An Agent-Based Model of Heterogeneous Driver Behaviour and Its Impact on Energy Consumption and Costs in Urban Space

Round 1

Reviewer 1 Report

Authors in this work present an energy consumption study, comparing EV/PHEVs and ICEVs simulating a fleet of vehicles in an artificial environment by an agent-based modelling (ABM) method and a 3D urban traffic simulator (UTS). The paper is interesting and well written, but I have some minor concerns for the authors:

a) Present better the emobpy model, as some readers could not know it.

b) Emphasize the novel aspects of your research at the end of the introduction section. They are not clear.

c) As you obtain similar results than the emobpy model, which is the main advantage of your method?

d) No weather parameters have been included in the model. Batteries behaviour are very sensitive to temperature. Justify why then your results could be considered valid.

e) Due to the computational effort, low number of agents and too many simplifications have been made to the simulation environment (only a flat system, etc.). Justify better the validity of your results. Can they be compared to those from a realistic scenario? What could we expect in a real case? Why not simulate a case with a lower number of agents, but a higher detail of the environment? And vice-versa?

According to the previous concerns, I suggest the Editor to propose a MAJOR REVISION until them are properly addressed.

Author Response

1. Present better the emobpy model, as some readers could not know it.
   
   Thank you for this comment. Regarding emobpy, we describe the model thoroughly in paragraphs starting from lines 250 to 269, where we cite the model multiple times, describe why it was created and what we expect when we compare it to our model.
   
   For the purposes of this study, we wanted to show how our model outputs compare to emobpy as it uses a mathematical model which can only present an aggregated depiction of energy consumption, while our solution using ABM is able to show each individual car and how they consume energy and show the global level, i.e., all cars. This is the main advantage, so therefore emobpy is mentioned here. We also provide links to the emobpy model. For the paper’s scope, emobpy is not very important; it is only used to demonstrate differences and comparison.
   
   
2. Emphasize the novel aspects of your research at the end of the introduction section. They are not clear.
   
   This is a very valid point. I managed to edit the final paragraph in the Introduction section; this now includes the main novelty points proposed by the article. Lines 46 to 57.
   
   
3. As you obtain similar results than the emobpy model, which is the main advantage of your method?
   
   A fair point, allow me to address this: as I discussed in the paper, emobpy is a mathematical model, so it presents you with an aggregated energy consumption of similar vehicle activity; this is because it is not agent-based. It doesn’t take into consideration the various entities that affect energy, such as the environment, other vehicles in the network and so on. We show that we can also produce similar outcomes as emobpy (refer to Figure 4) using an agent-based model while allowing practitioners to see individual level vehicle characteristics and demonstrate how energy is impacted by vehicle behaviour and vehicle density (refer to Figure 3). You cannot do this with emobpy. Furthermore, we were able to quantify the price for petrol consumption as well, which is also not available in emobpy refer to Figure 8.
   
   
   
4. No weather parameters have been included in the model. Batteries behaviour are very sensitive to temperature. Justify why then your results could be considered valid.
   
   A fair point. In lines 271 to 275, we talk about the limitations of our model and not including weather is one of those. We do not have the computational hardware to simulate weather in the model, as trying to do this with the current hardware would skew the results and produce invalid ones (this is because we have an agent-based model that is 3D and vast). As we wrote in lines 274-275, an area of future development is to explore how we can add weather to the model without it affecting the results. The results from this article are estimates and hypothetical scenarios. Therefore, it cannot be expected to reflect the real world but can infer a rough idea of how things can look if we have N number of vehicles in our street network. In paragraphs 342 to 353, we show how valid our results are compared to the official manufacturer's statistics from Mitsubishi and our model outputs are highly similar. In the Conclusion section, we discuss our study's limitations to make it clear to the reader.
   
   
5.  Due to the computational effort, low number of agents and too many simplifications have been made to the simulation environment (only a flat system, etc.). Justify better the validity of your results. Can they be compared to those from a realistic scenario? What could we expect in a real case? Why not simulate a case with a lower number of agents, but a higher detail of the environment? And vice-versa?
   
