High-fidelity concurrent multidisciplinary design optimization based on arbitrary hybrid turbulence modelling and fully coupled FSI

Faculty-development competitive research grants program for 2023-2025

Significant advancements have been made in the design and optimization of wind turbine blades using numerical approaches during the last few decades. Since the introduction of adjoint-based methods [1 - 2], whose computational cost is independent of the number of design variables, researchers have been able to tackle a variety of large-scale and practical problems, such as aerodynamic and aerostructural optimizations of complete aircraft configurations [3, 4]. In most applications, the underlying physical issue is assumed to be in a steady state, as shown by the abundance of literature on aerodynamic form optimization using stable Euler and Reynolds-averaged Navier-Stokes equations (RANS). Nevertheless, many aerospace engineering applications involves issues which are inherently unstable, such as active flow management, turbomachinery, aeroelastic flutter, biologically inspired flying, and aeroacoustics.
Unsteady adjoint-based optimization has not gotten its well-deserved attention as its steady equivalent, and thus, there are less established approaches for addressing this issue. This is partly because there are less efficient fully coupled fluid-structure interaction and turbulence modelling methods available, as well as prohibitively high computational costs are incurred due to a general lack of efficient and powerful computing equipment and huge quantity of solution data is needed to solve the unstable adjoint equation. Moreover, aerodynamic surfaces in motion are involved in a multitude of unstable issues. The necessity to precisely account for the required mesh movement in the governing equations and related adjoint equations creates extra complications. In spite of this, the increase in processing power and the enhancement of time-accurate numerical techniques have resulted in an increase in the study in this field during the last decade. The growing focus in reducing airplane noise – a phenomena intrinsically unstable due to the ever-stricter aviation noise restrictions – also acts as a required and timely impetus.
This proposed project relies on experience from an ongoing project which was mainly focused on developing a new whole-field approach for the aerodynamic analysis of wind turbines using the state-of-the–art computational technologies, developing an advanced parallel CFD code with RANS and hybrid turbulence models for fluid-structure interaction for single blades and whole rotors and developing a fast and effective physics-based optimization method for the aerodynamic and structural dynamic design of wind turbine blades. And our team has obtained fruitful results from each of the objectives. As a continuation of the project, we propose to extend our research to 3H high fidelity (3H: high fidelity, high dimensionality and high multi-disciplinarity) concurrent transient multidisciplinary design optimization as shown in Fig. 1. In the project, the interactions of aerodynamics and structural systems will be carefully modelled with tightly-coupled two-way interaction for multidisciplinary analyses and transient multidisciplinary design optimization. The software system will be based on Openfoam as in the current project in order to develop an open source fully integrated MDO system to serve the wind energy engineering community.

Fiure 1. 3H transient MDO with FSI
To accelerate the development of these new technologies, computational tools are essential to capture the multi-physics interactions that drive the performance and the structural sizing of these complex systems. The traditional “sequential” design of aerodynamic shape and internal structure falls short of taking full advantage of the interaction between fluid and structural dynamics [5–7].
We thus propose to develop our high-fidelity multidisciplinary design optimization methods developed in-house and implement them in open source software to perform aero-structural optimization studies of a large wind turbine configurations, using an efficient, highly-scalable, gradient-based approach based on a coupled CFD/FEM model with arbitrary hybrid turbulence modelling (VLES), ALE (Arbitrary Lagrangian Eulerian) method for rotors only and OM (Overset Mesh) method for complete wind turbine systems.
In our current project, we tried to investigate the trade-offs between steady-state aerodynamic efficiency and structural cost of a benchmark rotor using more than around 200 structural and geometric design variables. Mass and torque are simultaneously used as performance metrics.
In summary, we aim to achieve the following objectives in this project:
1) To develop and implement URANS/VLES-based high-fidelity concurrent aerodynamic design optimization
2) To develop and implement the discrete 2-way FSI based on ALE and OM methods and VLES for the concurrent transient design optimization
3) To develop and implement 3H (high disciplinary, fidelity and dimensionality) concurrent transient multidisciplinary design optimization.
In wind turbine design, the optimization of the blades is essential to enhance the aerodynamic as well as mechanical performance except increasing the power generated by the wind turbines. Though, due to the complicated nature of aerodynamics as well as the interaction between the air and the blade structure, blade design optimization is the hardest point of the wind turbine optimization.
The PI and Co-PIs investigated the advantages of a RANS-based approach to enhance the performance of a turbine rotor using a large set of design variables to tailor both its planform and cross-sectional shape as well as the tightly-coupled steady state aerostructural optimization. This project will be built on top of that work, extending the analysis and optimization to a tightly-coupled, transient aerostructural model.
The combination of CFD and Finite Element Method (FEM) tools extends the set of design variables that can be used in optimization and circumvents some of the assumptions that limit the modeling capabilities of conventional Blade Element Momentum Theory (BEMT) and beam theory-based design tools. This approach has the potential to maximize the performance and minimize the design costs of the next generation of wind turbines considering transient performance and operations. While wind turbine optimization is not a new concept, we propose the development of the first high-fidelity transient aerostructural optimization in the design cycle of a large turbine rotor.
During the previous project, the PI and the Co-PIs have successfully developed arbitrary hybrid turbulence modeling approach for high fidelity wind turbine CFD simulation, high-fidelity 2-way FSI simulation of a wind turbine using fully structured multiblock meshes in OpenFoam for Accurate Aero-Elastic Analysis, and adjoint-based high fidelity concurrent aerodynamic design optimization of wind turbine and adjoint-based high fidelity steady state multidisciplinary design optimization. And some of the representative results from each branch of our research are shown below [8 - 13].
An accurate two-way FSI model for wind turbine research is developed and implemented for the first time in Openfoam to investigate the FSI effects on the NREL Phase VI wind turbine. Fully structured MG mesh method is used for the fluid and solid domains to achieve good accuracy. Coupling method based on the ALE is developed to ensure rotation and deformation can happen simultaneously and smoothly. The simulation results shows that hi-fidelity CFD and CSD based 2-way FSI simulation could provide high accurate results for wind turbine simulation and multi-disciplinary design optimization (MDO).