static subs still emits noise?

[Dr.Sid]

I did some simulations .. just in 2D, and just with masking object of shape of circle (ie approximation of sphere in 3D) .. and it seems that even waves twice the size of the masking object leave some shadow. At wavelengths at same size the shadow is rather well defined and protrudes many wavelengths behind the masking object.



Sound comes from left, as planar waves. I've tried point source too, but it looks about the same. Te wavelength is 8 pixels, so is diameter of the masking object. The brightness is average amplitude over time. Ie you don't see individual waves, without any interference the picture would look like flat are with brightness slowly dropping with distance.
In these kind of simulation, sound reflects of any part which does not 'move' .. ie. also from borders of the simulation. To prevent it, border 50 pixels are 'attenuation zones' .. amplitude there is artificially attenuated so the sound is absorbed, rather then reflected. Even so some of the sound can be seen being reflected from top, bottom, or even right border. The circles in front of the blocking object are interferences of incoming and reflected waves, and they nicely show wavelength of the sound.

If we take Kilo as 10m diameter, and if we look for frequency with such wavelength, it would be about 150Hz. IMHO we can take the picture as rather good simulation of sound at 150Hz bending around Kilo sub.
150Hz is rather present in usual background noise. Higher frequencies would be masked even better, but background noise drops quickly with frequency.

I know sub guys usually go ultra silent when someone talks about 'how does Kilo look on sonar' .. especially when someone mentions 'hole in the water' .. but my bet is at 130m and silent, Kilo would be spotted as shadow on BB. Or anything at that size.

Sure my simulation is pretty crude, and I don't have anything 'official' to compare it with .. it might be totally off .. but it's my 2 cents ..

[magicstix]

Originally Posted by Dr.Sid

I did some simulations .. just in 2D, and just with masking object of shape of circle (ie approximation of sphere in 3D) .. and it seems that even waves twice the size of the masking object leave some shadow. At wavelengths at same size the shadow is rather well defined and protrudes many wavelengths behind the masking object.



Sound comes from left, as planar waves. I've tried point source too, but it looks about the same. Te wavelength is 8 pixels, so is diameter of the masking object. The brightness is average amplitude over time. Ie you don't see individual waves, without any interference the picture would look like flat are with brightness slowly dropping with distance.
In these kind of simulation, sound reflects of any part which does not 'move' .. ie. also from borders of the simulation. To prevent it, border 50 pixels are 'attenuation zones' .. amplitude there is artificially attenuated so the sound is absorbed, rather then reflected. Even so some of the sound can be seen being reflected from top, bottom, or even right border. The circles in front of the blocking object are interferences of incoming and reflected waves, and they nicely show wavelength of the sound.

If we take Kilo as 10m diameter, and if we look for frequency with such wavelength, it would be about 150Hz. IMHO we can take the picture as rather good simulation of sound at 150Hz bending around Kilo sub.
150Hz is rather present in usual background noise. Higher frequencies would be masked even better, but background noise drops quickly with frequency.

I know sub guys usually go ultra silent when someone talks about 'how does Kilo look on sonar' .. especially when someone mentions 'hole in the water' .. but my bet is at 130m and silent, Kilo would be spotted as shadow on BB. Or anything at that size.

Sure my simulation is pretty crude, and I don't have anything 'official' to compare it with .. it might be totally off .. but it's my 2 cents ..

A few minor problems:
- Close in sound does not behave as a planar wave at all, wavefront curvature effects are extremely important.
- The amount of shadowing and diffraction is very frequency dependent.
- Broadband sonars don't operate on a single frequency, but instead the average amount of energy in a band.
- Broadband sonars have complex autogain and normalization algorithms that affect how the data will appear to the operator.
- You're ignoring beamforming and array effects.
- You're mostly ignoring multipath effects around the occluder.
- Your sim treats the kilo as a perfect occluder with no transparency.
- Your sim assumes noise sources occluded by the kilo behave with strong directional correlation. This is not the case in ambient sea noise.
- Ray based models don't simulate low frequencies properly at all, especially at your 150 Hz test case. Sound doesn't behave like a ray at low frequencies, it behaves more akin to electrons in a waveguide, with "fuzziness" and a distinct lack of directionality.
- The "kilos are a black hole in the ocean" thing is a myth, started by a nefarious source linked in a wikipedia article.

That said, there has been research into using ambient noise as a sort of "acoustic daylight" to "see" quiet objects through shadowing and correlation coming from reflections off the object, but those are all very special arrays with complex processing that isn't employed in your run-of-the-mill passive broadband sonar, and they don't work very well at all yet (and tend to only work at extremely close ranges of tens of feet).

