UmbralRaptor changed the topic of #kspacademia to: https://gist.github.com/pdn4kd/164b9b85435d87afbec0c3a7e69d3e6d | Dogs are cats. Spiders are cat interferometers. | Космизм сегодня! | Document well, for tomorrow you may get mauled by a ネコバス. | <UmbralRaptor> … one of the other grad students just compared me to nomal O_o | <ferram4> I shall beat my problems to death with an engineer.
<kmath>
<MiraVylash> imagine seeing the words "no more nazis" and reacting like this https://t.co/5L9b8CPm3s
e_14159 has quit [Ping timeout: 204 seconds]
e_14159 has joined #kspacademia
<egg|afk|egg>
bofh: unrelated: due to a sequence of silly identifier choices I have something at work that I could reasonably call a catbus
<egg|afk|egg>
bofh: it's an omnibus of catalogues, you see,
<bofh>
egg|afk|egg: ...I am disturbed that that's sensible as a descriptor there.
<UmbralRaptor>
egg|afk|egg: Please do not create documentation that mauls the user.
* UmbralRaptor
likes the name.
<SnoopJeDi>
My Neighbor Tycho
<egg|afk|egg>
bofh: so tbh I have nothing formally called a catbus, I just have an Omnibus (it took us a week to find that name :-p), which has Catalogues
<egg|afk|egg>
but maybe I can call the whole API the catbus :D
* egg|afk|egg
stares at the ネコノミコン
<UmbralRaptor>
Summoning eldritch felines is very appropriate this month.
<egg|afk|egg>
is Fiora an eldritch feline
<UmbralRaptor>
Depends on how lovecraftian^W numeric code she's working with.
<kmath>
<diodelass> PEOPLE: please stop, we like headphone jacks APPLE: no you don't GOOGLE: see, they said you don't PEOPLE: no
<soundnfury>
egg|afk|egg: how long 'til you have a conifer API?
<SnoopJeDi>
Iö! Iö! Fh.flushtagn()!
<soundnfury>
in his process at r'lyeh, thread Cthulhu sleep()s dreaming
<SnoopJeDi>
LOL
<SnoopJeDi>
from space import colour
<soundnfury>
no, no
<soundnfury>
actually, what I was about to say is unfunny and insufficiently related, so nvm
<Ellied>
UmbralRaptor: come again?
<UmbralRaptor>
Ellied: As in asking about why $thing_people_use was removed and being told no one uses it.
* UmbralRaptor
?
<Ellied>
oh, google's saying people say they don't
<UmbralRaptor>
Not sure in this case, but was totally told that about keyboards, sdcards, and removable batteries.
<Ellied>
lol ugh
<Ellied>
"Hey, this thing we used to put on our devices that is almost universally liked is cramping our style for reasons you don't care about. How about we just insist you were the ones who wanted it removed until you get tired of arguing and give up?"
<SnoopJeDi>
in defense of people who do not deserve defending in the slightest, the smartphone market has always been about brands telling you what it is you need
<Ellied>
my friend at school holds that the whole reason phones are constantly made thinner was never about customers actually liking thin phones and just about being able to ship more of them for cheaper
<Ellied>
presumably it costs about the same to ship a crate of 500 iPhones or 200 TI graphing calculators.
<whitequark>
APlayer: you don't understand anxiety disorders
<whitequark>
it's *literally impossible* for me to voluntarily submit myself to dental treatment
<whitequark>
I'd rather get some horrible complication and die
<whitequark>
(sedation means involuntary of sorts which is why it works)
<soundnfury>
whitequark: I can understand being terrified of being dentisted
<soundnfury>
but I struggle to understand not being _more_ terrified of dentisting oneself :/
<soundnfury>
SnoopJeDi: this is one of the reasons I stubbornly still don't have a smartphone
<SnoopJeDi>
well, it's not particularly unique to phones either, but yea I hear you
<whitequark>
soundnfury: what's the big deal
<whitequark>
I don't have to wait, I don't have exposure to the environment, all of which make anxiety much worse
<whitequark>
I feel in control (regardless of the reality of the situation)
<whitequark>
seems very straightforward
<SnoopJeDi>
isn't being in control a known thing with fear of pain specifically?
