Shed engines with fuel tanks or keep them during launch?

[Kelderek]

Is it more efficient to shed both engines and fuel tanks in the early stages of a launch or just fuel tanks alone, keeping all of the engines (and their thrust) until orbit?

Let me give an example of what I mean. Let's say I have a central stack with an engine at the bottom and also 4 smaller stacks in a 4-way symmetry formation along the sides (2 pairs of 2) and each of those has its own engine too. The way I usually see this is with asparagus fuel lines and decouplers so that you eject the first pair of side stacks, then later on the second pair, leaving only the middle stack for the remainder of the ascent. With this design, you will lose thrust at each stage because you are losing the engines and the fuel tanks together.

Now, alternatively, if I were to move the 4 engines from the side stacks and attach them directly to the center stack (separating them from the side fuel tanks and using fuel lines to stay functional), then as I eject the side fuel tanks I would keep the same amount of thrust all the way to orbit. I would still use the usual asparagus setup on the side fuel tanks. The trade-off is that I keep the same constant amount of thrust all the way to orbit, but I also keep the mass of those side engines all the way too. Also, with more engines active during the entire ascent, I would burn through fuel quicker. The real question is whether the additional thrust would make up for keeping the mass and burning the fuel faster.

I'm not sure if being more specific about certain engine and fuel tank parts matters, but it might. I was mainly wondering what works best in a general sense.

Edited December 12, 2014 by Kelderek

[Taki117]

It is best to drop both engines and fuel tanks, this way you don't have an excessive TWR as you burn fuel, but instead your TWR remains about constant (if planned correctly)or drops off late enough that it doesn't matter.

[Kelderek]

That makes sense. After I made my post I made a couple of test examples in the VAB and was able to verify that they have different Delta-V values with the same number and type of parts attached.

Javascript is disabled. View full album

[Starman4308]

Is it more efficient to shed both engines and fuel tanks in the early stages of a launch or just fuel tanks alone, keeping all of the engines (and their thrust) until orbit?

Like almost everything with ascent, the answer is "it depends".

Delta-V favors shedding every kilogram of mass as soon as you possibly can. However, a higher TWR lets you get to orbit with less fuel/dV wasted to atmo/gravity*. The mass/cost-optimal solution will thus be a matter of trial-and-error, trying to find the right amount of engine (and how best to arrange engines/tanks) to get you to orbit with the lowest cost or mass. The more engine, the less dV is wasted, but the less engine, the more dV you have to start with.

*Within certain constraints, such as vertical ascents staying at or below terminal velocity.

In practice, you try to start with a TWR around 1.6-2.0, with upper stages starting at 0.8-1.0 TWR, but these are more suggestions than hard-and-fast rules.

Edited December 12, 2014 by Starman4308

[Kelderek]

Delta-V favors shedding every kilogram of mass as soon as you possibly can. However, a higher TWR lets you get to orbit with less fuel/dV wasted to atmo/gravity*. The mass/cost-optimal solution will thus be a matter of trial-and-error, trying to find the right amount of engine (and how best to arrange engines/tanks) to get you to orbit with the lowest cost or mass. The more engine, the less dV is wasted, but the less engine, the more dV you have to start with.

This is exactly how this all got started for me. I think I had a few designs where the TWR for the center stack alone was low enough that it felt really slow and sluggish after I had decoupled all of the auxiliary tanks and engines. I assumed I needed more thrust for that stage, but it also could mean that my auxiliary tanks were too small and needed to last longer. I'm still very new to this, so I also might have had my gravity turn not exactly ideal too.

[SRV Ron]

This is exactly how this all got started for me. I think I had a few designs where the TWR for the center stack alone was low enough that it felt really slow and sluggish after I had decoupled all of the auxiliary tanks and engines. I assumed I needed more thrust for that stage, but it also could mean that my auxiliary tanks were too small and needed to last longer. I'm still very new to this, so I also might have had my gravity turn not exactly ideal too.

Resist the urge to pile too much fuel into the boosters. You will only end up making huge rockets that have no additional advantage in delta v.

