[1.3.1] Ferram Aerospace Research: v0.15.9.1 "Liepmann" 4/2/18

[soggymushrooms]

XPlane is the benchmark for flight simulation accuracy, and it provides essentially true to life simulation of flight dynamics in realtime. Beyond XPlane, you'd probably have to start simulating fluid physics to be more accurate. That being said, XPlane has been in development for much longer than FAR, and I don't expect FAR to be as accurate as XPlane, but I'm confident it'll get pretty close.

[StevenRS11]

Yea, I was thinking more about a CFD solver like Stallion 3D, but you know, not $24K.

[ss8913]

FAR is easier then stock always has been, if your a trained pilot. I could never fly or design anything that flew anywhere remotely like a plane in the old placeholder aero. People always said the old placeholder was like flying in soup, but I disagree it was like flying in some state of matter not currently known to man. Closest I could ever come to describing it, was like flying through silly-string. FAR was much, much better but still off a bit. What ferram has accomplished with FAR is nothing short of amazing to me, when you realize he did it all in his free time, and the handicaps he had to work with. Without FAR I would never have been able to play with anything that looked remotely like a plane in KSP.

I agree. I am a trained pilot as well, and I expect air to act like air, and that's why FAR is the most important KSP mod for me. Looking forward to trying out the new USB control support in 1.0 with both my yoke/stick/throttles/rudder pedals and the TrackIR4. Raster Prop Monitor improves my life significantly in this regard as well since it adds useful IVA...

- - - Updated - - -

I wonder if there is some super-accurate but slow aerodynamics simulator out there that can simulate arbitrary 3D models that's not some ultra-pricey piece of software. I could export KSP ships as a model (maybe .stl?), have it simulate them, and compare the results to what FAR gives.

The idea being that the separate piece of software doesn't have to work in real time, so it could afford to use far more computationally expensive but more accurate (and maybe simpler) methods. It might be useful for finding edge cases that FAR doesn't handle well, and if FAR does a really good job, then it would definitely be a source of bragging rights.

"Hey, I made a mod for KSP that simulates aerodynamics as accurately as this research tool. Oh, and mine does it in real time, too."

Or you may already do this, who knows. I have access to a university license of Solidworks Flow, so I may fiddle with that. Flow is geared more towards lower speed stuff like fluid in pipes, though. I know it does some aerodynamic stuff, but I'm not sure if it will do stuff like supersonic wave drag or compression lift, though. Depends on how general it's fluid dynamics solver is, I guess. One way to find out!

X-Plane 10 from Laminar Research does exactly this, and has been used to design a few real world aircraft. You design the 3D model and it simulates the laminar airflow over that surface. It's not *cheap* but it's under $100 and includes global scenery. With the optional unlock key (which does cost a bit more), it can also be used as an FAA-approved PCATD.

[TeddyDD]

nuFAR hype train, anyone?

Oh yes! This will be amazing!

And Ferram4 I would just ignore anyone who says FAR is easier than stock.

But it is! (or rather was, because new stock aero is hyper easy to fly planes) Even if you are not a pilot. Of course, until you try to do something very unrealistic.

And in my opinion this is the advantage

Edited April 29, 2015 by TeddyDD

[AndreyATGB]

XPlane is the benchmark for flight simulation accuracy, and it provides essentially true to life simulation of flight dynamics in realtime. Beyond XPlane, you'd probably have to start simulating fluid physics to be more accurate. That being said, XPlane has been in development for much longer than FAR, and I don't expect FAR to be as accurate as XPlane, but I'm confident it'll get pretty close.

It's also not developed by a single person.. Not really a fair comparison consider planes in XPlane aren't made from pieces that don't bend as well.

[SacaSoh]

It's also not developed by a single person.. Not really a fair comparison consider planes in XPlane aren't made from pieces that don't bend as well.

I'm not sure, but I think Xplane just generates a "lookup table" with the aerodynamics variables for a given plane geometry (i.e. it will calculate the aerodynamics of a plane you designed, but not in real time, as your plane wont be subject to geometry changes, as things in KSP usually do).

[camlost]

Although Xplane has been out for long, it used to be one man's project for a great many years. AFAIK, its aerodynamic algorithm has been there for almost a decade. Similar to FAR, XP also applies some kind of classic aerodynamic calculation instead of CFD, so I wouldn't call it 'super accurate air simulator'.

Up to now, no one has tried to incorporate CFD into a flight simulator.

So I think XP and FAR are actually comparable.

- - - Updated - - -

I'm not sure, but I think Xplane just generates a "lookup table" with the aerodynamics variables for a given plane geometry (i.e. it will calculate the aerodynamics of a plane you designed, but not in real time, as your plane wont be subject to geometry changes, as things in KSP usually do).

In XP aerodynamic forces are calculated dynamically based on geometry. Other sims like FSX uses lookup table.

[ss8913]

I'm not sure, but I think Xplane just generates a "lookup table" with the aerodynamics variables for a given plane geometry (i.e. it will calculate the aerodynamics of a plane you designed, but not in real time, as your plane wont be subject to geometry changes, as things in KSP usually do).

I think this is how Microsoft Flight Simulator works. X-plane, however, simulates the laminar airflow over the surface of the craft in real-time, which is what makes it more accurate and can be used to fly unbuilt prototypes based on their dimensions - ie the "Carter Copter" was designed that way using X-Plane.

