Re: [Flightgear-devel] YASim 747-400 climb performance
Jim Wilson <[EMAIL PROTECTED]> said: > Andy Ross <[EMAIL PROTECTED]> said: > > > > > Yes! That's it! > > > Yes, I'm seeing a big difference. > Andy, That sure did the trick. To make a long story short, I screwed up the update the first time (had a two year old doing some major distracting :)). The YASim 747 is working great! Thanks, Jim ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim 747-400 climb performance
Andy Ross <[EMAIL PROTECTED]> said: > > Yes! That's it! > Yes, I'm seeing a big difference. Still maybe not quite up to spec (but don't want to say anything without doing more tests). We've gained at least 10k feet in the last week though :-) Very nice! Best, Jim ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
RE: [Flightgear-devel] YASim 747-400 climb performance
> Not at all unlike being a test pilot. :) This is the analogy I like to use, as well. It's one reason I'd really like to see us model EAFB well. Where else would we test our models? We wouldn't want to cause virtual property damage anywhere else! ;-) Jon smime.p7s Description: application/pkcs7-signature
Re: [Flightgear-devel] YASim 747-400 climb performance
I lied: > There's another 747.xml file attached. Sigh... someday I'll get one of these right on the first try. Andy -- Andrew J. RossNextBus Information Systems Senior Software Engineer Emeryville, CA [EMAIL PROTECTED] http://www.nextbus.com "Men go crazy in conflagrations. They only get better one by one." - Sting (misquoted)
Re: [Flightgear-devel] YASim 747-400 climb performance
Jim Wilson wrote: > David Megginson said: > > What happens when you reduce the fuel load? > > Hmmm...nothing because it won't reduce. I'm showing full 19k+ lbs > for three tanks no matter what I do with the "fraction" figure in > the config and can't dump during flight it either. Yes! That's it! Actually, the bug report is bogus -- /sim/fuel-fraction works just fine to modulate gross weight, and always has (do you maybe have a typo in your configuration file?). But it got me thinking about checking the weight during solution, and that's where the problem was. For efficiency, YASim separates the computation of the inertia tensor and gross weight from the normal iteration-to-iteration operations on the RigidBody object. The recalculation is handled outside, as part of the RK4 integrator. But the solver isn't using RK4 -- it's solving for instantaneous force only. So while it was setting the fuel fraction correctly, the change in fuel weight was *not* being applied to the aircraft. Basically, every solution produced was for an aircraft at zero fuel weight. So every non-zero fuel configuration was heavier than the cruise setting and thus incapable of reaching its performance numbers. Bingo. There's another 747.xml file attached. It's needed to work with the new changes (which changed a lot -- the older model doesn't have enough tail authority to trim for cruise with the fixed code). The other aircraft seem to be OK, but I didn't fly them extensively. Folks should try them out and look for stuff that might need to be retuned. No doubt there are still bugs in there, but this definitely fixes the climb performance problems on the 747. This is actually kinda fun -- flying a fundamentally broken aircraft very carefully, and looking for tiny clues as to its performance difficulties so they can be fixed. Not at all unlike being a test pilot. :) Bring 'em on. Andy -- Andrew J. RossNextBus Information Systems Senior Software Engineer Emeryville, CA [EMAIL PROTECTED] http://www.nextbus.com "Men go crazy in conflagrations. They only get better one by one." - Sting (misquoted) ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim 747-400 climb performance
David Megginson <[EMAIL PROTECTED]> said: > What happens when you reduce the fuel load? Hmmm...nothing because it won't reduce. I'm showing full 19k+ lbs for three tanks no matter what I do with the "fraction" figure in the config and can't dump during flight it either. Reducing the tank capacity by half causes it to take off like a rocket and then it'll climb to 38200ft. Best, Jim ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim 747-400 climb performance
Jim Wilson writes: > In any case, flying very carefully I can barely get it up to FL320 > with the change in preferences.xml. Going further just doesn't > seem possible. I tried serveral AoAs and just couldn't get it to > go up much without falling back. What happens when you reduce the fuel load? All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim 747-400 climb performance
Andy Ross <[EMAIL PROTECTED]> said: > I'm pretty sure we're seeing the same behavior here, actually. I'll With Dave's change to prefrences.xml I can get up to FL320 before running into trouble. Really tried to go higher...no luck. > take a look at the reduction in climb performance, but I'm pretty sure > you're just being fooled by the slower speed (once you drop below 260 > knots or so, it takes a *long* time to get back to 300, even if you > dive). No, I don't think so and yes I know. But haven't had a chance to verify this with the latest changes. > Here's my plan for testing 747 climb: > > 1.) Set autopilot to runway heading before takeoff, then disengage > with ^H. Verify that all trim controls are zero. > 2.) Throttle up. > 3.) Rotate at 190 knots. Gear up. Engage autopilot heading mode. > 4.) Hold the nose at 15° until the aircraft reaches 250 kias. > 5.) Hand fly the aircraft at 250 knots until 10k feet. > 6.) Reduce back pressure on the yoke as the aircraft accelerates to > 300-310 knots. Trim up very (!) slightly to hold this speed. > Use very small movements of the stick to damp the phugoid. I flew this exact protocol in the last test. BTW I might have mentioned before, but wasn't clear that when I do use the autopilot I'm tweaking the config at various flight levels with the property picker (/autopilot/config). Unfortunately the autopilot's "min climb speed" setting doesn't work quite right, so you have to watch the ias. Anyway this is a lot different than just entering an elevation like we do with the cessnas. In any case, flying very carefully I can barely get it up to FL320 with the change in preferences.xml. Going further just doesn't seem possible. I tried serveral AoAs and just couldn't get it to go up much without falling back. Best, Jim ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim 747-400 climb performance
