Pretty straightforward if rather nerdy question.
Most of the internet gossip on these things http://en.wikipedia.org/wiki/Pintle_injector suggest that they are the greatest thing since sliced bread. The linked PDFs off the wiki page show that TRW had some monster engines testing out very successfully (250,000 lb thrust) in the late sixties, with very low cost and high reliability being touted as primary benefits.
Forty some years later, they are still the ‘next big thing’ according to SpaceX and bloggers - so how come they didn’t take off already? The people who work on this stuff are proverbially the smartest folks around so I assume there must be good reasons. Are there compelling technical disadvantages, economic effects (such as everybody being invested in alternate designs) or philosophical differences (everybody focusing on maxing out performance/efficiency at the cost of unreliable & expensive designs)?
Do you mean replacing the internal combustion engine in cars? Being used for electricity production? Being used in planes? Being used in weapons? Being used in spaceships?
The pintle injector has certain advantages, specifically with regard to combustion stability with high degrees of throttleability. This is desirable for engines that have to be throttled or restarted repeatedly, but it isn’t the optimum configuration for best fuel and oxidizer mixing at any given throttle rate. Another problem with pintle engines is that because combustion tends to occur in the “surface” of a frustum you end up with a lot of localized thermal impingement on the combustion chamber wall rather than a more evenly distributed combustion across the chamber section and more even heating. This means you get high heating (and the angle changes with adjustment of the pintle) and can get unstable shock phenomena downstream even if the combustion kernel is stable near the injector. There are a number of other considerations with pintle-type injectors which don’t make them particularly well suited for gas phase systems or even liquid phase cryogenics, but for RP-1/LOX and liquid hypergolics they do have certain advantages. However, the real advantage of pintle injectors is that they are relatively simpler and easier to manufacture than “pieplate” and “showerhead” type injectors, and therefore are desirable if you want to develop and built a new engine quickly with less less of a design/test/redesign cycle.
It should also be noted that “pintle injector engines” are not all of a same type in terms of functionality and organization. Without going into detail, current series of SpaceX Merlin engines, while having a heritage from the TRW TR-106 Low Cost Pintle Engine (Tom Mueller, VP of Propulsion for SpaceX was the lead engineer for the TR-106), is a very different beast having gone through substantial evolution since the original Merlin-1A engine. Nor will the purported future SpaceX line of engines, the CH[SUB]4[/SUB]/LOX Raptor engine, which is hypothesized to be a full flow staged combustion cycle, likely use a pintle injector.
The linked Wikipedia article is not particularly useful in explaining function, advantages, and disadvantages of the pintle injector. Most of what you read on space geek blogs, message boards, and websites like Space.com are often incomplete, misguided, or just completely wrong. Many of the people espousing particular technologies or vehicle configurations have absolutely no practical experience with launch vehicles, rocket propulsion, or often even basic engineering principles. The pintle injector engine is not some magic technology that will make rocket engines into cheap, easily built devices which can be cranked out on an assembly line without high precision manufacture and acceptance testing.
Thank you for the excellent summary. I figured this had to be one of those “if it sounds too good to be true…” scenarios but I was struggling to find any signal in amongst the noise of a zillion geeks promoting their pet technologies and, I suspect, a lot of repackaged PR material from TRW and others.
I just wanted to applaud SOAT’s post above. Interesting and informative, on a niche esoteric subject, within a day of the original post. Bravo sir. Have an internet.
Probably why the Apollo program used them for the Lunar Module landing engines- they could throttle them up and down a lot. So even back then, the NASA guys knew about them, and only chose them for one very small part of the system.
That counts as “monster”? It’s 1/6th of the thrust of an F-1 (the Saturn V used five of them), and around 1/2 the thrust of the SSME (the Shuttle used three).
