Tag Archives: powerpack

J-2X Progress: Once Upon a Time at Stennis…

Posted on by .

I enjoy movies.  I don’t get to watch much television due to other endeavors that consume much of my time, but if I do it’ll almost always be one of four things on the screen:  some news program, a sporting event, a history program, or a movie.  And I like lots of different kinds of movies.  Some of my favorites include: The Hustler, Singing in the Rain, Rocky, Schindler’s List, Barfly, Hannah and Her Sisters, Fargo, The Apartment, The Godfather, Leaving Las Vegas, The Deer Hunter, Hoosiers, Nobody’s Fool (the Paul Newman one).  I don’t believe that one could decipher a pattern from that list other than the fact they all follow the classic narrative structure:

Think of the classic “stranger comes to town” story.  (1) It’s a quiet little town and all is peaceful.  (2) Then a stranger comes to town and stirs up all kinds of trouble.  (3a) In the end, the stranger marries and settles down with the prom queen and everyone learns to live with one another.  Or, (3b) in the end, the stranger ends up mysteriously dead and lying in the gutter along the road leading out of town and they secretly bury him promising never to mention it to anyone from out of town.  Or, (3c) in the end, the stranger ends up mayor of the town by exposing and driving out the secretly corrupt sheriff.  Obviously, the possibilities are endless and that’s why there are thousands and thousands of stories to be told.  But the root of all of this is the middle block, “…something disturbs that situation and troubles ensue…”  Nobody ever tells an interesting story where nothing happens.  And with no “troubles” of some sort, nobody cares about the resolution.

So, that brings me to rocket engine testing and the fact that it is always interesting.  This article is intended to bring you up to date on the status of our J-2X test campaign at the NASA Stennis Space Center in southern Mississippi.  Remember, we last left our heroes on test stand A1 with PowerPack-2…

Test A1J015, J-2X PowerPack-2: It ran 340 seconds of a planned 655 seconds duration.  The test profile called for simulated primary mode and secondary mode (i.e., throttled) operation.  Also, throughout the test, turbomachinery speed sweeps were planned meaning that we systematically varied turbine power, increasing and decreasing, to force the pumps through a broad range of conditions.  It was during one of these sweeps that the fuel turbopump crossed a minimum speed redline and the test was cut short.  Before the test, we knew that it would be close and the analytical prediction was just enough off from reality to cause the early cut.  Nevertheless, most of the primary objectives were achieved and the test was a success. 

One of the things that we often talk about when discussing an engine test is the “test profile” or sometimes the “thrust profile.”  The test/thrust profile is the plan for what you’re going to do during the test.  When we say that we had a planned duration of 655 seconds, that value comes from the test profile that is agreed upon prior to the test.  Usually a test/thrust profile is a single page showing engine power levels and propellant inlet conditions, but for these complex PPA-2 tests, the test profile can be expanded to include such things as these turbomachinery speed sweeps.  To give you an idea of what an engine test/thrust profile looks like, here is one for a Space Shuttle Main Engine (SSME) test performed back in 2001.  It contains a wealth of knowledge about the test to be run.

Test A1J016, J-2X PowerPack-2:  It ran 32 seconds of a planned 1,130 seconds duration.  In this case, unlike the previous test, because we cut so early we can’t really say that it was mostly a success.  However, every time that you chill an engine, successfully get it started, and shut it down safely, you have accomplished something significant and you are always collecting data and learning.  The early cut in this case had nothing to do with the PowerPack-2 performance.  Rather, it was a facility issue, a hydrogen fire due to a leak.  As I’ve said before, the PowerPack-2 is an oddball test article in that it is half engine and half facility.  That makes the interfaces technically difficult in some cases due to thermal and structural loads.  The leak and fire in this case was on the facility side near one of these difficult interfaces. 

Below is a picture captured off a video taken during the test and behind the structure and the piping you can see the bright orange flame that resulted in the early cut.  This issue of hydrogen leaks and fires has been somewhat recurring so a team of NASA and contractor folks stepped forward to work towards a resolution of the issue.

Test A1J017, J-2X PowerPack-2:  It ran the full, planned 1,150 seconds duration.  That’s over 19 minutes of continuous rocket engine operation and that’s pretty amazing.  It was the longest, most complex engine test ever conducted across the long history of the NASA Stennis Space Center A Complex.  We did some wacky stuff on test stand A1 during the XRS-2200 (linear aerospike engine) development effort and there were a couple of longer SSME tests in the B Complex twenty-some years ago, but test A1J017 stands out for the combination of complexity and duration.  The test profile contained over a dozen unique, steady state “set points,” i.e., prearranged combinations of engine operational conditions and facility boundary conditions.  The objectives of this test included speed sweeps for the oxidizer turbopump and an examination of cavitation performance for both the oxidizer pump and the fuel pump.  Pulling off this test was a dazzling success with many people deserving credit.

