So the reason why parts far enough are deleted in low atmosphere…

[Cesrate]

I guess, as the "collision plate" of planet surface has a limited radius, if the parts weren't deleted they will fall from the "ground" or, "planet plate". Correct?

All right, I'd like to know, considering the structure of KSP, where's the limit about how far in distance can a ship be physically presented without falling off or falling apart?

Edited December 17, 2013 by Cesrate

[KasperVld]

This is actually a deliberate choice: parts get deleted when they reach a certain atmospheric pressure and aren't loaded to simulate orbital decay. It's crude, but it works

[Kasuha]

Planet do not have a collision box, but the game just does not want to care about resolving terrain details at each place where random piece of junk falls.

The current method is effective (i.e. does not cause unclear pauses for the player) and results are reasonable. We may get improvements on it in the future or it may stay the way it is.

The limit of physics simulation is 2.5 km. Anything that gets further away from the active ship in Kerbin's atmosphere below 23 km disappears unless it is landed. Above 23 km it will follow its usual orbital trajectory as if the atmosphere is not there.

Different planets have this (23 km on Kerbin) limit different.

[tntristan12]

It's crude, but it works

Except it doesn't. If a major portion of career mode is to be recovery and reusability for the sake of keeping economic costs down, then having debris (potentially manned spacecraft with parachutes and heat shields equipped and set to deploy) simply burn up without any effort to simulate whether or not they can be recovered undermines that mechanic significantly. I know there are some basic models out there that can simulate where an object will land with some basic assumptions made. This is one area where a little effort would go a long way. I just hope it doesn't get lost in all the noise.

[Cesrate]

Humm, how about just running the simulation in two parts? It's not efficient but may work and usually players won't put too much things in atmosphere. And you only need to load the ship with simplified physics model and a precise "collision plate" on surface. Without rendering it.

[Stargate525]

tntristan, that's a BIG if. This is how it works right now, and it's a stopgap. Doesn't mean it won't change. I believe that the mission controller mod had a fairly decent simulation of spent stage recovery based on drag and parachutes versus weight. So it is possible.

[Murph]

Except it doesn't. If a major portion of career mode is to be recovery and reusability for the sake of keeping economic costs down, then having debris (potentially manned spacecraft with parachutes and heat shields equipped and set to deploy) simply burn up without any effort to simulate whether or not they can be recovered undermines that mechanic significantly. I know there are some basic models out there that can simulate where an object will land with some basic assumptions made. This is one area where a little effort would go a long way. I just hope it doesn't get lost in all the noise.

But it does work for now, as the economics have yet to be implemented. When they do come along, it's fairly likely that the behaviour will change. A trivial fix as part of the economics implementation would be the following pseudo-code run when the object is currently deleted:

if (object.recoverable())
    object.recover();
else
    object.terminate();

Something like that would add very little overhead over the current behaviour, just needing a fairly simple function to determine if the object is recoverable, something like checking for sufficient parachutes relative to the total mass and drag, which could be determined in advance to minimise calculations during flight.

[makinyashikino]

If the debris didn't get deleted, you'd have quite a lot more debris surrounding KSC, especially if you asparagus stage. Also, you if you left debris in a 40,000,000km x 30km orbit, you would be driven insane because it would kick you out of time-warp in order to perform the physics. Unless the debris just passes through like when you high-warp through a planet.

[Kasuha]

Except it doesn't. If a major portion of career mode is to be recovery and reusability for the sake of keeping economic costs down, then having debris (potentially manned spacecraft with parachutes and heat shields equipped and set to deploy) simply burn up without any effort to simulate whether or not they can be recovered undermines that mechanic significantly. I know there are some basic models out there that can simulate where an object will land with some basic assumptions made. This is one area where a little effort would go a long way. I just hope it doesn't get lost in all the noise.

It works, meaning it makes sense. If you want to recover your craft, you need to make sure it landed safely. Unattended craft has a parachute failure, decompression, structural failure, crashes into terrain and gets destroyed losing all its scientific value.

If it was the game's task to make sure all your ships landed safely, you'd be able to recover them from orbit.

For the game it is of course much simpler to avoid running numerous independent physics simulations at once. For you it means less lagging.

[bsalis]

Except it doesn't. If a major portion of career mode is to be recovery and reusability for the sake of keeping economic costs down, -snip-

I agree with you. However when we do get money and performance tuning, then proper physics (PhysX) may be made to run on detached parts of your currently focused vessel until they land/crash.

