The rapid developments of geotechnical constructions have greatly expedited the applications of 3D numerical modeling. Large-scale 3D finite element computations may lead to a series of large linear systems of equations, and solving these linear equations may significantly influence the efficiency of entire finite element computation. Complex geotechnical problems usually involve multi-body and multiphase coupling interactions, and the remarkable contrast in material properties for different phases or materials may lead to slow convergence or even failure of a Krylov subspace iterative method. To improve the solution efficiency and reliability of a Krylov subspace iterative method, a robust preconditioner should be employed. Based on some typical geotechnical problems, a partitioned block preconditioner is developed. By using some numerical examples, performances of the preconditioners are examined and their applicability is evaluated.