Flow Coefficient and Reduced Frequency Effects on Wake-Boundary Layer Interaction in Highly Accelerated LPT Cascade

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.:
N/A

Recommended Requested Changes:Recommended Changes: Changes will improve the quality of the paper. Authors are strongly encouraged to comply with these requests.:
The paper reports PIV measurements on wake-boundary layer interaction on highly accelerated LPT cascade.
The study has been carried out on linear cascade considering two different flow coefficient and two different reduced frequency associate to wake passing
Here below some questions/comments required:
1- Check minor typos
2- Add nomenclature
3- The authors claim that tests have been run at Re 160000. How the Re has been considered? Is it reasonable to consider the RE for C1 and C2 is double respect to the case C3 and C4? Do the test being run at the same chords lengths?
4- What the reason of f* 2.4 rather than 0.6?
5- Which rationale is behind the choice of using 300 snapshots?

Author Response

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.: N/A Recommended Requested Changes:Recommended Changes: Changes will improve the quality of the paper. Authors are strongly encouraged to comply with these requests.:
The paper reports PIV measurements on wake-boundary layer interaction on highly accelerated LPT cascade. The study has been carried out on linear cascade considering two different flow coefficient and two different reduced frequency associate to wake passing Here below some questions/comments required:

1- Check minor typos

The manuscript has been revised in order to improve the grammar quality, correcting some typos and language errors.

2- Add nomenclature

In the revised version of the paper, nomenclature has been added.

3- The authors claim that tests have been run at Re 160000. How the Re has been considered? Is it reasonable to consider the RE for C1 and C2 is double respect to the case C3 and C4? Do the test being run at the same chords lengths?

All the tests (C1 to C4) have been performed at fixed Reynolds number (i.e. 160k). Reynolds was based on the flow velocity at the exit of the cascade and on the blade chord. In the revised version of the paper, the definition of the Reynolds number is reported in the nomenclature to make clearer this point.

4- What the reason of f* 2.4 rather than 0.6?

The values of the flow coefficient and the reduced frequency were chosen to investigate wide parameter variations around their nominal (design) values. This is now clearly expressed in the test.

5- Which rationale is behind the choice of using 300 snapshots?

The number of the images per phase was chosen in order to ensure the convergence of the first (most energetic) POD modes, which are discussed in the text. Text have been revised to clarify this concept, including an extra reference to support the choice.

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.:
The paper investigates the flow evolution on the suction side of a highly accelerated low-pressure turbine cascade. Four different operating conditions are selected with different flow coefficient and reduced frequency at fixed Reynolds number. The experimental setup includes a bar wake generator to produce the upstream wakes that are convected within the blade passage and interact with the blade boundary layer. PIV technique is used to acquire the flow velocity field. The data are post-processed using the POD method to gain insight of the loss production mechanisms and how they are affected by the operating parameters.

The topic of the paper is of interest for the aerodynamic designers since link the unsteady loss production mechanisms to basic operating parameters like the reduced frequency and the flow coefficient. The detailed POD analysis highlights the regions and the mechanisms responsible for loss generation. Except some typos, that is recommended to fix, the paper is well written and uses appropriate arguments to discuss the results.

1. Please provide the definition of reduced frequency (there are several definitions in the literature) and discuss the rationale behind the choice of the operating parameters (range of flow coeff. and reduced frequency). A reduced frequency of 2.4 seems quite high for LPT operation.
2. The Reynolds number should be defined (based on outlet velocity, chord, ...)
3. Please specify where the value of inlet turbulence of 5% was measured.
4. Please specify the meaning of all the acronyms at their first occurrence in the text (e.g. TTL ..)
5. Please identify the control volume used for the loss balance in the cascade domain.
6. Fig. 4: Tu intensity between two wakes seems variable between 6% (C2) and 7% (C3) while it was declared 5%
7. A shape factor of 1.7 is out of the range in the scale of Fig. 8. However this means the fully turbulent flow is never achieved with these operating conditions (H=1.4)? For case C4 in Fig. 8 six wakes are actually visible (not seven).