   As I mentioned earlier, our model is the only model to produce outputs similar to official statistics from vehicle manufacturers, i.e., Mitsubishi Outlander paragraph 342 to 353. Furthermore, we included details such as braking energy recovery (Figure A1), which also shows a strong correlation with real-world vehicle behaviour, such as the faster vehicles travel, when they break, they recover more energy and vice-versa. Furthermore, we know electric vehicles are more energy-efficient and are cheaper to run than petrol vehicles. Our model also shows this correlation in Figure 8. Currently, it would be too early to compare model outputs to real-world scenarios as the elevation of cities and road networks means vehicles must work harder or less depending on the road they are travelling on, which impacts energy consumption. However, for the scope of this paper, we build the foundation of the model, which can be further developed in future studies to include elevation, real-world street networks and so on.
   
   If we had reduced the number of agents but added real-world elevation data, our model would still become too complex to run, as elevation data has millions of data points and representing that in 3D space would be difficult. However, if we had only one agent, this may be possible, but then the results would not consider other drivers, and we could not compare our vehicles with one another, so the results would be meaningless. The strength of this model is that it allows the user to simulate heterogeneity among driver behaviour. This is not possible in other pre-existing mathematical models.

Thank you very much for your review. I found your sophisticated comments highly useful and hope the points I have made above and changes to the paper satisfy your requests.

Reviewer 2 Report

The manuscript focuses on a well-known problem of vehicle energy consumption in urban areas through simulation with an agent-based model. The authors do good work, but it has many mistakes and inaccuracies: methodological descriptions, many references are obsolete or with incomplete information, some figures have a too-small font, table 1 reports the vehicle number from 1 to 500 while Figure 5 has only 100 vehicles tested over the simulations, the number of figures within the text is not consecutive (for ex. figure 62 at line 334 page 12).
Moreover, many things are not adequately described: driving cycles, the fuel and energy consumption should always be compared in kWh/km or l/km, while the authors reach the results of different models with cumulative consumptions.
There is at least a severe mistake in the mathematical formulation with a heavy impact on the result reliability; the force balance of the vehicle misses the rolling resistance, affecting the energy consumption directly.
A few comments are added to the PDF.

Comments for author File: Comments.pdf

Author Response

Thank you for providing a thorough review. Your support for all aspects of the paper has been well received. With regards to comments, I managed to complete all comments you made. For all comments made in the pdf review, I commented and highlighted the location within the text that has changed.

Overview:

• methodological descriptions
  
  We made changes in the paper to make the methods clearer including changes to the physics formulae to include rolling resistance and clarification of various parameter selections.
  
  
• many references are obsolete or with incomplete information
  
  I updated the references that were obsolete and included the date accessed for those that did not include it.
  
  
• some figures have a too-small font
  
  I increased the font size for these figures.
  
  
• table 1 reports the vehicle number from 1 to 500, while Figure 5 has only 100 vehicles tested over the simulations
  
  We provided a description of why only 500 vehicles could be simulated, furthermore, the conclusion section dissects the limitations of the study which is related to this point. See PDF attachment for clarity.
  
  
• the number of figures within the text is not consecutive (for ex. figure 62 at line 334 page 12).
  
  We changed these to better reflect the subfigure being referenced. Thanks for pointing this issue out.
  
  
• many things are not adequately described: driving cycles, the fuel and energy consumption should always be compared in kWh/km or l/km, while the authors reach the results of different models with cumulative consumptions.
  
  I added a definition for drive cycle on lines 48-49, we also managed to change kWh to better represent the distance to kWh/km and l/km.
  
  

There is at least a severe mistake in the mathematical formulation with a heavy impact on the result reliability; the force balance of the vehicle misses the rolling resistance, affecting the energy consumption directly.

Thank you for highlighting this issue. We managed to change the formula to include the rolling resistance and also re-ran the simulations to include these updated results. Updated the (Appendix A Energy Calculation Extension)

Thank you again for your useful comments, your expertise has made this paper much better and more accurate. 

Author Response File: Author Response.pdf

Reviewer 3 Report

The topic addressed in the paper is potentially interesting, however, in the reviewer's opinion there are some major comments in the paper which should be addressed by the authors in detail. Below is a list of some comments:

1. What is driver behavior? What is the relationship between driver behavior and individual vehicle behavior? This question should be further explained in this paper.
2. There are eight phrases listed in the Keywords part. Some of them are repetitive, such as electric vehicle and electric mobility. Meanwhile, energy intake and fuel cost are not consistent with electric energy consumption which appears in the title.
3. In this manuscript, the main contribution is not clear enough and should be rearranged.
4. How many ABM output variables are there, 12 or 13? The Table 2 is not matched with Figure 1, so that the authors should modify them.
5. The title format of the picture in this article is inconsistent, and the picture references in this article are confused. For example “Figure 62” (page 12, line 334), it is easy to cause misunderstandings. Besides, the title of Figure 7-1 is not fully displayed.
6. In the result analysis, the author uses a large number of bar diagrams, which can not directly explain the performance differences of different models. Authors are advised to use other forms of presentation of results, such as charts.
7. There are some typos. Please check the whole paper carefully.

