A Comparative Study of Conventional and FEM-Based Approach for Dynamic Analysis of Large Reciprocating Compressor Foundation Including Soil-Foundation Interaction

Abstract

This paper aims to present a comparative study of dynamic response of a machine foundation supporting gas-lift reciprocating compressor using a conventional lumped parameter method and an FEM-based approach. Evaluation of dynamic response of machine foundation involves dynamic analysis of a rigid body supported on elastic half-space (soil mass). In general, conventional approaches involve idealization of the machine-foundation system as a rigid body and soil mass as a set of frequency independent (or dependent) elastic springs. However, for a foundation supporting large reciprocating machinery subjected to unusually high dynamic unbalanced loads acting at a significant height and foundation mat supporting several numbers of relatively flexible pedestals, due importance shall be attributed to the actual behavior of the machine-foundation system considering the soil-foundation interaction as well as the effect of boundary conditions. It is therefore necessary to substantiate the analysis result of conventional method with an alternative approach. As a part of the assessment of acceptability of lumped parameter output, i.e., natural frequencies and response amplitudes, a finite-element method–based approach was adopted. ANSYS (version 17) software was employed to create three-dimensional (3D) finite-element model where, apart from modeling the foundation-machine mass, soil layers were also modeled to include the kinematic and inertial interactions to simulate the real behavior of soil-foundation interaction as precisely as possible. Output results from both approaches were critically compared for all relevant parameters. This study indicates meaningful correlation among results obtained from both approaches in terms of amplitudes of vibration and natural frequencies of the foundation. For the governing coupled rocking-sliding mode, amplitude of vibration using conventional lumped parameter approach is obtained as 74.17 μm while the FEM approach results in an amplitude of 72.90 μm, indicating a difference of less than 2%. The frequency response curve, i.e., plot of variation of amplitude over a range of frequencies also reflects a significant amount of similarity in terms of maximum amplitude of response. The study also highlights the effect of soil improvement on the dynamic response.