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:
• Introduction: boundary layer transition and separation are associated with low-Reynolds number operation in high-lift configurations, this must be clearly stated.
• The cascade should be better described from an aerodynamic point of view, for example:
o it would be useful if the authors could provide the SS diffusion factor and the diffusion rate of such a cascade since they are key parameters for the boundary layer development.
o Please provide some aerodynamic parameter to quantify the high acceleration of this LPT cascade, the reader should be in the conditions to appreciate the amount of acceleration. Without these parameters the information contained in the manuscript are not so meaningful for a designer

1. From these s-t plots (Figs. 8) usually characteristic velocity of turbulent patches and calmed region may be deduced. From these plots any evidence of boundary layer separation is not visible (while in the introduction a boundary layer separation condition was introduced for these profiles). Please add some details/comments to better identify the aerodynamics of this cascade.
2. Please add labels for Q1 and Q2, when present, in Figs. 10, 11, 12, 13. In the captions please write the flow coeff. and the reduced frequency for the reader’s convenience.

Comments for author File: Comments.pdf

Author Response

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 investigates the flow evolution on the suction side of a highly accelerated low-pressure turbine cascade. Four different operating conditions are selected with different flow coefficient and reduced frequency at fixed Reynolds number. The experimental setup includes a bar wake generator to produce the upstream wakes that are convected within the blade passage and interact with the blade boundary layer. PIV technique is used to acquire the flow velocity field. The data are post-processed using the POD method to gain insight of the loss production mechanisms and how they are affected by the operating parameters.
The topic of the paper is of interest for the aerodynamic designers since link the unsteady loss production mechanisms to basic operating parameters like the reduced frequency and the flow coefficient. The detailed POD analysis highlights the regions and the mechanisms responsible for loss generation. Except some typos, that is recommended to fix, the paper is well written and uses appropriate arguments to discuss the results.
1. Please provide the definition of reduced frequency (there are several definitions in the literature) and discuss the rationale behind the choice of the operating parameters (range of flow coeff. and reduced frequency). A reduced frequency of 2.4 seems quite high for LPT operation.

In the revised version of the paper, the reduced frequency definition has been added in the nomenclature. The values of the flow coefficient and the reduced frequency were chosen to investigate wide parameter variations around their nominal (design) values (that is close to the C3 case). The motivation of the choice of the operating parameters is now provided in the test.
2. The Reynolds number should be defined (based on outlet velocity, chord, ...) The Reynolds number has been defined in the nomenclature.
3. Please specify where the value of inlet turbulence of 5% was measured. Turbulence intensity of about Tu=6% was measured at the inlet of the cascade in the steady condition.
4. Please specify the meaning of all the acronyms at their first occurrence in the text (e.g. TTL ..)

All acronyms have been specified in the revised version of the paper.
5. Please identify the control volume used for the loss balance in the cascade domain.

The control volume used to quantify the losses was the whole PIV measuring region. This is now stated in the text.
6. Fig. 4: Tu intensity between two wakes seems variable between 6% (C2) and 7% (C3) while it was declared 5%

We wish to apologize for the erroneous value. Actually, the free-stream Tu is about 6% as documented in the plot. We have corrected the declared value in the revised version of the paper.
7. A shape factor of 1.7 is out of the range in the scale of Fig. 8. However this means the fully turbulent flow is never achieved with these operating conditions (H=1.4)? For case C4 in Fig. 8 six wakes are actually visible (not seven).

We agree with the Reviewer observation. The text has been modified to avoid discussing an out of scale value. The values of the shape factor suggest that a fully turbulent condition of H=1.4 is never achieved. About the C4 case, as the Reviewer observed six wakes are visible in the plot, thus we have corrected this point.
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:

• Introduction: boundary layer transition and separation are associated with low-Reynolds number operation in high-lift configurations, this must be clearly stated.

The introduction has been revised to clarify that transition and, especially, separation processes are usually observed in high-lift configurations at low Reynolds number, whereas in this work a highly accelerated profile is investigated.
• The cascade should be better described from an aerodynamic point of view, for example:

o it would be useful if the authors could provide the SS diffusion factor and the diffusion rate of such a cascade since they are key parameters for the boundary layer development.

o Please provide some aerodynamic parameter to quantify the high acceleration of this LPT cascade, the reader should be in the conditions to appreciate the amount of acceleration. Without these parameters the information contained in the manuscript are not so meaningful for a designer

In the revised version of the paper, the diffusion factor, the diffusion rate and the velocity ratio are provided in order to accomplish with the Reviewer’s suggestion.
1. From these s-t plots (Figs. 8) usually characteristic velocity of turbulent patches and calmed region may be deduced. From these plots any evidence of boundary layer separation is not visible (while in the introduction a boundary layer separation condition was introduced for these profiles). Please add some details/comments to better identify the aerodynamics of this cascade.

