A Frequency–Power Droop Coefficient Determination Method of Mixed Line-Commutated and Voltage-Sourced Converter Multi-Infeed, High-Voltage, Direct Current Systems: An Actual Case Study in Korea

Round  1

Reviewer 1 Report

The paper is focused on an electrical island with HVDC injections from LCC and VSC. The primary frequency control droops of the HVDC injections are determined via a “graphic” method, in such a way as to minimise the losses in the steady state of the primary regulation. The proposed approach is applied to a test system and to a model of Jeju island power system.

The objective of the paper is clearly stated, however the reader might wonder if minimising the losses is relevant in this context, as the situation referred to (i.e. steady state of primary frequency regulation at off-nominal frequency) might not last long (other regulations will shortly take over), moreover it may be not very far from the original operating condition (chosen according to some optimality criterion, whether market-based or other). Please discuss.

Section 2.1 presents an optimal reactive power flow formulation suitable for the configuration under analysis (island plus HVDC injections). It is functional to section 2.2.

Conversion losses are not included in the optimisation problem presented in section 2.1. This might appear a serious limit of the optimisation approach, please discuss.

The formulation is presented for one LCC and one VSC however in the examples there are 2 VSC, it may be worth saying here (2.1) that it be generalised.

Eq. (5) clarify the “<”< p="">

Sect. 2.1 “The interior-point method cannot guarantee a global optimum because of non-linearity, but many studies have adopted the solver in conventional optimization to find a better solution” please clarify the last part of the sentence.

Section 2.2 The method presentation could be improved e.g. by more systematic listing of the procedure, the flow chart in figure 2 helps but might be anticipated and further detailed. Please consider in depth revision of this section. Moreover You may anticipate that (if I understand well) the Monte Carlo load profiles are generated as small variations of the base case load profile, since they represent possible realisations in presence of forecast errors.

>assumption that only HVDCs participate in frequency regulation

This is a weakness of the approach though it may probably be overcome, please discuss.

Clarify how (17) is obtained and why it is a “compensation”.

“we can determine the slopes via a graphical methodology”. The expression “graphic method” is present several times though it appears not only graphic as reported in 2.2: ” the frequency-power droop can be represented by the slope of the blue line, to minimize the root mean square error between the points above the blue line and the grey dots.” Please clarify.

Section 3

First part of the results is just the application of the optimal reactive power flow, the comparison of the impact on voltage of the different strategies is probably little relevant.

“calculated droop coefficients are verified by comparison with the conventional case, which adopts an optimization algorithm in steady-state”, please clarify what is the conventional case and algorithm.

Editing: put figures after you have referred to them.

Conclusions

“analyzes the results graphically to reduce computation burdens”, please clarify

English: some instances of “different to”, check expression.

Author Response

We are grateful for your comments. We uploaded the response to your comments by MS word file.

Author Response File: Author Response.pdf

Reviewer 2 Report

Overall, the manuscript has a very good flow, and it’s easy to follow and understand. This manuscript proposed a method to determine frequency-power droop coefficients for mixed LLC and VSC multi-infeed HVDC systems. The proposed method effectively minimized the power loss in the IEEE 14-bus the system and the Jeju Island case study. The below items are suggested:

1.     Suggest changing the Figure 4, 5 and 8 to the bar diagrams, because bus numbers are discrete and directly correlated.

2.     Suggest reducing the fundamental description in Section 2.1.1

3.     Suggest adding further description explain the process of minimization of the power loss in both test cases.

Author Response

We are grateful for your comments. We uploaded the response to your comments by MS word file.

Author Response File: Author Response.pdf

Reviewer 3 Report

Figure 5 shows three results, one of which is the result of the proposed method. Table1 shows that the result of the proposed method is best in the sense of power loss. Confusing here is the fact that VC Mode gives the haigher voltage profile compared to that by the proposed. Usually the power loss minimization leads to the higher voltage profile and the higher the voltage is, the better the loss characteristics are.

The authors discuss that the VSC mode gives the solution where bigger reactive flow appears, and it looks like that this greater reactive flow gives bigger line loss. However, the reviewer thinks that this small difference would be overcompensated by the difference in voltage level in the obtained states. The reviewer requests that the authors should clearly explains why the proposed method gives better performance than VC Mode based on physical interpretation, not only that it is the result of the optimization process.‍

Author Response

We are grateful for your comments. We uploaded the response to your comments by MS word file.

Author Response File: Author Response.pdf

Round  2

Reviewer 1 Report

Here are Reviewer 1’s replies 

Point 1

Maybe you may consider this proposed variant:  

"It is practically important in case of islanded network, because the power disturbances in an island system are comparatively larger than in large systems. Therefore, off-nominal frequency situations occur more frequently and the frequency deviations are more severe than in conventional large power networks.  "

References to inertia should be checked and possibly eliminated, in fact inertia is involved only in frequency transients, while at steady state the frequency only depends on the network power frequency characteristic and on the disturbance.

Point 2

Loss minimisation is the objective of the optimisation presented in the paper. Neglecting VSC losses appears a limit from the overall (global) system optimisation viewpoint, it could deserve future investigation.

Best regards

Author Response

We are grateful for your comments. We add MS word file to respond your comment.

Author Response File: Author Response.pdf