Launch Vehicle Optimization Test Results

[OhioBob]

sardia said:

I'm not sure what this implies for our space program. Am I just too tired to get it? What's the cheapest and reliable way to get a payload to space via rockets?

There are too many possible configurations to test them all. I personally prefer two-stage liquid fueled launchers, so that is what I tested. I wanted to see what the sweet spot was for TWR in order the get my payloads to orbit as economically as possible. Based on my tests, it looks like I should target a TWR of between 1.2 and 1.5 for my first stage, and be around 1.1 for my second stage. It doesn't look like there is any cost advantage to using solids, though I think they are a good option to augment first stage thrust if you need a little extra oomph.

None of this means there aren't better ways to get to orbit. It just means that if two-stage liquid fueled rockets is your thing, then the cheapest way to get your payload to orbit it to load up on fuel, operate at a fairly low TWR, and don't be overly concerned about Δv inefficiency.

(edit) My answer assumes you are talking about KSP and not real life.

Edited December 21, 2015 by OhioBob

[Norcalplanner]

If you mean what it implies for our *kerbal* space programs, then, well, OhioBob just demonstrated fairly well that in 1.0.4 it's no longer the case that solid+sustainer is cheaper than 2stage liquid.

[clears throat]

Allow me to respectfully disagree. The trick is to remove a lot of the expensive bits, reduce drag, and generally optimize the design.

Javascript is disabled. View full album

1,089 funds/ton in the final configuration.

[sardia]

I dunno how fair it is to strip out all those "unneeded" parts. With a low TWR, are fins entirely unneeded?

And yes, I meant KSP.

[Norcalplanner]

I dunno how fair it is to strip out all those "unneeded" parts. With a low TWR, are fins entirely unneeded?

My understanding is that radial parts, especially struts and docking ports, are something that really ups the drag in 1.0.4. NathanKell can probably elaborate a whole lot better than I can - I just like running tests in the the game. And I can't come anywhere close to OhioBob's level of analysis.

With a gimbaling engine, a proper ascent profile, and a relatively aerodynamic craft, you can safely omit the fins. The trick is making sure you never get more than a few degrees off of prograde, which means a proper gravity turn. If you look at the MJ ascent settings, you'll see that it's set to limit AoA to 8 degrees.

I seem to recall reading somewhere that if the Apollo program had continued, they were going to omit the fins from the S-IC stage because they weren't needed. More modern lifters like Delta IV and Falcon 9 also omit the fins.

[OhioBob]

Norcalplanner said:

[clears throat]

Allow me to respectfully disagree. The trick is to remove a lot of the expensive bits, reduce drag, and generally optimize the design.

http://imgur.com/a/0FR4x

1,089 funds/ton in the final configuration.

Several comments:

1) The reason you couldn't get my unaltered design to reach orbit is because a full X200-32 tank brings the payload mass to a little over the theoretical maximum. I had to offload a tiny bit of propellant to get it to reach orbit.

2) I agree the fins were going overboard. I thought they could probably be excluded but I keep them in the design only because I included them on liquid-fueled test vehicle. I have since removed them and re-ran the computer simulations. This significantly reduces the cost per tonne of payload, but it is still higher than the all liquid vehicle. (I've edited my previous post to show the new results.)

3) I prefer to keep the reaction wheels in the analysis because they were included in the liquid fueled design to which I'm comparing.

4) I didn't use "Kickback" SRBs on purpose, simply because they are the best option. I wasn't trying build a vehicle that was the pinnacle of performance, I was trying to build a vehicle that fell somewhere in the middle in terms of performance. This was a situation where a few changes resulted in a highly efficient design, but that may not always be the case.

5) When I wrote "it's also likely that a more cost effective SRB design can be found than the sample tested here", the changes I was thinking about were exactly the ones that you made.

6) The main point, I think, is that using SRBs doesn't mean that cost is automatically going to be reduced. It's going to take the right situation and a good design.

Edited December 22, 2015 by OhioBob

[OhioBob]

Something else I noticed in doing this is that Kerbal Engineer computes the combined I_sp of different engines incorrectly. I don't know what the game does, but if the game computes it the same way as KER, then there's a problem.

For example, if you look at the image in post #19 you see that the I_sp is given as 211.9 s. Of course this is a little above sea level; in the VAB the sea level I_sp is given as 211.4 s. The formula for specific impulse is,

I_sp = F / (ṁ * g_o)

where F is thrust, ṁ is the mass flow rate, and g_o is standard gravity (9.80665 m/s²). When we have multiple engines the formula becomes,

I_sp = ΣF / (Σṁ * g_o)

where we simply have to sum up the thrusts and mass flow rates of the different engines.

