The effect of gravity is usually negligible unless your part/assembly has high mass. The question to ask is: will the part/assembly mass impose significant stress/deformation? If not, ignore it. You can easily do a test with and without gravity load and see if it makes a difference to the results.
You should pretty much always use gravity as a load. One case that might not need it is if you were to design a machine that would be constructed out of parts that were already in a zero-g state, and the machine would always be in a zero g state. So unless you work for the International space station, gravity is a good idea. The benefit of including it is that if the effects are not negligible, you don't need to re-run your study. Also, adding gravity to a study does not add any extra time to the solver stage or the mesher stage.
We had a guy at our work here who ran a study for 8 hours (overnight). In the morning, he declared, "well, it looks like everything is working, so I'll just turn on gravity and run it again!" What a waste of time.
thanks stuart and david got a fair hold on the topic thanks a lot