Load Restoration Flexible Optimization in Wind Power Integrated System Based on Conditional Value at Risk
Round 1
Reviewer 1 Report
The authors propose a method of load restoration flexible optimization nn order to better accommodate the uncertainty induced from high penetration of wind power during load recovery.
Questions:
1. The proposed method is not well explained. It is difficult to understand how the restoration will be done. If at every instant the wind regime changes and so does the power generated by the wind farms, how will restoration be carried out if the power generated will always be different? Do synchronous generation plants have their redispatch model changed during the search process for the proposed method? There are many variables that are not defined.
2. Page 6, Line 215: "When making LRS with the proposed method in this paper, Available wind condition must be got in advance." It is not very clear. The wind power plant will always be injecting power into the grid according to its wind regime. Do the authors expect that when the load is restored, the wind regime will change and the wind farm will start to inject more power into the grid? This strategy seems kind of unworkable. It is not possible to control the power generated by wind farms.
3. Should not power flow equations be included in the optimization model? It is necessary to guarantee the existence of a balance point in the system.
4. Does success in load restoration depend on the location of wind farms? The further away from the wind farm, the greater the difficulty of restoring that load? The test system adopted by the authors presents well distributed wind farms, but this may not happen in all power systems. Can this be considered a limiting factor of the proposed method?
5. It is not clear from the case studies which loads need to be restored. Will all system loads be restored at the same rate?
6. The authors indicate research in this area, but do not present comparative results with methods already existing in the literature. It would be interesting to present comparative results in order to assess the weaknesses and strengths of the proposed method.
Some mistakes:
7. Page 1, Line 38: "... redundancy and 2. Materials and Methods recovery ahead ..."
8. Page 5, Line 187: "The object function ..." (The objective function)
9. Page 5, Lines 187 to 192: there are two duplicate paragraphs.
10. Page 7, Line 258: "... are compared in Table . Accordingly, ..."
11. Page 10, Line 322: "Figure 4. it can be seen in the green part ..."
Author Response
Response to Reviewer 1 Comments
We appreciate your careful review of manuscript. The following is the response to comments.
Question 1. The proposed method is not well explained. It is difficult to understand how the restoration will be done. If at every instant the wind regime changes and so does the power generated by the wind farms, how will restoration be carried out if the power generated will always be different? Do synchronous generation plants have their redispatch model changed during the search process for the proposed method? There are many variables that are not defined.
Response 1: In order to explain the proposed method clearly, the contents of part 4 are reorganized. First of all, the primary model of load restoration optimization for making initial load recovery scheme is described in detail. Secondly, the model of load restoration flexible optimization is constructed through loosing the operation constraints with OCSF. Note that the wind power increment at the expected wind speed within the confidence level is adopted to optimize the initial load recovery scheme. In order to deal with the fluctuation of wind power occuring during the scheme is carried out, the upward reserve capacity is reserved by synchronous generators. In the revised version, the constrains of synchronous generator ramping and reserve capacity dispatching are supplemented (constraint 11 and 13). In addition, variables used in the model are explained in detail.
Question 2. Page 6, Line 215: "When making LRS with the proposed method in this paper, Available wind condition must be got in advance." It is not very clear. The wind power plant will always be injecting power into the grid according to its wind regime. Do the authors expect that when the load is restored, the wind regime will change and the wind farm will start to inject more power into the grid? This strategy seems kind of unworkable. It is not possible to control the power generated by wind farms.
Response 2: We are sorry for the difficulty in understanding caused by the improper expression. In the revised version, the requirement of wind power output being qualified for participating in load recovery is explained in detail. The specific contents are: “Before making LRS with the proposed method in this paper, the requirement of wind power output being qualified for participating in load recovery must be confirmed first. In the load recovery phase following the blackout, if the fault occurs again due to improper operation or random disturbance, the load recovery process will be further delayed. Therefore, the scenario of excessive wind speed that may trigger the shutdown of wind power should be avoided in load recovery. Meanwhile, the wind power output less than 10 percent of rated power is suggested not to integrate with the system because it has the limit benefit for recovery. Moreover, the wind power output meeting the above output range should be able to last for a period of time which is the minimum time guarantee for load recovery, such as 2 hours. After that the predicted wind conditions in the next 2-3 hours are confirmed in terms of above requirements, the LRS can be formulated according to the following flow.”