Example, this cluster rocket;

After staging while all rockets, including the core, are under power, there is no difference in performance.

However, if properly balanced to engines under asparagus staging, the launch to payload in orbit ratio can become very efficient.

[Red Iron Crown]

From a delta-V and payload fraction perspective it is better to drop tanks and engines, shed every gram of unnecessary mass as soon as possible.

From a financial perspective it is better to shed nothing and recover the entire craft intact.

[capi3101]

This is exactly how this all got started for me. I think I had a few designs where the TWR for the center stack alone was low enough that it felt really slow and sluggish after I had decoupled all of the auxiliary tanks and engines. I assumed I needed more thrust for that stage, but it also could mean that my auxiliary tanks were too small and needed to last longer. I'm still very new to this, so I also might have had my gravity turn not exactly ideal too.

I'ma gonna point you to an oldie but goodie, Temstar's '>Asparagus design philosophy. Yes, that was under 0.20 (0.19?), but the general principles are still valid - and these days they're far easier to implement (no thrust limiters back then). What his philosophy boils down to is having a 1.6-1.7 TWR at launch and having a core stage that's more powerful than any of the individual booster engines (he uses a 22%:13% core-to-booster engine thrust ratio for three booster pairs or 22%:9.75% for four booster pairs). Each stage gets up to 2.2-2.3 TWR before it's discarded, dropping the TWR back down towards 1.5-1.6; the final core stage is around 1.3 when it finally has to start working on its own. I've used Temstar's philosophy for a long time now, and it's almost always brought my payloads to orbit successfully (when it doesn't, it's almost always because I've screwed up the piloting).

You always need the most thrust at launch. Thrust becomes less and less critical as you ascend and you never want too much.

Edited December 12, 2014 by capi3101

[Daid]

From a financial perspective it is better to shed nothing and recover the entire craft intact.

Also if that uses more fuel? At some point the fuel costs of recovering everything must outweigh the recovery "profit" right?

[Red Iron Crown]

Also if that uses more fuel? At some point the fuel costs of recovering everything must outweigh the recovery "profit" right?

Not as the game is currently balanced. Fuel is cheap compared to tanks and engines, and Kerbin's requirements to orbit are low enough that an SSTO can deliver a good enough payload fraction even if not using jets.

[Starman4308]

Also if that uses more fuel? At some point the fuel costs of recovering everything must outweigh the recovery "profit" right?

Unlikely.

As mentioned, in KSP, fuel is cheap: you can use a huge amount of fuel for an SSTO and still come out with a big profit. In real life, fuel is even cheaper: Elon Musk estimated that fuel accounts for ~0.3% of launch costs.

[Guest]

That's actually pretty realistic, too. Fuel is cheap IRL as well. While recoverable SSTOs have not been made yet, there are rockets that use "single stage to almost-orbit" philosophy to avoid expanding more than one engine. This lets them use a (very cheap) kick motor for final insertion while not littering orbit with large fuel tanks.

[Kelderek]

I'ma gonna point you to an oldie but goodie, Temstar's Asparagus design philosophy.

Thanks for that link, I found that very helpful.

All of this makes me think of a second question I had: how viable is it to recover spent stages? Can you put parachutes on them and recover them that way? Isn't that how the real space shuttle launches worked? I thought that the main fuel tank and SRBs were fished out of the Atlantic and re-used. Or does the way the game handles debris cloud the issue? I expect you would have to semi-deploy parachutes really high up and that may cause them to fail.

[cantab]

In the stock game anything that drops below about 22 km when you aren't flying it is deleted. So you have two options:

Design your rockets so you can fly the dropped stages down while the final stage minds its own business. "Single stage to almost-orbit" as Dragon01 mentions is what you need here, you want to get the payload in a stable orbit before the boosters drop too low.

Use a mod. StageRecovery will give you your money back if you put enough parachutes on. FMRS will let you actively fly your boosters down so will work for winged or powered-landing designs.

[Norcalplanner]

Kerbal Construction Time also works well, and I think it's written by the same author as stage recovery. With KCT, it makes even more sense to recover boosters, as the recovered parts can be used to greatly speed up construction of the next rocket.

