Dynamic rocket engine regulation

Dear readers,

On Saturday February 4th we had the great pleasure to make our “Dynamic Pressure Regulation” truly dynamic. Until Saturday we have utilised the DPR-system to maintain a constant pressure in the tanks, and hence a constant pressure in the rocket engine. During this test we planned to use the system more actively, and let the engine go through a number of operating points, with varying chamber pressure and varying O/F-ratio (Oxydizer to Fuel ratio), by varying the pressure in the propellant tanks.

By recording data from different O/F-ratios and chamber pressures we gain a more detailed knowledge regarding fuel consumption, chamber pressure, and thrust as a function of feed pressure. This is an important prerequisite for being able to empty the propellant tanks syncronously during the Nexø II flight this coming summer. This is why we tested with BPM 5 #3 (BiPropellant Motor 5kN), as this is the engine that will power Nexø II.

IMG_0138

BPM 5 ready for test inside the test container. Photo: Mads Stenfatt.

We have had access to a capacitive fluid level sensor for a couple of the latest tests, which we have used to measure the level in the LOX (Liquid Oxygen) tank. Unfirtunately we only had it on loan, and have recently been asked to deliver it back to the owner. As the cost of the instrument is around 12.000 kroner (1,700USD) we have chosen to develop our own version, and the engineering model was tested this past Saturday. I won’t go into more detail as a blog about this project is forthcoming.

IMG_20170203_210357

The LOX level sensor developed in-house is calibrated in a LOX filled aluminium tube. Photo: Emil Møller.

All in all the tests went really well. If you followed the live stream you probably noticed a couple of minor problems. In both instances errors that we had not encountered before. The first error was related to the main valves which refused to open during our standard pre test check procedure. The error was rectified by power cycling the test stand. The other error concerned the regulating valves which controls the amount of helium loaded into the propellant tanks. During pressurization at both test 3 and test 4 one of them refused to open correctly. This meant the booster team had to return to the test stand, to assess the situation and debug the system. It was soon realised that the pressurisation system was very cold after two test runs, we actually found ice on the regulating valves. This hasn’t been observed previously, but then the test conditions we had this Saturday differed somewhat from what we are used to. Doing several very long burns on the same day was a first, and we just have to accept that the pressurisation system will be severely chilled after a long engine burn. It should hardly be surprising that while expanding a gas from 200 bar to around 20 bar, the piping and valves will be chilled down. This is just the first time we do several long burns, one right after the other, which is why we haven’t experienced this phenomenon before. The solution was straightforward, warming the valves a bit made them operational again.

Below we have compiled a video of how the day and the four tests went.

 

BPM5 #3 is the least used engine we have built. Most of our development work have been done utilising BPM5 #1 (Franken 5), while BPM5 #2 had several runs before its flight on Nexø I. But BPM5 #3 has only been run once before this test, meaning just 7.5 seconds of running time and 33 kNs total impulse up until the test this weekend. Utilising the upgraded propellant tanks and the well behaved DPR-system on the test stand we were able to clock 122 seconds of additional operation and a staggering 617 kNs impulse! Now we need to have a look at the data we aquired!

Test #1

For the first test of the day we aimed at maintaining an O/F-ratio around 1.3, and go from 12 to 16 bar in chamber pressure. We increase the pressure in steps every 6 seconds during all tests described here. The step size varies somewhat from test to test.

Data from Test #1.

Data from Test #1. Click for larger view.

