Load balance when handling a great amount of parcels

[DongWangChina]

In my spray cases, to ensure computational stability, there are lots of parcels in the computation domain (up to 535752). Though I use 640 cores in this simulation, there might be only tens of them computing the spray cloud, while others just wait. The consequence is that simulation on the spray cloud takes half of the simulation time. Is there any solution for that issue?