By David C. Copley
June 29, 2002
One of the most common subjects discussed on flight simulator forums is that of flight models.
"Which is most accurate? Stock or 1%? (or other developer)" are common questions.
I believe the basic premise of the 1% files is to achieve 1% of certain flight characteristics through the application of engineering principles and virtual flight testing. But, outside of MS development staff, we are making educated guesses at best on flight model parameters; our understanding of the overall flight model/dynamics "engine" is incomplete.
So we really have a quandary. On one had we have the makers of the overall flight model/dynamics "engine" and their stock planes are reportedly based on accurate data. In principle, this combination should produce the "most accurate" models.
Yet on the other other hand, we have third-party designers (the 1%-ers, BA, simTech, and others) trying to decipher the overall flight model/dynamics "engine" based on hours and hours of experimentation, solid engineering principles, and quite a bit of demonstrated research from original sources. Because it is likely these developers have spent more time researching (compared to MS developers who are likely constrained to development schedules set by marketing groups), these models could potentially be far superior if the dynamics engine were well understood.
But here's my take, and the guiding principle of my flight modeling. There are many, many aspects of flight dynamics we simply cannot model well, such as weather, manufacturing variability, engine and other system and degradation, effects of maintenance (positive and negative) etc. While the published specs of a plane make for interesting comparisons, etc. the composite uncertainty of all the aspects of flying probably result in the actual planes themselves not matching their very own specifications to within 1%.
Consider this 'thought experiment.' Take one actual WWII plane, say a P-40, and let a pilot fly it on a certain day. He takes the plane through take-off, some rolls, max speed tests at certain altitude, dives, turns, etc. After the flight, all the performance data are recorded. Then ship that same P-40 to another location, with different fuel supplies, different mechanics, different weather conditions, etc. and have a different pilot perform the same maneuvers. Then do that another 100 times. Then do that with 100 planes. Perform statistical analysis on these results and I would imagine the standard deviation of each and every flight spec would be well outside 1%.
I applaud the 1% methodology and I benefit greatly from the research and development of all those who continue to share their knowledge and discoveries of the basic dynamics engine. But if these complex physical systems themselves in real life did not consistently perform within 1% of their own specifications, we really cannot crown that approach and its models as king. I don't believe those directly involved with 1% ever intended that. It seems that their files are being used well beyond their original context (to provide equalize models for historical re-creation) and many assume the 1% files are the most accurate.
I now approach flight simulation as a combination of art and science. As soon as an artistic element is introduced, we open everything up to interpretation. Designers often place emphasis on some flight dynamics aspects more than others. For instance, I have tried to place emphasis on certain non-measurable flight characteristics in my P-38 models. But, in order to achieve a certain feel for a few characteristics I willingly sacrificed others.
One specific example: P-38 pilots that I have interviewed and printed first-hand accounts state that it did not take much effort to lift the P-38 off the ground once you reached about 100 mph. And it should only take about 1,100 ft of runway. To obtain this effect I made adjustments that made it so it could never reach the specified top speed at certain altitudes. My reasoning was this: I take off a lot more than I fly top speed for an extended duration. Compare this to the 1% methodology, where published top speeds are one of the primary design constraints. But the 1% P-38 jerks off the runway well beyond 100 mph (and beyond 1,100 ft) and requires considerable pull on the yoke. My P-38s lift of easily right around 100 mph near 1,100 ft. Yet my P-38s do not attain published max speeds at published altitudes. And, as of this writing, I have yet to model certain aspects of the flight characteristics as well as the 1% folks.
Thus, my philosophy is to capture and recreate the "essence" of the flight model, while making it fun for me and for those who fly my planes. After all, this is really just entertainment.
(PS. Again, I emphasize that this article is not a slam on the 1%
group nor any other group attempting realistic flight models.
I'm just stating my philosophy compared to others.)
During the ten years I worked on FS P-38 models, I found myself having to defend my approach to flight dymnamics on more than one occasion, so I wrote this piece and included it in the documentation for all the models.
There was, at the time, a group of highly talented indivisuals claiming the ability to match aircraft performance to within 1% of published data. While I found their approach intriguing and their tools useful, I never felt the need to adhere strictly to this approach (and others who followed similar methods).