Hello Angelo,

Angelo Farina wrote:

Ok reduced 'Time resolution' in Ramsete Tracer to 0.1 msec, and I see a 2.6 msec (26 lines) delay between the two peaks in bcove-02.__a (using the model I gave).

But I don't understand this, as 2.6 msec is around 0.9 m, while microphone 1 is at least 7 m for the reflecting wall... so I expect a second peak at 2 * 7 m :: 40 msec ...
The first peak I see is indeed at ~38 msec :: 13m (so my calculation method work there...)

Where do I go wrong with this 2.6 msec?
Thanks for helping me.

All the best,

Victor

I made the calculation with a time resolution of 0.0001 s too. The results make perfectly sense for me.
You entered a geometry with a lot of reflecting surfaces (actually 6, not just one). The first reflection you find is coming, always, from the floor. Then you will get the reflection from the wall, much more delayed. And finally, the double-reflection (floor+wall).
I show you here the pictures of the impulse responses and of the ray paths, for receiver n.1:


Please notice the thin red lin connecting the source with reciver n.1. This is the direct sound, not touching any surface.


This is the sound being reflected by the floor. It arrives with a delay of approximately 1ms from the direct sound. Please notice how the floor surface has been evidenced by a green border.


This is the second reflection, coming from the vertical wall (there was no need to use 5 surfaces for drawing it, one had been enough). Now the delay is much bigger (41 ms).


And finally here you see the fourth, and last, reflection, making a double-bounce over the wall and the floor.

A manual inspection of the .__A file reveals, as expected, that just 4 lines are non-zero, at the same time delays as the pictures above show.
I posted all the data files for this test case on:
http://www.ramsete.com/Public/Ramsete/Samples/Vrejis/

Bye

Understood. That is at around 38 msec

Understood, this is the 2.6 msec second peak I see in my graph...
And this is the last one in my .__a files.

I now scanned the dcove-02.__a file twice (the one with 0.0001 sec) and it ONLY has two lines of non zero. The one I download (uing 0.01 sec) has also two (at 38 msec and (38 + 2.6) msec ). You can also check then in the file I gave.
I also don't see the peaks around 78 msec using rams2wav.exe on my 0.0001 sec file

So why do I see in my .__a files only two lines (as I expect indeed four).

This experiment was done in real live and I see indeed the 38 and 78 msec peaks (also in a simulation using CARA), so why not in Ramsete...

Sorry of I made somewhere anobvious error, but at this moment I don;t see it.



All the best,


Victor

Your wall is small, and you are firing just 16 pyramids (demo mode), so it can well happen that none of them hits the wall...
Try with a larger wall (as you cannot fire more pyramids)...

This is a very clear example of the type of errors caused by the limited accuracy available in demo mode...

Did (to the size of the real megalithic stones (4*4 m), but still no second batch of reflections. The problem is that in my work (archaeoacoustics I can't manipulate the stone size;-)

Thanks, I am glad that is the reason then.

All the best,


Victor

A 4x4 wall is indeed quite small.
Please consider that, for ray tracing programs, the "geometrical acoustics" assumptions are considered valid. This means that any surface must be much larger than the wavelenght. This is OK for a theater, where the walls are, say, 10x8m, or even larger. A 4mx4m surface starts to be a bit on the small side, even when you launch thousands of pyramids, as the "geometrical acoustics" assumptions start to be unsatisfied for such a small wall. Try with a 40mx40m wall, and you will see the reflection!

I understand, but in megalithic stone circles, etc.; even 4*4 m is very big (this one is near Avebury). Normally megalithic standing stones are say 1 or 2 m wide and perhaps 1 to 4 m high.
This just gives me a usage boundary which I can't change.

All the best,

Victor

Given the type of geometry you are dealing with, I suppose that ANY simulator based on geometrical acoustics is going to cause errors. You should employ a software which takes really into account the wave equation, considering edge scattering, diffraction, etc... Programs based on ray tracing only take into account the energy of the sound wave, not its internal "wavy" nature...

I suggest a Boundary Element software, such as V-noise or Sysnoise. Or those new time-domain finite-difference programs, being pioneered in Japan (prof. Morimoto, Kobe) and in Spain (prof. Lopez, Valencia)....

Bye

Thanks for this advice. I was using CARA and I can simulate some of the effects we measured in real live.
But CARA is in principle for rooms and even simple rooms (otherwise the calculation times become too big).
Knowing the ins and outs of CARA, I was looking at this and was adviced to look at CATT and Ramsete.
Indeed the effect of edge scattering and diffraction are essentaia (no flat surfaces there;-).
I will follow up your leads.

All the best,


Victor

Ramsete, Catt, Odeon, Ulyxes, Ease, Raynoise, etc., are all programs based on diverging beams (cones, pyramids, or the like). They take into account edge scattering and diffraction by means of tricks (mostly based on the Maekawa or Kurze-Anderson formulation).
This works reasonably well for room acoustics simulations, where after an few reflections the sound merges in a diffuse reverberant field.
For outdoor simulations, in presence of a few, discrete reflections, all these algorithms suffer significantly of reduced accuracy and lack of proper modeling of large-distance curved-ray propagation effects, and of grazing-incidence effects over absorbing surfaces such as grass.

In these cases, if you do not want to go for the heavy task of solving the wave equations, you could consider ray racing programs specially designed for outdoor calculations, such as Lima, Predictor, Soundplan or Mythra. Albeit these programs are still of the "ray tracing" type, they employ proper tricks for dealing with the typical problems of outdoor calculations: a small number of discrete reflections, a lot of scattering/diffraction, the grazing effect over grass and soft soil, curved rays due to temperature gradients and wind, etc...

Bye!

Thanks for this advice! Have work to do to get experience with these!

All the best,


Victor