Machine learning algorithms running on quantum computers are being used to improve Li-ion battery performance and safety.
The prodigious potential of quantum computing is being applied to a critical energy storage problem: Improving battery simulation models that could help accelerate research into safer, more efficient energy storage along with new battery materials for electric vehicle and other consumer applications.
Cambridge Quantum Computing, a developer of quantum computing algorithms, is working with the German Aerospace Center to investigate how quantum machines can be used to boost the fidelity of battery simulations. The quantum algorithms are being applied to solve partial differential equations, rendering an initial one-dimensional simulation of a lithium-ion battery cell.
Once processed, the machine learning framework provides the foundation for rendering full 3D battery simulations that can be executed on “Noisy Intermediate-Scale Quantum” computers. The German research center said it would run its quantum simulations on an IBM Q quantum machine.
“We would like to improve the performance of the [Li-ion] battery, which means how much energy they can store without compromising on safety,” said Mattia Fiorentini, head of machine learning and quantum algorithms at Cambridge Quantum.
The research also addresses battery durability and supply chain issues such as reducing reliance on lithium used in commercial batteries. Research could also focus on alternative battery materials like plentiful, high-energy zinc.
The 3D battery simulation models also provide a new use case for emerging quantum computing platforms, including noisy systems like IBM’s Q. Those platforms are being augmented by machine learning algorithms like Cambridge Quantum’s. The combination would lay the foundation for high-resolution, multi-scale simulations models, the partners said, including simulations of entire battery cells.
“The hardware will get better,” Fiorentini said in an interview.
Along with solving partial differential equations associated with simulation models, quantum computing also can be used to describe a full system by encoding information on a quantum computer. “That’s extremely useful,” Fiorentini added.
“Quantum computers can indeed do a better job compared to classical ones. What we are looking to do in this collaboration with the German Aerospace Center would be first of all to prove that the mathematical problem that describes batteries can be solved by a quantum computer.” Fiorentini is confident they can.
“The high prize we are aiming at is really to try to prove something similar to the computational fluid dynamics domain, which is to find an aspect of a battery or a particular design of the battery where… we can prove [quantum computers] have an advantage.”
Hence, the goal of the collaboration is using a quantum-based machines to simulate an entire battery system. The resulting model would provide a detailed profile of battery performance while also being used to focus on specific aspects of battery design.
That would enable researchers to “tune and improve all aspects of a battery,” including rapid prototyping of new materials and assessing their behavior. That opens the possibility of speeding assessment of performance characteristics like energy densities that can be safely stored, toxicity, durability and battery behavior measured over a longer time span, said Fiorentini.
Aerospace researchers at the German center have previously used classical computer modelling to study a range of different battery types, including lithium-ion and “beyond-lithium technologies” like zinc thin-films.
The latest collaboration illustrates the evolution of quantum computing and machine learning for addressing real-world problems in key areas like energy storage and battery safety. “This is one of the earliest works combining partial differential equation models for battery simulation and near-term quantum computing,” the research center said in announcing its collaboration with Cambridge Quantum Computing.
Fiorentini said the collaboration would help accelerate the transformation of quantum computing from a theoretical tool to a computational platform offering key advantages in areas like 3D simulation modeling. As quantum computing evolves, he concluded, “We need to play the long game.”
This article was originally published on EE Times.
George Leopold has written about science and technology from Washington, D.C., since 1986. Besides EE Times, Leopold’s work has appeared in The New York Times, New Scientist, and other publications. He resides in Reston, Va.