An adaptive model for vanadium redox flow battery and its application for online peak power estimation

Zhongbao Wei, Shujuan Meng, King Jet Tseng, Tuti Mariana Lim*, Boon Hee Soong, Maria Skyllas-Kazacos

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

75 Citations (Scopus)

Abstract

An accurate battery model is the prerequisite for reliable state estimate of vanadium redox battery (VRB). As the battery model parameters are time varying with operating condition variation and battery aging, the common methods where model parameters are empirical or prescribed offline lacks accuracy and robustness. To address this issue, this paper proposes to use an online adaptive battery model to reproduce the VRB dynamics accurately. The model parameters are online identified with both the recursive least squares (RLS) and the extended Kalman filter (EKF). Performance comparison shows that the RLS is superior with respect to the modeling accuracy, convergence property, and computational complexity. Based on the online identified battery model, an adaptive peak power estimator which incorporates the constraints of voltage limit, SOC limit and design limit of current is proposed to fully exploit the potential of the VRB. Experiments are conducted on a lab-scale VRB system and the proposed peak power estimator is verified with a specifically designed “two-step verification” method. It is shown that different constraints dominate the allowable peak power at different stages of cycling. The influence of prediction time horizon selection on the peak power is also analyzed.

Original languageEnglish
Pages (from-to)195-207
Number of pages13
JournalJournal of Power Sources
Volume344
DOIs
Publication statusPublished - 2017
Externally publishedYes

Keywords

  • Battery model
  • Estimation
  • Model parameters identification
  • Peak power
  • Vanadium redox flow battery

Fingerprint

Dive into the research topics of 'An adaptive model for vanadium redox flow battery and its application for online peak power estimation'. Together they form a unique fingerprint.

Cite this