UFlow progress

Moderator: Mike Everman

Mike Everman
Posts: 4939
Joined: Fri Oct 31, 2003 7:25 am
Antipspambot question: 0
Location: santa barbara, CA
Contact:

re: UFlow progress

Post by Mike Everman » Mon Oct 31, 2005 7:01 pm

Great, Sam.
Pre and post processing with excel would be a neat solution for me; a step closer to automated iteration. Does your code animate the results?

RE: 2d: I thought that would be key to separation and depiction of vortices?
Mike
__________________________
Follow my technical science blog at: http://mikeeverman.com/
Get alerts for the above on twitter at: http://twitter.com/mikeeverman

sam
Posts: 80
Joined: Mon Jan 26, 2004 5:40 pm
Antipspambot question: 0
Location: Stoke-on-Trent, England

re: UFlow progress

Post by sam » Mon Oct 31, 2005 8:06 pm

Mike,
The results aren't animated. It wouldn't be too hard to write a script that would do it from MATLAB though. I'm sure you'll think of some way of viewing the results.
You're right of course - 2 dimensions are absolutely necessary to model any vortical motion. However, 1D wave mechanics captures the dominant fluid motion in the inlet and tailpipe very adequately. If you wanna model combustion properly you need a full 3D viscous solution (= very very slow).

sam
Posts: 80
Joined: Mon Jan 26, 2004 5:40 pm
Antipspambot question: 0
Location: Stoke-on-Trent, England

re: UFlow progress

Post by sam » Mon Oct 31, 2005 9:05 pm

Mike and Ben,
Software attached. Just unzip it into any directory. I've also thrown together some instructions. It may take you a while to get used to it but its fairly straight foward once you get used to it. In general it outputs a lot more information than UFLOW does, so file sizes can get quite large if you're not careful.

Let me know how you get on. I'm sure you'll have a million questions!

Sam
Attachments
code.zip
(35.52 KiB) Downloaded 199 times

Mike Everman
Posts: 4939
Joined: Fri Oct 31, 2003 7:25 am
Antipspambot question: 0
Location: santa barbara, CA
Contact:

re: UFlow progress

Post by Mike Everman » Tue Nov 01, 2005 1:47 am

Cool. Still reviewing it.
Here's some of the waveforms plotted. For those following along, each hump is the pressure wave at a different point in time. I think I can make a slider in Excel so you can drag it to animate like in UFLOW. I'll give it a shot.
Attachments
sams output.JPG
sams output.JPG (23.48 KiB) Viewed 4706 times
Mike
__________________________
Follow my technical science blog at: http://mikeeverman.com/
Get alerts for the above on twitter at: http://twitter.com/mikeeverman

Mike Everman
Posts: 4939
Joined: Fri Oct 31, 2003 7:25 am
Antipspambot question: 0
Location: santa barbara, CA
Contact:

re: UFlow progress

Post by Mike Everman » Tue Nov 01, 2005 7:45 am

Sam and Ben,
Attached is a first cut of animating the output in excel. Works great after importing p.dat. I'll tweak it a bit so you can select p,t,v, etc. It'll be a learning experience to be sure, doing the macros to automate the importing and chart scaling, but this works. It's fun to watch. I enjoyed figuring this out while staying home with a flu ridden 7 year old. ;-)
Attachments
nudis post processor demo.xls
(210 KiB) Downloaded 245 times
Mike
__________________________
Follow my technical science blog at: http://mikeeverman.com/
Get alerts for the above on twitter at: http://twitter.com/mikeeverman

sam
Posts: 80
Joined: Mon Jan 26, 2004 5:40 pm
Antipspambot question: 0
Location: Stoke-on-Trent, England

re: UFlow progress

Post by sam » Tue Nov 01, 2005 10:29 am

I'm really impressed with your animation tool Mike. Didn't realise you could do that in Excel. You can increase the temporal resolution in inputs.dat. On line 2 change 50 to 5 and you'll get 10 times as much output, but covering the same time interval.
You'll notice that the wave decreases in strength over time. This is partly because of friction and heat transfer, but mainly because the boundary conditions are not perfectly reflective as they were in UFLOW. This is physically more correct and simulates jetting during outflow and sink flow during inflow.
Have you worked out how to change the geometry yet?
Sam

sam
Posts: 80
Joined: Mon Jan 26, 2004 5:40 pm
Antipspambot question: 0
Location: Stoke-on-Trent, England

re: UFlow progress

Post by sam » Tue Nov 01, 2005 10:41 am

One of the great things about MATLAB though (and presumably SCILAB too) is that you can plot all the data at once.
For example the attached plot shows the pressure. The horizontal axis is cell number and the vertical axis is time. The magnitude of the pressure is interpreted as a colour. Red is positive, Blue is negative. You can see the compression waves move outwards from the centre and reflect from the open ends as blue expansion waves. The lines are staright which indicates that the waves move at constant velocity. This is because the speed of sound is constant (because the temperature is constant)
Sam
Attachments
p.jpg
(65.9 KiB) Downloaded 243 times

