Aerodynamic Optimization of a Reduced Scale Model of a Ground Vehicle with a Shape Morphing Technique

Round 1

Reviewer 1 Report

In the present manuscript, authors presents an aerodynamic optmization of a reduced scale model of a Sport Utility Vehicles (SUV) using a Kriging model.

The subject is interesting and worth of attention, but I noted some shortcomings that must be addressed by the authors before continuing with the review process.

The Introduction is rather complete and the problem at hand is clearly stated, I suggest to rephrase from line 55 to line 76 in order to clearly declare the goals of the work. Methods used seems to be appropriate to the aims of the study and clearly described. Section devoted to Mockup and experimental setup (section 2) description is quite exaustive. The description of the GPU calculations (lines 174-178) is very interesting, but it is necessary to give more details about this question and also to the questions related to the CFD software used. Sections devoted to Results (section 3) is interesting, figures and tables are easy to interpret. Maybe the section 4 (Discussion) should be renamed “Conclusions”.

Author Response

Dear reviewer,

We greatly thank you for your comments in order to improve and clarify the text of this paper. Answers to the comments are in red in the revised paper. We hope that these corrections will answer to the required modifications. We have also carefully read the text in order to correct maximum syntax and grammatical errors.

Thank you

Yours sincerely,

Philippe Gilotte

Answers to the comments :

The Introduction is rather complete and the problem at hand is clearly stated, I suggest to rephrase from line 55 to line 76 in order to clearly declare the goals of the work.

This part has been rephrase and integrates comments of other reviewers

Methods used seems to be appropriate to the aims of the study and clearly described. Section devoted to Mockup and experimental setup (section 2) description is quite exaustive. The description of the GPU calculations (lines 174-178) is very interesting, but it is necessary to give more details about this question and also to the questions related to the CFD software used.

CFD domain and boundary conditions are now described and a figure has been added

Sections devoted to Results (section 3) is interesting, figures and tables are easy to interpret. Maybe the section 4 (Discussion) should be renamed “Conclusions”.

The name of this chapter has been changed

Author Response File: Author Response.docx

Reviewer 2 Report

Optimization by adjoint method has been well-researched for some time now, and proves its efficiency either for finite volume or least-square Galerkin software. However, this method is still under development for direct coupling with Lattice Boltzmann solvers. It is therefore necessary to choose other methods dealing with design-of-experiments (statistically-designed models) and with optimization techniques applied on surrogate models. The manuscript reports on work to associate radial-basis-function morphing techniques with a Kriging model and includes guidance from a wind-tunnel test campaign focusing on shape optimization of the front undertray to minimize vehicle drag coefficient as the objective for a generic SUV.  Computations were performed using a commercial GPU-based Lattice Boltzmann solver that underscores the increase in computing capability, although the compute time for the volume of cases required for the optimization technique is still non-negligible (1 month cited); nonetheless, it is a good milestone to report for latest capabilities.

The word appears to be a sound application with novel integration of a CAE toolset.  However, this reviewer recommends a more-than-minor revision to improve the poor quality of images and often detracting technical writing that must be clarified to be acceptable for publication.

Detailed comments:

