I had to learn FORTRAN for my engineering classes back in college. I do NOT miss those days of compiling on a Wyse terminal. I feel especially bad for all the people that had to use punch cards.
The Aerospace Industry Discussion Thread
- Thread starter Bad Monkey!
- Start date
I wonder if I still have my Fortan 77 code for that modeling...
Wish I'd known about that at the time, I like FORTRAN. I don't really remember it, but it wasn't hard to pick up when I was helping El Nino researchers back in 1990.
I think it's important to mention that there's not actually much 'innovation' in the pure simulation part of things. It's all just a mesh in patran/nastran evolving with some specific mesh solver. There are actual analytical optimums you can choose for lots of the parameters in the mesh, from time step to mesh density. There are no shortcuts you can do for this kind of stuff, because especially fluid dynamic simulations can fundamentally not be reduced in time complexity.
The innovation that has dominated the industry since I was studying aerospace engineering all has to do with scaling across many compute nodes. Around the early 2000s, the big challenge was that when you distribute a mesh across thousands of compute nodes, because the entire mesh is still interconnected, you needed to exchange boundary information between nodes which put a fundamental limit on how useful it is to break a mesh up in smaller parts. Later on, as supercomputers started massively increasing the bandwidth between nodes and increasing local memory, you could do things like progressive solvers that would simulate an increasing mesh size as time goes on, synchronizing only every couple of time steps instead of every time step. Nowadays, the holy grail is dynamic meshing across many nodes, where the mesh changes to improve e.g. mesh orientation across moving features where deflections are larger than a mesh cell.
But underneath, the solvers are still the same. All of the real work is trying to get a few square meters of silicon working together efficiently.
So one thing that seems interesting to me and I wanted to get into, but never really had the time, is using something like SIREN to represent the fields, and to do the solving by gradient descent (where it'd calculate the error as deviation from your PDE at randomly chosen samples).
Mostly the reason I want to do that is to avoid meshing altogether, on high dimensional problems like general relativity. I am a very "I have to write some code" learner when it comes to physics. If I didn't write some kind of simulator then I don't feel I learned the subject. The other possible application idea I had for that is lighting inside clouds, my first commercial success of sorts was an atmospheric effects renderer that I used in some TV ads back in the 'noughts.
Anyways circling back to the topic of aerospace, I think there may be similar applications for gradient descent based mesh free solving for aerodynamics, eventually.
edit: or perhaps gradient based, meshed-ish solver whereby nodes appear/disappear, to parallel what happened in 3D scene reconstruction with Gaussian splatting. (Of course, the actual optimizer is usually not just plain gradient descent, but utilizes momentum and an estimate of higher order derivatives via last several steps)
Mostly the reason I want to do that is to avoid meshing altogether, on high dimensional problems like general relativity. I am a very "I have to write some code" learner when it comes to physics. If I didn't write some kind of simulator then I don't feel I learned the subject. The other possible application idea I had for that is lighting inside clouds, my first commercial success of sorts was an atmospheric effects renderer that I used in some TV ads back in the 'noughts.
Anyways circling back to the topic of aerospace, I think there may be similar applications for gradient descent based mesh free solving for aerodynamics, eventually.
edit: or perhaps gradient based, meshed-ish solver whereby nodes appear/disappear, to parallel what happened in 3D scene reconstruction with Gaussian splatting. (Of course, the actual optimizer is usually not just plain gradient descent, but utilizes momentum and an estimate of higher order derivatives via last several steps)
Fundamentally, that limits you to e.g. potential flow calculations. Potential flow is what happens in laminar flow systems without any turbulence, low Reynolds numbers, etc.. Potential flow is linearizable, and there aren't just gradient descent solvers - there are pseudoanalytic solvers for it and it's one of the few flow simulation types you see in e.g. games.
Generalized Navier-Stokes however is decidedly not linearizable. OK, to be fair, linearized solvers are used for short-term simulations, but only if you really know what you're doing. It's not even generalized N-S that's your problem, it's things like phase changes and the need to apply lookup tables for specific heat and highly nonlinear friction effects that really ruins your day.
All that being said, it's really fun to make physics solvers. I used to do that all the time in javascript, because it was easy to make instant visualizations. It also quickly gives you an appreciation for the complexity of it all
Generalized Navier-Stokes however is decidedly not linearizable. OK, to be fair, linearized solvers are used for short-term simulations, but only if you really know what you're doing. It's not even generalized N-S that's your problem, it's things like phase changes and the need to apply lookup tables for specific heat and highly nonlinear friction effects that really ruins your day.
