Inside The J-2X Doghouse: A2J006 &Turbopump Thrust Balance

Okay, before I even get to the subject of thrust balance, I want to report that J-2X development engine E10001 got back into the stand and that the next test, A2J006, was conducted.  The test went the full planned duration of 40 seconds and the issues that caused us to remove the engine after the previous test all appear to be fixed.  In other words, it was a complete success.  Here’s a video for your viewing pleasure:


Now, speaking of issues on E10001 that appear to be resolved, I want to talk about turbopump thrust balance.  While I hadn’t previously mentioned it, the data from the first several tests showed that there was a thrust balance issue with the oxidizer turbopump (OTP).  So, that naturally leads to the question:  What is “thrust balance?”


Here is a crude drawing of a turbopump for this discussion (And, please, save your cards and letters telling me that I am a terrible artist.  I will already catch plenty of grief from the turbo guys over this drawing.).  The arrows are intended to illustrate fluid flows.



I’ve drawn this in three colors because a turbopump can be thought of as having three fundamental elements.  The first element, the one in blue, is the part that spins.  This is the rotor onto which is attached the pumping stuff (i.e., the inducer and the impeller) on the pump side and the turbine stuff (i.e., the disks and blades) on the turbine side.


The second element, shown in green, is the part that doesn’t spin.  This is the housing.  It is the shell within which the rotor spins.

The third element, shown in red, is the bearings. 


Whenever you’ve got one thing spinning and another thing stationary, you’ve got to have some way of communicating between the two pieces.  Maybe the rotor just spins and slides around in some kind of restraint holding it in place.  Well that might work except that you can’t put grease into a cryogenic oxygen environment and, should too much heat build-up due to friction, then you’ve got a very combustible situation.  Mix heat with a pure oxygen environment and anything and everything will burn, including metal. 


What we do instead is typically use ball bearings between the rotor and the housing.  In some other pumps we use roller bearings that look like little cylinders rather than little balls, and there are other solutions as well.  These bearings do two things, they allow the rotor to spin and they hold the rotor in place.  So they need to be loose enough to spin yet tight enough to maintain control of a heavy hunk of metal spinning at over 10,000 rpm.  How much force the bearing put on the rotor to maintain control and, conversely, how much the rotor puts back on the bearings is called the bearing load. 

When you are designing a turbopump, you design it so that the bearings have a specific load, and that means more than just a value, it also means a direction as shown in the sketch above.  You design the whole bearing package assuming that you know the magnitude and direction of the forces loaded through the package.  This then allows you to do some calculations that tell you, “Yep, I’ve got the rotor properly and sufficiently controlled.”  Those calculations are part of what is often referred to as rotordynamics (and, by the way, NASA MSFC has some of the world’s best experts in that area).


Question: How do you get different bearing loads in a turbopump?  Answer: Pressures.


Here is a nifty little fact: Pressure is all around us.  We don’t feel the atmospheric pressure of 14.7 pounds-force over every square inch of your body because, well, it’s always been there.  That’s the environment within which we’ve evolved.  Now, go to the bottom of the ocean and you’ll find a whole other pressure environment, one that would crush our delicate bodies into mess.  The basic principle to remember is this: Pressure applied over an area equals force.


Within a turbopump, you are explicitly manipulating pressure.  That’s the whole purpose of a pump: to move fluid through the use of pressure.  So, in the pump end you have a low pressure coming in at the inlet and a very high pressure at the outlet.  On the turbine end, it’s the opposite since on that end you’re extracting energy (and therefore pressure) from the fluid.  So, on the turbine end you have high pressure at the inlet and lower pressure at the outlet.  Thus, throughout the turbopump, you’ve got pressures of all sorts and those pressures are pushing on the different pieces of the rotor.  This means that you’ve got all kinds of forces pushing on the rotor.  Some of the forces push the rotor towards the turbine end.  Some of the forces push the rotor towards the pump end.  Getting the right balance of forces is called your thrust balance.