<whitequark>
it wasn't painful
<whitequark>
it was the opposite of painful
<whitequark>
because root canal
<whitequark>
every time I took out a chunk of the nerve I instantly felt *much* better
<whitequark>
it's an incredible positive feedback loop
<SnoopJeDi>
oh, my mistake, I assumed the anxiety over the dentistry was related to anticipation of pain in general
<whitequark>
not quite
<soundnfury>
(and yes, I also "don't own a television", and am smug about that too https://xkcd.com/1299/ )
<whitequark>
it's an anxiety *disorder*. i don't have a problem with *surgeries* per se, be it on my teeth or on my whatever else
<SnoopJeDi>
I should've probably read earlier scrollback before jumping in
<whitequark>
i'll gladly watch it as it proceeds, provided the involuntary bodily response gets shut down
<SnoopJeDi>
but yea, anxiety is kinda by definition not rational
<whitequark>
there's no real pain in dentistry, they use local anesthetics these days...
<whitequark>
like if there is find a better dentist
<whitequark>
it's not even disgusting anymore. i don't know how it's called in english but the dentist has shown me some weird kinda contraption that isolates the tooth from the rest of oral cavity
<soundnfury>
whitequark: so, thinking back to when I had a root canal done, _yes_ it's scary not really knowing what's happening
<soundnfury>
but for me at least, I _really_ don't trust myself to know what I'm doing
<whitequark>
there's not exactly a lot to it
<whitequark>
there's pulp inside tooth. get the pulp out.
<whitequark>
try not to destroy hard tissues of the tooth too much
<whitequark>
so, use some relatively soft wire and be careful.
<soundnfury>
it'd be like a more scary version of fiddling with computer hardware
<soundnfury>
"I'm going to break something, I just know it..."
<whitequark>
it generally helps to look at tooth anatomy for that tooth beforehand.
<whitequark>
oh.
<whitequark>
i don't have that
<soundnfury>
Evidently ;)
<whitequark>
i have "I'm going to need a JTAG cable again, I just know it..."
<whitequark>
also I have a lot of dissociation
<whitequark>
generally
<soundnfury>
heh. At work when it gets to "need a JTAG cable" levels, there's someone else I can drop it on :)
<whitequark>
but that's slow!
<soundnfury>
hmm. If I'm not misunderstanding what dissociation is… it's possibly the scariest thing I've experienced.
<whitequark>
I've also experienced *really* severe levels of dissociation on [redacted] and that was quite fun actually
<soundnfury>
do you get used to it if it happens a lot? I get — I think — derealisation maybe two-three times a year, lasting about twenty minutes. Having it as a regular feature seems like it would be hella debilitating
<whitequark>
honestly, people report all sorts of experiences as unpleasant but that's because they're fucking junkies and have no understanding of the underlying biochemistry, no idea of correct dosage, no culture of safety, and just an incredibly wrong approach overall
<whitequark>
hm
<whitequark>
I rarely get derealization anymore, it mostly came with severe bipolar anxious depression
<whitequark>
my roommate has spent her entire life derealized
<whitequark>
she says it's vastly preferable to being realized
<soundnfury>
… yikes
<soundnfury>
um, if she's always been derealised, how does she have a comparison datum?
<whitequark>
ketamine (a dissociative. ironically.) consistently induces realization in her
<whitequark>
"jesus christ IS THIS ALL HAPPENING FOR REAL? NO I WANT BACK" is the usual response when it happens to fall along that axis
<soundnfury>
0_o
<soundnfury>
actually thinking more about it, what I get seems to have all the indicia of derealisation (unfamiliarity, jamais vu, rootless anxiety) except for the unreality itself
<kmath>
<Alex_Parker> Colleagues caught the shadow of Neptune's largest moon Triton, a world that started out as a sibling to Pluto. https://t.co/BZ4VQfvkMG
<Ellied>
mostly I'd describe it as a sudden intrusive doubt that the situation I'm in is real. I start trying to think of ways of testing whether it's real or not, but consistently conclude that all of them would require trusting sensory feedback alone and won't work if such is to be doubted. ultimately I agree with myself that my ability to conclude this and think lucidly about it means that my brain is
<Ellied>
functioning well above the level I'd expect for dreaming, which is the most likely non-reality I'd be in, and kinda give up and move on to other things.
