Re: [Flightgear-devel] Engine models: start-up and commonality between FDMs
On 11/13/02 at 12:16 AM Julian Foad wrote: David Luff wrote: It looks to me like you've got 2 too many curly brackets in doEnginePower, although I could be misunderstanding what you're doing there. Yes, I have got too many. This is the friction that was applied only when starting; I was making it permanent but haven't finished with it. Do you agree that it should be permanently in? Does a constant torque sound about right? That sounds more likely than constant power (which means decreasing torque), to me. Conventional friction would give constant torque; I'm not sure how oil and air viscosity behave, but I'd expect the torque to increase at higher speeds rather than decrease. Yes, friction mean effective pressure (basically torque non-dimensionalised by engine volume) goes up with speed. Typically it could be about 3 times as great at 2500rpm as at idle. IO360.cxx in LaRCsim has data from a published friction model in it that reflects this, and this ought to be put into FGPiston.cpp to provide resistance to windmilling operation and cranking instead of the crappy -x HP that I had in there as a hack. In fact, I'm not entirely sure why it got left out - I think it went into IO360.cxx after the JSBSim port and then never made the jump across. In the long term I'd like to see it go in during running as well, together with a model of how efficiently the fuel energy is converted to mechanical energy, which should be quite portable across most GA piston engines. However, the present power correlation we have is *brake* power, from Phil's Cessna flight management computor (Phil being the guy who originally started writing the file). Hence we can't put the friction in during normal running, or the power at the prop will be wrong, which will be noticable in the normal flight envelope. I don't understand how it could have worked with no resistance implemented. A propeller hardly provides any resistance at low speeds, so I would have thought you would have needed to tweak the developed power down to almost zero at idle. Yes, it was hard to get the engine to idle slowly - in fact I'm not what we did in the end! I do recall that my first guess at starter motor torque had to be revised downwards considerably since the imbalance caused the speed to go unstable in the first time-step. What I am concerned about is the throttle minimum being set to 0.2. ... Ah, thank you for explaining this. I had not understood the mapping onto manifold pressure and the power correlation. It certainly sounds like the power correlation is the thing to un-tweak instead! This puzzled me: the manifold pressure seems to be modelled as (for a given throttle position) independent of speed. When a real engine is running fast and you cut the throttle, the fast air flow will cause a very low manifold pressure which will then rise to its new steady value as the engine slows down. Without this effect, throttle changes will not take effect as quickly as they should and the speed variation with load changes will not be right. Maybe the effect is too small to be important? I might be attempting too much here; I know how car engines work but don't have data to work from (or a lab), and I don't have experience of modelling them either. I will tread carefully and check with you again when I make some more progress. During engine running, manifold pressure is primarily a function of throttle position, but yes, also affected by engine speed. However, the throttle position effect is considerably larger, and is currently all we have in the model. The effect with speed should be modelled though, in order to make it harder to set a given engine speed and man pressure with a variable pitch prop. Obviously, in the limit when the engine speed is cut to zero, the effect of speed dominates completely, and the man pressure goes to ambient. This is implemented. Cheers - Dave ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Engine models: start-up and commonality between FDMs
On 11/12/02 at 12:28 AM Julian Foad wrote: Ah, glad you're there. If you're interested and have time to look, my current attempt is at http://www.btinternet.com/~julianfoad/fgfs/JSB_piston_engine.diff http://www.btinternet.com/~julianfoad/fgfs/engine_sound.diff but, as I said, not finished. (Well, it will never be finished. I mean not completely satisfactory as a patch yet.) It removes some of the arbitrary bits - especially the non-linear bits like if RPM N then ... - and makes starting and idling nicer (especially at throttle less than the minimum allowed idle setting - it was fun running it below 500 RPM, on the unstable side of its power curve, after I put the friction always present but before I put the idle adjust constant in there). It looks good, although I haven't tried it out since I don't have any sound drivers installed. A couple of points though. It looks to me like you've got 2 too many curly brackets in