the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Oct 2022
Aeroelastic Validation of the Offshore Wind Energy Simulator for Vertical-Axis Wind Turbines
Abstract. Vertical-axis wind turbines (VAWTs) offer some unique advantages over traditional designs, particularly for floating offshore and certain distributed wind applications. However, the modeling capabilities that exist for VAWT designs greatly lags those for the traditional horizontal-axis wind turbines (HAWTs). Differences between vertical and horizontal turbines necessitates several key additions in modeling, including the aerodynamic model, as well as solving a fundamentally different structural mesh. The Offshore Wind ENergy Simulator (OWENS) is specifically formulated to fulfill these requirements. This paper presents validation cases of this tool for modal, centrifugal, gravitational, startup, normal operation, and shutdown analyses. The aeroelastic validation is performed with increasing complexity from analytical test cases to an experimental VAWT. Validation data are taken from the Sandia National Laboratories 34 meter research turbine. The results of the validation cases are presented and examined.
- Preprint
(2677 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on wes-2022-91', Anonymous Referee #1, 10 Nov 2022
This study deals with the aeroelastic validation of OWENS code for VAWTs by a series of code-to-code and code-to-experimental comparisons. In general, the topic is of great interest and is useful for validation of similar codes.
General comments
- The structural solver is verified by comparison against GXBeam. A brief description of the GXBeam code is recommended before presenting the verification.
- Aeroelastic simulations include coupling between the structural solver and the aerodynamic solver. The present study verified the structural solver comprehensively but did not pay attention to the aerodynamic simulation methods and associated correction. The coupling between the structural solver and aerodynamic method should be briefly presented, though they have been studied in other papers.
Detailed comments:
- Line 54: the authors mentioned a new unsteady method (RPI). Please briefly clarify this method.
- Page 4: it is recommended to present a figure to show the structural model. Table 2 and Table 3 can be replaced by two equations. Parameters denoted by formulas are also recommended to be presented by equations.
- Section 3.1: beams with and without a bend are used for different cases. Please clarify the beam shape (e.g. by figures) to avoid possible confusion.
- The discrepancies shown in Figs. 4-6 are significant. The authors attribute the discrepancies to methods of applying structural damping. Please clearly state the damping coefficient applied in the two codes. Is there any other reason that can cause the discrepancies? For instance, the time marching scheme?
- Section 3.4: control modeling is important for the dynamic behavior of VAWT, it is recommended to give a more detailed description of the modeling of the controller for the normal operating, startup, normal stop and emergency stop conditions.
Citation: https://doi.org/10.5194/wes-2022-91-RC1 -
AC1: 'Reply on RC1', Kevin Moore, 11 Nov 2022
General comments
- The structural solver is verified by comparison against GXBeam. A brief description of the GXBeam code is recommended before presenting the verification. -> Agreed, a description has been added
- Aeroelastic simulations include coupling between the structural solver and the aerodynamic solver. The present study verified the structural solver comprehensively but did not pay attention to the aerodynamic simulation methods and associated correction. The coupling between the structural solver and aerodynamic method should be briefly presented, though they have been studied in other papers. -> Agreed, more details on the coupling method have been added.
Detailed comments:
- Line 54: the authors mentioned a new unsteady method (RPI). Please briefly clarify this method. -> The RPI method has been clarified
- Page 4: it is recommended to present a figure to show the structural model. Table 2 and Table 3 can be replaced by two equations. Parameters denoted by formulas are also recommended to be presented by equations. -> Tables have been replaced with equations, the formulas changed to numbers where appropriate, and a clearer reference to the structural model figure is added.
- Section 3.1: beams with and without a bend are used for different cases. Please clarify the beam shape (e.g. by figures) to avoid possible confusion. -> Clearer references and more detail is added to highlight the relative figures and distinguish the different analyses.
- The discrepancies shown in Figs. 4-6 are significant. The authors attribute the discrepancies to methods of applying structural damping. Please clearly state the damping coefficient applied in the two codes. Is there any other reason that can cause the discrepancies? For instance, the time marching scheme? -> More detail and clarity added as recommended.
- Section 3.4: control modeling is important for the dynamic behavior of VAWT, it is recommended to give a more detailed description of the modeling of the controller for the normal operating, startup, normal stop and emergency stop conditions. -> Agreed, more detail and clarity was added.
