k-Wave

1. vilppus
   

   Member
   Joined: Mar '14
   Posts: 2

   Hi

   I have been studying k-Wave and I have a problem. My simulation is pretty straightforward at the moment. I have a transducer and a steel block (width 1 mm). There is 2 mm of water in between of the transducer and steel. The transducer is also the receiver. I detect the initial signal and the reflected one. When I calculate the speed of sound in water from the simulation data (from maximum peak to maximum) I get approximately 1600 m/s. I have set the speed to 1500 m/s. I don't know what's wrong. If I only have water I get speed value of 1507 m/s.

   I have tried many things to fix this. I have tried to adjust the CFL value, no change. I have increased the number of calculation points even up to 1024 x 1024, no change. I have also changed the properties of steel to closer water, no change. Can anyone look at my code and come up with a solution. Link: https://www.dropbox.com/s/o5klnd3ul218hk7/speedofsound.zip

   Posted 10 years ago #

2. ouelletn
   

   Member
   Joined: May '13
   Posts: 18

   Looking at the grid you have defined there is a mistake in the code. You have 160 grid points of water before the steel, which is .6250 of the total grid(1.875mm water). You seem to have mixed up the x and y dimensions at some point in your code. This accounts for the dependency mentioned exactly.

   On a side note, your grid should be redefined. As you have stated that the PML is outside the computational grid, the grid is expanded before computation to add in the PML dimensions, this will result in a calculation grid that is not efficient with the fft(multiples of prime numbers).

   -Andrew

   Posted 10 years ago #

3. vilppus
   

   Member
   Joined: Mar '14
   Posts: 2

   Thanks Andrew. That did the trick. Now I get a speed value of 1506 m/s. This was my first simulation so I wasn't used to the "matrix-notation" of coordinates used here (also in Matlab). When I initially noticed my x and y coordinates where wrong, I remembered to change all the values expect the division by dx. That should have been divided by dy.

   Thanks also for the tip about the size of the computational grid. I expanded my grid to 284 x 284 points, plus 40 for the PML, so it's 324x324 which has a prime factor of 3. Computational time pretty much stayed the same because of the increase in grid and time points. Nevertheless it's good to keep that in mind in the future.

   I'm sure that I'm going to ask more questions but I hope that they won't be as trivial as this.

   Posted 10 years ago #

RSS feed for this topic

Reply

You must log in to post.