| description abstract | A novel approach is developed to approximate the full mass matrix in the rotational-coordinate-based beam formulation, which can improve the efficiency of calculating its inverse in dynamic analyses. While the rotational-coordinate-based beam formulation can reduce numbers of elements and generalized coordinates, its mass matrix is a full matrix, such that corresponding Jacobian matrix is also full, and it is time-consuming to calculate its inverse. To increase efficiency of calculating its inverse, the full mass matrix is approximated in this work. Two approximations are adopted: (1) a double integral is approximated by a single integral; and (2) a full matrix is approximated by a sum of several rank-one matrices. Through this way, the approximate mass matrix can be decomposed as a band-diagonal sparse matrix and multiplication of low-rank matrices, and its inverse can be efficiently calculated using Sherman–Woodbury formula. Through this way, the approximate mass matrix can be efficiently calculated. Several numerical examples are presented to demonstrate the performance of the current approach, and its accuracy and efficiency are analyzed. | |