Author Response

Firstly, thanks for taking the time to conduct this excellent reviewer report. Your comments have been well received and all changes suggested have been made. I provided some brief explanations below point-by-point:

1. What is driver behaviour? What is the relationship between driver behaviour and individual vehicle behaviour? This question should be further explained in this paper.
   
   Thanks for this important point, I made it clearer throughout the paper and this point was made by previous reviewers as well. We made it clearer throughout the paper, if you take a look at lines 61 to 65, we define what driver behaviour is for context.
   
   
2. There are eight phrases listed in the Keywords part. Some of them are repetitive, such as electric vehicles and electric mobility. Meanwhile, energy intake and fuel cost are not consistent with electric energy consumption which appears in the title.
   
   This is a very valid point. I removed the repetitive keywords and edited the paper title to ensure “costs” are also present. Now there is no ambiguity in what to expect from the article. – The title is now: “AN AGENT-BASED MODEL OF HETEROGENEOUS DRIVER BEHAVIOUR AND ITS IMPACT ON ELECTRIC ENERGY CONSUMPTION AND COSTS IN URBAN SPACE”

3. In this manuscript, the main contribution is not clear enough and should be rearranged.
   
   We rearranged several parts of the manuscript upon comments from the previous reviewers, and the paper’s main contributions are clearer. In Paragraphs 41 to 65, we make it clear step by step what the paper is contributing, how it does it and the purpose behind it. The sentence: “This article contributes an energy calculation extension (Figure 1) which can be used in conjunction with the agent-based model [13] to quantify EV energy usage.” is very clear… and then we talk about the things that can be done i.e., quantify costs…. The purpose is to enable policymakers to forecast the future landscape of EV uptake in their relative cities to observe how much demand for electricity would have.
   
   
   
4. How many ABM output variables are there, 12 or 13? Table 2 is not matched with Figure 1 so the authors should modify them.
   
   Thanks for noticing this; there are 13 output variables; we missed one in the table, which was re-added.
   
   
   
5. The title format of the picture in this article is inconsistent, and the picture references in this article are confused. For example, in “Figure 62” (page 12, line 334), it is easy to cause misunderstandings. Besides, the title of Figure 7-1 is not fully displayed.
   
   Another reviewer spotted this; we changed the inconsistent subfigure labels to letters, so now it shows as figure 6a or 6b…Furthermore, for Figures 6 and 7, we fixed the overlapped captions so it's clearer now.
   
   
   
6. In the result analysis, the author uses many bar diagrams, which cannot directly explain the performance differences of different models. Authors are advised to use other forms of presentation of results, such as charts.
   
   In Figure 5, we show the distribution of energy consumption from different experiment scenarios; this is important to demonstrate how driver behaviour and density impact the energy consumption demonstrating the model’s validity to explore this phenomenon. Regarding the bar graphs that explore costs. Figure 8 clearly shows that PHEV technology is more cost-efficient than ICEV vehicles which we know is true from the literature; we can clearly see how for each experiment, this is true. Lastly, just for presentation purposes, we added costs from each vehicle agent to demonstrate the benefits of ABM over mathematical models. Figures 9 and 10 in the main body show that we can access each vehicle specifically. For the purposes of the paper, these bar graphs are useful and explore the model outputs effectively. The bar graphs in the appendix are only there for reference and are not used as core findings in the body of the paper. For the above purpose, we feel that these graphs effectively convey the message clearly and easily to average readers.
   
   
7. There are some typos. Please check the whole paper carefully.
   
   We managed to fix all the typos on the surface and used a text editor to help with this process.

Thanks again for your reviewer comments; if there is anything that is unclear, please do let us know.

Round 2

Reviewer 1 Report

The authors have improved their original manuscript, but my main concerns regarding the research, which are mainly that the proposed model assumes so many simplifications that, appart from the drivers behaviour, make the results not comparable to a real-world test, are not addressed by the authors. Then, I keep my original recomendation to the Editor (major revision), as, in my opinion, the model must be more realistic in order to show a real contribution to the state of the art.

