Chance-Constrained Based Voltage Control Framework to Deal with Model Uncertainties in MV Distribution Systems
Round 1
Reviewer 1 Report
Chance-constrained based voltage control framework to deal 2 with model uncertainties in MV distribution systems
14 Please rephrase: Then, model uncertainty impacts on the above linear optimization formulation are evaluated.
15 Please rephrase: This analysis defines that the voltage control problem subject to model uncertainties constitutes a joint CC formulation.
Please rehprase:
27 They are not basically prepared for incorporating intermittent powers of distributed generation (DG) units, which boost the short-circuit power, create bidirectional power flows, and lead to voltage rise issues [1]
What is the role of System identification and AI approaches to modelling?
119 What guarantees convexity? (having a convex optimization
Please provide algorithmic representation of the following procedure:
nts of the transformer tap changer, respectively. 145 The voltage sensitivity data needed in (6) are obtained as follows. The coefficients 146 giving the voltage sensitivity with respect to active or reactive power changes of DGs can 147 be extracted from the inverse Jacobian matrix in the Newton-Raphson load flow (NRLF) 148 study. The sensitivity of voltages with respect to the (substation) transformer tap changes 149 is obtained using the perturb-and-observe method from the results of two consecutive 150 load flow studies subject to one step change of the transformer tap position [2].
Please improve desription of Figure 3:
373 As it can be seen in figure 3
451 Please explain: disregarding the model uncertainties (= 6.044).
Validation should be provided on real data.
Aporopriate number of real world situations should be compared to Monte carlo simulation. What is appropriate the statistial test should tell.
Author Response
Response to the reviewer’s comments regarding submission energies-1280782 entitled:
‘‘Chance-constrained based voltage control framework to deal with model uncertainties in MV distribution systems’’
By: Bashir Bakhshideh Zad, Jean-François Toubeau, and François Vallée
Introduction: The authors would like to appreciate the reviewer for giving the following valuable comments, thanks to which we could improve the quality and readability of the paper. The current document presents a point-by-point response to the reviewer’s comments. Also, according to the given comments and suggestions, the paper has been revised and resubmitted. The authors hope that the current revision of the paper would adequately address the questions and concerns regarding the previous submission. All the changes and modifications relating to the raised comments are applied to the revised paper in the track mode for a better visibility.
Reviewer’s comments
Chance-constrained based voltage control framework to deal 2 with model uncertainties in MV distribution systems
Comment 1: 14 Please rephrase: Then, model uncertainty impacts on the above linear optimization formulation are evaluated.
Comment 2: 15 Please rephrase: This analysis defines that the voltage control problem subject to model uncertainties constitutes a joint CC formulation.
Comment 3: Please rehprase: 27 They are not basically prepared for incorporating intermittent powers of distributed generation (DG) units, which boost the short-circuit power, create bidirectional power flows, and lead to voltage rise issues [1]
Response to comments 1 to 3: The requested changes have been applied and the mentioned sentences have been rephrased as follows:
“Then, model uncertainty impacts on the above linear optimization problem are evaluated. This analysis defines that the voltage control problem subject to model uncertainties should be modelled with a joint CC formulation”
“They are not prepared to host the distributed generation (DG) units, which boost the short-circuit power, create bidirectional power flows, and induce voltage rise issues”
Comment 4: What is the role of System identification and AI approaches to modelling?
Response to comments 4: The authors would like to appreciate the reviewer for this insightful comment. The modelling procedure of electric power system components usually takes advantage of measurements, simulations, and analytical approaches. By doing specific experiments and practical tests on the field, the modelling procedure is initiated. The simulations are performed to extend the measurement results to the points that are not measurable due to physical or technical limits. The analytical analyses can also be employed to generalize the defined relations and models by doing for instance, linearization, interpolation, regression, etc. The machine learning and artificial intelligence techniques can help to extract the relations between input and output spaces by exploring big historical or measured data. The system identification can also be employed to model the input-output relations from measurement data when those relations are governed by the dynamic behaviour.
It should be noted that this paper does not focus on the modelling techniques. Regardless of the modelling approach that is employed, the available models (of loads, lines, transformers, etc.) in the electric power systems are not perfect and accurate. To investigate the model inaccuracy impacts, we consider that the parameters of system component models are varying with the predefined bounds (around their nominal values). In the submitted paper, we consider impacts of model uncertainty on the voltage control problem, and we develop a voltage control algorithm based on the chance-constrained optimization to efficiently deal with this phenomenon.
