Combined Experimental and CFD Investigation of Flat Plate Film Cooling through Fan Shaped Holes

Round  1

Reviewer 1 Report

Mandatory Request 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.:

note: overall really nice well-done piece of work and how you wrote it up -given the space restrictions for this conference

> in abstract please add that the measurements were carried out at near unity density ratio (this is important for interpretation of the blowing ratio) - or quote it exactly - calc based on page 3 suggests DR = 1.12

> page 6 and page 7 - please look at Fig. 12 - remembering the following - you have defined X coord to start at the mid plane of the hole exit

>> (1) you see that the asymptotic value at X/D to be different for BR=1 and BR=2 - and you recall from above my request for explicitly quoting t/P - because this value is the theoretical asymptote at exit - and anything below that typically suggests a separated flow at exit - for your experiment t/P is 0.47 and you are seeing for BR = 1 the asymptote is being displayed in Fig 12 as about 0.52 - not possible - and for BR=2 about 0.46 - so this seems to be some sort of experimental error - especially for the BR=1 more-attached case

>> (2) now turn your attention to Fig.14 - and look at the CFD simulatons for RKE and for RSM - note the most-likely expected flat behavior of the CFD in the region around X/D=0 and compare that with the linearly decreasing behavior that RKE exhibits compared to RSM - this suggests maybe a flow segregation and/or separation in the part of the Gritsch hole on the leeward (downstream side) - and most probably because for whatever reason Gritsch (and therefore you all) used a non-radiused transition between the metering hole exit and the start of the diffuser

note: you might be able to incorporate these observations if you so choose to accept them

Recommended Requested Changes: Changes will improve the quality of the paper. Authors are strongly encouraged to comply with these requests.:

> page 3 - "... coordinate system shown in Fig.2." - consider adding "and Fig.3" - because the"positive" x-y-z coordinate system is better displayed in Fig. 3

> page 3 - in the description of the geometry at the top of the page - could you add two ratios that are important to film cooling designers - t/P and AR - where t/P would be the 13.98/(5x6) and AR is ratio of cross-sectional area at exit to that of the metering Holes

> page 3 - "... and a X-wire Hot Wire anemometry (HW)." - need to change "anemometry" to "anemometer"

MAY I SAY - excellent that you delineate or show difference in uncertainty for eff of PSP at two levels

- thanks

> page 3 - "... turbulent length scale for the mainstream was set at ..." - ?? did you arbitrarily set it or was it measured or just assumed

> page 4 - "... thermal behavior of film cooling jet ..." - should use "the" after word "of"

> page 4 - suggestion only - you use (actually in two places in paper) the phrase "whatever the BR" -based on your paper data being reported it seems more appropriate to use "for both BR" - since you are only reporting two BR datasets

> page 5 - at bottom of page you are (I think) first mentioning having used flow visualization - could you mention briefly what you used for this

> page 6 - ".. injecting nitrogen as coolant stream, ..." - add "the" after "as"

> page 6 - "These findings are in general agreement with Gritsch ..." - the words "general agreement" do not seem to be in keeping with the high quality of this paper - is there any more you can say ?

> page 8 - "Jet stays close ..." - change to "The jet stays ..."

Author Response

Dear Reviewer,

Many thanks for your comments and suggestions on our manuscript. You may find our responses in the following:

MR

In abstract please add that the measurements were carried out at near unity density ratio (this is important for interpretation of the blowing ratio) - or quote it exactly - calc based on page 3 suggests DR = 1.12.

The phrase “at near unity density ratio” has been added to the abstract.

Page 6 and page 7 - please look at Fig. 12 - remembering the following -you have defined X coord to start at the mid plane of the hole exit:

(1) you see that the asymptotic value at X/D to be different for BR=1 and BR=2 - and you recall from above my request for explicitly quoting t/P -because this value is the theoretical asymptote at exit - and anything below that typically suggests a separated flow at exit - for your experiment t/P is 0.47 and you are seeing for BR = 1 the asymptote is being displayed in Fig 12 as about 0.52 - not possible - and for BR=2 about 0.46 - so this seems to be some sort of experimental error - especially for the BR=1 more-attached case.

Fig. 12 has been modified along with Fig.13. The error in effectiveness values at the hole exit for both cases happened during image post-processing, while determining the exact exit hole point, i.e. X/D=0.

(2) now turn your attention to Fig.14 - and look at the CFD simulations for RKE and for RSM - note the most-likely expected flat behavior of the CFD in the region around X/D=0 and compare that with the linearly decreasing behavior that RKE exhibits compared to RSM - this suggests maybe a flow segregation and/or separation in the part of the Gritsch hole on the leeward (downstream side) - and most probably because for whatever reason Gritsch (and therefore you all) used a non-radiused transition between the metering hole exit and the start of the diffuser.

note: you might be able to incorporate these observations if you so choose to accept them. 