Let's start with the Skipper. The volumetric flow rate is given in the game as 41.426/s, which gives us a mass flow rate of 207.13 kg/s. We can also compute the mass flow rate from the known thrust and I_sp. I doesn't matter whether we use the sea level or vacuum values because ṁ is supposed to be constant and either set of values should give the same result.

568750 N / 280 s / 9.80665 m/s² = 207.130 kg/s

650000 N / 300 s / 9.80665 m/s² = 207.130 kg/s

So all the math works out and we know the mass flow rate of the Skipper engine. Doing the same thing for the Thumper SRB gives us a mass flow rate of 145.674 kg/s.

In my example I have 1 Skipper, 2 Thumper @ 100%, and 4 Thumper at 50.5%. Therefore, we have

ΣF = 568750 + 250000 * 2 + 250000 * 4 * 0.505 = 1,573,750 N (sea level)

Σṁ = 207.130 + 145.674 * 2 + 145.674 * 4 * 0.505 = 792.74 kg/s

And the combined specific impulse is,

I_sp = 1573750 / (792.74 * 9.80665) = 202.4 s (sea level)

So where does 211.4 s come from? Honestly, I don't know. I've tried some different ideas but I can't come up with that number. Just to prove to myself that this isn't some one time thing, I've also tested other engine combinations and have got similar results. The same error occurs in the calculation of vacuum I_sp. For my test vehicle, KER gives a vacuum I_sp of 248.1 s while my calculations come to 238.7 s.

If we are to believe that the I_sp figures given by KER are correct, then we get mass flow rates of,

Σṁ = 1573750 / 211.4 / 9.80665 = 759.12 kg/s (sea level)

Σṁ = 1856000 / 248.1 / 9.80665 = 762.83 kg/s (vacuum)

So much for the idea that mass flow rate is supposed to be constant. Not only do these values not equal the supposed ṁ of 792.74 kg/s, they don't equal each other. There is definitely something wonky going on here.

Edited December 21, 2015 by OhioBob

[NathanKell]

The game doesn't calculate at all, it integrates, so that won't be an issue. Worth mentioning in the KER thread so Padishar can fix it, though!

[OhioBob]

The game doesn't calculate at all, it integrates, so that won't be an issue.

Good. I was hoping it was just an issue in KER and not in the game.

Worth mentioning in the KER thread so Padishar can fix it, though!

I'm planning to do that.

[sardia]

Curse this optimization thread. I was browsing through my oldest rockets to find the one that works best for my payload when I started to notice how outdated they allk were. Unshielded unoccluded cores, struts everywhere, half a dozen fins per stage... I ended up getting lost redesigning all the rockets that I never launched my polar satellite.

[MalfunctionM1Ke]

Any suggestions on when or when to not use "moar boosters" ? Specifically soilds as I do not like to use asparagus.

I tend to use Solids-to-Space as well, especially for small payloads such as Satelites.

It might be inefficient as hell, but it is soooo cheap compared to a Liquid-Fuel-Rocket.

[GoSlash27]

OhioBob,

6) The main point, I think, is that using SRBs doesn't mean that cost is automatically going to be reduced. It's going to take the right situation and a good design.

IME a disposable LFB first stage can nearly always be outperformed by a SRB in terms of cost/ tonne. Even the KR-1x2 can't keep up with the Kickback or BACC.

In your cost- optimized example the second stage presents a payload of 64 tonnes.

Your LFB costs $33,800 according to my model.

The same job can be done by 7 Kickbacks for $18,900.

For 1800 m/sec DV you simply take the upper stage mass in tonnes and divide it by 9.23. This gives you the required number of Kickbacks. The throttle setting will generally be in the 75% range. You can stagger the throttle settings to maintain a more even thrust through the burn.

BACCs are slightly cheaper than any LFB in smaller payloads. Not a night and day difference, but they are cheaper.

IMO nearly every disposable first stage should be SRB if cost is important.

*edit* having said that, I understand that it was not your intention to create a super- economical lifter, but rather to create some data points and they are very useful. Right down the middle of where I expected them to be, so that confirms that I'm not far off- course in my designs.

Best,

-Slashy

Edited September 18, 2015 by GoSlash27

[GoSlash27]

Any suggestions on when or when to not use "moar boosters" ? Specifically soilds as I do not like to use asparagus.

abh,

Best I can tell you is this: It depends.

SRBs excel at being cheap disposable lower stages. Anything beyond that.... they're probably not your best choice.

