Kerbal – Realism Overhaul

As with the more common launches of SpaceX, for example we had recently the first manned mission of SpaceX go to the ISS and back, and a lot of folks are getting interested on why rockets are reused, why it hasn’t done before, etc. I’m not a space nerd of the level of Scott Manley or Everyday Astronaut. If you have time, I hugely recommend checking out these two Youtube channels.

Kerbal Space Program would be ideal to explain these things to others, as Scott has done in the past. But of course, we’re getting here into economics, staging in Earth, recovery… these things work quite differently in Kerbal.

So I thought, let’s install Realism Overhaul mod so we have the real thing here, play a bit and maybe I can write an article about different stuff of space tech. Hah!

Months later and a hundred of hours, I’m still not confident if I know how to play this game with Realism Overhaul. It is one of the most challenging mods I ever tried for Kerbal. Let me show you what (pain) it has to offer:

Advantages of Realism Overhaul over Stock

Realistic experience in every sense!

  • Lots of fuel types that match in density, performance and prices to the real world.
  • Lots of real engines with real performance. Also they have tons of TWR!
  • Realistic tank weights.
  • Real Earth and Real Solar System replaces the Kerbol Solar system.
  • Most rockets can be built to match specs and they perform quite close to the real ones.

Disadvantages of Realism Overhaul

It’s too real to be “playable” or enjoyable in any sense. Real life it’s not only hard, it’s overwhelming.

  • Low Earth Orbit (LEO) requires now 10,000m/s of DeltaV under manual control. If using a carefully tuned kOS script or similar automation, it can be brought down to 9,000m/s.
  • Too much complexity: Too many fuels, too many engine options. Also the ISP of the engines is quite lower (Except for LH2 engines). The easiest way to get to orbit is a three-stage rocket.
  • Engines now suffer from ullage and lack of pressure, depending on the engine. Also they have limited ignitions (most of them, just one) and suffer from turbo-pump delays.
  • Reaction wheels now have real-life torque (insufficient for most purposes). Despite this, they’re still not real because they cannot be saturated as the real ones.
  • Ion engines have now zero newtons of thrust (milli-newtons, but the game UI rounds to zero). Unusable unless you add some sort of persistent thrust mod.
  • Part sizes are real and don’t match up between them. Unusable unless you add procedural parts mod.
  • Water physics and propeller physics no longer work properly. I’m unable to make a boat or an helicopter.
  • Sometimes ships start spinning out of control when we exit warp. This happens usually when using fast settings of warp (i.e. going to moon).

I have suffered a lot of bugs, kraken and realism slaps. If you’re still with me, I’ll explain what I did so far to make it work.

Step 0 – Downgrade KSP to 1.8.1

This is the first test to see if you’re brave enough. It doesn’t work in the latest KSP, so I had to go back two versions, to 1.8.x. I’m not quite sure if all the mods will work in this version, but at least my basic ones did.

Then I could install it via CKAN, and worked like a charm. It made all my other games unreadable, so had to start a new sandbox to play around.

I removed most mods and just left with my basic set:

  • Kerbal Engineer Redux (DeltaV prediction + UI info assortment)
  • Trajectories (to predict descent paths with aerodynamics)
  • Module Manager (basic mod to glue mods together)
  • TweakScale (allows resizing some parts)

After this, I requested CKAN to install:

  • Realism Overhaul
  • Real Solar System
  • Real Solar System Textures – 8192×4096

And a huge amount of mods were required and recommended. Follow the recommendations as much as possible.

The only thing to be careful about: Kopernicus Planetary System Modifier has been updated and now it works only with 1.9.1. At the time of this writing the other mods do not support this version, so you’ll have to downgrade Kopernicus to 2:release-1.8.1-1.

Step 1 – Re-learn the tanks, fuels and engines

The engines are a complete mess. The sizes are completely random; everything is huge. And almost anything has a bizarre fuel consumption.

The tanks have now a GUI to choose the propellant from a selection of more than 50 different fuels. Luckily when you plug an engine the tank gets a button to fill it with the propellant needed for the engine, in the right proportions.

I spent hours trying to figure out how to assemble anything here because the tank parts now are scaled into real-world sizes and they don’t match together. Any kind of rocket looks like the frankestein monster.

Also the engines are either tiny or huge; there’s almost nothing in between. Some engines are huge in size but light.

Note that on right-click on an engine there’s now a button called “Engine – Show GUI”. This is very important. This dialog allows in a lot of engines to select different variants that might allow you throttling, vacuum enhanced versions or several ignitions.

In the end I did a very stupid 3 stage rocket that was able to put a satellite into orbit:

The cost of that rocket is 108,449√ of which 12,690√ is the payload (antenna and other stuff to put in orbit) and 15√ is the propellant for all the stages. So the empty rocket costs 95,744√.

I thought this had to be wrong. The propellant cannot be so cheap compared to the full rocket; Sure, propellant is cheap compared to labor and high precision manufacturing, but is it that much? So I searched the internet and found this image:

Cost breakdown of Delta IV Heavy launch - Space Exploration Stack ...

