Intelligent Parametric Optimization of Building Atrium Design: A Case Study for a Sustainable and Comfortable Environment

Round 1

Reviewer 1 Report

This paper deals with a parametric intelligent performance optimization framework for atrium design to optimize its geometry through the balance between the effects of the lighting environment, energy use intensity and thermal comfort in terms of UDI (useful daylight illuminance), EUI (energy use intensity) and PPD (predicted percentage dissatisfied). The Rhino and Grasshopper platform is used to simulate the above evaluation indexes. The following five parameters are chosen as atrium parameters related to the three evaluation indexes, i.e. area ratio, shape ratio, volume ratio, position index X and position index Y. Through the case study in a real building with atrium, the paper presents the effectiveness of the proposed framework for supporting the designer to optimizing building atrium design from trade-offs between the three evaluation parameters. For possible publication, it requires to improve the manuscript under consideration of the following comments.

1. It needs a discussion on the deviation of the simulated results due to simplification of the building model from a polyhedron to a rectangle.

2. It is necessary to clearly define whether the simulated evaluation indexes (UDI, PPD) are averaged results in the analyzed building space.

3. A more detailed guideline is required to optimize building geometry based on tradeoffs between the simulated results for the three evaluation indexes (UDI, EUI, PPD).

4. Simulated results from the case study show that variation ranges of the three evaluation indexes are 55.0-57.0% for UDI, 121-125kWh for EUI, 23.9-24.1% for PPD, respectively. Variation potential for each evaluation index is too small to limit optimization possibility. A discussion on this problem should be added.

5. Whether is possible to provide simulated results of the evaluation indexes for each zone when the analyzed building is divided into several zones on a floor.

Author Response

Reviewer #1: This paper deals with a parametric intelligent performance optimization framework for atrium design to optimize its geometry through the balance between the effects of the lighting environment, energy use intensity and thermal comfort in terms of UDI (useful daylight illuminance), EUI (energy use intensity) and PPD (predicted percentage dissatisfied). The Rhino and Grasshopper platform is used to simulate the above evaluation indexes. The following five parameters are chosen as atrium parameters related to the three evaluation indexes, i.e. area ratio, shape ratio, volume ratio, position index X and position index Y. Through the case study in a real building with atrium, the paper presents the effectiveness of the proposed framework for supporting the designer to optimizing building atrium design from trade-offs between the three evaluation parameters. For possible publication, it requires to improve the manuscript under consideration of the following comments.

Response: Many thanks for your comment. We have revised the manuscript carefully and marked all the changes in red. Please refer to the detailed response to each comment and the revised manuscript.

1. It needs a discussion on the deviation of the simulated results due to simplification of the building model from a polyhedron to a rectangle.

Response: Thanks for your comment. Since we proposed a novel geometry mapping method for the building model, which can simplificate the irregular atriums into rectangular ones to facilitate the simulation and optimization process. The deviation of the simulated results is added as Section 3.5 ‘Mapping validation for the tested building’. We select three models for simulation and compare the simulation result of the original model and the model after mapping. Add the figure like follows.

Through the validation mapping experiment, the coefficient of determination (R²) is 0.991, 0.818, 0.960 for UDI, wEUI and sDTP, respectively. The mapping method basically controls the simulation deviation within a constant range. Please refer to lines 458-493 page 14-15 of the revised manuscript.

2. It is necessary to clearly define whether the simulated evaluation indexes (UDI, PPD) are averaged results in the analyzed building space.

Response: Thanks for your comment. Considering the PPD value may have extremely large values, the average value cannot represent the general condition of the occupants’ thermal comfort; we select the discomfort time percentage (DTP) as the thermal comfort simulation metric instead, which calculate the percentage of the time when PPD value is larger than 20%. The evaluation indexes (UDI, EUI and DPT) all use the average value of the simulation period. We declare that in the lines 355-356 page 10 and 409-414 page 12 of the revised manuscript.

3. A more detailed guideline is required to optimize building geometry based on tradeoffs between the simulated results for the three evaluation indexes (UDI, EUI, PPD).

Response: Thank you for the advice. We add a calculation method to find the most optimal solution based on the tradeoffs between the simulated results. The solution that achieves the highest score can be considered as the overall best solution, which gains the balance of the three objective improvement. Please refer to lines 594-596 page 20 of the revised manuscript.

4. Simulated results from the case study show that variation ranges of the three evaluation indexes are 55.0-57.0% for UDI, 121-125kWh for EUI, 23.9-24.1% for PPD, respectively. Variation potential for each evaluation index is too small to limit optimization possibility. A discussion on this problem should be added.

