The topic of noise, vibration and harshness (NVH) is often an important design consideration for both the aerospace and automotive industries. But this is just one of many different engineering situations where finite element analyses (FEA) can be employed to accurately simulate the structural and/or acoustic behavior of a system. However, these systems generally result in computationally expensive FEA models that require some trade-offs between accuracy and time to solution. In this presentation, a reduced order model will be demonstrated that can be used for low to medium frequency ranges, undamped (or minimally damped), structural, acoustic, or fluid-structural interaction (FSI) models with frequency independent properties. The vectors which form the orthogonal basis for the reduced order model are termed as Krylov subspace vectors, these are computed using the Arnoldi algorithm. The reduced order model is generated by projecting the harmonic system of equations from a higher-dimensional FEA space to a lower-dimensional Krylov subspace. This reduction can achieve efficient solution times while preserving the important properties of the original system. Once the reduced harmonic solutions are generated, the final results can be obtained by projecting them back to the higher-dimensional FEA space. This presentation will show some results of the implementation of this approach in the Ansys Mechanical program. Some results will be given demonstrating the accuracy of this method on several models along with the performance characteristics relative to simply solving the entire system of equations in the FEA space at each frequency point. It will be shown that the reduced order method using the Krylov subspace, in combination with distributed memory parallel processing, will be significantly faster in most applications while preserving a sufficiently accurate solution for most engineering purposes. The results demonstrate that this reduced order model technique can be a valuable option in engineering applications where a very fine frequency resolution is desired.