Are Nuclear Rockets "Cheating"?

[einsteiner]

Small changes in Isp cause large changes in the total mass of your craft, this demonstrates this perfectly. For a recent mission the total starting mass of the craft with a LV-N was 58.5 tons. With a Poodle it would be 580 tons. That's a lot to launch into orbit before your mission can even start. So they are useful.

Is it cheating? I say no if you use it with some restrictions. I don't use it as a takeoff/landing engine, to avoid irradiating planetary surfaces. I also don't use it for any burn less than 100 m/s, as I imagine starting and stopping a nuclear reactor is non-trivial.

Edited August 28, 2014 by einsteiner

[Sky_walker]

What is actually sorta cheating (in terms of realism) now is the ion engines, which have ridiculously over-powered thrust to weight ratios compared to their real life counterparts.

Good point about them, though it's setup this way due to the game limitations (you cannot simply turn on the engine for XXXX seconds and compress time). Nukes are setup the way they are cause it was the quickest way to do it.

[Renegrade]

Your Mun landers also have less payload than 2.71 tonnes, so they're not comparable. Efficiency is about completing the given mission as efficiently as possible, instead of completing a different mission more efficiently.

Well, we obviously have differing goals in terms of missions. I'm always looking for science returns, and you only need one kerbal to max those out, but you do need to carry at least one goo pod and a science jr (plus the massless 'doodads', but they're massless nonsense again..). So by my definition, the lighter mk1 designs I posted before are actually accomplishing that mission, but that 2-man can thing you posted is probably missing out on some of the science and thereby partly failing the mission.

What are your goals exactly? I'm guessing you're possibly playing sandbox...?

You'd better not say 'make an ion lander'

That particular lander was originally intended to demonstrate that the new ion engines were suitable for real missions, instead of just ultra-light stuff. You can obviously land on the Mun with one ion engine and a command chair, or two engines and a Mk1 lander can. The twin ion engine lander should weight something like 1.6-1.7 tonnes without abusing massless parts. With abuse, we could get it down to 1.22 tonnes.

Well, leaving aside the commander chair AS being ultra-light and not 'real', slapping two ions onto a lander can with a single inline ion tank gives 1220kg and 2435/2.01 dv/local twr, assuming you abuse the crap out of massless components (I'm imagining booms of ox-stats made out of those cubic struts, coated in Z-100s and Z-400s #lolnophysics). That is indeed delightfully light (although it's payload includes no goo pods or science jrs, which are definitely not massless).

Making a less cheaty, more sensible design would erase a lot of that advantage, however. That's about 3.3 m/sec^2 accel, and the ions use 17.46 e/sec, so you'll need to store ~3174e for an apollo-style/high efficiency landing at night (assuming about 600 accel at the very end, for around 182 seconds), or if you're using like, one or two ox-stats to recharge (having one to four massless oxstats is a fairly minor cheatiness that I would definitely permit). Stacking on three Z-1000s gives us 1370kg 2161/1.79.

Replacing the batteries with the OX-4L/W panels results in a slightly heavier craft, upwards of 1378 kg, so we'll ignore these as they're less efficient. These ion craft cost in the range of 20k funds in case anybody's interested.

Or we could just slap three round8 tanks and a 48-7S onto the mk1 can and have a 1226/16.63 craft that weighs only 1108 kg and costs 2862 and call it a 100-kg lighter day. Or replace the round8s with a FL-T100 and get a 1731/14.59 craft weighing only 1263, costing 2032. A less dv, but still lighter, and capable of landing on medium-sized muns. Bringing it up to ion-level range would put it up to 1399 or 1500.

So, even for this lighter payload, 48-7S can still provide a competitive solution. It's kinda short-legged in terms of dv, but it has enough TWR to handle any meaningful grav field in the game. Plus that TWR lets you use less-than-optimal approaches for very little penalty.

By the way, I might actually build a flying ion chair for fun, and possibly a mass-abusing ion mun lander.

(NB: if you're running my Horrible Nerf, or Stupid_chris' Stock Rebalance, you might find the sun a little weak in the outer system, although moreso in SR than HN. )

[Jouni]

Well, we obviously have differing goals in terms of missions. I'm always looking for science returns, and you only need one kerbal to max those out, but you do need to carry at least one goo pod and a science jr (plus the massless 'doodads', but they're massless nonsense again..). So by my definition, the lighter mk1 designs I posted before are actually accomplishing that mission, but that 2-man can thing you posted is probably missing out on some of the science and thereby partly failing the mission.

The basic idea is that you first define mission goals. Those goals determine the payload you need. Then, with the payload fixed, you build a propulsion system that gets the payload where it's needed. Because there can be different goals for the same mission profile (e.g. Mun landing), comparing total masses is pointless. Comparing payload fractions for the mission profile is more useful.

