UFlow progress

[hinote]

Hello all:

I've made more progress, in my effort to understand pulsejets--and now (in particular) I'm concentrating on the Escopeta type.

lnteresting note here: About this time last year I stated that I didn't consider the Escopeta design to be worth pursuing--what a difference a year makes!!

For the disbelievers, I'm now able to make design decisions in UFlow to an amazing degree of accuracy. The timing created by the tailpipe configuration in particular, can be refined to a value of 2 millionths of a second!. Of course, that disregards the accuracy of the other variables involved (such as the operating temps), so it's a useless refinement.

Regarding the above, configuration accuracy can still be reduced to very low values--less than 0.5 mm in most cases.

I'm in the process of creating yet another engine, based on my latest findings. I have a high degree of optimism that it can demonstrate intended thrust and TSFC values.

I would encourage those with great interest in valveless pulsejets to jump in, with both feet. This field is wide open, and there's lots of development potential for the amateur engineer.

Bill H.
Acoustic Propulsion Concepts

".......some day soon we'll be flying airplanes powered by pulsejets."

[milisavljevic]

Faaaaassscinating...

With great interest,
M.
__________________
"Hey! That's Mikey's line!"

[Eric]

oooooooooohhhh..... aaaaaaaahhhh....

When do we get to see some brightly glowing pictures? :)

[hinote]

oooooooooohhhh..... aaaaaaaahhhh....

Eric has enough experience to know that my statement about millionths of a second isn't really such a big deal.

It's just UFlow, failing to round-off its math.

Stated again--the other variables involved are still somewhat in question--so the accuracy available is reduced somewhat. Still, it's fun to run the numbers and see how close you can get to the design target.

Also stated again--UFlow is capable of a lot more than many are giving it credit for.

When do we get to see some brightly glowing pictures?

Several more weeks. I'm collecting suitable high-quality materials, and will take my time assembling it.

Bill H.
Acoustic Propulsion Concepts

".......some day soon we'll be flying airplanes powered by pulsejets."

[Mike Everman]

Faaaaassscinating...

With great interest,
M.
__________________
"Hey! That's Mikey's line!"

LOL!

Iiiiiiiiinteresting!

[larry cottrill]

Also stated again--UFlow is capable of a lot more than many are giving it credit for.

I'm currently working on a Java applet animation of the UFLOW1D 'Density' curve for a run of a standard size FWE (with intake out in front). The density variations are in shades of gray. So far I have only tabulated 1300 or so points into data files from the UFLOW worksheets; I will need at least twice that for it to look good. The only part that's really smooth so far is the intake, which I did in 25mm slices - you ought to see how it shows the cold air pouring into the front end! I'll start up a thread for it if I ever get done with it. Transcribing the worksheet data for each slice of the pipe is a real drag.

Stay tuned ...

L Cottrill

[Eric]

If I had some free time I would write a script to export the data into a 3d mesh in autocad. Doing anything more than a handful of data points becomes impossible. Especially when you want to do something like 100,000,000 polygon model. *sigh*

The problem with U flow is, like Bill says the rounding. Peaks are never really that sharp, or at least as sharp as they are displayed.

If we could get 100 or 1000 sample points at equal lengths (instead of uflows 10 limit) in the engine we could see a very clear depiction of what would happen inside the engine as far as wave reflection and such, at least in a pipe with non combusting gas.

Even the ones I did back a while ago shows the wave propogation and reflection pretty well. If you could do that instantly for each test run you could get a lot more useful information out of it.

Although it really doesnt matter since the combustion poses a whole new set of problems to account for.

Eric

[hinote]

I'm currently working on a Java applet animation of the UFLOW1D 'Density' curve for a run of a standard size FWE (with intake out in front). The density variations are in shades of gray. So far I have only tabulated 1300 or so points into data files from the UFLOW worksheets; I will need at least twice that for it to look good. The only part that's really smooth so far is the intake, which I did in 25mm slices - you ought to see how it shows the cold air pouring into the front end! I'll start up a thread for it if I ever get done with it. Transcribing the worksheet data for each slice of the pipe is a real drag.

Go, Larry!!..............

Actually, I'm a real fan of graphic representations, and of animations. For me, they're a shortcut to visualization during design and iteration.

