How do the v1.4 & DLC engines rank against our old favorites?

[GoSlash27]

17 hours ago, Temeter said:

I've went through much experimenting, and there is just not way to use for a poodle upper stage without wasting a bunch of money/weight. Kerbins circularization burns are just too short to support something with thrust that weak; and the Skipper is obviously too heavy and powerful for an upper stage.

Temeter,
 My design parameters for a 2.5m upper stage are 1,650 m/sec DV and minimum t/w of 0.7.

The Poodle is the lightest and cheapest stage for 14-20t payloads.
The Poodle is the cheapest (but not lightest) stage from 7-14t payloads.

The Skipper actually makes the cheapest and lightest stage for 21-45t payloads.

Both engines are supremely useful as upper stages.

Best,
-Slashy

 

[blakemw]

18 hours ago, Temeter said:

IMO the Poodle was always a bit underpowered, even after the buff. Sure it makes a solid orbital engine, but IMO the poodle should rather be an upper stage launcher engine. Think a 2.5m stack, the lower stage a Mainsail, the upper stage a Poodle. Sounds natural, doesn't it? Smart design to teach beginners how to do rockets.

But no! I've went through much experimenting, and there is just not way to use for a poodle upper stage without wasting a bunch of money/weight. Kerbins circularization burns are just too short to support something with thrust that weak; and the Skipper is obviously too heavy and powerful for an upper stage.

I somewhat agree, somewhat disagree. The thing about KSP is it is a punyverse. IRL the first stage gets a rocket 40% of the way to orbit and the second stage the other 60%. In KSP due to the tinyness of Kerbin, the first stage will get the rocket 80% of the way to orbit, and the second stage can then eject the payload to Dres or Moho. Kerbin tends to favor single-stage rockets while Earth tends to favor two-stage rockets, in the sense that if you load up an engine with enough fuel to justify its existence (usually 2-4 minutes of burn time), on Kerbin that engine will burn most or all the way to orbit.

Now with that out of the way, I'll just assume we're happy to use the second stage for an ejection burn so we get some real mileage out of it. Real rockets tend to have a 5:1 to 7:1 ratio between first stage and second stage thrust (for example Saturn V or Falcon 9 fit this pattern) and this also works quite well in KSP. The reason we need a ratio like this, is if say the second stage only has 30% less thrust than the first stage (say a Mainsail on a Twin-Boar), you'd be better off just burning the first stage engine for longer and eliminating the extra weight and expense of the second engine and decouplers, at the other end of the spectrum if the second stage engine is too weak you're not pushing enough mass with it (i.e. the rocket is way bottom heavy) or it'll be very low TWR, which isn't always bad, but does tend to be tedious.

A Twin-Boar:Skipper is a 3:1 ratio, a Mammoth:Skipper is a 6:1 ratio.

A Skipper:Poodle is 2.6:1, Mainsail:Poodle is 6:1 and Twin-Boar:Poodle is 8:1

So a Skipper is a bit overpowered on a Twin-Boar and you might want to just burn the Twin-Boar for longer (i.e. make it a SSTO), though this can also be partly remedied by boosting the Twin-Boar with SRBs. A poodle is very overpowered on a Skipper and you'd probably just skip the Poodle and burn that fuel in the Skipper. A Poodle is only slightly underpowered on a Mainsail, but is noticeably underpowered on a Twin-Boar - the Mainsail thing would be good, except the Twin-Boar is better in absolute terms than a Mainsail, being cheaper with 33% more thrust and about the same weight, so a rocket will almost always be better with a Twin-Boar than a Mainsail, and the Poodle will be underpowered.

So anyway, for an engine like a Twin-Boar, a Skipper is too powerful but a Poodle is too weak. What'd be ideal is an engine with a thrust of about 400kN. There's also a pretty wide gulf between the Terrier and Poodle, this is partly covered by the Aerospike, but something with a thrust of about 130kN would be most ideal for a second stage on a Skipper.

Edited March 13, 2018 by blakemw

[Tyko]

14 hours ago, Cunjo Carl said:

Could be that your system doesn't like my image host, Postimage? I don't know why that would be, but I'll be updating the pictures tomorrow, so if they're still missing after that I'll host a direct download link to the pictures too

I'm not even seeing them as links. The .png links you included are just showing up as plain text...tried 3 browsers on 2 computer.

Thanks for looking into this

[Cunjo Carl]

On 3/13/2018 at 10:47 AM, Tyko said:

I'm not even seeing them as links. The .png links you included are just showing up as plain text...tried 3 browsers on 2 computer.

Thanks for looking into this

For sure! Sorry it's not working.  It almost sounds like the trouble is specific to either your forum account or maybe your firewall somehow? Anyways, I've made a direct download link for the pictures available in a small edit at the end of the topic post. I'll be sure to keep it up to date (or atleast as up to date as I keep everything else ).

 

On 3/13/2018 at 1:05 AM, KerikBalm said:

I still don't see how you're arriving at "effective Isp" (which is a term I'm familiar being used for jet engine specific impulse, and a related effective exhaust velocity).

