Designing Launch Vehicles

[Mad_Maelstrom]

I'm curious about what approach people take to designing their launch vehicles. I personally have 3 criterion I like to constrain.

1) 2-stage to orbit design (this is personal preference, primarily for simplicity)

2) Delta V to orbit of >=4500 m/s (this is based on what I have found online and experience. This assumes SL Isp for the 1st stage and the avg between SL and vac Isp for the 2nd stage)

3) A TWR between 1.5 and 2 at the firing of each stage (A TWR of 2 is almost perfect because it balances gravity loss and air drag losses. I like to start a little lower than 2 because the TWR will rise as the vehicle burns fuel)

In addition, I use a spreadsheet to try to pinch-pennies when I'm in career mode. The spreadsheet allows me to consider many different designs of 2-stage launch vehicles rapidly. The spreadsheet only works for liquid fueled rockets and only handles parallel staging with fuel lines (onion staging) or series staging. A link to the excel spreadsheet is here.

Let me know what you think of the spreadsheet and what approach you take when designing a launch vehicle.

Edited January 19, 2015 by Mad_Maelstrom
repair images

[Sovek]

My approach?

Priorities are realistic, and no orbital trash floating around unless I screw up. Since I run with FAR I need to make sure the rocket is stable. Shorty stubby rockets are very, VERY unstable, if its short, add stability fins. As far as T:W, I look for about 1.3, with KIDS (Kerbal ISP Difficulty Scaler, set for a mostly stock but thrust varies with ISP so I have to be careful on the ground. 1.3 and the rocket probably wont lift off the pad), 1.5 T:W is the highest I'll go. I'll use a single or dual stage as needed. For instance my Quackshot rockets in this current playthrough can just about get to orbit on the first stage with the second stage to boost the Pe above the atmosphere and transfer to the Mun and Minmus, and its final mission, escape Kerbins SoI for a bit. Buckshot rockets though have a two stage to transfer because of the weight. 2.5m parts get heavy pretty quick.

Other concerns are as mission dictates.

[Solivagant]

I don't use spreadsheets or back of the napkin calculations, I just fiddle with something that seems good, then find its upper limit and spend a few hours tweaking it to be optimized. My criteria are 2 stages only, not counting boosters, of which I usually use liquid fuel for and there can be no more than 4 boosters.Another must-have is that it has to look nice and realistic. There's nothing wrong with rockets wider than they are tall, but I like sleek, SLS-style stuff with nosecones. Additionally, I'll be set when the aerodynamics overhaul comes.

[Budgie]

Honestly, aesthetics matter to me too. Sure, I could design a cylinder with a cone on the end, but there's no fun it that. I first make a rocket design that I like the look of, then tweak and tweak and change and change until the mission parameters are met. In the end, the good-looking rockets often work out well.

[OhioBob]

Let me first stipulate that I use the stock game, i.e. no FAR/NEAR. The rules of thumb that I use that have given me good results are:

1) Two stages to get to LKO (additional stages on top of that for whatever my mission is)

2) Vacuum ÃŽâ€V of 4550 m/s (actual ÃŽâ€V about 4400 m/s)

3) Stage 1 TWR at liftoff ≈ 1.65

4) Stage 2 TWR at ignition ≈ 1.30

5) Ratio of Stage 2 thrust to Stage 1 thrust ≈ 0.35

Using these guidelines I typically get a payload fraction of about 0.16. I also frequently use strap-on SRBs to get my liftoff TWR right.

(edited to add)

Now that I’ve given you my rules of thumb, I might as well show how I put them to use to design a launch vehicle. This might serve as a tutorial for new or inexperienced users who may be struggling to design a good rocket.

If you always use these guidelines you’ll produce launch vehicles that perform to a fairly consistent standard. I’ve done this enough times that I’m confident in the 0.16 payload fraction. So let’s say I want to design a rocket to launch a 20-tonne payload. My total rocket mass will be approximately,

20 / 0.16 = 125 t

Since I want a Stage 1 TWR of 1.65, my first stage thrust should be,

125 * 9.81 * 1.65 = 2023 kN

Therefore I choose to use a LFB KR 1x2 (2000 kN).

