In recent years, a number of authors have studied entropy generation in Wells turbines. This is potentially a very interesting topic, as it can provide important insights into the irreversibilities of the system, as well as a methodology for identifying, and possibly minimizing, the main sources of loss. Unfortunately, the approach used in these studies contains some crude simplifications that lead to a severe underestimation of entropy generation and, more importantly, to misleading conclusions. This paper contains a re-examination of the mechanisms for entropy generation in fluid flow, with a particular emphasis on RANS equations. An appropriate methodology for estimating entropy generation in isolated airfoils and Wells turbines is presented. Results are verified for different flow conditions, and a comparison with theoretical values is presented.
1Wells turbines have been studied extensively, both experimentally [15,31,24,63,74,53,57,48] and numerically [40,62,17,76,30,72,32,27], but mainly from a first-law perspective. More recently, a number of authors [66,67,68,69,71] have focused on the irreversibilities in this component, trying to link the amount of available work (exergy) and the entropy produced.The application of second-law analyses is not novel, as such analyses have been done by many authors, but mainly from a system perspective: Bejan [8,9] presents an extensive review of the application to different thermodynamic systems. Similar approaches have been used in the analysis of thermal power plants [39], wind turbines [56,6,7,60,52], and vortex generators [33].Computational Fluid Dynamics (CFD), aside from making possible a more rapid and economic evaluation of many systems (compared to experimental testing), can provide a better and more detailed understanding of many phenomena, by capturing a level of detail that would be extremely difficult in an experimental study. It allows a very fine decomposition of the overall problem that can be used to locate the sources of irreversibility within the system. Entropy generation by thermal and viscous sources can be calculated directly by post-processing the fields of thermo-and fluid-dynamic variables available from the numerical solution [8]. However, particular attention needs to be paid to the nature of the flow and to the solution approach. While in laminar flows all entropy generation is directly found in the CFD solution [18,2,70,33,79,59], this is not true for turbulent flows, unless a Direct Numerical Simulation (or DNS) is used. However, with current computational resources, the use of DNS is still impractical even for moderately large Reynolds numbers. To make high-Reynolds-number flows treatable, the governing equations are either averaged (leading to the Reynolds Averaged Navier Stokes equations, or RANS) or filtered (leading to the Large Eddy Simulation approach, or LES).Moore and Moore [49, 50] present a methodology for calculating entropy production in viscous flows, using RANS equations. They divide the mean entropy production into the contr...