Author Response

Dear Reviewer,

Thank you again for providing a subsequent review of your earlier comments. I have broken down your comments and replied to all comments in full:

- the proposed model assumes so many simplifications that, apart from the driver's behaviour, make the results not comparable to a real-world test, which are not addressed by the authors.

Our response:

There are simplifications this is true; however, like all models produced, no model is 100% a representation of a real-world system. For example, emobpy, which we described, does not include individual driver behaviour. Our model does not simulate weather as this is impossible to simulate or include in a model that utilises 3D modelling. No computer we can use can simulate this. We, as mentioned before, are proposing the building blocks for simulating real-world driver behaviour and the subsequent energy intake estimation. We justify why your proposal of simulating a real-world or comparing our model to the real world is not within the scope of the paper in the Conclusion section:

"A significant limitation of this work is computational tractability. The compute demand exponentially grows as we increase the number of vehicles or induce complex environmental settings. This can prevent users from simulating greater capacity of vehicles or more complex cities. Secondly, we assume the world as a flat plane in the model. However, this diverges from the real world. This assumption was due to computational demand, and we tried to configure the most simple environmental setting to ensure computation was not hampered. However, as cloud computing technologies become mainstream, this problem can be overcome." - 520 to 527

We also talk about the parts of the real world that we were able to include in our simulation:

"The environment consists of three road types with varying speed limits and intersections with right of way rules. The model environment is a simplification of the real world. Therefore, it does not capture all intersection types. However, it does contain the basic characteristics of an urban street network which have also been observed in several cities across the United States \cite{doi:10.1177/2399808318784595,Porta2006TheApproach, Thompson2020AStudy}. The vehicles also adhere to stop-go rules (conceptualisation of traffic lights) enforced at junctions. These rules are present in the vehicle's decision-making algorithm \cite{Olmez2021ExploringApproach}." - 239 to 245

This article is purely experimental and observational. We have shown that outputs from our model can be compared to real-world acquired statistics, such as the energy intake from Mitsubishi Outlander; we wrote:

"According to official Mitsubishi statistics \cite{OutlanderStatistics2021}, the range of the Outlander (kWh/km) is 0.169. To compare, the outputs in Figure~\ref{fig:energy_distance_both} show that as adherence increases (Experiment Conditions 3, 6, 9), the cumulative energy consumed aligns with the manufacturer's statistics. For example, in Experiment Condition 3, the average distance travelled is roughly 7 km. Therefore, given 0.169 kWh/km, the scenario mentioned above gives an average energy consumption of 1.183 kWh/km; this is observed in Figure~\ref{fig:energy_consumption_conditions}." - Lines 362 to 367

- the model must be more realistic in order to show a real contribution to state of the art.

This comment is highly ambiguous; how realistic can a model be? How do you quantify the realism of a model? We have mentioned that our work is exploratory and not representative of the real world. However, it can be later developed by practitioners who have access to HPCs to simulate real-world environments. We have presented novel contributions in the paper, firstly a 3D agent-based model that simulates traffic activity at various spatio-temporal resolutions using Unity (this has not been done before); secondly, we propose an energy calculation extension that not only quantifies fuel intake but also quantifies the relative costs which can be changed to represent the fuel costs of the current market, this has also not been done before. We show how powerful ABM is over mathematical models such as emobpy, which is also novel. 

As evidenced above, there is a substantial amount of novelty in this work; some of the things you have proposed are not within the scope of the paper. However, in future research, we will definitely take a look at these.

I hope the above has clarified the article's position and made it clearer that we are contributing novel techniques in the vehicle energy activity domain.

Reviewer 2 Report

The authors achieve significant improvements in the scientific soundness of the manuscripts; they partially bridge the gap and solve the mistake highlighted in the first turn by running new simulations. They still confuse the energy consumption (kWh or Wh) with the specific consumption (kWh/km).