All that being said, it's really fun to make physics solvers. I used to do that all the time in javascript, because it was easy to make instant visualizations. It also quickly gives you an appreciation for the complexity of it all
Quick question. Just read an article on the RLV C5 being developed in Europe says it doesn't need tiles and I was curious why. The article says:
Recovery would be very different from Starship, though. RLV C5 would instead descend into the atmosphere using wings to slow itself to sub-orbital speeds. At the appropriate speed and altitude, it would then be captured by a large, subsonic craft.
This approach would allow RLV C5 to carry up to 74% of its total mass into orbit, rather than the mere 40% payload mass of Starship. It would be smaller, a mere third of the overall weight, and likely far cheaper, too. Without the need for re-entry tiles, additional fuel, and full-reusability, the RLV C5 could fill a niche that Starship would just be overbuilt for, the researchers argue.
https://www.extremetech.com/aerospa...d-mid-air-recovery?utm_source=spiceworks-snap
Recovery would be very different from Starship, though. RLV C5 would instead descend into the atmosphere using wings to slow itself to sub-orbital speeds. At the appropriate speed and altitude, it would then be captured by a large, subsonic craft.
This approach would allow RLV C5 to carry up to 74% of its total mass into orbit, rather than the mere 40% payload mass of Starship. It would be smaller, a mere third of the overall weight, and likely far cheaper, too. Without the need for re-entry tiles, additional fuel, and full-reusability, the RLV C5 could fill a niche that Starship would just be overbuilt for, the researchers argue.
https://www.extremetech.com/aerospa...d-mid-air-recovery?utm_source=spiceworks-snap
I've never heard about this particular piece of space hardware, but if you're not going for full reusability, you could just use an abalative heat shield for reentry and then replace it as necessary. I think Artemis capsules are going to work this way? Not sure if that actually saves mass, but it probably saves engineering/maintenance costs for refurbishment. (My FiL worked on the shuttle tiles, and replacing them after each mission never went as well as the marketing said it would...) If the article uses 'weight' instead of 'mass', though, I'm not sure I really trust it. But maybe it's a translation?
Last edited:
I'm pretty sure it doesn't have an ablative shield of any description across its structure. It's a pure hypersonic wing.
I think they're going for some kind of heat shield on the nosecone and doing shockwave management to keep the rest of the structure out of the brunt of the ionized flow. It's theoretically possible (the whole concept of hypersonic scramjets) to remain low hypersonic all the way down from near-orbital speeds.
The thing I'm not immediately excited by is the technology tradeoff you're doing here. Instead of using a cheap steel fairing/structure and putting on ablative tiles, they're going to have to go for some super high temp titanium alloy hull. Is that an improvement? Maybe we should wait for a bit more info.
I think they're going for some kind of heat shield on the nosecone and doing shockwave management to keep the rest of the structure out of the brunt of the ionized flow. It's theoretically possible (the whole concept of hypersonic scramjets) to remain low hypersonic all the way down from near-orbital speeds.
The thing I'm not immediately excited by is the technology tradeoff you're doing here. Instead of using a cheap steel fairing/structure and putting on ablative tiles, they're going to have to go for some super high temp titanium alloy hull. Is that an improvement? Maybe we should wait for a bit more info.
Since they are recovering the craft via aircraft I assumed it was to be reusable.
The article made it seem like the craft would not need heat shielding at all. Maybe it is poorly worded for my brain.
The article made it seem like the craft would not need heat shielding at all. Maybe it is poorly worded for my brain.
Thanks for giving me some context! As far as some kind of super inconel vs tiles, like I just said, NASA spent a lot of engineering time trying to make the tile replacements for shuttles not such a big pain in the ass, without much success.
Thinking about Columbia, having a single point of failure (the one nosecone), is probably better than having to worry about the entire leading edge, but I don't know enough about the engineering of superalloys (or hypersonic drag) to know if that's even a possible safety tradeoff.
The full text paper is linked and open access, and is interesting reading. Note that the second stage is expendable and the first stage is what will do the aero capture. My layman's understanding is that this minimizes the mass used for first stage recovery, it has wings that let it glide at some high subsonic speed where it can be hooked on to a cable towed by a standard aircraft that will bring it down for a landing. Compared to the SpaceX SH booster, this concept won't need fuel for reentry or landing burns, but it will need added structure for wings that can glide in the subsonic regime. I suppose the tow capture idea minimizes the wing size? It just needs enough glide performance to rendezvous with another aircraft, and doesn't need oversized wings to land at same speeds at a runway that's not perfectly located downrange.