[dr_jt]

Recovery for use again could be a pretty complex thing to include.

Do you do it to the point that the player has to go and recover the part?

Do you do it so that a part can be damaged but still usable?

How would you handle pricing the part? Does a recovered part cost more, less or the same? The first thought is less but should recovery and refurbishing costs be included in the cost?

Does the reusable part get added to the part editor? An entry for each part because they may all be priced differently.

Do we want the part editor to have page after page of the same SRB because it was recoverable?

How do you cost sub-assemblies? By the parts or by the whole assembly? Do you include labor costs?

Do you tag all parts as reusable or only certain ones?

Do you allow recovering junk left in orbit or on other bodies? If so do you include the cost of the mission as part of the recovery cost? Would it be worth spending 20 million to recover 20 thousand worth of parts?

These are just a few of the things I can think of as I write this and I think it only scratches the surface of doing a complete re-usability interface.

It could also be handled very simply by just tagging some parts, say SRBs, as reusable and if they crash to Kerbin you get a certain amount of credits added to your account. Done that way it doesn't involve many code changes to the current debris deletion code. Maybe a display similar to the science gained for a mission could be used to tell which parts got recovered and how much they were worth. Have some randomness with parts not being recovered and gaining less credits for badly damaged parts needing a lot of repair. Maybe even a cost based on difficulty in recovery. An SRB a few kilometers from KSC would be a lot cheaper to recover than one in the ocean or on another part of Kerbin.

Recovery and re-usability is very much like resources gathering and thinking about it I can see the difficulty and effort that would be required for resource gathering or re-usability. It just might be to big and complex of a thing to be worth including. Both would be nice but both may have a low return on investment for the the dev team.

[tntristan12]

It's not just about recovery. How many times have you sent a piece of debris on a free return trajectory around the Mun hoping that it'd go back home and burn up in the atmosphere, only to discover that its periapse is at 26km instead of the 24 you predicted? That 2km of difference means you have a piece of debris perpetually orbiting in a highly eccentric and (what should be) totally unstable orbit. So then you either have to switch focus to it manually and watch it burn up, or remove it from the tracking station - both of which break the flow of gameplay for the player, while the latter also undermines any efforts the player might be making to "play realistically."

I know that it works well enough for now, but it does need to be improved before the final release; there's just no discounting that fact. A model to simulate aero-drag on unfocused parts need not be very complex or too computationally intensive! Consider that the only times the player will really encounter situations where unfocused debris enters the atmosphere are shortly after launch (i.e. the initial stages of a rocket), returning from an interplanetary mission (i.e. the example I mentioned previously where debris comes back on a free return trajectory / coming back from Duna and detaching the lander before re-entry), and setting jettisoned stages to de-orbit immediately after stage separation (i.e. ejecting the escape tower once your craft gets into orbit).

Notice that all of these cases require direct player intervention, and so it's highly unlikely that two or more of these will be occurring simultaneously (and even less likely that they will be occurring on different worlds). Therefore it is entirely reasonable to expect the total number of non-focused debris being acted upon by aero-drag to be in the single digits. For simplicity you could even group all pieces of debris that are within a certain distance of each other, i.e. 2.5km to be one piece of debris that de-orbits together, dropping this number even further. The devs could very easily add a slider bar in the options menu to set the maximum number of pieces of debris that get "simulated" in upper atmospheres, with a default value at 10 or so. That way, if 10 or less pieces/groups of debris are in the upper atmosphere, the game could run a much simplified series of calculations to determine approximately how they will de-orbit based on the average drag coefficient, some basic assumptions about exposed cross-sectional area (in all cases the mass is known so they could substitute some reasonable value for area based on that), and the altitude the piece/group happens to be at at the time (the density of atmosphere as a function of height is known).

It wouldn't be perfect, but it would give players a better sense of things going on outside of their 2.5km physics bubble. It'd remove the blatantly unphysical behavior of de-orbited pieces of debris continuing to orbit with a periapse of 26km. It would allow a more realistic simulation for the survivability and recovery of re-entry parts. And most importantly to me, it would mean that you could send multiple manned flights home without having the ship and crew suddenly disappear, despite the fact that I had pre-deployed the parachutes and was currently focused on another ship only 2.5km away. If I had any modding skill of my own I could probably make this happen, but as it stands I can really only hope the devs are out there reading this thread. :\