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

[Camacha]

I need the payload, most crucially, to re-enter; I'm sending it to Eve and I want it all down in the same place at the same time. This is what it looks like:

Hence my need for a big fairing.

To be honest, I think shuffling your configuration around a bit might help you a lot. No need for the rovers to radiate like that. Do not shape the problem to the solution but the other way around

Edited May 31, 2013 by Camacha

[Doomydoom]

For a change of pace, I decided to make something that doesn't break FAR completely. This proof of concept blended wing-body plane is only capable of 20 G turns. Wait, woops. http://i.imgur.com/rx08nDk.jpg

[foamyesque]

Use Payload factory to create a custom fairing that fits your payload. Or still assemble in space (sending up main hitchhiker and rovers up in separate launches). Considering your "small" rovers, I doubt the docking nodes will not survive a reentry. If that's the case, use quantum or Docking Struts to stabilize your connected payloads.

Using this approach I dropped entire space-assembled colony bases on Eve, being in the 100t+ range...

It's not really a question of getting the Hitchhiker to survive; it's a question of preventing it from doing this to the return vehicle:

Fairing Factory, after some fiddling, works... but even maxed out is still too narrow.

EDIT: Wait there's a zoom out button

To be honest, I think shuffling your configuration around a bit might help you a lot. No need for the rovers to radiate like that. Do not shape the problem to the solution but the other way around

What would you suggest?

Edited June 1, 2013 by foamyesque

[ferram4]

@Doomydoom: That is more the effect of having very large lifting surfaces with very little in the way of mass (at least from how it appears to me). Add to that the fact that the part connections are over-engineered because they have to be set to control deflections rather than the ultimate forces themselves and I can see that level of over-engineering happening. I think most modern fighters can come close to those g-forces but are limited because the pilots can't maintain proper situational awareness and control if the plane is executing a 9g+ maneuver.

As for trying to detect if there's a surface that the shock can attach to... It's difficult to find the absolute front-most point of a vehicle (it's close to the origin of the front-most part, but a little further due to the model); then I'd have to attempt raycasting the cone created by that point to find any places where the shock would impact the vehicle (this would be very expensive computationally, btw). Then I'd have to try and determine whether a part being impacted by the ray meant that it wasn't riding a shock (e.g. a vertical stabilizer that reaches outside the Mach cone, but doesn't ride the shock) or if a part that wasn't impacted meant that is wasn't riding a shock (e.g. a vertical stabilizer that remains completely inside the cone, but doesn't ride the shock). Attempting to use the normal vector of the impact point only does so much good and doesn't tell me anything about the vehicle geometry in between the ray origin and the impact point, so even using that to help differentiate is difficult.

The main problem is less about the physics themselves and more about trying to detect the actual aerodynamic configuration of the vehicle. The current part-by-part solution doesn't need me to figure out the configuration of the whole vehicle; I only need to worry about the properties of one wing part, or one nosecone, or one fuel tank when I'm attempting to code the aerodynamics. Yes, it makes the assumption that the flowfield is linear (i.e. that I can combine the individual effects of each part and the linear combination will be accurate), which is very, very wrong, but it is the simplest and most robust solution to the problem of indeterminate vehicle geometry.

I'd honestly love to do a more advanced aerodynamic simulation that handled the aircraft as a whole rather than part-by-part like I'm doing now. The first problem is that I would need a robust algorithm to determine all the pertinent details of a vehicle's geometry (which could range from identifying wings, fuselages, etc. to try and handle individually or trying to create a mesh that could be used to do proper CFD using the Reynolds-averaged Navier-Stokes equations with some kind of simple turbulence model). The second problem is trying to implement the simulation in an efficient and robust manner, by which I mean it doesn't lag your computer to hell and it doesn't vomit when you try to fly an Abrams tank because it doesn't know what to make of that. The third problem is trying to apply the forces and moments calculated for the vehicle as a whole to the proper parts so that you can get structural failures from aerodynamic loading without causing unrealistic behavior.

So there's a mini-rant on the subject; believe me, I've spent a lot of time thinking about this topic and a reliable solution to the geometry configuration problem still eludes me, especially because it has to be able to update in real-time to handle vehicle deformations.

