The Control of Corner Separation with Parametric Suction Side Corner Profiling on a High-Load Compressor Cascade

Round 1

Reviewer 1 Report

This paper investigates the corner separation issue and proposes a parametric suction side corner profiling method to supress the corner separation and reduce the corresponding loss. While some of the results presented in the paper are interesting, this paper requires a major improvement to be considered for the publication. The reviewer’s observations are listed below.

 

1. The literature review is weak. It requires more in-depth reviews on the significance of the issue and existing both experimental and numerical studies including CFD works.
2. A 3D geometry of the cascade blade should be shown.
3. CFD modelling section, in particular, needs to be majorly revised. It lacks major information including, but are not limited to:
   1. Governing equations
   2. Details of computational domain and grid including mesh details and boundary layer mesh
   3. The reason for the choice of the mesh size is not clear. The authors performed a mesh sensitivity study based on three mesh sizes; however, the difference between them is not significant. Please discuss more on the reason for selecting this size and also what would happen beyond the largest size discussed in this paper. Are the results going to be improved further?
   4. Are the CFD simulations steady or unsteady? This is not stated surprisingly.
   5. Numerical scheme used for the simulations
   6. The reason for the choice of the turbulence model
   7. Solution convergency
4. Do the authors mean Fig. 2 in line 129 on page 4?
5. Fig. 5 needs more detailed physical discussions and explanations.
6. An overall discussion/conclusion based on observations for different cases in Fig. 8 should be provided.
7. Figs. 11 and 15 are not clear. The authors use different colours for the streamlines and contours, but only the colourmap for the contour is provided, which makes it difficult for the readers. Although the authors explain it in the text, the figures should be in a clear format with enough data. The parameters used for the contours and streamlines should be described in detail.
8. The presentation/discussions of the figures are confusing. For example, which figure are the authors referring to in lines 357-362 on page 13? Is it Fig. 11? They should be clearly referred. A clear colourmap is also required for Fig. 13.
9. As this paper investigates, the corner separation, the pressure coefficient distribution over the blade surfaces would improve the understanding.
10. Flow visualisations using pressure or velocity contours are also necessary to support the discussion and conclusions.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper shows an interesting approach to optimize blading for reduced pressure loss at a range of conditions through parametric profile adaption to reduce corner separation. 

It took me too long to understand that that three different profile adaption methods were used at the same time: "corner profiling", "end wall profiling", "ss profiling". The first method uses an interest sinusoidal function approach while the other two use a more conventional Belzier spline method. I still don't understand why the sinusoidal approach was not used for all three. The three methods were used at once and then the correlation between the different parameters and the effects studied to assess which method was more effective. This approach needs to be made clear from the methodology section. Further, the "spearman correlation" approach must be introduced. 

I did not find Figure 11 very informative and could not find the definition of phi, so could not assess those plots either. Can entropy be used? 

More detailed comments (numbers are line reference):

19: "second flow" - secondary flow?
21-24: This part of abstract is hard to understand. I believe the authors are saying that "Design point and near-stall operation have different optimal profile distributions. The suction-side corner profiling method allows a profile distribution to be found with adequate performance at a range of conditions". Please note that "to compromise" in English can also have negative connotations, so I believe it is confusing here.  
103: Which preliminary work? Can it be referenced? If it was done as part of this work, please explain the justification for the selection of the three periods. 
105: Please use "amplitude" rather than weight factor. 
107: Please detail the smooth function. 
124: Check this heading.
125: "employs the traditional way" change to "employ the traditional method"
129: "green area in Fig.3"- Please check what figure this refers to. 
151: Which research?
171: Do the authors believe lamniar separation is important in this work, hence requiring a transition model? I believe some discussion is warranted regarding the need for transition modelling. Could this work not be carried out with a simpler SST model?
196-197: Can the authors indicate how transition is detected in the images? Also, if separation, and not transition, is the main goal, wouldn't SST turbulence model be enough? This may also allow better near-wall resolution. In addition, the authors should specify the y-plus in the region of interest and discuss its relevance with regards to the turbulence model employed.
Eq. 9: It is more common to use lower case omega for loss coefficient. I assume P* means stagnation (total) pressure.
218: "surrogate CFD solver" is confusing, better say "...neural network in place of the CFD solution..."
222: PEW?
Section 2.4: Fig. 6 and parts of the description mention "process 1" and "process 2" but other parts mention internal and external iteration. I believe these are the same. 
Eq. 10: better say only "case" rather than "certain case".
240: decrease instead of increase?
262: The Spearman correlation approach needs to be described or a reference given. 
334: Has phi been defined?
514-524: This parragraph seems to say one thing and then the opposite. I do not understand.
538: "compromising improvement" sounds like a negative result.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors investigate numerically the effect of wall profiling in a axial compressor. The results are compared to experiments.

When performing a grid sensitivity study, please use a solid methodology. Following the approach suggested by Celik et al. >https://asmedigitalcollection.asme.org/fluidsengineering/article/130/7/078001/444689/Procedure-for-Estimation-and-Reporting-of<, the authors can also estimate error bars due to the numerical approach.

The number of references is very low. Maybe the authors can find some more. For example, nothing has been stated about unsteady effects:

• Zambonini et al. "Corner separation dynamics in a linear compressor cascade"
• Wei et al. "Intermittent corner separation in a linear compressor cascade"

In line 171, the autors state that the k-epsilon and the gamma-theta turbulence models have been used. It is difficult for me to understand later which turbulence model is used where. Please be more specific.

In some of the plots, the axis description is very close to the numbers. Please leave a bit more space such that everything can be well read. For example, figures 7, 8, 10, and 12.

Please keep the font size as large as possible. Parts of figure 9 are difficult to read.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I am happy with the authors' responses and therefore would like to recommend this paper for publication. There are some reference source errors and typos. Please check the proof carefully before publishing. 

Reviewer 2 Report

Thank you for the corrections. The paper is much clearer now.