An analysis of engine performances on Eve

[MirageNL]

Given the extreme environment of Eve, I’ve always wondered what effect this has on engine performances. The Vector, Dart and Mammoth are often recommended for Eve, but I never really saw any quantitative numbers backing them up. Fortunately, the staging interface in the VAB lets us set the environment to Eve at sea level and it’s only a matter of using math to derive Isp values from the deltaV values. As expected, I found that the Vector, Dart and Mammoth do pretty well while most other engines suck. Interestingly, the Thud also performs pretty okay.

The highest Isp values for Eve are: Dart: 267; Vector: 152; Thud: 101; Mammoth-II: 97; Mainsail: 96; Spider: 74. All others are below 50 (except possibly the Rapier; I forgot to test that one).

The Isp values translate to the highest thrust values: Mammoth: 1332; Mainsail: 501; Vector: 411; Bottle-Rocket: 230; Clydesdale: 228; Dart: 142; Kickback: 115; Thud: 92.

Using Eve’s gravity of 16.68 m/s² (1.7x Kerbin), the highest TWR values are: Dart: 8.5; Vector: 6.2; Hammer: 5.6; Mammoth-II: 5.3; Mainsail: 5.0; Thud: 3.5; Kickback: 3.3.

Optimal launch configuration

While the Dart has the highest efficiency and TWR, it lacks absolute thrust and an efficient engine is useless if it can’t get its payload off the ground. Since the delta-V calculation doesn’t account for gravity pulling the rocket down, I find that instead the most useful quantity for a launch is the total change in momentum (or impulse) that an engine can deliver, which is equal to the net upward force integrated over time: 

J = integral (F - g*(m_payload + m_engine+ m_fuselage + m_fuel - R*t)) dt

Here F is the engine thrust, g is the local specific gravity and R is the fuel burn rate in kg/s. We can assume that for most cases m_fuselage = 0.125 * m_fuel . The total burn time can be calculated from t = m_fuel /R. The equation then results in:

J = (g/R) * ((F/g - m_payload - m_engine)*m_fuel - 0.625*m_fuel²)

By solving for dJ/dm_fuel = 0, we can find the amount of fuel for which the maximum impulse is achieved:

m_fuel = 0.8*(F/g - m_payload - m_engine)

Interestingly, this means that for every ton of payload, you need to substract 800kg of fuel to keep the impulse maximized. From this, we also get the optimal launch TWR:

TWR = 1 + (F/g - m_payload - m_engine) / (9*F/g + m_payload + m_engine)

This means optimal launch TWR is always <1.111, getting lower with increasing payloads and gravity, depending on the engine. By adding the optimal fuel mass to the impulse equation, we find the maximum impulse:

J_max = 0.4*(g/R)*(F/g - m_payload - m_engine)²

Since g=16.68 m/s² for Eve, and F, R and m_engine are constant for each engine, the only remaining free variable is m_payload.

Engine comparison

As you can see, the Mammoth-II can potentially deliver the most impulse by far for any payload. In second place is the Vector for payloads below 12t, but above 12t the Mainsail would be a better second choice. Without any payload, the Dart has almost as much maximum impulse as the Mainsail, but that quickly drops off.

However, to get the most out of the Mammoth, you’d need an enormous amount of fuel. Without local production, this would all need to be brought in from Kerbin and you would need to manage to land it on Eve without burning up in the dense atmosphere or smashing too hard into the surface due to the high gravity. So, maybe the best value to look at would instead be the maximum impulse per kg of starting mass. The math becomes a bit more complicated at this point, but the Dart would now become the best choice for payloads below 2.5t. Between 2.5t and 5.3t the best choice would be the Vector and for payloads above that the Mammoth-II brings the most impulse per kg:

Now, do keep in mind that these are the values per engine. Given the LG size of the Mammoth-II, you could argue it should actually be compared to 7 SM or 3 MD engines for similar footprints. In that case, the Mammoth becomes completely inferior to 7 Vectors and would only be better than 7 Darts for impractically heavy payloads of over 40t. It would perform about the same as 3 Mainsails or 36 Thuds:

7 Vectors would however require much more fuel for maximum impulse than a single Mammoth. So yet another way is to compare the amounts of engines that need a similar starting mass to achieve their optimal impulses. For very large payloads, that would be the case for either 7 Darts, 3 Vectors or 1 Mammoth-II. For smaller payloads, the 7 Darts would deliver far more momentum, followed by the 3 Vectors:

Staging configurations

For a final comparison, I considered a payload of about 3t (a command pod, a Terrier, sufficient fuel for a circularization burn and some appendices) and an asparagus staging configuration. Using 7 Dart engines would require 3.6t of fuel for the center engine and 13t of fuel for each of the 3 outer stages (so 6.5t per engine), giving a total of 24,000 kNs of impulse and a starting mass of 58t. Using 3 Vectors would require 14t of fuel for the center engine and 34t of fuel for the outer stage (so 17t per engine), giving a total of 22,500 kNs of impulse and a starting mass of 70t. A single Mammoth-II would require 50t of fuel for a total impulse of 18,300 kNs, with a starting mass of 74t.

Again the Dart comes out on top, but I do have to note that I used sea-level values for all stages. Performances of the Vector and the Mammoth would especially improve a lot while gaining altitude, while the Dart would only improve a bit. You could consider using a Vector at the center stage with 6 Darts on the outer stage, but the additional fuel for the Vector would then count as a higher payload for the prior stages. This makes the Darts much less effective and it would only result in a total of 19,300 kNs of impulse, while having a starting mass of 74t. In fact, since the Dart suffers so much from higher payloads, asparagus staging is probably not even the most efficient way to use it. Just using 7 Darts without staging would give us a much larger impulse of 43,200 KNs, for a starting mass of just 55t. Using drop tanks while keeping all the engines would be even better. Of course, this doesn’t account for the effects of drag as a result of the wider rocket and the increased acceleration, so the best results might actually be achieved by an in-between solution.

Conclusion

Given all these results, I would at least have to conclude that the Mammoth-II and the Mainsail are never good picks, at least not when they have to be brought in from Kerbin. The optimal choice would be to use Darts. For larger payloads the Vector is a viable choice to lower the amount of engines, or when stabilizers aren't enough and you really need thrust vectoring (which the Dart doesn’t have). The only advantage that the Thud brings is that it’s radially attachable, but you would need a lot of them to make them work.

This was pretty interesting to work out as preparation towards the Under Pressure mission, but it's all just theoretical. I don't have a lot of actual experience with Eve, so I'm wondering how this all corresponds with your experiences.

Edited March 29 by MirageNL

[MechBFP]

The first 5 to 10 kilometers of Eve really are a tricky thing. What is most efficient at sea level rapidly changes over what is really a handful of seconds as you gain altitude. I find it is more efficient to be less efficient at sea level in order to simply get through the thickest part of the atmosphere as quickly as possible. Any losses at those lower altitudes are quickly made up with increased efficiency of the following stages thanks to the increased ISP.

For all my sea level Eve launchers I have always used darts+vector for the initial launch, and then drop the darts+tanks soon after launch since as your pointed out their benefit compared to the other engines quickly diminishes.

[The Aziz]

Once colonies arrive I'm sure I'm gonna come up with something to lift a single Kerbal off the ground for the first time after a decade of playing.

My old big rocket never made it to the surface in one piece.

[magnemoe]

I found an cluster of to work well 

9 darts two in the boosters and one in core, works well from sea level, could probably cut it back quite a bit as core is a bit to heavy and thing is overbuild. 
Main issue for me is landing accuracy and stability. Here I eject the transfer stage after deorbit burn. 
Granted this is one Kerbal on a rover seat inside an faring. Side booster has crossfeeed, second stage with an terrier, then an oscar tank with 3 spider engines. 
Now for 10 kerbals landed at Eve, the vector looks promising. 

