New Mission-Based Rocket Designer + Simulator

[JSideris]

This tool has been succeeded by the Advanced KSP Mission Tool. Please check out the new thread.

I've spent the last few weekends doing some rocket science, and I'd like to present my KSP Mission-Based Rocket Designer & Simulator. The aim is to abstract away as many design decisions as possible that don't have any bearing, per se, on the mission you're trying to accomplish. Instead, the calculator will attempt to optimize for the lightest possible rocket able to get the job done. 

Try it: https://jsideris.github.io/KSP-Mission-Based-Rocket-Designer/

The tool will algorithmically design two-stage (and now three-stage) rockets, typically. The lower stage is for takeoff and may include anywhere from 0 to 32 radially detachable boosters. Most of the ΔV is intended to be carried out by the vacuum stage, and sometimes the tool will add an intermediate stage to assist with achieving orbit. Each rocket configuration's launch is quickly simulated with varying amounts of fuel to test for feasibility, and the best rocket is presented to the user. Some of the results can be surprising (for instance, using low-tier or weak rockets for takeoff), but all that you need to know is that the program is giving you the lightest rocket design that is able to at least meet the minimum requirements for your mission. This is great, for instance, for designing light-weight return landers.

The app is also not very picky about what type of detacher, fins, nosecones, or other peripherals that you use. It adds a small safety factor to the payload so that you can dress up your rocket with whatever parts you have available.

All in all the tool is fairly accurate and very usable, however there are a few corner cases that I'm working to address. For instance, it's not smart enough to know how to get to Eve and back yet, Overall, it's going quite well; the tool is now smart enough to get a small payload to Eve and back!

Give it a go and let me know your thoughts, questions, concerns, and suggestions. The tool will remain updated and receive upgrades as long as it is well received and people use it.

Notable Updates

2016-07-24

• Fixed a "snowball effect" glitch (discussed below) that caused the simulation to preferentially select more and more massive rockets as the simulation ran.
• Allow the simulator to try stealing some ΔV from the vacuum stage.
• Simulator will now be willing to trade off 1% extra mass for an additional 100 m/s ΔV when choosing between two rocket designs.
• Gave simulator the ability to tack more fuel onto the vacuum engine to assist with orbit.
• Added orbital stages (3-stage rockets) that will sometimes be explored if the rocket is re-entering the atmosphere too early.

Edited August 24, 2016 by JSideris
Updates

[Red Iron Crown]

That's an interesting tool but it yields some odd results. I asked it to put a 5t payload in orbit around Kerbin with 100m/s remaining and all techs available, just to see what the result would be. Here's what it thinks is an optimal lifter would be:

This strikes me as a rather overbuilt rocket to put 5t in low orbit. What am I doing wrong? Here are the settings used to get this result:

[VaPaL]

I liked the idea, sounds very usefull, but I found some problems. Aside from what @Red Iron Crown said (the first stage being superestimated), I got a Rhino engine for the first stage, that's a vaccum engine not a atmospheric one.

Other than that, this is preety good, good visual, clean design, easy to use, unfortunately I can't see the videos (at work). Good work, congrats!

[Red Iron Crown]

@VaPaL Rhino's a sustainer engine more than a pure vacuum one. Vacuum performance generally trumps atmo performance for an ascent engine, assuming the atmo Isp/thrust is not totally awful like with a Poodle or Terrier. Very generally speaking, it's efficient to use a sustainer-style motor for the main stage of an ascent (Swivel, Skipper, Rhino), assuming it can deliver enough thrust to get it moving.

[VaPaL]

@Red Iron Crown So, like an Ariane 5 where the booster generate 92% of the lift, the Vulcan engine only sustain them with the remaining 8%?

[Red Iron Crown]

@VaPaL Basically, yes. For smaller rockets you can do without the boosters if the sustainer gives sufficient TWR on the pad.

[JSideris]

7 hours ago, Red Iron Crown said:

This strikes me as a rather overbuilt rocket to put 5t in low orbit. What am I doing wrong? Here are the settings used to get this result:

Thanks for the feedback, I'll push a patch for that right away.

So what's happening here is that when the tool is comparing between rockets, extra  ΔV does have a bit of weight in the decision. I noticed in some early tests that if I only optimize mass alone, the tool will ignore rocket configs that are twice as efficient even if they only weigh 1 extra kg. So the comparison I made (somewhat arbitrarily since this is a matter of preference) is that if rocket A and rocket B are both successful launches, I will choose rocket B over rocket A if:

ΔV_B/M_B² > ΔV_A/M_A²

However as it goes through the list of rockets and simulates them, that M_A value usually snowballs, which is unintended. I'll play around with some different decision functions. And I'm open to suggestions.

7 hours ago, VaPaL said:

I got a Rhino engine for the first stage, that's a vaccum engine not a atmospheric one.

Yea, like I said, the results are sometimes surprising. If I change the optimizer function from ΔV_B/M_B² > ΔV_A/M_A² to M_B < M_A the tool will sometimes choose a bunch of spark engines for blastoff from Kerbin (like in the thumbnail of the video). A weird choice but the resulting stage only weighs a few tons.

Edited July 19, 2016 by [email protected]

[Red Iron Crown]

@[email protected] Perhaps optimize for mass first, then consider rockets that mass+some.fixed.percentage when looking at additional delta-V. That would permit paying a small mass penalty for much greater capability while not letting it go nuts.

[JSideris]

Just released an awesome new version of the tool. It is now able to design rockets capable of a return trip from Eve. Fixes the snowball bug discussed above and prevents over-building by having the ability to:

• Steal some ΔV from the vacuum stage (or adding additional tanks to this stage to assist with orbit).
  • Allows for small ΔV debts subtracted from the mission, in exchange for huge mass penalties (aka lighter rockets: each 100 m/s stolen = 20% shaved off). Will experiment with this feature further.
• Utilize a new orbital stage between the vacuum stage and takeoff stage.
• Trading 1% extra mass for each additional 100 m/s ΔV.

Here is a screen shot of the 5T/100m/s benchmark:

Be sure to clear your cache when re-loading the page in order to get the latest version.

Edited July 29, 2016 by JSideris