If the SEER 15 unit is single speed and the SEER 16 is two speed, and if there are ducts in unconditioned space, the SEER 16 run will have higher duct losses due to longer run times (see field testing evidence for this). You might re-run your test cases w/ ducts in conditioned space to isolate the effect of a SEER change. You should also make sure that nothing else is changing in the runs where you change SEER; for example, changing the water heater can result in different internal gains that the HVAC system would see.

There is also some similar discussion here.

EDIT:

Also, make sure you are looking at the combined result for both space cooling energy and cooling fan energy. There are built-in assumptions about when the fan efficiency increases as a function of SEER, so it is possible to see increased space cooling energy while the combined space cooling plus cooling fan energy is lower as expected.