The purpose of model order reduction (MOR) techniques is to provide a fast response approximate solution of a given system. Naturally the error induced due the approximation should be as low as possible. The paper presents the benefits of using a Krylov subspace based MOR for solution of large scale finite element (FE) dynamical systems. The main advantage of using a Krylov MOR is significantly shortening of the computational time while an induced error is almost negligible or very low. The process will be presented on structural harmonic analysis oflarge machine tool and a transient thermal analysis of thermal subsea equipment.