Effect of inlet rotating swirl on endwall film cooling for two representative hole arrangements

Yang ZHOU,Yang ZHANG,Xinrong SU,Xin YUAN

Key Laboratory for Thermal Science and Power Engineering of Ministry of Education,Tsinghua University,Beijing 100084,China

1.Introduction

In modern turbomachinery,the constantly increasing turbine inlet temperature proposes stringent challenge to the design,and advanced cooling technology is heavily employed to protect the high temperature turbine components.1To reduce the NOxemission and preserve the combustion stability,swirl generator is widely used in lean-burn combustor.2Inside the combustor,the swirling flow is not fully smeared and contributes to the non-uniform flow field at the combustorturbine interface.As a result,the real inlet flow condition of turbine is non-uniform and the non-uniformity would have negative impact on the aerodynamics and heat transfer features of the turbine.Consequently,the investigation on inlet swirl and Inlet Guide Vane(IGV)endwall film-cooling3is exceedingly indispensable.

In the last decade,several related studies concerning the combustor-turbine interactions were made.Shih and Lin4studied the interaction mechanism between inlet swirl and leading-edge airfoil fillet.They controlled secondary flow using these two methods to reduce aerodynamic loss and vane surface heat transfer and found that the swirls and fillets increased the size of the stagnation region on the endwall around the airfoil’s leading edge.Fillets accomplished this by geometry

Nomenclature

contouring and swirl flow achieved this by more kinetic energy.But they did not discuss the effect of inlet swirl on the film covering on the endwall.Researchers in Oxford University proposed the Integrated Combustor Vane(ICV)concept,Rosic et al.5investigated numerically the interaction between the combustor and the first vane in an industrial turbine inside which 16 can-com bus tors and 32 vanes were installed.It was claimed that the combustor walls had dreadful effects on the first vane film cooling.Promising combustor-vane integration minimized the axial distance between the combustor and vane.In addition,the cooling and aerodynamics performance of the turbo system was not detrimentally affected.Beneficial in reducing secondary flows in the ICV vane,the integrated design also saved up to 25%of the total coolant.Jacobi and Rosic6compared a ero thermal experimental data and their inhouse Computational Fluid Dynamics(CFD)code TBLOCK simulation to analyse the decrease in total pressure loss coefficient and heat transfer coefficient levels of integrated vanes.Their method halved the number of vanes,eliminated the coolant need of leading edge and considerably reduced manufacturing and development costs.Furthermore,Khanal et al.studied the effect of combustor hot streak and swirl on the unsteady aerothermal performance of high pressure turbine stage based on MT1,7then proposed a combined performance parameter which enabled aerodynamics and heat transfer behaviour to be quanti fied and marked in their solver HYDRA.According to He et al.8,the combined interplay of hot-streak and swirl resulted in crucial variations in heat transfer and aerodynamic performance.Qureshi et al.9investigated the effect of residual swirl from combustor exit on thermodynamics by comparing with and without inlet swirl.Inside their research,experimental part presented time-averaged heat transfer and static pressure measurement in an unshrouded turbine,and numerical simulation part based on Rolls-Royce in-house code HYDRA also were proceeded.In view of Oxford colleagues’research,Beard et al.10dig deeper to study the relationship between inlet swirl and stage efficiency.They conducted the experiment about the effect of swirl on the turbine stage efficiency with an efficiency measurement system capable of resolving efficiency changes within±0.16%.However,their current research hardly focused on endwall film cooling which is very significant in modern gas turbine system.Yang et al.11predicted film cooling and heat transfer characteristic by numeric al simulations with consideration of stator-rotor purge flow and discrete hole flows.The calculation was carried out based on Reynolds stress turbulence model with a non-equilibrium wall function.Las kowski et al.12conducted several research cases about turbine vanes with leading edge film cooling in the conjugate calculation method which can be in analogy with endw all film cooling.Dong et al.13constructed a CFD model to study the impact of vane wakes and leading edge bow wakes on ingestion,meanwhile,predicted turbine wheels pace cooling flow interactions with transonic hot gas path.These simulation research was meaningful in the study of industrial complicated calculations,especially for actual turbine working conditions,such as inlet swirl.

