discretization, simulation, compressor, refrigeration
The current digital era has been providing great advantages for every task that relies on data processing. Engineering is not an exception, numerical simulation methods have been used more and more for product design. It is basically a predictive method, where the most relevant physical phenomena taking place inside the product are described by equations. They are numerically solved demanding a great amount of data processing especially for the simulation of large systems in transient regimes. One of the most challenging cases are those where the phenomena is highly dependent on spatial aspects, demanding a three dimensional (3D) approach to the analysis. The space or component is discretized into small volumes, or elements, and a set of equations have to be solved for each one of them. This might be challenging even for the latest generation computers concerning the processing time, because the amount of elements might be as much as millions. On the other hand, solutions modeling the geometry and associated physics as one dimensional (1D) volumes drastically reduce the processing time, but sometimes are less accurate. In this work we simulated a hermetic refrigeration compressor in a transient stable condition developing a hybrid solution, a 1D or 3D approach was applied to each separate component of the system according to its characteristics. Different levels of discretization were employed for the 3D parts and their effect on the results was evaluated. Among the results we could clearly see that some phenomena demand a certain level of discretization to be modeled, which was not possible with a pure 1D approach. However a very high level of discretization is sometimes not necessary as a lower level is able to properly describe the phenomena. Three components in a compressor were modeled with three different levels of discretization generating many possible configurations for the whole compressor; the results of some of them were compared with each other and with the experimental data showing good agreement.