[1.3.1] Ferram Aerospace Research: v0.15.9.1 "Liepmann" 4/2/18

[smjjames]

It sounds like they MAY have fixed the issue with clipping and aerodynamics in the new stock aero.

[Van Disaster]

@Ferram: So FAR thinks the front node of the second piece is in front of the nosecone, effectively? ( or merges colliders or something and ends up with an open front node, I guess ). I wonder if just tweaking the nosecone attach node Z would help... guess I'll see.

Cantab, I'll second the procedural wing suggestion, especially for deltas, you can need some huge control surfaces.

[blowfish]

hey ferram, I'm curious if parts are affected by oblique shocks created by other parts. e.g. if the nose creates an oblique shock which the rest of the craft stays inside, then do the rest of the parts experience a lower mach number but higher pressure? It looks like in FARBasicDragModel, the mach number is assumed to be the same for every part, though I haven't checked wings.

- - - Updated - - -

My biggest problem throughout has been getting enough pitch control authority. The built-in trim on the wing trailing edge helped but it still struggles.

ferram's right that procedurals would make things a lot easier, but you could probably get a fair bit more control authority just by adding canards. this is made with all stock parts, and controls reasonably well (not exactly fighter maneuverability but manageable) - and it's quite a bit bigger than yours. Generous thrust vectoring on the rapiers helps a lot, but for a smaller craft canards should do the trick.

[ferram4]

@VanDisaster: No, it thinks that the front node of the second piece is clipped inside the nosecone, which is (for most other parts) a sign that the nosecone is not actually on the front of it. It is a very hacky attempt at getting around the deadly reentry heat shields being clipped into the built-in shield on the Mk1 pod by unobservant players causing issues combined with issues caused by Squad's unwillingness to address the failings of the part clipping prevention. A lot of it is simply due to the way that less and less can be assumed about how vehicles are put together.

[BlueTiger12]

Hello Ferram

I really love this mod.

But since a few installs (even clean installs; i think since version 24.2) i have a problem in the sph.

Whenever i right click on a control surface or a wing, the game is getting stuck for ~2 seconds. After this it is really really slow (about 3 fps)

Even tweaking in the right click menu is going that slow.

As soon as you click anywhere and the menu closes this problem is gone

Problem stopped with uninstalling far.

Do you know about this allready?

Otherwise i would take a deep look if it is a problem with some mods i have installed

[cantab]

The stock control surfaces are simply not designed for vehicles that large; you'll likely need to move to procedural wings to get the control surface area you need for controlling that thing.

Well as I see it I'm uneasy with adding new parts specifically to get a design I've already started to work, it smacks of moving the goalposts. If I'd been using PWings from the start it would be another matter. And it's not like I'm out of things to try with the parts I already have. Letting the elevons overlap is in any case motivated by looks not performance.

Speaking of things to try, I've seen people put elevons on the leading edge. What kind of things are they set up as, is it good for control or just as leading-edge flaps/slats?

[tetryds]

Well as I see it I'm uneasy with adding new parts specifically to get a design I've already started to work, it smacks of moving the goalposts. If I'd been using PWings from the start it would be another matter. And it's not like I'm out of things to try with the parts I already have. Letting the elevons overlap is in any case motivated by looks not performance.

Speaking of things to try, I've seen people put elevons on the leading edge. What kind of things are they set up as, is it good for control or just as leading-edge flaps/slats?

They are only good to control if you really know what you are doing, and give them a very small deflection.

I would recommend going for flaps with them at first, the gain is not too high, but it's good for big crafts.

You can also set them to have AoA% parameters to add artificial stability to your craft, but again, if you know what you are doing

Experimenting is a good way to learn.

[Van Disaster]

Well as I see it I'm uneasy with adding new parts specifically to get a design I've already started to work, it smacks of moving the goalposts. If I'd been using PWings from the start it would be another matter. And it's not like I'm out of things to try with the parts I already have. Letting the elevons overlap is in any case motivated by looks not performance.

Speaking of things to try, I've seen people put elevons on the leading edge. What kind of things are they set up as, is it good for control or just as leading-edge flaps/slats?

