Can you please describe how you took the test data. How did you apply load to the alternator to ensure that you were running on the power curve? Did you just put the throttle wide open and lug the engine? If the engine was truly under the same load for both tests (engine operating under the same conditions, with same throttle position and fuelling etc) then there's an error with the way you've measured power in one instance.
On your power curve for both tests, plot:
- fuel flow rate
- air mass flow rate
- exhaust temperature
Then we'll see whether the power was different in both instances or whether it's just a measurement problem.