Response: Thanks for your comment. The optimization potential of the previous optimization is relatively low for the sake of the simulation setting and the climate zone. We modify the optimized parameters, simulation settings and the simulation metrics and conduct a new performance optimization.  The specific modifications and thoughts as follows:

• In order to improve the optimization potential, the search space should be enlarged. We introduce the ‘well index’ to substitute the ‘area ratio’, which may lead to bigger area variation of the atrium. Additionally, the inner interface window-to-wall ratio is also introduced as the optimized parameter, as it affects the window size and may significantly influence the optimization result. Please refer to lines 221-240 page6 and lines 277-282 page8 for details.
• To investigate the building atrium design’s influence on EUI, we choose the heating season as the simulation period, which may show more dominant changes in the average value of the simulation period, and the metric is recorded as winter EUI (wEUI). And the simulation period for discomfort time percentage (DTP) is set as the non-heating season, the simulation metric is recorded as summer DTP (sDTP). Please refer to lines 400-416 page12 for details.
• Since the PPD value may have extreme large values, the average value cannot represent the general condition of the occupants’ thermal comfort; we select the discomfort time percentage (DTP) as the thermal comfort simulation metric instead. DTP is defined as the percentage of time when PPD is larger than 20%. From the simulation result of DTP, it shows more clear changes, that will contribute to achieving more accurate optimization results. Please refer to lines 324-341 page9 for details.

     For the new conducted performance optimization experiments, the result shows that the UDI value varies from 24.7 -36.03%, with an improvement of 43.20%; wEUI varies from 121.05 - 139.46 kWh, with an improvement of 15.52%; sDPT varies from 56.15 - 58.53%, with an improvement of 3.89%. The UDI and wEUI show significant improvement, but sDPT is still low due to the limitation of the climate. Since Poland is in a moderate climate and it has a long and cold winter, the occupants’ thermal comfort is more rely on the HVAC system. That explains why the improvement of sDPT is relatively low in this case study. We add the discussion in the result analysis part and the conclusion. If we select the studied buildings located in cold winter and hot summer climate, the optimization improvement may be improved. This can be studied in future research.

5. Whether is possible to provide simulated results of the evaluation indexes for each zone when the analyzed building is divided into several zones on a floor.

Response: Thanks for your comment. To achieve high efficiency of the optimization, we simply divide the analyzed building into two zones; each floor is a zone. The simulation result of each zone is represented in Section 3.5. It takes the lighting simulation as an example. Please refer to Figure 8 page 14 of the revised manuscript.

Author Response File: Author Response.docx

Reviewer 2 Report

The specific modification suggestions are as follows:

(1) In the introduction, the innovation of this study is lacking. In other words, what is the difference between this study and existing studies? Which way did you innovate?

(2) Some English sentences need further verification. For example, in line605, is it little or a little? Please check it.

(3) There are many abbreviations in the article, such as UDI/RUI/PPD/GA/PMV/SR/VR/AR. It is suggested to make a unified explanation at the end of the text.

(4) In Figure 10, it is difficult to determine the specific value from the height of the column. In order to facilitate readers' understanding, the author should set the vertical coordinate in the figure to the unified maximum value, and mark the specific value on the column chart.

Author Response

Dear Editor and Reviewers:

We have received your letter and carefully read it.

We highly appreciate your valuable advice for improving the quality of our manuscript. Based on your comments and suggestions, we have revised the manuscript carefully and marked all the changes in red.

Responses are listed below:

Comments from the reviewers:

Reviewer #2:

1. In the introduction, the innovation of this study is lacking. In other words, what is the difference between this study and existing studies? Which way did you innovate?

Response: Thank you for your comments.  We have revised the manuscript carefully and marked all the changes in red. Please refer to the detailed response to each comment and the revised manuscript.

We have reworked the ‘Introduction’ section to better highlight the innovations of our research. Firstly, we listed three research gaps based on the brief literature review, and then come up with the related way we considered to fulfill the gaps, which also constitute the aim and work of this research. After that, we further conclude the innovation points in the following sentences, from line 139-144 page 4, as follows.

“In addition to achieving comprehensive cross-platform simulations and optimizations based on these three objectives, this study also stands out due to its parametric modeling method for geometry systems, which can be applied to general buildings, even with complex and irregular shapes, to efficiently search for the optimal design solutions during optimization process. It also has high generalization capability in dealing with the related atrium performance optimization problems. ”

2. Some English sentences need further verification. For example, in line605, is it little or a little? Please check it.

Response: Thank you for your kind reminder. We have carefully verified the sentences and words and asked someone excellent at English writing to help us review and modify the language.

