Actuator line modelling of wind turbines and their wakes
<p>Wind energy has a central role in the green energy transition, with a projected surge in installed capacity in the coming decades. Understanding and modelling wind turbine aerodynamics and their wakes is key for the economical deployment of wind energy at large scale. This is only possible...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
2024
|
| Subjects: |
| _version_ | 1826314681721028608 |
|---|---|
| author | Zormpa, M |
| author2 | Willden, R |
| author_facet | Willden, R Zormpa, M |
| author_sort | Zormpa, M |
| collection | OXFORD |
| description | <p>Wind energy has a central role in the green energy transition, with a projected surge in installed capacity in the coming decades. Understanding and modelling wind turbine aerodynamics and their wakes is key for the economical deployment of wind energy at large scale. This is only possible if high-fidelity flow field data are available, often obtained by numerically simulating rotors using high resolution methods. One such simulation method is the actuator line method (ALM), which is the focus of this thesis. This thesis addresses evaluating and improving the numerical efficiency and accuracy of the ALM and explores the potential to use the generated high-fidelity data to validate and inspire low-fidelity models.</p>
<br>
<p>The ALM uses a virtual representation of each turbine blade, avoiding the complex and costly processes of meshing the blade surface. This thesis first addresses the robustness of the ALM in the prediction of rotor forces. An important step in any ALM implementation is the flow sampling: the process of obtaining a characteristic local velocity vector at each collocation point of the actuator line in order to calculate blade forces through reference to pre-defined aerofoil coefficients. It is shown that the widely-used method of sampling the flow at the collocation point that lies along the actuator line introduces a large dependence of the rotor forces on the time step. This has given rise to stringent time step limitations, typically employed by ALM practitioners. Using a line average sampling method that samples the flow at locations circumscribing the collocation point and outside the region where computed blade forces are imposed, improves robustness of predicted rotor power and thrust across a large range of time steps. This enables the use of larger time steps than those typically used in collocation point sampling ALM simulations, improving the numerical efficiency of the ALM.</p>
<br>
<p>The ALM is embedded in a large eddy simulation (LES) solver to perform ALM-LES of an experimental rotor and its wake. The numerical sensitivity of the rotor power and thrust to ALM parameters is first established. The sensitivity of the predicted wake on the mesh size is tested, with highly-resolved LES simulations used as a reference. It is demonstrated that in cases where the mesh is insufficient to resolve the tip vortices, the vortex breakdown mechanism is altered, with lower resolution simulations resulting in vortices merging into shear layers that then themselves develop instabilities.</p>
<br>
<p>The importance of the turbine support structure (nacelle and tower) for accurately simulating the turbine wake is explored. Three methods for introducing low-fidelity representation of the support structures are assessed for their ability to capture the aerodynamic interaction between the support structure and the rotor wake observed in experimental measurements. Good agreement with experiments is achieved when a low-resolution body-fitted representation of the support structure is used. Additionally, the effect that the support structure has on the far wake aerodynamics is investigated: when including the tower, the wake centre typically drops with streamwise distance. This deviation from an axisymmetric wake profile, as commonly assumed in engineering models, is quantified by comparing a Gaussian wake model against the ALM-LES data.</p>
<br>
<p>In order to explore the potential of reducing the dimensionality and numerical cost of simulating wake flows, the ALM-LES spatio-temporal flow data are analysed using a dynamic mode decomposition (DMD) algorithm. The merger between two turbine wakes in tandem is explored: the lower frequency background flow due to the upstream turbine is found to be more dynamically persistent, dominating local rotor dynamics.</p>
<br>
<p>This work evaluates and extends actuator line modelling of wind turbines and their support structures for the simulation of wind turbine wake aerodynamics. It thus supports more efficient, robust, and accurate turbine and wake modelling for wind energy research.</p> |
| first_indexed | 2024-12-09T03:09:07Z |
| format | Thesis |
| id | oxford-uuid:299200b0-d106-4d79-ac5e-f55dafa13adf |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-12-09T03:09:07Z |
| publishDate | 2024 |
| record_format | dspace |