[Dr.Sid]

- Close in sound does not behave as a planar wave at all, wavefront curvature effects are extremely important.

I'm talking about masking distant noise .. thats why I used planar waves. What else do you sugest ?

- The amount of shadowing and diffraction is very frequency dependent.
- Broadband sonars don't operate on a single frequency, but instead the average amount of energy in a band.

That's what I'm talking about. This is simulation for frequency with wavelength of size of the object. Lower frequencies will be masked less, higher will be masked more.

- Broadband sonars have complex autogain and normalization algorithms that affect how the data will appear to the operator.

Is the autogain different for different bearings ? I seriously doubt that, except for things like calibration.

- You're ignoring beamforming and array effects.

Imho irrelevant. Goal of beamforming is to present outer sound as true as possible. If it can pick gains in background noises, it should pick drops in it too.

- You're mostly ignoring multipath effects around the occluder.

I do ignore bottom and surface reflection. I can run the simulation with them, but again, it does not look much different. Other multipaths should not matter at 130m.

- Your sim treats the kilo as a perfect occluder with no transparency.

Now that is interesting problem. The hull of the sub could transfer some of the sound, and it is out of reach of my simulation. Anyway at least all reflected sound wont be passed through. And we know that subs without coating reflect sound rather well. So it cannot pass it well at the same time.
Kilo has rubber coating though (I believe). It would work by absorbing, not passing through (I believe). I guess it's mostly aimed against high frequency active sonar, but it could also work at 150Hz range, also for masking sounds coming from inside the sub.

- Your sim assumes noise sources occluded by the kilo behave with strong directional correlation. This is not the case in ambient sea noise.

At 150Hz the background noise should be rather directional in horizontal plane. I could not find much about coherence, but since it can be said that it is directional, it also must be rather coherent.

- Ray based models don't simulate low frequencies properly at all, especially at your 150 Hz test case. Sound doesn't behave like a ray at low frequencies, it behaves more akin to electrons in a waveguide, with "fuzziness" and a distinct lack of directionality.

This is not ray model. It's completely wave based simulation. It should simulate diffraction and interference. You can see the shadow has rather fuzzy edges, and it is filled with energy over distance. That is the diffraction in work. It's just not that strong, so it could erase any disturbance left by the object.

- The "kilos are a black hole in the ocean" thing is a myth, started by a nefarious source linked in a wikipedia article.

Have the link ?

Btw. you really seem to know a lot .. what's your relation to underwater sound ?

[Dr.Sid]

Hmm .. just had an idea. Instead of constant wave source I used random source. Or better 'sources'. Every pixel on the 50 column is 'sourced' with random numbers. Ie all wavelength from 1px and lower will be present in the noise. Result:



Very poor shadow, not extruding over 2 diameters. Even if there are higher frequencies present, then there was in the first case. The coherence indeed did it, I guess. There are also a lot lower frequencies present now. Unfortunately I can't easily create random source with predefined spectrum.
Ok .. I lower my statement into: at 130m Kilo might, but might not, create a shadow on BB.

[Rip]

Originally Posted by Dr.Sid

Hmm .. just had an idea. Instead of constant wave source I used random source. Or better 'sources'. Every pixel on the 50 column is 'sourced' with random numbers. Ie all wavelength from 1px and lower will be present in the noise. Result:



Very poor shadow, not extruding over 2 diameters. Even if there are higher frequencies present, then there was in the first case. The coherence indeed did it, I guess. There are also a lot lower frequencies present now. Unfortunately I can't easily create random source with predefined spectrum.
Ok .. I lower my statement into: at 130m Kilo might, but might not, create a shadow on BB.

Having seen the tender mask broadband noise as we neared it on a BQQ5 I can say it can certainly happen. Given the small size of a kilo I would guess that is pretty close to the range it would become obvious on the display. A Typhoon would block out probably 3x more distant. Of course you should have detected one of those well before that distance. I doubt even an aircraft carrier would create a noticeable shadow beyond 1000 yards.

Of course you have to add to that the need to have been at the appropriate depth and/or D/E to get a shadow at all.

[MRV]

Ok, this brings a question to my mind: on a nuke sub, a generator is running to provide on-board power to my systems, that generator has mostly a frequenzy of 50 hz, as it generates the same power as provided at home, i believe. (it would make sense though) this is modeled in-game as most ships and subs have the first line on or next to 50 hz on NB.

Now the kilo is a diesel-electric-sub, and my common sense, correct me if im wrong, tells me that all the systems are not powered by a generator but from the batteries. Only problem I see are transformators running to provide a different voltage. it should still be a lot quieter than the nuke-sub generator?