The best way I've devised to be sure is to put together a spreadsheet like Ohio Bob has done here.

You want to take the rocket equation and turn it on it's ear. Instead of answering "how much DV will this stage produce", you want to set it up to answer " what is my wet to dry ratio with this engine for a given amount of DV".

This takes the form Rwd= e^(DV/9.81Isp) where

e= 2.718

DV is your desired DV

Isp is the mean Isp of your engine in the given situation.

You apply some algebra-fu to work out fuel, tanks, engine mass, and payload to this wet to dry ratio, then you estimate the cost to build the stage.

If you do this with all engines simultaneously, you will see how much each proposed stage will weigh and cost and will be able to pick the best option. Upper stages place a premium on weight, while lower stages place a premium on cost.

tl;dr:

For cheep thrillz, the answer is "never". SSTO fully recoverable lifters are always cheaper than staged disposables in KSP and SSTO air breathing spaceplanes are always the cheapest option.

Best,

-Slashy

[OhioBob]

GoSlash27 said:

IMO nearly every disposable first stage should be SRB if cost is important.

I agree that using solids in-line as a first stage can be very economical. I do this frequently for small payloads where I can use a RT-10 or BACC. I have also sometimes clustered 2,3 or 4 SRBs together using stack adapters. Using them radially, however, appears to negates most, if not all, of the cost advantage. The problem is in the way the costs are balanced. A BACC cost only 850, while a TT-70 decoupler and advanced nose cone together costs 1020. Even a cheaper TT-38K decoupler and aerodynamic nose cone costs 840. That's just crazy out of balanced. There is no way the decoupler and nose cone should cost as much or more than the SRB. It's even worse for a RT-10, where the cost of the accessories is more than double the SRB cost. The tests we've performed in this thread seem to show that the only SRB to provide a definite cost advantage when radially attached is the Kickback. More tests are probably needed to confirm, but right now I think the rules for SRBs use might be:

1) In-line as a first stage = YES

2) RT-5, RT-10 & BACC radially attached = NO

3) Kickback radially attached = YES

By the way, note that in the post where I reported my launcher w/SRBs results, I wrote

OhioBob said:

If we add up the cost of just the propellant, tanks, engines and SRBs, then the SRB launcher is less expensive - 917/t vs. 1035/t.

So I was indicating that SRBs are cheaper if we can eliminate the costly accessories. I never intended to suggest that SRBs are never economical.

 

Edited December 22, 2015 by OhioBob
fixed formatting

[GoSlash27]

I agree that using solids in-line as a first stage can be very economical. I do this frequently for small payloads where I can use a RT-10 or BACC. I have also sometimes clustered 2,3 or 4 SRBs together using stack adapters. Using them radially, however, appears to negates most, if not all, of the cost advantage. The problem is in the way the costs are balanced. A BACC cost only 850, while a TT-70 decoupler and advanced nose cone together costs 1020. Even a cheaper TT-38K decoupler and aerodynamic nose cone costs 840. That's just crazy out of balanced. There is no way the decoupler and nose cone should cost as much or more than the SRB. It's even worse for a RT-10, where the cost of the accessories is more than double the SRB cost. The tests we've performed in this thread seem to show that the only SRB to provide a definite cost advantage when radially attached is the Kickback. More tests are probably needed to confirm, but right now I think the rules for SRBs use might be:

1) In-line as a first stage = YES

2) RT-5, RT-10 & BACC radially attached = NO

3) Kickback radially attached = YES

By the way, note that in post where I reported my launcher w/SRBs results, I wrote

So I was indicating that SRBs are cheaper if we can eliminate the costly accessories. I never intended to suggest that SRBs are never economical.

OhioBob,

Oh, I realize that was never your intent and I agree with all of this. There was just some confusion (not on your part) about whether SRBs were cheaper or more expensive than LFBs and I was trying to clear that up. I think your response here does a much better job of that than mine did.

*edit* to clarify, SRBs can be surface attached to each other or a central part, so radial decouplers (or stack adapters) aren't actually necessary. For example:

And I've found it's rare that the RT-10 can outperform LFBs even in this manner.

Best,

-Slashy

Edited September 19, 2015 by GoSlash27

[OhioBob]

*edit* to clarify, SRBs can be surface attached to each other or a central part, so radial decouplers (or stack adapters) aren't actually necessary.

Yes, I've though of that. As long as they are clustered together to work as a single unit, the method you illustrate looks like a very inexpensive option. However, I have a couple concerns. First, the 1.25m attachment point looks awful flimsy. Would you typically add struts to strengthen the connection? Second, how does something like that steer without having a gimbaled engine? Are the reaction wheels enough, do you add fins?