In this type of high accelerated LPT cascade, flow separation never occurs. Transition is induced locally in the unsteady operation by the impinging wakes, as observed in Fig. 8. In the revised version of the paper the text has been modified in order to clarify that flow separation never occurs.
2. Please add labels for Q1 and Q2, when present, in Figs. 10, 11, 12, 13. In the captions please write the flow coeff. and the reduced frequency for the reader’s convenience.

Some labels to identify Q1 and Q2 vortices have been added in Fig. 10, 11, 12 and 13, and the caption has been modified according with the Reviewer’s suggestion.

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.:
Not really applicable for conference presentation. If the authors target a journal publication, I have made a few remarks and raised points of discussion plus requests for extra data/analysis in the following section (recommended changes).

Recommended Requested Changes:Recommended Changes: Changes will improve the quality of the paper. Authors are strongly encouraged to comply with these requests.:
The paper reports an interesting experimental study on the effect of the flow coefficient and strouhal number on the wake-boundary layer interaction of a low-speed low-pressure turbine cascade. PIV time-resolved measurements are used to deduce the loss mechanisms in its viscous dissipation and turbulent kinetic energy production terms.
I think the paper has merit in that investigates an important effect, that of the flow coefficient, that often is neglected in not fully representative experimental rigs. Furthermore, the experimental data and the authors' analysis through POD confirms similar findings in the literature and provide experimental evidence of the complex wake-boundary layer interaction.
The paper makes certainly a valid contribution to the conference. To be considered of archival quality, I feel there are a few points that the authors should address in a revised manuscript or in an improved and extended version of the manuscript (post-conference?):
1. The introduction should perhaps include some extra reference on the effect of flow coefficient on the wake-BL interactions and loss generations in LP turbines.
2. Please motivate on the reasons behind the selection of the proposed cases. Why not one extra flow coefficient at multiple strouhal numbers? Why the Reynolds was selected to be quite high, at 160,000 whereas major difference in wake-BL interaction and losses might be expected to be larger at lower Reynolds for which typical LPT are designed for? Same for the inlet turbulence intensity, why a relatively high value of Tu=5%? Please elaborate on this in the paper.
3. I would like to see more details and information about the experimental setup to enhance the experiment replication and provide useful info to read for a non-trivial experiment:
a. How and where was seeding applied and to what fraction of the passage massflow does the seeding flow rate correspond?
b. Explain how the number of images and phase-lock number were selected. Why not more images and a much higher temporal resolution of the wake period (e.g. 20 phase periods rather only 6)? More details about the PIV imaging settings (exposure, delays, window duration, etc..).
c. Sketch in Figure 1 should also show the “bar shadow” issue! It took me a while to see what the authors meant about their strategy to get rid of this issue (well thought!) and the reader could benefit from visual support for this.
d. Elaborate on the criteria used to deem 300 images sufficient for application of POD
e. Please report the effect of the bar-to-bar variations on the statistics of the turbulent parameters from phase-locked images. Can the bar-to-bar variability contribute to artificially increase the measured TKE loss? How is this accounted for in the image processing?
4. What is the error made by the use of a 2D planar PIV (2 velocity components only) on the estimation of the TKE production loss term? Any reference to support a small error for (at least) profile-only turbine loss estimation?
5. For a better understanding of the turbine test case, can the authors report a Mach/pressure distribution on the airfoil?
6. How are the boundary layer thickness and momentum calculated from the PIV images? Please report the method and states a level of uncertainty.
7. I am a bit puzzled by the conclusion that the overall loss for the case C1-C3 are almost unchanged (see Fig. 9 and comments in the text), whereas Fig. 7 and 8-top show what seems to be a non-negligible change in momentum thickness of the rear SS boundary layer…That change in momentum boundary layer should be associated with a change in viscous dissipation loss. I would invite the authors to discuss further this point and to add more evidence that the loss coefficient is much unaffected by the flow coefficient change (at a same reduced frequency of the incoming wakes). Is it perhaps the scale in plot 7/8 that is misleading (one can infer significant change of loss) from what can be understood looking at plot 8 for instance. A non-negligible loss change can be also expected as the viscous dissipation loss dominates the overall loss with a 2/3 contribution.
8. In general, there is little analysis of the viscous dissipation loss from the PIV image analysis. And looking at the data reported in Fig7-8 I would be cautious about the conclusion that the major effect on loss is to be found on different rates of TKE production.
9. The interesting work would be more complete and of large impact, if wake measurements could be added to the analysis.

