Smart-Grid-Aware Load Regulation of Multiple Datacenters towards the Variable Generation of Renewable Energy

Round  1

Reviewer 1 Report

 How to control the power consumption of a datacenter and it’s response and effects should be described in detail. 

line 135-137 should be modified, since battery storage systems can effectively increase grid stability.

Wind power is not simulated in your experiment. About the description of wind power should be deleted.

It is not possible to use only NN for forecasting of sunny solar power generation. The rainy weather conditions also need to be simulated, analyzed and compared to verify the feasibility of the proposed method.

In a smart grid, grid stability, demand response, power management, etc., far more important than reduce power loss.

Author Response

How to control the power consumption of a datacenter and it’s response and effects should be described in detail. 

According to the comments and suggestions, we have added necessary references including our own prior work in Section 2.2 to describe how to adjust and control the power consumption of the datacenter and its response and effects.

line 135-137 should be modified, since battery storage systems can effectively increase grid stability.

Thanks for the comments and suggestions. We have modified the sentences and added some necessary relevant references about battery storage and also other broader systems in Section 2.3.

Wind power is not simulated in your experiment. About the description of wind power should be deleted.

Thanks. We have deleted the descriptions about wind power and the model in Section 3.2.

It is not possible to use only NN for forecasting of sunny solar power generation. The rainy weather conditions also need to be simulated, analyzed and compared to verify the feasibility of the proposed method.

Thanks for the comments and suggestions. Accordingly, we have also conducted experiments under the cloudy/snowy weather conditions, analyzed the results comparing with the sunny condition. Results of power losses under accurate responses are illustrated in Section 5.2 and the results of improvement based on forecasting methods and the examination of the time cost are discussed together in Section 5.4. The extended results show that out proposed method is feasible for both sunny and rainy weather conditions. 

In a smart grid, grid stability, demand response, power management, etc., far more important than reduce power loss.

Thanks for the comments. Indeed, we keep in mind that grid stability, demand response, power management are all important issues in the smart grid. During our experiments, in fact we consider the grid stability as well. In the modified manuscript, we added the corresponding description as constraints in the problem formulation part into Section 3.4, and all the results were obtained under the premise of ensuring the stability of the power grid, specifically, avoiding voltage violation and line volume overloading when generating possible solutions. In addition, our method has show the potential of datacenters participating in the demand response of the smart grid. We had done some preliminary work about design and implementation of datacenter demand response strategies, as introduced in Section 2.2, and now we are still focusing on the combination of multiple kinds of energy adjustment approaches and doing some studies about demand response and power management issues from the view of datacenters. We have put the relevant discussion into Section 6 as the future work. 

At the same time, we are considering that since the power losses also have impact on the transmission efficiency of the electricity grid, the reduction of losses is still meaningful for avoiding extra wastes and capital cost for the grid. To make it clearer, we have added some corresponding references stating the importance of electric power transmission and distribution losses into Section 1.

Finally, we want to summarize major revisions and extensions made based on the comments and suggestions of the reviewers, including:

(1)  adding relevant references as suggested about other options to maintain the grid stability, while discussing the importance of electric power transmission and distribution losses in the introduction;

(2)  adding a number of relevant references including our previous work about the methods to control the power consumption of a datacenter and the response and effects in Section 2.2;

(3)  providing some relevant references about battery storage and other broader systems and discussing their limitations in Section 2.3;

(4)  deleting the unnecessary wind generation model in Section 3.2, as suggested;

(5)  clarifying the bus voltage and branch power constraints in Section 3.4, which states the stability requirements of the problem we defined and solved; 

(6)  adding more experiments under the cloudy/snowy weather condition in Section 5.2;

(7)  adding the forecasting experiments under the cloudy/rainy/snowy weather condition, and obtaining different forecasting results under different forecasting accuracy and the corresponding loss in Section 5.4;

(8)  discussing about the relationship of forecasting accuracy and the time cost in Section 5.4.

Author Response File: Author Response.pdf

Reviewer 2 Report

This is an interesting paper analyzing how to use datacenters active in the integration of renewable power production.

I recommend for publication after considering the following suggestions for improvements:

1.

Several times, the paper refer to that datacenters can actively adjust their load, but there is not really any refs to this claim. Since it is essential for the whole idea of the paper, I suggest to explain better and to include some kind of refs and/or justifications…

2.

On page 3 lines 130-140 is mentioned the inclusion of batteries and hybrid systems to integrate and balance renewable energy into the power system. However, a broader systems approach including more sectors than the power sector will allow for better solutions, see e.g.

See e.g.: Smart energy and smart energy systems. Energy, May 2017.

