Experimental and Numerical Analysis of a Compressor Stage under Flow Distortion
, Benjamin Godard 2, Pierric Joseph 1, Olivier Roussette 1, Francesco Romanò 1, Raphael Barrier 2 and Antoine Dazin 1,*Round 1
Reviewer 1 Report
Mandatory Request Changes:Mandatory Changes: Requested changes which are essential for the understanding and completeness of the paper. Paper of author(s) who have not complied with these requests may be rejected.:
- Labels of Figures 5 and 6 to be corrected
Recommended Requested Changes:Recommended Changes: Changes will improve the quality of the paper. Authors are strongly encouraged to comply with these requests.:
- Nomenclature to be completed
- Additional information about the numerical simulations and the choice of the distortion setup
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Mandatory Request Changes:Mandatory Changes: Requested changes which are essential for the understanding and completeness of the paper. Paper of author(s) who have not complied with these requests may be rejected.:
* "This comparison shows that even the numerical simulated pressure drop is sufficient to introduce new modes in the signal." : unfortunately, I'm not agree with this conclusion because, when you look at the two first minimum peaks (-2500 Pa) for the case "without screen", pDMD is not quite good vs numerical signal. Perhaps, you also need 13 modes like for the case "with screen".
* "The DMD reconstruction is good but it loses some accuracy when some non-periodic peaks occur" : at first glance, this conclusion is not obvious. On Figure 4, for 5 kg/s, plot the error between the 2 signals rather the signals. Or, plot only the 4.2 kg/s signals to illustrate the purpose.
* "The minimum number of DMD modes required to well reproduce the experimental signal is 9 for the non-distorted case at a mass flow rate of 5 kg/s. The required minimum number of modes grows up to 19 when the gris is introduced" : What is your conclusion?
How can you determine the minimum number of DMD modes in these 2 cases ? Can you precise what quantity did you use to assess this point? Perhaps, you might apply the same incremental analysis for the 4.2 kg/s operating point and conclude that you have to use 50 modes ? It would be interesting.
* "the required number of modes to well reproduce the signal is 11 without the grid, and 13 with the grid" : Figure 6 is a qualitative approach. What kind of quantification do you use to build this conclusion?
* Figure 10: There is no comment of this figure in your text. It is not obvious to understand.
What do you mean with "four cases" ? Is the rotation speed the same ? Do you look for the repeatability of the phenomenon ?
Why the distance between 2 black rectangles is not equal to 1 revolution ? Why this distance is not the same for the cases ?
Some explanations would be useful.
Recommended Requested Changes:Recommended Changes: Changes will improve the quality of the paper. Authors are strongly encouraged to comply with these requests.:
* The numbering of the figures needs to be improved.
* The punctuation to.
* Is there any measurement dowstream the screen to compare experimental and numerical total pressure field ? It would be interesting to assess the numerical distorsion.
* Is there any information about the angle distorsion downstream the screen?
* At the beginning of the "STABLE OPERATION" parts, to get an overview, it would be interesting to plot the all compressor characteristic (with and without screen distortion) for the studied iso-speed and to materialize these 2 operating points. Likewise, the 2 simulation points should be added in this graph to quantify the solver predictivity and to gauge the quality of the comparison exp/num.
* "Since numerical signals are perfectly periodic, in the SVD only one mode is
present" : the signal may be periodic, but this doesn't mean that it is a simple sinusoid represented by a single mode. Can you precise this point?
* Figure 7: please, can you use the same scale for the 2 "little" graphs?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Mandatory Request Changes:Mandatory Changes: Requested changes which are essential for the understanding and completeness of the paper. Paper of author(s) who have not complied with these requests may be rejected.:
The paper discusses the effects of an inlet flow distortion on the stall inception of an axial compressor stage tested experimentally. The azimuthal flow distortion is generated by means of a perforated grid located upstream of the test rig inlet. Numerical analyses have been carried out using an URANS for comparison by imposing an azimuthal total pressure defect as inlet boundary condition. Numerical tools like singular value decomposition and dynamic mode decomposition are used to analyze the measured and predicted pressure signal in fixed location on the compressor casing.
