the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 26 Apr 2024
Analysis of a 30 GW offshore wind power scenario in Norway using time-series computed from numerical weather model data
Abstract. This article investigates implications of integrating 30 GW offshore wind in Norway. Wind power time series for relevant locations are analysed using 30 years of hourly numerical weather model reanalysis data. The study presents key statistical properties of the wind power time series. The emphasis lies on correlation, geographical smoothing, and variability across different time scales. These findings hold significant relevance for the strategic planning of offshore wind farm development, and for effectively preparing the energy system to accommodate this extensive wind power deployment that would mean a doubling of the Norwegian electricity generation.
This preprint has been withdrawn.
-
Withdrawal notice
This preprint has been withdrawn.
-
Preprint
(1472 KB)
Interactive discussion
Status: closed
-
RC1: 'Comment on wes-2024-45', Anonymous Referee #1, 23 May 2024
The paper asks interesting questions, and, at a high level, the methodology is fine, but this paper could have been so, so much better if the authors had not made some very weakly justified decisions. It would not have been overly difficult to do things properly, and it is not at all clear why the authors have not done so. 20-30 years ago a similar type of paper would have been written for offshore wind, and, at that time, perhaps some generosity could have been given regarding the technical rigour of such a paper, but it seems very difficult to offer such generosity nowadays, given large advances in knowledge and understanding.
The justification for the wind farm power curve (Figure 2) is very weak, which undermines everything else that follows in the paper. The paper also focuses on offshore wind power variability, which is important, but only when considered alongside onshore wind and solar PV and hydro variability, and perhaps considering net demand variability. Some acknowledgement of where the offshore wind farms are located, and proximity to the electrical grid, and major load centres, is not mentioned, even in passing, with potential implications for the likelihood of certain offshore wind farms being built, and their relative size.
Figure 1 – rather strange to start numbering at 0 – why not 1? Start numbering at 1 for Figure 1 and later correlation charts
Figure 1 – figure title implies 2 non-Norway wind farms rather than 3
There isn’t any comment on whether sub-hourly (rather than hourly) modelling should really have been implemented. Given that 30 GW of offshore wind is being discussed, sub-hourly ramp rates could well be a challenge.
It is acknowledged that the wind farm power curve is crude, so why does it not follow that the results (particularly in 20-30 m/s speed range) are not crude? It would help to tease things out further, and comment on the likelihood of occurrence of different wind speeds. Given that the interest here is in variability, plotting the derivative of the two power curves would also be informative – further highlighting the weakness of the adopted approach.
It is not clear why the authors didn’t take some “off the shelf” wind farm power curve, and use that as the basis for their later analysis. Surely, that would not have been difficult to do, and would have made the results more plausible?
Fig. 3 – a few days is very short, and doesn’t support the choice of wind farm power curve. It would also be helpful to give quantitative values for hourly variability – the model output seems to be a lot smoother than the actual wind farm output
While 30 GW of offshore wind is of interest, what really is of interest is the net demand variability, implying that onshore wind, solar PV and hydro also need to be included. Not clear what the value is in looking at offshore wind in isolation?
Fig.6 – some lines are green, and some are blue – what is the difference?
Fig.s 7-9 – it would be helpful if the wording for the figure titles (notably Fig. 7) followed the same format, making it easier to quickly work out what the difference was between the 3 figures
As the authors note % ramps reduce, relative to capacity, with increasing offshore wind farms, but what the power system cares about is the absolute (MW/hr) ramps, and the implications for the “balancing” plant – so, not sure that % and pu values are actually that informative
Section 5.1 – wind ramps are bigger than demand ramps, and hence what? The authors should really discuss the implications of this observation
Fig.s. 13 and 14 have the same title – make the titles (slightly) different, so that the 2 cases can be easily distinguished
For the various wind farm locations there is no mention (even in passing) of grid connection aspects, closeness to load centres, and hence linking with the likelihood that all offshore wind farms will actually be built and that they will be of the same size.
Fig. 18 – what does 1 pu mean?
As a general comment, the authors have chosen to use pu values rather than MW/GW values. There are advantages in using pu quantities, but is it actually helpful here, particularly given that the base values keep changing from figure to figure? By quoting values in MW, and perhaps assuming hydro/thermal units of 500 MW (just as an example) hourly ramps and inter-annual variations can be conveniently converted into how many units need to be switched on/off to counteract offshore wind variability ... just a thought. It is not easy to convert the pu values quoted in the paper into something meaningful.
Are Fig. 19 and 20 using the same base value – can the figures be directly compared?