There are more details to be explained:

• In the description of the parameters on page 4, the authors have right about heavy vehicles that do not have a hybrid or electric powertrain; besides, those vehicles still drive the same roads and often reduce the average speed of traffic flow. Avoiding them from simulation may lead to non-real results. How do you consider this problem?
• The font of figure 1 is still too small. Figure 5 fonts have been increased, but you can enlarge the y-axis scale or choose a different scale for each sub-graph. 
• The Comparison between UTS and Emobpy (page 8) is confusing; you may highlight it better with a table or bullet point with pros and cons.
• The figure 4 caption says, "Model Output Comparison: Electric Energy Consumption (kWh/km) for both UTS," while the results are reported in kWh. The same mistake affects Figure 5; the cumulative consumption is the total energy (which must be measured in kWh). In my previous revision, I suggested using energy specific consumption (kWh/km) to compare the results, so you have to divide each simulation's respective total (or cumulative) energy for all kilometres run by them. The same mistake is in lines 366-367, where you wrongly compare the energy-specific consumption (0.169 kWh/km, it is a reasonable value) with the energy consumption of 1.189 kWh?! Where is the similarity?
• In figure 6, the authors say they re-run all simulations to consider the rolling resistance (before missing); the results shown in Figure 6 are the same as the previous version (while Figure 5 has been updated). How could it be possible? Please consider avoiding separate graphs for cumulative energy and distance travelled (fig. 6 and 7) instead of the unique plot with specific energy consumption obtained by the division between the first and second.
• Line 414, what do you mean by the sentence "if the vehicle has driven 13km/s, the expected fuel consumption would be roughly 0.6l". Maybe there are typos in the measuring unit, or it needs further explanations.
• Line 418, please explain better why your model underestimates the fuel consumption when the adherence rise. 
• The discussion over the future trends of the vehicle market and its implications (autonomous vehicles, ride-sharing and so on) is fascinating, but it is off-topic in the manuscript. Please, consider concentrating on the simulations and clarifying the above points instead of focusing on unrelated topics (not mentioned in the title or abstract and in the introduction).
• I suggest improving the conclusions; this paragraph is the most important of the paper; it should include the shortcomings (as you do) and the main results achieved in your work and a few comments on them.

Author Response

Thank you again for providing us with a thorough review, we are glad the previous changes did contribute to the material. I have broken down your comments like before and made the necessary changes as suggested.  

• In the description of the parameters on page 4, the authors have right about heavy vehicles that do not have a hybrid or electric powertrain; besides, those vehicles still drive the same roads and often reduce the average speed of traffic flow. Avoiding them from simulation may lead to non-real results. How do you consider this problem?
  
  Great question, our article is focussing on only PHEV vehicles, we are aware that some heavier vehicles such as HGVs do not have electric or hybrid engines now, but in the future, they will. Furthermore, in the following article we see that the world’s trajectory is heading towards EV motor HGVs and heavy vehicles: https://www.greenbiz.com/article/8-electric-truck-and-van-companies-watch-2020 lastly, in our research title and discussion, we state that we are only interested in demonstrating energy expenditure of electric vehicles and/or ICEVs. The work is experimental, so the results are not used to influence policy but instead be used to develop further and incorporate these non-electric vehicles in the simulation results. The simple answer is non-EV vehicles is not within the scope of the paper.
  
  Also, on lines 144 to 146 we mention we said: “The rationale behind this distribution was to try intersect the EV and ICEV vehicle types which larger vehicles such as vans or trucks are not part of; the model distributes vehicles” outlining the scope is not including trucks and large vehicles.
  
  
• The font of figure 1 is still too small. Figure 5 fonts have been increased, but you can enlarge the y-axis scale or choose a different scale for each sub-graph. 
  
  Thank you for picking this up, we increased the text size for Figure 1 as requested. With regards to Figure 5, the scale captures a greater number of values on the y-axis, this helps with our messaging about showing the subtle differences between each simulation condition. Keeping it this way helps the paper tell the story better.
  
  
• The Comparison between UTS and Emobpy (page 8) is confusing; you may highlight it better with a table or bullet point with pros and cons.
  
  With regards to emobpy, it is only placed in the paper to demonstrate the similarities and differences between the two approaches. This section is a very small part of the article. To make it clearer, however, we did add bullet points from lines 282 to 289.
  
   
• The figure 4 caption says, "Model Output Comparison: Electric Energy Consumption (kWh/km) for both UTS," while the results are reported in kWh. The same mistake affects Figure 5; the cumulative consumption is the total energy (which must be measured in kWh). In my previous revision, I suggested using energy specific consumption (kWh/km) to compare the results, so you have to divide each simulation's respective total (or cumulative) energy for all kilometres run by them.
  The same mistake is in lines 366-367, where you wrongly compare the energy-specific consumption (0.169 kWh/km, it is a reasonable value) with the energy consumption of 1.189 kWh?! Where is the similarity?
  
  Apologies regarding this; we managed to change all the necessary graphs to compare kWh/km. Please refer to Figures 3, 4, 6 and 7.

