Board index FlightGear Development Aircraft

New historically realistic Sopwith Camel JSBSim FDM released

Questions and discussion about creating aircraft. Flight dynamics, 3d models, cockpits, systems, animation, textures.

Re: New historically realistic Sopwith Camel JSBSim FDM rele

Postby Richard » Fri Mar 10, 2017 5:16 am

Simon runs with FDM frequency at 50hz instead of the default of 120hz.

This is causing problems when the aircraft is on the ground.

I'd guess that something in the spring/damping for the wheels is slightly wrong; as the ground handling part of JSBSim is particularly sensitive to FDM frame rate (frequency).

I once spent hours retuning the ground handling on the F-14 after it became weird and mushy; only to discover that the reason for this was a change in the FDM frequency for some much earlier testing that I'd never noticed.

To investigate the problem on the Camel run with --model-hz=50
Richard
 
Posts: 726
Joined: Sun Nov 02, 2014 10:17 pm
Version: Git
OS: Win10

Re: New historically realistic Sopwith Camel JSBSim FDM rele

Postby AndersG » Fri Mar 10, 2017 11:34 am

I would choose one FDM rate and stick with it - some tuning parameters are FDM rate dependent so one can't expect a FDM rate independent optima, and, in particular, things go pear shaped when decreasing the FDM rate from the one the craft was tuned at.
Callsign: SE-AG
Aircraft (uhm...): Submarine Scout, Zeppelin NT, ZF Navy free balloon, Nordstern, Hindenburg, Short Empire flying-boat, ZNP-K, North Sea class, MTB T21 class, U.S.S. Monitor, MFI-9B, Type UB I submarine, Gokstad ship, Renault FT.
AndersG
 
Posts: 2448
Joined: Wed Nov 29, 2006 9:20 am
Location: Göteborg, Sweden
Callsign: SE-AG
OS: Debian GNU Linux

Re: New historically realistic Sopwith Camel JSBSim FDM rele

Postby Bomber » Fri Mar 10, 2017 12:09 pm

Ok yes flightgears 'default' frequency is set at 120Hz and Richards correct that I use 50Hz.. I didn't change to this value, historically Flightgears default wasn't 120Hz. If I remember it correctly after an iteration there became a problem with JSBsim ground handling and to resolve this it was suggested to raise the frequency to 120Hz..

As to tuning parameters, specifically 'inflight' ones being frequency dependent, I think it wouldn't do any harm to identify which ones these are as it's the first time I've heard of this.
"If anyone ever tells you anything about an aeroplane which is so bloody complicated you can't understand it, take it from me - it's all balls" - R J Mitchel
Bomber
 
Posts: 1934
Joined: Fri Dec 14, 2007 7:06 pm
OS: Windows XP and 10

Re: New historically realistic Sopwith Camel JSBSim FDM rele

Postby AndersG » Fri Mar 10, 2017 3:23 pm

sim/model-hz has been 120, at least, since commit bf7349f0c19c3bad3ec9de916b76f29f793802f8 in January 2001....

For your other question all of them but to a varying degree - when you decrease the simulation rate (increase the time step length) you increase the error accumulation rate and, eventually, any feedback loop can blow up - but most likely ones to blow up are those that include linear or angular accelerations. For an aircraft the ground interaction is typically a much stiffer system than the aerodynamics so that is why you first see problems there.
Callsign: SE-AG
Aircraft (uhm...): Submarine Scout, Zeppelin NT, ZF Navy free balloon, Nordstern, Hindenburg, Short Empire flying-boat, ZNP-K, North Sea class, MTB T21 class, U.S.S. Monitor, MFI-9B, Type UB I submarine, Gokstad ship, Renault FT.
AndersG
 
Posts: 2448
Joined: Wed Nov 29, 2006 9:20 am
Location: Göteborg, Sweden
Callsign: SE-AG
OS: Debian GNU Linux

Re: New historically realistic Sopwith Camel JSBSim FDM rele

Postby Bomber » Fri Mar 10, 2017 8:18 pm

I just remember a lot of talk about the rate needing to be 120Hz and it was a long time after 2001...