@foamyesque: I'd actually create a stack of them on top of the Hitchhiker, give each one a pair of radial chutes (this is Eve, correct?) and release them in sequence before pulling the chutes for the Hitchhiker itself. I'd also switch out two of those batteries for a pair of toroidal fuel tanks and then add a pair of radial engines just in case you need them; I've found that they can be used for asymmetric thrust can be very good for saving rovers that have flipped due to reckless driving as well as safely landing at your destination.

And if you're intending to return them home then the fuel tanks + engines will allow you to re-stack everything to pack it up and go home.

[Doomydoom]

Would it be possible to calculate things beforehand and write it all to a file for that craft to reference in realtime? Ignoring that KSP craft tend to tear apart.

[foamyesque]

@foamyesque: I'd actually create a stack of them on top of the Hitchhiker, give each one a pair of radial chutes (this is Eve, correct?) and release them in sequence before pulling the chutes for the Hitchhiker itself. I'd also switch out two of those batteries for a pair of toroidal fuel tanks and then add a pair of radial engines just in case you need them; I've found that they can be used for asymmetric thrust can be very good for saving rovers that have flipped due to reckless driving as well as safely landing at your destination.

Hm. I suppose I could do a series of sequential ejections with stack separators between them and then dump the hitchhiker. It would leave them a little scattered, but if done low and slow it shouldn't be too bad and would allow a much saner looking fairing.

Whilst I can flip the rovers on Kerbin (sharp turns at max speed), Eve's higher gravity makes them absolutely stable. Only time I managed to tip one was when it was hauling the Hitchhiker around (very top heavy, short wheelbase. I'd like to be able to chain the rovers together to create one long rover and move things as a unit, but the steering bugs out )

And if you're intending to return them home then the fuel tanks + engines will allow you to re-stack everything to pack it up and go home.

The rovers and hitchhikers are gonna stay behind, they're going back in the 3-man pod at the front of the plane.

Edited June 1, 2013 by foamyesque

[s20dan]

How does X-Plane do it?

In case you don't know, X-plane is a relatively realistic (perhaps the most realistic non-commercial) flight sim with an aircraft editor/constructor program for creating your own craft kind of like KSP, only much more complicated. Its also been around for a LONG time.

I always *thought* that it simulated a percentage of air particles and collided them against the aircraft and extrapolated from that, but I can't find any definitive info that says exactly how it does it.

But my point is, however aerodynamics are calculated in X-Plane, they feel pretty accurate and pretty snappy.

[camlost]

How does X-Plane do it?

In case you don't know, X-plane is a relatively realistic (perhaps the most realistic non-commercial) flight sim with an aircraft editor/constructor program for creating your own craft kind of like KSP, only much more complicated. Its also been around for a LONG time.

I always *thought* that it simulated a percentage of air particles and collided them against the aircraft and extrapolated from that, but I can't find any definitive info that says exactly how it does it.

But my point is, however aerodynamics are calculated in X-Plane, they feel pretty accurate and pretty snappy.

I haven't played X-plane, but seems they use Blade Element Theory, which is kind of similar to what FAR does

[hiul]

arg. this mods is great but i can't get why my plane is rolling when in takeof it's the most simple there is and I don't get it u_u.

http://speedy.sh/PWmrv/Auto-Saved-Ship.craft

someone can help me to get it ?

[Doomydoom]

arg. this mods is great but i can't get why my plane is rolling when in takeof it's the most simple there is and I don't get it u_u.

http://speedy.sh/PWmrv/Auto-Saved-Ship.craft

someone can help me to get it ?

It isn't compatible with the latest version. Can you just post a screenshot?

[Camacha]

@Doomydoom:

I'd honestly love to do a more advanced aerodynamic simulation that handled the aircraft as a whole rather than part-by-part like I'm doing now. The first problem is that I would need a robust algorithm to determine all the pertinent details of a vehicle's geometry (which could range from identifying wings, fuselages, etc. to try and handle individually or trying to create a mesh that could be used to do proper CFD using the Reynolds-averaged Navier-Stokes equations with some kind of simple turbulence model). The second problem is trying to implement the simulation in an efficient and robust manner, by which I mean it doesn't lag your computer to hell and it doesn't vomit when you try to fly an Abrams tank because it doesn't know what to make of that. The third problem is trying to apply the forces and moments calculated for the vehicle as a whole to the proper parts so that you can get structural failures from aerodynamic loading without causing unrealistic behavior.

