Why does the game hate middle stages?

[Fraktal]

Like the title says. Every time I take a rocket made up of 1.25m parts and try to squeeze in a middle stage, I end up with not only a massive loss in TWR but a net loss in dV as well, compared to me just giving the middle stage's fuel to the first stage. What's even the point of upper-atmosphere-but-not-yet-vacuum engines like the Swivel if the game punishes me for wanting to use them for anything other than giving the first stage gimbal without movable fins?

[Vanamonde]

Not sure what you mean, but as a rule, do not add anything to the upper stages. Start with the necessary payload for the mission, then add stages below it until it can reach the destination. Adding anything to upper stages reduces the productivity of lower stages and can be a net loss. To make the ship go farther, only ever add to lower stages and remove from upper stages. 

[Grogs]

56 minutes ago, Fraktal said:

Like the title says. Every time I take a rocket made up of 1.25m parts and try to squeeze in a middle stage, I end up with not only a massive loss in TWR but a net loss in dV as well, compared to me just giving the middle stage's fuel to the first stage. What's even the point of upper-atmosphere-but-not-yet-vacuum engines like the Swivel if the game punishes me for wanting to use them for anything other than giving the first stage gimbal without movable fins?

I think the answer is in what you wrote. Instead of pushing two fuel tanks and the upper stage, the first stage now has to push the extra mass of the second stage engine as well. That means a lower TWR at liftoff -> more losses to gravity during the initial upward push. If the extra mass of the second stage engine is large enough, then you end up worse off.

For 1.25-m parts, I usually build outward with solid rockets for the launch stage and then use the middle stage to finish the orbit. Once I have larger tanks and engines, then it makes more sense to have multiple stages in the center, e.g., a 2.5-m tank/engine on the bottom, then a 1.875-m stage, and finally a 1.25-m stage at the top.

[Snark]

2 hours ago, Fraktal said:

Like the title says. Every time I take a rocket made up of 1.25m parts and try to squeeze in a middle stage, I end up with not only a massive loss in TWR but a net loss in dV as well, compared to me just giving the middle stage's fuel to the first stage. What's even the point of upper-atmosphere-but-not-yet-vacuum engines like the Swivel if the game punishes me for wanting to use them for anything other than giving the first stage gimbal without movable fins?

Not sure what you're talking about-- I use 'em all the time and they work great for me.

The math does matter, though.  Not sure how you may be Doing It Wrong™... could you provide a screenshot of a typical attempt at a ship of yours?  It's a lot harder to answer "what am I doing wrong" questions if we can't actually see what you're doing. 

[HvP]

This depends in large part on how you plan your delta-v expenditure.

Personally, I like to prevent space trash from staying in orbit so I design spent stages to fall back into the atmosphere or crash into the Mun/Minmus. This means that a stage large enough get me to the Mun will be over half empty if I also used it as the second stage after booster separation. It would also have needed a more powerful engine than is necessary for a transfer stage, and I don't want to carry all that to the Mun with me if I don't have to. And I REALLY don't want it running out of fuel on the way to the Mun where I'd have to leave junk floating around on an eccentric orbit.

 My solution is to have a sustainer stage with about 1000-dv to finish getting me out of the atmosphere that is calculated to be empty about at about 100 to 200-dv short of orbit. Then I can drop that stage back into the atmosphere and finish circularizing with what will now by my transfer stage - with an even lighter and more efficient engine. That transfer stage is basically used only to round off the last leg of Kerbin orbit, and then get me all the way to the Mun or Minmus where I will either use it to also circularize there or separate it to crash into the moon's surface.

Edited May 31, 2020 by HvP

[Reactordrone]

It's very much going to depend on the mass of you vehicle. If you're making an 18t ship another 1.5t of engine is going to reduce the delta v. The first stage has to be designed to account for the second stage and third stage mass and you're going to end up with a larger vehicle to have the same delta v or more delta v.

[Fraktal]

So middle stage means even bigger first stage that could lift the same amount of mass even without that middle stage.

[Jacke]

Thinking back to my memory of astronautics texts from the 1960's as well as other things I've picked up, there's a few rules of thumb:

1. Optimized staging tends to having the same delta-V per stage.
2. All this interacts with choosing the ascent profile for best performance.
3. Each stage needs enough angular authority throughout the stage to keep the rocket guiding.
4. The 1st stage needs a TWR of about 1.25 or better.
5. Later stages can get away with a TWR of 1 or even a bit less.
6. But lower TWR can mean the burn time gets too long and causes trajectory issues like getting too near or passing apoapsis.
7. Adding stages adds fixed mass for engines and separation that will offset any gain from adding a stage to the point of being worse.
8. Adding mass to stages has greater impact on overall performance the higher the stage it's added to, offsetting any gain.

