Actions for Stabilization of Turbulent Separated Flow Fields for Adjoint Implementation
Stabilization of Turbulent Separated Flow Fields for Adjoint Implementation
- Author
- Maul, Ian
- Published
- [University Park, Pennsylvania] : Pennsylvania State University, 2024.
- Physical Description
- 1 electronic document
- Additional Creators
- Coder, James George
Access Online
- etda.libraries.psu.edu , Connect to this object online.
- Graduate Program
- Restrictions on Access
- Open Access.
- Summary
- Advanced active flow control methods play a crucial role towards enhanced aircraft efficiency. To optimize these control strategies, adjoint methods can be used as they provide a general framework for analyzing sensitivities to flow separation, which is critical for enhanced design. In the presence of chaotic dynamics characteristic of turbulent separated flows, adjoint solutions are subject to exponential growth and fail to produce meaningful gradients, requiring added stabilization. Most proposed stabilization methods, however, are computationally prohibitive for practical flows. This thesis presents a stabilization strategy of mean flow fields obtained with scale-resolved methods for use with adjoint methods to compute sensitivities to flow separation. Using the in-house implicit LES solver WRBLES, test cases exhibiting turbulent separated flows over an NACA 0012 airfoil at M=0.2 were simulated. The employed stabilization strategy utilized a time-averaged mean flow as input and made use of a turbulent viscosity operator trained by high fidelity scale-resolved flow fields to damp chaotic dynamics. In addition, a body force method was constructed to help preserve salient flow physics characteristic of the initial mean flow. The results presented in this work were obtained using two RANS eddy viscosity formulations: the standard k-[epsilon] eddy viscosity and its Realizable variant. The effectiveness of the body force method coupled with these added viscosity fields is also assessed on a case-by-case basis. The first case, denoted Case 1, exhibited a fully stalled flow field at Re=2,400 and an angle of attack of 20 degrees with large laminar vortices, and was a relatively inexpensive case for testing. Cases 2 and 3 were simulated with Re=100,000 at an angle of attack of 10 degrees and 16 degrees respectively. Case 2 presented a laminar separation bubble with turbulent reattachment, and Case 3 exhibited a stalled flow field with large-scale vortex shedding and turbulence. For all three cases, each stabilization method incorporating the turbulent viscosity operator successfully stabilized the flow field, but the Realizable k-[epsilon] model coupled with the body force method provided results most similar to the input mean flow, preserving the majority of the salient flow physics. However, a linear correction factor of 0.75 applied to the Realizable eddy viscosity field had to be introduced in order to preserve the separation bubble in Case 2. The resulting stabilized states represent quasi-RANS solutions accurately accounting for separated flow physics, and can be used in future work to obtain useful adjoint solutions aiming to enhance the understanding of the underlying flow physics of separation for flow control applications.
- Other Subject(s)
- Genre(s)
- Dissertation Note
- M.S. Pennsylvania State University 2024.
- Technical Details
- The full text of the dissertation is available as an Adobe Acrobat .pdf file ; Adobe Acrobat Reader required to view the file.
View MARC record | catkey: 46451355