I suggest to include a discussion of the consequences of such limitation of the study in the introduction, discussion and conclusions. This also means something for the use of storage:

Energy Storage and Smart Energy Systems, International Journal of Sustainable Energy Planning and Management Vol. 11, March 2016.

3.

In the introduction (page 1, line 39-41) batteries are mentioned as being potential provides of grid stability. However, many more options exists and was already introduced several years ago, see eg.

Lund, H. Electric grid stability and the design of sustainable energy systems. International Journal of Sustainable Energy, Vol 24 (1), pp. 45-54, March 2005.

4.

Since the paper is an expanded version of a paper in press I suggest to double check that there will not be too much overlap in the text..

Please note that the above mentioned litt´s are only examples. Include only what you find relevant. The main point is to relate and discuss better to previous findings and state-of-the-art.

Apart from that it is a really fine paper with interesting results and relevant methodologies and tools.

Author Response

Several times, the paper refer to that datacenters can actively adjust their load, but there is not really any refs to this claim. Since it is essential for the whole idea of the paper, I suggest to explain better and to include some kind of refs and/or justifications…

Thanks for the comments and suggestions. Accordingly,we have added somereferences including our own prior work about how the datacenter adjust their loads and actively control its power consumption towards certain targets. The relevant contents are discussed in Section 2.2. 

On page 3 lines 130-140 is mentioned the inclusion of batteries and hybrid systems to integrate and balance renewable energy into the power system. However, a broader systems approach including more sectors than the power sector will allow for better solutions, see e.g.

I suggest to include a discussion of the consequences of such limitation of the study in the introduction, discussion and conclusions. This also means something for the use of storage:

Thanks for the comments and suggestions. We have modified the sentences and provided some relevant references about battery storage and other broader systems in Section 2.3. Besides, discussion of their limitations have also been included in the modified manuscript.

In the introduction (page 1, line 39-41) batteries are mentioned as being potential provides of grid stability. However, many more options exists and was already introduced several years ago.

Thanks. As suggested, some references about other options to maintain the grid stability have been added into the introduction (Section 1). 

Since the paper is an expanded version of a paper in press I suggest to double check that there will not be too much overlap in the text..

Thanks. In terms of text length, the current version has been expanded by more than 105% compared to the last published one.

Specifically, compared to the conference paper, we have made considerable additions and modifications in the current manuscript, including plenty of references about the background and basis of this paper, experiments under the cloudy/rainy weather condition, explorations of three-datacenter scenarios which verify the effectiveness of the optimal load allocation strategy, the whole Section 5.4 about the results of forecasting under different accuracy, and the examination of the relationship between forecasting accuracy and the time cost.Furthermore, we have double checked the words and sentences throughout the whole manuscript, in order to correct the mistakes and avoid text overlaps with the original paper as much as possible.

Finally, we want to summarize major revisions and extensions made based on the comments and suggestions of the reviewers, including:

(1)  adding relevant references as suggested about other options to maintain the grid stability, while discussing the importance of electric power transmission and distribution losses in the introduction;

(2)  adding a number of relevant references including our previous work about the methods to control the power consumption of a datacenter and the response and effects in Section 2.2;

(3)  providing some relevant references about battery storage and other broader systems and discussing their limitations in Section 2.3;

(4)  deleting the unnecessary wind generation model in Section 3.2, as suggested;

(5)  clarifying the bus voltage and branch power constraints in Section 3.4, which states the stability requirements of the problem we defined and solved; 

(6)  adding more experiments under the cloudy/snowy weather condition in Section 5.2;

(7)  adding the forecasting experiments under the cloudy/rainy/snowy weather condition, and obtaining different forecasting results under different forecasting accuracy and the corresponding loss in Section 5.4;

(8)  discussing about the relationship of forecasting accuracy and the time cost in Section 5.4.

Author Response File: Author Response.pdf

Round  2

Reviewer 1 Report

NA

Author Response

Thanks.

According to the comments, we have done extensive English editing sentence by sentence, including checking mistakes of grammar/spelling/punctuation and reorganizing some words. The detailed revisions of the current manuscript can be displayed by the "Track changes" function in Microsoft Word.

Reviewer 2 Report

Please check ref 56, I belive it should be

H Lund, PA Østergaard, D Connolly, I Ridjan, BV Mathiesen, F Hvelplund, JZ Thellufsen, P Sorknæs: Energy Storage and Smart Energy Systems, International Journal of Sustainable Energy Planning and Management Vol. 11, March 2016.

Author Response

Thanks for the comments.

We have revised the format of ref 56 and in addition, we have also completed the necessary information of ref 54 and ref 55.

Besides, we have checked the spelling/grammar mistakes and typos throughout the whole paper sentence by sentence, as displayed by "Track changes" in the revise version.