As for experimental investigation,Cha et al.14measured temperature and flow fields of full-annual,rich-burn com bus tors and high pressure turbines using passive scalar tracing methods which were isothermal and non-reacting.The complementary results generally provided a datasets benchmark for conventional turbines design in the future.Koupper et al.15developed a kind of engine representative combustor simulator to generate hot streak pro file.This trisector rig was equipped with an automatic control system which allowed the facility to operate at stable situations for long time runs.The research results mentioned above were conducted on the industrial sized rigs,it is more difficult to monitor the exact mechanism about the flow phenomenon,than the same experiments on simple cascades.Especially for catalytic and dry low NOxcombustor,Ames et al.16measured endwall heat transfer distributions under turbulence from the combustor exit in a large-scale,low speed linear cascade facility.They made the assumption that the individual vortex had less evident influences.With consideration of a large range of Re numbers and turbulence conditions,the database of the exercise could be an assessment of endwall thermal simulation capabilities.Hedlund et al.17observed heat transfer and flow phenomena in a swirl combustor and relative turbine blade passage,and found that surface Nusselt numbers and time-averaged flow characteristics had important variations because of arrays of passage vortex pairs.Papa et al.18used naphthalene-saturated air and oil-dot visual-ization to measure the mass/heat transfer on the endwall surface in a linear cascade.The experimental results by Barringer et al.19indicated that the inlet total pressure pro files affected the aerodynamics loading by as much as 10%,revealing that the combination of different total pressure and total temperature pro files had significant effect.The bold exercise was conducted with the endwall cooling hole sets located along is ovelocity lines,this idea was further analogized to holes sets decorated along is o-pressure lines in this paper.

Some novel experimental methodologies and new test facilities were developed to measure more information of a erothermal fields in detail.Luque et al.20,21proposed a new technique to accurately evaluate cooling system performance.Taking engine-representative geometry,and internal and external cooling flow into consideration,the facility based on thermochromics liquid crystal provided a totally integrated approach to assessing the cooling system.Besides the above method,Luque et al.also made use of five-hole probes and transient infrared thermography technique to monitor flow information such as:turbulence,swirl,leakage flow,shed wake distributions,and surface coolant flows.Krichbaum et al.22introduced an extensively redesigned large scale turbine rig,which was a full annular 1.5-stage axial low speed turbine and equipped with combustor simulator modules including swirl generators.

As is known,endwall is always the emphasis of turbine cooling.Due to the difficulty of extremely accurate CFD prediction of endwall film cooling phenomenon,most of the investigation on endwall film cooling were conducted based on experimental measurements.Zhang and Jaiswal23compared the cooling results of double-row hole injection and singlerow discrete slot injection using Pressure Sensitive Paint(PSP).They argued the secondary flow which dominated the near endwall flow is at low mass flow ratio.On the contrary,the cooling film flow with a higher momentum dominated the near wall flow field and suppressed the secondary flow at higher mass flow ratios.Still based on PSP technique,Li et al.24investigated experimental impacts of swirl purge flows on platform cooling and phantom cooling of suction surface.The main parameters on cooling effectiveness were mass flow ratio and swirl ratio,which could change coolant distribution on platform close to suction side and phantom cooling result.On a multi-stage rotating test facility,Suryanarayanan et al.imitated real rotating conditions to measure cooling effectiveness on blade platform25,and suggested that film cooling effectiveness increased with an increase in the coolant-to mainstream blowing ratio(M)for different all rotating speeds,respectively.26

