Regarding the definition of the input parameter "dedt" in turbulence simulation

[YaoguangPei]

I attempted to determine the true meaning of "dedt" by examining both the code and the documentation on turbulence drive. However, I arrived at two conflicting conclusions. From the code, I inferred that "dedt" likely represents the energy injection rate per cell. Conversely, based on the turbulence drive description referencing "the energy injection rate so that the total kinetic energy injected at every driving step," I believe "dedt" might represent the energy injection rate for the entire simulation domain. I wish to clarify the precise meaning of "dedt".

[tomidakn]

It seems to me that the injection rate in the code is not per cell but something global, but I am not sure. @changgoo, can you comment on it?

[changgoo]

dedt is the total kinetic energy injection rate for the entire domain. The code does the right thing. I'm not sure which part is confusing, but if you specifically point out the part that confuses you, I will take a look at it.

[YaoguangPei]

First of all, I'm very grateful for your answer. My confusion lies in the following code from the 'src/fft/turbulence.cpp' file:

if (pm->turb_flag > 1) {
    // driven turbulence
    de = dedt * dt;
} else {
    // decaying turbulence (all in one shot)
    de = dedt;
}
aa = 0.5 * m[0];
aa = std::max(aa, static_cast<Real>(1.0e-20));
b = m[1];
c = -de / dvol;

where ‘dvol = pmy_fb->dx1 * pmy_fb->dx2 * pmy_fb->dx3'. I tried to interpret the physical meaning of 'c' to determine the true meaning of 'dedt'. Therefore, I believe that if 'dedt' is the energy injection rate per cell, then 'c' can be interpreted as the energy injection rate per unit volume. However, this conflicts with the description in the "turbulence drive" entry, which mentions "the energy injection rate so that the total kinetic energy injected at every driving step." Hence, I am confused.

[changgoo]

dvol is the volume if a cell, not the total volume. The confusion here is that dvol should actually be multiplied to aa and b such that they are integrals of rhodv^2 and rhov*dv. Since the turbulence driver assumes a uniform grid, dvol factor is moved to the denominator of c for convenience.

[changgoo]

I wrote up a note for this. https://www.overleaf.com/read/kypfnvsqbwhw#2d9e53

I will add this to the wiki when I have time.

[YaoguangPei]

Thank you very much for your answer. I have fully grasped the meaning in the code. Once again, sincere thanks for taking the time to clarify this for me.