Inside The J-2X Doghouse: Performance Measurement, Part 2 of 2

In the last article, we talked about the measurement of propellant flow during a test.  Propellant is the stuff we put into the rocket engine.  What we get out of the rocket engine is thrust.  We get propulsion.  Or, in the immortal words of Salt-n-Pepa, 1987, we get push… “Push it real good.”

But how do you measure “push,” or in other words, force?  The simple answer can be found through one of the most frightening household appliances any of us own:  The dreaded bathroom scale…

A bathroom scale works by pushing back at your weight when you stand on it.  Your weight is a force caused by your body mass and the Earth’s characteristic acceleration of gravity.  The scale pushes back with a spring system but deflects slightly under the load.  The scale is then measuring the deflection allowed by the springs at the equilibrium where the spring force exactly counteracts your weight.  More weight pushing down results in more deflection to the equilibrium point and that thereby results in a bigger reading (i.e., think: the Monday after Thanksgiving).

The way that we measure rocket engine thrust is basically the same thing except that instead of measuring something between zero and — as in the bathroom scale picture above — 300 pounds, we’re measuring hundreds of thousands of pounds of force.  Or, in the case of very large engines like the F-1 or the RD-170, we’re measuring over a million pounds of thrust.  That requires a system just a bit more rugged even if the principles remain the same.  What we use rather than bathroom scale and springs are things called “load cells.”  Below is an example of a generic load cell design:

The gray object with the funky cut through it is a metal piece.  As you can imagine, when forces are applied as shown, that slot on the right-hand side will tend to close slightly.  In turn, that would cause the metal on the left-hand side to stretch slightly as the whole thing bends a very little amount.  We measure that slight stretching of the metal on the left-hand side with a strain gauge bonded to the surface of the metal.  A strain gauge is a small electrical device that changes resistance when stretched.  Using an electrical circuit known as a Wheatstone Bridge, we measure small changes in electrical resistance caused by the slight deformations of the load cell.  The amount of stretch can be astonishingly small and yet good strain gauges and good electrical interpretation of the output can yield very accurate data.  The load cell is then calibrated using a known applied load and measuring the resulting strain (i.e., metal stretch).  You now have a more rugged version of a bathroom scale.  Apply a load, get a reading, and, ta-da, you’ve measured push.

Actual load cells used for rocket engines can take different forms from the generic cell shown here.  Any way that you can get an applied load to result in a slight, measureable stretching of metal (while obviously avoiding yielding or buckling) is a valid load cell design.

Above is a picture of a vertical load cell arrangement on test stand A-2 at the NASA Stennis Space Center, where we’ve been testing J-2X development engine E10001.  There are two pieces in series.  The bottom piece, the big chunk of metal with a bunch of crazy holes and slashing cuts through it, is called a “flexure,” which, to me, seems to be a silly name since it doesn’t look very flexible at all.  What it does, however, is effectively make sure that the load entering the load cell is properly directed through the intended vector.  Any skewing of the input off from the intended axis and your results could be erroneous.  The brown cylindrical thing above the flexure is the actual load cell.  You can see the strain gauge wires coming out of it that are fed into the data acquisition system.  This two-piece combination is effectively analogous to a spring in your bathroom scale.

The next item to discuss is how you put load cells into the structure of the test stand so that they can do their job.  On a bathroom scale, the thing that you step onto is essentially a platform “floating” above the base.  It has to be free to move so that the springs can compress honestly.  If there was some interference with this movement, then the reading would be wrong.  The same is true on the test stand when measuring engine thrust.  It is necessary to use a free-floating platform.  The picture below is a drawing of the platform used on test stand A-2.

The engine has a single input point as shown — the gimbal bearing that we’ve discussed before in previous articles — and there are three load cells above the platform.  This is not the only possible way to do it.  Other test stands use a rectangular pattern of four cells.  Or, if it’s a smaller system, you might be able to use just a single load cell.  The important point is that the load cells are in between the pushing engine and the resisting test stand.  Put into the structure of the test stand, and viewed from the side as in the picture below, you can see the whole stack up.  On the bottom is where you attach the engine.  In the middle is the platform into which the engine pushes.  And then the platform is connected to the structure of the stand through the load cells.  The structure of the stand has to be strong enough to absorb the thrust of the engine without distorting.  It has to be fixed (the mythical “immovable object” from physics class).  So, as you can imagine, when you’re talking about hundreds of thousands or even millions of pounds of force, the test stand structure is pretty darn stout.  We usually refer to the primary structure responsible for resisting the force of the engine as the “thrust take-out” structure. 