Citation: https://doi.org/10.5194/wes-2022-91-AC1
-
RC2: 'Comment on wes-2022-91', Anonymous Referee #2, 22 Dec 2022
The "References" section in very poor and lacks a large number of papers published in the last 15 years on Darrieus VAWT modeling. The authors in particular are recommended to analyze the works of the research groups of TU Delft (Prof. Ferreira), TU Berlin (Prof. Nayeri) and Florence (Prof. Bianchini).
The aerodynamic part is vaguely described. Please add more on the type/setting of dynamic stall models, virtual camber and virtual incidence correction, pitching moment treatment, post-stall polar extrapolation, struts’ resistant torque and tower shadow.
The description of other modules is also very short and does not allow the reader to understand which kind of results should be expected.
Overall, the impression is that the analysis is presented in a too synthetic fashion. Insight on the results is often missing. The simple “validation” has a scarce scientific interest for the reader if it is not connected with physical insights or with a critical analysis of the models used.
Discrepancies in figures 4-6 are also not negligible and their motivation is quite weak.
The discrepancies seen in Figure 9 are only marginally addressed and would deserve more analysis.
Even more critical is the analysis of Figure 13. Beyond the absolute wind speed values (but Re should be anyway high enough) predicting high AoAs is generally easier since the flow is attached. The results are quite poor and hardly connected only on the way TSR is achieved.
Citation: https://doi.org/10.5194/wes-2022-91-RC2 -
AC2: 'Reply on RC2', Kevin Moore, 08 Jan 2023
The "References" section in very poor and lacks a large number of papers published in the last 15 years on Darrieus VAWT modeling. The authors in particular are recommended to analyze the works of the research groups of TU Delft (Prof. Ferreira), TU Berlin (Prof. Nayeri) and Florence (Prof. Bianchini).
Thank you for your valuable feedback, the intro section has been updated with relevant citations from these suggested authors as well as significant additional clarification on the state of the art and needs for the paper’s primary topic on two-way aero-elastic verification/validation.
The aerodynamic part is vaguely described. Please add more on the type/setting of dynamic stall models, virtual camber and virtual incidence correction, pitching moment treatment, post-stall polar extrapolation, struts’ resistant torque and tower shadow.
Thank you, agreed. Additional descriptions have been added regarding the dynamic stall model type/setting, angle of attack corrections, extrapolation, tower, and strut considerations.
The description of other modules is also very short and does not allow the reader to understand which kind of results should be expected.
Agreed, additional clarification regarding the level of fidelity of the OWENS code, its core modules and brief clarification on expected level of accuracy in addition to better citations pointing to the specific formulation details have been added.
Overall, the impression is that the analysis is presented in a too synthetic fashion. Insight on the results is often missing. The simple “validation” has a scarce scientific interest for the reader if it is not connected with physical insights or with a critical analysis of the models used.
True, the initial verification cases of the simple beam models are synthetic and intentionally brief to match their purpose of building up to the full turbine validation. The OWENS structural solver has not been officially verified for these types of simple cases and were necessary prior to the complex full-turbine unsteady experimental validation. The comparisons with the experimental unsteady strain gauge data later in the paper, though not commented on, are well connected with physical insights and offer a much more detailed critical analysis of the OWENS code and its limitations. This has been clarified and updated in the verification/validation sections.
Discrepancies in figures 4-6 are also not negligible and their motivation is quite weak.
In the surrounding text, we describe that we can manually tune the different damping models in the two different codes to give resulting loads which match within a root mean square of less than 5% for all loads, giving a much better visual result. However, the development of a justifiable relationship between differing damping models was deemed out of scope for the primary objective of experimental validation; the simple setup and results were deemed adequate to show the ability to simulate structural damping of a forced resonate beam. We have revised this section to make this decision clearer and hope it to be satisfactory.
The discrepancies seen in Figure 9 are only marginally addressed and would deserve more analysis.
Agreed, additional detail on the differences and their potential origins have been added to this section including quantification of error for parked and spinning loads between the two simulations and the experimental data.
Even more critical is the analysis of Figure 13. Beyond the absolute wind speed values (but Re should be anyway high enough) predicting high AoAs is generally easier since the flow is attached. The results are quite poor and hardly connected only on the way TSR is achieved.
Thank you, the statement regarding attached flow at high angles of attack is confusing, however, this section has been revised for clarity. The errors associated with the experimental torque sensor at low loads are more clearly discussed in addition to better clarification between the appearance of errors between CP and Torque when viewed for a constant RPM TSR sweep.