[MirageNL]

I thought of another way to compare the performances. I realized getting out of Eve's atmosphere is not actually the real problem; the challenge is landing a rocket capable of doing so. Given the difficulties involving aerobraking and landing, having a low starting mass means everything. So, I changed the equation for impulse to a function of the total mass:

J = (g/R) * ((8/9)*(F/g - m_payload - m_engine)*(m_tot - m_payload - m_engine) - (40/81)*(m_tot - m_payload - m_engine)²)

Solving for dJ/dm_tot = 0 gives us the mass for maximum impulse:

m_tot = 0.1*(9*F/g + m_payload + m_engine)

So, for each tonne of payload, optimal starting mass increases by 100 kg. I plotted the impulse graphs for 0, 3 and 10 tonne payloads:

The question to ask here is how much impulse you need. My guess is somewhere around 20-30 kNs would be a good aim, which no single engine can achieve and so multiple engines will be necessary. Below is a comparison between the previous number of engines with similar optimal starting masses:

For actually equal optimal starting masses we would need to use fractions of engines:

Since the payload mass affects all optimal starting masses equally, the engine number ratios stay the same for increasing payloads. All the curves just get lower and narrower. Increasing the number of engines increases the impulse, but also the optimal starting mass.

Edited April 3 by MirageNL

[steveman0]

So if I'm reading this right, the dart in some kind of multi-engine/asparagus staging appears to be the best. This matches what I found emperically when I experimented with a sample return build in the VAB. 

[KincaidFrankMF]

2 hours ago, steveman0 said:

So if I'm reading this right, the dart in some kind of multi-engine/asparagus staging appears to be the best. This matches what I found emperically when I experimented with a sample return build in the VAB. 

Good old-fashioned trial and error

5 hours ago, MirageNL said:

I thought of another way to compare the performances. I realized getting out of Eve's atmosphere is not actually the real problem; the challenge is landing a rocket capable of doing so.

Very much the case. You need something capable of remaining perfectly stable during entry, that's nevertheless light and still stable when going the other way for blast off.

I found dart-asparagus best for a one-kerbal science mission. Under Pressure is a whole other challenge - the lack of gimbal was a real problem for me, so I ended up using a vector sustainer core surrounded by dart asparagus boosters. But huge reaction wheels might have worked too (must admit I didn't try).

The biggest surprise was how small you can make a craft for UP. My ten-kerbal lander was maybe three times the mass of the one-kerbal version (172t for atmospheric entry, 154t at take-off - and I'm sure it can be done lighter!)

[herbal space program]

On 3/29/2024 at 7:12 AM, MirageNL said:

This was pretty interesting to work out as preparation towards the Under Pressure mission, but it's all just theoretical. I don't have a lot of actual experience with Eve, so I'm wondering how this all corresponds with your experiences.

I haven't done an Eve return mission in KSP2 yet, but in KSP1 I found the Vector to be hands-down the best engine for lifting larger payloads from near Eve SL.  The Dart may have the most favorable ISP profile, but it has a lower TWR than the Vector (OK, perhaps just about equal at Eve SL), but more importantly, a much lower thrust per cross-sectional area than the Vector, like 4.5 times lower. When you're trying to take off through that endless soupy atmosphere, lowering your drag is a very important consideration that is not addressed by these calculations, and that requires the lowest cross-sectional area possible. Unless of course you want to shamelessly clip engines into one another to cheat the drag model, but that's not my style.

...And I should add that while these laudably rigorous calculations are great for figuring out what works best for a given payload at Eve SL, actually ascending to orbit only happens in that regime for the first few seconds. Pretty soon you're in substantially thinner air, so a lot of this stuff doesn't matter so much anymore. For that reason (h/t @magnemoe), I have a SWERV as the core stage on my Eve 10 lander. Its decent TWR and far superior ISP make it the best choice IMO for the large dV push from being high in Eve's atmo to actually being in orbit.

Edited April 4 by herbal space program

[magnemoe]

6 hours ago, herbal space program said:

I haven't done an Eve return mission in KSP2 yet, but in KSP1 I found the Vector to be hands-down the best engine for lifting larger payloads from near Eve SL.  The Dart may have the most favorable ISP profile, but it has a lower TWR than the Vector (OK, perhaps just about equal at Eve SL), but more importantly, a much lower thrust per cross-sectional area than the Vector, like 4.5 times lower. When you're trying to take off through that endless soupy atmosphere, lowering your drag is a very important consideration that is not addressed by these calculations, and that requires the lowest cross-sectional area possible. Unless of course you want to shamelessly clip engines into one another to cheat the drag model, but that's not my style.

