Implementation of Loss Correlations into a Trough-Flow Program for Axial Turbines

Abstract

The design of turbines, which involves a lot of parameters, requires numerical methods. At DLR in Göttingen 2 through flow programs for axial turbines were developed, which can compute the flow through axial turbines in 2D. Both are written with FORTRAN and use loss correlations, obtained by researchers during the last 60 years through experiments on conventional blades, to model 3D effects. The aim of my placement has been first to study the loss correlations in order to implement some of them into the program STAC4T and secondly I had to analyse and validate the program, by comparing the models, also with experimental results. The implementation of the loss correlations took place in 3 steps. Firstly, the study of the documentation for the loss correlations and the code of the 2 programs SLC4T and STAC4T. Secondly, the implementation of the parts of the code from SLC4T referring to the appropriate losses, by redefining the quantities and by adapting this code for STAC4T. And thirdly, the “debugging” of the program. The original method to estimate the loss coefficients in an axial turbine comes from Ainley and Mathieson, deviding the total losses in 3 parts : the profile loss, the secondary flow loss and the tip clearance loss. I have plotted the loss coefficients at different rotational speeds and the three models show very similar results. The Traupel model was initially implemented in STAC4T and seems the most realistic concerning the isentropic efficiency. The DCKO (Dunhame-Came and Kacker-Okappu) is the first Ainley-Mathieson version with improvements from Dunhame-Came (1970) and Kacker-Okapuu (1980) and we can see a similitude between Traupel and DCKO concerning the general variations of entropy. The AMDCKOBS is the most recent version with a new secondary loss model from Benner-Sjolander (2006). This program is satisfying, as it gives reasonable results, so we can say that the implementation of the loss correlations into STAC4T was successful. By designing a turbine, a program like that can give very fast performance results so it can be helpful for the turbine manufacturers during the first steps of designing a turbine.