As long as the existing controls move together it should be ok, I think - if they seperate at some point you might end up with biplane effects. Leading edge controls work just like simple flaps as far as I can tell, basically just a section of wing at a different AoA. I think whether that section is stalled will affect the section directly behind, so it does sound like a way of increasing AoA without stalling, as leading edge devices should.

@blowfish: from using aero vis at various mach numbers with one of my craft with too-long span, I'm pretty sure mach angle doesn't come into any drag calcs. Mach angle at mach 5 is something like 11 degrees, my outer engines should be like parachutes...

[FourGreenFields]

Do stalls also render control surfaces useless?

In RL, a stall will also make your ailerons pretty much worthless, which is one reason why spins are dangerous. Is that modelled?

[tetryds]

Do stalls also render control surfaces useless?

In RL, a stall will also make your ailerons pretty much worthless, which is one reason why spins are dangerous. Is that modelled?

Yes, it does.

@Van Disaster: Yes it increases stall angles, not super efficient but helps.

On smaller planes you can pull some very cool tricks with them, but it's a lot harder to tweak.

-100% on AoA% is a rule of thumb if you will, given its in front of the COM.

[Van Disaster]

Do stalls also render control surfaces useless?

In RL, a stall will also make your ailerons pretty much worthless, which is one reason why spins are dangerous. Is that modelled?

You can stall control surfaces seperately from whatever you attached them to, which can get interesting when you're trying to stop something back-flipping... but also helpful if you want decent airbrakes.

@tetrydis: presumably that works the other way then, so if I wanted to stall a wing I could just stall the leading edge control?

[tetryds]

@tetrydis: presumably that works the other way then, so if I wanted to stall a wing I could just stall the leading edge control?

I have never tried doing that, but it should work.

Why would you want to stall your wings on purpose, by the way?

[Van Disaster]

VTOL transition to hover, in this case; I have endless trouble trying to stop flying in a controlled fashion, usually because one wing stalls first. You could use it for lift dumping if you're trying to land in a hurry, but it might be more sensible to just go round again

[tetryds]

VTOL transition to hover, in this case; I have endless trouble trying to stop flying in a controlled fashion, usually because one wing stalls first. You could use it for lift dumping if you're trying to land in a hurry, but it might be more sensible to just go round again

Then, disable all controls and set them to flaps with a negative flap deflection.

Have fun trying to maintain control.

A good tip for breaking is also having powerful rudders and turning sideways, using roll to point your engines retrograde, it requires a lot of practice but works.

If you have lots of control authority turning backwards and using the main engines to stop also work.

Also, are you using procedural wings? Stall is a bit weird on them lately.

[Van Disaster]

Control should be ok if the engines have transitioned to something vertical and the rcs is active, as long as the wings hand over lift together.

I use B9 pWings at the moment, there's something a bit odd going on there with MAC changes too, and I've even been through the source of that dll to try and trace it and can't see anything...

[EzinX]

So, Ferram. Or anybody, really. You've got a big chunk of source code that does all the work. However, what isn't there is any documentation as to what any of it is for. Skimming the code, I saw a lot of methods that looked quite familiar from books on various 3d math, but what I don't know is, what method or technique is tying it all together.

What's the big idea here? Does a book or papers exist that describe the modeling technique you're using?

Thing is, you can get Autodesk or other tools to tell you quite accurate numbers for parts in isolation. But how do you modify these calculated drags to take into account adjacent parts? That, to me, seems to be the fundamental problem that your code is solving over many thousands of lines - but how does the basic algorithm even work? Surely you're not approximating navier stokes in realtime...

Also, instabilities. How's it done? Moreover, you've got a problem that the physics model of this game does not update as quickly as the control fins on, say, a fly by wire aircraft (200 hz or so). This means that a user may get a large torque applied to their craft due to instabilities that a real aircraft could correct for - but the game's physics model doesn't even inform them until it's already integrated that torque over a timestep.