...And I should add that while these laudably rigorous calculations are great for figuring out what works best for a given payload at Eve SL, actually ascending to orbit only happens in that regime for the first few seconds. Pretty soon you're in substantially thinner air, so a lot of this stuff doesn't matter so much anymore. For that reason (h/t @magnemoe), I have a SWERV as the core stage on my Eve 10 lander. Its decent TWR and far superior ISP make it the best choice IMO for the large dV push from being high in Eve's atmo to actually being in orbit.

With larger payloads I agree, for low payloads as in single kerbal, I say it would be overkill on core stage. My rocket is overbuild as it has 1000 m/s left then in orbit, but did not bother optimizing more. Its based on an KSP 1 design I used. 
In KSP 1 the decent was much simpler, inflatable heat shield and and two layers of plates in the rear to keep rear more draggy than front. 

[MirageNL]

16 hours ago, steveman0 said:

So if I'm reading this right, the dart in some kind of multi-engine/asparagus staging appears to be the best. This matches what I found emperically when I experimented with a sample return build in the VAB. 

Without drag effects, it would be better to keep the engines:

On 3/29/2024 at 3:12 PM, MirageNL said:

Using 7 Dart engines would require 3.6t of fuel for the center engine and 13t of fuel for each of the 3 outer stages (so 6.5t per engine), giving a total of 24,000 kNs of impulse and a starting mass of 58t.

On 3/29/2024 at 3:12 PM, MirageNL said:

Just using 7 Darts without staging would give us a much larger impulse of 43,200 KNs, for a starting mass of just 55t.

That's a huge difference, but drag definitely is a factor. I tested these setups yesterday: the 2-stage Terrier+7xDart setup easily achieved an Ap of over 150 km. However, by the time it was high enough to start the gravity turn, its speed was already so high that my stabilizers started to overheat. Turning too fast would flip the craft over, but within seconds the Ap was already above 90 km and I didn't have enough fuel left to do the full circularization. The Terrier+7xDart asparagus setup was much more manageable , though the core Dart stage struggled a bit with the gravity turn due to the low TWR. Still, after circularizing I had over 100 dV left.

A core stage of 3 Darts with the other 4 Darts in asparagus setup might yield better results. That would mean climbing gradually to 20-30 km until starting the gravity turn, after which the speed could pick up.

9 hours ago, herbal space program said:

I haven't done an Eve return mission in KSP2 yet, but in KSP1 I found the Vector to be hands-down the best engine for lifting larger payloads from near Eve SL.  The Dart may have the most favorable ISP profile, but it has a lower TWR than the Vector (OK, perhaps just about equal at Eve SL), but more importantly, a much lower thrust per cross-sectional area than the Vector, like 4.5 times lower. When you're trying to take off through that endless soupy atmosphere, lowering your drag is a very important consideration that is not addressed by these calculations, and that requires the lowest cross-sectional area possible. Unless of course you want to shamelessly clip engines into one another to cheat the drag model, but that's not my style.

...And I should add that while these laudably rigorous calculations are great for figuring out what works best for a given payload at Eve SL, actually ascending to orbit only happens in that regime for the first few seconds. Pretty soon you're in substantially thinner air, so a lot of this stuff doesn't matter so much anymore. For that reason (h/t @magnemoe), I have a SWERV as the core stage on my Eve 10 lander. Its decent TWR and far superior ISP make it the best choice IMO for the large dV push from being high in Eve's atmo to actually being in orbit.

Actually at Eve SL the Dart TWR is 8.5, while the Vector TWR is just 6.2. I do agree though hat the thrust-to-size ratio (TSR?) of the Vector is unparalleled, so I understand the preference when dealing with heavier payloads.

The Swerv however is 10 tonnes alone, requiring plenty of additional fuel and payload to justify. It's also LG sized, adding to drag. The amount of fuel for the Vectors would be massive, so what kind of payload could you possibly need to lift from Eve?

[KincaidFrankMF]

14 minutes ago, MirageNL said:

The Swerv however is 10 tonnes alone, requiring plenty of additional fuel and payload to justify. It's also LG sized, adding to drag. The amount of fuel for the Vectors would be massive, so what kind of payload could you possibly need to lift from Eve?

