Vibroacoustic Optimization Study for the Volute Casing of a Centrifugal Fan
Round 1
Reviewer 1 Report
In this study, parametric study on the geometry of centrifugal fan volute is conducted to minimize the vibrational noise. Overall, a reviewer feels that the authors do not understand how to write a paper.
* The composition of the introduction is bad. The first paragraph is too long.
* Units should not be written in italics.
* Fig.2: Is spatial accuracy maintained at multiple places where mesh size suddenly changes?
* When the authors use RBF, is not the number of initial samples insufficient? Also, rather than a parametric study, it is appropriate to set up a response surface to the extent that sensitivity analysis can be performed.
* How did the authors set various parameters of RBF? It is a bottleneck of RBF that it is difficult to tune these.
* How did the authors change the mesh when changing the geometry?
* Do not the authors have the problem that affect the solution by the scheme and turbulence model used in URANS?
* Fig.30: How do the authors mathematically guarantee that the obtained non-dominated surface is the Pareto surface in the problem?
* In Section 5 that summarizes the results, only the results obtained are listed; there is no consideration or discussion at all. This is not a paper but a mere report. It is not a paper unless the authors comprehensively present the findings that can be seen from the results obtained this time, such as how to obtain such a result, how to redesign the geometry to obtain better performance.
Author Response
Response to Reviewer 1 Comments
Point 1: * The composition of the introduction is bad. The first paragraph is too long.
Response 1: Thanks for your comments, the introduction comprises background (the first paragraph), the literature review of production of vibro-acoustic (the second and third paragraph), the literature review of control method of vibro-acoustic (the fourth paragraph), the works description of this paper (the fifth paragraph). Although the introduction section is longer, the author believes that the content cannot be deleted.
Besides, the duplicate and verbose part of the background have been revised and deleted, and marked with green highlight in the first paragraph.
Point 2: * Units should not be written in italics.
Response 2: Thanks for your advice, the units and others in italics have been revised and marked with yellow highlight.
Point 3: * Fig.2: Is spatial accuracy maintained at multiple places where mesh size suddenly changes?
Response 3: Thanks for your comment, firstly, I must explain the main research contents of this article to the reviewer. The numerical optimization in this paper is divided into three main parts: the first part is based on CFD to obtain the vibrational excitation source. The unsteady pressure fluctuations on the casing is constant throughout the optimization process.
The second part, the vibrations of the casing are solved based on the finite element method (FEM), and the vibrational sound radiation is solved by the vibrational acceleration of the casing surface.
The third part, the casing thickness is designed as variable, and the radiated sound power and total mass of the casing are considered as the objective of optimization. One point should be noted that the variables (each panel thickness of three-part volute) are specified as shell63 element, and the variables change if the structural properties of the each panel thickness change, such as assigning different thickness values in the designed order, so the grids change for all the parts are not involved.
Point 4: * When the authors use RBF, is not the number of initial samples insufficient? Also, rather than a parametric study, it is appropriate to set up a response surface to the extent that sensitivity analysis can be performed.
Response 4: Thanks for your advice, to ensure a sufficient number of samples, the full-factor method was adopted to collect sample points of the design space constructed by the three variables collected; thus, a total of 125 sample points was collected. In addition, the verification of the RBF model in section 5.2 also indirectly proves that the number of sampling points is sufficient.
The reviewer's suggestion is very good (it is appropriate to set up a response surface to the extent that sensitivity analysis can be performed), but the author believes that in order to determine the relationship between the variables and the objective function, a combination of parametric analysis and sensitivity analysis is advantageous. So the main influencing variables and the range of values are qualitatively obtained using this method.
Point 5: * How did the authors set various parameters of RBF? It is a bottleneck of RBF that it is difficult to tune these.
Response 5: Thanks for your comment, the various parameters of RBF are assigned by the platform of Isight, such as the three volute panel thickness variables (ST, FT, BT ) are defined as input vector, the volute radiated acoustical power and volute total mass are defined as output vector. In addition, this paper focus on the proposed optimization method to reduce the vibrational noise radiation of the casing. The RBF model is used to improve the efficiency and speed of this optimization merely.
Point 6: * How did the authors change the mesh when changing the geometry?
Response 6: Thanks for your advice, firstly, I must explain the main research contents of this article to the reviewer. The numerical optimization in this paper is divided into three main parts: the first part is based on CFD to obtain the vibrational excitation source. The unsteady pressure fluctuations on the casing is constant throughout the optimization process.
The second part, the vibrations of the casing are solved based on the finite element method (FEM), and the vibrational sound radiation is solved by the vibrational acceleration of the casing surface.
The third part, the casing thickness is designed as variable, and the radiated sound power and total mass of the casing are considered as the objective of optimization. One point should be noted that the variables (panel thickness of three-part volute) are specified as shell63 element, and the variables change if the structural properties of the each panel thickness change, such as assigning different thickness values in the designed order, so the grids change for all the parts are not involved.
Point 7: * Do not the authors have the problem that affect the solution by the scheme and turbulence model used in URANS?
Response 7: Thanks for your comment, the aerodynamic performance and pressure fluctuations results obtained using this CFD method (the scheme and turbulence model used in URANS) have been compared with experiments, showing a good agreement. The detail description and analysis can be obtained in the previous published paper [1]. It indicates that the CFD method using in this paper is reasonable and effective, there is not have the problem that affect the solution.
