Slashdot Mirror
SpaceX's New Combustion Technologies
An anonymous reader shares this story that takes a look at some of the advances SpaceX is working on. "Getting a small group of human beings to Mars and back is no easy task, we learned at the recent GPU Technology Conference in San Jose hosted graphics chip and accelerator maker Nvidia. One of the problems with such a mission is that you need a very large and efficient rocket engine to get the amount of material into orbit for the mission, explained Adam Lichtl, who is director of research at SpaceX and who with a team of a few dozen programmers is try to crack the particularly difficult task of better simulating the combustion inside of a rocket engine. You need a large engine to shorten the trip to Mars, too....Not only do you need a lot of stuff to get to Mars and sustain a colony there, but you also need a way to generate fuel on Mars to come back to Earth. All of these factors affect the design of the rocket engine....As if these were not problems enough, there is another really big issue. The computational fluid dynamics, or CFD, software that is used to simulate the movement of fluids and gases and their ignition inside of all kinds of engines is particularly bad at assisting in rocket engine design. 'Methane is a fairly simple hydrocarbon that is perfectly good as a fuel,' Lichtl said. 'The challenge here is to design an engine that works efficiently with such a compound. But rocket engine CFD is hard. Really hard.'"
This is the first article I've seen that explains well how GPUs can/are being used for practical applications along with what can be achieved and some of the issues. Well worth the read even if you're not into this stuff.
I'm sure that there is a significant cost in developing this new approach to CFD (as well as pushing the envelope on GPU operation) but the result is going to be usable for different applications. TFA says there's irony in what SpaceX is doing here as it has applications with automotive Internal combustion engines but I see that as SpaceX/Musk having a secondary revenue stream for this work that doesn't mean he's helping out his direct competitors.
Along with that, they are driving the development of high speed inter GPU communications which I'm sure has value as well.
All this means is that Musk returns to his home planet, not only is the trip going to be fully funded, but he's going to have some money to throw around when he gets there.
myke
Mimetics Inc. Twitter
It's kind of the nature of research that you have to do research. Plus, if you think its hard to design rockets and "rolling your own" CFD models, just imagine how it was when you didn't have the computing power to go beyond the rudiments.
Those troglodytes designing say, the F1 might have had a bit harder of a time designing with limited knowledge and experience than now when can try rolling out likely designs based on good CFD models. Channeled explosions are not terribly forgiving of bad design in that way.
Rather, Spacex is able to make better rockets, less expensively, by using extensive CFD, modeling using models they developed. This will allow for optimization of designs, and reduce reliance on the old process dreaming up a design, building and test firing it, and if it didn't explode, analyze the parts for incremental improvements.
And rather than whine about how hard it is, maybe we should marvel at just how damn cool it is to have modern computing power and new flow and design models to work with to make better rockets. Too freaking cool!
The shepherds did so well protecting the flock that the sheep no longer believed that wolves existed.
Umm, rocket science is...rocket science?
Combustion CFD is a very difficult area. The problem is that there are so many interlinked phenomena all requiring special modeling methods that one really isn't quite certain of the accuracy of the result unless they can compare it to a physical model test, which is what is frequently done. Simply getting the correct boundary conditions can be very challenging. Failing to apply appropriate modeling and boundary situations leads to a garbage in/garbage out situation, but the numerical solution may look plausibly correct.
CFD is not use exclusively in design work except for very basic cases where the modeling accuracy is well understood. However, CFD for more complicated situations is still useful as it may illustrate behaviors and trends in performance in situations where physical observations are difficult (like in a rocket nozzle). The CFD results can be used to guide and interpret the results of physical testing.
Understanding CFD really requires PhDs who understand fluid dynamics as well as the limitations of the numerical models used. This is true in many industries, not just rocket surgery.
Left MS Windows for Linux Mint and never looked back!
Vote for Bernie in 2016!
All real fluids have a finite Reynolds number, which tells you offhand how much grid refinement it takes to resolve the smallest scale directly. Since for supersonic flow in a rocket engine R is usually stupid high, the small scale turbulence is too small for direct resolution so you resort to turbulence models (e.g. RANS - Reynolds Averaged Navier Stokes) which is in itself an entire industry.
That part is relatively well developed and it's actually approaching the point that (with a team of experts who can recognize the defects on sight) things like CFD wing design are approaching predictive rather than "hey, the CFD actually got it right for a change. Woo!" The challenge for rocket engines is that you're not considering a single fluid, or even a two-phase flow, but a reactive flow which (if you look at all the paths even methane combustion goes through) contains about a hundred components, meaning a hundred flows, with 100 godawfully stiff nonlinear rate equations coupling them - in every single cell! This is the crux that largely stymies effective CFD of combusting flows.
I'm an astrophysics guy so I mostly get to watch from a distance and cringe in horror. We consider ourselves to be Doing Well if we look at gas/dust or neutrals/ions. Really good is looking at neutrals/ions/electrons. We do have our own 100-coupled-rate-equations horror show in examining the nucleosynthesis going on behind a supernova blast front.
The matter of computability is this: Watch a river flow, a prototypical turbulent system if ever there was one. Below the mercurial, ever fluctuating turbulence, you notice persistent, standing structures. Many flows of interest have a similar structure. The flow of water in to a nuclear reactor plenum, air over a car, the atmosphere - Turbulence superimposed on a coherent larger structure. Trying to model the exact turbulence is, as you say, chaotic and pointless: Paths depart exponentially. But if you can model the chaotic part you can still learn about the underlying nonchaotic structure.
In spectrum space, what I'm describing are systems where the turbulence lives in high-wavenumber modes and interacts in some relatively predictable way with the lower wavenumber modes describing the structures of interest. When something breaks down into complete turbulence (e.g. a Rayleigh-Taylor unstable turnover in the atmosphere - Have you ever been in a placid afternoon, then out of nowhere, huge gusts in random directions out of nowhere? R-T overturn), whole new animal...
I hope they simulate propane too, not just methane. Propane has some really interesting properties as rocket fuel but have (like methane) never gotten much research. But now there's a big rush to research methane as fuel based on the concept of generating it on Mars - so propane still gets left in the dark.
Methane's ISP is only very slightly better than propane's - 364,6 vs. 368,3 at a 100:1 expansion into vacuum and 20MPa chamber pressure. But propane at around 100K (note: not at its boiling point, 230K) has far higher density (782 kg/m^3), closer to that of room temperature RP-1 (820 kg/m) then that of boiling point methane (423 kg/m^3), which reduces tankage mass and cost. 100K propane's ISP is of course better than RP-1's 354.6 in the same conditions as above. Plus, its temperature is similar enough to your LOX that they can share a common bulkhead, which reduces mass further and simplifies construction.
Hydrogen generally is the easiest fuel to synthesize offworld. Methane is generally second, and propane third. Hydrogen is often rejected as a martian fuel because of the tankage and cooling requirements. Methane can be kept as liquid on Mars with little cooling in properly designed reflective / insulated containers - but so can 100K propane, in similar conditions, but with significantly smaller tankage requirements.
It seriously warrants more research, I tell you what.
"99 dead duelists of Dios on the wall. 99 dead duelists of Dios! Take one's ring, pass it around..."