Optimization Approach for Long-Term Planning of Charging Infrastructure for Fixed-Route Transportation Systems
Round 1
Reviewer 1 Report
This paper has presented well and helped the state of the art. I would like to congratulate the authors for this work.
- Kindly maintain the caption format of table 2 to match the rest of the paper.
2. Please use parenthesis for “See Chapter 2” on page: 5, line: 202.
3. Please revise the paper and check the grammar and use of punctuations throughout the paper, as there are some mistakes.
4. Kindly re-write the sentence in page 8, line 301 and 326 with proper punctuation and grammar.
5. Please write the theta symbol in page 13, line 501 and page 14, line 544 correctly. Kindly revise the paper and check the grammar and use of punctuations throughout the paper, as there are some mistakes.
6. The focus of the paper and the approached methodologies and contributions are not well synchronized. Kindly mention the contributions properly and align with the main objective of the paper. You may refer to the paper "A Comprehensive Study of Key Electric Vehicle (EV) Components, Technologies, Challenges, Impacts, and Future Direction of Development" to bolster the concept.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
This paper is interesting to read and very relevant to the journal. It summarises the research in the topic of charging infrastructure planning for electric buses and also presents an optimisation model which I have doubts on its novelty compared to previous research, since the reader is presented with very little results and a weak motivation on the research gap covered (see more specific comment on this aspect below).
Another important issue to address is the lack of conclusions. I think this lack of specific conclusions is strongly connected to the general lack of specific aim and research questions that needs to be adressed for this paper to be ready for publishing. Please, update by providing specific aim and research questions in the beginning of the paper and link to specific conclusions in the end.
The lack of research questions and a clear aim is also affecting the structure: is this paper a literature review? The literature review covers 7 of the 20 pages of the pdf. Is this paper a model implementation paper? We never get to really see the results of the model in detail. I think the paper would benefit a lot if the authors consider deciding on shortening some sections and tightening the scope.
Some more specific comments below:
Abstract: Please present in more detail the research questions, results, and conclusions of the paper. There abstract seems to be too general at this point.
Line 38: “The fluctuating character of renewable energies needs to be compensated.” Compensated with what? Please clarify this sentence.
Line 90-92: “As can be seen in Table 1, a list of stakeholders is complemented by perspectives, objectives, and literature references. The stakeholders are affiliated to either the transportation or the energy sector (first column).” Why was this choice made? Please explain for the reader.
Table 1: It is a great idea to provide this table, but please complement with relevant references to all optimisation objectives, otherwise the purpose of including the reference column is not that clear. This is done well for example in Table 2 where the reference column is filled for all rows.
Line 117-119: “The present paper will focus on the optimization of a single objective: minimizing the monetary cost related to the operation of a vehicle fleet including the construction of charging infrastructure. “ Why has this choice been made? It is important the authors clarify and motivate that choice when they have just presented how wide the variety of optimisation objectives is.
General comment on Section 1: It is scientifically paramount that the specific research questions this paper addresses are presented in a clear way to the reader. I do not see these questions, please provide them.
Line 205-206: “increasing the individual budgets available in each stage can never worsen the optimization results.” This concept might be unfamiliar to the readers (it is to me). I think it is worth allocating a sentence or two on clarifying.
Figure 1: I do not see the benefit of including the upper part of the figure on stakeholders and perspectives. If the factors are each not allocated to perspectives and stakeholders then there is not a purpose to include this part and I suggest to remove or, alternatively, match perspectives to factors (which they don’t as-is in this figure).
Line 272-273: do you mean in terms of planning operations to avoid accelerated battery aging?
Line 284-289: Do the authors mean that Fixed-route transport systems by a single operator have not been studied? Because fixed-route systems have been studied in the literature, albeit without discussing from the operator’s perspective and usually having the city/region limits as the system boundary. Please clarify accordingly, since it is misleading to argue that fixed-route systems have not been studied and there is a research gap. What is the research gap exactly?
Line 336: ICEVs? Do you mean ICEs? Please revise throughout the paper.
Line 351: “end-terminus-charging”: Do you mean “end-stop charging”? Please revise throughout the paper.
Line 428-429:”Further, all computations are performed once for every representative day for reproducing the impact of differing climate conditions. “ Taking into account temperatures you mean?
Line 486-487: “we consider the maximum aggregate power which is drawn in a certain period as the proxy for this share of the cost Cpower contract” Please provide more specific information on how this parameter is calculated. Most power contracts include demand charges based on peak-time power use. If a vehicle charges at noon, the cost will be higher than charging at night, for example. Has this been included in the calculation? If not, please address this aspect in a revised version.
