Advanced numerical modeling of reciprocating compressors

The quantification of pressure losses generated during the suction and discharge phases is essential to predict the working cycle of a reciprocating compressor. Beyond automatic valves, which are the main sources of fluid-dynamic losses, a relevant contribution is provided by all of the other components along the cylinder suction and discharge paths. Low-order numerical models for the simulation of the working cycle account for pressure losses through the imposition of global flow coefficients, which must be known from either experiments or three-dimensional CFD simulations. Therefore, the accuracy of the prediction of the absorbed power is strongly related to the methodology for the evaluation of the flow coefficients for all of the components. The geometrical features of the gas path have a great impact on the flow coefficient, which also depends on the piston motion during the cycle: indeed, the compression chamber shape, the pocket section and the flow pattern vary as a function of the piston position. Moreover, the valve and the working fluid used, as well as the operating conditions have a strong influence on the flow coefficient evaluation. For this reason, the number of configurations that should be simulated is extremely high and the evaluation of the pressure losses in the suction and discharge gas path with a 3D CFD steady state simulation is impossible without a substantial reduction of the set-up times and of the number of simulations. Within this scenario, the aim of this thesis is the introduction of a new methodology for the evaluation of the pressure losses occurring along the overall gas path in order to obtain a strong reduction of both the set-up and computational times for the realization of steady state CFD simulations. In more detail, the proposed approach starts by the idea to obtain the computational domain by means of parametric CAD models of the fluid domain directly without the need of starting from the more complex 3D drawings of the solid metal components and assemblies. This solution requires a greater initial effort for the realization of a parametric CAD model but allows an important reduction of the set-up times when simulating more cylinders starting from the same parametric model. The idea of the CFD routine described in this thesis is then to use parametric CAD models for the simulations of a large number of cylinders in order to reduce the global computational times. Indeed, the parameterization comprises also the mesh definition and the simulation setup phases, thus leading to a fully automatable approach. Nevertheless, even in case of exploiting the use of parametric models, the number of simulation cases that should be analyzed for each machine is still too high. For this reason, a detailed analysis was carried out to define the minimum number of simulations needed to obtain accurate results with the minimal computational times. Different solutions are then proposed in order to reduce the number of configurations to simulate; the evaluation of all the possible operating conditions of the large size reciprocating compressors analyzed in this thesis was obtained by means of 48 simulations for each machine. Moreover, the results obtained by the application of the defined approach to an industrial case are shown. In the second part, the utility of the simplified parametric routine to enhance the predictability in reciprocating compressors is highlighted by applying the results achieved to different numerical approaches. In particular, the effect of the approximations introduced in the flow coefficient evaluation on the performance prediction of large size reciprocating compressors is investigated. The numerical model used for the analysis is an in-house one-dimensional model based on the finite volume method (FVM). The study is carried out for three different compressor sizes and three different working gases, showing that the requirements in terms of flow coefficient accuracy depend on the specific case. The comparison between the numerical results using different flow coefficients and the measurements collected on a dedicated test bench allowed highlighting the importance to accurately predict the flow coefficient along the overall suction and discharge gas path. Moreover, a methodology to perform a 2D CFD simulation of the working cycle of a reciprocating compressor is developed in order to provide more accurate results than low-order models and, at the same time, to guarantee the reduction of the computational effort with respect to unsteady 3D CFD simulations. In more detail, the analysis presented in this work is carried out for a double-acting large bore cast iron cylinder. The simplifications required to allow the reduction of the three-dimensional fluid domain to an equivalent two-dimensional configuration are shown. In particular, the importance to use the flow coefficients obtained by means of the simplified CFD parametric routine for the pressure losses evaluation in the suction and discharge gas-path is highlighted. Finally, the possibility to use the parametric CFD routine to increase the predictability of the heat transfer process in a reciprocating compressor is shown. A brief description of a conjugate heat transfer (CHT) approach is given and the suitability to use the parametric routine for the forced convection modeling inside suction and discharge gas ducts is underlined. Furthermore, the utility of the parametric routine as well as the accuracy of both 2D and CHT models is highlighted by means of the comparison between the numerical results and the measurements collected during experimental campaigns.