Comment 5: 119 What guarantees convexity? (having a convex optimization)
Response to comments 5: The convexity of the sensitivity-based voltage control formulation is guaranteed since both its objective function and its constraints are linear. The convexity is one of the inherent features of the linear optimization problems.
Comment 6: Please provide algorithmic representation of the following procedure:
nts of the transformer tap changer, respectively. 145 The voltage sensitivity data needed in (6) are obtained as follows. The coefficients 146 giving the voltage sensitivity with respect to active or reactive power changes of DGs can 147 be extracted from the inverse Jacobian matrix in the Newton-Raphson load flow (NRLF) 148 study. The sensitivity of voltages with respect to the (substation) transformer tap changes 149 is obtained using the perturb-and-observe method from the results of two consecutive 150 load flow studies subject to one step change of the transformer tap position [2].
Response to comments 6: As requested by the reviewer, the procedure to obtain the voltage sensitivity coefficients is described in an algorithmic manner using the bullet points:
“
- The voltage sensitivity with respect to nodal power changes: it is extracted from the inverse Jacobian matrix in the Newton-Raphson load flow (NRLF) study as explained in [19].
- The voltage sensitivity with respect to (substation) transformer tap changes: it is obtained using the perturb-and-observe method. In the latter, two consecutive load flow studies are performed subject to one step change of the transformer tap position [2]. The voltage variation at the observed node subject to the step change applied to the perturbation node (the transformer tap position) is evaluated to derive the sensitivity of voltage at the observed node with respect to the transformer tap changer action.”
Comment 7: Please improve description of Figure 3: 373 As it can be seen in figure 3
Response to comments 7: As suggested by the reviewer, the description of figure 3 has been completed as follows:
“Figure 3. Impacts of considered model uncertainties on nodal voltages (solid blue line represents the voltages obtained considering the nominal parameters of system components, solid red line shows the predefined voltage limit i.e., 1.03 pu, and the boxplots demonstrate the model uncertainty impacts on nodal voltages”
Comment 8: 451 Please explain: disregarding the model uncertainties (= 6.044).
Response to comments 8: ‘‘disregarding the model uncertainties” implies that the voltage control problem considers the nominal parameters of system components; thus, the uncertainties arisen from e.g., voltage dependency of loads, thermal dependency of line resistances, etc. are disregarded. Solution of the voltage control algorithm disregarding the model uncertainties, called deterministic voltage control algorithm (DVCA) (presented in (5)-(9)), is immunized against the nominal parameters of system components, but if the system components take any other value than the nominal one, the solution of the DVCA cannot correctly remove the initial voltage violations. In figure 3, in the submitted paper, it is shown that due to uncertainties in parameters of network components, the voltage violations can have up to 0.01 pu variation with respect to the voltages obtained by neglecting uncertainties using nominal parameters (see figure 3, nodal boxplot of voltages shows uncertainty impacts and the blue solid line represents the voltages obtained by nominal parameters). Neglecting such variations in initial points (voltages) will mislead the DVCA and result in infeasible control solutions. The DVCA disregarding the model uncertainties requires smaller control effort (= 6.044) to manage the violated voltages since it considers the nominal parameters of system components, but its obtained solution will not be immunized against any other realization of uncertainty sources.
In order to clarify this aspect, the following explanation is added to the revised paper:
“In table 2, one can observe that the smallest objective function corresponds to the DVCA (= 6.044) that disregards the model uncertainty impacts by considering the nominal parameters of network component models”
Comment 9: Validation should be provided on real data.
Aporopriate number of real world situations should be compared to Monte carlo simulation. What is appropriate the statistial test should tell.
Response to comments 9: The authors would like to appreciate the reviewer for this important comment. Indeed, in order to validate the proposed voltage control algorithm, it is required to consider several network operating points on which the performance of the voltage control algorithm is tested. Such an extensive analysis however necessities to repeat the section 6.2 of the submitted paper for each of the considered working points, which would not be technically possible due to the space limit. In order to address this challenge, we validate the proposed voltage control algorithm on a working point that corresponds to maximum generation – minimum load situation. The latter point constitutes the most difficult voltage management task having the highest possible voltage violations. By demonstrating the appropriate performance of the proposed voltage control algorithm on this selected working point, it can be concluded that the proposed algorithm would be able to manage other network working points too (having naturally smaller voltage violations).