So, trouble ensues (hydrogen fire on test #16) and the combined team of NASA Stennis, NASA Marshall, test operations and support contractors, and Rocketdyne worked through to a resolution of the issue and a new situation of unprecedented success has been achieved.  It’s easy to write a blog like this when reality lines up so conveniently in the narrative form. 

But back at the ranch, our heroes find J-2X development engine E10001 on test stand A2…

To refresh your memory, we’d last tested E10001 on stand A2 back in December of last year.  Back then, we were testing the engine without a nozzle extension and not using the passive diffuser system on the stand.  This year, we were going to get back to testing E10001 but now with a nozzle extension so that necessitated use of the passive diffuser.  The Stennis folks installed a clamshell and seal apparatus that connects the engine to the diffuser thereby allowing the diffuser to “suck down” to pressures lower than sea level ambient.  In my crude sketch below, I try to show you how this fits together.

A key piece in this arrangement is the clamshell seal.  Whereas the engine is obviously metal and the clamshell and diffuser and big pieces of structural metal, the clamshell seal is a fibrous/rubber-ish piece that has to provide the seal that allows the whole thing to work together and simulate altitude operation when the engine is running.  It has to be strong yet compliant so as to accommodate movements of the nozzle during hot fire.  To give you an idea of how strong it needs to be, let’s calculate the force imposed on the seal during operation.  Ambient sea level pressure is 14.7 psia (pounds per square inch, absolute).  Let’s say that in the diffuser, during operation, it will be about 10 psi lower than sea level ambient.  In reality, the pressure will be slightly lower than that, but 10 is a nice round number to work with.  Let’s further say that the diameter of the nozzle at which the seal is attached is about five feet (or, 60 inches).  That’s pretty close to reality, give or take a bit.  And, let’s say that the seal itself is about six inches in width.  So, the total area of the seal is:

So, if the pressure differential across the seal is 10 pounds per square inch and you have 1,244 square inches of surface area, then that makes for over 12,000 pounds of force — or more than 6 tons!  Wow, so that seal and the brackets that holds it in place still needs to be pretty darn tough.

Test A2J011, J-2X E10001: It ran 3 seconds of a planned 7 seconds duration.  The early cut was due to a facility redline violation; specifically, the measured pressure within the clamshell did not drop down the way that it was supposed to.  Post-test inspections quickly revealed why this redline violation occurred.  The clamshell seal was torn up.  If the seal doesn’t seal, then the pressure differential is not maintained and so, appropriately, we tripped a redline.

An informal team was assembled of NASA, contractor, and Rocketdyne folks and the design deficiency was quickly identified.  New parts were designed and fabricated and, in a matter of just a couple of weeks, we were once again ready for test.

Test A2J012, J-2X E10001:  It ran the full, planned 7 seconds duration.  The objectives for this test were to demonstrate that the clamshell, seal, and diffuser arrangement was properly working and to perform a bomb test in the main chamber.  The testing arrangement worked perfectly and the bomb test did not reveal any combustion stability issues. 

Test A2J013, J-2X E10001:  It ran the full, planned 40 seconds duration.  This was yet another bomb test and again there was no combustion stability issue uncovered.  The neato thing on this test was that while the engine started to primary mode operation (i.e., 100% throttle), it switched to secondary mode operation (i.e., throttled) mid-test.  This was the first operation of the complete J-2X engine (as opposed to just the powerpack portions) in secondary mode. 

Test A2J014, J-2X E10001:  It ran the full, planned 260 seconds duration.  This test represented several more “firsts” for J-2X.  This was the first time that the J-2X was started directly to secondary mode.  It was the first time that the J-2X switched, in run, from secondary mode to primary mode.  This was the first J-2X test with a stub nozzle extension that offered the opportunity to perform an in-run calibration of the facility flow meters and, in so doing, provide for a good estimation of engine performance.  It turns out that E10001 is, to our best understanding, exceeding expectations in terms of required performance.

Again, the old narrative structure holds:  New guy comes to town (the stub nozzle extension).  The situation changes (new test stand configuration to accommodate the stub).  Troubles ensue (the clamshell seal gets torn up).  Resolution is found (new design for clamshell seal attachments).  And a new situation is achieved (we’re knocking off successful test after successful test). 