Author Response

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.:

Not really applicable for conference presentation. If the authors target a journal publication, I have made a few remarks and raised points of discussion plus requests for extra data/analysis in the following section (recommended changes).

Recommended Requested Changes:Recommended Changes: Changes will improve the quality of the paper. Authors are strongly encouraged to comply with these requests.: The paper reports an interesting experimental study on the effect of the flow coefficient and strouhal number on the wake-boundary layer interaction of a low-speed low-pressure turbine cascade. PIV time-resolved measurements are used to deduce the loss mechanisms in its viscous dissipation and turbulent kinetic energy production terms.

I think the paper has merit in that investigates an important effect, that of the flow coefficient, that often is neglected in not fully representative experimental rigs. Furthermore, the experimental data and the authors' analysis through POD confirms similar findings in the literature and provide experimental evidence of the complex wake-boundary layer interaction.

The paper makes certainly a valid contribution to the conference. To be considered of archival quality, I feel there are a few points that the authors should address in a revised manuscript or in an improved and extended version of the manuscript (post-conference?):

We wish to thank the Reviewer for his/her appreciation.
1. The introduction should perhaps include some extra reference on the effect of flow coefficient on the wake-BL interactions and loss generations in LP turbines. Some extra references have been included in the introduction.
2. Please motivate on the reasons behind the selection of the proposed cases. Why not one extra flow coefficient at multiple strouhal numbers? Why the Reynolds was selected to be quite high, at 160,000 whereas major difference in wake-BL interaction and losses might be expected to be larger at lower Reynolds for which typical LPT are designed for? Same for the inlet turbulence intensity, why a relatively high value of Tu=5%? Please elaborate on this in the paper.

The Reynolds number is slightly lower than the nominal design value of this cascade. Indeed, this highly accelerated cascade is designed to operate at high Reynolds number. The choice of Re=160k is motivated by the maximization of the PIV measurement spatial resolution. The values of the flow coefficient and the reduced frequency are instead chosen in order to investigate a wide range of parameter variation around their nominal point (close to the C3 case). All these points have been discussed in the revised version of the paper.
3. I would like to see more details and information about the experimental setup to enhance the experiment replication and provide useful info to read for a non-trivial experiment:

a. How and where was seeding applied and to what fraction of the passage massflow does the seeding flow rate correspond?

The seeding flow rate was chosen in order to obtain a particle concentration of around 4/5 particles/investigation area, thus ensuring to maximize PIV performances. The information has been added in the text.
b. Explain how the number of images and phase-lock number were selected. Why not more images and a much higher temporal resolution of the wake period (e.g. 20 phase periods rather only 6)? More details about the PIV imaging settings (exposure, delays, window duration, etc..).

The number of total images was limited by the complexity of the experimental setup e by the memory of the cameras. Therefore, the number of bins used to discretize the wake passing period was chosen in order to have a number of PIV images per phase that ensure the convergence of the first (most energetic) POD modes. More details about PIV settings have been added.
c. Sketch in Figure 1 should also show the “bar shadow” issue! It took me a while to see what the authors meant about their strategy to get rid of this issue (well thought!) and the reader could benefit from visual support for this.