So there's a mini-rant on the subject; believe me, I've spent a lot of time thinking about this topic and a reliable solution to the geometry configuration problem still eludes me, especially because it has to be able to update in real-time to handle vehicle deformations.

I assume you are talking to the developers about these kinds of things?

@foamyesque: I'd actually create a stack of them on top of the Hitchhiker, give each one a pair of radial chutes (this is Eve, correct?) and release them in sequence before pulling the chutes for the Hitchhiker itself.

This would have been my solution. After fitting a couple of unwieldy loads into fairings you kind of get used to compact thinking. It is a bit of a shame Damned Robotics is broken at the moment, since that is a perfect mod for getting complicated stuff to fit inside a cone. It should be back in operation sometimes in the future, though.

[ferram4]

@s20dan: I believe that camlost is correct and that they use Blade Element Theory for their propellers and wings; I also think that they do some very simple simulation of the general flow-field around the plane to handle the aerodynamics of the fuselage and to implement the downwash calculated through the Blade Element Theory implementation.

The main difficulty is the geometry configuration problem: how do I identify what is what on this vehicle? In X-Plane I believe the creator of any particular vehicle asset determines the proper simulation grid and how many blade elements need to be used, which simplifies the problem greatly; that cannot be done in KSP without making aerodynamics subject to the player's whims (knowledgeable players will be able to manipulate and glitch the physics to their advantage) or relying on some minor form of geometry identification to control the player's choices, at which point most of the hard part is done already and asking the player to handle the problem would be annoying.

@Camacha: I've thought about it, but I don't know how well it would be received. I know that they've stated they want to try to dance on the line between "game" and "simulation" and I know that I'd probably push KSP much closer to "simulation" than anything else. I suppose at a later time I can try contacting them, but I assume that they have enough technical know-how to implement an adequate aerodynamics model without me. I'd be happy to help, I just don't think I'm really needed to solve that problem.

[foamyesque]

This would have been my solution. After fitting a couple of unwieldy loads into fairings you kind of get used to compact thinking. It is a bit of a shame Damned Robotics is broken at the moment, since that is a perfect mod for getting complicated stuff to fit inside a cone. It should be back in operation sometimes in the future, though.

Mm. I was going for compact vertically, but it seems to've been irrelevant.

I might wind up switching back to a FF part though... I'm having a devil of a time getting those to separate cleanly.

[aquilux]

@ferram4

"that cannot be done in KSP without making aerodynamics subject to the player's whims"

Considering how simple the .cfg files are to understand and the sea of mods out there, I'm sure that anyone using your mod would be into more realism. Considering that any knowledgeable player could just bring up the debug menu and cheat his/her way to anywhere, I'm sure anyone who is going to use your mod will use it properly as that will probably be their intent.

"I don't know how well it would be received." I may be imagining things, but I'm pretty sure they've said that they plan on making a more realistic aerodynamic model at some point, just that they haven't gotten around to it yet. (In fact, I think it was mentioned when they were talking about implementing reentry heat)

As for not wanting your help, just look at the people they've hired on in the past. They have a tendency of snapping up the best modders out there that are working on stuff they haven't gotten around to. It seems like while they know that they could get something to work, they will happily let someone else work as part of the team if they feel that someone will be better at it then themselves. Just promise to post a mod to bring it to your level of realism if you think they fall short.

[hiul]

It isn't compatible with the latest version. Can you just post a screenshot?

sorry for the wait but here is the screen shot of the plane:

this is the simplest plane I did and each and every one I do seems to have the same problem...

I looked into the joystick entries but nothing is entered and even when I only activate the keyBoard it does the same... I realy don't understand.

[Huelander]

Ferram4 remember this bug I posted a while before? Now that is possible to change the starting state of the cargo bays on B9 it seems to work as intended again, with the parts on that pack at least. Sometimes I need to reload/launch and go back to SHP for FAR recognize the right values, other than that it works fine.

[Doomydoom]

sorry for the wait but here is the screen shot of the plane:

this is the simplest plane I did and each and every one I do seems to have the same problem...

I looked into the joystick entries but nothing is entered and even when I only activate the keyBoard it does the same... I realy don't understand.

