If you could precisely model every single component within the aircraft from a mass and size perspective, down to the rivets and connectors, then a 3D model may begin to approach real - world for behavior.
The problem is, it would only run on something on the orderof a Blue Gene/P, what with all the calculations necessary not only from real - time mass and balance calculations, but individual fluid dynamics for every square < insert unit of choice here > for that entire model. That’s before you add humidity, differences in air density, winds, or turbulences.
It’d be a heck of a lot of fun to try, though. You know, for somebody who actually likes programming.
As someone who knows almost nothing about programming (yeah - you should probably just stop reading here), I’ve thought for a while that surely there will come a point when we have enough computing power and programming tools to create a ‘standard’ general simulation environment with physics modelled into it incl. materials, temperatures etc.
Want plane X / car X / building X? Plonk the detailed CAD model into this standard simulation environment, specify all the materials etc. and it behaves like it would in real life (or close enough that it doesn’t matter).
Whether or not we are 90% there or 0.001% there, I have no clue, in terms of computing power etc…but we are already separately simulating all these different things (from building structural strength to hypersonic temperatures, viscosity etc.), surely a point will come where you will just have a SimEarth where you can freely test just about anything on the fly.
I’m sure the real-life programmer will come out of the woodwork and tell me to sit down at the back of the class and stop talking rubbish.
@Bearhedge, There is a hint of an answer in the wiki about x-plane above. The most important part of the airplane (or helicopter), the wing, is given just 4 sections for a total of 40 subsections. The granularity of what the designer wants modeled with each update must be very limited even with current computing power. And so much more happens below the surface. The variability among even the simplest of single-engine planes is staggering. And I don’t mean the little bits like rivets. Things like aileron cord thickness, balance tabs, counterweights, gap seals can make two identical looking planes fly very differently. Stock planes and helicopters in early versions of the sim flew well enough but everything seemed a touch off. I think that was because modeling surfaces alone wasn’t enough. Plane Maker and plugins were needed to massage the math to make the model better match reality. It takes both to get it right in real time.
Granularity is one of the major reasons to not worry about a true “real world” total environmental simulation. How close to reality do you need the modeled behavior to be? In what range of environments? Computationally and coding wise it’s a question of diminishing returns, if it takes X number of hours to write a simulation that’s 80% accurate, X^2 to get to 90%, X^3 to get to 95%, X^4 to get to 97.5%, and so on, when does it become not worth it? Even your professional clientele probably looses the ability to discriminate accuracy past a certain point below 99% (possibly well below depending on what it is).
Enough challenge and complexity to make a career out of it.
If you want to pull back the curtain a little bit the NASA Glenn Research Center has a great set of pages on the science behind flight: Index of Aerodynamics Slides
Don’t let the K-12 in the URL fool you- those are great plain-English explanations. Reading the “Aircraft Motion”, “Aircraft Parts”, “Aircraft Forces”, and “Aerodynamics” sections should be a decent primer on the math that flight modeling code needs to either execute or approximate.
These statements are all orbiting around the same issue which simplifies to something like this:
Even with infinite computing power, some poor sucker still needs to model, program, and validate the thing.
To take that a step further, a lesson frequently learned the hard way is that having an accurate computer model or simulation of parts A through Z does not necessarily mean you can combine the lot and have a good model/sim of the thing made up of parts A through Z.
Flight dynamics and aerodynamics tend to be defined by interactions rather than singular variables.
Which makes it all the wonder as to how a couple of bicycle mechanics got in the air. I suppose if you want something badly enough, that you will either figure it out or die trying.
I think the Wright brothers brought to the table a combination of scientific rigor, mechanical proficiency, and innovative spirit that hadn’t been seen in that era of flight yet. Right guys, right time.
If you read about their process there’s quite a few points where they could have gotten stuck but they instead distilled the issue down to a single problem or unknown, designed an experiment, got data, created a new principle, and went ahead based on that until they found a new problem. Rinse and repeat until controlled powered flight.
Haha, yeah thats a really funny story.
…but then that was in 1983 so there was some development in computer technology since then, allowing marginally better computer simulations.
it is easy, we should do it like decentralized blockchain computing. All the PCs not siming at the exact moment will do calculations for the siming PCs.
That will make for enough computing power. We just need fast enough internet
On my first flight to Mexico City the captain told me about a night he had an emergency going there in an MD-80. They’d had a hydraulic failure scenario which gave them gear but no flaps. They landed, blew out all the tires but stayed on the runway with no fire. Good airmanship no doubt. But the guy was pretty full of himself. His coup de grace was excitedly claiming that Douglas tried same scenario in the simulator and reported that it couldn’t be done. That was many years ago and simulator technology has come a long way since. I still marvel at how unlike a 737 my company’s 737 simulators fly. There is just too much happening in this madly complex physical world of ours to be convincingly simulated unless your expectations are sufficiently low. That’s why I don’t give a poop so long as the basic laws of physics seem to be followed.
I remember when I started flying the Fokker 50 and how different the sim was to the real deal.
Then I went on to fly the Dash8. Same thing.
You can simulate the systems, speed vs. power, etc. perfectly by the numbers, but the real aircraft will still have a different feel to it.
And very few airliners are tested to the edge of the envelope. So simulating their departure behaviour is pure guesswork. Educated guesswork, but still…
I know- I wasn’t entirely serious.