Trying to add pressure sensors to LDC testing.

[WobblyHand]

Well then, that's even more interesting, as that is an effect with supersonic flow.  Hmm, even more reason to get this CFD working.  Made a little progress on it, the WorkBench is installed and comes up, but I'm less certain about all the libraries.  Kind of super geeky software installation, which generated a lot of warnings (usually not fatal) but it did generate an error.  In particular, I'm not sure the HiSA (high speed aerodynamic solver) library installed right.  https://hisa.gitlab.io/doc.html. HiSA is supposed to work in the trans sonic as well as the supersonic domains.

But slowly plugging away at it.  The picture below is not intended as anything that works, it's just a test structure to see if anything works at all.

[WhatUPSbox?]

I would consider starting with just a model of a pipe with an exit to ambient (like in the video). Tricky part may be how to properly model the uncorking. The pressure behind the pellet may be high enough to set up a brief choked condition and that's where what appear to be shock waves are generated.

[WobblyHand]

I would consider starting with just a model of a pipe with an exit to ambient (like in the video). Tricky part may be how to properly model the uncorking. The pressure behind the pellet may be high enough to set up a brief choked condition and that's where what appear to be shock waves are generated.

Before that, I simply hope to run some of the canned examples!  At least I will (probably) know that they were set up right. 

Have no idea how to model the uncorking just yet.  Maybe something like a small high pressure volume where one wall vanishes at t=0.  How to do that in the tool is unknown at this time. 

The LDC model is easy, all the set up for the simulation will be what will take a lot of time to learn.  That and the inevitable issues of convergence and it's friend insufficient memory!

[WhatUPSbox?]

There may be canned examples of a shock tube that may close enough to the uncorking. It would not have the velocity of the gas behind the pellet but maybe increasing the static pressure to get it up to stagnation equivalent may be good enough to support visualization.
Sounds like you are on the right track.

[WhatUPSbox?]

I looked on Google Scholar and there are a number of papers on CFD modelling of gun barrels. They are mostly PB applications so I don't want to pull this in that direction with links but I think the modelling approach would be similar.

[WhatUPSbox?]

I took a couple of the measurements I made with the hollow (no cones) segments and used the measurement at the muzzle and the measurement in the first adjacent segment to approximate the upstream/downstream pressure ratio (Pu/Pd) used as the criteria for choked flow. Big caveat, I doubt my geometry and measurement location are exactly what is called for, but it may indicate the trend. The hollow segment configuration is the closest I have to the video I posted. Wiki shows that Pu/Pd = 1.83 is the minimum for choked flow. My chart shows about 1 milli-sec where that is met with a shot at about 1/2 the speed of sound.
Real? Who knows.

[WhatUPSbox?]

A mystery result.
In the result for the hollow (no cones) segments there was a lot of oscillation in the measurement at the cap (post 32). I wanted to check that it was not the pressure sensor responding to mechanical vibration. I printed a version of the cap with the pressure port blocked off with about a 1.1 mm wall (see picture). I reran the test shots and there was a smooth, delayed response in the output. I did do a check with the CH1 sensor removed to make sure there was no electronic cross talk between the channels and that checked out.
So, my guess is porosity in the 3D printed cap (by luck the print seam runs through that wall). I tried blowing through the blocked port and it is blocked. On the plus side there was no significant response to vibration.
Mystery.

[WobblyHand]

Can't help with solving your mystery, at least not right now.  This stuff can be quite puzzling.  All I can say is keep on plugging.

Sort of like installing this CfdOF software.  Sort of muddled through some of it, but things are still quirky.  But did run an example program, of an elbow with a porous section in it.  It allegedly simulated, but the software connection to the viewer is broken, I can see the simulation residuals, but not the results!  But it's allegedly using 4 cores in parallel for solving.  (I can change that number.)  But if I can't get the visualizer running, it will all be for naught.  This is CFD software using OpenFOAM.  One of the more complicated bits of software I have ever run into.  Shoot, I can barely understand the demos.  Honestly, this module is not that well documented.  Hope I can use it for something useful.

One of the demos is called projectile, although it looks a lot more like a rocket to me than a projectile.  I'll be trying that once I get the durn visualizer to work.  The thing is, it does work by itself, but is not being invoked correctly (for a Mac) by the CfdOF module in FreeCAD.  It's invoking a script file that doesn't exist on my machine.

[WhatUPSbox?]

I never got into CFD codes. I do remember several decades ago when a small company down the road from you a bit developed the first Fluent code that is now part of ANSYS.

