Abstract
With recent advances in second-generation high temperature superconductors (2G HTS) and cable technologies, various numerical models based on finite-element method (FEM) have been proposed to help interpret measured AC loss and assist cable design. The T-A formulation, implemented in COMSOL, shows great potential for reducing the overall computation costs. In this paper, the performance of the T-A formulation for calculating the AC loss of coated superconductors and cables were assessed and compared against the widely accepted H formulation, with benchmark model of a single REBCO tape in 2D/3D and a 14-strand Roebel cable. Evaluation and comparison on key metrics including the computation time, the number of degrees of freedom and the numerical accuracy were presented, which could provide a reference for researchers in applying the T-A formulation for AC loss calculation.
Highlights
With recent advances in commercialized second-generation high temperature superconductors (2G HTS), proposals of new cable structures and upgrades of previous ones are currently under way, including Roebel assembled coated conductor [1], conductor on round core (CORC R ) [2] and twisted-stacked-tape cable (TSTC) [3]
A solenoid demo magnet with CORC R cables retained the critical current over 4 kA in a background field of 14 T, generating a total center magnetic field of 15.86 T [6]
We focus on the performance of the T-A formulation for calculating the AC losses of HTS coated conductors and cables, with comparison against the H formulation, VOLUME 9, 2021
Summary
With recent advances in commercialized second-generation high temperature superconductors (2G HTS), proposals of new cable structures and upgrades of previous ones are currently under way, including Roebel assembled coated conductor [1], conductor on round core (CORC R ) [2] and twisted-stacked-tape cable (TSTC) [3]. With the T-A, it is valid to apply the thin-strip approximation to our test case of a single tape, and convenient to impose the transport current constraint by a set of Dirichlet boundary conditions, as proposed in [27] (Fig. 1(a)). ELEMENT ORDER As presented in several reports [37], [39], the common practice when modeling with H formulation is to use edge elements (or ‘curl elements’ in COMSOL R ) In our study, both first and second order curl elements were considered in 2D, and first-order curl elements were tested in the case of 3D. With the T-A formulation, it was reported in [29] and [31] that with first or second order Lagrange elements for both T. and A, spurious oscillations could appear in the current density J profile at certain study steps. The relative tolerance in 2D and 3D models are 1e-5 and 1e-3, respectively, and Minimal update frequency is used, unless otherwise stated
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call