Rentering with passenger cabin problems

[planet]

I built a space tourism bus with two cabin modules stacked but I had problems with it on re-entry: it kept flipping and heading nose first and there wasn't enough drag to slow the whole thing and deploy parachutes.

So I hit on the idea of using decouplers to split the whole thing up, attaching parachutes to each module individually. This worked the first time I tried it; the separate units floated down: the command module landed first and then I could watch the passenger cabins splash down and I recovered them and got my cash.

Then on one trip I was a bit high when I split the modules. I was watching the passenger modules float down from the command module view and at some point on the way down (pretty close to the ground actually) the view flipped by itself and the passenger modules disappeared. When I went to the tracking station they had gone. The tourists were gone, nowhere to be seen. Not in the astronaut complex or anything.

I concluded that the modules must have drifted too far away from the command module and that ksp "cleaned them up" or something, so on the next trip I waited until the last minute before splitting the modules. It worked and I recovered the passenger cabins fine.

Next trip, I did the same thing. The cabins were never more than 1200 meters away from the command module, but this time the view flipped again and the passenger cabins disappeared. I went to the tracking station and they were there though, but some time after (before I could recover them) they disappeared from there as well.

I am aware there is some problem/feature that prevents you from recovering parts, but I thought it was related to the distance between the part and the main rocket, but that appears not to be the case here.

Any idea what is going on?

Edited November 24, 2018 by planet

[Foxster]

Do you have a probe core on the decoupled part(s)? If not then KSP thinks its junk. 

Oh and this works...

Edited November 24, 2018 by Foxster

[OHara]

Welcome, @planet.  Even cabins with no probe core, thus shown as 'debris' with the fuel-tank icon on the map screen, I would expect them to stay in the simulation with you all the way to the ground.  When I drop uncontrollable packages under parachute, it works for me.

There is a distance, as you concluded, beyond which KSP stops simulating things.  It just leaves those things in their current orbit, and if that orbit has put them inside the atmosphere KSP assumes they burned up on re-entry and does delete them.   But the relevant distance, when the craft under control is inside the atmosphere, is 22.5km (from the file in Physics.cfg).  

If you are outside the atmosphere when separating, KSP might consider you in 'orbit', and there Physics.cfg says it "pack"s any craft more than 350m away.  I have never experimented to find what effect 'packing' has.  The view-flip usually indicates a switch between 'flying' and 'suborbital' and 'orbit' status, so maybe the switching of status is causing your cabins to be deleted as if re-entered.   (You might see a clue in the last few lines of the text file 'KSP.log' if you are so inclined to look.)

[Gargamel]

1 hour ago, Foxster said:

Oh and this works...

I did the same thing for my 39 seat tourist bus.    It was tumbling upon re-entry at bad times, so I slapped some airbrakes on it, and deploy them as needed.    If they start to overheat, I just put some roll into the craft and the exposed ones turn away and the cold ones gets exposed.    So it becomes a Kerbal Rotisserie. 

[bewing]

OHara explained it pretty well, but I think the answer to why your problem happened is a little more complicated than what he said.

Please understand that the game is designed to carefully simulate only one craft at a time. There are times when it needs to simulate more (e.g. rendezvous, decoupling, and crashing into each other), and the game tries hard to do it. But there is something called a physics bubble that extends outward in a sphere from the craft that currently has focus. Things that are inside the physics bubble get carefully modeled: deploying parachutes, reacting to atmospheric drag and SAS modes -- and things that are outside the bubble either go "on rails" or get deleted if the game thinks they have crashed or burned up.

So, the point is that the size of the physics bubble depends on what your current vessel is doing. If you look at the bottom of your physics.cfg file, it has all the scenarios listed out, with various distances attached. As OHara said, when you are in the air, the physics bubble size is 22.5km. As long as everything stays within that radius of you it will all be carefully modeled, because it's within the physics bubble.

However, you said there was one special thing about your situation. You said your command pod had the focus and had just landed. The default physics bubble size for a landed vessel is only 300 meters.

So, a few seconds after you land, the game changes the physics bubble size. And, following the game's standard logic -- anything that is outside the physics bubble, and below the autodeletion altitude (25km) gets deleted.

To make this not happen, either change the landed & splashed phsyics bubble size in your physics.cfg file, or, when your command pod lands, immediately hit the [ or ] key to switch focus to one of the floating passenger cabins that's still in the air (to retain the 22.5km physics bubble size), or, set the full deployment altitude for the parachutes on your command pod to 2000 meters so that your command pod lands last.

 

 

Edited November 24, 2018 by bewing

[overkill13]

Use two heat shields to shift the CoM towards the shielded end of the craft.  I have a line of Finch Class rockets that consist of a Mk1 Pod with passenger cabins along with the nose parachute plus other chutes, solar panels and batteries.  The single cabin line reenters fine, but the double and triple cabin crafts require a second heatshield to remain stable on reentry.  They will sometimes want to flip around high up in the atmosphere, but around 30-40km altitude it will lock in retrograde and only drift slightly out of line, never turning right over.  There's enough drag this way to forgo any additional braking measures, but I like to bring a radial drogue chute, just in case, along with the nose parachute and one radial standard chute per crew cabin.