Jim Wilson wrote: > But I'm still hitting top at just over FL300. When I switch to the > old weathercm to see if I can get the FL360 yasim won't initialize the > 747 (tried dc3 and c172 and they both come up). Sorry, I wasn't clear. Not the weathercm stuff -- YASim used to use its own Atmosphere class before the environment was exported through properties. The changes are back about 1-2 revisions in YASim.cxx. But as David points out, setting the sea level pressure to 29.92 in the preferences.xml file will work just as well. > Any chance we're looking at another 2.95.2 math issue here...for both > this bug and the differences in what we are seeing for flight > performance? > [...] > The evidence for my "time" question is that after I hit a stall and > drop down several thousand to recover full speed, that attainable > altitude ceiling decreases by 5-10% or so. I'm pretty sure we're seeing the same behavior here, actually. I'll take a look at the reduction in climb performance, but I'm pretty sure you're just being fooled by the slower speed (once you drop below 260 knots or so, it takes a *long* time to get back to 300, even if you dive). Here's my plan for testing 747 climb: 1.) Set autopilot to runway heading before takeoff, then disengage with ^H. Verify that all trim controls are zero. 2.) Throttle up. 3.) Rotate at 190 knots. Gear up. Engage autopilot heading mode. 4.) Hold the nose at 15° until the aircraft reaches 250 kias. 5.) Hand fly the aircraft at 250 knots until 10k feet. 6.) Reduce back pressure on the yoke as the aircraft accelerates to 300-310 knots. Trim up very (!) slightly to hold this speed. Use very small movements of the stick to damp the phugoid. If the pressure fix is in, you should see 1000 fpm up through FL300, and a service ceiling of about FL350-360. Fly it back down to sea level and repeat, making sure the speeds are correct. It worked for me a few times in succession. There's still a bug in there; you can see it by the difference between the solution target and actual performance at 36000 ft. Andy -- Andrew J. RossNextBus Information Systems Senior Software Engineer Emeryville, CA [EMAIL PROTECTED] http://www.nextbus.com "Men go crazy in conflagrations. They only get better one by one." - Sting (misquoted) ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim 747-400 climb performance
Andy Ross <[EMAIL PROTECTED]> said: > Jim Wilson wrote: > > Andy Ross said: > > > The bad pilot technique theory is looking pretty good. If YASim can > > > get to 1400fpm at 300 knots, then I move that it be declared > > > officially innocent on all counts. :) > > > > Well maybe not yet...it is better with today's patch...but still run > > out of steam in the 29000 range. It also may be time related. In > > otherwords the situation gets worse over time. > > It's not time related. First off, ditch the autopilot. It gets stuck > behind the power curve every time I use it. When it does this, it > very rapidly (in the space of 30 seconds or so) gets you trimmed all > the way down to 200 knots, at which point the engines can't keep you > in the air anymore and you have to dive to regain airspeed. Just trim > for speed; this is really easy to do in an aircraft this big. You can > verify that the climb performance drops slowly off toward zero as the > altitude increases, and recovers as you descend. I've done a zillion > of these this afternoon. :) Actually, I'm very familiar with how the autopilot works and the issue you are describing :-) For example's sake however, 300fpm and full throttle should have been maintainable with the autoplot code as it is. Believe me I'm babying this thing to get it up as high as possible. That said, my latest test (trimming for speed, no autopilot) shows a slight increase in the attainable altitude ceiling with your xml adjustment. But I'm still hitting top at just over FL300. When I switch to the old weathercm to see if I can get the FL360 yasim won't initialize the 747 (tried dc3 and c172 and they both come up). Any chance we're looking at another 2.95.2 math issue here...for both this bug and the differences in what we are seeing for flight performance? The evidence for my "time" question is that after I hit a stall and drop down several thousand to recover full speed, that attainable altitude ceiling decreases by 5-10% or so. It's only a question about possible accumulating error trouble, but you know the code and if my question doesn't turn on a light then don't worry about it. Just trying to help :-) Best, Jim ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim 747-400 climb performance
