Abstract
Capytaine is an open-source Python-based Boundary Element Method (BEM) program designed to simulate wave-structure interactions employing linear potential flow theory in frequency domain. A long term work is currently underway to improve its performance and accuracy, which is behind its commercial counterparts. The present paper focuses on two steps of the numerical resolution: the choice between the direct and indirect boundary integral equation and the accuracy of the tabulation used as surrogate model of the Green function. The results are illustrated on two test cases for which issues had been previously reported, namely the floating cylinder with a thin heave plate and the DeepCWind floating offshore wind turbine platform reference model. The inaccuracy of previous versions of Capytaine on these two test cases is explained and new results are presented in comparison with WAMIT and HAMS.