I haven't tried using purely vectors rather than darts (reduced part count might be another factor to throw into the mix) but can confirm the SWERV is amazing in this build. It's overkill, sure, what we really need is a mini SWERV with the same ISP but even so... it needs so little fuel that it ends up far, far lighter than a methalox alternative.

[astrobond]

Excellent thread

Here is the last version of my Eve 10K Lander:

144.85 Tons with 8 Vectors aspargus and a SWERV, Ok From Eve Sea Level with about 300m/s left in 100x100km Orbit

Ascent is vertical until SWERV stage at about 20km, then aim 40 degrees inclination, and prograde until AP 100km.

(Edit: Hydrogen tank set to 5.6 t of fuel)

Edited April 4 by astrobond
Hydrogen tank tunning :)

[KincaidFrankMF]

Ah, grumble seats. My guys travelled in comfort 

Did a bit of experimenting, and I still can't make vectors come out better than darts - even with the improved aerodynamics they're about 350dv short. Could just be a less efficient setup though - adapting a design isn't the same as building it that way from the ground up...

[astrobond]

15 minutes ago, KincaidFrankMF said:

Ah, grumble seats. My guys travelled in comfort 

Did a bit of experimenting, and I still can't make vectors come out better than darts - even with the improved aerodynamics they're about 350dv short. Could just be a less efficient setup though - adapting a design isn't the same as building it that way from the ground up...

You mean you have a 154t Eve 10k launcher without grumble seats ? WOW i hope you will post a picture of your lander 

Here is another tip i use for my eve Launchers:

I use two small fins with some angle to separate last boosters without colliding the core stage... lighter and more efficient on Eve than sepratrons

Edited April 4 by astrobond
typo

[astrobond]

55 minutes ago, KincaidFrankMF said:

Ah, grumble seats. My guys travelled in comfort 

Did a bit of experimenting, and I still can't make vectors come out better than darts - even with the improved aerodynamics they're about 350dv short. Could just be a less efficient setup though - adapting a design isn't the same as building it that way from the ground up...

Hmmmm... yes, Tourist Class for Kerbals is better !!! I just tried a new 151.58 t version, with Hydrogen tank Full,  30  degrees prograde at 20km and... 400m/s left with this big boy full of Kebals, SWERV is awesome and Vectors boosters definately good in aerodynamics

[herbal space program]

Looking at the above designs, I see I really overbuilt my Eve 10 lander! It looks lot a lot like what @astrobond posted above, but it has 4 large side boosters with 7 Vectors each instead  of eight with just one each. I blame the completely Borked dV readout in the VAB! I have to re-install KER and see how much dV my lander really has. My guess now is it could probably  make it all the way back to LKO without even aerobraking. 

[KincaidFrankMF]

1 hour ago, astrobond said:

You mean you have a 154t Eve 10k launcher without grumble seats ? WOW i hope you will post a picture of your lander 

Actually, after a bit of tweaking I can do better - 144t, made it to orbit with 600dv spare.

I start the gravity turn much earlier - around 4-5k but keeping it very gentle.

And all wrapped up for atmo entry...

1 hour ago, astrobond said:

I use two small fins with some angle to separate last boosters without colliding the core stage... lighter and more efficient on Eve than sepratrons

Love it. Mine seemed to separate ok without, but they often don't.

Edited April 4 by KincaidFrankMF

[KincaidFrankMF]

I have absolutely no idea why I decided to wrap the docking port up in a faring, btw. I thiiiiiiiiink it's a leftover from an earlier design, back when I didn't realise you could still use docking ports as decouplers, and then I just liked the way it looks? But it's adding drag for no reason lol

[astrobond]

21 minutes ago, KincaidFrankMF said:

Actually, after a bit of tweaking I can do better - 144t, made it to orbit with 600dv spare.

I start the gravity turn much earlier - around 4-5k but keeping it very gentle.

 

Awesome Design !!! Good idea to use exactly 10 seats with 4 parts,  lighter than my 11 seats two parts  Really optimized Lander.... Congrats

[steveman0]

Really awesome designs here. I've struggled with the landing plan. What's the preferred strategy here? Parachutes and/or powered touchdown? I don't see any landing legs. Landing on the heat shields? Decoupled legs?