A final subject to mention is that matter of tares.  Tares are corrections to measured data.  For example, when you step on the bathroom scale and you’re dressed, do you subtract off an estimate for the weight of your clothing and your shoes?  If so, then you’re making a correction for a particular tare.  Of course, you have to do this accurately (honestly).  If I assume that my clothes and shoes are made of lead, for instance, then I can declare that I weigh the same as when Salt-n-Pepa were releasing their first albums.  But that’s not quite the truth.  Getting your tares correct is important for interpreting your data correctly.

When measuring rocket engine thrust you have lots and lots of corrections to the raw data that you measure with your load cells.  This is because, in truth, the gimbal bearing is not the only connection between the engine and the test stand.  While you’d really like to have that perfectly free-floating platform situation, you’ve got to have, for example, propellant feedlines hooked up to the engine.  Flexible bellows are built into the line so that they’re not completely stiff and thereby interfering with the movement of the platform, but they still absorb some of the thrust load and, therefore, make the raw thrust reading skewed.  There are a number of other such corrections that need to be made such that the whole calculation process related to tares can get a bit cumbersome with all its many pieces, but nobody ever said that developing rocket engines was supposed to be easy, right?

Now, between this article and the previous one, you have a good idea of how we get basic performance data from rocket engine testing and also the necessary configuration of the test stands that allow us to gather this information.  The smoke and fire and rumbling roar of an engine test is all very impressive, but for us Datadogs, it’s the data that matters most.  We get lots and lots of data from every test, but propellant flow rates and engine thrust are the most important in terms of understanding how an engine fits into a vehicle and a mission.

10 thoughts on “Inside The J-2X Doghouse: Performance Measurement, Part 2 of 2”

  1. In answer to your poll, I remember using “Brylcreem — A Little Dab’ll Do Ya!” for the popular greasy hair look during high school and college. Question: How do you go about calibrating load cells for the extremely high forces that are to be measured on your test stand?

  2. @Steve: Good question, as usual.

    There is a two-phase answer to your question. I know the answer to the first part. The answer to the second part will require some research on my part.

    The first part is this: How are the three vertical load cells on stand A-2 calibrated? This is a good question because they’re not easy to get in and out of the test stand. Matt Strickland is the helpful gentleman who took the photos of the test stand components for the last two blog articles and he had to do some crawling around to get to the load cells. They’re deeply embedded in the structure of the stand. So, you would not want to have to pull these things out and calibrate them off of the test stand very often. Instead, we have a calibration system built directly into the stand.

    The calibration arrangement involves a hydraulic ram and a single calibration load cell. The hydraulic ram is built into the structure so that it applies a load to the platform and the load measured by the three vertical load cells is compared to the load measured by the calibration cell. You can do this calibration test with the engine removed, with the engine installed, and with the engine install and cryogenic propellant loaded. In this way, you can not only calibrate the load cells by comparison to the calibration cell; you can also get a measure of the tares resulting from engine installation and the impacts of the cryogenics on the feed system.

    The calibration of the vertical load cells is then tracked over time to make sure that they remain reasonably consistent. If a calibration point for a cell starts deviating, it might be due to a failure in the cell, or it might have to do with a change in the test stand configuration. I remember years ago we saw something odd in the measured thrust data that kicked off a minor investigation. It turned out that a technician had welded a small, rigid line to the thrust platform no realizing that this massive structure is actually supposed to “float” free of such encumbrances. It’s kind of like holding onto something when you step on the bathroom scale. You’re interfering with the function. The welded line was removed and all went back to normal.

    Now, the second part of the answer to your question is this: How is the calibration cell calibrated? It is, after all, the referee for the whole system. The answer is that we hand it over to a calibration lab and they take care of it. Yes, unsatisfactory answer. I honestly don’t know how they apply a controlled load to the cell. Do they stack a bunch of weight on it? Do they use less weight but a lever system? Or, do they apply a controlled load hydraulically somehow? It’s a lot of load, upwards of a million pounds, so it involves some thought. I will ask around and see what I can dig up.

  3. @ Steve: After a bit of poking around and talking to people, it would seem that the answer as to the question “How do you calibrate a calibration load cell?” is to use another, “gold standard” calibrated load cell. And so I asked, “Okay, but how do you calibrate the ‘gold standard’?” Well, by another standard one. Duh. And so on.