The Isp is fixed. Engine TWR is fixed. Payload is variable. More importantly, Payload as a fraction of total dry mass is variable. dV required for the payload is variable. You're going to have to give a more detailed mathematical explanation, because I really have no idea what you're trying to show here. You haven't given a rigorous definition of "effective Isp".

 

The definition is just what was in the little blerb and spoiler in the main post. This 'effective Isp' and the one from bypass jet engines aren't related. If I'd remembered the term was used for bypass, I would have given it a different name . This effective Isp is a measurement of mass efficiency, and the lines shown in the plots are the best possible mass efficiencies for that engine and TWR requirement. This optimum happens at a specific payload fraction, which is found by the following:

 

Spoiler

 

deltaV = -g₀*Isp*ln(m_dry/m_wet)       Let's solve for this in terms of rocket properties

Fuel tanks have a drymass ratio D such that
D = m_emptyfuel/m_fullfuel = 1/9

And other unitless variables which let us generalize our equation to any sized rocket.
E = m_eng/m_wet = TWR_required/TWR_engine
P = m_pay/m_wet
F = m_fullfuel/m_wet

Now we solve for mdry/mwet
m_wet = m_eng + m_fullfuel + m_pay
m_dry = m_eng + D*m_fullfuel + m_pay

m_dry = m_wet - (1-D)*m_fullfuel
m_dry/m_wet  = 1 - (1-D)*F

F, the fueltank massfraction, isn't terribly useful as a variable, so we'll replace it with engine and payload fractions.
m_wet/m_wet = 1 = E + F + P
F = 1-E-P
m_dry/m_wet  = 1-(1-D)*(1-E-P)     Final form
m_dry/m_wet  = (1-D)*(E+P) + D    Alternate form

deltaV = -g₀*Isp*ln(1-(1-D)*(1-E-P))  The working equation, deltaV for a single stage in terms of rocket properties.

 

Now we want a way to measure mass efficiency for both single-stages and full rockets in a way that takes into account both DeltaV and Mass Fraction. The above deltaV equation is a handy relationship between these for individual stages and tells us deltaV as a function of P (that stage's mass fraction). Here's the tricky part- when you have rocket made of n-many stages, we'll have a total deltaV that's the sum of all stages, but a total mass fraction that's the product of all stages. There's only one function that could let us compare the two, and that's the log, which can turn products into sums. Here's a quick, general example comparing two ways of combining deltaV and massfraction using an example of a single- and 2-stage rocket with X and Y as stand ins for the deltaV and payload fraction of the single stage. Notice how the bottom option for measuring mass efficiency is consistent between single and multiple stages.

characteristic                     1 stage     2 stage rocket
  deltaV                                     X                  2X
  payloadFrac                           Y                   Y²
  deltaV*payloadFrac            X*Y               2X*Y²                               ... While intuitive, this way of measuring isn't consistent between stages and rockets
  deltaV/-ln(payloadFrac)   X/-ln(Y)         2X/-ln(Y²)  =  X/-ln(Y)           Success! The value stays the same.

So our one and only consistent way to measure the tradeoff between deltaV and massfraction is   -deltaV/ln(payloadFrac) . I've been calling it 'deltaV per e-fold scale', which is ... well, mighty confusing probably. It's very useful though!

deltaV/-ln(P) = g₀*Isp*ln(1-(1-D)*(1-E-P)) / ln(P)         Mass efficiency equation (single stage, or rocket made of n equivalent stages)
deltaV/-ln(payloadFrac) = sum(deltaV_i)/sum(-ln(P_i))                      Mass efficiency equation (rocket made of different stages, subscript i)

You can use these formulas directly as a way of comparing rockets, and my main rocket spreadsheet does this. It's fairly easy to work towards the optimum rockets just by hand and eye, but of course we'd like to have solvers and plots instead. While the second equation for mixed stages doesn't have any convenient solutions, you can find optimal payload fractions for the top equation as a close second best.

Optimize   deltaV/-ln(P) = g₀*Isp*ln(1-(1-D)*(1-E-P)) / ln(P)  for P as the dependent variable, with given D,E.

The results are easy to plot, and for normal situations winds up quite close to the actual (very complicated) optimum!

The real question in the end is always 'does it work?' , and the answer is yeah! It works well in practice, I've gotten a lot of use from these. It's improved the deltaV on my previously best optimized rockets, so it passes the test. It's closer to the truth than maximum dV, but still not quite as good as finding the optimum to the generalized deltaV/-ln(payloadFrac) = sum(deltaV_i)/sum(-ln(P_i))  equation. Like I mention in the initial post, finding this final optimum requires you to put more deltaV in your better stages, and less deltaV in your worse stages.

 'DeltaV per e-fold scale' is super confusing so there's one final step for easy consumption. Define a new term called effective Isp such that:

deltaV = -g₀*Isp_eff*ln(P)           (for both single stages and rockets made of n equivalent stages) So...
deltaV/-ln(P) = g₀*Isp_eff
( deltaV/-ln(P) ) / g₀ = Isp_eff    

And this is the final equation for effective Isp. It's just another way of presenting the mass efficiency from before, but it's scaled in such a way that it looks generally familiar because the numbers are similar to normal Isp. Also, the word for it is a lot simpler! I guess I need to find another one though . If you made it this far, thanks for reading

 

 

On 3/13/2018 at 2:40 AM, Frozen_Heart said:

Which engine is the Skiff from the DLC?