<Ellied>
I've probably gone through the same approximate routine of thoughts hundreds or maybe thousands of times over the course of my life.
icefire has quit [Read error: -0x1: UNKNOWN ERROR CODE (0001)]
awang has quit [Read error: Connection reset by peer]
<kmath>
<marcelsalathe> Fellow Swiss, you know what the Nobel to J. Dubochet means: we must eat more chocolate to maintain the correlation!… https://t.co/jRcZ3gyjn8
<UmbralRaptor>
egg|work|egg: sell more chocolate to Swedes?
<rqou>
maybe you really should explain to me how to do everything with tensor products
<rqou>
e.g. i have no idea why there is a differential of a matrix
<rqou>
or how a matrix can turn a position into a velocity
<rqou>
and then somehow jacobians are involved
<egg>
rqou: I'll get back to you after dinner
<rqou>
yeah, I have a continuous slew of classes coming up anyways
awang has quit [Read error: Connection reset by peer]
awang has joined #kspacademia
icefire has joined #kspacademia
<egg>
whitequark: random weirdness of en-FR: in french, nuclear fuel is "combustible nucléaire", so you might see the rather nonsensical "nuclear combustible" under a french pen
<kmath>
<AdamBienkov> Goodhart blames rise of left on more young people going to university then adds inexplicably "it's all there in the Captain Underpants film"
<Iskierka>
does Ignition! have a section on nuclears?
<UmbralRaptor>
ICBM propellants only.
<Iskierka>
somewhat disappointed
<Iskierka>
clearly it needs a sequel
<bofh>
I'm pretty sure such a sequel would be classified heavily.
<whitequark>
egg: huh
APlayer has joined #kspacademia
<APlayer>
Oh god. Guys, I need some scientific babble to calm my mind
<APlayer>
So there I go peacefully to YouTube because I saw a video of a high altitude plane and thought thered be some nice shots. Then I go into the comments and see people argue that there is no curvature visible, so the earth is proven flat
<UmbralRaptor>
You can use uBlock to block all comments.
<APlayer>
And I just couldn't stop reading for an hour or so. It's like you know, some sort of addictive. You can't stop once you started. Only with a lot of effort I managed to stop reading and did other stuff. But that trash was in my mind for the whole evening now
<APlayer>
UmbralRaptor: Well, I learned my lesson for today. YouTube comments are legit psychologically harmful
<APlayer>
Anyway, hoped I could find some mind fuel here to get something else to think of
<APlayer>
(!wpn everyone)
<APlayer>
ferram4: Sorry for pinging you without a true issue... Would you have some time to try to explain me Bernoulli's principle intuitively?
<APlayer>
If not, I am fine with asking someone else or asking you at a different time
<ferram4>
What part of it do you have difficulty with?
<APlayer>
Why is pressure dependant on the velocity?
<ferram4>
Alright, so let me ask you this: what IS pressure, on a molecular level?
<APlayer>
Molecules colliding and pushing themselves in all kinds of directions, statistically outward, though
<ferram4>
Yes. And those molecules have momentum. Momentum must be conserved.
<ferram4>
When you increase the fluid velocity, you have bulk momentum in a certain direction. That momentum needs to come from somewhere.
<APlayer>
So they simply trigonometrically move at a shallower angle against a rigid moving body?
<ferram4>
When you work out the math, that should be what happens.
<ferram4>
It gets more complicated when you include variable density though.
<ferram4>
Then you need to account for the change in count of molecules as pressure and density increase.
<APlayer>
And (did no research here, though) that is what the Navier-Stokes Equations describe? What makes them so impossible to solve?
<ferram4>
They are conservation of mass, momentum, and energy for a continuum fluid (read: we assume that molecules can be averaged away completely) accounting for ALL forces.
<APlayer>
That sounds like an exciting opportunity to compress aerodynamics as a subject into code that may simulate it
<ferram4>
The problem is that they're nonlinear partial differential equations.
<ferram4>
Which are horrendous to solve.
<APlayer>
But thay actually work? Or is that some kind of approximation to get a general idea of how things work? Like patched conics is an approximation for orbital mechanics?
<ferram4>
The best we have right now are numerical solutions on a grid, with fucktons of iterations to solve them.
<ferram4>
They actually work, up until you go too far down in scale.
<ferram4>
To the point where the size of molecules is important.
<APlayer>
Oohh
<ferram4>
Problem is, the effect of turbulence tends to lead to the grid size needing to get smaller and smaller at certain points.
<ferram4>
Which 1) make things very expensive and 2) at some level starts getting to the point where maybe the continuum assumption isn't actually perfectly valid anymore.