    This was not a pleasing or sufficient answer. The whole thing was starting to get a bit like the chicken-or-the-egg paradox. Finally, though, it became apparent with a little research that the ultimate answer – the egg before the first chicken – is that calibration is done in a lab using a deadweight machine. What’s that? Well, it’s a machine that actually loads weight onto the load cell. Yes, that much. Lots and lots of weight. And, in terms of precision, they take into account local variations in the Earth’s gravity field and even buoyancy of the weights in air in their calibration procedures.

    According to the National Institute of Standards and Technology website, and some of the reference material that they cite, they can go up to 1,000,000 pounds-force using these deadweight machines, in increments of 50,000. Have you ever used a machine in the gym with a built-in weight stack? Their description kind of makes me think of something like that but on a much, much larger scale. That’s pretty cool.

    Note that for calibrations over 1,000,000 pounds-force they revert to using a series of deadweight-calibrated load cells and, I assume, a hydraulic ram. They said that they can go up to 12,000,000 pounds-force. Now, if I ran NASA, I’d take that as a challenge to build a rocket engine that goes right up to the limit. And that, right there, is a clear illustration why we shouldn’t let rocket guys like me run the agency! We’re all a little nuts.

    Thanks for the question that prompted me to go off and learn something new. Truly, no day is wasted in which you learn something new.

  4. Didn’t NASA get close to that 12,000,000 pound force when they were testing all 5 F-1 engines at once for the Saturn rocket?

  5. The Saturn V vehicle first stage had five F-1 engines. Each engine put out 1.5 millions pounds-force thrust. That’s a total of 7.5 million pounds-force. So, yes, that’s getting up there. I’ve been told that back when they tested these stages locally, they’d inform the entire region of North Alabama so that nobody would panic and assume that it was an earthquake. The tests would rattle windows throughout the county.

  6. I know this is getting quite off topic. My apologies. The SLS heavy lift rocket will eventually be designed to lift 130 tons (plus itself). That’s an incredible amount of weight to lift off of earth.

    How many millions of pounds-force thrust do you believe will be needed to get this massive rocket off the ground?

    And, I hope this question isn’t too dumb. But, what is the difference between a rocket engine and a rocket booster?

    Thank you for taking the time to answer all these questions and making “rocket science” so interesting!!

  7. The Saturn V vehicle weighed about 6.5 million pounds sitting on the pad before liftoff. The thrust necessary to get that moving was on the order of 7.5 million pounds-force, i.e., five F-1 engines at 1.5 million pounds-force each.

    The final configuration for the ultimate SLS Block 2 vehicle is not entirely resolved. There remains the open issue of whether the boosters are to be solid or liquid propellant. At the most extreme, it would seem that the weight of the vehicle on the pad could be as much as 7.5 million pounds. That would mean the liftoff thrust has to be better than 8.5 million pounds-force. Again, these are the biggest numbers that I’ve seen. Some variations show slightly lower numbers.

    Now, “booster” versus “engine”… I’ve illustrated in the blog what we typically think of as a “rocket engine.” It uses liquid propellants and it generates its own pumping power to pull in the propellants and generate thrust. The term “booster” is used a bit more loosely. We often refer to the large solid-propellant rockets on the sides of Space Shuttle stack as boosters but there are SLS vehicle options where these boosters are replaced with liquid propellant boosters propelled by liquid propellant rocket engines. So, “booster” is more of a generic term for a first stage, oomph-off-the-pad stage that could be either solid or liquid propellant.

  8. Off topic but sent in request for answer but no answer received …

    J-2X engine removed from A-2 Test Stand
    Pages 3-4 LAGNIAPPE October 2012
    NASA removed J-2X engine No. 10001 from the A-2 Test Stand in early October.

    Can you tell me the date(s) it was removed.

    Also if possible Lagniappe article on Blue Origin firing gave no date and I asked for it also with no response. Can you provide?

    Any and all help appreciated. THANKS!

  9. Please keep up the excellent Blog installments! I find them extremely interesting and informative and look forward to each and every one. Facinating reading and very, very good Blogging! You sir, are a natural! I find the develpoment of the J-2X endlessly interesting and educational. Please more! Thank you! – Jeffery Skinner (spaceflight nerd).

Comments are closed.