My ideal Stage 2 thrust is,

2000 * 0.35 = 700 kN

I choose the engine the gets me closest to this thrust, which is the Rockomax Skipper (650 kN).

Since I want a Stage 2 TWR of 1.30, the mass of the second stage and payload should be,

650 / 9.81 / 1.30 = 51 t

Therefore the ideal Stage 1 mass is,

125 – 51 = 74 t

Since my first stage already consists of a LFB KR 1x2 (42 t), and I require a decoupler (0.4 t), I can add the following in the form of additional tanks and fuel,

74 – 42 – 0.4 = 31.6 t

This is done by adding one X200-32 (18 t), one X200-16 (9 t) and one X200-8 (4.5 t).

I now must figure out what fuel tanks to add to the second stage. I have 51 t to work with, from which I subtract the engine (3 t), decoupler (0.4 t) and payload (20 t),

51 – 3 – 0.4 – 20 = 27.6 t

So I add one X200-32 (18 t) and one X200-16 (9 t) fuel tank.

My launch vehicle is now complete. Computing the ∆V I find that it will produce 4,546 m/s, just what I need to get to low Kerbin orbit.

Things don’t always work out as nicely as they did in this example, so compromises sometimes have to be made. However, if you follow these design guidelines as closely as possible, you should get good results.

Also note that the upcoming aero changes will likely mean I'll have to revise my rules of thumb.

Edited January 19, 2015 by OhioBob
added example

[Van Disaster]

"Will it fit in a plane"

- Yes - throw in a plane. Thankfully most things will... largest spaceplane I have will lift more than it's entire fuelled mass to medium Kerbin orbit, biggest payload was 285t out of 552t total. ( TWR of about 0.32 at launch! )

-- Can it make orbit on it's own?

--- Yes, put in sub-orbital spaceplane. This is usually for satellites & probes, the delivery system is a spaceplane that lacks any space control bits like RCS as all it has to do is hop out of the atmosphere for a few mins. I worked out that it costs 600 creds at most to fuel a launch in that, although obviously that doesn't include the 1-200 dV the payload burns to circularize. For ion powered craft that is not really worth worrying about.

- No -

-- Will it go up on one stage?

--- Yes - use a ssto/recoverable rocket, probably 3600-4k dV in total ( I run FAR ) including re-entry burn. Usually has fins large enough to be called wings which saves a couple of hundred dV.

--- No - usual design is three roughly similar stacks with the two outer crossfeeding to the centre one. Depending on where it's going, the payload may include a high ISP transfer stage which is used for the final part of ascent also. Heavier payloads generally just mean adding another outer stack. If that doesn't work I'll just throw any ideas out & build something that best fits from a clean sheet. I still like this one - even had functional tiny srbs which saved a noticeable amount of dV by giving it a kick off the pad - but nowadays the lifter would be a spaceplane.

Rocket TWR I try and keep below 1.3 on the pad, although given I use KIDS with the ISP fixer that has been really hard to guess.

[NecroBones]

Let me first stipulate that I use the stock game, i.e. no FAR/NEAR. The rules of thumb that I use that have given me good results are:

1) Two stages to get to LKO (additional stages on top of that for whatever my mission is)

2) Vacuum ÃŽâ€V of 4550 m/s (actual ÃŽâ€V about 4400 m/s)

3) Stage 1 TWR at liftoff ≈ 1.65

4) Stage 2 TWR at ignition ≈ 1.30

5) Ratio of Stage 2 thrust to Stage 1 thrust ≈ 0.35

Using these guidelines I typically get a payload fraction of about 0.16. I also frequently use strap-on SRBs to get my liftoff TWR ratio right.