[1] J. Zhang, W. Chu, H. Zhang, Y. Wu, X. Dong, Numerical and experimental investigations of the unsteady aerodynamics and aero-acoustics characteristics of a backward curved blade centrifugal fan,Appl Acoust, 2016, 110, 256-267.
Point 8: * Fig.30: How do the authors mathematically guarantee that the obtained non-dominated surface is the Pareto surface in the problem?
Response 8: Thanks for your comment, the pareto front solutions are nondominated in the sense that no other solution is “better” with respect to all objectives, and no improvement can be made in any objective without trading off (worsening) the other. In mathematical terms, given N functions fi, i=1…N, a system variable vector x* is Pareto-optimal if there is no vector x for which for fi(x)< fi(x*) all i and fi(x)< fi(x*) for at least one objective.
The Pareto curve is the set of x* where there are no other solutions for which simultaneous improvements in all objectives can occur (see schematic in Fig. 1 in the attachment)
Point 9: * In Section 5 that summarizes the results, only the results obtained are listed; there is no consideration or discussion at all. This is not a paper but a mere report. It is not a paper unless the authors comprehensively present the findings that can be seen from the results obtained this time, such as how to obtain such a result, how to redesign the geometry to obtain better performance.
Response 9: Thanks for your comment, I have reorganized the section 4 and section 5, and added the section 6 (the part of results and discussion). The results discussion of parametric and optimization parts were integrated into section 6.
Besides, the optimization results discussion were marked with green highlight.
Generally, the noise-reduction mechanism of single and multi-objective optimization is attributed to the major factor: the normal vibration velocity of three-part volute surface changed smaller by controlling the thickness combination with the unsteady aerodynamics determined and volute geometry fixed. The previous study [36] showed that the normal vibration velocity of the volute surface was the decisive factor that determined the radiated sound power. To comment on this major noise-reduction mechanism, the comparison of the normal vibration velocity of the volute surface before and after optimization were presented in Figure 18 and Figure 33. It indicates that the normal vibration velocity of the volute surface was greatly diminished (especially the tongue region) after optimization, which inevitably leads to a significant reduction for the radiated sound power of the volute surface, just as shown in Figure 32. In addition, the parametric analysis indicates that the ST had the largest influence on sound radiated power of volute surface, followed by FT, then BT. The sound radiated power of volute surface could be sharply reduced if the ST was much larger (larger than 7mm for this fan), BT and FT were designed as lower value (BT was less than 8.5 mm and the FT was less than 7.0 mm for this fan). Less noise radiation could be achieved, if the ST is increased over a certain limit (the setting extremum), But this can lead to unpredictable mass gain and increased costs.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The work presents an interesting optimization of the casing of a centrifugal fan, considering the externally radiated sound power as target to minimise and the casing thicknesses as design variables.
Although the work is a revised version of a previously submitted one, it still requires several relevant improvements.
All the highlighted sentences in the document are either incomprehensible (English must be improved) or pointless or redundant.
The manuscript seems to be report rather than a scientific article. It must be reorganised in such a way to be more compliant with the standards of a scientific journal. As instance, §4.2 “Volute FEM Model and Validation” comprises FEM model, tests description and numerical-experimental comparisons. It could be much better organised.
Several minor modifications are also required, e.g.:
figure 6 could be improved positioning the text outside the figure;
some typos are spread throughout the article, e.g. eq. 34 at page 24;
all the acronyms spread throughout the article have to be clearly defined, e.g. FT, ST, BT are not defined in the nomenclature and are presented in the abstract without any definition;
etc…
Please modify the manuscript accordingly.
Author Response
Response to Reviewer 2 Comments
Point 1: All the highlighted sentences in the document are either incomprehensible (English must be improved) or pointless or redundant.
Response 1: Thanks for your advice, all the highlighted sentences in the document are revised and marked with “green” highlight.
Point 2: The manuscript seems to be report rather than a scientific article. It must be reorganised in such a way to be more compliant with the standards of a scientific journal. As instance, §4.2 “Volute FEM Model and Validation” comprises FEM model, tests description and numerical-experimental comparisons. It could be much better organised.
Response 2: Thanks for your advice, I have reorganized the section 4 and section 5, and added the section 6 (the part of results and discussion). Please review the revised parts in the resubmit paper “applsci-450302-revised”
Point 3: Several minor modifications are also required, e.g.: figure 6 could be improved positioning the text outside the figure; some typos are spread throughout the article, e.g. eq. 34 at page 24;
Response 3: Thanks for your advice, the figure 6 have been improved, please see the revised Figure in the submited revised paper.
Besides, I have carefully checked that each equation is arranged from small to large, and eq. 34 is within the range of numbers.
Point 4: all the acronyms spread throughout the article have to be clearly defined, e.g. FT, ST, BT are not defined in the nomenclature and are presented in the abstract without any definition;.
Response 4: Thanks for your comment, I have added the definition of ST, BT, FT in the abstract and nomenclature.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
Thank you very much for your appropriate modifications.
As an example on the application of RBF to the actual problem which takes enormous time to evaluate the objective function, a reviewer considers that this paper is beneficial to readers.