Something also to consider... if a modellers flight model is so effected by the rate shouldn't the rate be set within the -set file for his flight model ?
Otherwise a user who unknowingly or simply has forgotten that he's set his rate differently could get less than satisfactory results, I don't think this is fair on the modeller.
"If anyone ever tells you anything about an aeroplane which is so bloody complicated you can't understand it, take it from me - it's all balls" - R J Mitchel
Bomber
 
Posts: 1934
Joined: Fri Dec 14, 2007 7:06 pm
OS: Windows XP and 10

Re: New historically realistic Sopwith Camel JSBSim FDM rele

Postby flug » Sat Mar 11, 2017 2:06 am

Bomber wrote in Wed Mar 08, 2017 7:30 pm:
Just did a drop test and it glides at 38kts... considering it's stall speed is 41kts I'd have expected a glide around the +20kts mark so 61kts.
It also has a lift to drag ratio of >14.... should be about 8.


Simon,

Would you mind outlining your procedure for a drop test? Or maybe there is somewhere else online or on the forums where it is already explained, and you could just refer me to it?
flug
 
Posts: 218
Joined: Wed Jun 17, 2009 1:06 am

Re: New historically realistic Sopwith Camel JSBSim FDM rele

Postby Bomber » Sat Mar 11, 2017 11:42 am

Flug as you know all planes have a natural trim, this being the speed they fly at where the moment from the wings are in balance with the moment from the H-stab. What we also know is that a plane will glide at a speed that produces a force that's equal to it's weight.

If a plane is spawned at 10kt with a velocity equal to its stall speed, with no engine turned on and the pilot not touching the controls what will happen is that the plane will attain its natural trim state. It will oscillate for a while but eventually settle down. Don't be confused that this will always be level flight (as we perceive it) from a pilots perspective it could well be that you're plummeting to the ground, but even so what you'll know is that there's a force being generated that's equal to the planes weight.

But we like to think of flying as gliding so we'll ignore the plumetting state and concentrate on the force equalling the weight of the plane being made up mainly of lift.

It's said that the act of observing an experiment influences the outcome of the experiment but this isn't true in the case, so touch nothing and just look at the outputs from the property tree.

As the plane weighs 1565lb, with a lift to drag ratio of 7.7 it's drag should be 203lb

So the glide speed should it be 38kts or it's reported stall speed +20kts ie 61kts.... who knows ?

But do we think that a pilot would want the planes natural trim to be 3 kts lower than it's stall speed, I don't think so.

Simon
"If anyone ever tells you anything about an aeroplane which is so bloody complicated you can't understand it, take it from me - it's all balls" - R J Mitchel
Bomber
 
Posts: 1934
Joined: Fri Dec 14, 2007 7:06 pm
OS: Windows XP and 10

Re: New historically realistic Sopwith Camel JSBSim FDM rele

Postby flug » Sat Mar 11, 2017 3:49 pm

Bomber wrote in Sat Mar 11, 2017 11:42 am:Flug as you know all planes have a natural trim, this being the speed they fly at where the moment from the wings are in balance with the moment from the H-stab. What we also know is that a plane will glide at a speed that produces a force that's equal to it's weight.

If a plane is spawned at 10kt with a velocity equal to its stall speed, with no engine turned on and the pilot not touching the controls what will happen is that the plane will attain its natural trim state. . . ..

Thanks - this is very helpful.

As the plane weighs 1565lb, with a lift to drag ratio of 7.7 it's drag should be 203lb

So the glide speed should it be 38kts or it's reported stall speed +20kts ie 61kts.... who knows ?

But do we think that a pilot would want the planes natural trim to be 3 kts lower than it's stall speed, I don't think so.

Thanks for this also--it gives a number of things to think about.

Regarding Trim: The Camel was notorious for pitching nose-up thanks to the weight of the fuel tank behind the pilot. Pilots had to maintain about 20 lbs continuous forward pressure on the stick just to maintain level flight. There are many reports of pilots stalling e.g. on take-off if they got distracted (diddling with the mixture, typically, which required adjustment soon after takeoff) and allowed the plane's natural nose-up attitude to take them into a stall. So actually, we know that for the Camel, the 'natural' hands-off attitude is actually pretty close to or even below stall.