We agree with the Reviewer. A detail sketch has been added in Fig. 1 to visually support to the text.
d. Elaborate on the criteria used to deem 300 images sufficient for application of POD

An amount of almost 300 images can be considered sufficient to ensure the convergence of the first POD modes, as documented by Lacarelle et al. (2009). The reference has been added in the paper to support the choice of the number of images per phase.
e. Please report the effect of the bar-to-bar variations on the statistics of the turbulent parameters from phase-locked images. Can the bar-to-bar variability contribute to artificially increase the measured TKE loss? How is this accounted for in the image processing?

In order to avoid the effect due to bar-to-bar variation, that can artificially increase the TKE production, POD has been applied as a filter. In particular, loss differences between the flow cases are shown due to the first POD modes, that are related to the flow physics, while effects due to bar-to-bar variation are captured by higher order modes that have been filtered out.
4. What is the error made by the use of a 2D planar PIV (2 velocity components only) on the estimation of the TKE production loss term? Any reference to support a small error for (at least) profile-only turbine loss estimation?

Actually, in a 2D planar PIV the turbulent component <w> is not measured. However, in a two-dimensional flow, the velocity gradient with respect to the z-axis is almost zero, thus the term <w> doesn’t significantly contribute to additional production.
5. For a better understanding of the turbine test case, can the authors report a Mach/pressure distribution on the airfoil?

In the revised version of the paper, more details on the test case have been added. In particular, diffusion rate and diffusion factor are now provided to characterize the acceleration on the blade suction side. Comments in the text have been modified to make clearer the test case.
6. How are the boundary layer thickness and momentum calculated from the PIV images? Please report the method and states a level of uncertainty.

Boundary layer thickness is directly evaluated from PIV snapshots, thanks to the high spatial resolution of the measurements (vector spacing=0.3 mm). Momentum thickness is calculated by trapezoidal rule. This is now specified in the text, with the related estimated accuracy.
7. I am a bit puzzled by the conclusion that the overall loss for the case C1-C3 are almost unchanged (see Fig. 9 and comments in the text), whereas Fig. 7 and 8-top show what seems to be a non-negligible change in momentum thickness of the rear SS boundary layer…That change in momentum boundary layer should be associated with a change in viscous dissipation loss. I would invite the authors to discuss further this point and to add more evidence that the loss coefficient is much unaffected by the flow coefficient change (at a same reduced frequency of the incoming wakes). Is it perhaps the scale in plot 7/8 that is misleading (one can infer significant change of loss) from what can be understood looking at plot 8 for instance. A non-negligible loss change can be also expected as the viscous dissipation loss dominates the overall loss with a 2/3 contribution.

As the Reviewer properly observed, slight inaccuracy is obtained in the loss comparison between the C1 and C3 flow cases, since the momentum thickness results higher in the C1 than in the C3 case and, at the same time, the calculated losses appears almost the same. However, differences are very small and difficult to quantify. The text has been modified in order to mitigate some conclusions.
8. In general, there is little analysis of the viscous dissipation loss from the PIV image analysis. And looking at the data reported in Fig7-8 I would be cautious about the conclusion that the major effect on loss is to be found on different rates of TKE production.

The analysis of the viscous term is not the main target of this work. We preferred to investigate in more detail the fluctuating part of the flow field, and the related turbulent kinetic energy production. The text has been modified in order to mitigate some conclusions in this regard too.
9. The interesting work would be more complete and of large impact, if wake measurements could be added to the analysis.

Unfortunately, wake measurements are not available. We take into account the Reviewer’s suggestion, planning further measurements to extend the work.

Round 2

Reviewer 1 Report

After a check of the updated paper the reviewer consider the paper worth of publication 

Reviewer 2 Report

The topic of the paper is of interest for the aerodynamic designers since it links the unsteady loss production mechanisms to basic operating parameters like the reduced frequency and the flow coefficient. The detailed POD analysis highlights the regions and the mechanisms responsible for loss generation. Thanks to the additional details provided about the cascade loading distribution, diffusion, and acceleration, the paper can be a reference for the design of highly accelerated LPT cascades.

Reviewer 3 Report

Based on the revision made by the authors, I deem the current paper worth journal publication. Most of my comments and remarks from the first round of review were addressed in the revised manuscript. Overall, I think the paper is a relevant addition to the literature about low-pressure turbine aerodynamics and investigate a quite important issue (scalability of rig results and aero effects of incoming wake flow coefficient) with high-quality, full-field PIV measurements.