Energy Cost Minimization with Hybrid Energy Storage System Using Optimization Algorithm
Round 1
Reviewer 1 Report
1. Contribution is not clear. Include contribution of this paper in Introduction section in a separate paragraph in the bulleted for form so that it can be easily followed by the readers.2. Due to the huge number of acronyms used, need there is a complete list of those. Please provide nomenclature with engineering units for each variable in the manuscript.
3. The title does not very well reflect the work presented in the manuscript. There is nothing new in the PV modeling presented in the manuscript.
4. The abstract and conclusion fail to explain the main target of this research. The authors are advised to rewrite the abstract considering: 1-2 lines as a description of the obtained results. 1-2 lines as a description of the significance in solving the problem and the future work.
5. Suggest to include the convergence curves of the proposed algorithm. Discuss the convergence speed of the proposed algorithm with other compared algorithms.
6. When using numerical methods, it is not just enough to employ them. We also need to discuss the reasons why this set of numerical methods are used.
7. There is no discussion on the cost effectiveness of the numerical methods surveyed. What is the computational complexity?
8. The review of optimization methods can be expanded, some works are recommended to be added, such as:
- Golden Search Optimization Algorithm "10.1109/ACCESS.2022.3162853
- Damping controller design for power system oscillations using hybrid GA-SQP
Author Response
Dear Reviewer,
Thank You for the careful revision of the manuscript.
We have referred to Your comments in the revised version of the manuscript.
Please find below answers to Your specific suggestions.
- Contribution is not clear.Include contribution of this paper in Introduction section in a separate paragraph in the bulleted for form so that it can be easily followed by the readers.
A bulleted list was added at the end of introduction to emphasise the contribution of the paper.
- Due to the huge number of acronyms used, need there is a complete list of those. Please provide nomenclature with engineering units for each variable in the manuscript.
A table which lists all abbreviations used in the article was added as supplementary material.
- The title does not very well reflect the work presented in the manuscript. There is nothing new in the PV modeling presented in the manuscript.
In our opinion the title “Energy Cost Minimization with Hybrid Energy Storage System Using Optimization Algorithm” refers precisely to the core of presented work. We do not claim in the title nor elsewhere that PV modelling is a novelty of the presented work, PVs are only present as the renewable source in the model.
4. The abstract and conclusion fail to explain the main target of this research. The authors are advised to rewrite the abstract considering: 1-2 lines as a description of the obtained results. 1-2 lines as a description of the significance in solving the problem and the future work.
The abstract has been updated according to the suggestions. Additional paragraphs regarding results and future work have been added to the discussion and conclusion.
5. Suggest to include the convergence curves of the proposed algorithm.Discuss the convergence speed of the proposed algorithm with other compared algorithms.
Additional paragraphs with analysis of complexity was added to the manuscript. Additional references have been added to address convergence discussion. The SLSQP was used in many publications in the domain of energy, e.g. [35-36]. The convergence and properties of the SLSQP are described in [37-38].
- When using numerical methods, it is not just enough to employ them. We also need to discuss the reasons why this set of numerical methods are used.
This work is focused on a particular application of a Hybrid Energy Storage System. Its goal is to verify applicability of the optimization- based algorithms in such a scenario. Comparison of different numerical methods is not within the scope of this paper – it is seen as a future work.
In this article only one optimisation method is used and it is an SLSQP algorithm. We chose this method as it is based on verified methods, it is sufficiently fast and for the considered problem it should find the optimal solution if there is one, without the need for finding parameters to make the algorithm converge. The goal function is not very complex one, although there are situation with multiple minimas.
- There is no discussion on the cost effectiveness of the numerical methods surveyed. What is the computational complexity?
Thank You for drawing our attention to this missing aspect in out article. Additional paragraphs were added to section 2.5 to solve this issue.
- The review of optimization methods can be expanded, some works are recommended to be added, such as:
- Golden Search Optimization Algorithm "10.1109/ACCESS.2022.3162853
- Damping controller design for power system oscillations using hybrid GA-SQP
The papers suggested by the reviewer are not directly relevant to this publication. Our goal is not to study the performance of different optimizers, but to check if the proposed method for HESS arbitrage has an economic justification. We however can see the origin of the comment and rewrote several parts of the article to further clarify this point.
Reviewer 2 Report
1. Why choose 15min as a time stamp instead of 20min or 10min?(in line 105) Does other referance have related instructions?
2. What are the input and output of the optimization model? Why choose the SLSQP algorithm instead of using intelligent optimization algorithms, such as particles swarm optimization(PSO) or genetic algorithm(GA)? What effect does the selection of different algorithms on optimization?
