Andy Ross wrote > Vivian Meazza wrote (in a CVS checkin): > > I've removed all the features that rely on the diff to YASim > > that I posted recently, I don't expect any reaction from Andy > > any time soon! I feel a bit inclined to remind him of his rant > > against Cygwin recently. I'm willing to be favourably > > surprised. > > Good grief. If you guys are going to snipe like this, at least keep > it out of the public record. And try giving me more than 24 hours to > reply next time. Easy stuff I can handle at work while I read the > mailing list, but some stuff requires that I get home and actually run > the simulator.
Ouch! Many apologies and humble grovelling for the public whinge!!!!! > This is decidedly not a trivial patch*, and takes time to test. No > one else reported trying it, so that means I need to manually load up > the engine definitions of every turbo/supercharged engine, verify that > the plane can't reach the non-physical regime, make sure the solver > still completes and that the parameters don't change too much, Already done, but you'll want to check, I expect. > only *then* worry about what the new features mean (example: why is > there a "cutout" control? Couldn't that be done more generally by > making the wastegate value settable? Did the Hurricane even have a > wastegate? What gadget the cutout lever control?) I would do this work if it didn't? :-) > * For one, I still hate the boost function that goes negative at high > RPM, and am 60% sure it's going to hurt someone somewhere. For > another, it's clearly modelling a supercharger; it doesn't > correspond well to turbocharger behavior, nor does it provide a sane > migration path to a simulation engine that supports both in a > general way (or splits them out into separate objects). I have revised the curve: now a Hoerl power function. It's a good fit over the rpm range up to x4 peak power rpm (unnecessary: x3 is too much imho) and tails off thereafter reflecting less output as more of the compressor stalls. The output remains positive for ALL values of rpm, and won't break under any circumstances. I agree that this is optimised for a supercharger. I don't believe that a compromise between turbo and gear driven is possible for maximum realism. On the other hand this is better than the present situation. I have a curve for a turbo up my sleeve. However, it is very definitely art not science, because turbo installations vary, and dealing with throttle opening is complex. A very general model should be possible if it is felt that a near linear response is not OK. It will be necessary to identify turbo or gear driven superchargers. > Now, of course, I am out of time before work and won't be able to work > on this more until tonight. If you want to help me out, stuff like > this would be really useful: > > + Fit a boost function that is asymptotic in the high RPM regime and > doesn't go negative. More than anything else, this is what freaks > me out the most about your patch. We discussed a few earlier, for > example. Note that it can be piecewise: you don't need just one > equation. Already the case: see above. I considered a spline, but the extra complication doesn't really give a better outcome. Asymptotic ... OK up to a point, but at some rpm the compressor goes supersonic (depending on the design) and output falls away. I attempt to model this in a general way, while ensuring that the output remains positive for all rpm values, no matter how unreasonable. I'll put the output on my website so that you can see. I'm satisfied with this solution, but others are, of course, possible. > + Try the other turbo/supercharged aircraft in the command line solver > and provide output for the before/after case to verify that nothing > weird is going on. I had already checked every propeller driven YASim models (supercharger and none, even with legacy code). As I said: so far as I could see there were no adverse effects. I should have been explicit. > + Explain better why you want the new CUTOUT control and didn't just > make the wastegate setting modifiable at runtime (which simplifies > the engine model and seems more general, IMHO). The Merlin (Hurricane, Spitfire and P51d) had a Boost Control which acted on the throttle to control the boost pressure: I briefly considered modelling that, but it is adequately modelled by the wastegate in YASim (as you yourself said here earlier). The Boost Control Cutout bypassed the Boost Control. The CUTOUT control seems to me to be simple to implement, reflects the way it worked in reality, and is applicable to several models. It avoids any complex interaction between it and the supercharger output curve. However, if this change gives you real difficulty, then I suppose setting the wastegate value to a very large value would be the equivalent. Bit of hack though and therefore feels unprofessional when a proper solution is so easy (and available). > + Convince other folks to try the changes and report success. That would be nice - the silence is deafening so far :-(. Anyone like to try???? > Just for the record: if this were an obvious fix or an > simple/orthogonal new feature, then I would just apply it like I apply > other fixes. It is neither, which means (I'm sorry) we are both going > to have to do more work. Pissing me off isn't helping. Apologies - I should have pursued you on this issue months ago, rather than letting it go dead and getting myself wound up. Work is no problem: onward and upward. V. _______________________________________________ Flightgear-devel mailing list [email protected] http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
