Hi Kai,
You are using the "Orifice (2P)" block to model the valve, right? Good choice, I'd say, but there are a few things you need to pay attention to.
1. the S signal to the orifice block is supposed to be the spool position, with a unit of length (m/cm/mm/inch, etc.). The parametrization of the block will translate this position to opening area. This is similar to how most of the hydraulic valve blocks are parametrized. Giving it command of 0 and 1 without specifying a unit consistent with the block parameters, may give you unexpected results.
2. Trying to open or close a valve instantaneously can cause numerical problems as well. And it's not really physical. Consider add a 1st order transfer function or add derivatives in the Simulink-PS converter to slow it down a bit.
3. A bang-bang type of control algorithm with no deadzone in the midde is usually not optimal for refrigeration loop, either. You may want add a threshold around your target (delta_T=0), beyond which the valves will act and switch direction.