The topic of the paper is of great interest for the community that investigates the occurrence of flow instability and the related physical mechanisms in axial compressors. The declared goal of the study is to investigate the effect of the screen geometry and the induced flow distortion. In particular, the reproduction of the screen effects in a numerical framework is investigated. Moreover, a detailed investigation of the dynamics of stall inception under inlet flow distortion is pursued.
1. Please add the number of struts in Tab. 1. Is the BPF of the struts visible in the rotor pressure signal or has been filtered out?
2. “Non dimensional pressure drop based on the velocity where the screen is placed (22 m/s) is 1.05.” possibly provide the corresponding Distortion Correlation parameter DC(60).
3. Please give more details for the numerical setup: specify the grid dimensions for each component (strut, rotor, stator) and the number of cells in the spanwise direction as well as in the rotor tip clearance.
4. It is not clear how the two simulations, with/without total pressure distortion at the inlet may converge to the same average mass flow (4.85 kg/s):
a. Please specify the other boundary conditions (inlet/outlet) as well.
b. Please specify how inlet turbulence boundary conditions were selected.
c. Is the same exit boundary condition (static pressure) imposed with and without inlet distortion?
d. Please explain how the time-periodicity (if any) of the numerical solution is assessed.
e. Please specify the time-period used for averaging.
5. Please note that the information provided in the abstract must be present in the paper too! In effect some of these information may be also superfluous in the abstract but are needed in the paper.
6. Figures are sometimes of not good quality, labels are superposed (Figs. 8 and 9), colors are not always able to highlight differences, please try to improve the quality. There are problems in figures numbering: Fig 3 citation is missing in the text. After Fig. 4, Figs. 1 and 2 are numbered (should be 5 and 6 instead).
Recommended Requested Changes:Recommended Changes: Changes will improve the quality of the paper. Authors are strongly encouraged to comply with these requests.:
This reviewer has the following concerns:
• In the introduction it is stated that “steady state and transient operation of an axial flow compressor have been compared …” however transient operations towards the instability condition are shown only for the experimental side and not for the numerical simulation. Moreover, it is not shown any numerical simulation of throttling. Please clarify this point in the introduction (and abstract).
• Since one of the goals is to compare the numerical results to the experimental data under inlet flow distortion it is not clear why the same mass flow was not targeted. The comparisons shown in the paper in terms of pressure signals are qualitative. Experimental and numerical data are not at the same flow coefficient so one may expect different harmonic content to some extent.
• In the “Stall Inception” section it is stated that “We anticipate, however, that the grid-geometry-induced features observed in the experimental pressure measurements play a major role in the onset and decay of rotating instabilities” but then this is not clearly shown in the following. It would be interesting to see an evidence of this discrepancy between numerical predictions and experimental data while approaching the rotating stall, but this is not shown.
• This reviewer wonders if a classical Fourier Transform of the signal may help to evaluate the harmonic content. Moreover, it may also indicate the origin of a singular harmonic in the signal (Fig. 5 and 6).
• In the conclusion: the fact that it is possible to find the footprint of the grid geometry in the experimental signal it is not per se a proof that this footprint has influence on the spike inception, please be clearer and give an evidence of this fact.
1. Modes with 5 and 10 oscillations per blade channel are highlighted in the measured signal, it is interesting to observe the occurrence of these modes, however it would be interesting to know the amplitude of these modes compared to the main harmonics?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The main points of the review have been accounted for, so that the current version seems more complete and easier to understand.
Reviewer 3 Report
This reviewer thanks the authors for their thorough answers.
As a further improvement, not mandatory, it is asked to add a figure with the characteristic of the fan stage (predicted by the steady RANS CFD) in terms of pressure ratio vs. mass flow and to indicate the operating points that are considered during the discussion. This would help the reader to identify the operating range and the margin before the instability onset.
After the changes and additions performed, the authors have improved the quality of the manuscript and for this reason the paper is recommended for publication.
