272K just to "give it some play". For boundary conditions, the edges of the ice block are adiabatic, and there is a 0.1kg/s mass flow in from the bottom edge of the water and out from the top - as if water is rising up - at 300K. The left edge is specified shear = 0 and also 300K. I initialized using hybrid initialization and patched the water interior at 272K, and the ice at 268K.
I changed Energy URF to 0.75 and liquid-factor URF to 0.2, because i was getting floating point error with the default values (is this even the best way to fix that problem?). My timestep is 0.1 and i have max 10 iterations per timestep.

RESULTS:
when i do mass-fraction contour at 20 sec ("time_20"), i see the result i expected, the water slowly melts away at the ice. But at 30 sec ("time_30"), there is a sudden phase change for the entire ice region!?
I also attached the temperature contoures at 20 ("temp_time_20") and 30 ("temp_time_30") sec, where it's very clear something is wrong as well. Do you have any ideas what i'm doing wrong and what might be causing this problem?

Answer
 

Just wanted to note to anyone interested that when i changed from SIMPLE to PISO and from second order pressure to PRESTO! i stopped getting this problem.

ANSYS recommends PISO for transient analysis and PRESTO! when there is large changes in momentum source/sink, which is very much present when using enthalpy-porosity method of modelling phase change, so i attribute the error to that.