the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
An Actuator Sector Model for Wind Power Applications: A Parametric Study
Mohammad Mehdi Mohammadi
Hugo Olivares-Espinosa
Gonzalo Pablo Navarro Diaz
Stefan Ivanell
Abstract. This paper investigates different actuator sector model implementation alternatives and how they compare to actuator line results. The velocity sampling method, tip correction, and time step are considered. A good agreement is seen between the line and sector model in the rotor plane and the wake flow. Using the sector model, it was possible to reduce the computational time by 75 % compared to actuator line model as it is possible to run the simulation with a larger time step without compromising the accuracy considerably. The results suggest that the proposed velocity sampling method produces the closest results to the line model with different tip speed ratios. Moreover, the vortex-based tip correction applied to the sector model results in the lowest error values, among the considered methods, to correct the radial load distributions. Also, it is seen that reducing the time step compared to the one used for actuator disk/sector does not provide an advantage considering the increased computational time.
- Preprint
(6667 KB) - Metadata XML
- BibTeX
- EndNote
Mohammad Mehdi Mohammadi et al.
Status: final response (author comments only)
-
RC1: 'Comment on wes-2023-16', Anonymous Referee #1, 21 Jun 2023
An Actuator Sector Model for Wind Power Applications: A Parametric Study
This manuscript presents a comparison between the actuator line model (ALM) and the actuator sector model (ASM). The article presents interesting insights on the effect of changing the sampling location and time-step in the ASM.
There is a fundamental drawback in the method/approach used and the comparisons presented in the manuscript. The authors fix the value of epsilon to epsilon=2dx in their study. This choice limits the validity of the study in 2 ways ( https://doi.org/10.1002/we.1747 ):
1. The results are not expected to converge because epsilon/dx=2 is not enough resolution to resolve the aerodynamics of the blade for that epsilon. The recommended values to be within less than 1% for converged quantities along the blades is epsilon/dx>=5.- By fixing epsilon/dx=2, the value of epsilon changes with the grid. Every time the grid is changed, the definition of the problem changes because epsilon is changed. This also leads to differences along the blade of at least a few percent.
I recommend the authors to assess these drawbacks and reevaluate the manuscript.
Specific comments:
The authors dive into the topic of tip corrections, but this is yet another source of error/difference for the simulations.
The results presented have many sources of error and they all contribute to the blade aerodynamics, which makes it too difficult to draw conclusions. The authors are trying to attribute the differences to the choice of sampling location or the width of the ALM, but there are also differences because of the grid resolution and size of epsilon.
“As can be seen in Fig. 2, the power and thrust values have decreased with each refinement for
the OP approach, as the forces become more concentrated due to the decreased value of ϵ which is proportionate to the cell
95 side length”
Comment: These figures are showing that there is a grid convergence problem and should not be confused with the old/new velocity sampling. Is epsilon changing as the grid is refined? Can you provide some details of the simulation before showing results?
“Regarding velocity sampling, in ALM, the velocities are sampled on the location of the blade points for each blade. However, in ASM, more methods are conceivable”
Comment: There are other methods to sample the velocity in the ALM. Please expand the literature review and cite the work in this area.
“Each case is run for 600 seconds as it is seen the thrust and power values do not change considerably after about 450 seconds which corresponds to flow passing through the entire domain about 3 times.”
Comment: This is not usual; these simulations typically converge in around 30-60 seconds of simulated time.
Comment: Fig 7 is referenced before 5 and 6. Please change the text or the order of the figures.
“Based on our investigation, at 0.7 the axial velocity matches best with the one from ALM for OP as shown in Fig. 10.”
Comment: What is 0.7? Please clarify in the text.“4.2 Velocity Sampling Method”
Comment: This section is again mixing the effects of velocity sampling with the vortex-based correction.
Fig 15
Comment: It is difficult to draw conclusions from these results. All the results differ, but which one is the correct one? The tangential force is different amongst all codes.
“The mean relative errors for radial distribution of axial and tangential forces are 0.57% and 1.20% and the results are presented in 240 Fig. 17 and Fig. 18.”
Comment: This measurement is misleading, max error would be a better metric. The errors are quite large in some parts of the blade, especially towards the tip.
Citation: https://doi.org/10.5194/wes-2023-16-RC1 - RC2: 'Comment on wes-2023-16', Anonymous Referee #2, 01 Aug 2023
- AC1: 'Comment on wes-2023-16, Final response', Mohammad Mehdi Mohammadi, 14 Sep 2023
Mohammad Mehdi Mohammadi et al.
Mohammad Mehdi Mohammadi et al.
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
617 | 101 | 14 | 732 | 10 | 6 |
- HTML: 617
- PDF: 101
- XML: 14
- Total: 732
- BibTeX: 10
- EndNote: 6
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1