3. There are many abbreviations in the article, such as UDI/RUI/PPD/GA/PMV/SR/VR/AR. It is suggested to make a unified explanation at the end of the text.

Response: Thank you for your advice. we have added a Nomenclature table at the end of the text. Please refer to lines 738-739 page 9 of the revised manuscript.

4. In Figure 10, it is difficult to determine the specific value from the height of the column. In order to facilitate readers' understanding, the author should set the vertical coordinate in the figure to the unified maximum value, and mark the specific value on the column chart.

Response: Thanks for your comment. We have re-illustrate the figure, and the vertical coordinate in the figure is set to the unified maximum value. Please refer to Figure 12 of the revised manuscript.

We hope these responses have clearly answered all questions.

Looking forward to hearing from you soon!

Best regards,

Yunzhu Ji

Author Response File: Author Response.doc

Reviewer 3 Report

This paper proposes a geometric mapping method, using software to simulate buildings with atrium.​ At the same time, the influence of factors include the area ratio, shape ratio, volume ratio and the position index of atrium on UDI, EUI, PPD of atrium building is studied. The logic of the article is clear, and the topic has good research value. However, there are some questions in the present work:

1.   In Table 5. Building atrium design variables, the initial value of V4: Position Index-Y is outside the range of V4. Please clarify whether this will have an impact on the results.

2.   Figure 6 named Standard deviation graph (top) and parallel coordinate plot (bottom) of MOO simulation. However, no parallel coordinate plot are seen in the figure.

3.   The content of lines 143 to 152 in the introduction is the same as the content of lines 163 to 171 in the Materials and Methods part. It is suggested to focus on the research methods in Chapter 2.

4.   Some of the software in Figure 1, such as Python and Honeybee, are not mentioned in the article. The authors should briefly introduce them.

5.   In Table 2, please explain the reasons for choosing PX and PY and the height of indoor light environment simulation.​

6.   In section 3.5.3, the author said “The Pareto front of EUI-Average UDI in Fig. 10 shows that EUI increases as Average UDI increases.” The conclusion does not correspond to the figure, please correct it.

Author Response

Response Letter

Dear Editor and Reviewers:

We have received your letter and carefully read it.

We highly appreciate your valuable advice for improving the quality of our manuscript. Based on your comments and suggestions, we have revised the manuscript carefully and marked all the changes in red.

Responses are listed below:

Comments from the reviewers:

Reviewer #3: This paper proposes a geometric mapping method, using software to simulate buildings with atrium.​ At the same time, the influence of factors include the area ratio, shape ratio, volume ratio and the position index of atrium on UDI, EUI, PPD of atrium building is studied. The logic of the article is clear, and the topic has good research value. However, there are some questions in the present work:

1. In Table 5. Building atrium design variables, the initial value of V4: Position Index-Y is outside the range of V4. Please clarify whether this will have an impact on the results.

Response: Sorry for the mistake. We have check and verify the initial value of the parameters. The newly calculated initial PY value is 0.04, and the modified variation range for PY is [-0.15,0.15], so the initial PY value is located in the range. Please refer to Table 6 and Table 7 on page 12 and page 14 of the revised manuscript.

2. Figure 6 named Standard deviation graph (top) and parallel coordinate plot (bottom) of MOO simulation. However, no parallel coordinate plot are seen in the figure.

Response: Sorry for the mistake. However the original Figure 6 was removed, because when describing the latest optimization results, we expressed the corresponding analysis results through other figures.

3. The content of lines 143 to 152 in the introduction is the same as the content of lines 163 to 171 in the Materials and Methods part. It is suggested to focus on the research methods in Chapter 2.

Response: Thank you for your comments. We have removed the duplicate contents in Section 1 and integrated the statements about research tools and platforms into Section 2.1. Please refer to line 158 -180 page 5 of the revised manuscript.

4. Some of the software in Figure 1, such as Python and Honeybee, are not mentioned in the article. The authors should briefly introduce them.

Response: Thank you for your comments. We have added the brief introduction to each tool involved in this research, which is in line with the content of Figure 1. Python is used in geometric calculation and geometric mapping, especially the custom implementation of some special calculation methods. Honeybee is the plugin in the Grasshopper platform, which is used to call the simulation engines, including EnergyPlus and Radiance. Please refer to line 160 – 168 page 5 of the revised manuscript.