I do something similar these days, except I'm happy with 3 or 4 stages (the latter usually only being the case if I have a few small SRBs for an extra kick off the pad). Though my target dV is usually around 5 km/s, since I can't assume I'll always have a perfect ascent profile, and the ~10% margin is quite useful. But after enough experience and forum discussions, the pad-TWR of 1.6 to 1.65 seems to work very, very well. I used to aim for 1.8 to 2, but have experimentally found that 1.6 on the pad is often slightly more efficient.

[OhioBob]

I do something similar these days, except I'm happy with 3 or 4 stages (the latter usually only being the case if I have a few small SRBs for an extra kick off the pad). Though my target dV is usually around 5 km/s, since I can't assume I'll always have a perfect ascent profile, and the ~10% margin is quite useful. But after enough experience and forum discussions, the pad-TWR of 1.6 to 1.65 seems to work very, very well. I used to aim for 1.8 to 2, but have experimentally found that 1.6 on the pad is often slightly more efficient.

I've flown the same basic rocket design (just in various sizes) long enough that I'm reasonable good at hitting the 4550 m/s number. Of course I still often fly with some safety margin because there aren't enough parts in the game to always be able to custom design a rocket that exactly matches my requirements, thus I usually end up rounding up little bit. For example, I might use a 20-t capable LV to launch a 18-t payload.

When I first started playing KSP, I also aimed for a liftoff TWR of around 2. That was before I understood how bad the drag model was. After I learned how aerodynamics worked in the game, I ran a bunch of computer simulations to try to come up with an optimized design, which is what led to my rules of thumb. When I put those rules to use in the game I was happy with the results and have since stuck to it. It's nice to discover that my results agree with what most of the KSP community has independently derived.

[Vanamonde]

We like to keep things organized, with the how-to questions in Gameplay. And so, thread moved.

[Mad_Maelstrom]

OhioBob,

The design of your rocket is pretty efficient but I'm having some trouble getting it into orbit. Below is an image of what I built based on your design. I'm running KSP stock and I can almost get it into orbit but I run out of fuel in the 2nd stage before I can get the periapsis out of the atmosphere. The payload is 19.9T and includes SAS and everything above it. What does your launch profile typically look like?

Edited January 19, 2015 by Mad_Maelstrom
Repair image

[OhioBob]

OhioBob,

The design of your rocket is pretty efficient but I'm having some trouble getting it into orbit. Below is an image of what I built based on your design. I'm running KSP stock and I can almost get it into orbit but I run out of fuel in the 2nd stage before I can get the periapsis out of the atmosphere. The payload is 19.9T and includes SAS and everything above it. What does your launch profile typically look like?

I'm not seeing your image but it's probably not necessary that I see it.

Part of my design simulations was also trying to optimize the ascent profile, so its a good thing that you asked. With a 19.9 t payload you don't have much margin for error, but you should be able to get into a 75 km orbit. I typically start the pitch over just after passing 5000 m altitude. I execute a slow gradual turn, always trying to keep the level indicator on the Navball within a few degrees of the prograde indicator. Avoid sudden sharp turns. As the prograde indicator approaches the horizon, I level off my vehicle horizontally. If all has gone well, I should be able to accelerate up to about 2300 m/s at a cutoff altitude of about 50 km. I cut the engine just as the apoapsis passes 75 km. Ideally the prograde indicator should be about +3 degrees at engine shutdown. The orbit circularization burn should be less than 100 m/s. The thing that takes practice is getting the prograde indicator to drop at the right rate to get an efficient turn. If the rate is too fast or too slow you have to correct by producing a fairly large separation between the level indicator and prograde marker, and that wastes ÃŽâ€V. If you make a really good turn, the level indicator never needs to go outside the yellow circle on the prograde indicator.

BTW, I just edited my design example. I wrote Mainsail for the second stage when I meant to write Skipper. If you're using a Mainsail then that could be your problem.