It should be noted that this paper does not aim to define the optimal confidence level to be used in the chance-constrained voltage control formulation. The current paper rather proposes the chance-constrained voltage control algorithm as an alternative to the robust voltage control algorithm that is too conservative and the deterministic voltage control algorithm that cannot guarantee feasible solutions. The accuracy of the proposed chance-constrained formulation is validated for various confidence levels (= 0.2, 0.1 and 0.05) in the post-processing analysis. The optimal risk level should be determined in accordance with on the one hand, the costs of corrective control actions, and on the other hand, the voltage violation costs.
Author Response File: Author Response.docx
Reviewer 2 Report
The paper presents an interesting technique for voltage control in distribution networks, taking into account the different uncertainties present in the problem. The paper is well written and organized, and both the proposed formulation and the reported results are of undoubted practical interest.
The following comments can be used to improve the paper:
1. Control techniques based on sensitivities are well known and there are recent publications on the subject that are not cited by the authors. For instance,
Zarco Soto, Francisco Javier, Martinez Ramos, Jose Luis, Zarco Periñan, Pedro, A Novel Formulation to Compute Sensitivities to Solve Congestions and Voltage Problems in Active Distribution Networks.
IEEE Access. 2021. Vol. 9. pp. 60713-60723. 10.1109 / Access.2021.3073082
2. The formulated problem is based on the linear approximation using sensitivities. However, this problem is valid in a relatively reduced environment with respect to the working point of the electrical system. In this sense, it is usual to limit the magnitude of the control actions to avoid large errors due to the linearization of the problem. The authors should comment in this regard in the paper.
Author Response
Response to the reviewer’s comments regarding submission energies-1280782 entitled:
‘‘Chance-constrained based voltage control framework to deal with model uncertainties in MV distribution systems’’
By: Bashir Bakhshideh Zad, Jean-François Toubeau, and François Vallée
Introduction: The authors would like to appreciate the reviewer for giving the following valuable comments, thanks to which we could improve the quality and readability of the paper. The current document presents a point-by-point response to the reviewer’s comments. Also, according to the given comments and suggestions, the paper has been revised and resubmitted. The authors hope that the current revision of the paper would adequately address the questions and concerns regarding the previous submission. All the changes and modifications relating to the raised comments are applied to the revised paper in the track mode for a better visibility.
Reviewer’s comments
The paper presents an interesting technique for voltage control in distribution networks, taking into account the different uncertainties present in the problem. The paper is well written and organized, and both the proposed formulation and the reported results are of undoubted practical interest.
The following comments can be used to improve the paper:
Comment 1: Control techniques based on sensitivities are well known and there are recent publications on the subject that are not cited by the authors. For instance,
Zarco Soto, Francisco Javier, Martinez Ramos, Jose Luis, Zarco Periñan, Pedro, A Novel Formulation to Compute Sensitivities to Solve Congestions and Voltage Problems in Active Distribution Networks.
IEEE Access. 2021. Vol. 9. pp. 60713-60723. 10.1109 / Access.2021.3073082
Response to comment 1: The authors would like to appreciate the reviewer for introducing this interesting reference. The latter has been added to the revised paper (as reference [26]) to represent the category of voltage control techniques based on the sensitivity analysis, alongside [1,2], [4,5], [7,8], [27].
Comment 2: The formulated problem is based on the linear approximation using sensitivities. However, this problem is valid in a relatively reduced environment with respect to the working point of the electrical system. In this sense, it is usual to limit the magnitude of the control actions to avoid large errors due to the linearization of the problem. The authors should comment in this regard in the paper.
Response to comment 2: The authors fully agree with the reviewer regarding this raised comment. Indeed, the accuracy of sensitivity analysis may be reduced when the network operating point is largely deviated from its initial position. In order to study this aspect, we evaluate the performance of the proposed voltage control technique on a working point that includes maximum generation from DG units and minimum load demands. To remove the voltage violations of this studied case, the initial working point of the system should be largely moved by the employed control measures. The conducted simulations in the paper confirm the accuracy of employed sensitivity analysis when applying the control actions requested by the voltage control algorithm. The accuracy of the sensitivity-based voltage control algorithm has been further investigated in our previous publications that can be found here:
- Bakhshideh Zad, B.; Hasanvand, H.; Lobry, J.; Vallée, F. Optimal reactive power control of DGs for voltage regulation of MV distribution systems using sensitivity analysis method and PSO algorithm. International Journal of Electric Power and Energy Systems, 2015, 68, 52-60.
