That means HTC has a 0.56% error from real life. Im betting that variation from plane to plane matches this.
That 0.24% error. To me it looks likes HTC has things pretty close.
I think you will find larger errors in the modeling of other planes, maybe HTC should worry about them first.
This raises an interesting point and a question in my mind.
Has HTC bothered to band the models such that they exhibit variation? My gut-level guess: I sincerely doubt it. However, consider how neat it would be that you might, for example, saddle up a D9 only knowing that its performance fell within a certain error band.
If so, it would also be really neat if you could "save" that ride - once you found one you liked. Of course, your saved ride would last until somebody turned you - and it - into a crispy critter. Then it'd be back to the bullpen.
You'd need some randomized constant on some of the key physicals, I think... e.g., power output or normalized drag (reflective of variable fit and finish) right? I mean, it doesn't seem like it would be that difficult.
Of course, getting the data for that variability, otoh, might be a serious pain. Do you reckon they kept SPC data at the FW plant? This was pre-Deming, after all. However, they might've done small sample tunnel tests. I can't imagine they would've seen any need to run anything more than a few models, though, and you'd need something like 30 production examples to get exemplary coefficient variability data.