A 250 kip engine is a medium/large class engine; by comparison, the RL-27A engine (used on the Delta II 7000 series) was ~200 kip[SUB]SL[/SUB], and the NK-33 engine (derived from the N1 rocket, used on the OSC Antares and intended for use on the RpK K-1) had a thrust of 338 kip[SUB]SL[/SUB]. This indicates that the pintle injector is scaleable for various thrust sizes, and it is even possible to have a multiple pintle injector engine just as you can have multiple inlet valves on a gasoline engine if you have a chamber too large to feed from a single injector or too difficult to make a single injector with enough throughput.
The TR-107 pintle injector engine (developed by TRW but not fired until it was absorbed by Northrop) was actually an 1100 kip[SUB]SL[/SUB] engine, which positions it between the F-1 and the RS-25 (Space Shuttle Main Engine). By the way, the F-1 is a truly massive engine; I’ve actually walked inside of one at Alamogordo. The RS-25 (418 kip[SUB]SL[/SUB]), while not nearly as large, is also a very large engine.
Does this matter as much with modern manufacturing methods? One can see that the F-1 injector plate was manufactured by, basically, a guy with a drill press (well, probably a manual mill). In this image (from this page) one can see a small defect in an injector hole where the guy got the offset wrong. Obviously this kind of mistake would never happen with a CNC machine and it would take far fewer man-hours to produce.
Yes, it still matters. Even a CNC machine needs to perform manipulations to drill those holes, which in a traditional injector is done via a purpose-designed jig. The bigger issue, however, is the amount of trial and error that goes into finding a jet configuration that achieves optimal performance in mixing while maintaining startup and combustion stability throughout the desired range of throttling. Even with modern hydrocode simulation tools being able to predict combustion phenomena, especially in multiphase flows with interacting shock fronts, is very challenging and is largely validated by informed guesswork and experimental data where available.
Fair enough, but I was speaking of the manufacturing aspects as opposed to design (from your statement “real advantage of pintle injectors is that they are relatively simpler and easier to manufacture”). I can certainly believe that even modern simulation methods are still not quite up to the task of such a complex flow and that a certain amount of empirical testing is still required. But once designed, it seems to me that showerhead-type injector plates are fairly straightforward pieces for a CNC machine to spit out copies of.
I’m not sure what you mean by “a CNC machine needs to perform manipulations to drill those holes”. One thing that all CNC machines are very, very good at is drilling holes. A simple 3-axis machine of sufficient size would be more than enough to machine the top side of the injector plate as long as the materials are not too exotic (I couldn’t say if the bottom side is more difficult, as I haven’t seen pictures). No jig required–just, possibly, some custom workholding.
I’m not sure if you’ve ever seen a typical rocket engine injector, but it isn’t just a simple plate with holes drilled in it. Injectors are often complex assemblies with orifaces, manifolds, feed passages, baffles, cooling channels, and other complex features. The orifaces have to be sized and angled for maximum diffusion of oxidizer and fuel over the range of throttlablility and yet prevent combustion kernels from growing past a critical size. The complexity of most injectors is such that, if tolerances could actually be held, it is one part that would really benefit from 3D metallic printing because designing one that can be easily fabricated and assembled is really challenging. The pintle injector, on the other hand, is mechanically pretty simple, and can generally be made from just two or three major components plus fastening hardware (more if it is designed to articulate, but still less complicated than a really complex showerhead.)
The injector is a key part of the functional technology of the rocket, and a failure to understand good injector design principles (along with a lack of knowledge about heat transfer at chamber walls) was in large measure why it took so long to develop reliable liquid propellant engines. Good injector design is crucial to getting optimal performance out of an engine. From Modern Engineering for Design of Liquid Propellant Rocket Engines (Huzel and Huang, et al):
No other component of a rocket engine has as great an impact upon engine performance as the inejctor. The meausre of delivered performance (specific impulse( is the number of pounds of thrust provided per pound of propellant consumed per second (see section 1.2). Each percentage point loss means a loss in injector combustion efficiency (c*, Eq. 1-32) means a loss of the same magnitude in overall I[SUB]S[/SUB] propulsive efficiency.