5. In Table 2, please explain the reasons for choosing PX and PY and the height of indoor light environment simulation.​

Response: Thank you for your comments. In Section 2.3, we introduced the simulation metrics selected in this research, and also chose design parameters that stated in Section 2.2.2 to run the experiment simulation. As for UDI, we select the position index as variables for different simulation experiments, because the location of the atrium will affect the distribution of natural daylighting. When the window of the building itself is not evenly distributed on all facades, different lighting distribution may affect the UDI_100-2000 value, so we choose to use PX and PY for the lighting simulation to reveal the possible impact of the geometric parameters of the atrium on UDI. The height for the indoor light simulation is set as 0.75m, which is a height of the working plane. Please refer to line 293 – 299 page 9 of the revised manuscript.

6. In section 3.5.3, the author said “The Pareto front of EUI-Average UDI in Fig. 10 shows that EUI increases as Average UDI increases.” The conclusion does not correspond to the figure, please correct it.

Response: Sorry for the mistake. We have corrected the Figure number.

We hope these responses have clearly answered all questions.

Looking forward to hearing from you soon!

Best regards,

Yunzhu Ji

Author Response File: Author Response.doc

Reviewer 4 Report

Dear Authors,

First, I would congratulate the authors for putting in the effort of preparing a manuscript and carrying out the study. In my opinion, the study was well conducted, which reflects the quality of the manuscript. I understood the motivation and aims of the study, as well as the methodology adopted. The presentation of the results and discussion were clear. Generally, the manuscript was well structured and presented. My comments focus more on the methodological part and the interpretation of the results. These are my remarks for the review to improve the quality of the manuscript:

Title

The word “self-adaptive” does not fit in the work developed, since the simulation did not take into account any dynamic adaptation of the building throughout the year. Besides that, the title describes the work developed and is coherent with the manuscript content.

Abstract

The relevance of the topic is well described along with the study’s main objective. The methodology adopted is presented accordingly. The abstract main findings focus on PPD, but other metrics could be also mentioned.

The keywords are adequate.

1.     Introduction

In my opinion, this narrative of the first section is complete. It introduces well the problem of the subject and identifies an opportunity to carry out the study (the need to develop more automated approaches to design atriums for architects). With the research gap identified, the authors summarize the up-to-date studies regarding atrium optimizations and conclude the first section by stating the objectives of the study.

I recommend improving the following parts only in what concerns written English:

L56- L57: This sentence does not sound well. The optimization should be required or similar.

“To have a better design result for the environment, the optimization process is involved.”

L66: The “For” is not needed.

L69-L70: The sentence “… and make the environmental performance effectively influence the design” does not seem correct. I understand the underlying idea of the design process being influenced by environmental performance, but the way the sentence is written feels unnatural.

L77: I would add the references for each point, instead of having only three citations together. In this way, it would be easier for the reader to identify which study addressed the separate topics.

Table 1: It is difficult to read some rows. For example, Wu et al. the design variables (not centered) while the Researcher column is vertically centered. I would uniformize everything to increase the readability, e.g., everything horizontally left aligned and vertically centered.

2.     Materials and Methods

L240-L241: This is more of a curiosity. Can the mapping approach be used or adapted for buildings with different roof heights? In other words, can it be adapted also for buildings that do not have flat roofs with constant heights? This would increase the range of applications of the approach. And in this case, also the volume ratio would need to be calculated differently.

L273: I would remove the sentence where glare is mentioned since there are other metrics to evaluate glare (for example, DGP). What the study is measuring is vertical illuminance, glare analysis requires other types of data.

L283-284: One thing that was not totally clear to me is if you were evaluating the minimum lighting requirements in spaces. Did the methodology include an adaptation of the artificial lighting to meet the lighting needs of the space considering the contribution of natural light? From what I read in Table 6, the authors assumed a constant lighting load (W/m2), meaning that the artificial light was always constant. And in this case, there in no optimization concerning light.

In my opinion, for the EUI this could be included if the authors want to consider the lighting environment and the savings in electricity from the atrium optimization. This would be more of an adaptive approach as mentioned in the title of the manuscript.

If the authors find that this would require thorough changes to the simulations, I would not consider lighting nor UDI, since the problem of the mapping approach would have a residual impact only on the thermal performance of the building. The focus would be solely on thermal comfort.