XP was originally designed by one guy, but that was long ago, and yes there is a team on it now.

- - - Updated - - -

so... ferram4: I'm assuming that when nuFAR releases it'll be available via CKAN? Will it be under FAR or is "nuFAR" the official name for the new version?

[monstah]

so... ferram4: I'm assuming that when nuFAR releases it'll be available via CKAN? Will it be under FAR or is "nuFAR" the official name for the new version?

Related question: oldFAR doesn't show up on CKAN anymore. Can I install it manually without breaking my game or do I wait until nuFAR is available?

Stock is an improvement, but I still like this one better.

[Crzyrndm]

FAR doesn't show because it's for 0.90 and IIRC CKAN filters by version. 0.90 FAR working in 1.0? Unlikely...

[blowfish]

Related question: oldFAR doesn't show up on CKAN anymore. Can I install it manually without breaking my game or do I wait until nuFAR is available?

If you have KSP 1.0 it won't show up because it's incompatible. If you want to keep using FAR you need to stay on 0.90, or download the experimental voxel aero branch, but of course there will be unresolved issues (which ferram4 has mentioned).

[ss8913]

What this means though is you can keep refreshing inside the CKAN interface to see when mods go official for 1.0

[monstah]

Ah yes, the good ole f5 spam.

Smells of the early 00s. *nostalgia*

[Jovus]

FAR is version-locked, so no previous version of FAR will work on 1.0.

Even if it weren't, I doubt it heavily, considering they completely changed aerodynamics.

[NocTempre]

I've been wondering, are aerodynamics affected at all by the scale differences between KSP planets and real planets?

Reynolds Number.

[Van Disaster]

Ferram, are you still using a single reference aerofoil in the new version?

[ss8913]

I'm assuming ferram4 hasn't posted much today because he's focusing on rolling nuFAR out. I kind of want to print out a hex dump of the object code, cover it in chocolate, and roll around on it.

[Wanderfound]

I'm assuming ferram4 hasn't posted much today because he's focusing on rolling nuFAR out. I kind of want to print out a hex dump of the object code, cover it in chocolate, and roll around on it.

If it works as promised, I think we should build an aircraft out of chocolate and send it to Ferram.

[duckofthemoon]

If it works as promised, I think we should build an aircraft out of chocolate and send it to Ferram.

seconded

[dc4bs]

Could be worse... ever tried to find a computer that is on your network, but you're not sure where it physically is?

Yes! Trying to locate a mystery server located somewhere in a haystack of over 1200 other servers distributed in about 220 racks...

Getting the switch name and port ID that was assigned from the networking group and then ever so gently tugging on network cables as you work your way slowly dragging a step ladder down the rows of racks to finally find a server that was renamed/repurposed but that no one ever bothered to actually change the physical label or update our server location database. I LIVE for such days... NOT!

[youkofoxy]

If it works as promised, I think we should build an aircraft out of chocolate and send it to Ferram.

better, make an aircraft out of no-toxic material and stuff the cargo bay with treats.

or maybe an aircraft router?

because some countries do not take it easy on supervision of food.

my being one of them (they confiscated a cupcake from a guy, a cupcake)

anyway, on V. FAR:

maybe step could do the trick? a mean, CPU 1 do frame 1 CPU 2 start to work on frame 2?

or pair work: CPU 1 gets numbers no multiple of 2 and CPU 2 get the numbers multiples of 2?

I probably sound stupid, after all, I don't writer code.

and the parachute problem... there any restrictions? besides: hacks not allow.

[Crzyrndm]

maybe step could do the trick? a mean, CPU 1 do frame 1 CPU 2 start to work on frame 2?

You can't start calculating the subsequent frame until you know how everything ends up in the current frame

or pair work: CPU 1 gets numbers no multiple of 2 and CPU 2 get the numbers multiples of 2?

This is what ferram is already doing essentially (multi-threading is batching up sets of calculations that can be run in parallel)

[diomedea]

It also holds true in data centers.

Me: "Say, what's this old server with no label on it?"

Other Admin: "Ummm... I think that might the temporary server for project XYZ that was brought online 3 years ago..."

Me: "So why is there no labeling of any kind on it?"

Other Admin: "Because it's temporary and should be gone in a few more weeks..."

Me: "Argh!!!"

Could be worse... ever tried to find a computer that is on your network, but you're not sure where it physically is?

Yes! Trying to locate a mystery server located somewhere in a haystack of over 1200 other servers distributed in about 220 racks...

Getting the switch name and port ID that was assigned from the networking group and then ever so gently tugging on network cables as you work your way slowly dragging a step ladder down the rows of racks to finally find a server that was renamed/repurposed but that no one ever bothered to actually change the physical label or update our server location database. I LIVE for such days... NOT!

Please, stay on topic (FAR in case it wasn't clear). This conversation is drifting away too much, would be better if continued on a thread of its own.

[rabzoue59]

every times i see ferram aerospace on the top of topics, i believe its the new version released lol, but fake joy lol

[Volodyuka]

Ok, i have a very strange request, is it ever will be possible that most of the instrumentation (like IAS and EAS, Stall and High dynamic pressure redouts) and tweakables for control surfaces will be available as a stand alone for stock? Cause i kinda like stock aero in 1.0 but i grown used to all this little things like possibility to tweak control surfaces or to see your speed in kph EAS...

Edited April 30, 2015 by Volodyuka