[WobblyHand]

Despite all my efforts and blunders to the contrary, eventually managed to get CFD to install on my Mac.  It runs all the known good test cases in the test suite.  Now, for the slow process to come up to speed in the simulation process.  But think I will make a separate thread on this.  Think CFD is complementary to Stan's stuff, but don't want to take away from his well regarded efforts in the test arena.

[WhatUPSbox?]

After some delays and outside distractions, I was able to get back to working on the combined pressure and sound instrumentation. To get enough data channels I am using the 4 channel Rigol and the 2 channel handheld DSO. I had hoped to trigger both with a single laser gate but had some noise issues from tying to a common ground. Instead I added a second laser gate at the LDC exit (cap) and kept the two DSO's separate. The handheld collects the two pressure signals and triggers (external trigger line) on the laser at the barrel exit. The Rigol collects three channels of sound data and uses the fourth channel to trigger on the laser at the LDC exit. To be able to overlay the two data sets, I measured the time delay between the two lasers for a typical shot and used that as an offset. The first image shows the configuration. The three microphones are at the bottom of the picture. The Sound Level Meter (SLM) has a convenient analog output line. At the top of the image is the LDC. It has a pressure probe and a laser gate at each end. The image has blue arrows to illustrate the direct sound paths and the numbers represent the speed of sound time delays for the geometries.

The second image shows the results for a single shot. The sound and pressure results were plotted at the same 1ms/div and offset by the pellet flight time between the two lasers. There is a lot going on in the combined plots, so I'll start at the top.
The two vertical orange lines are the two laser trigger events, one at the barrel and the other at the LDC exit. They are offset by the pellet flight time.
The short horizontal lines attached to the orange trigger lines are the speed of sound delays for the geometries for that Mic trace (seen on the previous image). The way to view these is if one made a ping sound at one of the laser locations, it should show up on a Mic trace with a delay represented by those bars.
The horizontal traces are for each of the Mics as labeled. The chart represents 12 millisec of data.
The bottom chart has the two pressure measurements. The chart is offset to line up properly with the laser trigger.
Observations:
First, hopefully these charts don't make your head hurt, it is pretty far down the rabbit hole.
Second, this is data from checking out the setup, not intended for conclusions. The microphones are intended to provide primarily timing information, not amplitude. I'm trying to see if one can discern where the sound is coming from in the system.
That said, it appears that the sound created at the barrel exit reaches the three Mics with the expected delays. There is also additional sound recorded before the sound from the pellet exiting the LDC arrives. Note, this checkout is being done on a hard table top near a wall, etc. At some point echoes start showing up in the trace
The pressure data will be useful once changes start being made. Also need to figure out how to add the quality Mic/Audacity combo to the mix. Also need to package this so I can take the tests outdoors.   

[WobblyHand]

All very interesting, Stan.  Good stuff.  Keep it up. 

I can see moving the setup outdoors would be a little awkward at the moment.  If you can, try to tie some stuff down to a board or plate.  Even hot gluing things down might help.  Or a weighted Panavise, or something like that to hold things still.

[WhatUPSbox?]

Yeah, I'm working on my strategy for the portable test.
In the mean time, as a comparison, I ran a test in the current configuration using the all hollow LDC segments (results below). With the hollow LDC, the pressure rise in the cap occurs much earlier. Pretty much at the speed of sound after the barrel uncorking. On the microphones, the initial noise spikes occur as before, consistent with geometry and a source at the uncorking. The second batch of noise occurs earlier, probably tied to the pressure reaching the cap earlier. It will be interesting if trends show up as changes are made.

[WhatUPSbox?]