<ferram4>
Generally that's where turbulence models come in to approximate things and get us a usable solution before the heat death of the universe.
<APlayer>
I mean, fluid dynamics always was a very weird thing to me. Seems to be governed by mostly simple and predictable laws (e.g. pressure always strives to distribute evenly) but so impossible to actually predict, it's weird
<APlayer>
turbulence models?
* APlayer
stops asking before using Google
<APlayer>
BRB, using Google :P
<ferram4>
Approximations of how viscosity, thermal conductivity, and sometimes other characteristics (generally some kind of turbulence factor) are transported through the fluid.
<ferram4>
It's a way of modelling what happens below the grid scale.
<APlayer>
Ah
<APlayer>
And can't those be somehow combined to prodice an efficient compromise model?
<APlayer>
produce*
<ferram4>
That is what's done.
<Iskierka>
efficient models are inaccurate models
<ferram4>
^^
<ferram4>
You come up with a moel that's fast enough to see if your design is at least, not completely retarded.
<APlayer>
Efficient as in CPU time to accuracy ratio
<ferram4>
And then use more expensive ones to refine things.
<ferram4>
And also spend lots of time checking your simulation grid to make sure that what you're finding is real and not just an artifact of the grid.
<Iskierka>
and I'm not sure if it's even mathematically possible to accurately model turbulence. Which is why, decades on and with absurd computing power, new aircraft simulations can still be as much as 25% out without wind tunnel calibrations
<APlayer>
That seems like a horrible weakness in humanity's knowledge
<APlayer>
But it's probably one of the few things nobody has tried to mathematically work out centuries ago
<ferram4>
Yep. Turbulence is less understood than quantum.
<Iskierka>
Not when you consider that turbulence comes from quantum interactions of E+stupid number of particles on a massive scale
<Iskierka>
which we then have to approximate with continuum assumption
<ferram4>
But which very likely has key aspects that go down below the scale where the continuum assumption is valid.
<Iskierka>
and there's some pretty early attempts to understand airflow, actually. While he came up with largely nonsense daVinci didn't come up from nothing
<Iskierka>
we've only been good at it for the past century-and-a-bit, but there's been millenia of at least thinking about wind. and we still don't really understand the finicky details
<APlayer>
Well, save for Davinci
<ferram4>
I still like Newtonian flow. Terrible for ship design. Excellent for hypersonic.
<ferram4>
Well, for first approximations anyway.
<APlayer>
ferram4: And how do "general public" simulations like FAR and KSP handle aerodynamics?
<Iskierka>
<ferram4> But which very likely has key aspects that go down below the scale where the continuum assumption is valid. <-- and this bit is why it's probably not mathematically possible to accurately model turbulence
<Iskierka>
because if it goes down that small, and in particular if that key aspect is quantum so hello heisenberg, shroedinger, glhf simulating
<ferram4>
Well, for a start, they take the Navier Stokes equations, say, "that's a very nice bit of theory you've got there, but I can't afford it"
<ferram4>
And then they look at the next steps up and say, "that's a very nice bit of theory there, but I'm pretty sure we still can't afoord it"
<APlayer>
Iskierka: Well, I trust we will have one day enough computing power to brute force simulate a fluid of molecules
<Iskierka>
It is very likely that would take a computer larger than the fluid being simulated, IF you could model the quantum effects (quantum theory says you can't)
<Iskierka>
barring the rather fun technicality that, well, we could just compute it with flow tunnels and models
<Iskierka>
then it's the same size as the fluid being simulated
<APlayer>
So aerodynamics on a macroscopic scale indeed directly depends on quantum mechanics?
<ferram4>
And then they start grabbing prelim design coorelations and theoretical solutions under rather restrictive assumptions. Then they try to cludge them together as best as possible.
<Iskierka>
As viscosity and thus turbulence come from electromagnetic interactions of the particles, yes, there's probably quantum aspects
<ferram4>
And then add some quick bullshit approximations to handle, "well shit, there's no data on the aerodynamics of a cancer-covered unicorn at Mach 0.94, I'll just make sure drag is positive"
<Iskierka>
good chance it's quantum aspects that are currently being proven NP rather than P, too
<APlayer>
ferram4: Well, some of them do it satisfyingly well
<ferram4>
And then, if you're KSP, you do things that even the local aero modder expert told you, "please, for the love of god, don't do this, it's terrible and wrong and evil"
<Iskierka>
To a large extent with real-time sim you just have a rough model of "this is maybe how much drag and lift an inefficient body produces" and a second model of "this is a wing, is designed to have efficiency, use this instead"
<Iskierka>
Which with good data, can do remarkably well. I'm reminded of an RC VTOL someone designed with the help of X-plane, which predicted some undesirable behaviour that the real model had
<ferram4>
A lot of the trouble is separating efficient and inefficient.
