When you are testing an airport for fps, you should only use a default airplane first. Do you see the same variation with a default airplane too ?
Not only that, you should test with AI at 0%, otherwise you are not testing the airport fps, you are testing the efficiency of the models of the AI airplanes you are using. Since AI can affect fps dramatically, due to parking layout, schedule (which means difference in time of the day and day of the year), if you are testing an airport for fps with AI turned on, you are likely to get confused to check where the problem is.
AFTER you tested with a default airplane, with AI off, IF you are still seeing such large difference in fps, THEN you can start looking at the scenery, either the airport itself, or any possible conflicts/problems with the rest of the installed scenery.