- Bakhshideh Zad, B.; Lobry, J.; Vallée, F. A centralized approach for voltage control of MV distribution systems using DGs power control and a direct sensitivity analysis method. Proceedings of 2016 IEEE International Energy Conference, Belgium, 2016.
- Bakhshideh Zad, B.; Vallée, F.; Lobry, J. A new voltage sensitivity analysis method for medium-voltage distribution systems incorporating power losses impact. Electric Power Components and Systems, 2018, 46, 1540-1553.
Author Response File: Author Response.docx
Reviewer 3 Report
This paper presents voltage level control considering uncertainties created by the lack of accurate information about network parameters. The method used is a chance-constrained framework.
The article begins with an introduction in which the authors present the problem and other solutions used in a clear and organized manner. They present contributions and discuss the content of the paper. Next, a further discussion on voltage control in the MV distribution system is presented. The third chapter describes the chance-constrained optimization algorithm. Chapter four discusses the applied algorithm. Chapter five presents and describes the analyzed MV distribution system. The sixth chapter was dedicated to the obtained results. The article ends with discussion and conclusions.
The article is well organized, interesting, and a good read. The structure is logical. In the introduction the authors present the problem by moving step by step to the performed study and research.
Minor comments to authors:
1. The contributions are clearly stated, however they can be lost in the text - the reviewer suggests to use boolet list to highlight them.
Author Response
Response to the reviewer’s comments regarding submission energies-1280782 entitled:
‘‘Chance-constrained based voltage control framework to deal with model uncertainties in MV distribution systems’’
By: Bashir Bakhshideh Zad, Jean-François Toubeau, and François Vallée
Introduction: The authors would like to appreciate the reviewer for giving the following valuable comments, thanks to which we could improve the quality and readability of the paper. The current document presents a point-by-point response to the reviewer’s comments. Also, according to the given comments and suggestions, the paper has been revised and resubmitted. The authors hope that the current revision of the paper would adequately address the questions and concerns regarding the previous submission. All the changes and modifications relating to the raised comments are applied to the revised paper in the track mode for a better visibility.
Reviewer’s comments
This paper presents voltage level control considering uncertainties created by the lack of accurate information about network parameters. The method used is a chance-constrained framework.
The article begins with an introduction in which the authors present the problem and other solutions used in a clear and organized manner. They present contributions and discuss the content of the paper. Next, a further discussion on voltage control in the MV distribution system is presented. The third chapter describes the chance-constrained optimization algorithm. Chapter four discusses the applied algorithm. Chapter five presents and describes the analyzed MV distribution system. The sixth chapter was dedicated to the obtained results. The article ends with discussion and conclusions.
The article is well organized, interesting, and a good read. The structure is logical. In the introduction the authors present the problem by moving step by step to the performed study and research.
Minor comments to authors:
Comment 1: The contributions are clearly stated, however they can be lost in the text - the reviewer suggests to use boolet list to highlight them.
Response to comment 1: The authors would like to thank the reviewer for this suggestion. The contributions of submitted paper are listed using the bullet points for better visibility (see below), as suggested by the reviewer:
“In view of the above discussion, the main contribution of this paper lies in the proposed formulation of the CC optimization for the voltage control task which has the following salient features.
- It preserves the linearity of the original voltage control problem.
- It effectively addresses the complex coupling uncertainties present in the voltage control problem.
- It leads to accurate voltage corrections as expected from the imposed confidence level, which allows us to efficiently cope with the considered uncertainty sources.”
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
Thank you for your replies. I would suggest to add some validation aspect in disussion section or where appropriate. Please rephrase the following paragraph and add it to the paper for interested readers. Your response regarding the convexity might also be added where apropriate.
->
in order to validate the proposed voltage control algorithm, it is required to consider several network operating points on which the performance of the voltage control algorithm is tested. Such an extensive analysis however necessities to repeat the section 6.2 of the submitted paper for each of the considered working points, which would not be technically possible due to the space limit. In order to address this challenge, we validate the proposed voltage control algorithm on a working point that corresponds to maximum generation – minimum load situation. The latter point constitutes the most difficult voltage management task having the highest possible voltage violations. By demonstrating the appropriate performance of the proposed voltage control algorithm on this selected working point, it can be concluded that the proposed algorithm would be able to manage other network working points too (having naturally smaller voltage violations).