L315-L316: When it is said: “The Honeybee model is used to simulate the value of the annual average PMV.”, I think it could be dangerous to evaluate annual averages of PMV since it varies from -3 to 3. Averaging could lead to a PMV average of zero, which indicates that the thermal environment is good, but actually, you could have several -3 and 3. I would count the number of hours for each PMV (-3, -2, …, 2, 3) and then convert that to the number of hours of PPD. The objective is to minimize the more severe bins (-3 and 3). Or evaluate PMV and PPD for each timestep, instead of annual averages.

The remaining part of the methodological section is clear to me and very well documented.

3.     Case study: Villa Reden in Katowice, Poland

L383: There are missing the units of clothing level (clo) and metabolic rate (met). Also in Table 6.

Table 6: What was the value of the assumed air velocity within spaces? This is also critical for the calculation of PMV.

Table 6: The envelope of the building seems very poor from a thermal perspective. Are the heat transfer coefficients of walls and glazing (single) the actual values of the real building? Are those characteristics typical in Poland or in accordance with the legislation?

These parameters have a considerable impact on the results of your work.

Table 6: Regarding the setpoints of temperature. I understand that there were used broad setpoints to evaluate the energy use of HVAC systems, but such setpoints are not used in the calculation of thermal loads for comfort standards. In other words, the study is not estimating the real energy use of the case study (so the conclusions drawn from such assumptions may not be valid).

My recommendation is to have two sets of simulations: (i) the first where you evaluate PMV and PPD without HVAC and their increase/decrease with the parametric analysis; and then, (ii) the second is to have a set of simulations with HVAC turned on with real setpoints (20 ºC and 25 ºC) and see the impact in terms of energy required.

In my opinion, it is not a good practice to evaluate PMV when you are already recurring to HVAC with poor temperature setpoints. If climate control systems are being used, they must ensure comfort with appropriate setpoints. This is why I recommend splitting up the simulations.

Section 3.5

I would split the EUI into two parameters, energy for cooling and heating since lighting is constant. In this way, it would be easier for the reader to understand where to maximize or minimize. For the climate of the case study, I am expecting that cooling loads are low when compared to heating ones, but it is not possible to evaluate this. Moreover, it is interesting to explore the optimal point of having glazing in this climate, since there are a lot of heating loads.

Regarding the average PPD, I am also afraid that using the average PMV, this metric is not the most adequate to make such an evaluation. Once again, there can be used other metrics, hours in discomfort (heating and cooling), or heating/cooling degree hours (ºC.degrees) where you can sum the difference between indoor temperatures and comfortable temperatures. In this way, it would be easier to evaluate where the optimization is going (reduce heating loads, increase light availability, etc.).

L469: Figure 8 instead of Fig. 10.

L470-L471: “This indicates that obtaining better indoor lighting performance will reduce the energy performance of the building” when this is stated, it can be confirmed if the approach included an adaptive dimming of artificial lighting according to the natural lighting and the required illuminance. The authors even mention this in L476, but it is not confirmed by numeric results.

Figure 8: I would change the axis of Figure 8 c), for the Pareto front to be in the lower part of the graph as it is in the remaining graphs a) and b).

L550: Figure 12 instead of Figure 14.

4.     Discussion

The discussion section content is rather important for the overall manuscript. A summary of the results was included along with the presentation of limitations. Everything is clear and straightforward.

If the authors do not carry out the lighting analysis considering the reduction of artificial light (and respective energy) it should be added as a limitation of the work.

5.     Conclusion

The conclusion is short. It could follow a more traditional structure and an independent section that could be read individually. It could include a sentence with the motivation of the work, the objectives, and a short description of the methodology. This would create added value for readers that want to have a short summary of the work.

Summarily, the content of the manuscript is interesting and adequate with the research gap. The study is well conducted, following a typical approach. Besides, there are minor points that are easily covered regarding the readiness of the manuscript, in English as well (mainly in the first section). Therefore, I recommend publishing after this original manuscript is changed, suggesting a major revision due to the mentioned reasons.

Author Response

Response Letter

Dear Editor and Reviewers:

We have received your letter and carefully read it.

We highly appreciate your valuable advice for improving the quality of our manuscript. Based on your comments and suggestions, we have revised the manuscript carefully and marked all the changes in red.

Responses are listed below:

Comments from the reviewers:

Reviewer #4: First, I would congratulate the authors for putting in the effort of preparing a manuscript and carrying out the study. In my opinion, the study was well conducted, which reflects the quality of the manuscript. I understood the motivation and aims of the study, as well as the methodology adopted. The presentation of the results and discussion were clear. Generally, the manuscript was well structured and presented. My comments focus more on the methodological part and the interpretation of the results. These are my remarks for the review to improve the quality of the manuscript.