Citation: https://doi.org/10.5194/wes-2022-91-AC2
-
AC2: 'Reply on RC2', Kevin Moore, 08 Jan 2023
Status: closed
-
RC1: 'Comment on wes-2022-91', Anonymous Referee #1, 10 Nov 2022
This study deals with the aeroelastic validation of OWENS code for VAWTs by a series of code-to-code and code-to-experimental comparisons. In general, the topic is of great interest and is useful for validation of similar codes.
General comments
- The structural solver is verified by comparison against GXBeam. A brief description of the GXBeam code is recommended before presenting the verification.
- Aeroelastic simulations include coupling between the structural solver and the aerodynamic solver. The present study verified the structural solver comprehensively but did not pay attention to the aerodynamic simulation methods and associated correction. The coupling between the structural solver and aerodynamic method should be briefly presented, though they have been studied in other papers.
Detailed comments:
- Line 54: the authors mentioned a new unsteady method (RPI). Please briefly clarify this method.
- Page 4: it is recommended to present a figure to show the structural model. Table 2 and Table 3 can be replaced by two equations. Parameters denoted by formulas are also recommended to be presented by equations.
- Section 3.1: beams with and without a bend are used for different cases. Please clarify the beam shape (e.g. by figures) to avoid possible confusion.
- The discrepancies shown in Figs. 4-6 are significant. The authors attribute the discrepancies to methods of applying structural damping. Please clearly state the damping coefficient applied in the two codes. Is there any other reason that can cause the discrepancies? For instance, the time marching scheme?
- Section 3.4: control modeling is important for the dynamic behavior of VAWT, it is recommended to give a more detailed description of the modeling of the controller for the normal operating, startup, normal stop and emergency stop conditions.
Citation: https://doi.org/10.5194/wes-2022-91-RC1 -
AC1: 'Reply on RC1', Kevin Moore, 11 Nov 2022
General comments
- The structural solver is verified by comparison against GXBeam. A brief description of the GXBeam code is recommended before presenting the verification. -> Agreed, a description has been added
- Aeroelastic simulations include coupling between the structural solver and the aerodynamic solver. The present study verified the structural solver comprehensively but did not pay attention to the aerodynamic simulation methods and associated correction. The coupling between the structural solver and aerodynamic method should be briefly presented, though they have been studied in other papers. -> Agreed, more details on the coupling method have been added.
Detailed comments:
- Line 54: the authors mentioned a new unsteady method (RPI). Please briefly clarify this method. -> The RPI method has been clarified
- Page 4: it is recommended to present a figure to show the structural model. Table 2 and Table 3 can be replaced by two equations. Parameters denoted by formulas are also recommended to be presented by equations. -> Tables have been replaced with equations, the formulas changed to numbers where appropriate, and a clearer reference to the structural model figure is added.
- Section 3.1: beams with and without a bend are used for different cases. Please clarify the beam shape (e.g. by figures) to avoid possible confusion. -> Clearer references and more detail is added to highlight the relative figures and distinguish the different analyses.
- The discrepancies shown in Figs. 4-6 are significant. The authors attribute the discrepancies to methods of applying structural damping. Please clearly state the damping coefficient applied in the two codes. Is there any other reason that can cause the discrepancies? For instance, the time marching scheme? -> More detail and clarity added as recommended.
- Section 3.4: control modeling is important for the dynamic behavior of VAWT, it is recommended to give a more detailed description of the modeling of the controller for the normal operating, startup, normal stop and emergency stop conditions. -> Agreed, more detail and clarity was added.
Citation: https://doi.org/10.5194/wes-2022-91-AC1
-
RC2: 'Comment on wes-2022-91', Anonymous Referee #2, 22 Dec 2022
The "References" section in very poor and lacks a large number of papers published in the last 15 years on Darrieus VAWT modeling. The authors in particular are recommended to analyze the works of the research groups of TU Delft (Prof. Ferreira), TU Berlin (Prof. Nayeri) and Florence (Prof. Bianchini).
The aerodynamic part is vaguely described. Please add more on the type/setting of dynamic stall models, virtual camber and virtual incidence correction, pitching moment treatment, post-stall polar extrapolation, struts’ resistant torque and tower shadow.
The description of other modules is also very short and does not allow the reader to understand which kind of results should be expected.