So far,related research literatures27mainly focused on the effect analysis of interaction between inlet non-uniform flow and the whole turbine aerothermal performance in the downstream.But few investigations were about the most important heat transfer feature,turbine component cooling.Moreover,scant studies about endwall film-cooling hole arrangement criterions were searchable in this field.These literatures did not explain clearly the mechanism of swirl effect on film cooling.There were almost no related research about the effect of inlet swirl on endwall cooling and how to suppress this turbulence influence with the method of changing holes arrangement strategy.Besides,limited by calculation ability,it must be admitted that current numerical simulation had less accuracy and less precision than experimental measurements.Thus,using experimental methods to clarify the effect of inlet swirl on endwall cooling is significant.A more in-depth arrangement plan comparing two representative cooling-hole to expound the reason of the difference is quite meaningful.Based on our view that specially designed holes arrangement could be beneficial to cooling effectiveness and the development of cooling film,it was in novatively inferred that new holes arrangement under inlet swirl conditions,are supposed to contribute better cooling effectiveness.In this paper,to resist the film lift-off effect of inlet flow non-uniformities and protect the endwall from being ablated by high temperature gas,a new hole arrangement was optimized and chosen by CFD simulations according to Design Guidelines that film-cooling holes lay along the isobars on the endwall.As for this experimental research,we used a clockwise swirl generator model embedded in a removable plexiglas plate to imitate swirl flow induced from the combustor in the real turbine set.The cooling effectiveness were measured under variable blowing ratios on two typical endwalls,then the mechanism of result physical phenomenon were analysed.The novel discoveries are that inlet swirl has hugely devastating effect on endwall film coverage,and the endwall on which holes lie on isobars has strong resistance against swirl effect.

As for the structure of this paper,experimental theory and test rig facility will be introduced firstly,then measurement plan in different groups will be clari fied;the final part is conclusion and discussion about the results.

2.Experimental methodology

In this experiment,PSP technique based on oxygen-quenched photoluminescence was used as the measurement method.PSP is a kind of contactless measurement paint,with the reacting theory resolved by Zhang and Jaiwal.23In current experiments,Light-Emitting Diode(LED)lights whose wavelength are 450 nm are used to activate PSP so that PSP could emit reflected light,then the light will be captured by a Charge-Coupled Device(CCD)camera with high spectral sensitivity.The CCD camera has a band pass filter whose traversing wavelength is 600 nm,to catch the 600 nm reflected light by PSP in the energized state.In this study,the main stream is air which contains about 21%oxygen.The coolant is industrial nitrogen for which the partial pressure of oxygen is almost 0.With the consideration of the analogy between mass transfer and heat transfer,oxygen concentration and partial pressure could express the film effectiveness in terms of Eq.(1)as below:

where η means the film cooling effectiveness,Cairrepresents the oxygen concentration of the main stream,Cmixrepresents the oxygen concentration of the mixture of air/nitrogen,CN2represents the oxygen concentration of the nitrogen,which equals to 0.Meanwhile,(PO2)airrepresents the oxygen partial pressure of the main stream,(PO2)mixrepresents the oxygen partial pressure of the mixture of air/nitrogen.With this technique,the measured film cooling effectiveness ranges between 0%(at far upstream and far downstream locations)and 100%(inside the film cooling hole).

In each PSP film-cooling test,CCD camera takes 4 pictures at the same main stream temperature value to measure the film-cooling effectiveness accurately.28The No.1 picture is a‘dark’image that is photographed without LED light and is the main stream flow condition.The No.2 picture is the‘reference’image photographed without the main stream flow,but with the LED light on.The No.3 picture is ‘air’image is photographed with both the main stream flow and LED light on,and the coolant is totally air injection as mainstream.The No.4 picture is ‘N2’image photographed in the same condition as No.3,except that the coolant is pure nitrogen injection instead of air.By visualization program,the effectiveness distribution could be obtained from the four figures for every test sequence.The information on the change in oxygen concentration due to the change in oxygen partial pressure,which cannot affect the magnitude of local film-cooling effectiveness.This could be provided by the distribution of reference ratio data in the‘air’image.Compared with the reference ratio data,the other one from the ‘nitrogen’image could make contributions to effectiveness by absolute oxygen concentrations.Referring to the analogy between mass transfer and heat transfer,the post algorithm obtains the local adiabatic film-cooling effectiveness distributions by these two groups of reference ratio distributions.