Kinda crazy sounding but maybe not that much more crazy than the SH tower catch before it was actually demonstrated. Definitely more crazy now that tower catch has actually been shown to work repeatedly!
Kinda crazy sounding but maybe not that much more crazy than the SH tower catch before it was actually demonstrated. Definitely more crazy now that tower catch has actually been shown to work repeatedly!
Kinda makes sense, but it isn't a Starship competitor as much as super heavy booster competitor in this case?
The US (and probably the USSR?) used to catch photo capsules out of the air, and there's also the fulton recovery system... But I'm pretty sure most of the catchy bits have been removed from the planes (and probably scrapped). And there's also finding pilots who think they've got enough of the right stuff to basically be the target of a seeking missile that doesn't have a warhead. Also, getting from close to orbital speed down to high subsonic is not exactly easy. Lots of engineering to do. I wish them luck!
More to the point, those photo capsules were pretty small, and were falling relatively slowly hanging from parachutes. I'm not sure there are many useful lessons that would apply to something much bigger and moving much faster. Looking for inspiration in air to air refueling systems would seem to be a closer analogy.
A big difference you've left out in that description is that Superheavy has to fly all the way back to the launch point. It doesn't just need fuel for re-entry and landing, but a big chunk of the remaining fuel is required for boostback. Superheavy's trajectory is a trade-off between how much fuel will be used to get to orbit and how much fuel will be reserved for boostback -- a more efficient trajectory for orbit would require more fuel reserved for boostback. RLV C5 can fly a more efficient trajectory because it will get back to the launch point through a combination of gliding and being towed.
As far as re-entry heat management, that is also related to the different trajectories. RLV C5 flies a shallower trajectory (although boosts for longer, so gets to a higher altitude before separation) compared to Superheavy. The RLV C5's trajectory is more fuel efficient to orbit, but also allows for lower heat loads during re-entry. It basically re-enters more slowly compared to Superheavy, spreading the energy as lower heat load over a longer period of time. I haven't been able to find enough detail to determine if the lower heat loads are what allows for not using tiles, or if it is more a matter of different design philosophies. For instance, the RLV C5 designers have talked about using transpirational cooling for some of the hotter leading edge areas, and they definitely have talked about different exterior alloys than what Superheavy uses.
For those who haven't read the paper, here's a helpful graphic. RLV C5 is obviously on the left. The yellow, blue and red thing is the second stage, showing the cargo fairing (upper yellow part), LH2 tank (blue), LOX tank (red) and then the tail end is yellow again for some reason.
I would also note that the paper doesn't really get at what the cost trade-offs between a cheaper single-use upper stage and a more expensive re-useable upper stage. The paper is an analysis of potential payload tonnage, and not trying to address costs.
Rocket Lab is carrying this as far as they can in the opposite direction with Neutron. Move as much hardware as possible from the to-be-expended upper stage to the recoverable lower stage. The payload fairing is the most visible component, but I bet there are other pieces that their engineers have cleverly shifted.
Reminds me of the old SR75/Mothership discussions a decade ago.
Yeah, I was one of those - my best friend was a wide black magic marker. Fortran77!
Sadly my parents threw all that away when I was at university - back in 1982.
First view of B-21 in-air refueling... Kudos to Sandboxx.
View: https://www.youtube.com/watch?v=A0xSQE29Sag
View: https://www.youtube.com/watch?v=A0xSQE29Sag
News reports I've read described close proximity flight testing, but not actual hookup or fuel transfer. In other words, making sure there's no aerodynamic impediment to refuelling with the current design. Does the video definitively show what the title claims?
https://www.airandspaceforces.com/b-21-spotted-flying-with-kc-135-in-precursor-to-refueling-test/
Indeed, the Air Force has explicitly said it was only close proximity, without connection or fuel transfer
Indeed, the Air Force has explicitly said it was only close proximity, without connection or fuel transfer
Crash at LaGuardia involving an Air Canada CRJ. Apparently an emergency vehicle responding to another incident was given clearance to cross a runway while it was landing. Pilots are dead, seems to be the only fatalities.