• Abstract: “Fluid Dynamic Computations (CFD)” – awkward arrangement of acronym; “locale” – local; “depending of the geometric parameters” – depending on; and later on P2 L44, “adjoin” – adjoint.  There are more grammatical/typographical errors throughout.  Please do a careful proofreading of manuscript.  It detracts from understanding the work.  Even immediately in the title, “technics” should be “techniques”!
• “According to the reference pressure taken at the Pitot 97 tube localized at a height of 1034 mm above the floor and at 630 mm ahead the front 98 bumper (figure 2), pressure measurements of these two rear surfaces lead to a partial Cd ratio of 45%. Therefore, a decrease of one of these two values must decrease significantly 100 the drag value of the mockup, in case the rear pressure ratio remains in the same range.” - The writing here is difficult to understand at this point.  Explain precisely what you mean for the readers.
• Figure 2: Poor image resolution; the labels are barely legible on my screen, and there is an obvious misspelling (“refence”).
• Figure 3: Clean up the labels with red underlining.
• P4 L128-129: “At time iteration…” – this sentence has grammatical error and is missing multiplication symbol on time step value.
• P4 L136: “These velocity profiles have been measured for the baseline configuration, with a step of 0.2 mm starting at 4 mm from the wall.”  The 4-mm offset from the wall should be more clearly indicated in Figure 4; it is misleading how the “relative” profiles appear to go to wall-normal distance of zero, including the statement of “increase of the boundary layer thickness from a relative normal distance of 5 mm for sensor 3 until 15 mm for sensor 8” (poor presentation).
• Figure 5: Image quality is not acceptable for publication.
• P5 L154: “remainder” – reminder.
• Figure 11: There are two different fonts used in the labels.  Red underlining present.  Choose better colors for color-impaired readers…
• Figure 12: Why isn’t the colorbar labeled?!
• P9 L270: “A better comparison between numerical and experimental rear pressure will help reducing this error in the vertical axis.”  This is carelessly offered without any insight into the systematic differences… please explain or remove the statement.
• Figure 14: Image quality is not acceptable for publication.  Labels are illegible!
• “CAESES” is never properly defined as to what it is. For completeness, please clarify for readers.  Proper copyright should be indicated with the registered (R) symbol.

Author Response

Dear reviewer,

We greatly thank you for your comments in order to improve and clarify the text of this paper. Answers to the comments are in green in the revised paper. We hope that these corrections will answer to the required modifications. We have also carefully read the text in order to correct maximum syntax and grammatical errors.

Thank you

Yours sincerely,

Philippe Gilotte

Answers to the comments :

Optimization by adjoint method has been well-researched for some time now, and proves its efficiency either for finite volume or least-square Galerkin software. However, this method is still under development for direct coupling with Lattice Boltzmann solvers. It is therefore necessary to choose other methods dealing with design-of-experiments (statistically-designed models) and with optimization techniques applied on surrogate models. The manuscript reports on work to associate radial-basis-function morphing techniques with a Kriging model and includes guidance from a wind-tunnel test campaign focusing on shape optimization of the front undertray to minimize vehicle drag coefficient as the objective for a generic SUV.  Computations were performed using a commercial GPU-based Lattice Boltzmann solver that underscores the increase in computing capability, although the compute time for the volume of cases required for the optimization technique is still non-negligible (1 month cited); nonetheless, it is a good milestone to report for latest capabilities.

Some sentences of this paragraph have been used to rephrase the introduction

Detailed comments:

• Abstract: “Fluid Dynamic Computations (CFD)” – awkward arrangement of acronym; “locale” – local; “depending of the geometric parameters” – depending on; and later on P2 L44, “adjoin” – adjoint.  There are more grammatical/typographical errors throughout.  Please do a careful proofreading of manuscript.  It detracts from understanding the work.  Even immediately in the title, “technics” should be “techniques”!

Reading has been more carefully read, with special care on the mentioned expressions

• “According to the reference pressure taken at the Pitot 97 tube localized at a height of 1034 mm above the floor and at 630 mm ahead the front 98 bumper (figure 2), pressure measurements of these two rear surfaces lead to a partial Cd ratio of 45%. Therefore, a decrease of one of these two partial Cd values must decrease significantly 100 the drag value of the mockup, in case the rear pressure ratio remains in the same range.” - The writing here is difficult to understand at this point.  Explain precisely what you mean for the readers.

The text has been modified in order to be clearer

• Figure 2: Poor image resolution; the labels are barely legible on my screen, and there is an obvious misspelling (“refence”).

Figure has been corrected

• Figure 3: Clean up the labels with red underlining.

done

• P4 L128-129: “At time iteration…” – this sentence has grammatical error and is missing multiplication symbol on time step value

An error occurs in the step value. This paragraph has been corrected.

• P4 L136: “These velocity profiles have been measured for the baseline configuration, with a step of 0.2 mm starting at 4 mm from the wall.”  The 4-mm offset from the wall should be more clearly indicated in Figure 4; it is misleading how the “relative” profiles appear to go to wall-normal distance of zero, including the statement of “increase of the boundary layer thickness from a relative normal distance of 5 mm for sensor 3 until 15 mm for sensor 8” (poor presentation).