It is necessary to calibrate the PSP employed along the test surface to obtain the relationships between light intensities and local partial pressures of oxygen,before every experimental measurement.For this goal,a calibration system is established which is shown in Fig.1.During the calibration of the PSP system,a PSP-coated copper coupon is employed as the model of the experimental surface,underneath which three thermocouples are mounted to monitor the unsteady change of surface temperature.This PSP-coated copper coupon is placed in a partial or total vacuum room,sealed by machinery.A heater located at the back of the coupon increases both the temperature of the copper coupon and the heater to keep the desired temperature of the sample with accuracy better than±0.5 K,continuously.Through a transparent plexiglas window,the sample coupon is in the vision of CCD camera.Since the surface temperatures in this system varies from 298 K to 308 K,the calibrations are conducted for surface temperatures of 276.5 K,298 K,and 308 K,in addition,for pressures from partial vacuum(almost 1/3 of an atmosphere)to one standard atmosphere,respectively.These three temperature points of calibration undertaken make sure the completeness of influence information of temperature variation.The interdependence between the light intensity factor and pressure factor is shown in Fig.1,which shows the calibration result curve.In this figure,the abscissa is dimensionless relative pressure ratio,where P means local nitrogen partial pressure,and Prefmeans the reference nitrogen partial pressure.The ordinate is dimensionless relative light intensity,where I means local light intensity,and Irefmeans reference light intensity.It is obvious that there is little dependence of calibration data on surface temperature,and the quantitative change of the three data sets of different temperature values are tiny.

Based on the calibration results,the conventional dimensionless temperature around the cooling areas also decided by light intensity,could be expressed in terms of Eq.(2)as below:

where θ means dimensionless relative temperature,T∞means temperature of freestream,Tcmeans temperature of injected coolant,T means local temperature.

For constant property and low speed flow,after importing the adiabatic wall temperature parameter,the equation yields dimensionless film-cooling effectiveness.Furthermore,the final adiabatic film effectiveness could be represented in terms of Eq.(3)as below:

where Tawmeans the temperature of adiabatic wall,η means the same cooling effectiveness as in Eq.(1).

As for the uncertainty and tolerance,the confidence of the calibration is on 95%extent.The measurement relative tolerance of the adiabatic film-cooling effectiveness is almost 3%,and the magnitude of uncertainty value is less than 0.5.However,if the effectiveness is close to 0,the non-confidence will increase,even to a magnitude of almost 20%when the value is about 0.05.The cumulative uncertainty is the superposition result from calibration error(4%)and deviation of capturing pictures(1%).Thus,the absolute error of effectiveness measurement changes during 0.01–0.02.As for extreme assumption,maybe the relative uncertainty for extremely low effectiveness(0.01)would be exceedingly high,about 100%.

3.Experimental facility

The test facility is composed of an inlet segment,a linear turbine cascade and an exit section.In the experiments,the free-stream air velocity is 24.3 m/s,which corresponds to a Mach number of 0.075.In the measurement section,four GE-E3guide vanes are mounted.The tested cascade29has a scaling ratio of 2.2,with its chord length equal to 79 mm and height 129 mm.In the current study,the effect of the non-uniform flow field at the combustor-turbine interface is generated with a swirl generator,as demonstrated in Fig.2.