1. Title: The word “self-adaptive” does not fit in the work developed, since the simulation did not take into account any dynamic adaptation of the building throughout the year. Besides that, the title describes the work developed and is coherent with the manuscript content.

Response: Sorry for the unclear expression. We use the word “self-adaptive” to refer generalization ability of the proposed parametric framework. It can be adapted to general buildings, especially dealing with irregular building forms based on the geometry mapping method, as mentioned in line 184-212 page 5-6. It seems may lead to misunderstanding about the contents, so we removed it from the title.

2. Abstract: The relevance of the topic is well described along with the study’s main objective. The methodology adopted is presented accordingly. The abstract main findings focus on PPD, but other metrics could be also mentioned. The keywords are adequate.

Response: Thanks for the comments. We have added the optimization performance of both three objectives in the abstract part. Please refer to lines 21 - 22 page 1 of the revised manuscript.

3. Introduction: In my opinion, this narrative of the first section is complete. It introduces well the problem of the subject and identifies an opportunity to carry out the study (the need to develop more automated approaches to design atriums for architects). With the research gap identified, the authors summarize the up-to-date studies regarding atrium optimizations and conclude the first section by stating the objectives of the study.

I recommend improving the following parts only in what concerns written English:

L56- L57: This sentence does not sound well. The optimization should be required or similar.

“To have a better design result for the environment, the optimization process is involved.”

Response: Thank you for your comments. This sentence does not sound well. But when I considered rewrite it, I found the meaning already narrated in this paragraph's first sentence. “Performance-based atrium design needs integrated comprehensive simulation and optimization of various quantifiable performances of atriums.” Please refer to lines 45 - 46 page 1 of the revised manuscript. So, I just removed this unnatural sentence here.

4. Introduction：L66: The “For” is not needed.

L69-L70: The sentence “… and make the environmental performance effectively influence the design” does not seem correct. I understand the underlying idea of the design process being influenced by environmental performance, but the way the sentence is written feels unnatural.

Response: Thank you for your comments. I have corrected the sentence into “To get rid of disadvantage of manual adjustment,  many studies combine the parametric design methods with building performance optimization, which can effectively locate the optimal design solutions by iterating the performance simulation and getting feedback from simulation results.” Please refer to lines 610 - 64 page 2 of the revised manuscript.

5. Introduction: L77: I would add the references for each point, instead of having only three citations together. In this way, it would be easier for the reader to identify which study addressed the separate topics.

Response: Thank you for your advice. I have add the references for each point to make it more easier to identify for the readers. Please refer to lines 71 - 72 page 2 of the revised manuscript.

6. Introduction: Table 1: It is difficult to read some rows. For example, Wu et al. the design variables (not centered) while the Researcher column is vertically centered. I would uniformize everything to increase the readability, e.g., everything horizontally left aligned and vertically centered.

Response: Thank you for your advice. I have modified the format of the table as you suggest. It looks much clear and easy to read. Please refer to Table 1 in page 3 of the revised manuscript.

7. Materials and Methods: L240-L241: This is more of a curiosity. Can the mapping approach be used or adapted for buildings with different roof heights? In other words, can it be adapted also for buildings that do not have flat roofs with constant heights? This would increase the range of applications of the approach. And in this case, also the volume ratio would need to be calculated differently.

Response: Sorry for the unclear expression about the geometry adaptive method. Here the geometry adaptive method just dealing with the atrium profile transformation in the 2d plane, it only changes the plan of the atrium and doesn’t deal with the geometry calculation in the height. In other words, we suppose the building has flat roofs. We can further extend the method in 3d level and try to be adapt to various kind of the roof in the future study. Of course, the volume ratio needs the be calculated differently under this condition. 

8. Materials and Methods: L273: I would remove the sentence where glare is mentioned since there are other metrics to evaluate glare (for example, DGP). What the study is measuring is vertical illuminance, glare analysis requires other types of data.

Response: Thank you for your advice. I remove the sentence here as you suggested, to avoid the confusion.

9. Materials and Methods: L283-284: One thing that was not totally clear to me is if you were evaluating the minimum lighting requirements in spaces. Did the methodology include an adaptation of the artificial lighting to meet the lighting needs of the space considering the contribution of natural light? From what I read in Table 6, the authors assumed a constant lighting load (W/m2), meaning that the artificial light was always constant. And in this case, there in no optimization concerning light. In my opinion, for the EUI this could be included if the authors want to consider the lighting environment and the savings in electricity from the atrium optimization. This would be more of an adaptive approach as mentioned in the title of the manuscript. If the authors find that this would require thorough changes to the simulations, I would not consider lighting nor UDI, since the problem of the mapping approach would have a residual impact only on the thermal performance of the building. The focus would be solely on thermal comfort.