<APlayer>
And doesn't aerodynamics boil down to pressures and their interactions?
<Iskierka>
it's just definitely best-guess at best
<Iskierka>
It boils down to that, and then to how that pressure is made
<Iskierka>
and then you're back to individual particles or wobbly continuum assumptions
<ferram4>
I mean, what's better in KSP, a rocket with a nosecone attached the correct way, or a rocket with a nose cone attached to the same part, but flipped 180 degrees so that it's blunt end forward?
<ferram4>
Stock KSP, I mean.
<Iskierka>
... I get the feeling the answer is "it depends, but the second CAN be as good as the first"
<ferram4>
The second is always better than the first, so long as you go fast than ~Mach 0.7 or so.
<Iskierka>
but there's probably ways to attach it that the engine goes "Oh! Wait, that's blunt! I'm supposed to drag here!"
<Iskierka>
Or maybe not :p
<APlayer>
I get the feeling that specifying what way a nosecone is attached is already weird
<APlayer>
:P
<ferram4>
Look, the basic lesson of KSP aero is: lego-brick-based aerodynamics is cancer.
<Iskierka>
I'm tempted to say that my original half-made mod around the same time FAR started, which just cut the drag multiplier so that air wasn't 8x denser than it should be, and gave wings v^2 lift, would possibly be better than current stock
<APlayer>
And wouldn't it make lottle difference for the typical rocket if you did not simulate turbulence at all, but only simulated drag of anything that is not covered by another part?
<Iskierka>
it would have all the oddities of back then with mass-based drag, but it was predictable and easy and you knew more parts was more drag
<Iskierka>
FAR is predictable and easy if you know real aero
<Iskierka>
current stock is "wait what?"
* APlayer
knows no real aero :/
<Iskierka>
You can probably guess it better than you can guess current stock
<Iskierka>
"That looks like a plane. It should plane"
<APlayer>
Well, perhaps
<ferram4>
APlayer, at real-time scale, turbulence is simulated as, "ALL FLOW IS TURBULENT!!! USE THE WELL KNOWN SKIN FRICTION AND THERMAL TRANSFER CORRELATIONS" and we ignore the rest
<ferram4>
*Well known flat plate
<APlayer>
One particular thing I wondered about for some time is the shape of payload fairings, that it's pretty blunt actually. Does that create some sort of turbulence that reduces skin drag or something?
<Iskierka>
Reduces nose lift and hypersonic heating
<ferram4>
Iskierka, one problem. THe capsules will have EVEN HIGHER terminal velocities.
<Iskierka>
though the latter could probably be tolerated
<Iskierka>
That's just part of the fun
<ferram4>
...you'll also need to revert the parachutes breaking then
<APlayer>
That's why you use RealChute
<APlayer>
Although I have to say, I hate its UI
<ferram4>
Then the parachutes will still break.
<APlayer>
But they will deploy smarter
<ferram4>
Believe it or not, parachutes don't like being openned in high-density supersonic flow
<APlayer>
So they will not deploy :P
<APlayer>
Iskierka: But how does it reduce lift? I mean, if the rocket flies a true gravity turn in the relevant part of the atmosphere, there should be no lift at all?
<ferram4>
APlayer, also back to your "Only simulated drag of anything that is not covered by anothe rpart" Trust me on this: determining if something is covered or not is non-trivial.
<APlayer>
Also, how do you know all of that? :D
<Iskierka>
a "true" gravity turn is a poop way to get into orbit and you would still get lift
<Iskierka>
also I do remember ideas of trying to take a render of the craft in airflow for calculating drag
<APlayer>
Well, no, not a cone actually. Just a diagonap from every point of the part's surface at an angle determined by velocity
<ferram4>
Iskierka, I remember doing something like that. Unfortunately, it results in lots of really weird things. Like airplanes getting a fair bit less stable if the wings eclipse the horizontal tail. And no good method for base/tail drag
<APlayer>
diagonal*
<Iskierka>
and hello computational complexity
<APlayer>
Well, seems easier than what I heard just before
<ferram4>
Every... point. How far apart we talkin'?