[Fraston]

turn on debris in map view or tracking station.

should fix things.

[Foxster]

6 hours ago, overkill13 said:

Use two heat shields to shift the CoM towards the shielded end of the craft.  

Bolting on extra mass to shift the CoM doesn't seem like a great solution. You'll waste dV carting that otherwise useless heatshield around. 

I'd say add some lightweight draggy bits to the top makes more sense. 

Edited November 25, 2018 by Foxster

[overkill13]

2 hours ago, Foxster said:

I'd say add some lightweight draggy bits to the top makes more sense. 

You have a point, but I've never found a reliable way to measure center of drag without wings and CoL.  Basic fins tend to burn up and AV-T1s attached to a command pod at the front of a vessel are terrible looking, airbrakes are too far into the tech tree and even radially attached batteries or other bits don't look good. 

For getting around Kerbin's SOI a sub-10% increase in mass has proven negligible.  Adding docking capability to the vessel adds about the same mass and seldom require's a change to the launch vehicle.

[FleshJeb]

8 hours ago, overkill13 said:

You have a point, but I've never found a reliable way to measure center of drag without wings and CoL.

CorrectCoL mod. I’d link but I’m on mobile

[Orc]

Hi all,

I'm not qualified to comment on the more technical side of KSP's handling of things, but to answer the OP directly:

I've had good results with building a tourist rocket with the two (or even three) crew cabins ABOVE the Mk1 Capsule. Yes, I know, sucky draggy thing on the way up. But I've sent tourists as far as Minmus Orbit and gotten them back safely. You don't even need the MK1 to be carrying a heat shield. This configuration wobbles and spins like crazy upon re-entry into the lower  atmosphere (sub-20K) but it holds together. Use one of the big central parachutes coupled with two or three radial drogue chutes for slowing down, or go with two radial regular chutes with two or three drogues. 

Two or three air-brakes help enormously but they don't become available when local tourist contracts are the primary way of earning the big bucks.

Regards

Orc

[magnemoe]

Use an 1.25 meter service module, then open it works like an air brake who don't overheat, put probe core, batteries and parachute inside. Heat shield in other end. Put on aerodynamic nosecone on top of service bay, unless you add other stuff like drop tank or an fairing to also lift satellites  
More service bays works even better and give more room for stuff, the 2.5 meter ones can even hold an small satellite and is obviously even better. 
Note that bay has to be open to activate parachutes inside it, this is very important  

[Loren Pechtel]

On 11/25/2018 at 9:58 PM, Orc said:

Hi all,

I'm not qualified to comment on the more technical side of KSP's handling of things, but to answer the OP directly:

I've had good results with building a tourist rocket with the two (or even three) crew cabins ABOVE the Mk1 Capsule. Yes, I know, sucky draggy thing on the way up. But I've sent tourists as far as Minmus Orbit and gotten them back safely. You don't even need the MK1 to be carrying a heat shield. This configuration wobbles and spins like crazy upon re-entry into the lower  atmosphere (sub-20K) but it holds together. Use one of the big central parachutes coupled with two or three radial drogue chutes for slowing down, or go with two radial regular chutes with two or three drogues. 

Two or three air-brakes help enormously but they don't become available when local tourist contracts are the primary way of earning the big bucks.

Regards

Orc

You don't even need a capsule at all.  It's perfectly acceptable to have everyone in passenger compartments.  Something I did once that worked reasonably well for a suborbital flight was to put decouplers on the top of the rocket and nose cones on them.  Blow the cones and re-enter sideways (so long as the reentering bit is symmetric head-to-tail you can do this pretty well) for as long as you have the control authority.  I never tried that with orbital flight, though.

Edited November 28, 2018 by Loren Pechtel

[Orc]

Greetings all,

Thanks, magnemoe and Loren Pechtel, you've given me some ideas to try out for my next tourist contract binge.

Regards

Orc

 

[planet]

Thanks for all of the useful suggestions.

In the end I did this: I took the step to modify the physics.cfg to increase the pack and unload values after landing/splashdown to around 5km. It's not quite clear to me why it worked a few times without doing this, but maybe I was just lucky. I also set the capsule parachute to open higher than the cabins so it lands last. I added some extra boosters to give me more fuel for a deceleration burn and that allowed me to leave activating the decouplers to the last minute (about 20km); the parts then usually stay within 1.5km of each other. Finally, after touchdown I wait a bit for things to settle, recover the capsule and then the two cabins via the tracking station. This I do as quickly as possible (you don't even have to finish confirming the first before recovering the second).

So far it has worked three times in a row without losing anyone, so fingers crossed!