Abstract. The Scaled Boundary Finite Element Method (SBFEM) is a semi-analytical method which combines the advantages of the Boundary Element Method (BEM) and the Finite Element Method (FEM). Like in the BEM, only the boundary needs to be discretized. On the other hand, the SBFEM is based on the virtual work principle and does not need any fundamental solutions. If the scalability condition is fulfilled, a separation of variables representation can be employed leading to a quadratic eigenvalue problem and a linear equation system which can be solved by standard methods. The SBFEM has proven its high efficiency and accuracy in the presence of stress singularities, especially in 2D fracture mechanics when the singularity is entirely located within the considered domain.However, in 3D elasticity problems, there can also be singularities on the discretized boundary itself. Then, the SBFEM suffers from drawbacks also well known from the standard FEM, i.e. moderate accuracy and bad convergence. To overcome these deficiencies in such 3D cases, we propose the enrichment of the standard separation of variables representation with analytical fields which are known to exactly fulfill the local boundary conditions:The examples of a single plane crack and two perpendicularly meeting cracks in an isotropic continuum are considered. It is demonstrated that the method's original excellent accuracy and convergence are regained, at a minimum cost of additional degrees of freedom (DOF). The normalized errors in the solution of the quadratic eigenvalue problem already become negligibly small for very coarse boundary meshes. The obtained convergence orders are often optimal and sometimes even superconvergence is observed.