Overall, the impression is that the analysis is presented in a too synthetic fashion. Insight on the results is often missing. The simple “validation” has a scarce scientific interest for the reader if it is not connected with physical insights or with a critical analysis of the models used.
Discrepancies in figures 4-6 are also not negligible and their motivation is quite weak.
The discrepancies seen in Figure 9 are only marginally addressed and would deserve more analysis.
Even more critical is the analysis of Figure 13. Beyond the absolute wind speed values (but Re should be anyway high enough) predicting high AoAs is generally easier since the flow is attached. The results are quite poor and hardly connected only on the way TSR is achieved.
Citation: https://doi.org/10.5194/wes-2022-91-RC2 -
AC2: 'Reply on RC2', Kevin Moore, 08 Jan 2023
The "References" section in very poor and lacks a large number of papers published in the last 15 years on Darrieus VAWT modeling. The authors in particular are recommended to analyze the works of the research groups of TU Delft (Prof. Ferreira), TU Berlin (Prof. Nayeri) and Florence (Prof. Bianchini).
Thank you for your valuable feedback, the intro section has been updated with relevant citations from these suggested authors as well as significant additional clarification on the state of the art and needs for the paper’s primary topic on two-way aero-elastic verification/validation.
The aerodynamic part is vaguely described. Please add more on the type/setting of dynamic stall models, virtual camber and virtual incidence correction, pitching moment treatment, post-stall polar extrapolation, struts’ resistant torque and tower shadow.
Thank you, agreed. Additional descriptions have been added regarding the dynamic stall model type/setting, angle of attack corrections, extrapolation, tower, and strut considerations.
The description of other modules is also very short and does not allow the reader to understand which kind of results should be expected.
Agreed, additional clarification regarding the level of fidelity of the OWENS code, its core modules and brief clarification on expected level of accuracy in addition to better citations pointing to the specific formulation details have been added.
Overall, the impression is that the analysis is presented in a too synthetic fashion. Insight on the results is often missing. The simple “validation” has a scarce scientific interest for the reader if it is not connected with physical insights or with a critical analysis of the models used.
True, the initial verification cases of the simple beam models are synthetic and intentionally brief to match their purpose of building up to the full turbine validation. The OWENS structural solver has not been officially verified for these types of simple cases and were necessary prior to the complex full-turbine unsteady experimental validation. The comparisons with the experimental unsteady strain gauge data later in the paper, though not commented on, are well connected with physical insights and offer a much more detailed critical analysis of the OWENS code and its limitations. This has been clarified and updated in the verification/validation sections.
Discrepancies in figures 4-6 are also not negligible and their motivation is quite weak.
In the surrounding text, we describe that we can manually tune the different damping models in the two different codes to give resulting loads which match within a root mean square of less than 5% for all loads, giving a much better visual result. However, the development of a justifiable relationship between differing damping models was deemed out of scope for the primary objective of experimental validation; the simple setup and results were deemed adequate to show the ability to simulate structural damping of a forced resonate beam. We have revised this section to make this decision clearer and hope it to be satisfactory.
The discrepancies seen in Figure 9 are only marginally addressed and would deserve more analysis.
Agreed, additional detail on the differences and their potential origins have been added to this section including quantification of error for parked and spinning loads between the two simulations and the experimental data.
Even more critical is the analysis of Figure 13. Beyond the absolute wind speed values (but Re should be anyway high enough) predicting high AoAs is generally easier since the flow is attached. The results are quite poor and hardly connected only on the way TSR is achieved.
Thank you, the statement regarding attached flow at high angles of attack is confusing, however, this section has been revised for clarity. The errors associated with the experimental torque sensor at low loads are more clearly discussed in addition to better clarification between the appearance of errors between CP and Torque when viewed for a constant RPM TSR sweep.
Citation: https://doi.org/10.5194/wes-2022-91-AC2
-
AC2: 'Reply on RC2', Kevin Moore, 08 Jan 2023
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 623 | 285 | 38 | 946 | 34 | 31 |
- HTML: 623
- PDF: 285
- XML: 38
- Total: 946
- BibTeX: 34
- EndNote: 31
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 476 | 51 |
| Germany | 2 | 67 | 7 |
| China | 3 | 56 | 6 |
| Ireland | 4 | 46 | 4 |
| Netherlands | 5 | 37 | 3 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 476