This isn’t some thing you just slap out over a weekend in the backyard shop. Most engine design projects have a major second devoted to injector design and ignition stability (the two go hand in hand), seperate from the internal ballistics, chamber wall/thermal jacket, feed system, nozzle entry/throat, and nozzle exit, mounting and thrust vector control, and health and status monitoring instrumentation, and of course the actual test team which fires the engine and collects data. Most liquid engine design teams are several dozen engineers working together around a set of detailed functional and performance requirements.
The page I linked to earlier has some nice pictures of the F-1 injector plate, and it has the features you mention. It’s by no means a simple part, but I’ve done enough CNC work to see that it’s a pretty straightforward job. Everything I can see can be machined from the top down, and although it is very detailed, the detail is fairly uniform–the kind of thing CNC machines are good at.
If I were told to replicate the part on a manual machine, I would suffer some kind of nervous breakdown, so I’m unbelievably impressed at the work they did at the time (on pretty much every aspect of that engine, from the injector plate to the welds). But freehand curves and zillions of holes are the bread and butter of CNC.
No doubt about that at all. The early combustion instability problems in the F-1 could have ended Apollo had they not solved them. Ultimately, it was a major benefit that the Saturn V had over the N1. The Russians could not build a large enough engine and they suffered from it in the form of additional failure modes from the large number of engines.
At any rate, I’m just thinking here about how different designs come into and out of play as technologies improve. The pintle architecture seemingly still has advantages in design simplicity but maybe not so much in manufacturability.
Getting back to the OP:
If SpaceX keeps delivering, pintle engines will more or less have taken over the world. They are building Merlin 1D engines at a tremendous rate and they are gaining flight hours equally rapidly. Already, more have been built than engines that have been around for decades, such as the RD-180.
The reason of course is that there are ten of them on each Falcon 9. And when the Falcon Heavy, they will use 28 of them on each flight. Even if SpaceX only has a small fraction of the total number of flights, the Merlin engine will dominate the engine counts.
I find it rather startling that one of the latest launch vehicles around is built on the back of soviet engines built for their moon program and which have sat for 40 years in a warehouse.
I assume a lot of this is down to avoiding manufacturing cost but the fact that these engines appear to be 100% competetive in performance terms also suggests that the basic technologies haven’t advanced an enormous amount and that those magnificient men and their slide rules really did an astounding job back in the sixties, both in the US and Soviet programs.
If a lot of empirical testing is required to fine-tune the designs then that would make sense, as I believe that will cost just as much (or more) now as it did back then, but without the budgets to support it.
I have personally examined a few of the refurbished NK-33/43 (now AJ26-58, -59) engines (but not the -62 engines used on Antares). These things are genuine works of art; the intricacy and quality of the welding alone is incredible. And getting back to injectors, the injector system on the NK-33/AJ-26 is a very impressive piece of engineering. It has been elsewhere misreported to also be a pintle-style injector which isn’t quite right but it is actually a coaxial injector with orthogonal fuel/oxidizer streams so the concept is somewhat similar, albeit the NK-33 is designed to enhance turbulent “swirl” for optimal mixing. (Unfortunately because of how the engine is constructed, i.e. it is impossible to non-destructively disassemble the welded chamber assembly, it is very difficult to actually see inside without using a borescope.) This engine is actually renown for its unparalleled combustion stability, the recent catastrophic failure of a -62 in ATP at Stennis (alleged to be an aging issue) notwithstanding. Aerojet has actually looked at reverse engineering the design for domestic production, and while there is no official word the general conclusion seems to be that they could not cost-effectively manufacture a clone. Whether the Russians could reopen production and produce NK-33 engines of similar quality is unknown.
A large amount of empirical testing is required for any rocket engine. The combustion conditions and environments that an engine is exposed to are just beyond our capability to effectively simulate without physical test data to tune and validate any model.
This seems to be a common theme. From this article on scanning the F1:
Presumably the fact that these things were cranked out in relatively large numbers during the space race must have helped build up a very impressive skill base.