Regarding stall speed, here is the comment from the Camel Notes (SopwithCamel-Realistic-JSBSim-Flight-Model-Notes.txt): Wikipedia and several other sources report 48 mph/41.7 knots. However, this does not match up with the Camel's reputed quick turn rate. Lower stall speed directly equates with quicker turn rate (both related to AoA where stall occurs). A replica Camel reports a stall speed under 40 MPH (35 knots) and Frank Tallman reports stall speed of 35-40 mph (30.4-34.8 knots). A stall speed in this range seems to fit better with the Camel's reported turn rate.

All the early reports on the Camel report speed in MPH. I think perhaps at some point 41 MPH was transformed into 41 KTS in some second-hand source, and then other sources simply copied this. Several first-hand reports put the put the stall speed in the 35-40 MPH range whereas I have yet to find a first-hand report or an actual original source document for the 41 KTS stall speed.

41 KTS=47 MPH. 47 MPH is quite dramatically different from the multiple first-hand reported stall speeds of 35-40 MPH.

The too-high lift/drag ratio worries me more--I'll look into that.

Thanks again.
flug
 
Posts: 218
Joined: Wed Jun 17, 2009 1:06 am

Re: New historically realistic Sopwith Camel JSBSim FDM rele

Postby flug » Thu Mar 16, 2017 4:33 pm

As the plane weighs 1565lb, with a lift to drag ratio of 7.7 it's


Thanks again to everyone for the thoughts and comments above. The stall speed, zero-lift drag coefficient, etc. usually quoted for the Camel seem to come from this very interesting source: Quest for Performance: The Evolution of Modern Aircraft by Laurence K. Loftin, Jr.

I don't have any particular reason to question the values given there, but the sources given there are three reference books from the 1960s--so written 50 years after the Camel was produced and (presumably) tested. So this is a at least a tertiary source and I haven't been able to trace it all the way back to the primary source. The value reported there is 41 kts (48 mph) for stall.

Meanwhile, I have at least 3 primary sources who report that the Camel stalls and/or takes off at 35-40 mph. (I realize the take off & stall speed are different, but it seems unlikely that an aircraft would lift off at 35 mph while stalling at 48 mph.)

I experimented by using the complete set of data given in Loftin (particularly zero-lift drag coefficient and wing surface area), and interestingly enough that really seemed to pull a lot of things together. Suddenly the stall speed was 48 mph, the multiplier I'd been using on the lift table wasn't needed, etc. I'm guessing all those values were worked out by some engineer using aircraft in some standard configuration and so in a sense they all 'go together'.

That was interesting, but why is the stall speed recorded there so wildly different from that reported by a number of pilots?

I re-read a few of these first-hand accounts, and realized that later pilots are often flying something of a stripped-down aircraft. They usually don't have machine guns, certainly not ammo, may be flying with less than a full tank of fuel for various reasons, may be flying with a different engine--perhaps one that weighs quite a bit less.

And of course it turns out if you knock a few hundred pounds off the loaded weight of the aircraft, stall speed and a bunch of other handling characteristics are quite different! Just for example, here is a modern Camel replica, empty weight 943 lbs, useful load 300 lbs, stall speed 40 mph. So fully loaded, about 1250 lbs - whereas in the JSBSim Camel I'm modelling a fully loaded & combat-ready aircraft that masses out at 1567 lbs. Obviously, that extra 300+ lbs is going to make quite a difference, including for details such as stall speed. Losing 300 lbs weight is (more or less) equivalent to gaining 300 lbs lift . . .

Additionally, it explains (in my mind) some other discrepancies I've been trying to work out, like the different max turn rates reported, and the discrepancy between stall speed and reported max turn rate. Both of those are related to maximum angle of attack, so they should be closely related. But . . . the difference in mass of a fully loaded plane vs (say) one low on fuel and ammo is significant--and would allow quite a difference in stall speed and turn rate in those different configurations. Interestingly the difference we would expect to see from an aircraft in those different configurations, and the difference we do see in the JSBSim version of the aircraft in the different configurations, tracks pretty well with reported differences in performance.

Most reports on performance, even if first hand, don't necessarily give exact details about every aspect of the aircraft tested. So it seems as though you are getting precise first-hand data, but in reality it has quite a bit of wiggle room in it because of exact details of aircraft configuration, which are often un-reported. (One reason I would like to track down the original source of the basic characteristics of the Camel rather than relying on various tertiary sources--they original could/should have more of the specifics of the conditions tested.)