Mike Everman
Posts: 4939
Joined: Fri Oct 31, 2003 7:25 am
Antipspambot question: 0
Location: santa barbara, CA
Contact:

re: UFlow progress

Post by Mike Everman » Tue Nov 01, 2005 2:50 pm

There's some methods of plotting all the data, I haven't explored them all. Here's one possible representation.
Attachments
cool plot.JPG
(92.21 KiB) Downloaded 246 times
Mike
__________________________
Follow my technical science blog at: http://mikeeverman.com/
Get alerts for the above on twitter at: http://twitter.com/mikeeverman

Mike Everman
Posts: 4939
Joined: Fri Oct 31, 2003 7:25 am
Antipspambot question: 0
Location: santa barbara, CA
Contact:

Re: re: UFlow progress

Post by Mike Everman » Tue Nov 01, 2005 2:54 pm

sam wrote: Have you worked out how to change the geometry yet?
Sam
Geometry input should be easy, you'll see a graphic representation as you build it.. I started with the semi hard part. I need to learn more about VB to make it really good, but I'll get it to limp along in the wee hours tonight. ;0)
Mike
__________________________
Follow my technical science blog at: http://mikeeverman.com/
Get alerts for the above on twitter at: http://twitter.com/mikeeverman

Mike Everman
Posts: 4939
Joined: Fri Oct 31, 2003 7:25 am
Antipspambot question: 0
Location: santa barbara, CA
Contact:

re: UFlow progress

Post by Mike Everman » Thu Nov 03, 2005 2:15 am

Here is the state of the post processor. Geometry entry sheet is next. I've got the pressure, velocity and temperature plots animating off of the slider at right.
Attachments
nudis post processor demo.xls
(1.39 MiB) Downloaded 256 times
Mike
__________________________
Follow my technical science blog at: http://mikeeverman.com/
Get alerts for the above on twitter at: http://twitter.com/mikeeverman

mk
Posts: 1053
Joined: Mon Jan 19, 2004 8:38 pm
Antipspambot question: 0
Location: FRG

re: UFlow progress

Post by mk » Fri Nov 04, 2005 7:58 pm

Amazing, amazing, guys!

Luckily I recived an email copy earlier while I did not know I would drive home this weekend.

I'll see whether I can effort some time donation to the most useful stuff provided since, well, quite some time here.
mk

larry cottrill
Posts: 4140
Joined: Sun Oct 05, 2003 1:17 am
Antipspambot question: 0
Location: Mingo, Iowa USA
Contact:

Re: re: UFlow progress

Post by larry cottrill » Wed Nov 09, 2005 4:44 pm

hinote wrote:Also stated again--UFlow is capable of a lot more than many are giving it credit for.
I just completed revising my "animated" FWE density applet, dividing the tailpipe up into twice as many slices as before (each slice now less than an inch long). This required eight new data files from the worksheet.

Guess what - now, you can actually see the little red hot zone that's about three inches in from the tailpipe flare! I swear I am not making this up. Amazingly, this narrow zone stays at a higher density than the gas around it for two separate, fairly lengthy segments of the cycle, which I suppose allows a lot of conduction to the pipe in that zone.

Another interesting detail that would have been "predicted" by UFLOW1D - the extreme concentration of heat in the chamber. I believe the evidence for this shows in all the pictures I posted earlier - the low density (i.e. superheated) gas in the chamber remains practically undisturbed throughout the entire cycle. This is verifiable by looking at the UFLOW Mach numbers and mass flows throughout the chamber zone - all very low values at almost every time point. In the FWE, the chamber is nothing but a static pump that applies pressure to the outer parts of the engine. (But, maybe this is just as true for every other pulsejet.)

L Cottrill

hagent
Posts: 413
Joined: Sun Feb 06, 2005 9:01 pm
Antipspambot question: 0
Location: Simi Valley CA

re: UFlow progress

Post by hagent » Wed Nov 09, 2005 6:57 pm

Hi Larry,

Nice to hear about UFLOW progress.

It would be great if you could upload a UFLOW file for us. I'm still having trouble trying to determine the best settings.

Thanks very much,
Hagen Tannberg

serverlan
Posts: 62
Joined: Tue Oct 05, 2004 11:31 am
Antipspambot question: 0
Location: Australia

re: UFlow progress

Post by serverlan » Thu Nov 10, 2005 2:25 am

Excellent stuff Larry,

I think I'll be looking for a copy RSN.

I think you have confirmed my view of an old argument, several years back.
People used to talk a log about hot slugs of gas moving throught the engine, and I would argue that it is really a pressure wave moving through the gas.
The pressure wave will move at the speed of the gas flow, plus the speed of sound in the gas (which unfortunately varies all over the place, making "rule of thumb" design very difficult, and making UFLOW a godsend.

The gas does move a bit each cycle, but only, say, a tenth of the length of the engine? this determines how much fresh intake charge we can get each cycle. If the gas flow is too low, then it doesn't breathe well enough, and will give low thrust.

A good analogy of wave motion through a medium that I remeber from my school days is this:
Put a wrinkle/loop in one end of a rug.
Then push that loop all the way to the other end of the rug.
The loop will have travelled the length of the rug, and the rug will have travelled the "length" of the loop.
The loop is a good analogy for the pressure pulse in the "flying carpet" jet engine.

Have another look at my UFLOW training model I posted a while back.
(I'll bring it to the front again)
This model was concerned solely with getting significant mass flow so as to get a good intake - more power. It keeps the pulse patterns as simple as possible for clarity.
The rest of the engine design is just a straight tube.

Don

Post Reply