3. The writing in the article should be consistent with the legend, like Figure3, not Fig.3. (in line 108) Please check the full text.
4. In Formula 2, Eboli writing is inconsistent with the text, it should be written as a bidding Ebol,i. Please check the full text.
5. How to explain when the battery has enough energy to meet the load, and still needs to buy electricity from the grid, as shown in Figure 7d.
6. What might be the reason behind the discharge energy generated on only one day in Figure 8c?
7. The title of line 368 is ‘3.2 Economic Optimization’, and the title of line 452 is ‘3.2. Modified Economic Optimization’. Are you sure, they are all 3.2?
8. The optimization results of the energy balancing method and the economic optimization method seem to be small in Figure 8d and Figure 15d, and the battery operations of the economic optimization method seem to increase the additional cost? Is it the lowest total cost after increasing the additional battery cost for the month of January?
9. In Figure 9d and Figure 16d, the economic optimization method seems to be more costly compared to the energy balancing method. Even if it is reduced after modified economic optimization, it is still not as good as the energy balancing method. Is it more economical to consider the combination optimization(energy balancing method + modified economic optimization)?
10. Frequent charging and discharge process will cause battery depreciation. Whether there is a threshold to prove that the system containing a battery is more competitive than buying electricity directly from the power grid.
11. Is there any consideration for the calculation time of each strategy, and is it a factor that has a greater impact?
12. This article choose 24h as T cyclical optimization. What effect does the optimization of the previous cycle have on the optimization of the next cycle?
13. Does the optimization result exist in multiple poles? What is the difference between them? Which optimization method is more stable and applicable?
Author Response
Dear Reviewer,
Thank You for the careful revision of the manuscript.
We have referred to Your comments in the revised version of the manuscript.
Please find below answers to Your specific suggestions.
- Why choose 15min as a time stamp instead of 20min or 10min?(in line 105) Does other referance have related instructions? x
The 15 minute time interval was chosen, as it is the standard resolution of energy meters in Poland. This strictly binds the model to the data gathered in real systems, and allows us to use the data without a non-trivial conversion.
- What are the input and output of the optimization model? Why choose the SLSQP algorithm instead of using intelligent optimization algorithms, such as particles swarm optimization(PSO) or genetic algorithm(GA)? What effect does the selection of different algorithms on optimization?
This paper does not focus on the optimization algorithm itself. In this article only one optimisation method is used and it is an SLSQP algorithm. We chose this method as it is based on verified methods, it is sufficiently fast and for the considered problem it should find the optimal solution if there is one, without the need for finding parameters to make the algorithm converge. Section 2.3 describes input and output of the algorithm.
- The writing in the article should be consistent with the legend, like Figure3, not Fig.3. (in line 108) Please check the full text.
The manuscript has been checked and such errors have been corrected.
- In Formula 2, Eboli writing is inconsistent with the text, it should be written as a bidding Ebol,i. Please check the full text.
The formula has been corrected.
- How to explain when the battery has enough energy to meet the load, and still needs to buy electricity from the grid, as shown in Figure 7d.
Figure 7d (Figure 9d in the revised version) represents the monthly aggregation, so there can be days where it is necessary to buy electricity and there can be days when there is an abundance of power from PV. It has been explained in lines 374-378.
- What might be the reason behind the discharge energy generated on only one day in Figure 8c?
The Figure 8c represents the operation of the battery in January, and at the beginning of the simulation the battery was charged to 50% (as explained in lines 386-388) - which was an arbitrary choice of the authors. Due to lack PV overproduction in January the battery discharges in the first day of the year and then remains inactive till the charging possibilities appear, according to the considered method.
- The title of line 368 is ‘3.2 Economic Optimization’, and the title of line 452 is ‘3.2. Modified Economic Optimization’. Are you sure, they are all 3.2?
It is an obvious mistake, the numbering has been corrected. “Modified Economic Optimization” is section 3.3.
- The optimization results of the energy balancing method and the economic optimization method seem to be small in Figure 8d and Figure 15d, and the battery operations of the economic optimization method seem to increase the additional cost? Is it the lowest total cost after increasing the additional battery cost for the month of January?
There is no surplus PV energy in January. The economic optimization method only provides some small savings using energy arbitrage, so the cost difference is relatively small. The method accounts for the battery depreciation, so the total cost is minimized. Please refer to the table 2, where results with and without battery depreciation are compared.
- In Figure 9d and Figure 16d, the economic optimization method seems to be more costly compared to the energy balancing method. Even if it is reduced after modified economic optimization, it is still not as good as the energy balancing method. Is it more economical to consider the combination optimization(energy balancing method + modified economic optimization)?
