[ycui]

Dear Foamer,

I want to calculate the velocity gradient tensor on a parcel for Lagrangian tracking. The code is in KinematicParcel.C:

// First calculate the velocity gradient for all fluid nodes
volTensorField gradU = fvc::grad(td.cloud().U());
// Access to the value at the cell where parcel locates
tensor gradUatPoint = gradU[this->cell()];
// Transform to the particle phase
tensor gradUatParticle = rotMatrix & gradUatPoint & rotMatrix.T();

The above 3 lines works, but has two disadvantages:
1. The first line cannot parallel computed, I don't the reason.
2. For every parcel, the first line will be executed. If tracking many parcels, its time consuming.

Is there any solution for:
1. Parallel compute fvc::grad()??
2. Or directly obtain the velocity gradient tensor at the cell point (remove line 1 and only keeps line 2 & 3)??

It bothers me for a long time. Thank you very much in advance!

Best,

ycui

[ycui]

I solve this problem, by:
1. put fvc::grad(U) in side CreateFields.H, read by the constructor of KinematicCloud.H, and set as vgt (velocity gradient tensor) a parameter.
2. set the interpolation function of vgt in KinematicParcle.H, otherwise cannot parallel computing.
3. use the interpolation function in KinematicParcle.C.

Then it runs very fast since for every time step fvc::grad(U) only calculate once, and allows for parallel computing.