Question 3. Should not power flow equations be included in the optimization model? It is necessary to guarantee the existence of a balance point in the system.
Response 3: Yes, power flow equations should be included in the optimization model. The load recovery operation is completed step by step. Based on a balance point of system operation achieved in the previous step, the proposed optimization method is to identify the optimal load location recovered and corresponding load amount picked up in current step. In order to facilitate the solution of the model, sensitivity analysis is employed to directly associate network security constraints with bus injection increments. In the original manuscript, the balance constraints of bus injection increment were carelessly omitted, and now they have been supplemented completely. Once again, thanks for pointing out our omissions.
Question 4. Does success in load restoration depend on the location of wind farms? The further away from the wind farm, the greater the difficulty of restoring that load? The test system adopted by the authors presents well distributed wind farms, but this may not happen in all power systems. Can this be considered a limiting factor of the proposed method?
Response 4: The success of load restoration depends on reasonable dispatching of the power both from the syschronous generators and wind farms. In other words, the power generated by wind farms is subject to unified dispatching together with the power generated by other syschronous units in terms of maximizing the value of load recovered in each step. Due to actual operation constraints such as transmission capacity and bus voltage, the more uneven distribution of power sources and loads, the greater the difficulty of optimal dispatching, which is common to optimization of load restoration and will not be a limiting factor of the proposed method. As far as the test system in this paper is concerned, the adjustment of wind farms location will further change the power flow when picking up loads, thus the situation in which the operation constraints exceed the limits and the use of OCSF may be different from the existing results.
- It is not clear from the case studies which loads need to be restored. Will all system loads be restored at the same rate?
Response 5: The details of loads need to be restored are supplemented in Table 1 in the revised version. The speed of picking up load is up to the comprehensive weight of the load. On the whole, the higher the comprehensive importance of nodes, the more likely they are to be recovered first. As for a specific node, the higher priority part of the load will be recovered first, which has been described in part 4.1 through Formula (7), namely the comprehensive weight of load node.
- The authors indicate research in this area, but do not present comparative results with methods already existing in the literature. It would be interesting to present comparative results in order to assess the weaknesses and strengths of the proposed method.
Response 6: A new part, “5.4. Comparison of of FLRS with LRS obtained from other optimization methods”, is presented in the revised version. First of all, load recovery results from chance constrained programming (CCP) model and robust optimization (RO) model are listed in Table 6 as well as the results from flexible optimization (FO) proposed in this paper. Secondly, the different ways of dealing with the wind power uncertainty in CCP, RO and FO are explained in details as the main reason for different restoration results. Further, the load recovery speed of FO model and CCP model in different stages are compared and analyzed.
Some mistakes:
- Page 1, Line 38: "... redundancy and 2. Materials and Methods recovery ahead ..."
Response 7: We are sorry for the clerical error caused by carelessness. In the revised version, the error has been corrected. The whole sentence is: “For the electrical power system with resilience, or resilient system, it should be qualified with key features of resistance, redundancy and recovery ahead, during and post the event respectively, which means the system is resistant to withstand initial shock, being capable to minimize the damage for the duration, and self-organized to normal state as soon as possible post event”.
- Page 5, Line 187: "The object function ..." (The objective function)
Response 8: In the revised version, the error has been corrected. In part 4.1, the expression is revised to: “Generally, the objective of load restoration optimization is to maximize the value of load recovered, which is expressed by (6).”
- Page 5, Lines 187 to 192: there are two duplicate paragraphs.
Response 9: Fig. 4 is used to show the recovery effect of FLRSs generated at different levels of wind power penetration. In order to show the amount of load recovery more clearly, two different perspectives, namely perspective A and perspective B, are provided at the same time. In the revised version, the subheadings for these two perspectives are supplemented.
- Page 7, Line 258: "... are compared in Table . Accordingly, ..."
Response 10: In the revised version, the number of table has been added. Before Table 2, there is the following description: “The amounts of load picked up corresponding to ILRS and FLRS are compared in Table 2.”
- Page 10, Line 322: "Figure 4. it can be seen in the green part ..."