Line 524-525: “Every year, a percentage capacity loss is implemented to account for the battery aging” If only one year is the time period of the model (Line 468), how exactly is this factor applied? Please clarify.
Table 3: Where are the maintenance costs for vehicles and infrastructure? The cost of CPC activation is the cost of one charging session I assume? Where is the investment cost for the charger? Please clarify.
Line 598: “…have either 50 kW or 100 kW maximum charging power.” Why is such low power assumed for the charging? Is this just the depot charging we are talking about? On-route charging has higher power capacity (300 kW and more) in the vast majority of cities implementing electric buses.
Section 4 - general comment: Where are the costs of the proposed solutions? If the objective function is cost minimisation then I certainly expect to see these costs compared to the business-as-usual case. This is a serious issue to address with this paper, i.e. a lot of time is spent on literature review and algorithm presentation but very little time is spent on presenting results.
Section 5: This section should be focusing on the actual conclusions of the research done in this paper but looks more like a general discussion of potential next steps. Please reformulate to provide clear conclusions/answers to specific research questions linked to the work presented in this paper.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
Dear authors,
Thank you very much for your diligent effort to improve the paper.
Several of my initial concerns are now clarified.
However, I still have major concerns on the novelty and relevance of the paper. The authors' responses to comments 14 to 20 of my initial review have not been fully addressed, please see my comments in italics and underlined below (now the comments are numbered 1 to 7 instead of the original 14 to 20 in the authors' responses). I hope these issues are comprehensively addressed in this review round.
- “Line 486-487: “we consider the maximum aggregate power which is drawn in a certain period as the proxy for this share of the cost Cpower contract” Please provide more specific information on how this parameter is calculated. Most power contracts include demand charges based on peak-time power use. If a vehicle charges at noon, the cost will be higher than charging at night, for example. Has this been included in the calculation? If not, please address this aspect in a revised version.”
This is a very important aspect when considering the costs that arise for a fleet operator. At the same time, this is subject to regional differences and a grid operator’s product portfolio. We have now included Equation (5) in the manuscript for clarifying on how this is taken into account. Based on our use‑case, we built the model so that in each period (in every year), the optimization can choose between different power contracts.
Intraday variations of the price structure for electricity are currently not considered, as they played no role for our use-case. Certainly, this represents an important aspect for future considerations. Especially for a V2G operation that relies on dynamic price formation.
Is this limitation of the analysis included in the recommendations for future research of this paper? Please include.
- “Line 524-525: “Every year, a percentage capacity loss is implemented to account for the battery aging” If only one year is the time period of the model (Line 468), how exactly is this factor applied? Please clarify.”
What is meant in line 468 of the first submission is that one period in the framework corresponds to one whole calendric year. If for a use-case we analyze 4 periods, that means we are looking at a time-horizon of 4 years.
But you are not running the model for 4 periods in this paper so why confuse the reader? Please clarify that it is possible to run the model for various time periods but choose here to run for one or remove this sentence to avoid confusion.
- “Table 3: Where are the maintenance costs for vehicles and infrastructure? The cost of CPC activation is the cost of one charging session I assume? Where is the investment cost for the charger? Please clarify.”
The cost for constructing charging points is the main part of the charging infrastructure optimization. This is the investment cost. The term activation is introduced by us to refer to the binary decision of the optimizer to incur the construction cost and enable a CPC.
The computation time is already very extensive for the given set of parameters and variables, so that we have chosen to neglect other aspects, for now. Yearly maintenance costs usually represent only a fraction of fuel / energy cost.
Please clarify/state what parameters are excluded in the text then. Operation and maintenance are quite often part of such calculations so if you are choosing to exclude them you have to motivate why and let the reader know.
- “Line 598: “…have either 50 kW or 100 kW maximum charging power.” Why is such low power assumed for the charging? Is this just the depot charging we are talking about? On-route charging has higher power capacity (300 kW and more) in the vast majority of cities implementing electric buses.”
This relates to the use-case of Darmstadt, where the electric buses (Mercedes Benz eCitaro) have a maximum charging power of 150kW. Therefore, higher charging powers are not included in the optimization.
I am quite concerned of the relevance of using such low charging power. The authors mention themselves that 150 kW is the maximum power for Darmstadt’s case and still choose to go for maximum 100 kW.