But, there is a twist (literally) to our denouement.  I’ll explain this twist by starting with a picture:

Can you see it?  This is a picture of the fuel inlet duct.  Remember, this duct has an inner and an outer shell (or bellows as we call them) so that in between there will be vacuum to keep the hydrogen cold, like a Thermos® bottle.  Between tests, one of the customary inspection techniques used to ensure that you’re good to go for the next test is to do a series of helium leak checks.  You systematically pressurize different portions of the engine and make sure that everything is still sealed up tight.  Well, when they pressurized this portion of E10001, they got what we’re calling “squirm.”  If you look closely at the duct you’ll see that on the left-hand side the convolutions are bunched together and on the right-hand side they’re spread apart.  This indicated that there was leak in the inner bellows of the duct so that the cavity between the two bellows was pressurizing with the leak-check helium.  The squirm effect was due to the outer shell was deforming — squirming — due to that pressurization of the vacuum cavity. 

Now, there are several important things to note about this.  First, this particular duct is a heritage piece of hardware.  It was not made for E10001.  It was made during the Apollo era for J-2 and J-2S, forty years ago.  It had seen its fair share of hot-fire history long before it reached E10001.  Second, the new ducts being built for J-2X have a design modification that ought to mitigate this kind of failure.  Third, we can see in the test data, with perfect hindsight, exactly when the leak occurred in test A2J014 and the engine ran for some time with the leak and nothing catastrophic happened.  Thus, while nobody is happy when something breaks, in this case there’s no need for overreaction.

Getting back to the narrative structure and this little twist at the end, I kind of think of this like a teaser — a cliff-hanger — that leads to a sequel.  Will our intrepid heroes dig their way out of this situation?  Will the test program recover and move ahead to new successes and glory?  Or will the monster creep up from the dark, dank Pearl River swamps and terrorize the test crew…?

…oops, wrong movie. 

[Hint:  We’ll be fine.  Already moving out at full speed.  In the immortal words of Journey (i.e., Jonathan Cain, Steve Perry, and Neal Schon) “Oh the movie never ends. It goes on and on and on and on…”]


 

J-2X Progress: A New Star on Our Horizon

Posted on by .

J-2X Progress:  A New Star on Our Horizon

For weeks and weeks (or months and months really), we’ve been going on and on about the star of our J-2X project, development engine E10001.  And there is every reason to focus much of our attention on this first example of our new engine.  It has really put on one a heck of a show, generating oodles of data, and we’re far from being finished with it.  


 
So, E10001 is unquestionably a star.  Beyond this, however, we have other potential stars waiting in the wings.  I would liken this situation to “American Idol” except that I’ve never actually seen that show and, further, all of our test articles are not in competition with each other.  Indeed, the whole point of a coordinated and integrated development plan is for all of the test plans and test articles to complement each other.  One big star that will soon be making an important contribution is called “PowerPack Assembly 2” (or “PPA2”).  Okay, you’re saying to yourself:  I know what an engine is, but what is a “powerpack assembly”?  And, why is this number two?  Good questions.  We’ll start with the first one…

A powerpack assembly — or simply a “powerpack” — is a subset of the total engine.  Specifically, it is the engine minus the thrust chamber assembly (i.e., the main injector, main combustion chamber, and nozzle/nozzle extension).  About a year ago, I wrote an article here in the J-2X development blog talking about what a gas-generator cycle rocket engine looks like.  The schematic of that cycle is shown below for reference and comparison:


Where:
MCC = Main Combustion Chamber
GG = Gas Generator
MFV = Main Fuel Valve
MOV = Main Oxidizer Valve
GGFV = Gas Generator Fuel Valve
GGOV = Gas Generator Oxidizer Valve
OTBV = Oxidizer Turbine Bypass Valve

The lines and arrows in red denote fuel (hydrogen) flow; the green lines and arrows denote oxidizer (oxygen) flow; and the gray lines and arrows denote the flow of combustion products.  Using the same abbreviations and same color schemes, here is the schematic for a gas-generator cycle powerpack:


See?  As I said, you simply pull off the whole thrust chamber assembly and there you go: powerpack.  If you think of the thrust chamber assembly as what you use to make thrust, then the powerpack portion of the engine is what you use to feed the thrust chamber assembly.  In other words, to be particular, it’s the gas generator, the turbopumps, and the full set of major control valves…plus, of course, the lines and ducts that connect everything together.

What this configuration allows you to do, far more so than the complete engine configuration, is “play games” with turbomachinery conditions and operations.  And here’s why.  On the full engine configuration, you have to feed the thrust chamber assembly a pretty steady diet of fuel and oxidizer.  If you deviate too far, things get too hot or too cold or you get too much pressure in the chamber or too little.  The thrust chamber assembly is a wonderful piece of equipment, astonishingly robust when functioning in their normal regimes, but it’s basically static and, to be honest, a bit persnickety when it comes to significantly off-nominal operations. 