What are your goals exactly? I'm guessing you're possibly playing sandbox...?

I try to find new interesting goals for a mission, and then complete them.

I completed the career mode once back in 0.22. I've tried it again after every update, but I always get bored after a while. The gameplay is still mostly the same. You fly the same basic missions with minor variations, and that's it.

So, even for this lighter payload, 48-7S can still provide a competitive solution. It's kinda short-legged in terms of dv, but it has enough TWR to handle any meaningful grav field in the game. Plus that TWR lets you use less-than-optimal approaches for very little penalty.

In other words, the 48-7S is barely competitive for simple single-landing missions. If you want to land multiple times, the extra fuel you need to carry in the lander or in the mothership makes it inefficient.

On the other hand, using a separate lander is often inefficient, unless it's part of mission goals. You already have engines suitable for the landing in the transfer stage, so you should use them instead. Just leave excess fuel tanks in orbit, and retrieve them after returning from the surface, if they still have any fuel left. As this increases the delta-v you need from the engines, it favors engines with high Isp over engines with high TWR.

[Renegrade]

The basic idea is that you first define mission goals. Those goals determine the payload you need. Then, with the payload fixed, you build a propulsion system that gets the payload where it's needed. Because there can be different goals for the same mission profile (e.g. Mun landing), comparing total masses is pointless. Comparing payload fractions for the mission profile is more useful.

Actually no, absolute masses are very important, as they impact the cost of the lower stages tremendously. Adding five tons in orbit requires 30-50 tons (non-FAR, stock) extra on the launchpad. It doesn't matter if that's like fifty goo pods or twenty ion engines, it still has to be lifted.

Anyhow, our landing goals are incompatible, there's no point in continuing. I'll just say that landing at 1.7 local TWR must be.. really "fun", and that gravity drag asked me to say 'hi' to you. Hermann says hi too.

[Jouni]

Actually no, absolute masses are very important, as they impact the cost of the lower stages tremendously. Adding five tons in orbit requires 30-50 tons (non-FAR, stock) extra on the launchpad. It doesn't matter if that's like fifty goo pods or twenty ion engines, it still has to be lifted.

Remember that we are talking about a scenario, where costs don't matter at all.

Anyhow, our landing goals are incompatible, there's no point in continuing. I'll just say that landing at 1.7 local TWR must be.. really "fun", and that gravity drag asked me to say 'hi' to you. Hermann says hi too.

I actually find low-TWR landings more fun, because they require more planning.

Gravity losses due to low TWR are actually quite insignificant on planets without atmospheres. You probably won't even notice them, unless your initial TWR is 1.3 or below. tavert made some interesting charts on that a while ago. Also, with high-Isp engines such as ions, it's easy to add 500 m/s or 1000 m/s of excess delta-v by accident, because fuel tanks don't come in small enough sizes.

[Renegrade]

Remember that we are talking about a scenario, where costs don't matter at all.

Well, we are talking about efficiency, and it has to be based on something. I usually take that to either mean launchpad mass (from before 0.24, that was the only metric) as there's an upper limit to part count (varies depending on system, drivers, parts in question, etc, but it's always present), and we only have up to 3.75m equipment in stock. Now that we have Funds, I also consider that too (although Funds are rather poorly balanced, so it's kinda of a, "that's a really heavy ship AND it's funds cost are high" rather than the other way around).

Gravity losses due to low TWR are actually quite insignificant on planets without atmospheres. You probably won't even notice them, unless your initial TWR is 1.3 or below. tavert made some interesting charts on that a while ago. Also, with high-Isp engines such as ions, it's easy to add 500 m/s or 1000 m/s of excess delta-v by accident, because fuel tanks don't come in small enough sizes.

I've seen those charts before, although I've never compared landing performance against them. I am able to perform that type of landing, although usually I'm well above 2 TWR (I independently developed the technique after reading about the Apollo lander's descent profile, for dealing with 'low twr landers' which actually meant 3-4 at that time).

Out of curiosity, I had created a non-cheaty dual-engine ion Mun lander after my last post, and using the efficient approach, I got about uh, 822 expended (1892/1.57 dv/twr @1560kg). Also, that took three attempts (although I have to admit I screwed up on the first one, I let it drop and tried to scrub 30m/sec in like two seconds just before touchdown when I could have just let it settle slowly) with a lot of scary mountain-scraping and one desperate mountain-evasion. Maintaining a constant altitude took enormous pitch-up towards the end as my orbital speed fell off, for a very long time. I don't think the 640 that the charts are showing is anywhere near realistic unless you're landing on a perfectly featureless sphere, or have a vast freedom in target selection.

I'm sending up a 3-engine, 2.0 twr model now to see if that's much better..