I can also see your image now. It certainly looks like the design I described, and it looks like you're using the Skipper engine. I see no reason why you shouldn't be able to get 20 t into orbit with that.

Edited January 19, 2015 by OhioBob
added comment

[Torquemadus]

Two stage to orbit.

All solid expendable first stage to minimise cost.

Liquid fuelled re-useable upper stage.

[Mad_Maelstrom]

I'm not seeing your image but it's probably not necessary that I see it.

Part of my design simulations was also trying to optimize the ascent profile, so its a good thing that you asked. With a 19.9 t payload you don't have much margin for error, but you should be able to get into a 75 km orbit. I typically start the pitch over just after passing 5000 m altitude. I execute a slow gradual turn, always trying to keep the level indicator on the Navball within a few degrees of the prograde indicator. Avoid sudden sharp turns. As the prograde indicator approaches the horizon, I level off my vehicle horizontally. If all has gone well, I should be able to accelerate up to about 2300 m/s at a cutoff altitude of about 50 km. I cut the engine just as the apoapsis passes 75 km. Ideally the prograde indicator should be about +3 degrees at engine shutdown. The orbit circularization burn should be less than 100 m/s. The thing that takes practice is getting the prograde indicator to drop at the right rate to get an efficient turn. If the rate is too fast or too slow you have to correct by producing a fairly large separation between the level indicator and prograde marker, and that wastes ÃŽâ€V. If you make a really good turn, the level indicator never needs to go outside the yellow circle on the prograde indicator.

BTW, I just edited my design example. I wrote Mainsail for the second stage when I meant to write Skipper. If you're using a Mainsail then that could be your problem.

I can also see your image now. It certainly looks like the design I described, and it looks like you're using the Skipper engine. I see no reason why you shouldn't be able to get 20 t into orbit with that.

I got it, the trick definitely was getting the ascent right.

[GoSlash27]

Early in the career I use a very similar approach to the OP; I use a spreadsheet to "build" the vehicle in 3 stages to orbit from the top down.

Once I've unlocked enough, Everything goes up in either jet powered SSTO mass lifters or spaceplanes.

Best,

-Slashy

[Jouni]

I usually build 2.5-stage rockets for stock and 6.4x. For stock planets with FAR, I usually replace the upper stage of the stock rocket with more payload. My payload targets are 15% for stock, 25% for FAR, and 4% for 6.4x.

The amount of fuel in a stage depends almost exclusively on the engine. The Poodle uses an X200-16 fuel tank, the Skipper uses an orange tank, the Mainsail uses 2-2.5 orange tanks, the LFB uses 1-2 additional orange tanks, the KR-2L uses 1-2 S3-14400 fuel tanks, and the KS-25x4 uses 2.5-4 S3-14400 fuel tanks.

The ideal upper stage for a stock rocket should have TWR at least 1, but as low values as 0.7 are acceptable. For FAR and 6.4x, the values are 0.7 and 0.5. If the engine is too weak, I switch to the next engine type (from Poodle to Skipper, from Skipper to KR-2L).

The lower stage is one category bigger than the upper stage. I pair a Poodle-powered upper stage with a Skipper-powered lower stage, a Skipper with a Mainsail, and a KR-2L with a KS-25x4.

The next step is adding boosters. Unless I'm building something really big, there will be two boosters. For a small rocket with a Skipper, the boosters can be SRBs or use Skippers or Mainsails. Mainsail-powered rockets use Mainsails or LFBs as boosters, while 3.75 m rockets use LFBs or KS-25x4s.

The final step is fine-tuning the rocket, which usually involves a few test launches. I add or remove fuel to reach the delta-v target (4600-4700 m/s for stock, 3500 m/s for FAR, 7500 m/s for 6.4x), and tweak engines to get the initial TWR to 1.2-1.4 and the TWR after booster separation above 1.

[SRV Ron]

Two stage career 18 ton limited to orbit.

