Using Computational Models and Simulations for Battery Safety

Ann Nguyen:
Hello, I'm Ann Nguyen, Senior Associate Conference Producer with Cambridge EnerTech. Today, we have a podcast interview for the Battery Safety conference running this November 3-4 in Bethesda, Maryland. I'm happy to be chatting with one of our featured speakers, Dr. John Turner, Group Leader in Computational Engineering and Energy Sciences at Oak Ridge National Laboratory.

John, thank you for carving out some time for this interview.

John Turner:
You're quite welcome.

Ann Nguyen:
Much of your work over the last 25 years has involved applying computational science to problems such as nuclear energy as well as battery safety. How has the use of models, sensors and simulations to address battery safety evolved over that time?

John Turner:
As the available computational resources have increased over the last couple of decades, really, simulation capabilities in general have evolved in three related ways. The first being increased spatial resolution -- that is, finer meshes and more particles, whatever's needed for the simulation. The second would be increased coupling between the physical phenomena -- that is, feedback between the relative physical phenomena such as heat transfer and material motion whether fluid flow or ion transport or heat transport. The third being coupling between scales -- that is, feedback between continual molecular and atomic behaviors. That means less reliance on pre-computed tables and empirical fits to idealize situations and more integration of sub-grid models directly in simulation. That's probably the area that's under the most active research from my perspective. It's really highly relevant to batteries. I think we're really in the early stages of learning how to apply this to batteries.

For example, we recently did a simulation of a full, automotive battery pack consisting of 10 modules, each with four pouch cells in which every single layer is represented. About 16 million degrees of freedom with full 3D coupling of electro-chemistry and heat transport. Now this was just a simple one C discharge just to demonstrate the capability and next we'll do more challenging charge/discharge cycles in abnormal conditions. But it's just an example of how this has evolved and where it potentially can go. I think we're really still just at the very beginning of seeing how this will really pay off for battery design and optimization.

Ann Nguyen:
Can you please describe some of the work currently being conducted by your Computational Engineering and Energy Sciences group at Oak Ridge National Lab?

John Turner:
Our work is really focused on three major application areas. The first being nuclear energy. Oak Ridge was the lead lab for the DOE Innovation Hub known as the Consortium for Advanced Simulation of Light Water Reactors or CASL, and that started in 2010 and is now in its second five-year period. That of course required a certain set of physics. And of course batteries, as you know. The third being additive manufacturing. That includes the manufacture of carbon composites and most recently additive manufacturing of 3D printing of metals and carbon fiber-infused polymers. Those are three kind of seemingly different areas, but the common themes between these are coupled multiscale physics and high-performance computing. Those aren't the only three areas that we'll do, it's just the ones we're focusing on now and we consider ourselves computational scientists even though we are mostly a collection of nuclear engineers, mechanical engineers, aerospace engineers, and we focus on applying simulation technology to different application areas that are of relevance particularly for energy subjects.

Ann Nguyen:
On November 3, you'll discuss balancing accuracy and computational costs in battery simulation, focusing on simplified versus complex 3D models of battery architectures. What's the main theme you'd like to convey to the R&D engineers, cell manufacturers, and other peers in the audience?

John Turner:
I think one of the main points is that there remains a lot of potential impact for simulation in this community and that computational resources really are relatively inexpensive. Even if the desktop or single work station level, a system with one or more fancy video cards or graphical processing units can be extremely powerful, but significantly more is possible with even a small cluster and of course the large systems. National labs such as ORNL provide even more potential for impact. Of course these efforts have to be collaborative with the experimental community, but this has really paid off in other communities, I really think we're just at the start of seeing where this will go.

Ann Nguyen:
What challenges still need to be resolved to optimize computational models of batteries to predict and prevent battery incidents?

John Turner:
There's a continuing need for microscale data. For example, transport properties. The models that we're building now are able to incorporate the full heterogeneity and anisotropies inherent in battery behavior. But we don't always have accurate property data for that, a number of ongoing activities to get better data in particular the anisotropic aspects based on imaging and microscopic simulations, that's an ongoing challenge. Again, I think we're only at the early stages here. The simplified models used until now have been extremely useful in describing the normal charge and discharge type scenarios, but there's tremendous potential for exploring advanced design, for example 3D architectures and especially the accident scenarios that I've spoken about recently.

Ann Nguyen:
Okay, with that we'll wrap up. Thank you again, John for sharing your experiences and perspective.

John Turner:
Certainly. Thank you very much.

Ann Nguyen:
That was Dr. John Turner of Oak Ridge National Laboratory. He'll be speaking during the Battery Safety conference taking place November 3-4 in Bethesda, Maryland. To learn more from him, visit www.knowledgefoundation.com/battery-safety for registration info and enter the keycode “Podcast”.

This is Ann Nguyen. Thanks for listening.