Permanent Magnetization and Corrosion-Related Field Prediction of Complex Structures

The objective of the proposed research is to further develop tools that can accurately and efficiently predict changes in induced and permanent magnetization of a Naval vessel due to dynamic changes in the applied magnetic field, to mechanical stress related to hydrostatic pressure, and to eddy currents which include corrosion-related effects. The electromagnetics group at the University of Kentucky has been developing Magström to predict the magnetic fields of ships as they move through the earth’s magnetic field and are subject to dynamic stresses and EMagström to predict corrosion-related electromagnetic phenomena. Magström/EMagström relies on a fast-direct solver called MFDlib, which is capable of efficiently handling large structures. In particular, this effort will further develop and validate Magström, EMagström and MFDlib as follows For Magström, the effort includes 1) incorporation of a novel Locally Corrected Nyström-to- Moment Method (LCN-MoM) for tetrahedral elements into the non-linear transient solver used to predict permanent magnetization in magnetostrictive materials and eddy currents in conducting materials, 2) improving the efficiency of the LCN-MoM field prediction, 3) a thorough validation of the LCN-MoM prediction of permanent magnetization of complex structures cycled through various dive profiles, and 4) addition of pyramidal mesh elements and pyramidal bases to allow for mixed hexahedral-tetrahedral modeling. For E-Magström, the effort includes 1) implementing analytics to improve system fill time efficiency for triangle elements, 2) implementing dual image planes to efficiently model the sea floor, 3) continued implementation of zoning for piecewise homogeneous electrolytes and investigation of alternative methods for modeling vertically stratified electrolytes, 4) investigation of the feasibility of modeling ac corrosion fields due to time-harmonic variations in impressed fields and mechanical structure, and 5) continued validation of results in cooperation with NSWCCD for increasingly complicated geometries and polarization curve profiles. For MFDlib, the effort includes 1) implementing methods to improve the accuracy of pseudoskeleton fill techniques (which have exhibited compromised accuracy for some E-Magström simulations), 2) adding improvements in the randomized projection methods (RPM) required by EMagström to handle non-linear polarization curves, 3) incorporating recompression techniques in the H2 fill to further reduce system memory, 4) continuing development of the overlapped diagonal factorization to reduce the non-linear solver iterative solution time, 5) implementing a deflated binormalization method to help further reduce system memory. Finally, the effort includes continuing technical support of the tools to NSWCCD as needed