No abstract

This paper presents a fast and robust algorithm for contact detection between elliptical and ellipsoidal particles. The algorithm belongs to the class of geometrical potential methods, which consider the solution of two minimization problems in order to determine a contact point between the particles. The efficiency of the algorithm relies on several ingredients, namely, a transformation that maps the pair of ellipses (ellipsoids) into an ellipse (ellipsoid) centered at the origin and a unit circle (sphere), the construction of an effective initial guess to the solution of the minimization problem, the use of Newton's method for the root finding problem, and the introduction of an additional constraint to guarantee convergence to the desired root. The performance of the new algorithm is compared with that of existing contact detection algorithms on large sets of randomly generated pairs of particles. In particular, the results from several numerical examples show that the present algorithm is several times faster than the existing algorithms for comparable accuracy.

No abstract

This paper presents a novel finite-element approach for the electromagnetic modeling of superconducting coated conductors with transport currents. We combine a thin-shell (TS) method to the H-Φ formulation to avoid the meshing difficulties related to the high aspect ratio of these conductors and reduce the computational burden in simulations. The interface boundary conditions in the TS method are defined using an auxiliary 1-D finite-element (FE) discretization of N elements along the thinnest dimension of the conductor. This procedure permits the approximation of the superconductor's nonlinearities inside the TS in a time-transient analysis. Four application examples of increasing complexity are discussed: (i) single coated conductor, (ii) two closely packed conductors carrying anti-parallel currents, (iii) a stack of twenty superconducting tapes and a (iv) full representation of a HTS tape comprising a stack of thin films. In all these examples, the profiles of both the tangential and normal components of the magnetic field show good agreement with a reference solution obtained with standard black2-D H-Φ formulation. Results are also compared with the widely used T-A formulation. This formulation is shown to be dual to the TS model with a single FE (N=1) in the auxiliary 1-D systems. The increase of N in the TS model is shown to be advantageous at small inter-tape separation and low transport current since it allows the tangential components of the magnetic field to penetrate the thin region. The reduction in computational cost without compromising accuracy makes the proposed model promising for the simulation of large-scale superconducting applications.

We consider nonclassical entropy solutions to scalar conservation laws with concaveconvex flux functions, whose set of left-and right-hand admissible states u l , ur across undercompressive shocks is selected by a kinetic function ur = ϕ ♭ (u l ). We introduce a new definition for the (generalized) strength of classical and nonclassical shocks, allowing us to propose a generalized notion of total variation functional. Relying only upon the natural assumption that the composite function ϕ ♭ • ϕ ♭ is uniformly contracting, we prove that the generalized total variation of front-tracking approximations is non-increasing in time, and we conclude with the existence of nonclassical solutions to the initial-value problem. We also propose two definitions of generalized interaction potentials which are adapted to handle nonclassical entropy solutions and we investigate their monotonicity properties. In particular, we exhibit an interaction functional which is globally non-increasing along a splitting-merging interaction pattern.