NYT rolling updates, gift link - https://www.nytimes.com/live/2026/0...e_code=1.VVA.TEGs.IVdK1EAK49gE&smid=url-share
I feel bad for the controller. He's gonna take the brunt of the heat, but this sort of thing is a political failing, and will continue to happen as we kick the can down the road instead of investing in our infrastructure.
If they try to scapegoat the controller when a person working solo is routine due to chronic understaffing the whole union should go on strike and dare Trump to fire them. If he tries to do a Reagan and fire them, air travel will be at a standstill for years. If he tries to privatize them, the same workers will turn around and demand 5x the pay from the airports.
I saw some comments on Reddit that the ATC was working solo for air & ground traffic. Which seems nuts for an airport as busy as La Guardia, even if it was the middle of the night.
Yup. Happens all the time, too. Roll the dice enough and a freak accident becomes a certainty.
Sounds like the Potomic crash last year, one ATC was tasked with handling both chopper & local control duties which was intended to be a two-person job during the day, and only to be combined late at night when the workload was at a low level.
Boggles my mind anyone would want to be ATC, the stress and having the lives of hundreds of people wavering on a single brief slip of the mind or mistake in hearing is something I could never handle. Having to dedicate years of training to get into it and that you'd be forced to work mandatory overtime and get shit pay/scheduling are not exactly positive incentives either.
I know a guy that was ATC in the LA basin. Owned a decent 911 and relatively new Ferrari. Hated LA and moved to Indiana. Still ATC but a much lower stress level (and much lower salary). Still has the fancy cars. The 911 is easy to work on, the F car would be high stress to me maintenance wise.
And you're not even guaranteed a steady paycheck either.
The qualification to get into it is "high school graduate". The training is fully paid. The hourly pay is actually quite good, especially if one works at the higher-level positions, at busy airports or TRACONs. The job security is quite good, even if one does make mistakes, so long as they're not persistent.
This is true, but keep in mind that the current partial federal funding lapse isn't hitting them, since they're in Dept of Transportation, not Homeland Security.
I'm aware of that, but I think we should assume anyone fit for the job (eg, not a complete idiot) can look at the track record and figure out it's probably going happen 2-3 times per decade, whether or not it's happening right this second. The pay isn't even that good given the working conditions and the qualifications needed.
If the US privatized ATC tomorrow, I have every expectation the existing workforce would turn around and say to airports "guess what, you're recognizing our union, and the base salary for all current qualified staff is a quarter mil". And faced with the prospect of years of disruptions if they refused, airports would cave.
Last edited:
I don't think you understand how hard it actually is to be an ATC, which is one of the problems with staffing actually (strict requirements are a good thing, in my opinion).
These aren't "just high school grads." The training is intense; failure rates on the testing is incredibly high; tolerances are low and have mandatory retirement age. $100k+ starting in many cases is really good for someone with a high school diploma and I know of plenty of stressful jobs that pay half that for people with Master's.
We need more of them and there's been big pushes in recent years to try and get them. There's bad advertisement (there needs to be social media reach outs; job fairs; high school involvement; etc...) so noone really thinks about it as a career option.
These aren't "just high school grads." The training is intense; failure rates on the testing is incredibly high; tolerances are low and have mandatory retirement age. $100k+ starting in many cases is really good for someone with a high school diploma and I know of plenty of stressful jobs that pay half that for people with Master's.
We need more of them and there's been big pushes in recent years to try and get them. There's bad advertisement (there needs to be social media reach outs; job fairs; high school involvement; etc...) so noone really thinks about it as a career option.
I seriously considered ATC as a career. Unfortunately at 32 I was already too old.
The maximum age requirement is a self-inflicted injury.
Is this a response to me? If so I'm not sure where I gave the impression it's easy. I just said I infer you can't be a complete idiot, obviously that's not the entirety of the requirements.
Yeah, and a lot of those people could turn around and work at Goldman Sachs for 7 figures if they wanted. I'm just commenting if ATC is privatized it should expect to jump a lot.
It's really not. There are sound human performance reasons to have mandatory retirement in this kind of line of work. Cognitive flexibility and response times degrade with advanced age. They could maybe move to a performance-based standard, where controllers are tested in high-stress simulations to make sure they're keeping up to the level expected.
The other reason for it is that the intense stress of the job comes with increased medical issues. They don't want to have could-have-retired-as-planned controllers instead dropping dead or disappearing on medical leave.
Maybe there should be a separate status for controllers too old or ill to actively operate, but still fully able to train and fulfill other support roles, to keep the experience available.