[GoSlash27]

Yes, I've though of that. As long as they are clustered together to work as a single unit, the method you illustrate looks like a very inexpensive option. However, I have a couple concerns.

First, the 1.25m attachment point looks awful flimsy. Would you typically add struts to strengthen the connection?

Nah. It's actually quite rigid; way less flimsy than the LFB, since the thrust vectoring tends to make it oscillate.

Second, how does something like that steer without having a gimbaled engine? Are the reaction wheels enough, do you add fins?

The reaction wheel is adequate. Adding fins would tend to make it overly- stable. That's also why I don't add nose cones to the boosters.

It is important to dial back the thrust to an appropriate level. Otherwise you can lose steering authority.

Best,

-Slashy

[Frozen_Heart]

SRBs can be made inline and pretty cheap by just using the quad adaptor. 4 of the largest ones packs a lot of power for little cost that way.

[selfish_meme]

They also clip together around a single one, I did that with my Ares 1 replica, 9 Kickbacks to provide enough thrust for the stack.

[More Boosters]

Kickbacks are downright horrible, I can't quite put my finger on what's wrong with them but there's no way they're strong enough to be the last SRBs you unlock. Guess it's thrust. They lack thrust. I'd rather get a good 2.0 TWR with 800 m/s Delta-V than be forced to place 4+ Kickbacks and have extra Delta-V I don't want just to reach the TWR required.

[GoSlash27]

Kickbacks are downright horrible, I can't quite put my finger on what's wrong with them but there's no way they're strong enough to be the last SRBs you unlock. Guess it's thrust. They lack thrust. I'd rather get a good 2.0 TWR with 800 m/s Delta-V than be forced to place 4+ Kickbacks and have extra Delta-V I don't want just to reach the TWR required.

More Boosters,

Keeping true to your name, I see

The Kickback is definitely the best SRB in the game (and for what it does the best engine in the game). It has the lowest cost per payload tonne and highest specific impulse. As for it's raw thrust, it has more than enough for it's intended role. If you want something disposable that will get a payload halfway to orbit for cheap, the Kickback is unbeatable.

2G off the pad and only 800 m/sec DV is an odd requirement. Most designers want a much lower initial thrust (it's going to get hectic as the fuel drains) and more DV. I'd imagine you could unload propellant to do that, but it kinda ruins the entire point of the Kickback.

Best,

-Slashy

[sardia]

OhioBob,

Oh, I realize that was never your intent and I agree with all of this. There was just some confusion (not on your part) about whether SRBs were cheaper or more expensive than LFBs and I was trying to clear that up. I think your response here does a much better job of that than mine did.

*edit* to clarify, SRBs can be surface attached to each other or a central part, so radial decouplers (or stack adapters) aren't actually necessary. For example:

http://i52.photobucket.com/albums/g13/GoSlash27/22t_zpsgltbnjln.jpg

And I've found it's rare that the RT-10 can outperform LFBs even in this manner.

Best,

-Slashy

I tried attaching the 4x SRBs to the adapter you used, and to a science module radially, but it was super noodly and fell apart. Can you elaborate on how it's suppose to look? How did you get it so strong?

[GoSlash27]

I tried attaching the 4x SRBs to the adapter you used, and to a science module radially, but it was super noodly and fell apart. Can you elaborate on how it's suppose to look? How did you get it so strong?

sardia,

I attached the adapter to the stack decoupler, then the center booster to the adapter. After that, I attached the remaining boosters to the center booster.

I'd need to see a pic of your setup before I could tell you why it's flimsy.

Best,

-Slashy

[OhioBob]

I attached the adapter to the stack decoupler, then the center booster to the adapter. After that, I attached the remaining boosters to the center booster.

Slashy,

What do you have the SRB thrust limiter set to?

[GoSlash27]

Slashy,

What do you have the SRB thrust limiter set to?

OhioBob,

I'll have to verify it after work, but iirc the center booster is set to 100%, 2 radials are set to 75% and the remaining 2 radials are set to 60%

*edit* Just checked.

The center is set to 100.

One pair is at 75 and the last pair is at 50.

http://wikisend.com/download/311120/22t test.craft

Best,

-Slashy

Edited September 21, 2015 by GoSlash27

[sardia]

Ah, there was a central booster, so it's really a 5 pack. What's the point of the central adapter than? To provide occlusion? Why not just do a central SRB surrounded by SRBs and have that central SRB decoupled?