Accelerated simulation of supercapacitors using order reduction techniques and waveform relaxation methods

Supercapacitors are promising energy storage devices due to their capability to deliver high peak current and to capture a huge amount of energy in a short time with very low internal power loss. They are candidate component to be used in as a secondary energy storage device in hybrid electric vehicles (HEV)). Precise model development and accelerated simulation environments are needed to minimize the design time (and subsequently the product-to-market time) and minimize the hardware redesign and retrofit.;Transmission line and RC parallel-branch equivalent models show relatively higher accuracy in predicting the dynamic behavior of the device. They also have shown features such as easy parameter identification from experimental data and easy adaptation to other system simulation tools. However, these circuit models have higher order compared to the other existing equivalent circuit model, and are computationally intensive.;This thesis provides an extensive characterization analysis for a commercialized supercapacitor by means of electrochemical impedance spectroscopy (EIS). Based on the EIS measurements, the circuit parameters for both circuit models (RC parallel-branch and transmission line models) have been extracted and validated. First, the performance of the order reduction techniques, when applied to high-order RC parallel-branch models, has been investigated. It has been shown that the high-order RC parallel-branch models, which show high accuracy in dynamic behavior of the supercapacitor, can be effectively reduced to low order models. The reduced-order models can be computed faster without compromising the model accuracy. Then, the features of the transmission line model of supercapacitor have been exploited by one of the waveform relaxation methods, which is an iterative method used for solving differential equations, in order to obtain fast time-domain simulation of supercapacitor. These aforementioned studies have been performed using numerical simulation and hardware measurement.