Alternatives to Ferram for realistic drag models?

[Sean Mirrsen]

Not "a craft". Not any given craft.

One, specific, baseline craft. A well-built, typical stock KSP rocket. Or the Kerbal-X, at least.

The point is to make the baseline craft perform in FAR about as well on takeoff as it does in stock. From there, craft built worse will perform worse, while craft built better - slimmer, longer, more aerodynamic (which isn't all that hard in case of the Kerbal-X), will perform better.

Once the values necessary for that behavior are established, everything else will work.

(We called it "tailor-balancing" in Total Annihilation modding. Picking balanced points of reference and making any and all new material have performance and costs balanced against them. It's a slightly different application of the principle, but it's still relevant.)

Edited October 16, 2013 by Sean Mirrsen

[HeadHunter67]

The point is to make the baseline craft perform in FAR about as well on takeoff as it does in stock.

OK, I'm starting to understand where you are coming from, now - though I hope you can see how I'd get the wrong impression from what had been written.

So, aside from the debatable dV savings, what should the measurable effects of a more aerodynamic build be? For the life of me, the only other metric I can imagine is atmospheric efficiency - such as what we can see in KER. And all that really affects is terminal velocity and how fast one can ascend. Is that alone reason enough for using FAR?

[Sean Mirrsen]

A more aerodynamic build is more stable and has a smoother velocity/air-resistance curve (i.e. what being "aerodynamic" is all about, being able to go faster in the needed direction). Rockets are fairly single-purpose, and "going fast in the needed direction" happens to be that purpose.

The reasons for using FAR are really primarily tied to spaceplanes - lift, control surfaces, aerodynamic shape, etc. Rockets are slightly secondary in that regard, which is all the more reason to try and not break them with the mod. Well-built aerodynamic rockets should benefit slightly from FAR, while typical barreloidal asparagus batches should suffer, and the spaceplanes will have the same changes going on with lifting and control surfaces.

As it is now, thick barrel-shaped rockets fly worse in FAR than aerodynamic stacks do - but they still fly better in FAR than they do in stock. That's not really promoting good and/or realistic rocket design.

And while I'm here, I think ferram's proposed balance solution to tweak engine Isp is honestly ridiculous. FAR is all about altering aerodynamic parameters, so why should it be balanced with engine efficiency? Why should VTOL rocketbases on Duna, that don't go very fast at all, suffer from increased fuel use intended to rein in suddenly overperforming lifters? Wouldn't tweaking the atmospheric values be a better way of going about it?

[Van Disaster]

This old argument again; the way to fix asparagus excess ( and any other design excess ) is for the game to make economical designs matter, not to bend physics any more than it has already.

What would your atmospheric tweaks entail?

[Sean Mirrsen]

Mine? Make atmosphere a little more obstructive at high velocities, add emullated turbulence to flat-top rockets, slightly increase overall perceived atmospheric density - probably. I don't really know how to change it yet, I'm no expert at this. Essentially shift the aerodynamic model's numbers around until an agreed-upon baseline rocket performs similarly both in FAR and stock, while at the same time realistically degrading relative performance of poorly-built rockets, and having a similar effect of well- and poorly-built airframes. The changes FAR makes to spaceplanes are a little too complex for me to quantify, but I am generally fine with what it does. Changing the numbers to balance rocket performance will likely mean that FAR spaceplanes will have a little more lift, and travel a little slower on the overall, and will have difficulty breaking the sound barrier close to the ground without being exceptionally streamlined.

It will all have to be balanced, of course, what I'm suggesting is more or less just a direction. The details would need to be experimentally established.

[asmi]

To be honest, I don't quite understand the point of this thread. Being "realistic" means "like in real life", so there is nothing to balance as you can't tweak laws of physics for your purposes. It's either "like in real life", or it isn't "realistic". There simply could not be two "realistic" aerodynamics since laws of physics are universal. "Balancing" it to craft is even more ridiculous since with FAR craft itself is just one variable in the total equation, piloting is just as important (if not more important), and you can't possibly come up with any baseline since underlined mechanics are totally different.

And as was pointed out before, testing aerodynamics with single SRB is just not very scientific.

[HeadHunter67]

A more aerodynamic build is more stable and has a smoother velocity/air-resistance curve (i.e. what being "aerodynamic" is all about, being able to go faster in the needed direction). Rockets are fairly single-purpose, and "going fast in the needed direction" happens to be that purpose.

But if you want them to use the same delta-V, then that purpose is purely academic for the point of the game - unless you're impatient. If you use the same amount of fuel to get to orbit either way, then a more aerodynamic build is purely academic - which was my original point.