Having checked the simulation results at the hole exit for RSM turbulence model, one could verify the flow separation on the leeward side of the hole based on RSM predictions, this observation has been added to the manuscript (Please see Page. 7).

RR

-page 3 - "... coordinate system shown in Fig.2." - consider adding "and Fig.3" - because the "positive" x-y-z coordinate system is better displayed in Fig. 3

This figure is replaced by Fig.3 which along with modified Fig. 1 may show the coordinate system clearly.

-page 3 - in the description of the geometry at the top of the page – could you add two ratios that are important to film cooling designers - t/P and AR - where t/P would be the 13.98/(5x6) and AR is ratio of cross-sectional area at exit to that of the metering Holes.

t/P and AR have been added to the manuscript (Please see page. 3).

Page 3 - "... and a X-wire Hot Wire anemometry (HW)." - need to change "anemometry" to "anemometer

modified.

page 3 - "... turbulent length scale for the mainstream was set at ..." -?? did you arbitrarily set it or was it measured or just assumed.

Turbulent length scale for mainstream (?) was calculated from ? = 0.4 ?, where ? is the boundary layer thickness available from the measured velocity profile.

Page 4 - "... thermal behavior of film cooling jet ..." - should use "the"after word "of"

modified.

Page 4 - suggestion only - you use (actually in two places in paper) the phrase "whatever the BR" - based on your paper data being reported it seems more appropriate to use "for both BR" - since you are only reporting two BR datasets

modified.

page 5 - at bottom of page you are (I think) first mentioning having used flow visualization - could you mention briefly what you used for this

For more details regarding flow visualization, the first author’s previous article was mentioned as a reference in the manuscript (Please see page. 5).

Page 6 - "... Injecting nitrogen as coolant stream,” - add "the" after "as"

Modified.

page 6 - "These findings are in general agreement with Gritsch ..." – the words "general agreement" do not seem to be in keeping with the high quality of this paper - is there any more you can say ?

The word “general” has been deleted. Unfortunately, a direct comparison with Gritsch data is not possible, due to the different testing conditions, more specifically the lower density ratio of the present data. Nevertheless, the general trend of effectiveness variation with BR is correctly replicated as well as the smooth decay going downstream.

page 8 - "Jet stays close ..." - change to "The jet stays ..."

Modified.

Reviewer 2 Report

Mandatory Request 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.:

Figure 1 is too small to be read, increase size to make it readable.

Figure 3 is better but still too small, increase size as well.

Recommended Requested Changes: Changes will improve the quality of the paper. Authors are strongly encouraged to comply with these requests.:

Please check file

Comments for author File: Comments.pdf

Author Response

Page 7, 2nd paragraph: "deviation from experimental data raises to about 10% at the highest

X/D of 18."The deviation is about 50%, relatively speaking. 10% is misleading. You should state

that the error is about 0.1 in film effectiveness instead. For the cooling engineer designing a

vane or blade, this deviation in average eta at x/d = 18 is not acceptable and from that point of

view, the RSM model is not providing satisfactory results.

Focusing on Fig. 13, at X/D=18, RSM provided ave = 0.225 to be compared with the measured ave of 0.195. So the difference is 0.03, corresponding to about 15% of the experimental ave value. Authors are aware that a deviation of 15% from experimental data may be relevant in terms of absolute vane temperature. But there is no doubt that RSM predictions of ave are the closest to PSP data.

Page 8, 2nd paragraph: "eta shows a narrowing of the coolant footprint on the surface at about

9D downstream..." Looking at figure 10a, the narrowing is quite uniform along the length. I

cannot see anything special at 9D. Can you please clarify? It seems that more mixing is present

in the experiment, which is also in line with the average eta comparison.

The authors agree with the reviewer. The sentence has been rewritten: “the experimental pattern

of  shows a slight narrowing of the coolant footprint on the surface, at increasing distance from the

hole exit, whereas the RSM model provides a constant width coolant trace travelling in the stream wise direction. This indicates that lateral mixing is underestimated by RSM simulation, consistently with the above mentioned profiles of ƞave.”

Page 8, 3rd paragraph: "However, it seems that RSM overpredicts the jet spreading in the

spanwise direction... This is in contradiction to the results shown in figures 16 + 10a and figure

13. Please clarify.

The authors apologize for the mistake. We wrote “spanwise” instead of vertical.

The sentence at pag.8 has been rewritten as following: “It seems that RSM slightly overpredicts

the vertical spreading of the jet when comparing to HW results.”

