YASim or JSBSim for a new aircraft

[KL-666]

Interesting discussion starting at the other forum. I also think, jsbsim, nice if you have all the wind tunnel data, but if not, you have an empty vessel. In such case you are better off with the generation of yasim.

Really, how often do you get at real wind tunnel data? Maybe for one or two planes? In other cases you either do not get the data at all, or if you do, it is restricted by license. So the dream of jsbsim is really a pipe dream.

YASim or JSBSim for a new aircraft.

Postby CaptB » Sun May 29, 2016 11:19 pm

Hello,

I'm looking for a word of advise/clarification from you guys as I have quite a bit of trouble to choose the future FDM for my aircraft project. Looking at the YASim wiki page there is the following sentance which made me lean twards YASim:

If you have solid flight data for your aircraft such as wind-tunnel data or you are looking to eventually generate a hyper-realistic simulation, JSBSim is probably a better approach.

However, I do see quite a lot of aircraft using JSBSim, and I have some personal doubts if wind tunnel data was available for use, so this seems to contradict the information metioned in the quote.

How feasable is using JSBSim for a jetliner without having the data mentioned?

Thank you.

http://forum.flightgear.org/viewtopic.php?f=49&t=29682#p286897

Re: YASim or JSBSim for a new aircraft.

Postby Bomber » Sun May 29, 2016 11:45 pm

How accurate do you want it

Kind regards, Vincent

[jwocky]

Well, wind tunnel data aside ...

- if you have a lot of weight points, maybe even dynamic ones ... Yasim can't go there
- if you want ground effect ... yasim can't go there
- if you have a plane with a relative big relation of wing surface to control surface area, Yasim can go there but does strange things
- if you have actual engine data, for example from the manufacturer, you can in Yasim go only based on approximations there (the thrust ratios are quite an iffy thing there)
- if you have an older engine, lets say some WWII propeller plane, you can't really do a detailed engine modelling in Yasim, regardless how much real data you have.

otherwise, a CoG is a CoG is a Cog ... so that's something, you have to deal with in both. Weights are the same (except for some rather invented balancing weight, you use in Yasim to get your balance back). Bottom line is, if you have limited data, you can have the same sucking result in both. If you got more data, JSB gives you the edge. If you expect at some point in the future to get more data, JSB gives you the edge.
The problem with JSB is however, that you can do much more with it. In many aspects, it's rather a physics machine. So, you have an almost infinite variety of options what to make it do and this confuses people easily. Another problem is, if you don't change a thing, JSB planes usually suffer from a realistic slide slip effect in turns, Yasims don't so much, which makes them feel more comfy to fly.

So Yasim is rather the pipe dream of those who get themselves toy planes. Nothing wrong with toy planes and in many cases, it's fine enough. However, with the currently on going attempts for example from Bomber, to get more technical quality in the plane pool, you will see planes more and more going to JSB. From a developer's point of view, it is also clearer structured and, once one has the learn curve mastered, easier to handle in big projects. Of course, if you have extraordinary data, like wind tunnel data, engine burn data and such fine data morsels, then, there is anyway no other way to go.

[KL-666]

Hello Jwocky,

Well, the positions are clear now. I expected you to say this. I could have written your response for you

There are ample examples of that jsbsim in the wrong hands leads to disaster. I have only to remember the disastrous fdm of the 707. No, i am not going to test it for the umpths time again, only to find out it is still a disaster.

Now i see really worrying posts about my beloved 777 on that other forum.

Re: Announcing NEW Boeing 777 'Seattle' - Pre-Release

Postby Hyde » Sun May 29, 2016 7:29 pm

swwon wrote in Wed May 25, 2016 7:00 am:
200LR:
VNAV doesn't stay at altitude, it picks a random alt and deceds to it (vnav speed, path nor vnav alt) must stay in flch and let it switch to alt once altitude is reached. cannot operate A/P based on vnav for decent or fixed climb at set waypoint altitudes. Have to hit flch in order to decend at "Top of Decent" and at proper decent speed autoadjustment; instead of vnav path working properly, vnav path will decend and overspeed the plane. Used to work right, I wonder what happened?
When will the jsbsim version be avail?

Please try the latest program and report if still not good.
And jsbsim is a still primitive state but you can test it. You can not run with current version though. You need to replace entire 777 directory.

https://forum.flightgear.org/viewtopic.php?f=4&t=21870&p=287154#p286882

Reading "And jsbsim is a still primitive state" makes me really nervous about that it will stay in that state. How much i respect Hydes work in general, on the flying front i had to save him several times.

- Forced 250 kts below fl 100 on autopilot. Insane, it is an agreement, not built in planes
- Glue the aircraft to the ground because some people do not understand x-wind. Insane, at high speeds you should apply a lot of rudder in x-wind.
- And now manual trim is discarded of, because we have auto trim. Insane, pilots use manual trim all the time when flying manual.

So overall, i think it is a very bad idea for Hyde to venture in the complex world of jsbsim, when there is already a more than decently tweaked yasim model.

Kind regards, vincent

[jwocky]

Soooo ... you don't know what will come, you don't know why it will come, but you are already afraid of it? See, the developers of planes in FG have bormally reasons to take what they take. Some reasons are handling reasons. If one like a certain Gentleman in France dishes out plane models with two Mickey Mouse tanks and the flight attitude of a wet sack in mass production, Yasim is certainly the right choice for him. If someone like Bomber goes and calculates down a propeller blade in segments, he simply can't use Yasim because Yasim can't even to begin to go for that level of simulation.

Another subject that comes always up again are the extras. In JSB, you can integrate a fuel system. Either to the level of primitivity of Yasim or to do anything a fuel system with pumps and connectors can do.
The same could be of course done for an electrical system. The only thing preventing this in a bigger scale is actually, that some people will cry then "our poor Yasim was left behind" and that is the reason why a lot of that stuff is still in Nasal .. with the inherent costs in system performance.
Try to hang a pushback on a Yasim ... big mess. In JSB, it's down to ten lines and a bit copying to get the model of the pushback cart in the right position.
See, almost everything, I can do in JSB, I can do with Yasim. I only need some ten thousand lines of Nasal to fill the gaps and the users with weaker computers will feel that.

So, honestly, I doubt, you could have written my answer for me. Because you know my opinion but you never bothered to delve into the reasons for this opinion.

[bomber]

The same could be of course done for an electrical system. The only thing preventing this in a bigger scale is actually, that some people will cry then "our poor Yasim was left behind" and that is the reason why a lot of that stuff is still in Nasal .. with the inherent costs in system performance.

Yep that's exactly how I feel about...

I could go on about this for pages and pages....

Nasal is a cancer infecting FG..

[jwocky]

Nasal is the language for masochistic nosebears?

[legoboyvdlp]

in the wrong hands

''wrong hands'' not being our resident JSBSim expert, Jwocky

By the way, I assume that 'primitive state' refers to the JSBSIM 777 -- not JSBSIM itself.

[jwocky]

I get slowly the feeling, Bomber is outracing me in JSB Just like It0uchPods outraces me in systems ... both feels, oddly enough, quite good on my end.

[bomber]

I've spent the last 6 years on jsbsim.... the last 4 on one single gliders fdm, redoing it over and over again, about 18 rewrites of a 7000+ line flight model.

I'm always open to looking at doing a jsbsim flight model from a different direction.

Heh baby it's my bag.

[bomber]

I know it's not flightgear....so I've never posted any video but it is a JSBsim flight model.

https://youtu.be/_sK_04Vl_xo