Linear Pulsejet Combustion simulation.

[Graham C. Williams]

Dear All.
I've developed a program that simulates (I'm not sure if that’s the right word) a running valveless pulsejet or not. The Gas dynamics are handled by Sam Masons NUDiS program. The rest (detection, combustion etc) are my own work and is done within MS VB and MS Excel. No way have I attempted to simulate combustion, only the results of combustion. I'll leave the simulation stuff to people as bright as Sam (Many thanks for the Coffee and chat).
The program is responding very well to my collection of known working motors, in that results follow experience and suggest improvements. Front-loaded motors and Rear-loaded all seem to respond well. Even those that run in the next mode up settle (those with a P-node approximately in the centre of the motor) after 1000 iterations. Unpacked Chinese motors are no problem either. Some of the Escopette motors are not so good but I can see a way forward. I think the problem concerns the double breathing of these motors. Finally, Combustion tracking is reasonable.

I'd like to test the program on some motors that are not known to me. If someone would be so kind as to let me have 2 designs (A and B don’t tell me which is which), one that is known to work and the other that has defied all attempts at starting. I’d like to attempt to predict the results. I’ll, of course let you have the full results/analysis of any tests I do along with an analysis excel sheet so that you can study them.

Graham.

[Dave]

Graham
Although I don't have designs worthy of submission at this time I will be watching with great interest.
Dave

[Mike Everman]

Hi Graham. Great work!
I have geometry that didn't run, but fueling was new to me at the time I tried running them, so maybe they're not "reliably unworking" ha.

I do think it would be telling to analyse two running embodiments of the same motor with merely a tail pipe length difference that made for 50% more thrust. This one is a straight tail linear.

Will your program bring in an existing geom.dat? It's not many dimensions in any case...

[Graham C. Williams]

Dear Mike.
I've not tried importing a Geom.dat file direct. I'll have to put that on the todo list.
I agree about the tail pipe change test. Can you send me the dims for the two versions of the motor?
(Note I'm onto the next rev of the Combustion and Analysis book, in advance to the one you have. The C.dat viewer will still work so you can view the results.)

Regards
Graham.

[Graham C. Williams]

Many thanks to all those who have responded, on and off Forum; a special thanks to Bill, ‘It's exactly what’s needed’. As a result of Bill's motors I'm redesigning the 'Mass of fuel' and Qf routines. The amount of charge compression Bill’s motors produce has made a big difference. It'll take a few days to test the new routines. I'll keep you informed.

Thanks again.
Graham

[hinote]

The amount of charge compression Bill’s motors produce has made a big difference.

Hell--that ain't nuthin'!!

Wait until you see the monster I've just started cutting metal for.

The sim numbers make our previous and current efforts look like child's play. I've made some interesting discoveries that can (hopefully) result in new performance standards for VPJ's.

Graham-can I send you the file for my currently successful M15E "simple-build" motor? Mike has had a look at it in the NUDis sim, and thinks it's a good project there, too.

[Graham C. Williams]

The sim numbers make our previous and current efforts look like child's play. I've made some interesting discoveries that can (hopefully) result in new performance standards for VPJ's.

Graham-can I send you the file for my currently successful M15E "simple-build" motor? Mike has had a look at it in the NUDis sim, and thinks it's a good project there, too.

Dear Bill.
Yeah yeah yeah. We'll see what's what when we get the Phoenix running at full chat - 100psi peak combustion pressure may be possible.
Back to reality - Please send the file for your M15E. It'll be a few days before I can look at it for you.

Regards
Graham

[Graham C. Williams]

I’ve rewritten the Mass of fuel routine and updated the program. Preliminary tests are running. These preliminary tests are run with the fuel set to deliver full power. Later tests will be at about half power. The fuel metering is very crude at the moment, I’m just changing the combustion efficiency to achieve this. A better routine is in the building.
So far our (Albion jets) Type 07 FWE and Type06 motors have started with no difficulty; these motors are very well behaved and the fact that they start only shows that the program has no real faults. That covers the simple rear-loaded motors.
Next I ran the Kentfield Lockwood, a front-loaded motor. For this type of motor, with the long tail cone it is remarkably well behaved. It started with no problem so I stopped the test after 5 cycles. The analysis shows that my centre of combustion routines are still not correct. This routine makes no allowance for the appreciable delay between the end of induction and the start of second phase combustion in these types of motors and so the routine goes off and finds the most likely centre of combustion – half way down the tailpipe. No way should this be allowed because of the fuel migration rate. The reason that the motor still started is due to the energy release ramp shape I’m using. This ramp assumes that until the charge has migrated to the centre of combustion energy release is very low. The ramp performs other functions like allowing time for masstransfer and mixing. By the time First Phase combustion is over the centre of combustion routine had moved the limits back to the correct area so all was well. Anyway the Kentfield Lockwood ran within acceptable limits.
Next Up. Mike E’s 50mm Type E motor. You’ll find some video of it running on this forum. Again no real problems save that associated with the ignition time delay set by the exhaust design. Again the CofC routine found the correct location in time. My real concern is that some energy (a very small amount) is being released into the pipe at a location that is not possible at that given time into combustion. It doesn’t have much effect and did not stop the motor running.

I know I’ve not completed the tests but I couldn’t help giving Bill’s motors a try.
Motor A has a 123mm dia. Combustion Chamber.
Motor B has a 145mm dia. Combustion Chamber.
So at full power startup:

Motor A stopped after 2 cycles. It was a large shock in the middle of the pipe that did it. This doesn’t mean much because the code tests are not finished. It may run at some lower fuel setting so I’ll test it again later. Bill, I’ll rerun it tonight and send you the results.