Leaving aside pure efficiency for a moment, there's a second issue with this lander: that approach is the ONLY approach it can take. A high TWR lander can do a high precision drop by simply scrubbing most of it's orbital velocity above a target, and then doing a strong burn just before landing. This allows precise target picking as well as landing vertically or nearly vertically, meaning it could be at the bottom of a north-south aligned valley etc. The rough cost of such a landing varies, but usually it's in the 600-700 range itself.

(For teh record, I think the 48-7S needs a good beating with the nerf bat in terms of thrust..)

[Jouni]

Well, we are talking about efficiency, and it has to be based on something. I usually take that to either mean launchpad mass (from before 0.24, that was the only metric) as there's an upper limit to part count (varies depending on system, drivers, parts in question, etc, but it's always present), and we only have up to 3.75m equipment in stock. Now that we have Funds, I also consider that too (although Funds are rather poorly balanced, so it's kinda of a, "that's a really heavy ship AND it's funds cost are high" rather than the other way around).

Efficiency is always relative to what you're trying to accomplish. A big rocket can be as efficient as a small rocket, if it's launching a respectively bigger payload.

The traditional efficiency metric in the KSP community was payload fraction. I didn't like it too much, because it favored extreme asparagus staging and large clusters of small engines over designs that felt more reasonable. My current metric is a bit informal: I favor using a few engines over large engine clusters, weak engines over powerful engines, simple staging over complex staging, and simple engines over complex engines. Even though I haven't assigned any numerical values, the metric describes how I feel launch costs should be.

Out of curiosity, I had created a non-cheaty dual-engine ion Mun lander after my last post, and using the efficient approach, I got about uh, 822 expended (1892/1.57 dv/twr @1560kg). Also, that took three attempts (although I have to admit I screwed up on the first one, I let it drop and tried to scrub 30m/sec in like two seconds just before touchdown when I could have just let it settle slowly) with a lot of scary mountain-scraping and one desperate mountain-evasion. Maintaining a constant altitude took enormous pitch-up towards the end as my orbital speed fell off, for a very long time. I don't think the 640 that the charts are showing is anywhere near realistic unless you're landing on a perfectly featureless sphere, or have a vast freedom in target selection.

Those charts assume that you start from a circular equatorial orbit at 0 km. In a real landing, you also want to get rid of your orbital altitude, which makes things more complicated.

My standard approach to low-TWR landings is that landing is the inverse of reaching orbit. When you launch with low TWR, you first climb vertically to clear any nearby obstacles. Then you start to turn, still increasing the ascent rate, while also trying to gain horizontal velocity. When the ascent rate is high enough, you start concentrating on the horizontal velocity, while maintaining the ascent rate roughly constant. As your apoapsis starts getting high enough, you can let the climb rate drop slowly and turn even closer to horizontal. Ideally you should be at a somewhat circular orbit at the right altitude by the time you reach the apoapsis. To land with a low TWR, you just do everything in reverse order.

Getting that right usually takes much more than three attempts, because the game doesn't have any mission planning tools.

[Caelib]

It's not OP or cheating ... fuel lines on the other hand ...

[corvustech]

I see a lot of calculations here to determine if the LV-N is balanced but there are other factors, too.

1. I can land gently on 909s. I can can gently on 45-js. I can even land on clustered poodles( now that's a Kerbal sentence). Anyone who has watched an LV-N snap off and roll downhill can tell you don't land on LV-Ns.

2. LV-N fairings were invented by Jeb to see how easy it is to anger pilots who use clustered engines. Make sure to lock gimbal when staging....

3. The travel stage is the easiest stage to fly, and LV-Ns only belong in vacuum. For usage on vacuum worlds see above.

4. Their mass means using more rcs fuel and thrust when docking or stabilizing.

[Red Iron Crown]

corvustech, mounting a pair or more of LV-Ns radially on a lander solves most of the issues you mention. Even the fairings, which I agree are a special form of torture.

[Laie]

corvusetech: or if you think that a pair of LV-Ns would be one too many, you can literally design your vessel around the engine.

[magnemoe]

I see a lot of calculations here to determine if the LV-N is balanced but there are other factors, too.

1. I can land gently on 909s. I can can gently on 45-js. I can even land on clustered poodles( now that's a Kerbal sentence). Anyone who has watched an LV-N snap off and roll downhill can tell you don't land on LV-Ns.

2. LV-N fairings were invented by Jeb to see how easy it is to anger pilots who use clustered engines. Make sure to lock gimbal when staging....

3. The travel stage is the easiest stage to fly, and LV-Ns only belong in vacuum. For usage on vacuum worlds see above.

4. Their mass means using more rcs fuel and thrust when docking or stabilizing.

1 The kethane and ore miner to the right has both clustered LV-N and clustered poodles.

It land with the LV-N, the poodles is for fully loaded takeoffs.

