Bluebottle is capable of simulating particle-laden fluid flows in configurations ranging from particles in Stokes flow to flows at moderate Reynolds number. Since particles are fully resolved within the flow solver, Bluebottle accurately captures the physical behavior of these fluid flows at arbitrary particle volume fraction. Bluebottle has been used to investigate particle behavior in channel flows, shear flows, sedimentation, and other canonical situations with up to 1,000,000 particles in domains consisting of over 7-billion degrees of freedom on up to 216 GPUs.
Constructed with the primary goal of bringing to bear the power inherent in today's GPU hardware on particle flow physics simulations, Bluebottle has been shown to operate up to 100-times faster than a legacy CPU implementation. The newest many-GPU implementation, Bluebottle-3.0, exhibits favorable strong and weak scaling.
Bluebottle-3.0, the new many-GPU implementation, has been released under the Apache License Version 2.0. Download it via Github.
The original single-GPU version of Bluebottle is also available under the Apache License Version 2.0 at Github.
- D.P. Willen, A.J. Sierakowski, G. Zhou, and A. Prosperetti, Contiuity waves in resolved-particle simulations of fluidized beds, Phys. Rev. Fluids 2 (2017) 114305, dx.doi.org/10.1103/PhysRevFluids.2.114305
- Y. Wang, A.J. Sierakowski, and A. Prosperetti, Fully resolved simulation of particulate flows with particles-fluid heat transfer, J. Comput. Phys. 350 (2017) 638-656, dx.doi.org/10.1016/j.jcp.2017.07.044
- A.J. Sierakowski, GPU-centric resolved-particle disperse two-phase flow simulation using the Physalis method, Comput. Phys. Comm. 207 (2016) 24-34, dx.doi.org/10.1016/j.cpc.2016.05.006
- A.J. Sierakowski, Numerical simulation of disperse particle flows on a graphics processing unit, Ph.D. thesis (2016), Download
- A.J. Sierakowski and A. Prosperetti, Resolved-particle simulation by the Physalis method: Enhancements and new capabilities, J. Comput. Phys. 309 (2016) 164-184, dx.doi.org/10.1016/j.jcp.2015.12.057
The following links should provide enough information to run simulations using Bluebottle.
Bluebottle consists of several pieces. The following links describe the code in some detail. The information here is meant to provide a good starting point for those interested in the details of the algorithms implemented in the code. It is best digested using both the Bluebottle code documentation and the source code itself.
The server at lucan.me.jhu.edu is a small GPU development cluster used to develop Bluebottle. Lucan has two compute nodes:
- Guinevere: A small node dedicated to development tasks. It contains four EVGA GeForce GTX 660 GPUs.
- Arthur: A high-powered compute node dedicated to running large production cases. It contains four NVIDIA Tesla K40 GPUs.
Note: The following documentation relates to the many-GPU implementation of Bluebottle, which has recently completed active development.