I am shocked. I thought they were cheap, but not by that much! So it seems clear why we would like to reuse rockets, and why we don’t care that much on engine efficiency. It has been there a lot of discussion regarding about development of aerospikes or other engines that are more fuel efficient. The cost of the propellant is not the problem, but the cost of the rocket itself.

Step 2 – This is Earth, not Kerbin

Well, this is the main point of Realism Overhaul, you launch from the Earth, on the real solar system from real locations. Want an equatorial orbit? Too bad! you can’t reach it from most launch sites.

The main problem is the Delta-V requirements to put things in orbit. In Kerbin, this is around 3200m/s; For the Earth, I need around 9900m/s. The atmosphere raises up to 140km of altitude, but feels the same as in Kerbin. So instead of going from a bit of atmosphere at 69km to zero at 70km, in the Earth it keeps decaying exponentially. I believe the atmosphere in Earth is slightly thicker at 20km than in Kerbin, but not by much. In Kerbin, flying above 20km is near impossible, while in the Earth this ceiling feels more like 25km.

But back to the Delta-V problem, now we have the problem of the rocket equation. The rocket grows exponentially as the Delta-V or the payload increases linearly. So imagine how stupid a rocket needs to be to get 10km/s of Delta-V:

As most engines don’t reach an ISP of 300 seconds, it grows crazily. In the above example I’m using the first LH2 engine I found, with an ISP of 403 seconds on vacuum. Thanks to it, it’s still “manageable”, so I could do the launch in one stage + boosters.

Step 3 – Real life engines are… “special”

Used to the stock engines, there are lots of things from RO engines that come completely unexpected. The only good thing compared to the Kerbal Stock counterparts is the Trust-to-weight ratio. They pack a lot of punch!

Most of them have a TWR of 20 or more. Some reach or exceed 100. Excellent, right? Well, yes.. but: They’re mostly unthrottleable. I mean, the throttle is stuck at 100%, with no way to dial it down.

Remember those crazy challenges of reaching Kerbin’s orbit with Solid Boosters? So it’s more or less the same in RO. At least you can stop the engine in this case…

…but you can’t restart it afterwards. Most engines have a finite amount of ignitions, a lot of them are rated for just one. So if you mistakenly cut the throttle to 0%, you lose the entire stage.

The engines also don’t ignite to 100% thrust instantaneously as they do in stock, they take around 3 seconds to power-up, which gives trouble when releasing the rocket from the pad and on some staging processes. Some engines do fire up quickly, it depends on the type. (If they have a turbo-pump, most likely they take time to ramp-up)

Then we get the additional problem of some engines being pressure-fed. This means you’ll have to remember to use special pressurized tanks for those, or they won’t work at all.

And for those engines that aren’t pressure-fed, most probably they have to deal with vapor in the fuel lines. So, if the tanks are being “shaked”, the fuel doesn’t sit on the bottom and instead of sucking fuel, they get gases. So the engine stops abruptly or fails to start. This does waste an ignition cycle. This is a very common problem when staging, so common that I started to add RCS thrusters just to push the vessel forward so the fuel gets towards the back of the ship. In this way at least I can recover from this issue.

Step 4 – Missing mods

I realized that this was just unplayable, because it includes FAR aerodynamic model, and without a way to seamlessly join parts, this is almost impossible. So I found there are a few mods that are recommended to go with Realism Overhaul:

  • Procedural Parts
  • Procedural Fairings
  • kOS (Kerbal Operating System)
  • TooManyOrbits
  • PersistentThrust

Finally – launching some rockets

Let’s start easy and try to launch a satellite. One of the first things we’ll need is an array of satellites for bouncing the connection around. Realism Overhaul does not have enough sites to cover for low altitude unmanned planes (around 15km and bellow).

Every gram counts here, so we have to design these properly. Start by using a “Delta Avionics Package [1.45m]” and scale it down to 0.625m. Don’t worry about this being “cheating”, it will not save you from failing 99.99% of the time. In the same fashion, add two “XT1 Solar Panel Array Mk2” and scale it down to 50%. Two antennas “Communotron 32”, and a procedural battery of 300x40mm. This weights 21kg.

Now we have the basic antenna. I want to send eight of these at the same time to put them in precise orbits. So we’ll need to clone it several times. To make thinks easier, add on top another Delta Avionics Package, scale it down, and make it the root of the craft. Now we can use Right Shift+Left Mouse Button to clone it.

Before that, we will need a decoupler on the bottom. The TD-06 does the job here, just rescale it down to 0.625 so it fits. And we can clone now the part 6 times, and it’s stackable if we keep turning each part around 30 degrees. Well, not exactly, I had to turn the antennas a bit outside and to the right. Here’s the final result:

You might be asking why 8 and not just 3. Simple. For best coverage we want a low orbit. And this means we don’t have much visibility, so we’ll need more to cover the sky.

We’re ready to move to the next stage. This will be the stage that will push each satellite in it’s own orbit. For this I have two engines that I like. The RL10 Vacuum is a good candidate. It’s a LH2 engine with 50 ignitions when configured as CECE-Base, and also good throttling capabilities. The ISP is 460, which is quite good. The other option is using 2.2/3.6kN Thrusters, which are basically RCS engines for main thrust.

