In the in-core direct solver, the solution algorithm used performing the factorization of the left-hand-side of the above equation is referred to as Crout elimination (K = LDL^T decomposition). Once the factorization is completed, the unknown displacements U are calculated by back-substitution [Hughes, T.J.R, 2003]. The current implementation adopts an in-core direct solver supported with band-width optimization [Siek, et. al., 2002], where the solution procedure utilizes a sky-line (column-wise) storage scheme to store only values below the sky-line of K . Thus, a sparse global stiffness matrix is obtained, provided that both storage requirements and analysis time to solve the above equation are optimized.

In addition, an in-core sparse solver (from INTEL-MKL, 2008) is also available, which is mainly used for high performance computations (ideal for big models).

In a typical RAM Frame run, the force matrix F contains a number of load cases. The corresponding displacement vector U also has as many columns as the number of load cases. Note that there is no limit on the number of load cases that can be handled in a single run.