Given the future timescales involved, what about climate change impacts on the available wind resource?
Citation: https://doi.org/10.5194/wes-2024-45-RC1 -
RC2: 'Comment on wes-2024-45', Anonymous Referee #2, 04 Jun 2024
The paper presents an analysis of a 30 GW offshore wind power scenario in Norway using the MERRA 2 reanalysis data in the Renewables.Ninja environment with a virtual wind farm (VWF) model. Here are the major issues:
ISSUE 1.)
pp. 2, line 40: "The main novelty of this study is the scenario that is being considered: 30 GW offshore wind in Norway in line with government plans, and located according to recently published areas for potential offshore wind development in Norway. Although the time series analysis methods are well-known, we believe the results to be of wider interest, with supporting material (Svendsen, 2023) that includes Python code used for downloading data and creating all the figures presented in this article."
– It is implicitly stated that the present paper is a technical report that has no novelty in terms of methodology, concepts, ideas, analyses, or data, and it states that the "time series analysis methods are well-known".
– The claimed novelty of the paper is the scenario itself, which is perceived to be too incremental for this journal when compared to other cited papers on the same topic, such as Solbrekke, Kvamstø and Sorteberg (2020), Hjelmeland and Nøland (2023), etc.ISSUE 2.)
pp. 16, line 210: Comparing Figure 19 and Figure 20 we see that the seasonal variation of demand and offshore wind power output are very similar when considering their average weekly values. This is a good thing. The variability within a given week of the year, however, is significantly different: Whereas demand varies within a fairly narrow band around the mean value, we see that wind power can have essentially any value between zero and full output."
– What the paper is showing is not new compared to what the other cited papers are showing. Figure 19 highlights the fact that the minimum guaranteed power is low. Consequently, the power adequacy problem cannot be solved by "the seasonal pattern of wind power matches very well with the power consumption", as stated in pp. 16, line 2015. Moreover, the seasonal energy profile in Figure 21 does not address the power adequacy problem either. Overall, the paper makes no attempt to solve this problem in a better way than in the earlier references other than stating that "the need for balancing on shorter time scales will increase due to wind power variability", which is unfortunately not enough to address the problem.ISSUE 3.)
Finally, the conclusion states that the seasonal energy output of Norwegian offshore wind matches well with the seasonal variation in power demand, which is not really a new finding compared to the other references. It is also emphasized that more geographically distributed buildouts, including mid-Norway and Northern locations, have benefits, which is not really a new finding compared to the existing findings in the cited literature. As a result, the present paper does not bring anything new beyond the state-of-the-art (SotA).MINOR ISSUE:
The ERA5 data are clearly the best reanalysis product, but the present paper use the MERRA2 reanalysis data which has to extrapolate the data to get the right inputs. It is not clear to what extent this simplification effect the analysis.Citation: https://doi.org/10.5194/wes-2024-45-RC2 - RC3: 'Comment on wes-2024-45', Anonymous Referee #3, 10 Jun 2024
-
AC1: 'Comment on wes-2024-45', Harald G. Svendsen, 05 Jul 2024
Thank you to all the reviewers for insightful comments to our manuscript. We would like to address these in an updated version of the paper, but need some weeks to do this. Together with a revised version we will also repsond to each of the specific comments.
On a general level we understand the reviewers’ concern about novelty, and that our paper brings little new compared to existing literature, such as “Solbrekke, Kvamstø and Sorteberg (2020), Hjelmeland and Nøland (2023), etc”. We do not agree with this. As stated in the paper, Solbrekke et al. discuss onshore wind power only, and Hjelmeland/Nøland focus on correlations and optimisation of wind power capacity locations. That latter paper uses outdated offshore wind sites for Norway, and MERRA2 data only. It could well be that their paper was in part inspired by our work, as it cites a project memo (https://sintef.brage.unit.no/sintef-xmlui/bitstream/handle/11250/3054972/Norway_30_GW_wind_v1_signed.pdf) we wrote with early results.
In any case, we would like to improve the novelty of the manuscript in an updated version where we propose to:
- Repeat the analysis with different underlying weather data, at least ERA5 and maybe others. Our hypothesis is that this makes little differene, but that needs to be investigated.
- Repeat the analysis with different wind farm power curves. Again our hypothesis is that this does not make a big difference, but needs to be investigated.
