CFD Predictions of the National Experimental Turbine
Open Access
- Author:
- Tien, Leland
- Graduate Program:
- Mechanical Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- May 17, 2024
- Committee Members:
- David Hall, Outside Unit & Field Member
Reid Berdanier, Major Field Member
Robert Kunz, Chair & Dissertation Advisor
Karen Thole, Major Field Member
Robert Kunz, Professor in Charge/Director of Graduate Studies - Keywords:
- CFD
gas turbine
LES
RANS
AM
turbomachinery - Abstract:
- Gas turbines have become the platform technology for modern society's energy and travel needs. They find application in power generation and nearly all forms of travel including aviation, marine, and terrestrial. The increasing cost of fossil fuels coupled with the climate crisis make improving gas turbine efficiency more important than ever. The National Experimental Turbine (NExT) program was created to address this critical technology drive. The primary mission of the NExT program is to develop a modern research turbine for U.S. manufacturers and institutions \cite{thole2021}. The program aims to provide a platform for acquiring detailed data for new design method development and new design concept validation. The NExT program is being conducted at the Penn State Steady Thermal Aero Research Turbine (START) Laboratory \cite{barringer2014design}, and is characterized by tight axial spacing and a richly complex film cooling scheme, similar to the first stage turbine in a modern aircraft engine. The NExT turbine stage is also additively manufactured (AM), representing the first time that an AM turbine has been spun in a U.S. university laboratory. In this thesis, several state-of-the-technology computational fluid dynamics (CFD) methods are applied and advanced in the context of modern axial turbine configurations. Specifically, a contribution of this thesis is the first application of a broad range of modern CFD methods to the NExT configuration. A stepwise approach has been pursued ranging from isolated blade row relative frame-of-reference Reynolds-Averaged-Navier-Stokes (RANS), through unsteady RANS (URANS), full stage harmonic balance methods, and finally to fully space-time resolved stage rotor-stator interaction. Ultimately, three-dimensional, time accurate full stage (stator+rotor), sublayer resolved RANS/URANS and wall resolved Large-Eddy-Simulation (LES) models are pursued. All internal flow paths, cooling holes, and detailed tip clearance geometry are resolved and incorporated into these high-fidelity models. The blade and endwall solid metal volumes are fully represented, and conjugate heat transfer (CHT) is accommodated through appropriate solid-gas boundary conditions and energy equation solution within the metal. The hierarchy of methods pursued, and their attendant comparisons to data, each other, and RANS performed by NExT's design firm, provide the community with an objective assessment of the comparative cost-to-accuracy of CFD methods for modern turbines, this being a second contribution of the work. Another major contribution of the thesis is that, to the author's knowledge, the analyses performed represent the first published work where wall resolved LES has been deployed on a cooled turbine stage that includes explicit resolution of internal flow paths. An attendant contribution is that in this context a new method for loosely coupled adaptive grid refinement for LES was developed. This new approach is applicable to any LES simulation and may help the greater CFD community efficiently incorporate LES in design. These LES studies have yielded a number of significant findings, inaccessible to the RANS based simulations. Surface temperature predictions from the LES simulations exhibit notable differences from the RANS models, emphasizing the difficulties that RANS-based heat transfer models have on airfoil leading and trailing edges. Turbulent pressure spectra from this case exhibit a -7/3 power law relationship adjacent to blade boundary layers and a -11/3 power law relationship in the near wake region. Interestingly, blade passing frequency harmonics also decay at a −7/3 slope. To the author's knowledge this phenomenon has never been observed numerically or experimentally, and therefore represents a turbulence physics discovery worthy of future study. New computational techniques have also been developed here to accommodate the emergence of AM turbine parts and the associated CFD application and validation challenges associated with design-intent vs. as-manufactured geometry. 3D structured light scans of the NExT turbine vanes and blades are used to create spatially averaged as-built CFD models. Because these 3D scans only characterize the exterior surface of the hardware, a new technique has been developed in this dissertation to incorporate internal cooling features from design intent models. These spatially averaged as-built models with design intent cooling are shown to significantly improve results. As-built vs. experimental comparisons of select key turbine performance parameters including stage pressure ratio, stage temperature ratio, and exit flow angles are shown to greatly improve predictions over design intent models. Also presented is a new technique to assess boundary layer thickness over the uneven as-built surfaces. Collectively, these new AM spatial scan averaging and the boundary layer analysis techniques represent a fifth contribution of the thesis. CFD modeling of heat transfer in turbines (and many other configurations) are well known to suffer severe accuracy deficiencies. Accordingly, an important opportunity associated with the NExT program is that the first-of-its-kind local heat transfer measurements being generated present an opportunity for the community to advance heat transfer modeling for turbomachinery CFD. Since infrared (IR) thermography of the NExT vanes and blades is just now becoming available, the timeline of this thesis did not permit quantitative comparison of CHT results against these data. Nonetheless, a final contribution of the thesis is the heat transfer studies presented that lay the groundwork for such comparisons in the future. Ultimately, all of the research presented in this thesis aims to capture a better understanding of the physics governing the NExT configuration. Through this pursuit, the specific contributions of this thesis should help direct future studies of NExT and the broader turbomachinery CFD community.