Also,for comparative study,the swirl generator module is removable and in this manner the experiment with uniform freestream can be conducted very easily.The swirl intensity is defined as Swirl Number in the terms of Eq.(4)as below,which is a widely used parameter in both research field and engineering:

where S represents Swirl Number,Rinrepresents the inside radius of swirl generator,Routrepresents the outside radius,and the φ represents the mounting angles of blades in the swirl generator.As for the current con figuration as given Fig.2,the inside radius is 14 mm and the outside radius is 48 mm,with which the swirl number is about 1 and this swirl intensity is representative in modern lean-burn combustor.

During the entire experimental process,the swirl generator was installed at the same position,the accurately middle of the span wise test-section width,to keep the swirl flow conditions constant.Almost all components of the test section can be replaced including endwall parts,swirl generator,cooled vanes and coolant supply cavities which is installed below endwall plane.Since the holes arrangement on the endwall is the primary concern during this research,no other components except two representative endwall parts is changed.

The test rig30is shown in Fig.3(the real system diagram)and Fig.4(the schematic diagram).

To compare the film-cooling effectiveness distribution accurately,the relative position between the CCD camera and the main endwall module is held stable constantly during the experiments.A high power compressor supplies enough coolant air for the secondary air system,and the compressor delivers coolant into cavities located under the wind tunnel test section.To realize the practical conditions inside industry turbine as real as possible,the structure of coolant cavities was designed as below:all secondary air supply share one communal passage which is connected with Air or Nitrogen gas cylinder,the total mass flow could be controlled by flow meter.After the shared general coolant passage,several cavities are linked between coolant supply outlet and cooling holes inlet,respectively.With the consideration of processing convenience,each row of cooling holes on endwall is allocated with one divided cavity,by which the coolant could go spouted outside the fan-shaped holes.Since all cavities share one coolant supply which means the same total pressure,the distribution of coolant would be dominated by pressure difference of each cavity,instead of active interference.During the entire experiments to all tested cases,the total coolant mass flow was changed to simulate mutative M,which is a macro variable,and no more active control on coolant distribution was adopted.Above is the working mechanism of secondary air system on this test rig,which is shown in Fig.4(a).Consequently,the coolant supply structure and working principle are permanent,no matter which tested part is installed.

As for the hole shapes,it is widely known that the jet de flection of fan-shaped holes could result in a wider coolant diffusion area.31Therefore,typical fan-shaped holes with the geometry features shown in Fig.5 were chosen as the holes on the experimental endwalls.

There are two kinds of endwalls used in the test.The first one is called ‘Type A’endwall.Taking the vortex sweeping coolant from endwall to the mid-span region into consideration,all holes are arranged in straight lines perpendicular to the axial to re flect the blowing-off phenomenon more clearly.To cover the gill region fully,the cooling structure consists of four rows of fan-shaped film-cooling holes.This endwall is shown in Fig.6(a).As shown in the figure,the first row of holes is fixed at upstream of the vane leading-edge,and the other three rows of holes are located in the passage.The angle between the axial of fan-shaped holes and platform surface is 30°,and the compound angles for the main axial direction are 0°,30°,45°and 60°for the four rows of holes,respectively.Both the lateral expansion and forward expansion into the platform are 10°.The hole diameter of the cylindrical part below the expansion is 1 mm,as suggested in open literatures and used widely in industry.

The second one is called ‘Type B’endwall,for the purpose of resisting the film blow-off effect.With the consideration of the same outlet back pressure making sure the coolant flow evenly,all holes are arranged along the isobars except the first row.Under the permission of structural strength,coolant flows out from five rows of holes to protect the whole gill region,as shown in Fig.6(b).The first row is located upstream of the leading-edge,and the row is separated to be two parts,the part around the pressure-side of channel is moved toward downstream to obtain better cooling effectiveness.The other holes is fixed up with both the consideration of isobars and the structural strength,so this endwall could be manufactured for industry.As shown in the figure,the second row includes ten holes,with three discrete holes located as a triangle near the pressure-side among the ten.The third row of holes starts from the head of suction-side in the passage and ends at the mid-Caxof pressure-side in the passage.The fourth and fifth rows are arranged downstream,nearly parallel to the main axial direction of the turbine stage.Overall,the cooling holes arrangement could be seen in the figure.For every single hole,the geometry size and expansion angles of the ‘Type A’endwall are the same as the ‘Type B’endwall.Under the platform with the holes,every row is installed correspondingly with a coolant cavity to simulate the secondary-air system,and all cavities share one coolant source,like real turbine cooling condition.The coolant flow could be changed by rotameters connected with the cavities and coolant source.