Response: Sorry for the mistake. Since the lighting environment is evaluated in this study, the simulation result will affect artificial lighting consumption. The energy consumption should use the lighting simulation result as the input and then calculate the artificial lighting consumption rather than setting a constant lighting load. I correct the energy simulation setting by linking the lighting simulation result as the input.  Please refer to Line 401 - 403 page 12 of the revised manuscript. Since one of the research focuses is to evaluate the comprehensive environments of the atrium design, we chose to keep the lighting evaluation and correct the related simulation settings. The overall simulation and optimization experiment were all conducted again. The simulation and optimization are all be modified and re-analyzed accordingly.

10. Materials and Methods: L315-L316: When it is said: “The Honeybee model is used to simulate the value of the annual average PMV.”, I think it could be dangerous to evaluate annual averages of PMV since it varies from -3 to 3. Averaging could lead to a PMV average of zero, which indicates that the thermal environment is good, but actually, you could have several -3 and 3. I would count the number of hours for each PMV (-3, -2, …, 2, 3) and then convert that to the number of hours of PPD. The objective is to minimize the more severe bins (-3 and 3). Or evaluate PMV and PPD for each timestep, instead of annual averages. The remaining part of the methodological section is clear to me and very well documented.

Response: Sorry for the mistake. It is not appropriate to calculate the average value of PMV as the simulation metric. As you suggested, we use the PPD to evaluate thermal comfort. We calculated the time percentage when PPD is greater than 20%, and recorded it as discomfort time percentage (DTP) for the thermal comfort simulation metric.

11. Case study: L383: There are missing the units of clothing level (clo) and metabolic rate (met). Also in Table 6.

Response: Thank you for your advice. We have corrected it in Table 6.

12. Case study: Table 6: What was the value of the assumed air velocity within spaces? This is also critical for the calculation of PMV.

Response: Thank you for your comments. We set the air velocity speed as 0.1m/s for the thermal comfort simulation. We added the declaration in Line 414 – 415 page 12, and also added it in Table 6.

13. Case study: Table 6: The envelope of the building seems very poor from a thermal perspective. Are the heat transfer coefficients of walls and glazing (single) the actual values of the real building? Are those characteristics typical in Poland or in accordance with the legislation? These parameters have a considerable impact on the results of your work.

Response: Sorry for the mistake. We have modified the building constructions and materials using the program of ‘Honeybee Construction set by climate’, the typical construction material for the Poland is generated by the program. We checked the relevant material coefficients against the ASHRAE 90.1 2019 standard and found that they all meet the requirements. Please refer the modification in Table 6 in page 13.

14. Case study: Table 6: Regarding the setpoints of temperature. I understand that there were used broad setpoints to evaluate the energy use of HVAC systems, but such setpoints are not used in the calculation of thermal loads for comfort standards. In other words, the study is not estimating the real energy use of the case study (so the conclusions drawn from such assumptions may not be valid).

My recommendation is to have two sets of simulations: (i) the first where you evaluate PMV and PPD without HVAC and their increase/decrease with the parametric analysis; and then, (ii) the second is to have a set of simulations with HVAC turned on with real setpoints (20 ºC and 25 ºC) and see the impact in terms of energy required.

In my opinion, it is not a good practice to evaluate PMV when you are already recurring to HVAC with poor temperature setpoints. If climate control systems are being used, they must ensure comfort with appropriate setpoints. This is why I recommend splitting up the simulations.

Response: Thank you for your advice. In order to estimate the real energy use of the case study, we correct the setpoint as 20 ºC (heating) and 25 ºC (cooling). At the same time, we divided the whole year as two simulation periods, the heating season, and the non-heating season, simulate for EUI and DTP, respectively. The reasons are as follows:

(1) Poland has a long and cold winter, and the occupants’ thermal comfort relies significantly on the air-conditioning system during the heating season. The building form can hardly influence the thermal comfort result in the heating season. Also, as you mentioned, evaluating the thermal comfort metric when the HVAC is already set is not a good practice because the HVAC system will ensure comfort in a certain range. It is better to evaluate the thermal comfort metric under the condition without the HVAC system, which can better reveal the influence of building atrium design on occupants' thermal comfort. So we modified it to simulate the comfort metric (DTP) only in the non-heating season without an HVAC system.