The economic optimization method in certain situations is more costly compared to the energy balancing method, which, as explained in the article, is due to the limitation of the optimization process to 24-hours. That is the reason for introducing the modified economic optimization. In this case the difference between the energy balancing method and the modified economic optimization method is very small, and is mainly caused by small oscillations of the solutions given by the optimizer when it failed to reach the optimum solution in the defined number of iterations.
An additional explanation is added in lines 472-476.
- Frequent charging and discharge process will cause battery depreciation. Whether there is a threshold to prove that the system containing a battery is more competitive than buying electricity directly from the power grid.
The model includes all costs related to battery operation (energy losses and depreciation related to cycling) as described in lines 121-130. The final economic results in the Table 2 include depreciation (captions have been changed for clarity). Even when battery depreciation is considered, the costs are lower.
- Is there any consideration for the calculation time of each strategy, and is it a factor that has a greater impact?
The calculation time is not considered, as it has no impact for the analysis. Additional complexity analysis has been added in section 2.5. The calculation time has not been an issue as whole year operation is done in rational time on a standard PC. When applying this method in practice, only a 24h window is calculated at a time.
- This article choose 24h as T cyclical optimization. What effect does the optimization of the previous cycle have on the optimization of the next cycle?
The SOC of the batteries is passed to the next iteration, as the starting condition for calculating the next day are dependent on the behavior of the battery the day before. The manuscript refers to these issues in lines 305 and 474. The modified economic optimization method improves on the possibility for the cycle of one day to improve the next.
- Does the optimization result exist in multiple poles? What is the difference between them? Which optimization method is more stable and applicable?
There are multiple solutions as for example discharging for 1 hour within one tariff zone is equivalent from a cost-perspective for all hours in this zone. As such, the solver will return one solution that provides us with a discharge planning, but all those solutions are equivalent from the application point of view. Particularly the COBYLA solver seemed more unstable and fluctuating between solutions, impacting its convergence to a good solution. The SLSQP solver behaves better, but most likely also exhibits this behavior. This to some extent explains why it also fails to finish within the set number of iterations, but the solution returned in those cases still proved to be adequate (economically good enough) for our purpose.
Reviewer 3 Report
This study investigates Energy Cost Minimization with Hybrid Energy Storage System Using Optimization Algorithm. Although the results attained in the present study show the importance of the paper, The authors should address the following comments:
1: In general, the introduction is light and does not represent this domain's state of the art. The amount of works in this area continues to rapidly rise. Important references on the subject covered in this article are missing. The authors are advised to strengthen their literature review section with supplementary material.
2: In general, the authors should give more details about the used optimization algorithms.
3: What are the advantages and disadvantages that should be highlighted?
4: Proof reading by a native English speaker should be conducted to improve both language and organization quality.
5: The conclusion section is too brief, please supplement it to make the article more organized.
Author Response
Dear Reviewer,
Thank You for the careful revision of the manuscript.
We have referred to Your comments in the revised version of the manuscript.
Please find below answers to Your specific suggestions.
- In general, the introduction is light and does not represent this domain's state of the art. The amount of works in this area continues to rapidly rise. Important references on the subject covered in this article are missing. The authors are advised to strengthen their literature review section with supplementary material.
The manuscript was extended a more details and additional references were added.
- In general, the authors should give more details about the used optimization algorithms.
We added details on the employed SLSQP optimizer and included the reference. Added: SLSQP optimizer is a sequential least squares programming algorithm, it applies the Han-Powell quasi-Newton method with a BFGS update of the B-matrix and an L1-test function in the step-length algorithm. It has implemented a modified version of Lawson and Hanson’s NNLS nonlinear least-squares solver. The original source code was provided by Dieter Kraft in "A software package for sequential quadratic programming", DFVLR-FB 88-28, 1988.
- What are the advantages and disadvantages that should be highlighted?
The conclusion was extended to better highlight the advantages and disadvantages. To quickly summarize, the advantages are: cheaper, and increased auto-consumption; disadvantages: none.
- Proof reading by a native English speaker should be conducted to improve both language and organization quality.
We went over the entire article and different sections were restructured or extended and many sentences were rewritten.
- The conclusion section is too brief, please supplement it to make the article more organized.
The conclusion was extended to summarize all findings and results from the article, as well as propose potential future work.
Round 2
Reviewer 1 Report
Publish
Reviewer 2 Report
It can be accepted.
Reviewer 3 Report
In the revised version of the manuscript, the authors have attempted the suggested changes.
