One of the main research areas for the Titan supercomputer, the elite Cray system installed at Oak Ridge Leadership Computing Facility, is energy, and in particular nuclear energy. At this point, about 20 percent of America’s energy is produced by nuclear power. Many experts say that increasing this percentage is key to meeting our nation’s growing energy needs while also maintaining our commitment to reducing fossil fuel emissions.

Researchers at Oak Ridge National Laboratory’s Consortium for the Advanced Simulation of Light-Water Reactors (CASL) are working to ensure an abundant nuclear energy supply with optimal safety margins. To support this mission, they are using modeling and simulation technology to both improve existing reactors and to innovate more efficient novel reactor designs.

In light-water reactors, the most common type of nuclear reactor, boiling plays a big role in the overall process. Fuel rods heat water and that water produces steam. The steam is then passed through turbines to generate electricity. In some reactors, the same water that cools the fuel rods also boils and produces steam, while in others the coolant water is kept under higher pressure to prevent boiling, and there is a lower-pressure intermediate stage where boiling produces steam.

Scientists are very interested in the boiling process because it is directly tied to the efficiency and safety of the reactor. Boiling water makes bubbles, and this is the real target of study: the bubbly flow, or turbulence, effected by the mixing together of liquid water and steam.

A team of researchers, led by Igor Bolotnov of North Carolina State University, is using Titan to perform direct numerical simulation (DNS) to characterize these turbulent bubbly flows and simulate boiling to the point of predictability. These extremely intensive calculations are well-suited to Titan’s scalable architecture and large total memory.

Yet, even with a system such as Titan, the fastest supercomputer in the US, the simulation challenges of such complex multiphase fluid dynamics are such that researchers must make approximations in simulating certain stages of the boiling process. Knowing where it’s acceptable to make these “short-cuts” is essential. The scientists use a technique called adaptive mesh control, which lets them automatically adjust grid resolution in certain regions of the simulation while it is occurring.

The detailed simulations provide fundamental data on the nature of turbulence in these systems that is leading to the proposal of new and improved models of bubbly flows that are far less computationally expensive than DNS simulations. An example of this is the parallel flow solver, PHASTA, developed by Bolotnov’s team. The solver was used to harness 46,000 of Titan’s CPU cores to study how bubbles form and evolve and how they affect the surrounding liquid.

The project highlights the advantages of computer modeling over physical experimentation. The combination of DNS and supercomputing let’s researchers alter variables, such as pressure and flow quickly and cheaply, whereas changing these conditions in a physical experiment would often not be feasible from a cost, time and/or resource standpoint.

“We want to be able to predict the behavior of the boiling,” said Bolotnov. “With simulation, you can experiment with any temperature and any pressure.”

Bolotnov’s team is using their understanding of flow to develop models for several reactor designs, both new and existing.

The study, which is featured in the April 3, 2013 issue of Journal of Fluids Engineering, reflects how computer simulation is fundamental to accelerating the discovery process. The payoff is that reactor designs can be brought to market faster and at less cost.

The next step for the team is developing a version of PHASTA that will be able to leverage Titan’s GPUs. This will allow them to perform multiphase flow simulations for pressurized water reactor geometries. The GPU support will also get them closer to their goal of simulating an entire reactor core, which they are aiming to do “in the near future.”