Integration within Fluid Dynamic Solvers of an Advanced Geometric Parameterization Based on Mesh Morphing

Round 1

Reviewer 1 Report

This work describes a procedure for the integration of shape optimization employing RBF-based mesh morphing in the commercial package STAR-CCM+. By reading the manuscript, it becomes clear that the authors master the techniques involved, but a clarification of the innovations introduced by this study should be made from the very beginning to the reader.

Specific comments follow:

1 – There may be too many acronyms in the title of the manuscript: RANS, RBF; however, the title is already long even without defining a less-standard acronym such as RBF. In addition, many other acronyms (CFD, CAD, CAE, …) are not defined in the text when these are used for the first time.

2 – Lines 66-68: Alternative methods are suggested, but a more detailed discussion of the advantages and disadvantages of RBF with respect to those other approaches is still missing in the Introduction.

3 – Lines 67-68: Free-Form Deformation (FFD); Radial Basis Functions (RBF); all terms should be uppercase.

4 – Line 79: Please use “such as” rather than “as”.

5 – Lines 176-180: The use of RBF raises concerns regarding computational cost; how does it compare with other approaches on this basis? In addition, at least a reference to previous work should be provided regarding parallel implementation of the method.

6 – Subsection 4.1.1: Despite given references, more information about the definition of the parameters used in the shape optimization should be presented so that the reader may adequately follow this work.

7 – Line 310: “A mesh slightly larger than 1.1 million of elements” – this sentence is not precise from the engineering point of view. Has this mesh resolution proven to provide mesh-independent solutions?

8 – Lines 353-354: “about one million”; “almost 200.000” – Same comment as above. In addition, beware that the use of the decimal point would mean “two hundred” (not “two hundred thousand”) in the latter case.

9 – Lines 357-370: The contents of the paragraph is not clear and it must therefore be improved. Why is a multiphase approach needed at this stage? Additionally, the grounds for the choice of various values selected in this paragraph should be given.

10 – Author names in the References list are not consistent in form. A careful revision is advised.

Author Response

We express our gratitude to the reviewer for the suggestions provided to improve the manuscript

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper has interesting elements, but some revision is needed. The English needs to be improved. 16 out of 39 references from the authors, and one of them is in Italian. Please check this and reduce the citations of own work.

Specific comments

Line 36, replace '.. user manual interactions' by '.. user intervention'

Line 58, Chimera methods are in aerospace used for very complex geometries, please correct the statement

Line 116, it is not clear what the authors want to say with ' ... gained computing ...' Please rephrase

Line 161/Table 1, for completeness, define r and Rsup

Line 168, Fast Multipole Method is not explained, please provide some more details

Line 172, I assume that the interaction radius is Rsup? It this is the case please provide this detail, if not please provide more details

Line 321, please explain the optimization objectives better, I guess that it should be a reduction in total drag and lift coefficients???

Line 328, please indicate the two optimal points in Fig. 6. Also, Fig 6 shows 8 optimal designs points, so this is confusing.

Line 360, boundary conditions are imposed, not established

Line 383, please provide more details on the optimization studies, the selection of the 25 candidates for the DoE, why these 25 candidates etc

Fig. 10, it is not possible to read the scales

Author Response

We express our gratitude to the reviewer for the suggestions provided to improve the manuscript

Author Response File: Author Response.pdf

Reviewer 3 Report

RE Integration Within RANS Solvers of an Advanced Geometric Parameterization Based on RBF Mesh Morphing for Shape Optimization

The authors present a paper that is great in premise but very poor in execution. After reading the introductory and theory part, I was particularly dissapointed with the amount of detail or lack of it given in the methodology and results sections. Because of these it is impossible to judge the quality of the work done as it appears the authors were withholding details. Yes this can be done but can be done by non-dimensionalising for example rahter than leaving wholesale useful detail out which will enable your results to be verified. Also no validation was given. For these reasons the amount of work needed is more than what will be required for a major revision, rather it is more of a case of resubmission. Hence, the reviewer suggests a rejection as in the current form, it is not written like a research paper but like an informational white paper. In addition to these comments, the authors should refer to the following to help them in revising the paper for resubmission here or elsewhere?

1. the introduction is too sudden, going straight into parametric modelling. A "big picture" background is needed.

2. There are too many commercial references to Volvo and Star CCM+. In a research paper, these need to be minimised to avoid the appearance of it being an "advertisement" or product placement

3. section 4.2.3: please start the paragraph with "Twenty-five" rather than "25".

4. The details of your CFD should be shown in figures and properly explained. For example line 311-313: An eight-layer O-Grid was generated on the body wall. A mesh refinement was created in the car wake region spanning approximately 60% of the car length and spreading with a 20° angle." These need to be shown.

5. fig. 5 is useless with no dimensions, the surfaces you showed are not proper boundary conditions as will be expected.

6. your mesh needs to be shown, and no mention of mesh independency was made

7. No validation was given, at least even comparing with the Ahmed body will suffice.

8. You mentioned a segregated flow solver was used. Which one? All these details are needed as to allow others replicate your results

9. Your DOE table needs to be shown, so that your 25 cases are clearly mapped to your parameters P1-3, E1 and 2.

10. the parameter illustraions shown in Fig. 8 are not clear, are they lengths? If yes the length value ranges for each parameter taken need to be shown and discussed

11. From the three point above, a large amount of detail required to make your methodology clear to the reader is lacking. It is at least usually good practice to tabulate all your CFD settings and matrix.

12. How did you define/measure the "film thickness" referenced in Figs 9 and 10 and which was used as your objective or cost function?

13. Fig 7b is difficult to know where that is in the mirror and hence it renders your fig. 10 rather meaningless. Better angles/illustrations are needed throughout.

14. No CFD results were at all shown, for example the efect of the parameterisation on the velocity, turbulence fields etc. You must not show contours or vector plots for all your 25 cases, but a representative sample should be shown.

15. in general, your results section is too scanty and cannot be acceptable for a research paper.

Author Response

We express our gratitude to the reviewer for the suggestions provided to improve the manuscript

Author Response File: Author Response.pdf

Reviewer 4 Report

RBF has been widely used in mesh morphing applications. The authors proposed an optimization approach by coupling the mesh morphing engine RBF mesh morph and the solver environment StarCCM+. The methodology is described in detail. The proposed approach has been verified and tested in two cases, which demonstrated the solver capability. The results discussion is clear and insightful. Overall, the paper is well written, and the reviewer found the idea presented interesting.

 

Although RBF has been incorporated in ANSYS Fluent to achieve similar objectives, the reviewer thinks that the idea presented in the paper represents a sizeable effort to extend and develop the tool in optimization and this work still lays a solid foundation. The reviewer suggests adding some description about the difference between this tool and ANSYS Fluent RBF, what’s the motivation in that perspective.

Author Response

We express our gratitude to the reviewer for the suggestions provided to improve the manuscript

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors have made honest efforts to address all the points raised in the first round of review. The paper was therefore improved accordingly, with many needed clarifications included in the text of the revised version. In my opinion, after minor text editing corrections, this manuscript will be in acceptable form to be published in Fluids.

Reviewer 3 Report

The authors have very well revised the paper and I am happy with the changes made as well as their satisfactory responses. It is now very clear what was done and how. Hence, the paper is of sufficient quality, a good addition to the literature, and may be accepted for publication. 

All the very best!