What this means is for smaller KSP rockets that orbit a payload (thus about 3400m/s total delta V), especially those from the 1.25m parts, the advantages of making a 3-stage launch vehicle are often offset by other factors.  Even suboptimal (in the sense of #1) 2-stage LVs can have better performance because of the effects of #6 to #8.

Because of this, for smaller LKO launch vehicles, more often than not, I've made them 2-stage, usually having the payload's mission stage complete orbit circularization.

[Spricigo]

15 hours ago, Fraktal said:

 What's even the point of upper-atmosphere-but-not-yet-vacuum engines like the Swivel if the game punishes me for wanting to use them for anything other than giving the first stage gimbal without movable fins?

Well, for me the whole point of the Swivel is "because the Relliant is not unlocked yet" but it is not  an "upper-atmosphere-but-not-yet-vacuum engine" since upper atmosphere is "pretty much vacuum" anyway. Take a look at how the stats of a vacuum engines change with altitude and you see it. 

As an example, command pod+crew cabin as payload, a FL-T400 fuel tank and a TLv-909 Terrier (4.6 total mass and 2,6 total mass):

• Sea level, Thrust 14.78kN , ISP 85s, TWR0.33, dV477m/s
• 10km,  Thrust 51.51kN, ISP 296s, TWR1.14, dV1662m/s
• 20km,  Thrust 58.85kN, ISP 338s, TWR1.31, dV1899m/s
• Vacuum,  Thrust 60kN, ISP 85s, TWR1.33, dV1936m/s

Now, when you design a vessel, do it in the reverse order of the use. Let's say is a ship to send 3 kerbals to orbit and back, design the part that will come back to kerbin (crew space, parachutes, heat protection,...), then the part that will make orbital maneuvers and deorbit (a small engine, some fuel tanks, RCS, electrical parts, science,..)  and so on until the last thing you design is the stuff need right at the launch pad.

Given some room for error, design each stage to be just capable enough to delivery the craft to where the next stage is supposed to assume. If your upper stage can  take over at 20km and 1500m/s,  don't build a "middle" stage that can place it on orbit, if you just need a pair o SRB to getting you "middle" stage of the launch pad don't build an unnecessary "lower" stage and, going back to the Swivel and why I consider it pointless, don't use a crappier engine with gimbal and movable fins if you can just point the rocket in the right direction at the launchpad and let it fly itself to space (I design my rockets to be launched that way, plenty of examples on my kebalX profile)

 

[Jacke]

19 hours ago, Fraktal said:

What's even the point of upper-atmosphere-but-not-yet-vacuum engines like the Swivel if the game punishes me for wanting to use them for anything other than giving the first stage gimbal without movable fins?

I missed the comment on the Swivel.  As @Spricigo pointed out, it's not an only-near-vacuum engine.  It's intended for larger rockets using many 1.25m engines, either in the 2nd stage possible with very large designs (especially many 1.25m tanks clustered or Making History engine plates on larger tanks) or the sustainer stage of a rocket with solid or liquid fueled strap-on boosters.

[Fraktal]

51 minutes ago, Jacke said:

or the sustainer stage of a rocket with solid or liquid fueled strap-on boosters.

That's what I generally use it for, actually: first stage boosters are radially mounted onto the second stage as core to make the rocket shorter and less flexy.

It usually works but this time, I was trying to restrict myself to a 2 Thumper + 2 Reliant + 1 Swivel lower stage, a 2 Swivel middle stage and a 2 Terrier upper stage, plus lander (which has two Terriers mounted radially to make the lander shorter and thus less likely to tip over if landing on a slope). This configuration only gave me around 6.5k m/s and adding on more fuel to the lower stage was no-go because it already had a 1.7 takeoff TWR without payload, 1.3 with the lander, so even more mass would just result in more gravity losses.

Edited June 1, 2020 by Fraktal

[The Aziz]

I see we talking mostly about 1.25m rockets, perfect, it's the best size.

Okay so, for me things are quite simple: if the payload is based on 0.625 parts OR if it's 1.25 like single mk1 capsule, I have second stage with the Terrier, it usually have enough power to carry on once the first stage drops around 15km. It's worth mentioning that I use Swivels for the first stage exclusively. I don't use reliants at all, too hard to steer, and I don't use fins.