The advantage of using 3.6kN Thrusters is that we can use a single propellant for everything. The engines are super-light weight, and can be ignited as many times as we want. But they don’t throttle down, and the ISP is around 320s at best.

On the other hand, going LH2 has a hidden problem: boil-off. The propellants will evaporate over time, and using them to correct orbits means days or months of throttle up, down and pauses to wait for orbit correction. We will need to use radiator panels to avoid this from happening.

So, which one? I say both. Because monopropellant (or in our case, hypergolic propellants) has to be carried anyway for attitude control (forget reaction wheels), we can design a small stage that will act as attitude control and final push.

Remember to click “Engine GUI” on both thrusters and RCS and select Aerozine50+NTO, and don’t forget to click on Tech Level until it maxes out. This should bring the ISP to 321s.

The tank, I’m using a procedural one of type “Fuselage”. Anything works as long as it shows “Highly Pressurized?: True”. If it says False, it doesn’t have enough pressure, and the thrusters will not work.

This part doesn’t need much DeltaV. It’s just for small adjustments and a final push. Don’t overdo it because it will make the next stages way heavier. Mine has around 1700m/s.

Next, let’s cover everything with a fairing. I’m using the stock 2.5m one. As tentative as it is to use the small 1.25m one (it fits beautifully), the next stage is LH2 based, and having a bigger diameter will help to reduce the total height of the rocket. LH2 requires lots of volume.

For the tank to hold LH2, the best one is the basic “Tank [Procedural]” with the Tank Type set to BaloonCryo. This is important as it will reduce Dry Mass. Also set Utilization to 100% and MLI Layers to 10 (maximum). MLI Layers will help reducing boil-off.

Under the tank, a RL10 engine configured as CECE-Base. My tank is 2.5x2m and gives me 6,212m/s. I plan to use around 2km/s from it for circularization. So the stage bellow needs around 8km/s of DeltaV.

Be careful, the RCS has staging in Realism Overhaul. We want them to be staged when this engine is firing. Also we need the fairing to be staged too. With this, we’re ready for the last stage.

For this, I’m going to use just the stock 3.75m fairing to cover it, and the same tank bellow, configured in the same way, BalloonCryo. And oh, I forgot, we need radiators to cool down the tank, two Radiator Fins is more than enough. I’m counting on the batteries and solar panels of the satellites to do the job. I was able to hide the radiators under the fairing.

Again, because we’re using LH2, I’m using wider tanks to avoid having a super tall rocket. It’s easier to design, but keep in mind that thinner rockets have less drag. The tank is 3.75x13m, and gives 8,027m/s of DeltaV.

For the engine, I’m using a LR87-LH2 with the SustainerUpgrade. It’s not a good engine, but it’s lightweight and provides a good amount of thrust. For bigger launches I do prefer the RS-25, but in this case it’s too much thrust, and too much weight.

Add the TT18-A Launch Stability Stabilizers, but remember: The engine needs to start first, then the de-clamp of these should happen seconds after. So have different stages and start the engine first, or the rocket may fall.

17,453m/s of DeltaV is quite a lot. And for sure more than we actually need. But let’s play easy and safe. Probably the same rocket can be used to deploy antennas on the moon if done properly, but it’s hard.

I also added a Big S-Wing Strake for stability when doing turns.

Launching the rocket

This is quite a science on itself. First give the engine a few seconds to start before releasing. Then you’ll notice that it’s slow. This is intended, if it exits too fast, it means that at the end of the stage the TWR will be too much. So we need to make it start slow. This can be simplified further with boosters to give it a small punch on the beginning.

Wait to reach 125m/s and give it a few touches to the right (around 80-85 degrees east), Once prograde moves to 80º, switch to prograde in surface mode, and let the rocket follow.

When apoapsis reaches 100km, switch to orbital mode (this is around 50km of altitude). Shortly after, apoapsis will reach 140km, switch back to manual and point to 0º pitch. Wait for the engine cutoff.

Try to stage fast and ignite the second stage as soon as possible. If you don’t do it fast enough, the fuel will bounce and the engine icon will turn red, igniting in this state will fail and you’ll need to push the ship forward using RCS before igniting the engine again.

Look on the time to apoapsis, and throttle down until it doesn’t raise anymore. We want to circularize, not increase the apoapsis even more. Be careful, if you throttle down back to 0%, you’ll cut off the engine and you’ll have to do again the start-up process I just described.

This is a good moment to start the radiators and deploy one antenna.

Remember, while you’re not touching anything, you can hibernate the cores. Click wake before having to do anything to it. This will save a lot of power while it’s not in use.

And this is it! Now use the stage in orbit to keep placing the satellites spaced as we usually do in stock. Decouple manually, because they’re out of order. Extend at least one antenna and both solar panels.

(It seems 2 small solar panels are not able to gather enough power for this probe. You might want to try bigger ones or 4 of them to avoid running out of battery. Anyway, they can hibernate at any time, but I’m not sure if they will relay properly)