I suggest they try to hoan in on the best performing set-up every weekend, step by step, not getting ahead of themselves. Then try to find the common denominator between the best performing set-ups and apply that knowledge to the simulator. The issue I have is these "experiments" seem more like stabs in the dark. It's trying to take three steps at once and more often than not (if not always) they end up back on square one.
It's 'hone in'! But seriously, changing more than one variable at a time means you have no way to understand how each variable affects the result and you are indeed taking stabs in the dark.
Things that can be tried during free practice are limited but add up to many possible changes. The team can quickly change ride height, camber and pressure (to Pirelli limits), toe, wings, shock (damper) settings, sway bars, suspension interactions, engine modes, etc, so there are an enormous number of combinations. This is where the step by step approach is required. Radical changes will tell you nothing. If we had unlimited testing these issue would probably be quickly solved; simulators are great, but they are not real life, especially with aero components when there are correlation issues.