So again I will ask - in what measurable way would the change in performance make a difference in the game?

[Sean Mirrsen]

Not "them". Just one craft to use a similar amount of delta-V, taken as baseline. A more aerodynamic shape than the baseline will be more efficient. A less aerodynamic shape than the baseline will be less efficient. More efficient rockets will get into space faster, lose less delta-V to air resistance, and be more stable and better controllable in flight. Less efficient, the opposite. The baseline, the "academic" build, would be a sort of a common rocket - not an outrageously bulbous stack of boosters, nor a slim space-bound needle. The "common rocket" should perform similarly in FAR and stock KSP both, allowing better designs to benefit and worse designs to suffer.

I think you keep misreading my intent here.

[HeadHunter67]

Well, that's because there's a lot of ambiguity. How do you propose to decide what's a "better" build? Who arbitrarily decides?This is why I keep asking you about what metric you are going for.

[Diazo]

The other issue in trying to make a direct comparison like that is the stock drag model favors the short, wide rocket design while FAR favors the tall, spindly design because it tries to apply aerodynamics to the drag model.

I'd argue that using a design with a single stack is not valid for comparison because when aerodynamics get applied, the single stack favors FAR quite heavily.

At the other end, a rocket 5 stacks wide but only 1 tank high favors the stock drag model too much and FAR comes out quite badly in comparison in that setup.

So where is the line drawn? I see this as a case of apples to oranges comparison and I'm not sure there is a valid line to be drawing at all.

Having said that, I'm going to use FAR and KIDS on all my games from here on out so we'll see how it goes.

D.

[InfiniteDice]

A more aerodynamic shape than the baseline will be more efficient. A less aerodynamic shape than the baseline will be less efficient. More efficient rockets will get into space faster, lose less delta-V to air resistance, and be more stable and better controllable in flight. Less efficient, the opposite.

Speaking loosely, I'm not a aerodynamics specialist.

Efficiency moving in the direction of least resistance is one thing, a sheet of paper parallel to flow vs perpendicular to flow. But control efficiency is the manipulation of that flow, in other words specific drag. You won't be able to turn at all without reducing aerodynamic efficiency.

What should be better in system of multiple parts (ksp) would simply be averaging the drag values of all the parts in the vessel. Then one part of a vessel could have a higher cd than that of the whole vessel. etc.

What shouldn't have any impact is a part that has no contact to the airflow and therefore wouldn't affect atmospheric drag, like the engine in your car... the seats... etc.

The tricky part is determining what parts are averaged and what parts are excluded when they are internal or inside the outer skin of the vessel.

I suppose before takeoff some fancy code could scan the outside of the entire vessel using a raycast, and mark all the outside parts as drag-affecting, then only average those....

Fun stuff!

[Sean Mirrsen]

Well, that's because there's a lot of ambiguity. How do you propose to decide what's a "better" build? Who arbitrarily decides?This is why I keep asking you about what metric you are going for.

Um.. the rocket's efficiency? What other metric is there to go for?

See, the way FAR works, there is a "perfect" rocket design. A perfect rocket shape, that, given the same TWR and delta-V, will achieve maximum performance. Stock KSP has no such preference, as only the mass and combined drag coefficients count - so long as the rocket is symmetrical.

So, FAR in the end decides what is a "better" build. Kinda like it does now. The task of choosing the baseline rocket falls to us, however. The stock Kerbal-X seems to be a decent example of a typical stock rocket - full asparagus with a wide-ish base, short main stack with a Mainsail, and a heavy-ish lander that can only barely go to the Mun and maybe back. It has some room for improvement performance-wise, but that's the whole point.

It's really a matter of experimentation. Maybe if FAR opened its aerodynamics coefficients in its config file...

[ferram4]

FAR does open its coefficients in the config file... if you read the readme file you would see the documentation for that (why does no one read the readme?). I used to manually specify the coefficients in the configs, but that doesn't work for thrid-party mod parts. You can still manually apply coefficients if you like.

However, you can get what you want quite easily: Make a plugin that grabs the atmospheric density functions for each planet and multiply each by ten. There, drag is increased, dV requirements are back where they were, and you don't have to go through each and every config file for each mod ever made to make sure it is compatible with FAR.

[HeadHunter67]

Um.. the rocket's efficiency? What other metric is there to go for?

In. the. game. I don't mean to sound contentious, but that's the only reckoning that matters to this discussion.