Probe SSTO

Mod two stage to orbit where the first stage is a SRB

And, last year 2 stage 3 kerbal to orbit

[OhioBob]

I got it, the trick definitely was getting the ascent right.

I’m glad you got it to fly. I generally judge how well I did on the ascent by how much ÃŽâ€V it takes to circularize the orbit. My objective is to be below 100 m/s, with 50 m/s being about the best I’ve been able to achieve. If I’m much more than that then I know I didn’t hit my cutoff conditions quite right and I need more practice. I’m by no means perfect as I still botch it every now and then.

When I first started playing KSP I used a spreadsheet similar to yours, mainly to compute the mass of my stages and to compute the ÃŽâ€V. Now that I’ve added KER I don’t do that anymore. I now just scratch out my designs on a piece of paper like I did in the example that I gave.

Early in a career game things change so fast with the acquisition of new technologies that I tend not to be very formal in my launch vehicle designs. I just throw something together on a case by case basis that gives me the ÃŽâ€V I need. I have enough experience at this point that I know how not to overdesign things too much.

After I’ve unlocked all the rocket parts I switch to using a fleet of stock designs that I have saved as subassemblies. I’ve got nearly 20 designs ranging from 1-tonne launchers all the way up to an 80-tonne launcher. After I’ve designed a payload I just snap on the smallest launcher that has a lift capacity equal to or greater than my payload mass (though in some cases cost may be the deciding factor).

The majority of the designs are 2-stage liquid, though about a third of them are augmented with 2 to 4 strap-on SRBs. I just recently added two new designs that use 6 strap-on SRBs as the first stage with an air-lit liquid core as the second stage. I get a payload fraction of only about 0.13 with these, but they’re cheap because of the extensive use of solids. I mostly like to avoid complex staging designs even though they have the potential to be more efficient than my preferred 2-stage approach. Complex staging, such as asparagus, I find to be more expensive than it's worth.

The best payload fraction I’ve able to achieve with any of my launchers has been 0.175, which is my 80-tonne launcher. I haven’t actually needed it yet in any of my games, though I’ve confirmed its capability by launching a dummy payload. This is what it looks like (shown without payload):

The first stage uses a cluster of five Mainsails and the second stage uses a single Kerbodyne KP-2L. It may look like I'm using asparagus staging, but that's not the case. The four outboard engines are rigidly attached without decouplers. The propellant feeds from the center tank to the outer tanks. When the center tank goes dry, the center engine cuts off to lower the TWR. The four outboard engines continue to run until they drain their individual tanks (in about 38 seconds), then the entire first stage drops away as a unit.

2) Delta V to orbit of >=4500 m/s (this is based on what I have found online and experience. This assumes SL Isp for the 1st stage and the avg between SL and vac Isp for the 2nd stage)

I just noticed the above in your OP. I initially didn’t read it close enough and thought the 4500 m/s was the vacuum ÃŽâ€V, which is how most people around here calculate it. The general consensus is that it requires a vacuum Isp of 4550 m/s to attain orbit, which my own experience agrees with. One thing I learned from my simulations is that the actual ÃŽâ€V needed to attain orbit, calculated using actual ISP, is about 4400 m/s. I’ve found that if you average SL Isp and vac Isp for the first stage, and use vac Isp for the second stage, you’ll come very close to calculating the actual ÃŽâ€V. I think this works better than the method you’ve described.

Edited April 2, 2015 by OhioBob

[moogoob]

I design backwards from the payload - work out what I need where and use Mechjeb's delta-v calculations help me out. (That's cheating!!! Boo! Hiss!) I'll often put a limit on the throttle to keep TWR at something that will keep my ascent nice and steady, depending on engines (K-advanced benefitsthis when launching smaller-than-gigantic payloads).

What I did was make a save game, in sandbox mode, for the creation and testing of launch vehicles. I call it "SpaceM Launch Systems", and made many, many launches. Cluttersats, station parts and giant fuel slugs ring Kerbin like a highly-caffeinated Saturn, and over lots of practice I got good at designing launchers that were effective and efficient enough to use in Career.