doEnginePower, although I could be misunderstanding what you're doing there. What I am concerned about is the throttle minimum being set to 0.2. The throttle values are meant to represent the available physical range of throttle movement using 0 - 1.0. I don't understand why you've put the 0.2 bottom limit in. I assume real planes will idle with the throttle on the stop in the same manner as a car. The air getting through in this case is set by the minimum manifold pressure in the engine model. By putting the 0.2 limit in you've effectively changed the range of throttle movement to be 0.2 - 1.0 and hence raised the minimum manifold pressure obtainable when running, which was set at about 0.3bar MAP (~6inHg), a value I measured from an idling car engine in the labs. If this was what you really intended then you should raise the min man pressure if you think its too low rather than munging the lower throttle value. If you think the min man pressure is OK but need more power at the bottom end to overcome the friction you should tweak the power correlation, which I had tweaked downwards in the idle region from Phils original values in order to get it to idle slow enough in the absence of friction. All IMHO of course :-) Cheers - Dave ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Engine models: start-up and commonality between FDMs
On 11/10/02 at 4:02 AM Julian Foad wrote: As for the guts of how the engines are modelled ... I first worked on the starting and stopping behaviour of the JSBsim engine. The thermodynamic model of the engine is probably very good Parts of it are, parts of it aren't and are overdue a re-visit. but there's lots of yucky stuff there to do with starting etc. I've done some stuff there, and in the sound configuration, but not finished. I'll go into that later. Ah yes, starting, I seem to recall a lot of hacking and kludging to get everything to work :-) There's a number of problems currently: 1/ My assumption of cranking speed at the time (480rpm!!!) was *way* too high. 2/ There's currently no friction modelled, which means there's not enough resistance to a proper starter motor torque at very low rpm when there's no prop resistance (I put a friction model into the LaRCsim IO360.[ch]xx to resist the prop when the engine was switched off in flight but I haven't brought it over to the JSBSim one yet). 3/ Need a proper starter motor torque curve. I did dig one out at one point but never put it in and now I've lost it. I'll have another look. Part of the problem is that I have to make sure that I'm working from publically available data and not anything internal. Have fun :-) Cheers - Dave ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
[Flightgear-devel] Engine models: start-up and commonality between FDMs
I was having a look at the piston engine start-up code. I absolutely love the way it chugs away for a second or two and then coughs into life - the sound effects really make it - but I wanted to make the speeds and stuff more realistic. Looking at the JSBSim engine code, it uses lots of arbitrary speeds and time delays and abrupt transitions from one mode to another. I have some improvements to this. Much of this code in JSBSim is very similar to FDM/IO360.cxx which seems to be only used by LARCsim even though it is not in the LARCsim subdirectory; presumably one was derived from the other. I really don't like duplication; I feel that any work I do on one of them is half wasted if it isn't automatically shared by the other. And then there is Yasim's engine code. Three piston engine models, each specific to its own FDM. So I started thinking about deriving them all from a common PistonEngine interface with the aim of making the engine models interchangeable. Haven't gone anywhere down this road yet, though. Thoughts? - Julian ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
re: [Flightgear-devel] Engine models: start-up and commonality between FDMs
Julian Foad writes: Much of this code in JSBSim is very similar to FDM/IO360.cxx which seems to be only used by LARCsim even though it is not in the LARCsim subdirectory; presumably one was derived from the other. I really don't like duplication; I feel that any work I do on one of them is half wasted if it isn't automatically shared by the other. And then there is Yasim's engine code. Three piston engine models, each specific to its own FDM. So I started thinking about deriving them all from a common PistonEngine interface with the aim of making the engine models interchangeable. Haven't gone anywhere down this road yet, though. Thoughts? I like the idea as well: it would be nice if the engine were its own subsystem and we could mix-and-match engines and FDMs (let's try the J3 cub with 180HP). Unfortunately, the FDM people haven't been too enthusiastic: in particular, JSBSim is supposed to run standalone as well as inside FlightGear, so they need some kind of internal engine model. 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