The RE-I2... I think it's supposed to be the J2 (SaturnV second stage)

 

These DLC engines have atm curves now! While the curves are the normal linear scaling we'd expect in the 0-1atm region, the part files describe a sharper drop at higher atm. I'll have to test it, but I think that means the otherwise amazing looking Mastadon won't be much use on Eve

 

 

Edited March 15, 2018 by Cunjo Carl

[PTNLemay]

I haven't done a whole lot of number crunching, but I have been able to see the result in the Kerbal Engineer deltaV indicator.  Looking at almost every payload choice, the Wolfhound or the Cheetah come out on top when it comes to deltaV.  But in some cases it's very close.

 

The only times I can see myself using the Poodle anymore is if I know the top stage will need good thrust, but not be too heavy.  Which is a rare necessity... if your top stage needs that thrust to get into orbit, it usually means your before-last stage didn't do it's job properly (like an over-burdened Skipper).

It's a shame, because the Poodle is an old favorite and I definitely have a soft spot for it.  But the new ones ones are just better.  I do love the look of the Cheetah, but the Wolfhound feels a bit meh.  That rectangular baseplate is very... bland.

Edited March 27, 2018 by PTNLemay

[AngryKitty]

19 hours ago, PTNLemay said:

It's a shame, because the Poodle is an old favorite and I definitely have a soft spot for it.  But the new ones ones are just better.  I do love the look of the Cheetah, but the Wolfhound feels a bit meh.  That rectangular baseplate is very... bland.

There's a reason for that, to be fair!

[klesh]

Yeah, the Wolfhound baseplate is spot on to me; looks like I would expect it to.  If you don't like it, you can always switch to the bare variant.

[PTNLemay]

@AngryKitty
OK I'll admit that is an impressive recreation.  I'm just so used to having nice plumbing and greebling sprinkled around the engine bell.  The Wolfhound is so minimalist it's a bit disorienting.

I wonder why the real Service Module main engine had so much paneling covering up it's gutty-works.  Were they worried lunar dust would threaten the components?

Edited March 28, 2018 by PTNLemay

[Skystorm]

@PTNLemay The service module would not have been affected by lunar dust as it remained in orbit the whole time.  However, they may have been concerned about how the cold temperatures in space would affect it as it had to be able to perform multiple burns over the course of the 9 day mission.

[fourfa]

Booster engines have to operate for about ten minutes max. The SPS needed to work for two weeks in hard vacuum, with extreme temperature gradients between full sun and utter cold darkness. So no surprise they opted to shield all the fiddly bits with a cover

EDIT: also, the complicated greebly bits on most of the other engines represent turbopumps and propellant feed lines, cooling apparatus, etc.  The SPS was a pressure-fed engine (no pumps at all), low-power (compared to the rest of Saturn anyway, so small feed lines), hypergolic (so no elaborate cryogenic insulation), no cooling required.  Just a couple of valves and an oversized engine bell.

Edited March 28, 2018 by fourfa

[DerekL1963]

2 hours ago, fourfa said:

Booster engines have to operate for about ten minutes max. The SPS needed to work for two weeks in hard vacuum, with extreme temperature gradients between full sun and utter cold darkness. So no surprise they opted to shield all the fiddly bits with a cover

*And* the SPS/CSM evolved from a direct ascent lander -  and I bet the configuration inherited features from that.  Tucking what few greebly bits it did have up into the SM would have saved height and protected them from any recirculated plume and dust.

Edit:  Here's what the parts inside the SM looked like, minus the TVC actuators - https://airandspace.si.edu/collection-objects/rocket-engine-liquid-chamber-apollo-service-module-propulsion-system-sps-0

Edited March 28, 2018 by DerekL1963
Add link to picture.

[RoboRay]

7 hours ago, Skystorm said:

The service module would not have been affected by lunar dust as it remained in orbit the whole time.  However, they may have been concerned about how the cold temperatures in space would affect it as it had to be able to perform multiple burns over the course of the 9 day mission.

The SPS motor for the service module was actually designed to land a larger version of the CSM on the Moon, before the Lunar Orbit Rendezvous approach with a separate lander replaced the original Direct Ascent plan.  It was seriously overpowered for the role it ended up playing, but it was already designed and tested so they didn't mess with it.  It was certainly designed with lunar surface conditions in mind.

Edited March 29, 2018 by RoboRay

[FleshJeb]

@Cunjo Carl I don't know if you're aware of the fantastic work that @tavert and @Meithan did on optimal engine charts, but here is the end result:

https://meithan.net/KSP/engines/

It looks like they're both AWOL, but perhaps you'd like to take over development?

https://github.com/meithan/engine_charts

I happen to love the style of chart they use, BTW.

Edited March 29, 2018 by FleshJeb