Andy Ross writes: > First, the air pressures returned from the environment system don't > agree with the standard atmosphere that YASim uses to do its > calibration. They match pretty well at sea level, but diverge as > altitude increase. At 35000 feet, they're too low by 20%, which is > substantial. Could someone check and see where the environment system > is getting its numbers? >From YASim, but I may have done it wrong (I converted from metric to more common aviation units). Look around line 257 in src/Environment/environment.cxx, or search for "Atmosphere model". It could even be an interpolation problem -- I'm using the regular SimGear interpolation table, while YASim uses a custom interpolation table. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim 747-400 climb performance
Jim Wilson wrote: > Andy Ross said: > > The bad pilot technique theory is looking pretty good. If YASim can > > get to 1400fpm at 300 knots, then I move that it be declared > > officially innocent on all counts. :) > > Well maybe not yet...it is better with today's patch...but still run > out of steam in the 29000 range. It also may be time related. In > otherwords the situation gets worse over time. It's not time related. First off, ditch the autopilot. It gets stuck behind the power curve every time I use it. When it does this, it very rapidly (in the space of 30 seconds or so) gets you trimmed all the way down to 200 knots, at which point the engines can't keep you in the air anymore and you have to dive to regain airspeed. Just trim for speed; this is really easy to do in an aircraft this big. You can verify that the climb performance drops slowly off toward zero as the altitude increases, and recovers as you descend. I've done a zillion of these this afternoon. :) That said, here are two more layers to this giant onion of a bug that won't die: First, the air pressures returned from the environment system don't agree with the standard atmosphere that YASim uses to do its calibration. They match pretty well at sea level, but diverge as altitude increase. At 35000 feet, they're too low by 20%, which is substantial. Could someone check and see where the environment system is getting its numbers? The ones YASim uses are typed in from McCormick, who got them from "The ARDC Model Atmosphere". I verified this afternoon that they are correct, so I'm pretty sure the bug lies elsewhere; some difference is fine, of course, but 20% difference at FL350 is a lower low pressure system than any hurricane. Ever. Replacing the new environment system with the old one yields significantly improved performance at altitude. Second, I realize now that the cruise performance numbers in the 747.xml file are tickling a problem. They are specified as a maximum speed of 530 KTAS, which I got out of a book somewhere. Problem is, this corresponds to 0.92 mach. At that speed, the plane should be seeing more drag than YASim calculates because of (as yet unmodelled) transsonic effects. But since YASim doesn't model this extra drag at high speed, it solves for an airplane with extra drag at all speeds. I've replaced the cruise setting in my .xml file with the long-range cruise numbers I found in David's link (the 75% throttle is just a guess), and this makes things better still. ... ... But, after all that, things still aren't quite right. The plane should, by the definition given in the solver output, fly at 490KTAS, FL360 with an AoA of 2.5 degrees. But it doesn't. At that altitude (which is attainable with the above modifications), it wants to fly much faster to stay in the air. I'm still working on it. And did I mention not to use the autopilot? If you must, watch it carefully. If it tries to trim the plane below 300 knots, shut it off! Andy -- Andrew J. RossNextBus Information Systems Senior Software Engineer Emeryville, CA [EMAIL PROTECTED] http://www.nextbus.com "Men go crazy in conflagrations. They only get better one by one." - Sting (misquoted) ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim 747-400 climb performance
Andy Ross <[EMAIL PROTECTED]> said: > The bad pilot technique theory is looking pretty good. If YASim can > get to 1400fpm at 300 knots, then I move that it be declared > officially innocent on all counts. :) > Well maybe not yet...it is better with today's patch...but still run out of steam in the 29000 range. It also may be time related. In otherwords the situation gets worse over time. Haven't quite nailed it down, but when I ran a test with the autopilot set at just 300 rpm it went into a stall around 24,000ft. Certainly at 300fpm we should be able to maintain a climb to cruise altitude. Best, Jim ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
[Flightgear-devel] YASim 747-400 climb performance
More evidence. The following quote is from an HTMLized google cache of a file named "747operations.pdf". The file itself is gone from the web, unfortunately. It appears to be a "POH" compiled for an MSFS virtual airline: > After climbing as described above to 10,000 feet, reduce climb to > 500fpm. Accelerate to 280-300kts. If using autothrottle, increase by > 10-20kts, wait until aircraft attains that speed, and increase another > 10-20kts. Once accelerated, increase climb back to 1,400 to 1,800 > fpm. Once you are stable and climbing, you may turn off the fasten > seatbelt sign. At about 12,000 feet, switch to mach hold and set to > .55-.60. Increase up to .70 if needed. Do not drop below 240kts. The bad pilot technique theory is looking pretty good. If YASim can get to 1400fpm at 300 knots, then I move that it be declared officially innocent on all counts. :) Andy -- Andrew J. RossNextBus Information Systems Senior Software Engineer Emeryville, CA [EMAIL PROTECTED] http://www.nextbus.com "Men go crazy in conflagrations. They only get better one by one." - Sting (misquoted) ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel