Static Analysis for Worst-Case Battery Utilization

• Published in: 2019 IEEE/ACM 7th International Conference on Formal Methods in Software Engineering (FormaliSE) pp. 21 - 30
• Main Authors: Ivanov, Dmitry, Schupp, Sibylle
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.05.2019
• Subjects: Analytical models; Batteries; battery utilization; fixed-point computation for total lattices; KiBaM; Kinetic theory; Lattices; Mathematical model; Static analysis; Transfer functions; worklist algorithm
• ISSN: 2575-5099
• DOI: 10.1109/FormaliSE.2019.00009

Lithium-ion batteries demonstrate two kinds of nonlinear behavior: rate capacity effects and recovery effects. These effects are dual to each other and show the dependency between the power consumption profile and battery utilization, i.e., the fraction of the battery charge that can be withdrawn.We propose a static analysis for computing the lower bound of battery utilization based on the Kinetic Battery Model (KiBaM), an analytical model capturing nonlinear battery behavior. Our method does not depend on the battery charge level and can be composed with other flow analyses and model checking techniques for improved accuracy. We propose a modification to the worklist algorithm for totally ordered semilattices with computable fixedpoints of transfer functions, which is necessary for computing our analysis. We prove the termination and correctness of our algorithm and introduce a nonforgetful extension to it for speeding up convergence. We implement the battery-utilization in the software verification tool CPAchecker by encoding the analysis together with our algorithm in a relaxed instance of a configurable program analysis (CPA). Our experiments show that the convergence speed up is sometimes achievable, but does not necessarily lead to a performance improvement.