Please either provide the results for 150 kW and/or 300 kW or motivate sufficiently and with regards to the current bus electrification trends why such low power is assumed. The study needs to be realistic and offer results relevant beyond one city as a case study. 100 kW is depot charging power levels, not for bus stop charging.
- “Section 4 - general comment: Where are the costs of the proposed solutions? If the objective function is cost minimisation then I certainly expect to see these costs compared to the business-as-usual case. This is a serious issue to address with this paper, i.e. a lot of time is spent on literature review and algorithm presentation but very little time is spent on presenting results.”
We have now re-run the optimizations for all seven scenarios increasing the maximum solving time from 24 h to 48 h per scenario. We have included the final objective values in Figure 4 also indicating the respective relative mixed integer problem gap (MIP‑gap). Please refer to the end of Chapter 3 for the optimization results, where we now address the specific results including the status‑quo scenario.
However, the differences in the total costs (the objective function) are not indicative of a best solution. This is because an exhaustive optimization would take much longer than 48 h per scenario, which did not fit our timeframe. Further, it is difficult to compare the scenarios, because they have slightly increasing complexities which lead to increased computation times. In theory, the additional deg. of freedom of having more CPC cannot worsen the result. But since computation time is limited to 48 h and the problem formulations that are more complex take longer to solve, they also take longer to find a comparable quality of results than models with a simpler structure would take.
A clear result we could identify is that the initial set of CPC – not necessarily the amount – strongly influences the whole optimization structure and duration.
We’d like to emphasize that the methodology itself is central to the paper. The results are based on the use‑case specific input parameters and, as stated in the paper, many assumptions had to be made. Price structures are simplified. The meaningfulness of these results is not considered as relevant as the methodology itself is for the academic community trying to tackle this kind of problem.
Thank you for your effort put in updating this section. What is meant by “did not fit our timeframe” that is mentioned above? Do I understand correctly that you have not found optimal solutions for all cases except the status quo? If the limited solving time is the reason it has to be solved in two ways: either increase solving time or restructure the problem. Presenting non-optimal results for an optimization problem is self-contradictory in my opinion.
- “Section 5: This section should be focusing on the actual conclusions of the research done in this paper but looks more like a general discussion of potential next steps. Please reformulate to provide clear conclusions/answers to specific research questions linked to the work presented in this paper.”
In accordance with our previous comment to point 18, we’d like to add that our conclusions don’t take the shape of specific numbers. At the beginning of Chapter 5 we mention our contribution to the academic world. Then, we mention important conclusions that other researchers or fleet operators should take into account when confronted with a charging infrastructure planning problem. This goes beyond our use‑case results. Important aspects are: there is no one‑fits‑all solution; fleet operators have significant experiential knowledge that can serve for narrowing down the problem; long computation times are to be expected; and, finally, in the bigger picture of the sector coupling domain, cost structures in general have to be used with care given the possible involvement of other stakeholders and contractual possibilities for cost allocation.
Again, there have been studies proving these results with more robust analyses in the past. The authors mention in the literature reviews such studies. An additional effort should be placed on identifying the unique originalities of this paper.
- Doubts on novelty, no motivation / lack of specific aim and conclusions. / Paper structure unclear; is it a literature review or model implementation paper?
A general comment on motivation, novelty, and structure of the paper:
We think the motivation for this kind of paper is strong enough. In the developed world, the electrification of the transportation sector is omnipresent. Few fleet operators have made their first steps towards this goal. The bigger share, however, will likely do so in the coming decade.
We think it is important to provide the reader with the necessary background to the problem (stakeholders, perspectives), in order to understand where the complexity arises. The novelty of the paper lies in the clever combination and integration of what is important for an efficient and meaningful problem solution. Jefferies et al. (MDPI, 2020) have also segmented their problem and combined different methodologies into an overall framework and supported their model implementation by a thorough literature review. The present paper is not as extensive, but the structure is similar.
The specific results for the use-case of Darmstadt are not of central relevance to the academic community. Nevertheless, we have now included the final objective values. For the conclusions see point 19.
If as you mention Jefferies et al, 2020 do a more comprehensive study of the same topic what is the purpose of this paper then? Please notice that Jefferies et al. use charging powers up to 450 kW, while here we have the very unrealistic assumption of maximum 100 kW charging power. Please motivate relevance, novelty, and scientific strength in a more comprehensive way so that it is evident that you are adding to the extant literature body.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Round 3
Reviewer 2 Report
Thank you for the clarifications and congratulations for the interesting study!