| spelling | oxford-uuid:299200b0-d106-4d79-ac5e-f55dafa13adf2024-10-03T14:25:03ZActuator line modelling of wind turbines and their wakesThesishttp://purl.org/coar/resource_type/c_db06uuid:299200b0-d106-4d79-ac5e-f55dafa13adfAerodynamicsWind turbinesComputational fluid dynamicsTurbulenceEnvironmental engineeringEnglishHyrax Deposit2024Zormpa, MWillden, RSmyth, AVogel, CApsley, D<p>Wind energy has a central role in the green energy transition, with a projected surge in installed capacity in the coming decades. Understanding and modelling wind turbine aerodynamics and their wakes is key for the economical deployment of wind energy at large scale. This is only possible if high-fidelity flow field data are available, often obtained by numerically simulating rotors using high resolution methods. One such simulation method is the actuator line method (ALM), which is the focus of this thesis. This thesis addresses evaluating and improving the numerical efficiency and accuracy of the ALM and explores the potential to use the generated high-fidelity data to validate and inspire low-fidelity models.</p> <br> <p>The ALM uses a virtual representation of each turbine blade, avoiding the complex and costly processes of meshing the blade surface. This thesis first addresses the robustness of the ALM in the prediction of rotor forces. An important step in any ALM implementation is the flow sampling: the process of obtaining a characteristic local velocity vector at each collocation point of the actuator line in order to calculate blade forces through reference to pre-defined aerofoil coefficients. It is shown that the widely-used method of sampling the flow at the collocation point that lies along the actuator line introduces a large dependence of the rotor forces on the time step. This has given rise to stringent time step limitations, typically employed by ALM practitioners. Using a line average sampling method that samples the flow at locations circumscribing the collocation point and outside the region where computed blade forces are imposed, improves robustness of predicted rotor power and thrust across a large range of time steps. This enables the use of larger time steps than those typically used in collocation point sampling ALM simulations, improving the numerical efficiency of the ALM.</p> <br> <p>The ALM is embedded in a large eddy simulation (LES) solver to perform ALM-LES of an experimental rotor and its wake. The numerical sensitivity of the rotor power and thrust to ALM parameters is first established. The sensitivity of the predicted wake on the mesh size is tested, with highly-resolved LES simulations used as a reference. It is demonstrated that in cases where the mesh is insufficient to resolve the tip vortices, the vortex breakdown mechanism is altered, with lower resolution simulations resulting in vortices merging into shear layers that then themselves develop instabilities.</p> <br> <p>The importance of the turbine support structure (nacelle and tower) for accurately simulating the turbine wake is explored. Three methods for introducing low-fidelity representation of the support structures are assessed for their ability to capture the aerodynamic interaction between the support structure and the rotor wake observed in experimental measurements. Good agreement with experiments is achieved when a low-resolution body-fitted representation of the support structure is used. Additionally, the effect that the support structure has on the far wake aerodynamics is investigated: when including the tower, the wake centre typically drops with streamwise distance. This deviation from an axisymmetric wake profile, as commonly assumed in engineering models, is quantified by comparing a Gaussian wake model against the ALM-LES data.</p> <br> <p>In order to explore the potential of reducing the dimensionality and numerical cost of simulating wake flows, the ALM-LES spatio-temporal flow data are analysed using a dynamic mode decomposition (DMD) algorithm. The merger between two turbine wakes in tandem is explored: the lower frequency background flow due to the upstream turbine is found to be more dynamically persistent, dominating local rotor dynamics.</p> <br> <p>This work evaluates and extends actuator line modelling of wind turbines and their support structures for the simulation of wind turbine wake aerodynamics. It thus supports more efficient, robust, and accurate turbine and wake modelling for wind energy research.</p> |
| spellingShingle | Aerodynamics Wind turbines Computational fluid dynamics Turbulence Environmental engineering Zormpa, M Actuator line modelling of wind turbines and their wakes |
| title | Actuator line modelling of wind turbines and their wakes |
| title_full | Actuator line modelling of wind turbines and their wakes |
| title_fullStr | Actuator line modelling of wind turbines and their wakes |
| title_full_unstemmed | Actuator line modelling of wind turbines and their wakes |
| title_short | Actuator line modelling of wind turbines and their wakes |
| title_sort | actuator line modelling of wind turbines and their wakes |
| topic | Aerodynamics Wind turbines Computational fluid dynamics Turbulence Environmental engineering |
| work_keys_str_mv | AT zormpam actuatorlinemodellingofwindturbinesandtheirwakes |