- Expanded discussion on implications for the power system
- Additional modifications that address the various detailed comments
Regarding the comment about whether offshore wind is interesting to study in its own – versus studying net demand, we believe both are intereting. The fact that net-demand is interesting does not mean that understanding offshore wind power charachteristics in isolation is not. And for a Norwegian scenario with 30 GW offshore wind, this will by far dominate over onshore wind and solar power.
Citation: https://doi.org/10.5194/wes-2024-45-AC1
Interactive discussion
Status: closed
-
RC1: 'Comment on wes-2024-45', Anonymous Referee #1, 23 May 2024
The paper asks interesting questions, and, at a high level, the methodology is fine, but this paper could have been so, so much better if the authors had not made some very weakly justified decisions. It would not have been overly difficult to do things properly, and it is not at all clear why the authors have not done so. 20-30 years ago a similar type of paper would have been written for offshore wind, and, at that time, perhaps some generosity could have been given regarding the technical rigour of such a paper, but it seems very difficult to offer such generosity nowadays, given large advances in knowledge and understanding.
The justification for the wind farm power curve (Figure 2) is very weak, which undermines everything else that follows in the paper. The paper also focuses on offshore wind power variability, which is important, but only when considered alongside onshore wind and solar PV and hydro variability, and perhaps considering net demand variability. Some acknowledgement of where the offshore wind farms are located, and proximity to the electrical grid, and major load centres, is not mentioned, even in passing, with potential implications for the likelihood of certain offshore wind farms being built, and their relative size.
Figure 1 – rather strange to start numbering at 0 – why not 1? Start numbering at 1 for Figure 1 and later correlation charts
Figure 1 – figure title implies 2 non-Norway wind farms rather than 3
There isn’t any comment on whether sub-hourly (rather than hourly) modelling should really have been implemented. Given that 30 GW of offshore wind is being discussed, sub-hourly ramp rates could well be a challenge.
It is acknowledged that the wind farm power curve is crude, so why does it not follow that the results (particularly in 20-30 m/s speed range) are not crude? It would help to tease things out further, and comment on the likelihood of occurrence of different wind speeds. Given that the interest here is in variability, plotting the derivative of the two power curves would also be informative – further highlighting the weakness of the adopted approach.
It is not clear why the authors didn’t take some “off the shelf” wind farm power curve, and use that as the basis for their later analysis. Surely, that would not have been difficult to do, and would have made the results more plausible?
Fig. 3 – a few days is very short, and doesn’t support the choice of wind farm power curve. It would also be helpful to give quantitative values for hourly variability – the model output seems to be a lot smoother than the actual wind farm output
While 30 GW of offshore wind is of interest, what really is of interest is the net demand variability, implying that onshore wind, solar PV and hydro also need to be included. Not clear what the value is in looking at offshore wind in isolation?
Fig.6 – some lines are green, and some are blue – what is the difference?
Fig.s 7-9 – it would be helpful if the wording for the figure titles (notably Fig. 7) followed the same format, making it easier to quickly work out what the difference was between the 3 figures
As the authors note % ramps reduce, relative to capacity, with increasing offshore wind farms, but what the power system cares about is the absolute (MW/hr) ramps, and the implications for the “balancing” plant – so, not sure that % and pu values are actually that informative
Section 5.1 – wind ramps are bigger than demand ramps, and hence what? The authors should really discuss the implications of this observation
Fig.s. 13 and 14 have the same title – make the titles (slightly) different, so that the 2 cases can be easily distinguished
For the various wind farm locations there is no mention (even in passing) of grid connection aspects, closeness to load centres, and hence linking with the likelihood that all offshore wind farms will actually be built and that they will be of the same size.
Fig. 18 – what does 1 pu mean?
As a general comment, the authors have chosen to use pu values rather than MW/GW values. There are advantages in using pu quantities, but is it actually helpful here, particularly given that the base values keep changing from figure to figure? By quoting values in MW, and perhaps assuming hydro/thermal units of 500 MW (just as an example) hourly ramps and inter-annual variations can be conveniently converted into how many units need to be switched on/off to counteract offshore wind variability ... just a thought. It is not easy to convert the pu values quoted in the paper into something meaningful.
Are Fig. 19 and 20 using the same base value – can the figures be directly compared?
Given the future timescales involved, what about climate change impacts on the available wind resource?
