How to Land Your Fat Rocket Safely in the Water

[problemecium]

How to Land Your Fat Rocket Safely in the Water

Greetings fellow Kerballers~!
As this took me literally all night to figure out, I thought I'd spare posterity the frustration I went through and spell out the way to recover spent lift stages (you know, the way SpaceX tries to) without accidentally blowing them up (cough).

Step 1: Use the water (believe it or not!)
Historically, the seas of Kerbin have been universally reviled for their uncanny ability to act like solid concrete as far as incoming rockets are concerned, a fact not helped by the rockets' surprising buoyancy and their parachutes' nasty habit of shutting off as soon as anything touches anything. However, KSP 1.0 has changed the way parachutes work, turning the water into a promising solution for landing large, topheavy rockets that would have trouble staying upright on land (even on the mildest of slopes).
Water has two major advantages:
- It's always flat. No mountainsides here.
- When something lands in water, the bottom parts don't suddenly have to carry the weight of the rest of the rocket and run the risk of snapping off - to a degree. In KSP 1.0.5, small parts can still break off if they find themselves beneath something heavy. Struts are a lifesaver in these situations.
There's also one caveat: water does NOT ignore collision tolerances, and that includes any part that touches it. Even if the rocket as a whole is stationary, if it rotates fast enough that one end hits the water at a higher speed than the collision tolerances of those parts, they will crash and explode. Therefore rockets that are meant to land in water must include enough parachutes, or reverse thrust, to slow down to below the lowest collision tolerance of any part on the craft - for most parts this is 6 m/s. If you're not sure you have enough parachutes, feel free to spam them for the first few launches.

Step 2: Find your rocket's "centers"
In the editor, as you most likely well know, you can display three important "centers" for your craft: the center of mass, center of thrust, and center of lift. We're not concerned with the center of thrust in this tutorial, but the other two are relevant.
The center of mass is fairly self-explanatory. As your rocket comes in to land and experiences the various aerodynamic forces, it will pivot more or less around this point. When you land, the rocket will rotate around the point of contact based on the distance between the center of mass and the ground, so if this point is high up, the rocket will be "topheavy" and tip more easily than if this point is near the bottom (yes, this does seem painfully simple, but it mustn't be forgotten).
The center of lift isn't really relevant here, but what is relevant is the "center of drag," which tends to follow the center of lift, particularly at low speeds (as you will hopefully be going when trying to land). When you're going up, this should of course in most cases be behind the center of mass, and consequently if your rocket made it to space then you'll find the center of lift is accordingly low on the rocket. This is important because if your rocket "magically" slowed down sans parachutes, it would naturally point "forward" and thus down, hitting the surface nose-first. Generally we don't want that when landing, because when the other end hits it will carry with it the engines and winglets, which are heavy and fragile respectively and thus likely to be damaged on impact (particularly if we miss the water and end up on land).
A third "center" that will be involved here is the "center of parachutes" (shown below in green). I doubt this is a real physics term, but hey, I'm not a physicist and this is what I've discovered. This point is not visible in the editor, but it represents the average point at which the drag force from the parachutes is applied, based on their locations, sizes, and state (inactive, partially open, or open). This point may well be the most important one in this tutorial, because it determines which way your rocket will be pointing when it touches down.

(In case they're hard to see, note that there are four radial parachutes mounted at the top.)
In order for the strategy later in this tutorial to work, most of your parachutes should be clustered around the center of mass so that the "center of parachutes", i.e. their average position, is roughly on the center of mass. More precisely, the main group of parachutes should be centered just a smidgen behind the center of mass so that the rocket has a small amount of torque tilting it nose-down. Not to worry, this will be countered in the next step. If your rocket has winglets, they will, as discussed earlier, cause it to tilt, so the "center of parachutes" may need to be moved a bit further forward to compensate.

Step 3: Asymmetrical parachutes
Here's where it gets interesting. Most likely you found that the centers of mass, drag, and parachutes were all near the back end of the rocket, since the back contains both the engines and the winglets whereas the front is just empty fuel tanks. On the opposite end of the rocket, add just a few parachutes in pairs without using symmetry, but arranged into symmetrical pairs (the Offset gizmo is useful for this). They will look symmetrical, and behave in a balanced fashion when undeployed, but by being placed singly they can be tweaked differently on different sides of the ship.