To make a long story short, I've spent the last few days rebalancing everything to these slightly different specs, which give a stall speed of 48 mph when fully laden and a lift/drag ratio close to the reported 7.7 (again, when fully loaded). I think this will be the best & most realistic version yet.

Another thing that makes me feel optimistic that we're settling in on something pretty close to the 'real thing' is that despite all the tweaks I've made since the 2.0gamma version of the Camel, they really are just tweaks. Flying it still feels like flying the same basic aircraft--just about the differences you might expect if you added or subtracted a few hundred pounds weight. Obviously it affects handling and specifics like stall speed noticeably, but it is still quite obviously the same basic aircraft.

I've tried tweaking some of the reported characteristics, where the behavior has been well documented but the exact values given in the FDM are by necessity something of a guess (ie, adverse yaw due to ailerons, exact gyroscopic moment due to rotary engine/prop). The interesting thing is you can tweak these things as much as +100% or -50% and, obviously, details of handling will different but what happens is that in a few minutes of flying you learn to adjust to the new situation and you just proceed. It is as though your particular aircraft happened to have slightly larger or smaller elevators, or you put a bigger/heavier prop on or whatever. Those things all do in fact affect aircraft performance and handling to some degree, but in real life, as long as the differences are within a certain range, you just adapt and proceed.

In fact we know that different variants of the Camel were in fact produced over time and undoubtedly those variants had differences in handling and performance to a certain degree. Yet it is still the same underlying aircraft.

Tweaking it to hit the reported max speeds at 6500, 10000, and 15000 ft, as well as the climb time to 6500, 10000, and 15000 feet (all within a reasonable margin of error) took some time and work, but the end result really was a fairly minor tweak to what it already does 'naturally' so to speak. And obviously those precise values all have an error bar around them in the first place and could be affected by something as small as carrying and extra 50 pounds of weight or outfitting a slightly different windscreen or gun mount.

All the above is a long-winded way of saying--new & improved FDM release coming soon! Different but not so different from the previous 2.0gamma version.
Last edited by flug on Thu Mar 16, 2017 5:01 pm, edited 3 times in total.
flug
 
Posts: 218
Joined: Wed Jun 17, 2009 1:06 am

Re: New historically realistic Sopwith Camel JSBSim FDM rele

Postby flug » Thu Mar 16, 2017 4:45 pm

AndersG wrote in Fri Mar 10, 2017 11:34 am:I would choose one FDM rate and stick with it.


Is there some way to set the FDM rate in the -set.xml file or otherwise, from the designer's end?

Or is it only something the user can set via a command line option?

FWIW I seem to recall that one of the issues early on with this model was inconsistent ground reactions, and that part of the solution was moving to 120 hz FDM rate and also making some other adjustments.

One of the issues with a model like this is that the original aircraft didn't really have any suspension per se. It's just a hard rubber tire mounted to a wheel, the wheel to an axle, the axle to some bracing, and the bracing directly to the aircraft fuselage.

So it's not like any more modern landing gear where there is likely to be some kind of damped spring arrangement that has X inches (or even feet) of compliance built into it.

Rather, with an aircraft like the Camel they were depending on the 'natural' inherent spring of the entire system--tire, wheel, axle, bracing, fuselage, etc--to absorb any forces. This is hard to get your head around modelling but more to the point of JSBSim, the entire system seems to act pretty much like a very short, stiff spring with quite a bit of damping and that is the very type of thing that JSBSim with its relatively slow modeling rate is going to have the most difficult time with.

120 hz is faster than it used to be I guess, but I would also guess that to model this type of ground reaction accurately you might want more like 1200 hz or even 2400 hz.

I doubt we're going to get those types of simulation speeds, even just for the ground reaction component, any time soon (though it is interesting to contemplate!) so in the meanwhile we are more in a situation of figuring out reasonable compromise solutions that will be reasonably realistic.

That's one reason, I'm guessing, that we can't always just come up with a solution that works the same for any arbitrary FDM simulation rate.
flug
 
Posts: 218
Joined: Wed Jun 17, 2009 1:06 am

Re: New historically realistic Sopwith Camel JSBSim FDM rele

Postby AndersG » Thu Mar 16, 2017 7:26 pm

flug wrote in Thu Mar 16, 2017 4:45 pm:
AndersG wrote in Fri Mar 10, 2017 11:34 am:I would choose one FDM rate and stick with it.