(2) Since Poland has a long heating season, heating consumption will become the most important part of the energy consumption. Energy consumption in heating season can better revel the impact from the building atrium design. So we modified it to simulate the EUI only in the heating season with HVAC system turned on.

Please refer to line 400 – 416 page 12 of the revised manuscript.

15. Case study: I would split the EUI into two parameters, energy for cooling and heating since lighting is constant. In this way, it would be easier for the reader to understand where to maximize or minimize. For the climate of the case study, I am expecting that cooling loads are low when compared to heating ones, but it is not possible to evaluate this. Moreover, it is interesting to explore the optimal point of having glazing in this climate, since there are a lot of heating loads.

Regarding the average PPD, I am also afraid that using the average PMV, this metric is not the most adequate to make such an evaluation. Once again, there can be used other metrics, hours in discomfort (heating and cooling), or heating/cooling degree hours (ºC.degrees) where you can sum the difference between indoor temperatures and comfortable temperatures. In this way, it would be easier to evaluate where the optimization is going (reduce heating loads, increase light availability, etc.)

Response: Thank you for your comments. Firstly, I should explain why we didn’t split the EUI into cooling and heating consumption. Because we decided to integrate the lighting simulation result in the evaluation and optimization, as I explained in the ‘review response no.9’. The energy consumption in this research should contain the lighting, heating, and cooling consumption in total. It is not appropriate to ignore the lighting consumption when the lighting performance is considered. So, we decided not to split the energy consumption into heating and cooling. Secondly, we modified the comfort metric as ‘DTP’, which is declared in ‘review response no.10’.

16. Case study: Figure 8 instead of Fig. 10.

Response: Sorry for the mistake. We have corrected the Figure number.

17. Case study: L470-L471: “This indicates that obtaining better indoor lighting performance will reduce the energy performance of the building” when this is stated, it can be confirmed if the approach included an adaptive dimming of artificial lighting according to the natural lighting and the required illuminance. The authors even mention this in L476, but it is not confirmed by numeric results.

Response: Thank you for your comments. As ‘review response no.9’ declared, we modified the simulation setting to have an adaptive dimming of artificial lighting. The optimization result of the objectives is illustrated in Figure 11; the numeric results are also carefully analyzed in Section 3.6.2. Please refer to line 510 – 532 page 16 of the revised manuscript.

18. Case study: Figure 8: I would change the axis of Figure 8 c), for the Pareto front to be in the lower part of the graph as it is in the remaining graphs a) and b).

Response: Thank you for the advice, we have modified the pareto front figure make the front to be in the lower part of the graph. Please refer to Figure 10 in page 15 of the revised manuscript. The original Figure 8 was deleted, because when describing the latest optimization results, we expressed the corresponding analysis results through other figures.

19. Case study: L550: Figure 12 instead of Figure 14.

Response: Sorry for the mistake. We have corrected the Figure number.

20. Discussion: The discussion section content is rather important for the overall manuscript. A summary of the results was included along with the presentation of limitations. Everything is clear and straightforward.

If the authors do not carry out the lighting analysis considering the reduction of artificial light (and respective energy) it should be added as a limitation of the work.

Response: Thank you for your comments. Since we have already considered the reduction of artificial lighting in the energy simulation, it should not be added as a limitation of the work.

21. Conclusion: The conclusion is short. It could follow a more traditional structure and an independent section that could be read individually. It could include a sentence with the motivation of the work, the objectives, and a short description of the methodology. This would create added value for readers that want to have a short summary of the work.

Summarily, the content of the manuscript is interesting and adequate with the research gap. The study is well conducted, following a typical approach. Besides, there are minor points that are easily covered regarding the readiness of the manuscript, in English as well (mainly in the first section). Therefore, I recommend publishing after this original manuscript is changed, suggesting a major revision due to the mentioned reasons.

Response: Thank you for your comments. We have modified the conclusion part and added the content to make it an independent section that could be read individually. We briefly describe the overall work and then declare the methodology, the innovative points, and the objectives. Please refer to line 678 – 692 page 22 of the revised manuscript.

We have carefully verified the sentences and words and asked someone excellent at English writing to help us review and modify the language.

We hope these responses have clearly answered all questions.

Looking forward to hearing from you soon!

Best regards,

Yunzhu Ji

Author Response File: Author Response.doc

Round 2

Reviewer 4 Report

Dear Authors,

Enough effort was put during the revision of the manuscript and simulation results. In my opinion, the study is now ready to be pusblished considering the changes made. 

My best regards