It should be noted that this paper does not aim to define the optimal confidence level to be used in the chance-constrained voltage control formulation. The current paper rather proposes the chance-constrained voltage control algorithm as an alternative to the robust voltage control algorithm that is too conservative and the deterministic voltage control algorithm that cannot guarantee feasible solutions. The accuracy of the proposed chance-constrained formulation is validated for various confidence levels (= 0.2, 0.1 and 0.05) in the post-processing analysis. The optimal risk level should be determined in accordance with on the one hand, the costs of corrective control actions, and on the other hand, the voltage violation costs.
Author Response
Response to the reviewer’s comments regarding submission energies-1280782 entitled:
‘‘Chance-constrained based voltage control framework to deal with model uncertainties in MV distribution systems’’
By: Bashir Bakhshideh Zad, Jean-François Toubeau, and François Vallée
Introduction: The authors would like to appreciate the reviewer for giving the following valuable comments, thanks to which we could improve the quality and readability of the paper. The current document presents a point-by-point response to the reviewer’s comments. Also, according to the given comments and suggestions, the paper has been revised and resubmitted. The authors hope that the current revision of the paper would adequately address the questions and concerns regarding the previous submission. All the changes and modifications relating to the raised comments are applied to the revised paper in the track mode for a better visibility.
Reviewer’s comment
Thank you for your replies. I would suggest to add some validation aspect in disussion section or where appropriate. Please rephrase the following paragraph and add it to the paper for interested readers. Your response regarding the convexity might also be added where apropriate.
in order to validate the proposed voltage control algorithm, it is required to consider several network operating points on which the performance of the voltage control algorithm is tested. Such an extensive analysis however necessities to repeat the section 6.2 of the submitted paper for each of the considered working points, which would not be technically possible due to the space limit. In order to address this challenge, we validate the proposed voltage control algorithm on a working point that corresponds to maximum generation – minimum load situation. The latter point constitutes the most difficult voltage management task having the highest possible voltage violations. By demonstrating the appropriate performance of the proposed voltage control algorithm on this selected working point, it can be concluded that the proposed algorithm would be able to manage other network working points too (having naturally smaller voltage violations).
It should be noted that this paper does not aim to define the optimal confidence level to be used in the chance-constrained voltage control formulation. The current paper rather proposes the chance-constrained voltage control algorithm as an alternative to the robust voltage control algorithm that is too conservative and the deterministic voltage control algorithm that cannot guarantee feasible solutions. The accuracy of the proposed chance-constrained formulation is validated for various confidence levels (= 0.2, 0.1 and 0.05) in the post-processing analysis. The optimal risk level should be determined in accordance with on the one hand, the costs of corrective control actions, and on the other hand, the voltage violation costs.
Authors’ Responses: The three points raised by the reviewer have been addressed in the new revised manuscript as follows.
- Regarding the first paragraph (choice of the selected network operating point), the following text has been added to the paper (lines 354-359):
“The selected network operating point constitutes the most difficult voltage management task having the highest possible voltage violations. By validating the performance of the proposed voltage control algorithm on the latter working point, it can be expected that the proposed algorithm would be able to manage other network operating points, having naturally smaller voltage violations.”
- Concerning the choice of the optimal confidence level for the chance-constrained voltage control problem, it has been already discussed in the first paragraph of the discussion section (lines 505-518) that can be found here:
“The simulations carried out in the previous section confirm that the proposed CC voltage control framework can effectively deal with the model uncertainty inherent in the distribution systems. It enables us to adjust the desired confidence level according to which we aim to immunize the voltage control solution subject to uncertainty realizations. While the solution of the voltage control method relying on deterministic simplified models might be insufficient to completely manage the voltage constraints, and on the other hand, the solution of the robust voltage control technique could appear too conservative, the proposed CC framework allows us to find a compromise between the voltage management costs and the conservatism degree. It should be noted that the desired conservatism degree depends not only on the costs of control decisions, but al-so on the costs associated with the voltage violations. In other words, the optimal risk factor should be determined in accordance with on the one hand, the costs of corrective control actions, and on the other hand, the voltage violation costs. The more crucial is the voltage violation management, the more justified will be the price of robust-ness to pay. ”
- Regarding the convexity feature of the linear optimization problem, it has been added to the revised paper (lines 119-121) as:
“It constitutes a convex optimization problem (having linear objective function and constraints) that will ensure the optimality of solutions obtained by the implemented voltage control algorithm. ”
Author Response File: Author Response.docx