[WEB] Online engine cluster calculator

[blizzy78]

It depends on the booster stack size. On a 3.75 m tank, you can fit a lot of Rockomax 48-7S at the outer spot.

The calculator currently does not allow to pick a booster stack size, it recommends one itself. This is mainly to prevent booster stacks clipping into each other. Of course you're not bound to use the recommended stack size, you can use any size you want. For example, if the calculator recommends using 2 booster stacks, you can practically use any size.

[blizzy78]

So, to get back to the "no clipping" setting. What it currently does is to prevent clipping between the center and outer engines. Which means that there should be no clipping occurring in a straight line from outer-center-outer. Of course it can be considered a bug that you can have so many outer engines that they will still clip into each other. You can see that pretty easily from the graphic.

[Streetwind]

One thought that occured to me... have you considered allowing for multi-adapters, even if only the stock Rockomax Quad Adapter if nothing else? That allows you to have four 1.25m engines under a 2.5m stack without part clipping enabled, at the cost of a little bit of extra mass. However, from personal experience that still results in less total mass than having to size up to, say, a mainsail because a cluster of 2 or 3 1.25m engines doesn't provide enough TWR.

Also, how's your experience with the dV of the designs that this spits out? Earlier I randomly got a "single stack with one mainsail" configuration while trying out different values. On a 12ton payload with 15% payload fraction, that turns into an 80t vehicle where up to 55.8t may be fuel ( (80-12-6)x0.9 ). Being doubtful that a mainsail equipped SSTO rocket with 15% payload fraction is even possible at all, I quickly plugged the numbers into the rocket equation (assuming an average 320 Isp over the entire burn) and got ~3750 dV. In other words, this lifter will fall short of reaching orbit, as expected. Is that a regular thing that suggested configurations don't survive a reality check, or did I just find a corner case here?

[blizzy78]

have you considered allowing for multi-adapters, even if only the stock Rockomax Quad Adapter if nothing else?

I have not. Personally I don't play with stock only, so I've not had a real need so far to use the adapters.

Also, how's your experience with the dV of the designs that this spits out? Earlier I randomly got a "single stack with one mainsail" configuration while trying out different values. ... In other words, this lifter will fall short of reaching orbit, as expected.

The problem with smaller payloads (and in turn, small lifters) is that the calculator will tend to suggest SSTOs, which makes sense from the mathematical standpoint. It does not make sense in reality because when the tanks get empty, you still carry around all that mass. The same goes for the engines: In the higher atmosphere, you don't need all that thrust, so any engine mass you haven't lost by then is practically dead mass.

In the end, the calculator is designed around the asparagus staging, shedding useless mass as soon as possible.

[Streetwind]

The quad adapter is useful even with mods - KW Rocketry's Wildcats (and, of course, the OP Vesta VR-1) work great, for example. But then again, I'm not sure how it would have to be implemented code-wise.

Regarding the Vesta VR-1, by the way, I'm surprised it doesn't show up more. For example, I know from personal experience that these things in a 7-cluster under a 3.75m core can give you a 15% payload fraction SSTO without the slightest effort. But I can't seem to force the calculator to show this. Even when selecting "best sea level Isp" and choosing my payload so it fits the expected thrust, it is recommending me to use 1 Vesta and 4 LV-T45, throttled down to match whatever TWR limits I try to give it. That's 6.6t of engines, with an average sea level Isp of 324. Alternately without throttling, it offers 1 Wildcat and 3 LV-T45, for 6 tons of engine mass and 320 sea level Isp at 830 kN thrust. A set of 7 Vestas however would have 4.2t engine weight and 350 sea level Isp at 840 kN thrust. It's clearly better in all of the metrics the calculator cares about, but for some reason the algorithm doesn't seem to like it. Does it maybe try to minimize engine count?

[blizzy78]

Does it maybe try to minimize engine count?

Yes, it does. The default priority list looks like this. These are per total rocket, not per stack nor per engine. Also, a "shell" is a spot where you can place an engine (center, outer, radial.)

var SORT_PRIORITIES = [ 'engineTWR', 'twr', 'isp', 'engineMass', 'numParts', 'numShells' ];

As a user, you can pick one of these and pull it to the front, but the rest stays the same. Note though that throttling may still give better TWR (engines only.) than using other engines and not throttling them.

I haven't looked into the specific engines you mentioned because I'm at work right now. Perhaps you can post the recommendations again with total thrust, mass, Isp, and throttle setting.

Edited March 31, 2014 by blizzy78

[Streetwind]

Well, the Vesta VR-1 has the second highest 'engineTWR' in a set that contains stock and KW Rocketry only, beaten by the 48-7S alone. As a result, the 7-cluster of them has the lowest possible total engine weight and thus the highest possible total vessel TWR for that amount of thrust; you would have to fit 28 of the small 48-7S on your rocket to do better, and there isn't enough space for that even under a 3.75m stack. It also has the second highest ASL Isp, beaten by the toroidal aerospike alone, an engine which has terrible TWR. The Vesta's ASL Isp is equal to the vacuum Isp of the 48-7S.