Edited March 12, 2015 by EzinX

[futrtrubl]

Also, instabilities. How's it done? Moreover, you've got a problem that the physics model of this game does not update as quickly as the control fins on, say, a fly by wire aircraft (200 hz or so). This means that a user may get a large torque applied to their craft due to instabilities that a real aircraft could correct for - but the game's physics model doesn't even inform them until it's already integrated that torque over a timestep.

Can't speak to the rest but yes, this is a problem in stock KSP and unfixable by mods. It can be seen with any form of autopilot even SAS as over-corrections and oscillations, especially during physics warp.

[silentdeth]

I'm working on a FAR tutorial for youtube and was wondering if anyone had any suggestion for what it should cover. Anything that they have, or had found it difficult to find the answers to. I already intend to cover the basics, and all the screens, ect. but I'm looking for things like Mach tuck, things that are not really covered in other tutorials and require a lot of searching to found out about. In essence I want to cover everything someone needs to know so that they don't have to go searching around for a bunch of different tutorials.

[cantab]

Use of the new superSAS tools in rocket ascents. Ie should you click "Hold Prograde" for a gravity turn or not. That's something old tutorials aren't going to discuss.

How FAR behaves with clipping, and the notorious DRE heatshield problems. Again, something old tutorials won't include - but equally something liable to be obsoleted.

As far as aircraft go, consider discussing different configurations and the pros and cons. For example "conventional" wings and tail, tailless deltas, tailed deltas, flying wings, even bi- and tri-planes.

How to flight test, how to relate issues in flight to what the analysis tools report, and what changes to make to a plane to correct what issues.

How to use the Flight Assistance, and how to tune it to work well. How to use trim too.

Techniques to shed speed - it's often a real nuisance in FAR I find.

Really most of those topics would be a video in themselves. Aeroplane design is complicated.

[Svm420]

Something you may be covering but that I'll mention. Talk about how to correct the most common derivates or w/e that end up red. Yaw, sideslip, and the few in the upper half when you are at high velocity and altitude. You run into when building space planes. Discuss how to improve flight characteristics on a more advanced level fine thing placement and flap/spoiler use especially if you cover more exotic techniques. Just some ideas I'd like to see covered in 1 place.

[Van Disaster]

Yes, it does.

@Van Disaster: Yes it increases stall angles, not super efficient but helps.

On smaller planes you can pull some very cool tricks with them, but it's a lot harder to tweak.

-100% on AoA% is a rule of thumb if you will, given its in front of the COM.

Well, certainly looks promising. Still stalls at 50m/s but some quite absurd AoA is possible.

Edited March 13, 2015 by Van Disaster

[cantab]

How can I work out the speed for best rate of climb for a plane? Since spending ages getting to altitude sacks. Wikipedia mentions it's the speed with maximum excess power; FAR shows a "Specific Excess Power" data point so do I just want that high?

[NoClass]

How can I work out the speed for best rate of climb for a plane? Since spending ages getting to altitude sacks. Wikipedia mentions it's the speed with maximum excess power; FAR shows a "Specific Excess Power" data point so do I just want that high?

My understanding is that it's also generally close to where you have maximum L/D, which is related to both a particular speed and AOA. (I use an AOA, because it shouldn't vary with altitude)

...but I don't know how much your actual thrust angle affects best rate of climb if you have high TWR.

What I'd love to predict is best range speeds/altitudes for craft while constructing.

[Van Disaster]

I would suspect there's an ideal EAS, flying a constant Spc Excess Power looks like it'd eventually have you pointing straight up given the way it climbs as you do.

[tetryds]

I would suspect there's an ideal EAS, flying a constant Spc Excess Power looks like it'd eventually have you pointing straight up given the way it climbs as you do.

Given the way jet engines work in KSP that is about what would happen, yes, and that would be exactly the expected behavior.

Not if you are using AJE, though.

But remember that jet powered aircraft have some pretty high climb rates.

The best climb rate is the highest speed you can be while not losing speed (or losing little) while aiming as high as you can.

There are a few methods to figure it out, but wouldn't it be more fun to experiment with your plane and create your own methods rather than following a fixed methodology and moving on?

Well, I think it is, but prove me wrong and I may give more tips

Edited March 14, 2015 by tetryds