This paragraph has been modified in order to focus more precisely on experimental and numerical comparison of velocity profiles.

• Figure 5: Image quality is not acceptable for publication. Image 5 have been replaced
• P5 L154: “remainder” – reminder. corrected
• Figure 11: There are two different fonts used in the labels.  Red underlining present.  Choose better colors for color-impaired readers…

Font is now homogeneous and dark colored of the background has been choosen

• Figure 12: Why isn’t the colorbar labeled? Label is added
• P9 L270: “A better comparison between numerical and experimental rear pressure will help reducing this error in the vertical axis.”  This is carelessly offered without any insight into the systematic differences… please explain or remove the statement.

This paragraph has been modified in order to explain this difference

• Figure 14: Image quality is not acceptable for publication.  Labels are illegible!

Figure has been arranged in order to read the labels

• “CAESES” is never properly defined as to what it is. For completeness, please clarify for readers.  Proper copyright should be indicated with the registered (R) symbol.

The register ® has been added saying also that it is a optimization software. Idem also for ultraFluidX

Author Response File: Author Response.docx

Reviewer 3 Report

DOE and MOGA was applied on the paraments optimization of the air dam for drag reduction, which is a pregnant ideal for drag optimization. The select designs was verified by experiment. The research is an interesting topic with great scientific significance, but many indistinct expressions including the figures decrease the brightness. Some issue could be improved are listed as following:

1. What the result expressed in abstract seems so plain and apparent(L21-L23). In other word, nothing innovative was found about the mechanism of air dam. Hence I suggest that the authors could emphasize more performance of Sobol algorithm or Kriging model rather than result of the result of air dam.
2. It seems no necessary to annotate references in abstract.
3. L158

         The difference can be explained due to pressure space averaging in CFD different from point averaging in experiments. Why not choose the same probe and space averaging to evaluate the difference.？

4. Annotations and labels in figure 3,4,5,7 and 14 could be more clear.
5. In Result section, the word “error” occurs such as error side pressure at L259. However, the concept or description is necessary. In addition, more interpretation about figure 12 and 13 needs to supply, at lease the meaning of label and color map.

Author Response

Dear reviewer,

We greatly thank you for your comments in order to improve and clarify the text of this paper. Answers to the comments are in blue in the revised paper. We hope that these corrections will answer to the required modifications. We have also carefully read the text in order to correct maximum syntax and grammatical errors.

Thank you

Yours sincerely,

Philippe Gilotte

Answers to the comments :

1. What the result expressed in abstract seems so plain and apparent(L21-L23). In other word, nothing innovative was found about the mechanism of air dam. Hence I suggest that the authors could emphasize more performance of Sobol algorithm or Kriging model rather than result of the result of air dam.

Some modifications have been made in paragraph 2.3 and in paragraph 3 in order to explain limitations of the optimization process and how to decrease uncertainty of the prediction.

1. It seems no necessary to annotate references in abstract.

We have difficulties to remove references from the abstract. We thought that it could help understanding the different optimization technics. We can of course remove them if necessary

1. L158  The difference can be explained due to pressure space averaging in CFD different from point averaging in experiments. Why not choose the same probe and space averaging to evaluate the difference

Comparison of the rear pressure at the probe location has been added. It enables to compare experimental and numerical rear pressure values. The difference is now small.

1. Annotations and labels in figure 3,4,5,7 and 14 could be more clear.

Figures 4, 5 and 7 and 14 has been modified to be more clear

1. In Result section, the word “error” occurs such as error side pressure at L259. However, the concept or description is necessary. In addition, more interpretation about figure 12 and 13 needs to supply, at lease the meaning of label and color map

Explanations have been added in the text related to figure 10 in order to explain the error definitions. Legend of the colorbar has been added in figure 12 and 13. The text related to figure 12 and 13 has been modified to be clearer

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The Authors have addressed all of my concerns with the original manuscript.

Reviewer 3 Report

All the comments were replied clearly and the whole paper was improved. As far as I can see,however, references hardly present in abstract part and I insist that the authors could remove them.

I recommend to accept after in present form  if English language and style could be accepted.