Speed of Sound, what should it actually be?


I am running some simulations where waves are impacting on an object. Using shallow water theory I see that the velocity is about 2.75 m/s of the waves - then I set speed of sound as 10*2.75. But since the waves are impacting on an object some particles become accelerated due to the shock front, and increase their velocity to above 2.75 m/s. Does the reduce the quality of my results?

Kind regards


  • It should not be a problem

    However you can compare results of two simulations: i) using c=27.5 and ii) using c=35

    First difference will be runtime for sure, but you can compare the output results.

    Please let us know


  • Hello Alex!

    I ran the simulation with coefsound = 10, which totalled in a speed of sound of 34 m/s and this is what I see in the Run.out file

    Looking at the Run.csv I see:

    So it seems like only particles were excluded due to leaving the rhop range of 700-1300. I don't understand the number 0.016971 though? I have 345256 particles in total.

    I am still a bit confused though, because theoretically a particle could get a velocity higher than 3.4 m/s due to impact on an object - would this particle then break compressibility locally?

    Or is it "fine" if a few particles go "rogue" but the main flow is less than 3.4 m/s?

    It seems like no particles left the domain due to velocity, so it seems like it does not strictly look at each particle, but at an "average" of particless?

    Kind regards

  • If 23 particles of a total of 345256 are leaving the domain.... that is not a big problem, right?

    What is more problematic is that time step was adjusted to DtMin 99 times....

    Questions for you:

    1) Why do you have a speed of sound of 34 m/s? is the speed of system 3.4?

    2) How are you opening Run.csv in excell? Please check the configuration and the use "." and "," since that value "0.016971" is not possible... otherwise we have an error there (please can you send the Run.csv?)


  • No, it is not a big problem at all, I just wanted to be sure I was not losing particles to velocity ^^

    Why is it problematic that the time step has been adjusted? I regard it as something good, since it means up to that point in the simulation I have used a "large" time step and simulated it efficiently and when it starts impacting the obstacle, the software it self figures out to reduce the time step - I think that is good?

    1) Approximately. The speed of the system is 2.75 m/s according to theory, so I could just have used c = 27.5 m/s. Since DualSPHysics uses Dambreak theory for the auto calc of c, I just took that value and multiplied with 10, so I got 34. That is also fine and gives it a tiny bit more safety in that regards

    2) I am just opening it in Excel but imported incorrectly as you hinted it. When I import it properly I see

    Which is an integer which makes sense. This value of 16971 particles is of course much, this is due to it being a failed simulation :-)

    So everything is good again and makes sense

    Kind regards, Ahmed - thank you very much!

Sign In or Register to comment.