Hi Michael,
I think there are a couple things at play here:
First, there is currently no volume at all between the two reservoirs. The hydraulic line connections do not imply any fluid volume, and the valves have no volume either.
And the valves can never be fully "closed". Instead, the valve passage area drops to leakage specified in the block. This is done both because it's physical (no zero-leakage valve exists) and numerically much more efficient and stable. "Shutting" a valve changes the system of equations and that's a crazy thing to allow to happen in the middle of a simulation.
Since there is no volume, there is not any fluid to "hold" the initial pressure. The system of equations is instead solving for the pressure drop under this tiny leakage flow. Adding some realistic volume between the valves should give you results that are more intuitive.