I got a chance to try an outside measurement. Still very breadboard-ish but somewhat portable. The first image show the overall range. The 1740 w/LDC and its local instrumentation are on a work stand platform in the foreground, the SLM mic is on a tripod in front of it, and down range is my "quiet" target trap.
The second image shows the location of the instrumentation on the LDC, the two vertical red lines represent the laser traps. Though it does not look it in the photo, mics 1 and 2 are underneath the LDC at the cap and barrel locations respectively. The Aud mic is at the same axial location as mic 1 and same distance from centerline. All of the mics are pointing up (Aud at an angle) to try to minimize any reflections. The two pressure sensors are located at each end of the LDC as before. The LDC is 3 segments with one cone each. The Rigol DSO and the laptop running Audacity are off to the side.
The third image is the geometry and speed of sound timing delay diagram. The timing delays are used in the results charts in image 4. I took the Rigol microphone data which is triggered by the laser at the LDC cap and added two vertical (orange) lines. One is at the Rigol trigger and the second is the offset back to when the laser at the barrel tripped. I got this offset (avg 0.889 msec) from separate tests with both lasers on the Rigol. I then aligned the pressure data (which is triggered at the barrel) with the mic data using the left trigger line. I then took the Audacity plot, adjusted it to the same timescale, and eyeball aligned it with the Mic 1 data since they are at the same distance from the LDC. So now all the plots are roughly on the same timescale.
I then added the timing delays from the previous chart and they are represented by the horizontal lines, scaled to the values. So as an example, looking at the SLM mic trace (purple), the line labeled 1.56, points to when sound from the barrel uncorking would be reaching (unobstructed) that mic. The line labeled 1.15, points to when the sound from the pellet exiting the LDC would reach the mic. So in that trace, there is minimal sound before the barrel uncorks (as expected) and the sound between the 1.56 and the 1.15 point is presumably coming through/from the LDC as opposed to the LDC exit. The sound after the 1.15 point is a mixture of both.
Fortunately, the timing in each of the traces sort of makes sense. For example looking at the mic 2 trace (light blue) most of the high peaks are from within the LDC. This makes sense since this mic is at the barrel end.
The last image shows the overall recording (75 msec highlighted) of the shot from the Aud mic. It show the main event, a hammer bounce, and either a second hammer bounce (likely) or the sound from the target trap (should be arriving around 45 msec). All of that would be captured in the SLM numerical value (92.1 db) using fast setting (125 msec).
So this was an incremental step. I think the outdoor setting provides cleaner data. I'm hoping that with the Audacity traces time tagged, one could then start processing that for what is before LDC exit and what is after. I'll see if it is possible to untangle the spaghetti. 

[OldCrow]

A mystery result.
In the result for the hollow (no cones) segments there was a lot of oscillation in the measurement at the cap (post 32). I wanted to check that it was not the pressure sensor responding to mechanical vibration. I printed a version of the cap with the pressure port blocked off with about a 1.1 mm wall (see picture). I reran the test shots and there was a smooth, delayed response in the output. I did do a check with the CH1 sensor removed to make sure there was no electronic cross talk between the channels and that checked out.
So, my guess is porosity in the 3D printed cap (by luck the print seam runs through that wall). I tried blowing through the blocked port and it is blocked. On the plus side there was no significant response to vibration.
Mystery.

Whistling?

[OldCrow]

A while back I had pulled some frames from that video and overlayed a rough scale in pellet diameters. That bottom image would roughly fit inside the segments I'm using in my setup. Good seed for flow theorizing. 

Those diamonds look to be about 4/5 of one caliber.  For a .177 that works out to about 95.8 kHz (as an audio wave front).  It looks like the standing waves are disrupted by dispersion and turbulence at about 10 wave lengths from the muzzle.  That might suggest dimensions for baffling material.

I would lay odds they were caliber specific.

Just speculating:
At STP MACH 1 is 1130 fps (344,424mm/sec)
Works out to approximately 95.8 kHz for .177,  77.0 kHz for .22 and  67.8 kHz for .25.

Wave guide calculations:=>   uspas.fnal.gov/materials/10MIT/Lecture5.pdf

Good information!

[WhatUPSbox?]

Playing the video at .25 speed in YouTube, it looks to me like you can see the stepping of the 15 khz frames. If that is true, then maybe 2 standing waves/frame or 30 khz. Even that may be hard to detect.

[OldCrow]

Playing the video at .25 speed in YouTube, it looks to me like you can see the stepping of the 15 khz frames. If that is true, then maybe 2 standing waves/frame or 30 khz. Even that may be hard to detect.

Umm, If the waves are "standing" then you can establish the frequency from the wavelength without worrying yourself about frame rates, or am I wrong about that?  The diamonds are 1 lambda, aren't they?

If all of that is correct then I stand by my estimate, otherwise correct my thinking, please.

[WhatUPSbox?]

I probably should not have used the term standing waves. I don't know what those features are and what the pellet speed is. My only observation was that two of them appear to emerge during one 15khz frame. This is not really part of what I hope to measure other than knowing that there are potential acoustic events up at those frequencies.

[OldCrow]

I see what caused the confusion now.  Yes sir, Those are "shock diamonds" or "thrust diamonds" and are called by other names as well.  They are standing waves.
I am not allowed (dont have enough posts) to post external links so:  en.wikipedia.org/wiki/Shock_diamond <= cut and past into your browser.  Someone linked some good links earlier and the wikipedia article provides additional references which are helpful as well.  In those references is additional information regarding the mechanism whereby they are formed.

Anyway, thanks for providing the video link and the static captures.  Very interesting path to run down.  I had not realized that these devices were principally active at such high frequencies.  That's going to drive some of my simulations if I ever get back to working on designs.