Response 11: In the revised version, the clarity of description is improved. Table 4 is followed by the following description: “From the green part of Fig. 4(a), it can be seen that the amount of load recovery in the early recovery period decreases slightly with wind power penetration rate in-creasing. …… From the blue part of Fig. 4(b), it can be seen in that when the wind power penetration rate increases from 0.1% to 19%, the relaxation of transmission capacity and unit out-put constraint boundary accelerates the load recovery.”
Reviewer 2 Report
1) in the introduction
"redundancy and 2. Materials and 38 Methods recovery ahead"
please explain this
2) Introduction include the comma sign.
In addition"," to infrastructure rein- 43 forcement, optimizing the strategy of dispatching and operation has been recognized a 44 preferred choice for improving the restorability of the existing grid [6, 7...
3) Please improve the references, they are too old, which may lead to thinking your work is out of date.
Author Response
Response to Reviewer 2 Comments
We appreciate your careful review of manuscript. The following is the response to comments.
1) in the introduction
"redundancy and 2. Materials and 38 Methods recovery ahead" please explain this.
Response: We are sorry for the clerical error caused by carelessness. In revised version, the error has been corrected. The whole sentence is: “For the electrical power system with resilience, or resilient system, it should be qualified with key features of resistance, redundancy and recovery ahead, during and post the event respectively, which means the system is resistant to withstand initial shock, being capable to minimize the damage for the duration, and self-organized to normal state as soon as possible post event”.
2) Introduction include the comma sign.
In addition"," to infrastructure rein- 43 forcement, optimizing the strategy of dispatching and operation has been recognized a 44 preferred choice for improving the restorability of the existing grid [6, 7...
Response: In the revised version, the description is changed to: “Besides infrastructure reinforcement, optimizing the strategy of dispatching and operation has been recognized a preferred choice for improving the restorability of the existing grid [6, 7].”
3) Please improve the references, they are too old, which may lead to thinking your work is out of date.
Response: In the revised version, Some literatures have been updated.
Reviewer 3 Report
The work is well done and the bibliography is well described. One already understands the purpose of the work from the introduction.
The non-predictability of wind is mentioned, but no information is given on how a possible unavailability of a generator is evaluated.
Only the reliability of the system in general is mentioned, but not the machines. In a work such as this, the impact of the unavailability of generators should be considered, this should at least be linked to a minimum study on maintenance, periodic verification and the resulting reliability.
The effects of mains faults should at least be commented on and integrated in this work, at least from a statistical point of view.
The conclusion is clear, the work is valid without this recommended small integration.
Author Response
Response to Reviewer 3 Comments
The work is well done and the bibliography is well described. One already understands the purpose of the work from the introduction.
The non-predictability of wind is mentioned, but no information is given on how a possible unavailability of a generator is evaluated.
Only the reliability of the system in general is mentioned, but not the machines. In a work such as this, the impact of the unavailability of generators should be considered, this should at least be linked to a minimum study on maintenance, periodic verification and the resulting reliability.
The effects of mains faults should at least be commented on and integrated in this work, at least from a statistical point of view.
The conclusion is clear, the work is valid without this recommended small integration.
Response: We appreciate your constructive suggestions. As for the research of wind power on the resilience of load recovery scheme following blackout, not only the fluctuation of wind output but also the impact of unit availability should be considered, which will make the load recovery scheme more adaptable to extreme adverse operating conditions. However, due to the complexity of modeling units’ avalability and the limit length of this article, related contents have not yet been reflected but been included in the next modeling improvement plan in the conclusion of the revised version.
Round 2
Reviewer 1 Report
The authors propose a method of load restoration flexible optimization nn order to better accommodate the uncertainty induced from high penetration of wind power during load recovery.
The article has been improved, the contribution is good and all questions have been effectively answered.
Author Response
We appreciate your careful review and comments.
According to your opinion, we have comprehensively checked the organization of the content and the spelling of words. In this revised version，the part of nomenclature and disadvantages of the proposed method are supplemented at the beginning and the end respectively. Meanwhile, Fig. 1 is redrawn and relevant descriptions have been modified in order to demonstrate the flow of optimization method more clearly.
