PRELIMINARY FOR NASA & LPT PROGRAM PARTICIPANTS INTERNAL USE ONLY NASA Grant NAG 2099 Large Eddy Simulation of Flow in Turbine Cascades Using LESTool and UNCLE Codes Final Progress Report P.G. Huang Department of Mechanical Engineering University of Kentucky Lexington, Kentucky 40506-0503 Submitted to: Dr. David Ashpis, Technical Monitor NASA Glenn Research Center Cleveland, OH 44 13 5 August 3 1, 2 004 Large Eddy Simulation of Flow in Turbine Cascades Using LESTool and UNCLE Codes P.G. Huang Department of Mechanical Engineering University of Kentucky Lexington, Kentucky 40506-0503 Progress Summary During the period December 23,1997 and December August 31,2004, we accomplished the development of 2 CFD codes for DNS/LES/RANS simulation of turbine cascade flows, namely LESTool and UNCLE. LESTool is a structured code making use of Sth order upwind differencing scheme and UNCLE is a second-order-accuracy unstructured code. LESTool has both Dynamic SGS and Sparlart’s DES models and UNCLE makes use of URANS and DES models. The current report provides a description of methodologies used in the codes. 1. Introduction Flow transition plays an important role in turbomachinery applications. The majority of boundary layer flows in turbomachines involve flow transition under the effects of freestream turbulence, diverse pressure gradients, wide range of Reynolds numbers, flow separation, and unsteady wake-boundary layer interactions. Prediction of this type of complex flows is an important element in analysis and performance evaluation of gas turbine engine components and ultimately in the design of more efficient jet engines. Especially, in low pressure turbine applications prediction of transition becomes pivotal in terms of efficiency. For low pressure turbines the flow is mostly turbulent at the high Reynolds number conditions encountered at take off and the efficiency is at its design maximum. However, at high altitudes and cruise speeds which correspond to lower Reynolds number conditions, unpredicted losses and substantial drops in efficiency have been observed. These losses are attributed to flow separation on the suction surface of the turbine blades. At low Reynolds numbers, the boundary layers on the airfoil surface have a tendency to remain laminar and hence the flow may separate before it becomes turbulent, causing increase in fuel consumption and drop in efficiency. The impact of such losses is directly felt on the operation costs. It has been estimated that a 1% improvement in the efficiency of a low pressure turbine would result in a saving of $52,000 per year on a typical airliner. In order to calculate the losses and heat transfer on various components of gas turbine engines, and to be able to improve component efficiencies and reduce losses through better designs, accurate prediction of transitional boundary layers is essential. When one deals with a complex fluid phenomena like a transition, separation and turbulence, several hundred millions grid points are needed to resolve boundary layers and other flow structures correctly. We have started to develop technology to make such large scale simulations not only possible at supercomputing centers like NCSA or NAS but on inexpensive, high-performance clusters of PCs, or “Beowulfs”. These clusters are specialized for CFD applications, using the novel approach that the hardware, operating system, and application code are optimized together rather than separately. A Honorable Mention in the PricePerformance Category of the Gordon Bell Prize was awarded for this approach at IEEEIACM SC2000 Conference on High-Performance Networking and Computing. Several turbulence test cases have been computed and an overview of the results is given. 2. Code Descriptions (1) A description of LESTool is give in Appendix 1. (2) A description of UNCLE is given in Appendix 2. 3. Publications by the PI (1) Th. Hauser, T. Mattox, R. LeBeau, H. Dietz and P. Huang. Code optimizations for complex microprocessors applied to CFD software. SUM Journal for ScienGc Computing (acceptedf or publication), 2003. (2) Th. Hauser, R. LeBeau, T. Mattox, P. Huang and H. Dietz. Improving the performance of Computational Fluid Dynamics codes on Linux Cluster Architectures. 16th AIAA Computational Fluid Dynamics Conference. Orlando, Florida, June 23-26,2003. AIAA. (3) Th. Hauser, T. Mattox, R. LeBeau, H. Dietz and P. Huang. Scrutinizing CFD performance on multiple Linux cluster architectures. Presentation at the Clusterworld Conference & Expo, June 23-26, San Jose, California, 2003. (4) Th. Hauser, T. Mattox, R. LeBeau, H. Dietz and P. Huang. A comparative study of the performance of a CFD program across different Linux cluster architectures. Proceedings of the third LCI international confernece on linux clusters. St. Petersburg, FL, 2002. (5) Hauser, T., R. LeBeau, T. Mattox, H. Dietz and P. Huang. High-Cost CFD on a Low- Cost Cluster. Proceedings of 8th National CFD Conference. 2001. Invited Keynote Talk, E-land Taiwan, Aug 18-20. (6) Th. Hauser, T. Mattox, R. LeBeau, H. Dietz and P. Huang. High-Cost CFD on a Low-cost Cluster. Regular paper at SC2000 and honorable mention for the Gordon Bell award in the category "price-performance". 2000. (7) Th. Hauser and P. Huang. Numerical simulation ofturbulent spots. 25th Annual Dayton-Cincinnati Aerospace Science Symposium, 2000. (8) Th. Hauser and P. Huang. A hierarchical parallelization concept for a high- performance Navier-Stokes solver. Proceedings of the International Conference on Parallel and Distributed Processing Techniques and Applications (PDFTA'99). 1999. (9) Th. Hauser and P. Huang. Large eddy simulation of low pressure turbine flow. 24th Annual Dayton-Cincinnati Aerospace Science Symposium, 1999. (10) Th. Hauser and P. Huang. Shared Memory Parallelization of an implicit ADI-type CFD code. In C. Lin and et. al, editors, Parallel computational fluid dynamics, evelopment and applications of parallel technology, proceedings of the Parallel CFD98 Conference, pages 145-152. 1999. Elsevier Science B.B. (1 I) R. Savaram, T. Hauser and P. Huang. DNS and LES of homogeneous turbulence. 24th Annual Dayton-Cincinnati Aerospace Science Symposium, 1999. (12) Th. Hauser and P. Huang. Shared Memory Parallelization of an implicit ADI-type CFD code. Technical report, NASA-CR-208688, NASA, 1998. 4. Award received by the PI A Honorable Mention in the PricePerformance Category of the Gordon Bell Prize was awarded for this approach at IEEEIACM SC2000 Conference on High-Performance Networking and Computing \cite(hauser2000). - Appendix I LESTool