<APlayer>
And that is more like ray casting than aerodynamics already
<Iskierka>
For base/tail drag, I think you could probably do something with just taking two renders of both ends? But inevitably you'll always have things mutually eclipsing so still weirdness
<Iskierka>
you'd have to have no drag-torque which makes no sense
<UmbralRaptor>
!wa conifold
<Qboid>
UmbralRaptor: conifold (English word): noun | (mathematics) a certain generalization of a manifold
<APlayer>
ferram4: Well, since parts in KSP are polygons, I'd take every corner point of every polygon
<ferram4>
APlayer, very well, I raise you this:
* UmbralRaptor
blames egg.
* Iskierka
attempts to parse what O() this will be
<APlayer>
Iskierka: O(n)?
<ferram4>
A long rectangular prism with vertices only on the ends, oriented in cross-flow. I now cover the ends in aerodynamic spheres. Do I get high drag from this setup?
<egg>
<ferram4> And then, if you're KSP, you do things that even the local aero modder expert told you, "please, for the love of god, don't do this, it's terrible and wrong and evil" << any eggsamples?
<APlayer>
I fail to see any sort of loop nesting needed herr
<APlayer>
here*
<Iskierka>
raycasting itself is a nested loop
<Iskierka>
it must test everything it MIGHT hit to see if it did
<ferram4>
egg: Besides the general lego-brick-aero is an abomination? Combo of alllowing clipping plus that? Using specific part modules for payload fairings and cargo bays with associated weird behavior on the ends?
<egg>
<ferram4> The second is always better than the first, so long as you go fast than ~Mach 0.7 or so. <<< aaaaaaaaaaaaaaAAAAAAAAA
<Iskierka>
this is why we don't render with raycasting
<APlayer>
ferram4: Sorry, my English at the limit here. Cross flow means perpendicular?
<egg>
ferram4: earlyFAR was legobrickbased, right?
<egg>
also what became of those C++ nuFAR plans :-p
<Iskierka>
as to render 1 polygon you must, for every pixel in that polygon, test a few thousand, versus just check the expanse of the one polygon and fill
<APlayer>
Iskierka: Okay, might need that stuff to determine whather a part is enclosed by that eclipsing diagonal-poly-cone
<ferram4>
APlayer, yes. Airflow perpendicular to the long axis.
<ferram4>
egg, yes, and abandoned on the basis of not being viable as well as missing key aspects. Also, too much trig for fast calculations.
<ferram4>
Also, too linear. Missed too many key behaviors. Too many assumptions violated for blunter things.
<Iskierka>
if you wanted to determine shadowing with some accuracy you *might* have an extremely heavy method in using renders of the part's cross section and overlaying to determine the overlap, but this will get complicated, stupidly intensive, and neglect skin drag
<APlayer>
ferram4: Well, the drag comes from mainly the front surface there, the skin drag is ignored due to the eclipse poly-cone amd as for the aero spheres, I am not quite sure what happens there
<egg>
ferram4: yeah, I remember
<Iskierka>
and therefore would allow something with a lower minimum plan-area-Cd than physically possible
<Iskierka>
APlayer, most drag actually comes from the back if your front isn't stupid but your back is practical and square
<Iskierka>
this is why new truck trailers have the curved back to reduce rear area
<APlayer>
Iskierka: Well, it seemed much easier, and at the very least more understandable, than simulating turbulences
<Iskierka>
It's also why things aren't automatically unstable at speed, they're not being primarily pushed from the front
<APlayer>
Uh, really? How does drag at the back look like?
<Iskierka>
very rough and unstable because the air's being pulled along with the car, so the car's not just moving itself
<APlayer>
Also, would skin drag still be a thing on a hypothetical perfectly even surface?
<Iskierka>
at the front, anything not absurdly stupid can split low-speed air reasonably well
<ferram4>
Short answer: in (inviscid) theory, the pressure at the back is just as high as the pressure at the front. In practice, viscosity prevents full pressure recovery and produces a nice, low pressure recirculating region.