Now if the payload is bigger, and the Terrier isn't powerful enough (eg when I need to get that mk1 capsule much higher than usual) I switch the Terrier with Swivel. But then the Swivel on the first stage isn't enough, since the upper part of the rocket is much much heavier. So to not overdo with the power, I make the first stage shorter, but add side boosters, just like Delta VI/Falcon Heavy do, copies of the core stage. And that do the thing.

This thing can flyby Moho, and still has plenty of d-v for possible future maneuvers.

 

I used SRBs here for first stage help, tiny Mun lander, still powered by Swivel+Terrier set.

 

This one's got a bigger and heavier upper stage, hidden in a fairing, of course with Terrier. Launches with three Swivels.

And all tthese rockets are dirt cheap, if you take off the science equipment, you can have a Duna orbiter in less than 10,000

Edited June 1, 2020 by The Aziz

[TheFlyingKerman]

To directly answer the OP directly, a LFO stage should have around 3000 DV to be efficient. It takes 3400 DV to go to LKO so there is no point squeezing too many stages.
For my standard single use LFO lifter I just use 1 Swivel all the way to orbit and 4 Thumpers to get it off the pad, that's enough for 7.5t of payload.

 

[Jacke]

42 minutes ago, TheFlyingKerman said:

...a LFO stage should have around 3000 DV to be efficient.

Your reasoning to support this?

[TheFlyingKerman]

Just now, Jacke said:

Your reasoning to support this?

Empirical rule of thumb. 

[Jacke]

13 minutes ago, TheFlyingKerman said:

Empirical rule of thumb. 

Then please show examples.  I don't think it quite applies.  I've mostly made 2-stage launch vehicles to LKO, depending on the payload with the 2nd stage circularizing or the payload mission stage doing it, sometimes split between them.  The division of delta-V between those 2 stages has varied widely.  For rare light payloads, the 2nd stage is also the mission stage.

[TheFlyingKerman]

Examples:

One Skipper as core stage, 2 SRBs. Rated for 28t to LKO, launch cost 21000 funds

http://www.imagehousing.com/image/ksp/1456577

 

This variant is closer to single stage, with the Hammers just getting the rocket off the pad. It is also fully reusable. Rated for 9.7t, launch cost 3200 funds

http://www.imagehousing.com/image/ksp/1456576

 

[Streetwind]

@Jacke @TheFlyingKerman There is actually proper math behind that "empirical rule of thumb". It's called the mass fraction. This is a number you can calculate for any given rocket stage, and if you multiply it with your effective exhaust velocity (specific impulse x standard gravity), you get your dV. Now, increasing the mass fraction is subject to diminishing returns for reasons of dry mass being a thing that exists. So there comes a point where trying to add more duel to a given stage becomes impractical. At the same time, you want your money's worth out of your engine, so adding too little fuel is also bad. And that is why you end up with a "healthy middle ground", from which derive a number of rules of thumb for sensible rocket building.

This is the TL;DR version of a wall of text I once wrote on this topic, which you can find here in case you're curious for more.

Edited June 2, 2020 by Streetwind

[Jacke]

28 minutes ago, Streetwind said:

@Jacke @TheFlyingKerman There is actually proper math behind that "empirical rule of thumb". It's called the mass fraction.

I remember reading about that as well.  I couldn't recall the formulas, but I thought for the case of KSP it wouldn't drive the range of efficient high enough to include delta-V as high as 3000m/s.  But in retrospect, I think the amount of decline going to 3000m/s wouldn't be a whole lot.

KSP parts are worse than the real world for having a larger fixed (engine and separation) and relative (tankage) dry mass for the amount of propellants in a stage.  The first increasing usually leads to wanting greater delta-V per stage, the later increasing wanting more stages.

So, depending on the vagaries of KSP parts' performances, masses, and funds costs (which as far as I know are still weird in many places), you're going to find different best designs.  But for the usual size of early career 1.25m-part stages, it's going to mostly be 2 or 3-stage launch vehicles, including the mission stage of the payload.  Lower payload will really push the 2-stage designs.  Higher payload, due to the greater dry mass of tankage, will likely be better as 3-stage designs.  It would need a massive launch vehicle to make a 4-stage design worth it.

Edited June 2, 2020 by Jacke

[Streetwind]

Yeah, that lines up with my experience. I actually launch single stage to orbit in early career/science mode a lot, when all I need to carry is a pod and some science instruments. Means I can delay spending science on the Terrier too, perhaps getting something more important before it. With two stages, I can do Mun and Minmus flybys. Three stages to land on them - maybe four to return, as having a detachable ascent stage a la Apollo is usually the most convenient route.