Having a check value (or any closed valve) with absolutely zero leakage is in general a bad idea. It basically changes the set of equations the solver needs to solve. If you are trying to resolve the issue of slow simulation by "fully" closing the valve, well, I suspect you may want to try the opposite, i.e. increase the minimal/leakage area. Let me explain.
Now the leakage. If it's set too small, what may happen is that the solver has to be able to resolve from a tiny flow, say 1e-10, to fully open (say 1e3). The difference in the order of magnitude (1) may not play well with your tolerance settings and (2) make the matrix to be solved really "stiff". I would suggest you increase the leakage area and see if the simulation speed improves, while keeping an eye on the leakage flow rate. A few orders of magnitude difference between open flow and closed is usually sufficient and realistic, but don't try too hard getting "zero" flow.
Hope this helps. If you could provide more info after you run the solver profiler or if you could share your model, we can take a closer look.