Oh, I just noticed the parachutes on the aero tail for orientation!

Edited April 4 by steveman0
Spotted the chutes

[astrobond]

Just now, steveman0 said:

Really awesome designs here. I've struggled with the landing plan. What's the preferred strategy here? Parachutes and/or powered touchdown? I don't see any landing legs. Landing on the heat shields? Decoupled legs?

Thankyou @steveman0 This is only the version to test to go orbit, i use only Parachutes, you must keep all the fuel for the ascent

I use landing legs that can be decoupled yes, same for the heatshields, they are decoupled before landing to save weight for the parachutes !

I also use heatshieds at the top of the lander to be autostable on reentry, and i have also twisted a little the upper shields to do an autorotation to keep parts not too much hot ^^

[KincaidFrankMF]

2 hours ago, steveman0 said:

Really awesome designs here. I've struggled with the landing plan. What's the preferred strategy here? Parachutes and/or powered touchdown? I don't see any landing legs. Landing on the heat shields? Decoupled legs?

Thanks! I found it was lighter to use extra chutes than try to convince landing legs not to snap... I just come down on the heat shields! They're tough as anything and give a reasonably wide, stable base. They handle 10m/s impacts no problem.

The other thing to consider is the overall take-off weight from Kerbin, including the strategy for getting back home. Rather than bring along a heavy 10k command module, I just dock the lander to a mini "parachute stage" once it's back in orbit and refuel the swerv. There's a heat shield underneath the hitchhiker, so the lander can just decouple its engines and land a second time with the new chutes.

Btw, is it just me or are fairings doing literally nothing at this point? Likewise cargo bays. Neither seem to help lifting off from Eve, where aerodynamic costs are really obvious.

Edited April 4 by KincaidFrankMF

[astrobond]

1 hour ago, KincaidFrankMF said:

Btw, is it just me or are fairings doing literally nothing at this point? Likewise cargo bays. Neither seem to help lifting off from Eve, where aerodynamic costs are really obvious.

Not only cargo bays don't help, but actually they induce a lot of drag !!! I wanted to do science on Eve, i ended with all science parts on a decoupler to let them on Eve after doing the science  and getting the datas in the command pod 

Tried also fairings, but no help for me too... Using just good aerodynamic parts are enough, and lighter is better

For landing legs if the lander tends to tip over , i have "home made" landing legs with xs truss and small heatshields at the bottom, really strong and no problems  

 

Edited April 4 by astrobond
picture of landing legs

[herbal space program]

2 hours ago, KincaidFrankMF said:

Thanks! I found it was lighter to use extra chutes than try to convince landing legs not to snap... I just come down on the heat shields! They're tough as anything and give a reasonably wide, stable base. They handle 10m/s impacts no problem.

I went with  the solution @magnemoe used for this, which was to put small heat shields as feet on the ends of 2.5m square XS trusses and attach them to the ship with small aerodynamic decouplers + struts.  I haven't actually landed it on Eve yet,  but testing it on Kerbin I landed a 300-ton  (yes, massively overbuilt) lander on ten of those at 16 m/s and amazingly nothing broke! I don't have the jumbo landing legs yet, but even if I did I was concerned those would burn off during re-entry if I mounted them to decouplers. The heat shield feet however, if offset slightly inwards, did a great job of protecting the struts and decouplers connecting them to the ship. 

 

Heh, looks like @astrobond posted a pic of exactly what I was talking about just above and I failed to notice! Not sure to whom the original credit for that solution actually belongs, but I can now attest from experience that it works amazingly well.

Edited April 4 by herbal space program

[KincaidFrankMF]

1 hour ago, astrobond said:

Tried also fairings, but no help for me too... Using just good aerodynamic parts are enough, and lighter is better

Yep, definitely. Hopefully they'll get worked on soon.

1 hour ago, astrobond said:

For landing legs if the lander tends to tip over , i have "home made" landing legs with xs truss and small heatshields at the bottom, really strong and no problems  

Ha! That's brilliant