[Chewy]

I've yet to save a launcher into a sub-assembly. Typically the way I design a launcher is top down similar to moogoob. I generally try to go small and light with the final stage depending on the payload, it has big paybacks if the upper stages are built efficiently (I don't really care if I litter space with loads of debris).

I use KER for delta-v calcs and NEAR. My rockets will usually be 2-4 stages depending on the payload size and destination but i customize each rocket for its intended use. For each stage I will choose a reasonable engine and then choose a fuel tank size that will cause me to hit the twr that I'm shooting for. I'll recover the 2nd to last stage for funds (assuming I decouple on a sub-orbital trajectory) and I cap it off with a launching stage of srb's.

[Janos1986]

Payload is king. You design your launch vehicle accordingly, with balanced TWRs.

My magic numbers are 1,7 for the launch stage, 1,5 for the stratospheric burn (which should bring me to an 80km Ap) and 0.50/0.70 for the circularization and vacuum travel.

First and second stage combined should have at least 4200m/s of dV.

Edit: Yay 200 posts.

Edited January 19, 2015 by Janos1986

[Mad_Maelstrom]

Mod two stage to orbit where the first stage is a SRB

http://i.imgur.com/d92Z8V2.jpg

http://i.imgur.com/hCNCm2h.jpg

The homemade rocket parts are awesome! What mod are these parts from?

[OhioBob]

I design backwards from the payload -

Typically the way I design a launcher is top down similar to moogoob.

That’s really the most logical method. If you’re custom designing for a specific payload, there’s probably no other good way to do it.

… but i customize each rocket for its intended use.

That’s what I use to do, and I still do it in early career mode. I also always custom design the upper stages to meet each mission’s specific requirements. However, I found that many of my lower stages ended up looking the same, so I decided to just formalize it with a group a standard designs that I could save as subassemblies. I've found that it saves me a lot of time.

Most of my subassemblies were built in the reverse of what I normally do, i.e. from the bottom up. I’d pick an engine to serve as my first stage and then build a launch vehicle around it. After I had a design I’d confirm its maximum lift capacity by launching dummy payloads. I’d then save it as an X-tonne LV. In order to build a bigger fleet of launch vehicles to meet a broadening range of payload sizes, I started adding SRBs to give me a large range of liftoff thrusts.

Even with the standard subassemblies I can still customize a little bit to my specific payload by off loading propellant and/or limiting thrust.

Edited January 19, 2015 by OhioBob

[MircoMars]

most of what you guys said, but I still go to 10km before I turn. thus 1500m/s is a magic number for me. I try to get my liftoff-stage to 1500km/s so it burns out at around 10km.

my second stage is about 4550 -1500 - 50 = 3000 m/s. circularisation is done with the payload.

TWR >1.5 liftoff and low; > 1.2 upper atmo; ~0.8 vacuum.

in my current career I use a recoverable LFO-SSTO for everything around 17.5t (more mass to orbit with SRBs).

[capi3101]

I'm still a fan of asparagus, despite changes over the years that make such designs more expensive and aerodynamically less practical. In career I'll use standard single or dual-stage boosters until Fuel Lines are unlocked, at which point I'll switch over. '>Temstar's guidelines, despite their age, still give me a reliable booster every time, though in this day and age I'm liable to just pick a single engine that gives me more thrust than I need and then use the thrust limiter to set it to where it needs to be. I've only used engine clusters once since the thrust limiters became available - and that was with a mega booster for a 450 tonne test payload.

[tsotha]

That’s [designing for payload] really the most logical method. If you’re custom designing for a specific payload, there’s probably no other good way to do it.

You can certainly be more efficient if you design around the payload, but I really liked Manley's series on "Completely reusable space program", where he had a standard launcher. Having extra dV doesn't hurt you very much if you land everything back at KSC, and it saved him gobs of time.