Bottlenecks for particle growth in turbulent aerosols
We ask for support for large-scale computing to study particle-laden turbulent flows. Our interest is to numerically study collision and possible coagulation of particles advected by turbulence. The focus of our study is to understand rain initiation in warm clouds and planetesimal formation in accretion disks. We shall calculate the stationary probability distribution function of the relative positions and velocities of the particles for various circumstances, e.g., in homogeneous turbulent flows, in turbulent flows with gravity, for identical and non-idential particles. We shall also model collisions between the particles. We use the pencil-code which is sixth-order in space and third-order in time. The code has been well-tested in all the swedish supercomputers and show linear scaling up to tens of thousand processors. This proposal was originally a large-scale proposal. The proposal was denied with the following evaluation: "Scientifically and technically very interesting research proposal and corresponding strong consortium. However, the requested time of 2 million core hours per months is very large for the small group of active researchers (which have another proposal in this round), and the usage of PI's medium scale allocation. Therefore, the recommendation is to continue on medium scale, to prove the needs, before applying for large allocation again." As suggested by the committee I am now applying for a medium scale allocation for the same physical problem.