There’s a fair amount of data collected form each of these tests, let’s summarise what is shown on the graphs.
Top left shows tank pressure, chamber pressure, and the target pressure for the DPR-system. Top center shows various pressure drops in the system, injector pressure drop for both fuel and LOX, pressure drop in the piping between tanks and engine, and fuel pressure drop across the cooling mantle. Top right is the thrust, both the raw 10 kHz signal and the filtered version. Middle left shows fuel temperature before and after the cooling mantle, and the difference. Middle center shows accumulated propellant flow from the flow meter outputs. Middle right is propellant flow and O/F-ratio. Bottom left shows specific impulse, based on thrust and fuel consumption. Bottom center is pressure in the high pressure system and gas consumption as a function of time. Finally bottom right showing LOX tank level. Please note that the time span for the last two graphs are longer than the rest, as pressurisation is included here.
Looking at chamber pressure and thrust, the pressure rise every sixth second is obvious. A rather nice of graphs, if I may say so myself. It is also clear that the chamber pressure is a bit higher than we aimed for, the pressure is between 13.1 and 16.8 bar, resulting in a thrust value between 4,500 and 6,000 N. A new record, if I’m not mistaken, for the BPM5 engine, which is designer for 5000 N thrust at 15 bar chamber pressure.
Keeping the O/F-ratio at around 1.3 has been less successful. We are actually more than a bit off, and end up with a ratio around 1.0. But that’s part of the reason for running these tests, to learn and trim the parameters. The low ratio led us to adjust the plan for the following tests, to increase the O/F-ratio.
Currently the DPR-system is implemented with only a P-term in the control loop. As can be seen this results in an offset between target pressure and resulting pressure in the tanks. The offset is a constant 2 bar approximately, which we allow for in the test planning, but we need to add an I-term to the control loop to get rid of this offset.
As mentioned above, the helium consumption is shown bottom center. During pressurisation of the two tanks, at T-50 s and T-40 s, consumption is high, peaking at 550 standard liters per second. But notice how swift pressurisation is performed! Just 5 seconds to pressurise each propellant tank, which is very cool. Also note the four peaks while the engine is running, this is the four changes in feed pressure showing up. We can also see that it takes 1.5 second to reach the new target pressure.
The propellant flow graph for the first test looks a bit fuzzy, at least for the fuel. The reason is unknown to me, but it doesn’t show up in the three subsequent tests.
The fuel temperature doesn’t vary significantly with increasing chamber pressure, maybe a slightly increasing tendency. This is ti be expected, as the heat flux to the chamber wall increases with increasing chamber pressure. It is however not very clear.
The signal from the capacitive LOX level sensor looks rock steady. There’s a small spike when the tank is pressurised. Exactly the same spike occured with the commercial sensor that we had on loan previously. A funny thing is the fact that the graph shows negative values. A bit weird, as the sensor doesn’t reach below the 17 liter level in the tank, which results in a 17 liter minimum. Still the signal goes negative. We will probably have to look at calibration once more. Emil and Mads will probably tell more about this in their forthcoming blog.
Despite running with an O/F-ratio below the preferred, the graph for specific impulse shows a result around 190 s at the largest pressure, which isn’t bad at all!
Comparing with data from earlier tests, one will note that we have shortened the pre stage duration cinsiderably, it is now down to 500 ms. There’s no reason for the rocket to expend fuel running idle, which is why we want to reduce pre stage to a minimum. The 500 ms is probably very close to a safe minimum.

Test #2

The procedure for test #2 was identical to test #1, apart from the fact that we increased the LOX tank pressure to obtain a higher O/F-ratio. That helped, but not sufficiently.

Data from Test #2.

Data from Test #2. Click for larger view.

The graphs are very similar to those from test #1, although the LOX tank presure has been increased, as mentioned. This results in an increased LOX consumption, and a higher O/F-ratio value between 1.12 and 1.25.

Test #3

The LOX tank pressure has been increased again, to obtain higher O/F-ratio values.

Data from Test #3.

Data from Test #3. Click for larger view.

The O/F-ratio is now at 1.26 to 1.36, which is closer to where we want it to be. The specific impulse does not concur however, as this has decreased to 183 for the largest pressure. A bit odd again.

Test #4

Last test of the day, the main change is again increased LOX tank pressure.

Data from Test #4.

Data from Test #4. Click for larger view.

Now the O/F-ratio is at 1.39 to 1.47, and specific impulse is 189 s at 16.5 bar. The temperature increase in the fuel through the cooling mantle has suddenly increased significantly, it is now at around 70 degrees where it was between 45 and 50 degrees at the previous test runs. A higher O/F-ratio will naturally lead lead to a larger increase (hotter combustion and less fuel to cool the mantle), but the difference is quite significant.
We aquired a lot of very useful data, as you can see, and measured in total impulse this was our ”biggest” test to date. There’s a bit of data crunching still to be done, in order to get the most out of the test. On top of that we want to upgrade the DPR controller before we launch Nexø II, to get rid of the obvious offset between target pressure and resulting pressure. This will likely lead to an additional static test before we launch Nexø II.
We are also contemplating a possible all up static test of Nexø II before launch. The fault on Nexø I might have been found and rectified if we had performed at static test of the complete rocket. This make it compelling to do a full test of the Nexø II rocket. On the other hand, if something akin to what happened to HEAT-2X happens, we will be without a rocket come summer. A tough decision .. but one we will have to take very soon.

Leave a Reply


*