So, what’s the “right” thrust balance?  It is the one to which you’ve designed your bearings.  You design your bearings to have a certain load and a big part of that load is determines by the thrust balance on the rotor.  So, if your thrust balance is off, i.e., not what you intended, then your bearing loads are off and you’re therefore not controlling the rotor the way that you wanted.  In the worst of circumstances, an out-of-control rotor represents a catastrophic situation.  In less-than-worst circumstances, operating with a thrust balance significantly different than the design intent could lead to unexpected wear on components and therefore limited operational life. 


The next question is this: How do you create the thrust balance that you want and need to accomplish all this? 


That’s a matter of designing the internal guts of the turbopump and analytically modeling all of the internal flows within these guts.  In addition to the big flows coming into and out of the pump and turbine, you’ve got internal flows around seals, between pump or turbine stage, and through the bearings (as coolant).  When operating, the whole turbopump is full of fluid.  And, everywhere you have fluid, you have a pressure; and everywhere you have pressure applied to an area, you have a force.  So you calculate pressure drops through small passages or through seals and figure out a whole detailed map of different pressures and forces throughout the whole unit and you can understand your thrust balance.  And then, as necessary, you manipulate your clearances and seals and passages designs until you achieve the thrust balance you need.




Getting back to the E10001 OTP, what appears to have been the issue was that we had some seals that were too tight.  The tight seals overly restricted internal flows and, basically, messed up the intended force balance on the rotor.  This could be seen in the data collected from the testing.  Remember when I said several articles ago that this engine is instrumented to the max so that we can learn all about how it is working.  Here is a case where special instrumentation identified an issue.  The interesting part is that these seals were kinda sorta fine in terms of design but the processing of the material of which they are made — it feels like hard plastic — was not quite refined enough so that when subjected to operational conditions it changes dimensions and actually tightened up. 


Prior to test A2J006, while we had the engine out of the stand, our contractor Pratt & Whitney Rocketdyne was able to partially disassemble the engine and the OTP just enough to change out the seals in question and then get the whole thing buttoned up and ready for test in record time.  They truly did some excellent work here.  And the success of test A2J006 is the proof!


8 thoughts on “Inside The J-2X Doghouse: A2J006 &Turbopump Thrust Balance”

  1. Your detailed and full explanation was enlightening. Thank you.

    However, ignoring weight, I wondered if the friction cost could be overcome by exploring magnetic bearings and using and testing these.

    I appreciate weight is always a limiting factor, but if the engine is, throughout a mission, to be used over again, then surely reliability of such a system could be considered useful.

    I would appreciate your thoughs.
    Regards, Antony

  2. Congrats! It was great to see the engine run a full test 🙂 🙂

    Did you do some kind of unit test to determine that the seal was the culprit? Did you have to chase down a lot of red herrings during the debug process?

    One other question..why was the engine run at 99% power for A2J006? (according to the video description).

  3. @Aaron: In order to help pinpoint the issue, the engineers followed a fault-tree approach. However, a big clue that it was most likely the seal came in the form of the post-test torque checks. It was taking more and more torque to move the turbopump rotor at ambient conditions suggesting that it was being rubbed or grabbed at those conditions. Also, the pump performance was higher than predicted and this too suggested a tighter than intended seal. So, there were lots of clues leading to the conclusion made and the action taken, but it was not until test A2J006 was conducted did we know for sure. Sometimes rocket science involves a little bit of spicy techno-drama.

    The reason that the power level was 99% was because we intentionally aimed a little low. If you remember, we’ve had a couple of early cuts due to having too much power (in part due to the high performance of the oxidizer pump). So we orificed the engine on the slightly low side to help ensure a full-duration test. Note, though, that anything between 97% and 103% is effectively full power level. An open-loop engine like J-2X doesn’t dial itself in like, say, SSME (see earlier article about engine control systems).

    @Antony: I didn’t talk about higher technology bearings because I didn’t want to make the article too complex. Talking about rolling elements as bearings is generally easier to picture for most people because they’ve seen them before. Anyone who has ever fiddled with skateboards or roller skates can picture ball bearings right away.