Motor B. Well, I’ve stopped it after 1600 iterations and it was still running. That’s 8 combustion cycles at this timestep. Much joy, It’s the first time I’ve had this type of motor running. The previous Rev(05) of the program did not want to look at the IAME motor. Bill’s B motor started under the same combustion set-up conditions I’ve used for all the tests so far. Now, this next bit doesn’t mean much because the system has not been tested against data gathered in the field and the CofC issue above will change the results but Peak combustion pressure is about 2.2 – 2.3 bar.

The previous Rev (05) code for measuring the mass of fuel involved directly measuring the amount of fuel that is in the combustion chamber. The start and end points for the measurement being governed by a set of logic switches set throughout the program by significant events. Start of induction, End of Induction, Start of combustion etc. I chose this method because I wanted to express the fact that fuel is inducted and some fuel could also be blown out without being burnt. I see the whole process as a flow in and out. Bill’s motors are generating an increase in charge density at about the time of combustion and it was this that made my method inaccurate.
The method I’m using now looks directly at the massflow of gas through a cell in the induction pipe at the fuel injection point. This ‘Flux’ method is less prone to the problems above but not so good at the fuel loss side of things. I’m sure I can make some adjustments and improve this as well.

Graham.

[Mike Everman]

That's just phenominal, Graham! Sounds like you're homing in on it, and have been very busy. I wish I could play too, but I'm just buried by a deadline for a few days more. ;-{
I'll be in touch...

[larry cottrill]

Graham -

Bravo! That sounds like wonderful progress - monumental, really. Amazing!

L Cottrill

[hinote]

Motor A stopped after 2 cycles.

Motor B. Well, I’ve stopped it after 1600 iterations and it was still running.

Graham:

You've certainly nailed the results with my 2 engines. Your evaluation exactly parallels the real-life experience I've had with them.

Motor A was the first version of the "mini-Escopeta" from just about a year ago. Motor B was the M6Ec--a modified "kentfield" style motor (4 intake tubes).

I'm excited with the progress you've made so far.

[Bruno Ogorelec]

Wow! And I thought 2005 was a very exciting year in the pulsejet field...

Great work Graham! Thanks for sharing this with us.

[Graham C. Williams]

Following up –

I found a UFLOW file for one of Mike Evermans Kazoo motors. ‘The short Knstein’(please correct the name) motor 49mm Dia CC and 1.036m long. Is this the motor you wanted me to test? Converted the UFLOW file, using settings that have worked with most of the other motors I went for a full power start-up. No problems, the motor started and run with no difficulties. Mike, do you want to see the files?
The success so far makes me wonder if I can change some of the variables into constants and remove them from the set-up sheet. While musing this prospect (on my favourite musing throne) Bill sent me a motor that changed things. Bill doesn’t want me to reveal any details of this motor, I’ll respect that but the underlying story is here.
When starting to develop this set of programs it soon became evident that detection routines and the Logic switches were going to be the backbone of the program. Testing for events, setting the switches that activate or not other sections of the program, that sort of thing. Speed of execution was another factor – 2000 iterations in a VB/Excel development environment takes time; some of the niceties of program writing and white space and extensive Rems were going to have to go by the wayside. I needed some sort of DATA dump to keep an eye on what’s going on.
A chance meeting with an old friend, Sunny. Sunny is now retired but still as sharp as they come and part of the team that developed the MRI scanner. He helped me set-up the data dump and made program suggestions concerning speed of execution and data gathering. Sonny doesn’t do the programming thing any more but as a mentor is ideal. Now, at each iteration and within iterations all the Boolean switches, the detection arrays and other variables and constants are dumped to a Worksheet. Some are also put on the screen between iterations but this slows the program down a lot. Organise the data dump correctly and it’s like watching ‘The Matrix’. More progress was made in a few hours from looking at these dumps than ever before and during the testing with known working motors I made good use of this technique to make the little corrections required.
So, all is looking very good. Bills two sample motors highlight some small problems with the detection routines and the mass of fuel, the details are above. More cogitation and a small change to the fuel calculation routine so that the Flux routine allows positive and a little negative fuel flow and a simplification to the CofC detection routine provide the answer. I still need to improve the combustion migration rates but feel it’s not too far from the truth.
Then Bill says, “Have a look at this one, it ran.â€

[Mike Everman]

Graham,
sounds like the combustion program is light-years ahead of my last peek!

that wasn't the motor i was talking about, but yes I'd love to see the data! it's the straight piper I built at burning grape. I've been trying to find the original dimensions, and may be reduced to analyzing one of the photos, ugh.

The 3 lb configuration is this:
intake: 7.5x1.18
flat head to throat: 8.5
cc dia 3"
back cone 3" long
straight tail length: 29x1.43

I'm trying to figure out how much longer the tail was for 2 lb thrust.

These are low thrust numbers, I know guys, but remember that it's a straight pipe exhaust. I am working on hero style turbine set-ups, so am wanting tubular exhausts with higher exhaust velocities than with expansive tails.

[Graham C. Williams]

Dear Mike.
I've only just finished running an unpacked model of Ed's Straight pipe Chinese. It works well. I can only assume that your motor will run in a similar way.
Funny you should be working on a hero style turbine Oliver has just made one at his R.I.institute lectures, it ran on liquid Nitrogen.

Regards
Graham.
P.S. Ed Knesl, If you'd like to see the data from this motor just ask, give me your email address and I'll send it to you.