Leave the parachute(s) on one side set to open along with the main group, and tweak the other(s) to open at 50 m. Generally I'd suggest also setting these to deploy early and setting the rest to deploy later, e.g. at 0.5 atm, so that you don't have to wait around for 20 minutes while your ship slowboats down from 20 km.
When the rocket reaches 500 m (or whatever altitude you picked), the main group of parachutes and the asymmetrical parachute(s) on one side will open, while the asymmetrical parachute(s) on the other side will remain partially open. This will cause the "center of parachutes" to be skewed to one side of the rocket, causing it to tilt away from vertical (nose-up or nose-down) and possibly even assume a horizontal position. It should remain tilted in this fashion until the last minute when the second group of parachutes opens. When it does, the "center of parachutes" will return to the center of the rocket, but since this occurs at a mere 50 m, there won't be enough time for the rocket to return all the way to vertical before landing. If everything goes well, the rocket will touch down at a significant angle, causing the main group of parachutes (or most of them at least) to despawn, but the asymmetrical group at the top end to remain open, guiding the top pieces down to a gentle repose and looking something like this:

Lying on its side, the rocket won't have to use its SAS to remain stable and will remain still until it is recovered at your leisure.

Step 4: Adjust and run tests
Oh no~! You followed all my steps to the letter and the rocket flipped all kinds of crazy ways and exploded! Or perhaps it turned fully vertical before landing, then tipped over and crashed. No worries, I saw this coming. Go ahead and revert to Vehicle Assembly. Most likely one of the following situations occurred:

The rocket nose-dived, then landed on its delicate little fins and/or the engines smacked into the water or land too hard:

This means that the true center of drag, i.e. the combination of drag from various parts, winglets, and parachutes, was just a bit behind the center of mass, so when the first group of parachutes opened, it assumed a tilted orientation with the nose pointing down, causing the nose to touch down first.
Imagine the center of mass as the fulcrum of a lever. On the short end is the main group of parachutes, and on the long end is the smaller asymmetrical group with half open and half partially open. If the nose points down, it means that the asymmetrical group is too "weak" and doesn't have enough leverage to turn the rocket nose-up. It should either be moved farther toward the front of the rocket, or if this is not possible, the main group should be moved a slight distance farther forward to give it a shorter lever on which to work and thus less leverage. Remember that the effects of any winglets may have given the main group a helping hand in pointing the rocket down, so the main group may have to be centered right on the center of mass or even possibly a short distance in front of it.

The rocket went completely vertical with the nose pointing down:

This means that the center of drag with the main group of parachutes open was too far behind the center of mass, and the first half of the asymmetrical group did not have enough leverage to tip the rocket more than a few degrees. Thus when the second group opened, those parachutes found themselves on the opposite side of the central axis of the rocket, as illustrated here. Consequently, when they opened, they turned the rocket back toward a nose-down orientation (instead of a nose-up orientation as intended).
The solution is to move the main group of parachutes forward, most likely closer to the center of mass. This will make the torque they generate smaller, allowing the smaller group of asymmetrical parachutes more control over the rocket, so that when the first half open, they will tilt the rocket far enough that the second half open on the same side of the center of mass. If this occurs properly, they will try to rotate the rocket toward a nose-up orientation through a horizontal orientation, which will leave it greatly tilted on impact and allow a gentle sideways landing.

The rocket went completely vertical with the nose pointing up, landed, then tipped over:

This means either that the asymmetrical group of parachutes is too powerful (perhaps there are too many in this group, or it is too far forward) or that the main group of parachutes is too far forward and thus pulls the rocket into a nose-up orientation before the second group opens.
To solve this, try moving the main group of parachutes back a small amount. If that doesn't work, and there is more than one pair of asymmetrical parachutes, try reducing the number of asymmetrical parachutes and increasing the number of parachutes in the main group. It's okay if the parachutes seem to be clumped around the bottom end of the rocket instead of the top as seems more intuitive - remember that despite what its shape suggests, the rocket's center of mass is probably very close to the back end.

I strongly advise performing several test launches, e.g. with an almost-empty rocket just hopping off to the ocean east of KSC, and adjusting the parachutes' positions a little at a time until everything works properly. I myself did about 50 test launches trying to figure all this out, so while you hopefully won't have to do anywhere near that many, it's entirely possible you'll need five or six tries to get everything balanced right.

Demonstration craft file: Download - Broken in KSP 1.0.5. A replacement may be uploaded soon.
Let me know if this helps, it's hard to understand, or anyone still has trouble after trying it!

Edited December 1, 2015 by parameciumkid
Forum upgrade problems xP

[hieywiey]

Interesting... I think this might also be a good guide for how to design boats that won't explode upon landing.

[mrclucks]

I remember reading this a week ago and noting how interesting it was, then I modified designs to emulate what you did. In practice I've found that parachutes still work when a craft has touched down but is still rotating. It's only a few seconds but that can make a big difference in total destruction and sticking the landing. This has worked in both land and sea for me

[basic.syntax]

Attempting in-physics recoveries was one of my favorite .90 activities & frustrations. I've been doing more forum reading or writing lately, than actually playing. Thanks for this great overview, it inspires me to 'get back in there'

[problemecium]

Update!
I've run some tests in KSP 1.0.5, and in principle this still works, but with the changes to buoyancy it is VERY unforgiving of part clipping and of rockets wherein a small, weak part lands before a large and heavy one. For the time being, the demo craft linked in the OP is unusable, so I'll probably post a new one soon.