I'm thinking it's your landing gear placement. Place the rear gear a little behind your center of mass. And make all the gear straight vertical.

[hiul]

hmmm. well i'll try. but when I do take off (in the extremly rare cases that I don't crash before) the plane is utterly out off control spinning like hell even at low combustion. about tier 1 an under. tier 1 is enought to take off.

so it might be something else too. but ill try to make the gears near the center of mass.

[Doomydoom]

There are probably other things wrong with it aerodynamically, but bad landing gear placement never helps.

[Gauss H2K]

Could I get some tips on aerobraking on other planets with this mod. What would be good heights that would slow the aircraft down without disintegrating it. I'm aiming for an aerocapture.

If a specific scenario is needed: Here is the one I'm currently facing, an SSTO is docked to a Space Stationy Interplanetary Booster Fuel Ring Thingy, basically an aerodynamic brick assembled in orbit, not meant for atmo. This vessel has way too much delta v for a Laythe Landing, and I'm looking for a way to aerocapture the entire station without wasting its precious fuel.

[Doomydoom]

Bill flew the Star Rocket on a sub-orbital trajectory and upon reentry ignited the engines again to reach a top speed of over Mach 13. Unfortunately, the craft was torn to pieces when it tried to maneuver towards land. RIP Bill. Your sacrifice for aerodynamic science will be forever remembered. http://i.imgur.com/cfvcJlT.jpg

[Camacha]

@s20dan: @Camacha: I've thought about it, but I don't know how well it would be received. I know that they've stated they want to try to dance on the line between "game" and "simulation" and I know that I'd probably push KSP much closer to "simulation" than anything else. I suppose at a later time I can try contacting them, but I assume that they have enough technical know-how to implement an adequate aerodynamics model without me. I'd be happy to help, I just don't think I'm really needed to solve that problem.

Well, considering Kerbal Space Program is intentionally very mod oriented and FAR being one of the major mods, I can not imagine the developers minding a dialog about improving the situation for you or both. Squad has a tendency to incorporate solutions from the community or even hire modders to do work for them. Even if they are perfectly capable of developing something like FAR, I think your extensive experience might be useful. They will have to deal with the problems you are facing right now at some point. Maybe they could make your life a lot easier with some internal adjustments, maybe they could even use parts of your work. Who knows?

You are pleasantly humble about your work , but I think you're selling yourself short. Communication is always a good thing and if Squad really does not want to speak to you they can always tell you to go away

[theflyingfish]

Bill flew the Star Rocket on a sub-orbital trajectory and upon reentry ignited the engines again to reach a top speed of over Mach 13. Unfortunately, the craft was torn to pieces when it tried to maneuver towards land. RIP Bill. Your sacrifice for aerodynamic science will be forever remembered. http://i.imgur.com/cfvcJlT.jpg

I see you're not using Deadly Re-entry.

[Firov]

Ferram, I looked through the last couple of pages and I didn't see any mention of this, so I'll go ahead and bring it to your attention, just in case.

There seem to be some minor incompatibilities between FAR and the latest version of B9 Aerospace in regards to the way FAR handles the cargo holds.

First, the cargo holds in that mod aren't actually recognized as cargo holds, since your mod looks for "Cargo Bay" in the object title and the B9 holds are specified as "Cargo Hold".

Secondly, even if that's fixed by manually changing the title it's not properly handling the shielding calculations for the bays. The objects in the bay are considered shielded even if the bay is open. Furthermore, and more importantly, the cargo bay will occasionally shield objects that are attached to the exterior of the cargo bay. In an earlier post, discussing unrelated cargo bays, it was explained that this could result from cargo bays starting in an open state, however, even configuring the B9 cargo bays to start in a closed state (it allows setting the start state from the action menu) doesn't fix the problem, it just causes it to manifest later during the flight. I had the forward section of a compound delta wing become "shielded" during transonic flight, for example. Not one of my better flights... It does appear that the B9 parts use a different method of animating the cargo bays, from what you normally see, which is likely causing the conflict. Not entirely sure what can be done about it.

Anyway, I just wanted to bring that to your attention if it hadn't been already. Thanks for all your work.

[Doomydoom]

I see you're not using Deadly Re-entry.

Oh it was deadly enough.