2 radial mounted engines don't get fairings, the problem is if you use a core engine and a ring of engines around, I tend to leave the center open as the center engine also overheat easy.

3 I agree that LV-N is marginal for landings, exception is where you need transfer burns too like an LKO to Mun surface shuttle.

[corvustech]

I think my point is, as demonstrated, if you have to build an entire ship to fit it, build special configurations so you don't have to deal with its staging, and increase your rcs and launch stage fuel and thrust to the point that the efficiency starts to be eaten by how much fuel and equipment it took to get into LKO with it in the first place and how much rcs for docking, one can't call it cheating. It's more like the vacuum answer to the Mainsail sub orbital niche.

[Renegrade]

The traditional efficiency metric in the KSP community was payload fraction. I didn't like it too much, because it favored extreme asparagus staging and large clusters of small engines over designs that felt more reasonable. My current metric is a bit informal: I favor using a few engines over large engine clusters, weak engines over powerful engines, simple staging over complex staging, and simple engines over complex engines. Even though I haven't assigned any numerical values, the metric describes how I feel launch costs should be.

Well, that IS what you would get if you turn the real world $-per-pound-to-orbit into KSP terms, where dollars didn't exist previously. Now that we have Funds, we have another metric, although I'll agree it's got some weird prices going on.

I do agree with your subjective measure though, the.. ugly asparagus staging (like the picture on KSP's wiki..currently down for maintenance) looks busy and expensive (and ugly) to me. That's one thing that Funds actually gets partly right: all of those decouplers and fuel lines are adding a lot to the funds cost, especially on 1.25m designs.

That being said, I feel that the 48-7S would be a much cheaper engine to build than an ion design. It would sorta be on the scale of a turbocharged car engine...

Those charts assume that you start from a circular equatorial orbit at 0 km. In a real landing, you also want to get rid of your orbital altitude, which makes things more complicated.

Okay yeah, that's definitely way optimistic there. I was starting with a 12km apoapsis/periapsis. At that point, I was decoupling from the injection stage, burning to bring the pe down to just above terrain at my chosen target, and then when I was close (I calculated a 180-190s burn for the various designs, so I was starting the final burn about 90s to peri), I was burning purely horizontal at first, and then slowly pitching up to maintain descent rate close to zero. This exposes the craft though to dangers, specifically because the Map view uses a somehow-different model (yay for faulty LOD) for the planet as the actual space view, and you can find yourself intersecting a mountain or crater rim that looks a lot shorter on the map...

My standard approach to low-TWR landings is that landing is the inverse of reaching orbit. When you launch with low TWR, you first climb vertically to clear any nearby obstacles. Then you start to turn, still increasing the ascent rate, while also trying to gain horizontal velocity. When the ascent rate is high enough, you start concentrating on the horizontal velocity, while maintaining the ascent rate roughly constant. As your apoapsis starts getting high enough, you can let the climb rate drop slowly and turn even closer to horizontal. Ideally you should be at a somewhat circular orbit at the right altitude by the time you reach the apoapsis. To land with a low TWR, you just do everything in reverse order.

That's.. roughly what I was doing for landing (in reverse. see above). In fact, I don't see any difference at all.

My high TWR takeoffs are basically similar to that, by the way, only the 'start to turn' is more like a 'violent jerk to 90 after 0.25 seconds'. It's only the landings that differ. Also, one other thing about the high TWR design - it's significantly faster not only in real world time, but also Kerbal time. That could be important for a life-support enabled game (and is, for BTSM).

Getting that right usually takes much more than three attempts, because the game doesn't have any mission planning tools.

Meh, I stuck my very first Mun landing using a high-ish TWR plan. But you're right about the lack of tools. Having a more accurate map would be a good start... I am going to give it more runs later to see if I can refine it, because I'm now curious about how accurate those charts are compared to actual landings.

corvusetech: or if you think that a pair of LV-Ns would be one too many, you can literally design your vessel around the engine.

http://www.schnobs.de/ksp/Asymmetric_surface.png

Geez, I didn't notice it last time, but man, that's one heck of an asymmetric ship. I love it!

Did you use RCS Build Aid for that, or just stock tools?

[Laie]

Geez, I didn't notice it last time, but man, that's one heck of an asymmetric ship. I love it!

Thank you very much.

Did you use RCS Build Aid for that, or just stock tools?

Just stock at the time. The most challenging bit was to find parts of suitable length so it would be kinda-sorta-balanced to begin with. From that point, I just launched it with SAS disabled and watched in which direction it would turn. Wasn't all that difficult, really... but some effort.

I've since moved on to a two engined Lab-Rover-Lander, but I like to plug this belter scout whenever someone complains that LV-Ns are too damn tall.

[samstarman5]

This is a game we all like to play... if it's in the game, it's not cheating (Except the Dev Console...)

Hey, Whack a Kerb is not cheating!