"Efficiency" is ambiguous. When I say "by what metric", I mean specifically what significant, measurable statistic that actually affects something in-game. I've already explained why aerodynamic efficiency is only relevant as regards fuel savings, because time-to-orbit doesn't matter if the same amount of fuel is expended.

Right now, there really aren't very many such significant metrics in-game - the most relevant one is, of course, delta-v. Is there another means of measuring "efficiency" that actually matters in-game?

[Camacha]

IMO, FAR shouldn't neglect the gameplay aspect of the game: making something realistic, with the consequence of making it easier, is counter intuitive and not something you seek when looking for realism.

I think you miss the point of FAR: properly (.ie they adhere roughly to real world requirements) built rockets are easier to launch and fly. Badly constructed vehicles are a disaster to operate. That this really works is evident; so many people complain that FAR is making KSP impossibly hard. Those are the people that are used to extensive asparagus staging and stuff like that. Also, building planes is a lot more challenging than in stock, but makes a lot more sense too.

If you really feel that KSP with FAR is too easy, try it combined with the new ferram4 mod. It adjusts the ISP to more real world numbers, making everything (possibly) a lot harder. It was made to adress similar issues to yours.

Edit: don't mind me, I am only stating things that have been said. Never mind, move along, nothing to see here.

Edited October 17, 2013 by Camacha

[Frederf]

I appreciate and perhaps more reassuringly understand what is desired by the original poster. I restate that FAR-like mechanisms and features are desired with overall craft performance be similar to standard design craft within the default environment. The result of this scheme is to have the desirable features and intracacies of FAR with comparable craft mission performance for difficulty balance purposes.

My reaction is to estimate that the kind of person that develops FAR is unwilling to compromise the physical fidelity of their system in order to achieve an arbitrary (even if common) desired level of challenge. Further the cause for mission performance of a higher fidelity scheme in a certain environment should not result in faulting the accurate modeling of the world but rather the world that when accuracy is applied to it results in an undesirable way.

[Van Disaster]

So to paraphrase, once again you want to replace one arbitrary set of physics with another arbitrary set? once again, who decides the rules of this arbitrary physics? who decides what is hard, balanced, challenging, etc? quite a few people have reported horrendous difficulty making spaceplanes, especially with FAR, yet I can throw one together in a few minutes. If I constructed a world ruleset that I'd struggle with where would that leave people who have it hard now? yet that would be *my* balance. Real physics are a neutral third party and the ultimate balance.

If you're making a simulation - let's note that outside atmospheres KSP *is* a simulation - it's utterly insane to not use real physics; there's no balancing work, no trying to work out some ruleset from scratch, you've had countless humans do it all for you. As Ferram has said, change the environment to make it harder if you must; nobody seems to appreciate how much work it is to completely define how objects interact with the environment and each other.

Lastly; if you're not happy with the mod, don't bitch about it in some attempt to get a personalized copy from the author, make your own. The source is right there to study.

[Sean Mirrsen]

"Make your own" implies making an entire new mod for what is essentially a tune-up of the existing one. I don't want FAR to permanently change. I just want to be able to make FAR comply with a specific set of requirements, without compromising it for other players.

KSP is set in a universe that is only an approximation of a real one, and it makes sense that physical laws in it would be, in at least some form, only approximations of real ones. Having fully realistic simulation may have undesired consequences. What FAR does alters the physics laws of the universe Squad have constructed, and at least in some respects, does it to the detriment of the game.

In this case, the undesired consequences include making rocketry far easier than is intended by the core game. Spaceplanes are more difficult to make owing to the more complex/realistic rules of aerodynamics that FAR introduces, and that is perfectly fine. That's actually a very good thing, spaceplanes and SSTOs aren't supposed to be easy. (now if only it applied its airflow-shielding mechanics to air intakes... unless it actually does it already) Increasing atmospheric density could be an answer to the rocket problem, as Ferram points out, but:

However, you can get what you want quite easily: Make a plugin that grabs the atmospheric density functions for each planet and multiply each by ten. There, drag is increased, dV requirements are back where they were, and you don't have to go through each and every config file for each mod ever made to make sure it is compatible with FAR.

Would it not be far easier for you, Ferram, to make controls for those changes in FAR, and expose them in the config? There is no need for a plugin separate from FAR that does this, and you already have the functions in place - all you need is a multiplier variable for atmosphere density in them. Just for the lift and drag, no need to supercharge air intakes and hobble LFEs.

I do admit that despite being a little bit of a programmer (stats in my sig are relatively up to date), I have next to no idea on how KSP plugins are put together yet, which is really why I am saying that. If I undertook that task I'd spend a week learning how basic things are done, and ultimately wouldn't do as good a job of it as the mod's original author could.