• In figure 6, the authors say they re-run all simulations to consider the rolling resistance (before missing); the results shown in Figure 6 are the same as the previous version (while Figure 5 has been updated). How could it be possible? Please consider avoiding separate graphs for cumulative energy and distance travelled (fig. 6 and 7) instead of the unique plot with specific energy consumption obtained by the division between the first and second.
  
  As requested, we changed all figures including 3, 4, 6 and 7 to show kWh/km and we merged distance with the consumption thus we better represent distance and its impact on energy efficiency.
  
  
• Line 414, what do you mean by the sentence "if the vehicle has driven 13km/s, the expected fuel consumption would be roughly 0.6l". Maybe there are typos in the measuring unit, or it needs further explanations.
  
  Apologies about that, we meant to describe km’s as km/s (we changed these lines 416 and 417). Our answer comes simple from the fact that the engine provider states that 5.1l/100km for this particular vehicle, therefore 10kms would be 0.51litres. Thus, we see in our figure 7a and 7b for condition 1, on average, vehicles have consumed roughly 0.60+ litres of fuel in 13kms of distance travelled.
  
  
• Line 418, please explain better why your model underestimates the fuel consumption when the adherence rise.
  
  We added further explanation for why we believe this to be the case; please take a look at the new paragraph from lines 410 to 424
  
  
• The discussion over the future trends of the vehicle market and its implications (autonomous vehicles, ride-sharing and so on) is fascinating, but it is off-topic in the manuscript. Please, consider concentrating on the simulations and clarifying the above points instead of focusing on unrelated topics (not mentioned in the title or abstract and in the introduction).
  
  This article focusses on the future city where the current trajectory shows that EV uptake has exponentially grown. With this, comes novel technologies such as driverless cars, automatic speed adherence, platooning. Which all impact energy consumption or demand for energy. These are very important topics as they set the scene for the future paper that will look at how autonomous vehicles can demonstrate better fuel efficiency due to behaviour compared to non-autonomous vehicles. Our conclusion sets the scene for future research. We have also explored every point made above.
  
  In our concluding remark, we highlight the point of the article is to try estimate the future city which in our introduction we talk about i.e., policies shifting to EV technology uptake, cop26… We wrote:
  
  “This study highlights the importance of individual-based modelling methods such as ABMs in investigating future transport systems in cities. As some of the most important global policy agendas focus on the diffusion of low carbon-emitting technologies, this research is well-timed and crucial in planning for the future city.” – Thus, the AV domain is crucial for this point. Lines 539 to 542.
  
  
• I suggest improving the conclusions; this paragraph is the most important of the paper; it should include the shortcomings (as you do) and the main results achieved in your work and a few comments on them.
  
  We added several sentences to the conclusion section describing some of the main findings, Lines 488 to 497.
  
  

Reviewer 3 Report

 The author has made appropriate revisions based on the reviewer's comments and agrees to accept it.

Author Response

Dear reviewer,

In response to your comment " The author has made appropriate revisions based on the reviewer's comments and agrees to accept it."

We would like to thank you for agreeing to accept our article for publication.

Kind Regards. 

Round 3

Reviewer 1 Report

I appreciate the work of the authors in this contribution but, in my opinion, I still consider that the proposed method assumes too many simplifications that make me doubt about the real validity of the model. I understand that modelling more realistic conditions means a substancial increment in the computational effort, but in my opinion, I think it is not a scientific justification. The approach and limitations of the model must be adequately presented and the title, abstract and introduction of the paper must not promise a model resolution that finally cannot be achieved. Moreover, cloud computing services are accesible nowadays and could have been implemented.

The authors compare his model with emobpy model and claim that they obtain similar results. Then, if emobpy is more simple and computationally efficient, they do not present adequately the advantages of their model. I suggest them to highlight this part.

In conclusion, I clearly differ with the authors' opinion on the scope of their work, and they seem not to be able to apply my suggestions. Thus, I suggest the Editor to consider this work for its publication but considering clearly its limitations.

Finally, I would like to highlight that I have no conflict of interests with this work and my suggestions are purely scientific and with the aim to improve the authors' work. I suggest them to reconsider their position regarding this aspect and be more respectful with other colleagues' opinions and reviewers' comments.

Reviewer 2 Report

The authors improve the manuscript content and clarify all points that I noticed.

The actual manuscript has now robust scientific soundness and is clearly explained, it is suitable for publication. Further improvements in the conclusions could make the article even better.