Just to clarify, RSM slightly over -predicts the jet vertical spreading (see Fig. 18) and underpredicts

the lateral mixing (see Fig. 16 vs. 10a and Fig. 13) when comparing to experimental

results. There is no contradiction between the two.

Reviewer 3 Report

Mandatory Request 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.:

See PDF

See comments starting with MR:

Recommended Requested Changes: Changes will improve the quality of the paper. Authors are strongly encouraged to comply with these requests.:

See PDF See comments starting with RR:

Author Response

RR: “13th” instead of “12th”

Modified.

MR: Ratios instead of “ratio”

Modified.

MR: Pressure instead of “pressure”

Modified.

MR: please add a separate sub group for abbreviations

Added.

RR: please add for all symbols the adequate dimension.

OK.

RR: The authors should provide information according the near wall solution methods of the

investigated turbulence models, e.g. if wall functions are applied or not

The following was added to the manuscript: “Since all turbulence models have been used with a

two-layer zonal model for near-wall treatment, values of y+ were on the order of unity.”

MR: Please add references for these correlations

Those correlations are commonly used in CFD applications: the reference has been added to

pag. 10.

RR: The authors should provide information according near wall resolution, i.e. y+ values for

the investigated meshes.

These values have been added to Table 2.

MR: The authors could give some more meshing details. For example does a refinement grid

size of 0.001 mean that about 5 mesh elements will cover the complete hole diameter

D=0.005?

Meshing details have been added to the manuscript (see fig. 3 and table 2). The refinement grid

size refers to grid spacing in XYZ direction since a cylindrical volumetric control was used to

refine the mesh near the desired hole. The cylindrical volumetric control covers -3D upstream of

the hole (including hole itself), and 20D downstream of the hole trailing edge in X direction, ±3D

in Z direction with its origin at hole center as shown in Fig. 2, and ±5D in Y direction starting

from the flat plate surface.

RR: Change to BR=1 and BR=2, as only these are presented.

Modified.

MR: Better if authors check the SST results and provide a non-speculative statement. Further

are similar experiences for STAR-CCM+ computations in conjunction with the SST available?

The velocity contours predicted by SST k-w at X/D=1 show a huge separation of the flow from

one side of the diffuser, thus leading to an asymmetrical jet coverage and a narrower coolant

trace on the surface, compared to the other two cases. This results in lower values of laterally

averaged and centerline adiabatic effectiveness. The following was reported at pag. 7: “This is

due to SST KW tendency to overestimate separation of the cooling jet, thus leading to a

narrower trace on the surface and, consequently, lower.”

With regard to similar experiences, the k-w model implemented in STAR-CCM was found to split

the cooling air jet inside the diffuser part of the hole, with strong vortices within the

corresponding region [Dickhoff J. et al., CFD Simulations for Film Cooling Holes: Comparison

Between Different Isotropic and Anisotropic Eddy Viscosity Models, GT2018-75543].

RR: Fig. 17 (at Z/D=0?) and Fig 18 (at X/D=5) Different U/Ue values in Fig. 17 and Fig. 18 for

same location (see lines below). Especially for CFD results these U/Ue should be identical!

Please re-check results. Please use identical ranges for Y/D (e.g. 0-3) and levels for U/Ue (e.g. 0-

1) for both figures.

Identical levels of U/Ue (i.e. 0-1) have been applied to Figs. 17 and 18. Concerning the Y/D scale,

we prefer to keep them as they are. Otherwise the size of Fig. 18 becomes too small.

RR: Please add Z/D position

Added.

RR: Please add X/D position

Added.

Reviewer 4 Report

Mandatory Request 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.:

Introduction:

“…to the off-the-wall flow features which determine the thermal performance.” What do the author

mean? Thermal performance is the amount of heat flux that passes at the surface of the wall.

Experimental setup:

What medium was used as cooling fuild? Air, CO2? To achieve engine representative density ratio, it is quite common practice for lab experiments to use CO2 as cooling medium.

Fig 1. Add the coordinate system X-Z at the center hole exit.

Fig 2. Eliminate the Z axis pointing diagonally. This is not a 3D view.

Computational setup:

What is the medium of the coolant used? Air, CO2? And what is the medium of the main stream?

Combustion gas representative fuild? Or standard air?

Computational grid:

What was the total height was the 8 prism layers.

Show the volumetric control used. Also show a figure displaying the mesh.

The authors mention 3 different grid settings we used but figure 4 of the BL velocity profile shows

the same number of points, with more than 8 points in the layer in contradiction with the mesh

description provided. Please clarify.

Nice experimental results in Fig 6 and 7. Please add the X/D location line or planes on a figure.