[Diazo]

KSP is set in a universe that is only an approximation of a real one, and it makes sense that physical laws in it would be, in at least some form, only approximations of real ones. Having fully realistic simulation may have undesired consequences. What FAR does alters the physics laws of the universe Squad have constructed, and at least in some respects, does it to the detriment of the game.

I'd argue that Stock KSP does not have a aerodynamic drag model, period.

Stock drag is "add up the drag values of each part to get the vessel's drag". A tall spindly rocket flying sideways has the same drag and a tall spindly rocket flying straight up in Stock KSP.

FAR then goes and adds aerodynamics which makes things easier because that is how rockets actually work.

Rockets are tall and spindly to reduce drag and FAR rewards this. Does that make the game easier for rockets? Yes, but I blame Stock KSP's drag model for that, not FAR.

As always, to each their own, FAR is a mod after all. If it makes rockets easier, take advantage and make bigger rockets. Or don't use it.

I'm going to keep using it as I like being rewarded for good design.

D.

[Tiron]

FAR's actually a crapshoot, difficulty wise. Yes, it cuts delta V to orbit by about a quarter. It also makes design and piloting much, much harder. Screw up either, and your rocket is going to end up with the wrong end pointing towards space, just from aerodynamic forces. You can literally end up with a completely intact rocket that's flying backwards because you gravity turned too hard. Or a really bad, inefficient trajectory, because you didn't get turned enough. I've had a lot of problems with that in 0.22, because my rocket designs so far have had really high thrust and speed, and thus really high aerodynamic forces that limit their ability to pull a gravity turn. Planes are an order of magnitude more difficult than rockets because stalling suddenly matters, going supersonic actually changes how your plane develops lift, any amount of stalling can put you in a spin that is completely unrecoverable...

There is one thing you don't see mentioned as often too, but it comes up: the changes in the drag modeling work both ways. FAR makes it possible to go much, much faster at low altitude. For example, if you're coming in to Kerbin from the Mun on a really steep trajectory, it's entirely possible to hit the ground still doing well over 1000 m/s. It's also entirely possible that your design won't ALLOW you to keep it in a high drag orientation: Planes for example, if designed to be stable, will correct themselves to fly nose-first once the drag gets high enough.

This makes re-entry somewhat more challenging, because you have to make sure that you can actually get slowed down enough by the greatly reduced drag. Pair it with deadly re-entry and you could have some serious problems, re-entry wise.

[Sean Mirrsen]

The design and the piloting I have absolutely no problem with. If it throws a challenge at me because of the improved aerodynamics system (or, yes, just aerodynamics system, as stock KSP doesn't really have one), then I will take it on and solve it.

What I dislike is the reduction in difficulty that results. I regularly build designs that fly well, FAR or no, but I can't keep turning FAR on and off if I want to alternate between launching a rocket and a spaceplane, because the difficulty is dramatically reduced.

I can put a non-aerodynamic 1-seat lander can on top of a small SRB, and that on top of a big SRB, and that configuration will easily achieve orbit with a periapsis halfway to the Mun. That's... really not encouraging good design. Not compared to stock. If a bad design flies better in FAR than in stock - and by "bad" I mean "shaped like a flat-top barrel" - then you aren't really improving anything. Yes, you are likely to need a few wings and control surfaces to keep the barrel-rocket from tipping over, but it simply shouldn't be as effective as it is, with FAR doing what it claims to.

(note: I may be misremembering details of that lander-can/SRB thing, but I did do several tests with that, and that was one of the results I got - I just don't remember which it was at the moment, and I uninstalled FAR in lieu of playing through the new career mode)

[rhoark]

The hard part of launching a rocket from Kerbin with FAR is not in having enough dV. It's possible to simply come to peace with that.

[Tiron]

What I dislike is the reduction in difficulty that results. I regularly build designs that fly well, FAR or no, but I can't keep turning FAR on and off if I want to alternate between launching a rocket and a spaceplane, because the difficulty is dramatically reduced.

I can put a non-aerodynamic 1-seat lander can on top of a small SRB, and that on top of a big SRB, and that configuration will easily achieve orbit with a periapsis halfway to the Mun. That's... really not encouraging good design. Not compared to stock. If a bad design flies better in FAR than in stock - and by "bad" I mean "shaped like a flat-top barrel" - then you aren't really improving anything. Yes, you are likely to need a few wings and control surfaces to keep the barrel-rocket from tipping over, but it simply shouldn't be as effective as it is, with FAR doing what it claims to.