I'll be looking forward to the fruits of your efforts.

Have you considered capturing the numbers with an Excel (or similar) spreadsheet? Then, although I don't really know for sure--I'm betting somebody out there has put a piece of software together that will do the animation for you.

Just a thought.

Bill H.
Acoustic Propulsion Concepts

".......some day soon we'll be flying airplanes powered by pulsejets."

[sam]

If anyone's really interested I can post a copy of my program which essentially does the same task as UFLOW. The solution method is pretty much identical. However, you can output the temporal variation of any variable you like at every grid point if you wish. You can also choose between open, closed or reservoir boundary conditions.
However, it doesn't have a nice GUI like UFLOW, and ideally requires MATLAB to import the data and create the geometry so its probably got a fairly limited appeal. As with UFLOW there is no combustion model.
Sam

[Mike Everman]

Hmm, I have a copy of MATLAB somewhere, what would be the learning curve for just the geometry prep tasks? Would SCILAB work (that's free)?

I'm very interested, Sam. Can you tell us a bit more? Can you input a cyclic pressure pulse, as in repeating combustion pops?

[Mike Everman]

Sam,
Is the input geometry and conditions a text file that I could possibly prepare with excel? THAT would be ideal...

[sam]

Hi Mike,
The program simulates 1D fluid mechanics in a duct of variable cross-sectional area. It can handle steep pressure and temperature non-uniformities easiliy, including shock waves.
The boundary conditions are better than those in UFLOW, which simulate an open end as a pressure node. In my program the pressure is reduced from atmospheric pressure during inflow, which is more realistic. You can also set atmospheric pressure and temperature to simulate altitude performance if you like.
It accounts for friction and convective and radiative heat transfer to the combustor walls, and the wall temperature is allowed to vary as a result.
Unfortunately you can't input a periodic pressure variation in the combustor. It would actually be very difficult to program something like that, mainly because its a non-physical process.
I'd never heard of Scilab but I had a quick check on the website and it appears to be a direct immitation of MATLAB, so should be fine. You might get away with Excel, but the output files get pretty big if your not careful. Creating geometry is really easy though.
In practice I'm not sure how much more useful it would be to anyone than UFLOW is. Both programs fail to capture periodic combustion. I think that numerical solutions are only really indicative of pulsejet performance if you can properly simulate the cyclic behaviour.

[Mike Everman]

Thanks Sam. The problem for me with UFLOW is the closed nature of it. The input file is not text, but some encrypted thing. If it were a text file, I could automate iterations to great benefit, I think. I've been taking the output files and crunching the numbers for thrust estimate purposes, but it is tedious and ultimately unrealistic in an absolute sense, but would be useful comparitively as part of an iterative run.
Just rambling. What do I want for free??

Dr. Corboran is not interested in giving the source for UFLOW, and I can't afford to have his group change it for my needs, so...

I've found a reasonably priced 2dD code called Answer Express, with a combustion module available. I haven't delved into enough to know if you can have cyclic heat release events. I think there is a trial version, but I want to wait until I can give it my complete attention, and that is not now.

Here's the Answer site:
http://www.acricfd.com/software/answerexpress/
I'd very much appreciate your assessment of it.

[mk]

Very interesting Bill!

Seems like you still donate a huge part of your life to pulse-jets.
However, I haven't heared from you onc in a while. Hope everything is fine though.

As to the general dicussion. U-Flow uses mesh points, but you can only enter ten locations to have a look at. So why not set-up identical models, altering the data point locations to values one desires?
Due the plot quality could be justified, what means Excel (direct connection) brought the same results as U-Flow, you could (a) have detailed looks at each section's concentrated model in U-Flow, or (b) sum up everything in a program like Excel or so for your purposes.

[sam]

I think a 2D simulation is overkill - the essence of pulsejet operation is one-dimensional.
I'll post a version of my code with some basic instructions sometime soon. I'm afraid it will be a compiled version. I don't want to submit the source code - its got a lot of stuff in it thats related to another project which I can't make available to the world in general.
Creating input files for my software should not be a problem with excel. Can excel handle matrices of ~500x256? If so then it should also be able to handle the output files, and you can do any analysis that way. You can see what every cell is doing if you like as a function of time.
Sam