    However, you are correct. There are other technologies that can be used for bearings. The question that is usually asked is whether the greater complexity of these approaches warrants any potential advantages. I have not personally seen magnetic bearings used in cryogenic pumps, but I am not a turbopump expert. Maybe they’re out there. But one bearing technology that has been used successfully on cryogenic turbomachinery is the use of hydrostatic bearings. Here, the rotor is effectively suspended on a cushion of high pressure fluid, the same fluid that being pumped. If you can get the rotor up and started and properly control the pressure distribution, this is an effective bearing approach.

    Regardless, though, of whether you have rolling elements or non-contact bearing technology, you still have to deal with thrust balance. The bearings hold the rotor in place laterally, but the thrust balance ensures that the rotor stays in the proper place axially.

  4. This was an excellent article. I really enjoy reading technical details about the inner workings of a rocket engine.

    It seems to me by reading this and several of the other posts that these turbopumps are really a big key in the whole engine.

    The fact that we have the best experts and you use vague descriptions in describing their inner workings (and in some of your previous posts you won’t show pictures of their insides) seems to indicate to me that this is what separates us from the rest of the world in terms of being able to build such powerful rocket engines.

    It also is fascinating to see that even some of the smallest parts have such a big impact on the overall workings of the engine.

    Not sure if you have an answer for this, but is it expected that the J2-X will have to work several times during a mission? or does it just fire once? if so, you mentioned in this post that, “In less-than-worst circumstances, operating with a thrust balance significantly different than the design intent could lead to unexpected wear on components and therefore limited operational life.” If it’s expected to only fire off once, can you indeed have less critical tolerances?

    Thank you for helping us understand this beautiful engine.

  5. I’ll bet it makes you appreciate those old engine design pioneers back in the 40’s and 50’s who had to design those first liquid fueled rocket engines without modern computers, and then build them without numerically controlled machine tools. I was visiting a Russian web site showing a cutaway of the guts of the NK-33 engine. Looking at all those seals, bearings and turbines I am still amazed that anything that complex can be built, much less be made to work without blowing itself to bits.
    We heard your 40 second test, 9.1 miles away inside a brick home with the TV playing in Slidell. At first, we thought it was thunder, until it went on and on. That new Infinity Museum should be fun during testing of the new rocket, should it be tested before shipment to Florida. It is 3.7 miles from the test stand. Those glass walls will shake a little, especially during the winter when the wind is blowing from the North.

  6. @Bruce: The reason that I sometimes use vague language to describe design details has two parts. First, because of the International Traffic in Arms Regulations (ITAR) law, it is illegal to not properly control technical data that could be used to inform weapons design. All of the pictures and diagrams and descriptions that get posted here must go through the filter of an export control expert (for whom I am greatly appreciative). Second, there are numerous cases where, beyond simple ITAR considerations, our contractor Pratt & Whitney Rocketdyne is using proprietary designs and/or manufacturing methods at the detailed level. One of the things that we agree to when we sign a contract with a company is that we agree to protect their trade secrets and intellectual property. Thus, everything that I post here also goes through a filter of checking for where I might have inadvertently included proprietary data.

    So, sorry if I am sometimes vague with regards to technical details, but in the interest of being a responsible citizen and a responsible customer, that’s what we have to do.

    And, yes, turbomachinery is a key element of high-power liquid rocket engines. Absolutely.

    Regarding the J-2X mission, it is designed to be capable of two starts during a given mission. However, remember that in addition to the engine starts during the mission, it has to be calibrated and green run on the test stand. So, a new engine out of the box could have several test stand tests by itself, then there might be a stage-level test, and then, after that, comes the mission. If during any one of those tests you get an early cut due to, say, a minor facility issue, then you’ve eaten up some life without accomplishing your objectives. Thus, even for expendable engines, we built them with more operational life than just the scope of the mission.

  7. Thanks so much for this. This is awesome post I ever seen on internet. This is rare to find that’swhy difficult to understand. Anyway, you are definitely someone that has something to say that people need to hear. Keep up the good work. Keep on inspiring the people.
    For more information regards:

Comments are closed.