The volumes of sparge water are to be measured at the temperature of addition, which means that the thermal expansion is configured into the volume. When you set the thermal expansion to zero, it will give you the volume at room temperature effectively lowering the volume predicted.
If you measure your wort from the mash tun or after pulling your bag of grains (for BIAB brewing) like I usually do, then the thermal expansion becomes a factor in measurement from BeerSmith to actual. If the volumes were calculated at room temperature, then there would always be an offset in measured (at wort temp) versus the anticipated (at room temp) and the calculation for mash efficiency would also have that offset.
At the end of the boil, BeerSmith backs out the volume for the thermal expansion of the wort because most people measure the volume as volume into the fermenter and excess for trub and loss at room temperature (or close to it).
I have my thermal expansion set at 2.5% which is close to the thermal expansion at mashing temperatures and is where I will do my actual volume measurement of volume from mash and from sparge.
The pre-boil gravity is adjusted based upon the volume of water, which reacts to the thermal expansion value. This may be, from my process calculations, an issue that should be addressed in the future by calculating the gravity based upon a 'room temperature' volume, especially since gravity readings should always be taken at room temperature (temperature of hydrometer calibration). In the end, it really is a minor change to most of my brews and the actual error pretty much disappears in the error of my volume measurements with hot wort.