Saeed Asgari

A linear parameter-varying model for HEV/EV battery thermal modeling

Battery thermal management for high power applications such as electrical/hybrid vehicles is crucial. Modeling is an indispensable tool to help engineers design better battery cooling systems. A fast and accurate battery thermal model capable of predicting volume-averaged cell temperature or temperature at user specified locations under transient heat dissipation and varying mass flow rate is proposed. In such an approach, several state space models are generated first from computational fluid dynamics (CFD) results for different mass flow rates. Then a linear parameter-varying (LPV) model is created out of the state space models to account for non-constant flow rates. The model is then shown to provide excellent results compared with those from CFD under transient heat dissipation and mass flow rate. Such a LPV model runs many orders of magnitude faster than the original CFD model.

A state space thermal model for HEV/EV battery using vector fitting

Battery thermal management for high power applications such as electrical/hybrid vehicles is crucial. Modeling is an indispensable tool to help engineers design better battery cooling systems. A fast and accurate battery thermal model using state space representation can be used to predict volume-averaged cell temperature or temperature at user specified locations under transient heat dissipation. The matrices of the state space model are calculated by curve-fitting the step response of the state space model to that of the battery thermal system computed from computational fluid dynamics (CFD) results. In this paper, the vector-fitting (VF) method in the frequency domain is proposed for the curve-fitting. The frequency domain VF method is shown to provide advantages over the time domain Levenberg-Marquardt (LM) method. For batteries using prismatic cells, the VF method is shown to be more accurate. For multiple-input multiple-output (MIMO) battery thermal systems, the VF method generates much smaller state space models compared with those using the LM method. Therefore, for MIMO battery thermal systems, models from the VF method run much faster than those from the LM method.

A transient reduced order model for battery thermal management based on singular value decomposition

Battery thermal management for high power applications such as electric vehicles and hybrid vehicles is crucial. Modeling is an indispensable tool to help engineers design better battery cooling systems. A reduced order model (ROM) for battery thermal management based on singular value decomposition (SVD) is proposed. The ROM generation process starts with computation fluid dynamics (CFD) results of a battery model, from which SVD is used to calculate the approximate subspace which temperature solution belongs to. Then a state space model is created for each dimension of the subspace. The state space model is created taking advantage of the linear and time-invariance (LTI) properties of the battery thermal system. The state space model, the ROM, is then shown to provide very similar results as those from CFD under arbitrary transient power dissipation. The ROM runs orders of magnitude faster than the original CFD model.

Analytical integration-based causality checking of tabulated S-parameters

Causality of tabulated data is detected by checking for Hilbert-consistency between datas real and imaginary parts. This consistency can be robustly checked if the generalized Hilbert transform is used. Existing algorithms evaluate an integral in this transform numerically. In this paper, a new algorithm for checking causality that involves only analytical integration is presented.

A computationally efficient transient thermal model for electric machines based on singular value decomposition

A transient reduced order model (ROM) for electric machines based on singular value decomposition (SVD) is proposed. The ROM generation process starts with computation fluid dynamics (CFD) results of an electric machine model, from which SVD is used to calculate the approximate subspace that the temperature solution belongs to. Then a ROM working in the subspace is created. The ROM is shown to provide very similar results as those from CFD under arbitrary transient loss profiles. The ROM runs orders of magnitude faster than the original CFD model.

Improved procedure to test causality of tabulated S-parameters

Tabulated S-parameters are often tested for causality to ensure successful transient simulation. Existing approaches for this test either are not robust or have limited resolution in detecting noncausality. In this paper, a new approach for this test is proposed. The proposed approach is more robust and has better resolution than existing approaches.

An efficient transient thermal model for electronics thermal management based on singular value decomposition

Electronics thermal management for high power applications is crucial. Modeling is an indispensable tool to help engineers design better electronics cooling systems. An efficient transient thermal model for electronics thermal management based on singular value decomposition (SVD) is proposed. The model generation process starts with computation fluid dynamics (CFD) results of an electronics model, from which SVD is used to calculate the approximate subspace which temperature solution belongs to. Then a state space model is created based on the CFD and SVD results to calculate the temperature field in the subspace. The state space model is very efficient and runs orders of magnitude faster because the dimension of the subspace is small. The state space model is then shown to provide very similar results as those from CFD under arbitrary transient power dissipation.

Accurate delay-based convolution macromodel using delayed least-squares convolution

Existing delay-based convolution macromodels for S-parameters of electrical interconnects can compromise accuracy. This paper solves this problem by extending the recently-proposed convolution-based macromodeling method called least-squares convolution (LSC) to model propagation delays. The new method is referred to as the delayed LSC (DLSC) method.