[Torquemadus]

In the stock game anything that drops below about 22 km when you aren't flying it is deleted. So you have two options:

Design your rockets so you can fly the dropped stages down while the final stage minds its own business. "Single stage to almost-orbit" as Dragon01 mentions is what you need here, you want to get the payload in a stable orbit before the boosters drop too low.

Use a mod. StageRecovery will give you your money back if you put enough parachutes on. FMRS will let you actively fly your boosters down so will work for winged or powered-landing designs.

My solution is to use a two stage launch vehicle. The first stage consists entirely of solids and is used to reach sub-orbit. Solid "trash bins full of boom" are worth very little when empty, so I don't worry about recovering them.

The second stage is a liquid fuelled winged fly-back stage. It is used to circularise into Kerbin orbit and also performs most of the transfer burn. After separating, the payload finishes the remainder of the transfer burn, while the fly-back stage is left in a highly elliptical orbit around Kerbin. This allows the fly-back stage to aerobrake back into LKO. From there, it deorbits and returns to the runway at the KSC.

[Kelderek]

Let me see if I have the gist of this...

Let's say I know the following as an example (using some numbers from that Temstar post):

Payload to LKO = 69 t

Desired Payload Mass Fraction of Total Vessel = 15%

Desired Launch Pad TWR = 1.6-1.7

I can calculate the total mass of my rocket:

69 x 100/15 = 460 t

I can calculate how much thrust I will need from engines to lift that mass:

460 x 1.6 x 9.81 = 7220 kN

460 x 1.7 x 9,81 = 7671 kN

So I would need to have enough engines in my center stack and boosters to give me between 7220 and 7671 kN of thrust. I would need to add enough fuel tanks and other hardware parts to bring my dV up over 4500m/s and that should be roughly 460 - 69 - engine mass. Is that right?

If you know your target thrust, then am I right in assuming that you want to achieve that thrust with the least amount of mass? For example, let's say I want 1000 kN of thrust on a booster, wouldn't I be better off using a thrust-limited mainsail engine instead of using five LV-T45 engines (6 t mass for the mainsail compared to 5 x 1.5 t = 7.5 t mass for five LV-T45 engines)? I would think that would save fuel, allowing your payload to be a higher fraction of your total rocket mass.

[SRV Ron]

That's actually pretty realistic, too. Fuel is cheap IRL as well. While recoverable SSTOs have not been made yet, there are rockets that use "single stage to almost-orbit" philosophy to avoid expanding more than one engine. This lets them use a (very cheap) kick motor for final insertion while not littering orbit with large fuel tanks.

Somewhat the idea behind the original Atlas rocket of the 60s. It used three engines for launch and dropped the two outboard ones when their thrust was no longer needed.

[Guest]

That was a bit different, Atlas was a "stage an a half" system with orbital capacity. If it wasn't for the primitive engine, it'd be a straight-up SSTO. The Atlas engines couldn't throttle (it gained that capability only after the introduction of RD-180), and full thrust with tanks nearly empty would've introduced unacceptable loads on the flimsy balloon tanks the rocket was made of. Dropping the engines was more like a poor man's throttling mechanism. It had the problem of littering the orbit (though being big and light, spent rockets decayed quickly) and wasn't very cheap, either. The reasons for doing the Atlas that way was to avoid igniting a large (turbopump-fed) liquid rocket engine at altitude, which was an unresolved problem at the time.

Atlas was an oddity, because normally, you'd want to drop spent tanks with engines (the way R-7 handled the same problem), but Atlas was made with a monocoque construction, and the tanks were very lightweight (the walls of the tank were also the hull). As such, dropping tanks designed like this wouldn't have saved enough to justify the added mass and complexity of a separation system. On the other hand, those tanks were problematic to handle and had to be kept pressurized when in storage, or they'd collapse. That's why they're not used anymore, except on Centaur (being an upper stage, it's smaller and its mass is at premium, so it's worth enduring the handling difficulties).