Thermal aspect of film cooling:

Here N2 is mentioned as cooling fluid. It would have been best to use CO2 to better capture engine

representative DR. Note that Wagner et all (2006) confirmed CO2 can be used as fluid in PSP

experiment.

Add experiment or analytical in legend of Fig12-15. Also add that all numerical results are for BR=1

The author claim RSM to be the most accurate model, but RKE is not too different too.

RSM validation against experimental data:

Fig 16. Mention BR=1.

To be complete this paper needs to show RSM comparison with experimental data for both BR of 1

and 2.

Recommended Requested Changes: Changes will improve the quality of the paper.  Authors are strongly encouraged to comply with these requests.:

Overall, interesting paper with good balance between experimental and numerical information.

Unfortunately incomplete at this time. The authors need to make at least the mandatory changes for the paper to be accepted for the conference.

Author Response

MR

Introduction:

“…to the off-the-wall flow features which determine the thermal performance.” What do the

author mean? Thermal performance is the amount of heat flux that passes at the surface of the

wall.

Rephrased to: “Flow field in the near-hole region which determines the thermal coverage”.

Experimental setup:

What medium was used as cooling fluid? Air, CO2? To achieve engine representative density

ratio, it is quite common practice for lab experiments to use CO2 as cooling medium.

The aerodynamic measurements have been done using Air as coolant while thermal

measurements have been performed with N2 as well as CO2. Given the page restrictions, and to

be consistent with aerodynamic data, only PSP results with N2 as coolant have been reported. In

the abstract, it was stated that the density ratio is near 1.

Fig 1. Add the coordinate system X-Z at the center hole exit.

The coordinate system X-Z at the center hole exit has been added.

Fig 2. Eliminate the Z axis pointing diagonally. This is not a 3D view.

This figure is replaced by Fig.3, which along with modified Fig. 1, may show the coordinate

system clearly.

Computational setup:

What is the medium of the coolant used? Air, CO2? And what is the medium of the main

stream? Combustion gas representative fluid? Or standard air?

The medium of both the mainstream and the coolant is standard air at 288K and 323K,

respectively. The following was specified at pag. 3: “The working fluid is assumed to be

incompressible air with temperature-dependent properties.”

Computational grid:

What was the total height was the 8 prism layers.

The total height of prism layers and mesh refinement was 0.44D, and 5D, starting from the flat

plate surface, respectively (see pag. 3). 5D is the radius of the applied cylindrical volumetric

control. To be clearer, pictures of prism layers and mesh refinement (i.e. Cylindrical Volumetric

control) have been reported in Fig. 3.

Show the volumetric control used. Also show a figure displaying the mesh.

Please refer to the new Fig. 3.

The authors mention 3 different grid settings we used but figure 4 of the BL velocity profile

shows the same number of points, with more than 8 points in the layer in contradiction with

the mesh description provided. Please clarify.

The points in the velocity profiles do not correspond to the prism layer spacing. They are just

chosen locations where the velocity U has been extracted from medium/fine grid simulations.

Nice experimental results in Fig 6 and 7. Please add the X/D location line or planes on a figure.

X/D location lines are added to Fig. 1.

Thermal aspect of film cooling:

Here N2 is mentioned as cooling fluid. It would have been best to use CO2 to better capture

engine representative DR. Note that Wagner et all (2006) confirmed CO2 can be used as fluid in

PSP experiment.

Thermal measurements have also been performed with CO2 as coolant. Although, only PSP

results with N2 as coolant have been reported given the page restrictions. As mentioned before,

this choice was motivated by the fact that PIV and LDV measurements could only be performed

using air as coolant. To be consistent with the aerodynamic data, only PSP results obtained

injecting nitrogen have been here reported.

Add experiment or analytical in legend of Fig12-15. Also add that all numerical results are for

BR=1.

Added.

The authors claim RSM to be the most accurate model, but RKE is not too different too.

Actually RSM and RKE deliver quite similar results but RSM can ensure the most accurate

prediction of laterally-averaged effectiveness and velocity profile at X=1D. This was specified at

pag. 7: “one could conclude that the RSM is the most appropriate model to choose for fanshaped

holes in this set of experiments, thus minimizing the error in the prediction of ƞave and

U/Ue, at least in the near vicinity of the hole exit (X/D =1).”

RSM validation against experimental data: Fig 16. Mention BR=1.

Done

To be complete this paper needs to show RSM comparison with experimental data for both BR

of 1 and 2.

Experimental data for stress components at X/D=1 and 5, for BR=2, were added to Fig. 16 with

some comments at pag. 9. Please consider that CFD simulations were run only for BR=1, which is

the best trade-off between coolant consumption and cooling effectiveness.