Is there some way to set the FDM rate in the -set.xml file or otherwise, from the designer's end?

Or is it only something the user can set via a command line option?


It is easy to set from the -set file, just set /sim/model-hz to your value of choice (but expect odd behaviour if you change it at run time):
Code: Select all
<PropertyList>

 <sim>

  <description>Swedish Navy 20m T21 class motor torpedo boat.</description>
  <author>Anders Gidenstam</author>
  <aircraft-version>2016-11-24</aircraft-version>
  ...
  <flight-model>jsb</flight-model>
  <aero>MTB_20m</aero>
  <!-- The higher rate seems to aide stability. -->
  <model-hz type="double">960.0</model-hz>


I suppose a user can override it by the specific command line argument (IIRC one exists) or by just setting the property with --prop but IMO that would be ill advised to do without having a very good understanding of the FDM in question.
Callsign: SE-AG
Aircraft (uhm...): Submarine Scout, Zeppelin NT, ZF Navy free balloon, Nordstern, Hindenburg, Short Empire flying-boat, ZNP-K, North Sea class, MTB T21 class, U.S.S. Monitor, MFI-9B, Type UB I submarine, Gokstad ship, Renault FT.
AndersG
 
Posts: 2448
Joined: Wed Nov 29, 2006 9:20 am
Location: Göteborg, Sweden
Callsign: SE-AG
OS: Debian GNU Linux

Re: New historically realistic Sopwith Camel JSBSim FDM rele

Postby Bomber » Thu Mar 16, 2017 8:44 pm

Flug,

I've just tested your engine and it's a bit weak....

Sea level 15degs C.
1288rpm
56.18hp
Fuel flow 15.757 gph

what you should expect to see is

1225rpm
130hp
Fuel flow 12.03 gph

Simon

p.s. If you want a correct engine and prop combination for this plane just pm me your email addy.
"If anyone ever tells you anything about an aeroplane which is so bloody complicated you can't understand it, take it from me - it's all balls" - R J Mitchel
Bomber
 
Posts: 1934
Joined: Fri Dec 14, 2007 7:06 pm
OS: Windows XP and 10

Re: New historically realistic Sopwith Camel JSBSim FDM rele

Postby Johan G » Fri Mar 17, 2017 8:45 am

Thank you flug for that interesting writeup about sources, weights and flight dynamics. :)
Low-level flying — It's all fun and games till someone looses an engine. (Paraphrased from a YouTube video)
Improving the Dassault Mirage F1 (Wiki, Forum, GitLab. Work in slow progress)
Johan G
Moderator
 
Posts: 5545
Joined: Fri Aug 06, 2010 5:33 pm
Location: Sweden
Callsign: SE-JG
IRC name: Johan_G
Version: 3.0.0
OS: Windows 7, 32 bit

Re: New historically realistic Sopwith Camel JSBSim FDM rele

Postby Bomber » Fri Mar 17, 2017 11:15 am

I'm not knocking Flugs work on the Camel, just trying to help by pointing out some issues.

I hope we all appreciate that you can't have a correct flight model if the engines not producing the correct power... the weights/CoG, engine/prop and aerodynamics all have to be correct (in that order). If the first two aren't correct you're going to have to compensate within the aerodynamic section and that takes more effort and time than it does to simply get the first two correct in the first place.

Regards

Simon
"If anyone ever tells you anything about an aeroplane which is so bloody complicated you can't understand it, take it from me - it's all balls" - R J Mitchel
Bomber
 
Posts: 1934
Joined: Fri Dec 14, 2007 7:06 pm
OS: Windows XP and 10

Re: New historically realistic Sopwith Camel JSBSim FDM rele

Postby Thorsten » Fri Mar 17, 2017 12:51 pm

I find it implausible that a factor 2 discrepancy in engine power would not have shown up in any other tests so far. So I would first double-check the procedure used to test.
Thorsten
 
Posts: 11190
Joined: Mon Nov 02, 2009 8:33 am

PreviousNext

Return to Aircraft

Who is online

Users browsing this forum: No registered users and 2 guests