Turbulent fluid flow impacts a wide variety of engineering problems, but turbulence is mathematically complicated and poorly understood. One of the most productive research techniques is numerical simulation, but simulations powerful enough to be realistic are also computationally intensive enough to require the largest supercomputers to run at all.
SciServer offers a solution: run the simulations in advance, then store the results online for easy distributed access. SciServer offers accerss to the ensemble of the Johns Hopkins Turbulence Databases, a growing collection of multi-Terabyte computational models of turbulent fluid flow. These databases preserve the entire spacetime history of high-performance computing simulations, including forced isotropic turbulence, magnetohydrodymaic turbulence, and channel flow. Users can access the data using a variety of data retrieval and immersive analysis interfaces.
We will continue to add additional simulations into these databases in the future – but more importantly, we will offer new ways of accessing turbulence simulation data. Starting in late 2015, SciServer will build off its existing MyDB framework to develop data sharing and collaborative visualization tools. SciServer will also develop GPU computing technology to offer full-field operations, making future simulations even more powerful for real-world research.