[Ghost Dog]

I'd say so. In a nuke sub, you also need to continuously run the coolant pumps as well...or shut down your reactor.

a diesel-electric sub on batteries should be VERY quiet.

[magicstix]

Originally Posted by Dr.Sid

I'm talking about masking distant noise .. thats why I used planar waves. What else do you sugest ?

The plane wave approximation breaks down the moment it hits the near-field occluder, as the occluder forms a new wavefront (think of the Huygens principle). To represent a truly physical situation, you would have neither a point source nor a plane wave, but an infinite number of point sources along the surfaces generating the noise, all with mostly random phase. It's more akin to ambient illumination or fog in graphics terms than a directional source. That said, your theoretical sub most likely wouldn't shadow ambient noise, but *could* shadow a coherent noise source behind it. However, again, multipath effects would probably negate the shadow zone created by the sub.

Originally Posted by Dr.Sid

Imho irrelevant. Goal of beamforming is to present outer sound as true as possible. If it can pick gains in background noises, it should pick drops in it too.

Very much not the case. Beamforming's prime goal is to reject uncorrelated noise and form correlation of what's left based on direction (as well as effect an increase in SNR, but that's somewhat irrelevant to the discussion at hand). Ambient noise is only weakly correlated in most cases. What you'd wind up with is not just a 'hole,' but a 'hole in a sea of holes.' It'd be like trying to find darkness in a room without any light.

Originally Posted by Dr.Sid

I do ignore bottom and surface reflection. I can run the simulation with them, but again, it does not look much different. Other multipaths should not matter at 130m.

Multipath matters at all ranges, especially in a shallow water environment. You would get a lot of extremely high angle paths that tend to break down most models in this close-range case. The roughness of the surface and bottom would also immediately make a mess of the pretty lines you have in your plot. ;P

Originally Posted by Dr.Sid

Now that is interesting problem. The hull of the sub could transfer some of the sound, and it is out of reach of my simulation. Anyway at least all reflected sound wont be passed through. And we know that subs without coating reflect sound rather well. So it cannot pass it well at the same time.
Kilo has rubber coating though (I believe). It would work by absorbing, not passing through (I believe). I guess it's mostly aimed against high frequency active sonar, but it could also work at 150Hz range, also for masking sounds coming from inside the sub.

Anechoic coatings are generally highly tuned to specific threat frequencies. Don't expect the coating to absorb everything across the band of interest. The reflectance of an object to sound depends on the impedance mismatch between the medium and the object, thus keep in mind the surface of the object is not necessarily it's "acoustic size," so you could be overestimating how big your occluder is.

Originally Posted by Dr.Sid

At 150Hz the background noise should be rather directional in horizontal plane. I could not find much about coherence, but since it can be said that it is directional, it also must be rather coherent.

Depends, but ambient noise tends to be very uncorrelated to begin with, unless you have a distant noise source such as a port, sealane, environment, etc.

Originally Posted by Dr.Sid

This is not ray model. It's completely wave based simulation. It should simulate diffraction and interference. You can see the shadow has rather fuzzy edges, and it is filled with energy over distance. That is the diffraction in work. It's just not that strong, so it could erase any disturbance left by the object.

What's the underlying approximation? Normal modes? Parabolic equation? Or is it finite element? Your plot seems to indicate an isovelocity environment, is this the case?

Originally Posted by Dr.Sid

Have the link ?

It appears the original link in the Wikipedia article has thankfully been removed and the article itself no longer mentions the kilo being a 'black hole.' However, I've heard that "black hole" quip applied to any number of quiet subs from Gotland to Los Angeles.

Originally Posted by Dr.Sid

Btw. you really seem to know a lot .. what's your relation to underwater sound ?

I develop underwater acoustic simulations using supercomputers for a certain employer.

[Dr.Sid]

What's the underlying approximation? Normal modes? Parabolic equation? Or is it finite element? Your plot seems to indicate an isovelocity environment, is this the case?

I's say finite element. It's just something I made up. Generally you hold speed and pressure for every pixel (ie element of space) and in every step you change those based on values in neighbors. This case is isovelocity, but it can do gradient (check left side sound channel caustics in my sig, that the same tool). It's not very realistic though with speed gradients .. this method is SLOW and it does not support different scale for x and y axis. So to create sound channel like I have in my sig, I had to change the velocity in range of 50% to 200% .. which then renders the interference totally different. It was just test if caustics and focusing would show up in such basic simulation (and it did).

I develop underwater acoustic simulations using supercomputers for a certain employer.

Sweet ..