Therefore by all accounts it should always win the combination of the first three categories, no matter which attribute I prioritize, so long as there is space to mount them.

payload 7t

payload fraction 15%

twr 1.75 to 2.0

center stack size 3.75m

boosters 0 - 0

max center outer engines 8

center radials 0 - 0

max booster outer engines 0

booster radial engines 0 - 0

true radial engines = TRUE

allow part clipping = FALSE

allow thrust limiting = FALSE

center stack thrust 0 - 100

targeting best sea level Isp

engine packs: Stock and KW Rocketry

Expected Result: 1 center Vesta VR-1 with 350 ASL Isp, 6 outer Vesta VR-1 with 350 ASL Isp

- Mass: 46.7 t (engines: 4.2 t)

- Thrust: 840 kN

- TWR: 1.83

- Number of engines: 7

Actual Result: 1 center Wildcat-V with 320 ASL Isp, 3 outer LV-45T with 320 ASL Isp

- Mass: 46.7 t (engines: 6 t)

- Thrust: 830 kN

- TWR: 1.81

- Number of engines: 4

Result with allow thrust limiting = TRUE: 1 center Vesta VR-1 with 350 ASL Isp at 95% throttle, 4 outer LV-45T with 320 ASL Isp

- Mass: 46.7 t (engines: 6.6 t)

- Thrust: 914 kN

- TWR: 2

- Number of engines: 5

Result when targeting minimum engine mass: 1 center Skipper with 300 ASL Isp, 6 outer 48-7S with 300 ASL Isp

- Mass: 46.7 t (engines: 4.6 t)

- Thrust: 830 kN

- TWR: 1.81

- Number of engines: 7

Interesting fact: removing stock and allowing only KW Rocketry will in fact yield the expected 7-cluster of Vestas. It's as if the algorithm is somehow preferring stock engines whenever possible and only resorts to mod engines if it absolutely can't get around it... either that, or the absence of asparagus booster stacks is thoroughly confusing it. That case is harder to test, but technically a set of 3.75m booster cores with 7-clusters of Vestas should still be very high on the priority list for all possible setups. Literally the only thing this solution doesn't have is an abundance of thrust, so your rockets can't get too big, but the engine is still so horribly unbalanced that it's the best choice for almost everything that has space to mount enough of them. Maybe I can try tinkering together a Vesta-based asparagus launcher when I get home myself.

Well, I hope this helps, in any case

Edited March 31, 2014 by Streetwind

[blizzy78]

targeting best sea level Isp

Expected Result: 1 center Vesta VR-1 with 350 ASL Isp, 6 outer Vesta VR-1 with 350 ASL Isp

Total Isp would be 350.

Actual Result: 1 center Wildcat-V with 320 ASL Isp, 3 outer LV-45T with 320 ASL Isp

Total Isp would be ~321.4.

This is odd indeed, the 350 Isp calculation should win over the 321. I shall be looking at the code again to see what's wrong.

It's as if the algorithm is somehow preferring stock engines whenever possible and only resorts to mod engines if it absolutely can't get around it...

It is not, I promise

[blizzy78]

Okay, I've fixed a logic error in the code. It should now work as expected.

[Nepos]

I just came across this today and I was wondering this -

How do you know what your stack size is? exmaple Stack size: 2.5 m

When I look at the wiki for the Rockomax Jumbo-64 Fuel Tank is does not mention the size.

Thanks

[Streetwind]

The sizes correspond to the ingame part size steps. The Jumbo64 is a 2.5m part. The stuff you start with is 1.25m, the tiny probe stuff (like the ion engine) is 0.625m. The other two sizes, 3.75m and 5m, are only available through mods right now, but in the upcoming Asteroid mission patch you will get stock 3.75m parts as well.

@blizzy78: time to test some stuff then! :3

[Streetwind]

Oh wow. This works so much better all of a sudden. I was prepared to give you more bug reports, like the one instance I managed to get where "best sea level Isp" gave me 5 engines and switch to "minimum engine count" gave me 20 engines all of a sudden (wtf!), but it seems like it was all down to the same bug you squashed. The results make so much more sense now. It also seems to come to a solution slightly more quickly, but that may just be me.

Now the only thing that's missing from my own personal point of view is clipping detection for engines within a shell, so it only suggests designs that are actually free of clipping if the checkbox is disabled But I'll still recommend this to my friend anyway.

It could probably be hardcoded if necessary. One size class difference between tank and engines allows a maximum of 2 in the outer shell, two size classes difference a maximum of 6, and so on. Let me see if I can find the rest.

EDIT: right, 3 classes difference is 16 max, and I have not successfully managed to get the calculator to put enough 48-7S on a 5m tank to clip but it's likely between 20 and 22. The most I got was 19, and that still had some space. Interestingly enough, there's easily room for more than one 0.625m outer shell on the 5m tank...

Edited March 31, 2014 by Streetwind

[blizzy78]

Considering that if you activate multiple engine packs the calculator has to look through literally hundreds of thousands of combinations, I'm still impressed that it remains pretty fast (more or less.) It does try to use all of your CPU cores, though.

[Streetwind]

I have four 4.2 GHz Haswell cores

I may have found a corner case where it gets stuck and doesn't finish at all despite no CPU load being shown, but I need to reproduce it first before I can tell you more.

[Nepos]

The sizes correspond to the ingame part size steps. The Jumbo64 is a 2.5m part. The stuff you start with is 1.25m, the tiny probe stuff (like the ion engine) is 0.625m. The other two sizes, 3.75m and 5m, are only available through mods right now, but in the upcoming Asteroid mission patch you will get stock 3.75m parts as well.

@blizzy78: time to test some stuff then! :3

Thanks (10 chars)

[Streetwind]

Nope, can't reproduce it. Must've been a fluke...

[CaelumEtAstra]

Im confused...How do I use this?

[blizzy78]

Im confused...How do I use this?

See this post. While not being the documentation to my calculator, I think it should clear things up.

[blizzy78]

I've just posted a thread to announce that I'm looking for a developer to take over this software.

Meanwhile, I'm halting all development on this. It's been a good time so far. Thank you all for your input and support.