I would use the dead load plus best estimate of actual live load. In storage structures like warehouses, grain elevators, and water tanks, the live load is very real and cannot be neglected.
I would use DL + LL, both unfactored for service loads, since your are evaluating serviciability. If you want to set a range of estimated settlement, you can also calculate it with factored loads.
Pretty sure US AASHTO Service I is DL + LL unfactored for settlement. Some consideration of how the live load is determined and applied to a foundation system is needed though. There are some worst case scenarios that may overstate the typical live load condition for a settlement analysis. I think of multi-lane bridges with various loading conditions that can yield higher pressures. I think one needs to look for the highest average pressure over a effective footing width and not some odd structural combination that only exists for brief moments in time.
Guys thank you. I also use G+Q. But I have to prove someone that this is true. Can you offer me any resource that proves your case?
Doctormo, yes, Service I state in AASHTO. Also they have a sentence that OVERALL STABILITY ANALYSIS SHOULD BE PERFORMED BASED ON THE SERVICE I. So this is the sentence I am looking for, only instead of overall settlements, it should say, SETTLEMENT CALCULATIONS.
Doctormo, that is the one I am looking for, but I couldn't see that on Caltrans Calofirnia Foundation Manual, either in 2015 or older versions. Can you specify exact location of that sentence?