So, you first get rid of the persnickety thrust chamber assembly to give yourself more flexibility and then, taking the next step, you get creative with the valves.  On the complete engine configuration for flight, the J-2X engine has pneumatically actuated valves.  As we’ve discussed in the past, this means that they have two positions to which they are actuated: open and close.  We can’t partially open or close them and hold them in intermediate positions thereby altering or directly controlling the propellant flows through the engine.  But for powerpack, we’re not so constrained.  For powerpack, we will use electro-mechanical valve actuators for the two gas generator valves (the GGFV and the GGOV) and we will use hydraulically-actuated facility valves to simulate the two main valves (the MFV and the MOV).  All four of these valves will then no longer be simply open/close.  They can be held as partially open or closed and, using these as control tools, we can vary temperatures, pressures, and flowrates throughout the powerpack.  We can vary the power with which we drive the turbines.  We can vary the downstream resistances seen by the pumps thereby altering the flows and pressure-rise profiles through the pumps.  The OTBV — the valve that we normally use to alter engine mixture ratio by applying differential power levels to the two turbines — will not be actively actuated for the powerpack testing, but it will be configured such that we can alter its fixed, incremental position from test to test.  In that manner, we can use the OTBV position variations to explore inlet mixture ratio deviations on powerpack that the full engine configuration simply couldn’t tolerate.

Thus, the powerpack assembly configuration is first and foremost (though not exclusively) a test bed for the turbomachinery.  Just as with the “bomb test” philosophy discussed in the previous article, we already know that the J-2X engine works, but now we need to further explore the detailed implications of the design.  We need to anchor and validate our analytical models, demonstrate operations across the spectrum of boundary conditions and environments, better characterize our margins, and exercise the full slate of design features and operational capabilities.  The powerpack assembly test series is one very important means for doing this.

Okay, so it’s a useful test article, but where does the actual Powerpack Assembly 2 stand?  Well, while we’ve all been heavily (and appropriately) focused on the testing of J-2X development engine E10001, our contractor, Pratt & Whitney Rocketdyne, has been also quietly assembling Powerpack Assembly 2 back in the engine assembly area.  Here is a picture of the complete Powerpack Assembly 2.


It kind of looks like an engine, almost, doesn’t it?  Well, that’s because we assembled it kind of like an engine but used a “dummy” thrust chamber assembly.  You should recognize the yellow thing that looks like a cage.  That’s the nozzle simulator that we used early on in the assembly of E10001.  Sitting on top of the nozzle simulator is a simulated main combustion chamber and a simulated main injector.  By making it look so much like a regular J-2X engine, it allows us to install the PowerPack Assembly 2 into the test stand much like we do a regular engine.  The only special adaptations are lines to catch the propellants coming out from the pumps and the discharge coming from the turbines.  In a regular, full configuration engine all of these flows get routed through the thrust chamber assembly to produce thrust.  For PowerPack Assembly 2 testing, these fluid streams are collected and disposed of off of the test stand.

Next is a picture of the PowerPack Assembly 2 being carefully loaded onto the truck to transport it out to the test stand.  Road trip!


PowerPack Assembly 2 will be tested on test stand A-1, which is the sister test stand to A-2 where E10001 is currently being tested.  Here, below, are a couple of pictures of PowerPack Assembly 2 being lifted onto and then sitting on “the porch” of A-1.  In the background you can see a portion of the canals that weave in and around the big test stands at the NASA Stennis Space Center.  Nowadays, these canals are mostly used just to transport barges full of propellants.  But back in the Apollo era, these canals were used to transport whole rocket stages in and out of the test facilities since they were too big for trucking.


And here, is Power Pack Assembly 2 installed into the test position on stand A-1.  Many kudos should be extended to our diligent contractor Pratt & Whitney Rocketdyne and our faithful partners at the NASA Stennis Space Center for making this milestone possible.  Great work guys!


Now, getting back to that other question regarding the “2” part of “PowerPack Assembly 2.”  That denotation is simply there because this is the second powerpack assembly we’ve tested as part of the J-2X development effort.  PowerPack Assembly 1 testing was conducted about four years ago using residual hardware from the XRS-2200 (linear aerospike) development project.  While that first PowerPack Assembly did not use any true J-2X hardware since that hardware was not yet designed or built, it did help inform the J-2X turbomachinery designs.  It used what were essentially J-2S turbopumps to explore J-2X-like operating regimes.  The J-2X turbopump designs then began with the J-2S designs and made the changes necessary to fulfill the J-2X mission.  Another way of looking at this is that PowerPack Assembly 1 was used to inform the design and PowerPack Assembly 2 will be used to validate and characterize the design.  To me, this sounds like a very nice pair of bookends on either side of the J-2X turbomachinery development effort.