The information about the geometric and position data of film holes on ‘Type A’and ‘Type B’endwall is shown in Tables 1 and 2,respectively.In these tables,X/Caxmeans relative position of each hole,Caxmeans actual chord length of scaled-up blade pro file at axial chord direction.D means diameter of each cooling hole.Meanwhile,the coolant flow in the experiments is listed in Table 3,the detail information about the geometric and flow conditions of test rig is shown in Table 4.In Table 3,M represents blowing ratio in the experiments plan.

4.Results and discussion

In this study,the two types of endwalls with different film cooling hole arrangements were tested with both uniform and swirling inlet conditions to study their performance within the practical turbine environment.A broad range of M value from 0.2 to 1.3 were investigated in the experiments,with emphasis on the off-design non-uniform performance of the endwall film cooling.

4.1.Cooling effect difference between two endwalls without swirl

The contours of film-cooling effectiveness on two gill regions with uniform freestream are shown in Figs.7–10.For the series of cooling effectiveness contours,abscissa is dimensionless chord length,X/Caxmeans the ratio of axial position and actual chord length of scaled-up blade pro file at axial chord direction.Ordinate is dimensionless pitchwise length,ZPmeans length of one pitch at the pitchwise direction.

It is pretty obvious that the ‘Type B’endwall could obtain a more satis fied cooling effectiveness distribution under the same conditions.For the designed working M situation,Fig.7 shows film-cooling effectiveness distribution on ‘Type B’endwall without inlet swirl when the M is 0.7.It is obvious in the image that almost the whole gill region around film-cooling holes could be protected by coolant coverage,except the single holes located closest to leading edge of pressure side,where the outlet backpressure on the platform surface is too high toallow the coolant flow out heavily.Thus,this area of gill region near leading-edge of pressure-side is crucial to the cooling blind area.With consideration of the blind area,the total outcome of film-cooling effectiveness is pretty satisfying.There is almost noplace without coolant coverage in the downstream of the first row.Fig.8 shows film-cooling effectiveness distribution on ‘Type A’endwall without inlet swirl when M is 0.7,and the film-cooling effectiveness distribution is also satisfying and there forms a successive coverage.However,besides the blind area around leading-edge of pressure-side,some area around holes in the middle of first and second rows cannot supply enough coolant to protect the endwall surface,especially for the 3rd,4th,6th,7th,9th hole in the second row.During the second row and the third row,the coolant film is thin and the cooling effectiveness is lower than downstream behind the third row.Figs.9 and 10 show film-cooling effectiveness distribution on ‘Type B’and ‘Type A’endwall without inlet swirl when the M is 0.3,respectively.This is for further off-designed work conditions,with decreasing coolant out flow.Compared with Figs.7 and 8,an intensive difference of filmcooling effectiveness could be observed.For ‘Type B’endwall,there still is a successive film of coolant,although one single hole in the blind area cannot allow coolant flow out heavily and the effectiveness of some areas near the pressure-side is somewhat poor.In general,the effect of film-cooling is pretty good,even M is so low as 0.3.In contrast,for ‘Type A’endwall,the film of coolant is not continuous when M is low,as shown in Fig.10.The pitchwise lines at 20%Caxand 75%Caxare chosen to reflect quantitatively the cooling effectiveness of upstream and downstream,and the line locations are shown in Figs.7–10,as 1–1,2–2,3–3,and 4–4.The cooling effectiveness distributions along pitchwise lines are shown in

