Overview
There isn’t much incentive in career mode to use reusable rockets, because of how much margin contracts give. The aim of this guide is to provide some tips to save you some time in getting good at a part of the game that isn’t very incentivized but is nevertheless a fun challenge. I will also try to add sections as time goes on and to regularly update for newer versions of KSP.
Theory
There are several factors that limit the possible configurations in KSP for reusable rockets. The main restriction is the fact that if a stage falls below somewhere around ~24 km in altitude, the game seems to delete them automatically. As a result, to recover those stages, you need to switch to them before they dip that low so that the game keeps on computing their trajectory. After doing so, however, you cannot switch again to another vehicle that is not in physics range (like a second stage in orbit) while in the atmosphere, until landed. The effect of this is that the second stage must have a high enough thrust to weight ratio to complete the orbital insertion burn before any stages fall too low in the atmosphere.
Based on my experience flying recoverable vehicles in KSP, there are several solutions that fit within the set of possible configurations:
1. Hover Slam Vertically Landing First Stage
2. Hover Slam Vertically Landing SSTO First Stage
3. Parallel Hover Slam Vertical Landing First Stages
4. Horizontal Landing First Stage
5. Parallel Horizontal Landing First Stages
6. Horizontal Landing SSTO First Stage
7. Parachute Landing First Stage
8. Parachute Landing SSTO First Stage
9. Parallel Parachute Landing First Stages
These would be paired with several options for second stages:
1. Parachute Landing Second Stage
2. Hover Slam Vertically Landing Second Stage
3. Expendable Second Stage
4. Horizontally Landing Second Stage
5. No Second Stage
There are of course more exotic second stage or first stage arrangments, such as parallel second stages, or parallel SSTO first stages that can reach orbit before needing to stage (at which point it ceases to be an SST( ). However, these are either highly specialized for specific payloads or completely unnecessary, as far as I can tell. Unnecessary complexity is important to avoid because reusable rockets aren’t really worth it if they are impossible to use at least relatively easily or require large amounts of time to successfully use.
The way I think of this is as three tiers of difficulty – parachutes are the easiest to use to recover stages, while propulsive landing (a “hover slam”) is significantly harder, and finally SSTOs are the single most difficult approach. The level of reward these options provide depends on your patience.
My hypothesis that I have extensively tested is that the easiest way to create a vertically landing reusable rocket is to divide the rocket into a reusable first stage and an optionally reusable high performance second stage that lands under parachutes. I should note that it is entirely possible to create a vertically landing SSTO rocket, but this is unnecessarily difficult due to the small margins and comparative inefficiency. Additionally, it is possible to create a vertically landing second stage that lands under active propulsion, but again this cuts into margins and impinges on overall performance to an unsatisfactory degree. It is comparatively easy to create a second stage that lands under parachutes. Because the second stage – with its small engine(s) and significantly smaller tanks – is significantly cheaper than the first stage, it isn’t exactly necessary to recover. Plus, for payloads going to the Mun or beyond, recovery of the second stage becomes more difficult with the addition of a heat shield for reentry and, again, cuts into performance excessively.
Due to these factors, the easiest to use design as far as I can tell is a 7:3 arrangement, but if you have more patience a 1:1 arrangement is a more interesting challenge; finally, if you have tons of time and really want to have fun, you could go straight for a pure SSTO (or beyond) design.
My personal favorite is a 3:3 Falcon Heavy style triple-core arrangement.
Hover Slam Design Philosophy
Boostback burns are extremely wasteful, and barge landings require extra margins for maneuvering onto the barge, not to mention the time required to get good at even landing on barges with a 50% reliability.
Instead it is better to balance the combined mass of the payload and second stage with the mass of the first stage to allow the first stage after being released by the second stage to follow a trajectory that sends it to the next continent over the ocean from the KSC. If this is done correctly, then the first stage has enough left-over fuel to land on the next continent, and the second stage has enough fuel left after orbital insertion to send the payload to the desired destination.
As one can see from the video, there was a bit of inefficiency in the profile – I staged the payload out while still in low Kerbin orbit with a large fuel margin. This means I could have added payload mass and reduced the tank size in the second stage. On the other hand, the configuration shown in the video was entirely capable of a boostback burn rather than landing on the next continent. This serves to illustrate how much more efficient landing on that peninsula is compared to an inefficient boostback burn, in view of how much fuel was left over.
SSTOs
SSTOs are as much a work of art as they are functional. (Debatably) they are the most fun single category of vehicles to design and fly, requiring the most attention to margins, performance and stability. As such SSTOs are the most time consuming, but also most rewarding.
As far as I can tell, the standard ascent profile depends on the thrust to weight ratio of the SSTO – if it is a low TWR SSTO, it is more efficient to fly at very low altitudes and start to climb after passing around ~400 m/s, which for the Rapier engines is the make or break point in the trajectory. After climbing to a point where the thrust of the engines drops off significantly, closed cycle mode plus nuclear thermal rockets provide the kick required to send the apoapsis into space. For a higher TWR SSTO, it is more efficient to immediately start climbing, if in a relatively steep climb 400 m/s and beyond can be reached before somewhere around 5km in altitude.
I would recommend this SSTO I created, as it possesses reasonable performance and does not suffer terribly from chronic instability, which I would admit is a problem many of my vessels run into…
[link]Future Work
An obvious next step for reusable rockets in KSP is either complete, or at least partial, automation, using the kOS mod. This and/or trajectory prediction mods can allow for a range of possibilities made possible by precision landing – such as on strategically placed barges in order to optimize for certain payloads or trajectories.
The first graph shows the linear decrease in horizontal velocity in an example first stage following retrograde down onto the ground. Note that there is a first linear segment and a second line on the graph after the landing legs deploy. This is because at low altitudes, stages fall at very close to terminal velocity. The stages decelerate because of the increase in thickness of the atmosphere, which means that the terminal velocity is not static. The graph is roughly linear, because as it happens the parabolic nature of the retrograde trajectory combined with the changing terminal velocity combines to lend a linear decrease in horizontal velocity. This is important because it means we can easily approximate for a given cross section the horizontal velocity required in order to reach a target landing site. This means that it is easy to create a kOS script where a certain horizontal velocity + horizontal distance triggers switching from a glide script to a retrograde lock and active propulsion maneuvering script for last-minute precision translation. A simplified version of this is shown on the graph – initially, horizontal velocity increases as the stage autonomously maneuvers towards the pad, before switching to retrograde lock. The key takeaway here is that a script is relatively easy to produce where there is active correction for built-up errors created in previous phases of flight relying on approximations.
I will update this section as I continue to work through this problem for situations where the stage starts with a higher horizontal velocity. Still, as it stands, this is how I approach this particular problem.