Citation: https://doi.org/10.5194/wes-2024-45-RC1 -
RC2: 'Comment on wes-2024-45', Anonymous Referee #2, 04 Jun 2024
The paper presents an analysis of a 30 GW offshore wind power scenario in Norway using the MERRA 2 reanalysis data in the Renewables.Ninja environment with a virtual wind farm (VWF) model. Here are the major issues:
ISSUE 1.)
pp. 2, line 40: "The main novelty of this study is the scenario that is being considered: 30 GW offshore wind in Norway in line with government plans, and located according to recently published areas for potential offshore wind development in Norway. Although the time series analysis methods are well-known, we believe the results to be of wider interest, with supporting material (Svendsen, 2023) that includes Python code used for downloading data and creating all the figures presented in this article."
– It is implicitly stated that the present paper is a technical report that has no novelty in terms of methodology, concepts, ideas, analyses, or data, and it states that the "time series analysis methods are well-known".
– The claimed novelty of the paper is the scenario itself, which is perceived to be too incremental for this journal when compared to other cited papers on the same topic, such as Solbrekke, Kvamstø and Sorteberg (2020), Hjelmeland and Nøland (2023), etc.ISSUE 2.)
pp. 16, line 210: Comparing Figure 19 and Figure 20 we see that the seasonal variation of demand and offshore wind power output are very similar when considering their average weekly values. This is a good thing. The variability within a given week of the year, however, is significantly different: Whereas demand varies within a fairly narrow band around the mean value, we see that wind power can have essentially any value between zero and full output."
– What the paper is showing is not new compared to what the other cited papers are showing. Figure 19 highlights the fact that the minimum guaranteed power is low. Consequently, the power adequacy problem cannot be solved by "the seasonal pattern of wind power matches very well with the power consumption", as stated in pp. 16, line 2015. Moreover, the seasonal energy profile in Figure 21 does not address the power adequacy problem either. Overall, the paper makes no attempt to solve this problem in a better way than in the earlier references other than stating that "the need for balancing on shorter time scales will increase due to wind power variability", which is unfortunately not enough to address the problem.ISSUE 3.)
Finally, the conclusion states that the seasonal energy output of Norwegian offshore wind matches well with the seasonal variation in power demand, which is not really a new finding compared to the other references. It is also emphasized that more geographically distributed buildouts, including mid-Norway and Northern locations, have benefits, which is not really a new finding compared to the existing findings in the cited literature. As a result, the present paper does not bring anything new beyond the state-of-the-art (SotA).MINOR ISSUE:
The ERA5 data are clearly the best reanalysis product, but the present paper use the MERRA2 reanalysis data which has to extrapolate the data to get the right inputs. It is not clear to what extent this simplification effect the analysis.Citation: https://doi.org/10.5194/wes-2024-45-RC2 - RC3: 'Comment on wes-2024-45', Anonymous Referee #3, 10 Jun 2024
-
AC1: 'Comment on wes-2024-45', Harald G. Svendsen, 05 Jul 2024
Thank you to all the reviewers for insightful comments to our manuscript. We would like to address these in an updated version of the paper, but need some weeks to do this. Together with a revised version we will also repsond to each of the specific comments.
On a general level we understand the reviewers’ concern about novelty, and that our paper brings little new compared to existing literature, such as “Solbrekke, Kvamstø and Sorteberg (2020), Hjelmeland and Nøland (2023), etc”. We do not agree with this. As stated in the paper, Solbrekke et al. discuss onshore wind power only, and Hjelmeland/Nøland focus on correlations and optimisation of wind power capacity locations. That latter paper uses outdated offshore wind sites for Norway, and MERRA2 data only. It could well be that their paper was in part inspired by our work, as it cites a project memo (https://sintef.brage.unit.no/sintef-xmlui/bitstream/handle/11250/3054972/Norway_30_GW_wind_v1_signed.pdf) we wrote with early results.
In any case, we would like to improve the novelty of the manuscript in an updated version where we propose to:
- Repeat the analysis with different underlying weather data, at least ERA5 and maybe others. Our hypothesis is that this makes little differene, but that needs to be investigated.
- Repeat the analysis with different wind farm power curves. Again our hypothesis is that this does not make a big difference, but needs to be investigated.
- Expanded discussion on implications for the power system
- Additional modifications that address the various detailed comments
Regarding the comment about whether offshore wind is interesting to study in its own – versus studying net demand, we believe both are intereting. The fact that net-demand is interesting does not mean that understanding offshore wind power charachteristics in isolation is not. And for a Norwegian scenario with 30 GW offshore wind, this will by far dominate over onshore wind and solar power.
Citation: https://doi.org/10.5194/wes-2024-45-AC1
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 350 | 101 | 26 | 477 | 17 | 14 |
- HTML: 350
- PDF: 101
- XML: 26
- Total: 477
- BibTeX: 17
- EndNote: 14
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1