(note: I may be misremembering details of that lander-can/SRB thing, but I did do several tests with that, and that was one of the results I got - I just don't remember which it was at the moment, and I uninstalled FAR in lieu of playing through the new career mode)

First, there is such a thing as 'Fake Difficulty'. In general, Fake Difficulty consists of things that make a task in a game difficult that aren't actually related to the difficulty of the task. Bad camera controls, lack of information, lack of feedback... a few examples. Sometimes developers add them on purpose, in order to extend gameplay time or make a task more difficult that they're either unwilling or unable to make legitimately difficult. KSP has a few of these, almost entirely because it's incomplete.

The high drag losses are an example. They're caused EXCLUSIVELY by the invalid and ridiculously unrealistic placeholder drag model the game uses. My original, stock aero, hypersonic transport plane had a top speed that increased from 1600 m/s to over 1900 m/s, and a maximum altitude for full throttle that increased from 23000m to over 26000m as it burned off fuel, because the fuel tanks have lower drag when empty than they do when full. This means that, for example, you could vastly reduce your drag losses by using the realfuels subbranch of modular fuels to switch to Liquid Hydrogen fueled rockets, which is absurd.

The sad thing is that, unrealistic as it is, it doesn't really substantially increase the difficulty. You just need a slightly bigger rocket, and the primary limitation on building bigger rockets is the single-threaded physics limiting the ability of your system to run a high part count. It also vastly reduces the effort required to design an effective rocket, increasing the utility of things like Asparagus Staging and making Pancake rockets viable.

As for the SRB thing...the lander can is really really light, and the SRBs have a really, really high TWR. Also, 'Flat Front increases drag' isn't universally true: At high enough speeds it can actually REDUCE drag. Many rockets these days use a device consisting of a small disk on a spike sticking out the front of the rocket. It reduces the overall drag by creating a sort of wake that encompasses almost the entire rocket. It's a principle similar to supercavitation, and also similar to the blunt-body method of designing re-entry vehicles. The lander can on top of SRBs would realistically do something similar, although not as efficiently as using a smaller surface further out in front: the lander can would have enourmous drag on the face of it, but everything behind the front face would be within the low-pressure wake and have very low drag, once it got up to speed. To a point, anyway.

It's also possible that it was a bug in FAR that got fixed in the meantime. I'll have to try this myself when I get back to Kerbin.

[rkman]

I have never heard FAR makes getting to space too easy.

FAR significantly reduces delta-v requirements (and thereby fuel requirements, and launch vehicle mass) for atmospheric ascent (by some 30%). That makes it easier. How much easier could it get before being to easy?

Umm, to me that means FAR is working?

FAR would also be working if it would cause an aerodynamically unfit rocket (the sort that is typical for non-FAR KSP; build wide, not necessarily with fairings, nose cones etc) to require more delta-v to reach LKO than it does with the stock drag model. Imo FAR should - relative to the stock drag model - 'punish' when you do it wrong, not make it easier when you do it right. Doing it right would of course still make it easier wrt fuel requirements, than doing it wrong.

I'd think it would require little more than adjusting either an atmospheric density constant or a surface friction constant.

[ferram4]

So the consensus here is that if only FAR would maintain the current delta-V requirements (and hence, drag characteristics) for an "ideal" rocket, everything would be fine. For most of our cases, the Kerbal X would be rather close to ideal; Sean Mirrsen's SRB stack is also fairly "good," in terms of having a high fineness ratio (length / diameter) and a lack of sudden changes in cross-section across the length, with the only significant problems being the lack of a nosecone to smooth out the drag at the top. And then the consensus seems to be that the standard asparagus monsters and wide-bodied pizza rockets should somehow have even higher drag.

My problem with this is that an increase in atmospheric density / increase in drag coefficients to keep the delta-V requirements will reintroduce the requirement of a 10km vertical climb before any type of pitch over maneuver can be considered, which is not necessary with FAR's current drag characteristics, since it is possible to begin a gravity turn under 1km up and still reach orbit. I honestly do not understand the gameplay benefits that would come from reintroducing this vertical climb, since I have always taken it as an opportunity to leave SAS in control of my rocket while I get a drink. If someone can honestly explain the gameplay benefits inherent in reintroducing that (IMO boring) segment of launch (which will be a necessary consequence of increasing atmospheric density / increasing drag coefficients), I'll look into adding a switch so that FAR can increase density for the people who want it.

I simply don't understand what benefits would come from that that wouldn't be better served by rescaling Isps; the delta-V requirements might drop, but with lowered Isps the same size rocket would be required to get to orbit.