<Iskierka>
also ^ for actual theory on symmetrical bodies
<Qboid>
3d 0h 0m 0s left to event #11: 「みちびき4号機」／H-IIAロケット36号機 [at 2017-10-09 22:00:00]. Say '!kountdown 11' for details
<Iskierka>
for ideal drag you want quite a long teardrop which it's quite awkward to make a vehicle shaped like
<kmath>
<FioraAeterna> @whitequark is a 3 legged cat a partial cat
<Iskierka>
cars can't deal with that long a tail, rockets need to put an engine there, planes can come closest but would *like* to make use of that volume
<ferram4>
And it'll actually end up with slightly more drag, because the flow will separate anyway but might end up being a larger region with a stupid teardrop shape.
<APlayer>
So it's the pressure difference between front and back that pulls backward?
<ferram4>
That's part of it. That's pressure drag.
<APlayer>
Iskierka: Wouldn't a double-cone be better? In the sense that a tear drop has a hemisphere at its front (or back) while a double cone would be nice and pointy?
<Iskierka>
if subsonic pointy front is unhelpful
<ferram4>
I think that, in a perfect zero-AoA-the-world-is-wonderful-and-happy situation they're end up abaout the same.
<Iskierka>
It never is, though
<ferram4>
The world is not wonderful and happy and you get fucktons of extra drag when something like that ends up off 0 AoA
* APlayer
feels like all his aero intuition is totally breaking diwn
<Iskierka>
you can see why from the big blue blob of low pressure that you really don't want
<Iskierka>
and the fact the prius as a much smaller butt to produce that blob with, while preventing flow separation
<ferram4>
Supersonically that works out great. Actually tends to make more sense to have a pointy front end and a chopped-off back end, though I'm pretty sure the chopping-off is to increase stability more than decrease drag.
<Iskierka>
Chopped off definitely increases drag, though possibly less dramatically. You can get compromises though with much less finicky geometry, just choose closing angle for pressure on short cone, or truncated area for corrective drag
<APlayer>
And what happens to a double cone minimally off 0 AoA when subsonic?
<APlayer>
Actually, not even a double cone, it'd have an edge in the middle. Let's round that off and make the surface smoother from front to back
<APlayer>
A teardrop with a pointy end at both sides
<Iskierka>
well known and studied geometry already
<ferram4>
APlayer, off 0 AoA, you see the big recirulating regions behind the hatchback? You get that forming on the away-side from the airflow and it proiduces a lot of drag. Nice amount of lift too, which you might not want.
<APlayer>
ferram4: But that's if it's off AoA by an angle that is at leat a big fraction of the cone angle?
<Iskierka>
I'd recommend trying that lift out. Make rocket in FAR, give very long pointy empty fairing, even throw on okay fins if you like, try fly it to high speed
<APlayer>
Iskierka: Ah, I will bookmark that!
<Iskierka>
this conveniently comes back to the earlier point of pointy fairings
<ferram4>
It works irritatingly well
<ferram4>
As in, "And now my rokcet is in many pieces"
<APlayer>
You already tried it? :o
<Iskierka>
Most people try it, because of course pointy is good, right?
<APlayer>
Well, yes :D
<APlayer>
I'd intuitively ideally build a needle rocket
<ferram4>
I'm the mod developer. I need to verify the behavior.
<ferram4>
"Does pointy result in happy flippouts at high dynamic pressure? Yes? Good!"
<ferram4>
"Users complain? Lrn2aero"
<Iskierka>
related: NEAR
<APlayer>
And why are conuc fairings a thing too then?
<APlayer>
conic*
<ferram4>
Manufacturing easy.
<ferram4>
Cylinders and cones are easy to build.
<ferram4>
Spheres also are, but less so.
<APlayer>
Wouldn't you just mold or press or whatever you do with fairings a certain shape?
<ferram4>
So build a fairing out of a combo of a section of cylinder, two conic sections near the top, one widenning conic section near the bottom, and a sphere on the top.
<Iskierka>
that would be a big press. Conic rollers are much smaller and cheaper
<APlayer>
Pretty much what I do with stock aero and fairings, rather for the looks than anything else, though
<APlayer>
ferram4: I was thinking Atlas V (4--)
<Iskierka>
Thinking about it it's kinda weird that the shuttle tank was curvy. Was gonna mention that you only put complex curves on hardware expensive enough to make use of it but there's a clear exception there
<Iskierka>
might be a fashion thing like most "streamlined" steam engines though
<ferram4>
It could also be an expansion using the existing hardware.
