[vishwesh]

Hi

I am creating a new solver which identifies a VOF bubble in a certain region at every time step, converts it to lagrangian bubble, writes its data to kinematicCloud file and then uses Lagrangian to track that bubble and InterFoam to track the VOF.

My solver is running fine in single processor, and now I want to convert it for parallel processors.

According to what I have searched so far, reduce operations allow compiling multiple processor data into a single data.

As of now, my every processor is doing the following things:

1. Finding out cell labels satisfying my alpha value constraint.
2. Using those cell labels to identify bubble parameters.
3. Writing those cell parameters to kinematicCloud file for Lagrangian tracking.

Now, since each processor is running the solver, each tries to create a bubble according to its mesh and thus instead of one complete bubble, I will end up with multiple smaller bubbles. Moreover, since each processor is running the solver, each of them tries to modify the kinematicCloud file simultaneously which finally results in crashing of the solver.

I wanted to ask if there is any way by which the following can be done:

1. Every processor identifies the celllabels satisfying the alpha constraint. This data is stored in a dynamicList.
2. The data obtained from the previous step is combined (or reduced) to form one dynamicList having the data.
3. Only one processor works on the compiled dynamicList, identifies the bubble, and writes it to kinematicCloud. No other processor should be able to do it.

Any help regarding this will be greatly appreciated.

Thanks

Vishwesh