<ferram4>
The 4m fairing still has the same cone angle as the original 3m fairing for Atlas-Centaur
<egg>
Iskierka: ferram4: is NEAR still a thing?
<ferram4>
No.
<APlayer>
So NASA would do a fashion thing despite high costs for it?
<ferram4>
Stock tries to be more advanced than NEAR. So no demand for it.
<Iskierka>
tries.
<Iskierka>
APlayer, the shuttle was meant to be fancy new tech
<Iskierka>
They did paint it white to begin with
<APlayer>
Yeah, I read that
<egg>
oh I seem to recall being involved in the acronym for NEAR :D
<APlayer>
IMHO the orange looks even cooler, though :D
<ferram4>
Remember that sex appeal gets Congressional funding.
<Iskierka>
The idea with making it fashionable would be to try drum up excitement for ^
<ferram4>
And blocky is less sexy than curvy
<Iskierka>
which is similar to streamlined steam engines in that steam engines were losing popularity because cars and such were gaining style, so they wanted to rebrand them
<APlayer>
Well, let's say the ET was teardrop shaped
<APlayer>
Or a Haack Body
<Iskierka>
the only ones that actually had any aerodynamic benefit were the Gresley A4 and possibly the Pennsylvania S1/T1 (more the latter)
<APlayer>
I see the problem that it'd be significantly higher than the shuttle itself
<ferram4>
Besides the fact that it would be very expensive to make the tanks
<ferram4>
And it would be heavier to support everything
<ferram4>
And it'd be a pain to attach the solids.
<Iskierka>
Which is why they only curved the ends
<Iskierka>
even though practicality says curve nothing
<APlayer>
And if you have the tip protruding well above the shuttle, only god know ls what kind of turbulence the deflected flow would cause when it hit the shuttle
<ferram4>
I think the bottom curve is actually from the hydrogen tank.
<Iskierka>
That's possibly true, but the nose isn't
<APlayer>
I still do not quite intuitively understand why blunt shapes have less drag than pointy ones... I always assumed tesrdrops assumed their shape because of surface tension and would have been pointy otherwise
<Iskierka>
well, they start as a ball due to surface tension
<APlayer>
That is, yes, be slightly off-AoA and it's bad, but the AoA is a negative feedback thing, isn't it?
<Iskierka>
and the point raised was that the front doesn't matter that much subsonic - pointy would work if you kept it pointed right. It's not used as much as you get less volume so less to work with and heavier
<Iskierka>
So, may as well go with practical and blunt-ish, and focus on the back, if you really care about subsonic drag
<APlayer>
So if you care about drag only (say for a challenge) and your CoM far at the front, a pointy tip would be viable?
<Iskierka>
Potentially. It doesn't hurt so long as you can reliably keep the AoA inside the cone
<APlayer>
Uh, please excuse those inconsistencies in my grammatical constructs and the typos. I am on mobile, sleepy and it's 0:30 AM
<Iskierka>
it's just that, in practical terms, you can't, so don't bother
<APlayer>
So a needle shape is bad, but a Haack Body as on the image on the Wikipedia page is okay?
<Iskierka>
Depending on the angles it has to deal with, yes.
<ferram4>
The Haack Body is the optimum for minimizing _wave_ drag.
<APlayer>
Well, it looks like there is a good 5 - 10 deg angle to either direction of it. I doubt that's too hard to control
<ferram4>
It is not optimum for minimizing other sources of drag.
<Iskierka>
It's a good base if you don't screw up the cone angle
<APlayer>
ferram4: Well, it looks quite aerodynamic in general. Though I see I should not trust my intuition anymore.
<Iskierka>
If you follow through to the Haack series nose cones that's what the body is based on two of and produces more options. Airliner noses are generally a modification of the LV-haack presented
<Iskierka>
(or possibly more towards the C=2/3 shape)
<APlayer>
Okay, thank you very much for making myself revise my KSP tactics and giving me some more insight in aerodynamics. It's certainly a very interesting topic, I only fear it's a bit too complex for me to truly deep into
<APlayer>
I am off for today then!
<Iskierka>
It's too complex for full-time aerodynamicists to truly deep into, given there's quantum theory lurking at the bottom
<APlayer>
Wait, just before I leave... Does one even say "deep into [sth]"?
<APlayer>
Or is that something my subconscious first cooked up and later rejected as weord phrasing?
<Iskierka>
I don't think it's normal english but it would work as memeable english