Figs.11 and 12.For these series of curve graphs,abscissa is pitchwise location,where Span means one pitch size,Z means local pitchwise coordinate.Ordinate is cooling effectiveness.At the upstream,the film-cooling effectiveness of the‘Type B’endwall is overwhelmingly more favourable than the ‘Type A’endwall.At the downstream,the averaged film-cooling effectiveness of‘Type A’endwall is slightly more positive than‘Type B’endwall.However,the effectiveness value of the ‘Type B’endwall is close to 0.3.This value is near that of the ‘Type A’endwall and is satis fied.Compared with Figs.11 and 12,it is found that the upstream effectiveness value is close to the downstream for the ‘Type B’endwall,the value is around 0.35,meaning the coolant flow out is generally very uniform.On the contrary,the situation for ‘Type A’is unsatisfied.The upstream effectiveness value is around 0.2 and the downstream effectiveness value is around 0.4,with consideration of effectiveness contours,obtaining the result that the coolant flow out is not uniform at all.In Fig.12,there are two peaks on the curve of‘Type B’endwall,because the 75%Caxpitchwise line passes through film-cooling holes on the platform.Thus,the maximum effectiveness appears without significant

meaning,the averaged effectiveness is more persuasive,therefore,‘Type B’endwall has more homogeneous cooling effect.

Table 2 Film holes location and orientation for ‘Type B’endwall.

Table 3 Injection coolant flow in experiments for two endwalls(irrelevant to swirl).

Table 4 Geometric and flow conditions.

To sum the performance comparison between two endwalls without inlet swirl, ‘Type B’generally overwhelms ‘Type A’,according to the detailed experimental results.It is mainly because with almost all holes lay on the isobars,the pressure gradients between secondary-air pressure and outlet back pressure on the isobars are close,so the flow out of cooling holes are more uniform,this phenomenon will lead to less coolant momentum difference,which means less interaction between coolant secondary flow and free stream,retaining film in longer distance and bigger lateral expansion area for shaped holes.More uniform coolant flow allocation also avoids the problem that some holes cannot appear unobstructed coolant flow.On the other side,for the ‘Type B’endwall,due to smaller distance between adjacent rows than ‘Type A’,the coolant from upstream rows is able to survive until covering the near downstream rows,this overlay consequent also makes contribution to better cooling effect for ‘Type B’endwall.

4.2.Effect of swirl on endwall cooling for each endwall

The effects of inlet non-uniform swirl to endwall film cooling are key points of this research.Two different endwall cooling structure design forms were studied respectively under representative swirl conditions whose intensity is equal to 1.

With the inlet swirl,generally on the endwall,there appears a massive blow away instead of continuous film.Fig.13 is the film-cooling effectiveness distribution on ‘Type A’endwall with inlet swirl of the M equal to 0.7,corresponding to the result without swirl shown in Fig.8.The comparison indicates the huge adverse influence of swirl disturbance when M is high.Under this situation,the effect of film cooling is unsatis fied,and there in no clear connection between adjacent rows,meaning the coolant film is not successive at all.Fig.14 is filmcooling effectiveness distribution on ‘Type A’endwall with inlet swirl of M equal to 0.3,corresponding to Fig.10.The comparison shows that inlet swirl has a devastating influence on the cooling effectiveness.For instance,half of the rows of holes cannot allow coolant to flow out,which means the mainstream with high temperatures would flow backward into endwall to melt the components.Laterally averaged film-cooling effectiveness distribution along the axial direction on the‘Type A’endwall with different M values without inlet swirl is shown in Fig.15.In the figure,the result of the averaged cooling effectiveness gets decreasing as M decreases,monotonically.The infomation with inlet swirl is shown in Fig.15,too.It is evident that the swirl has a huge influence on end wall film cooling effectiveness,especially at the upstream of the gill region,the effectiveness is sensitive to swirl,and the sensitivity at downstream is slight,relatively.Also,the comparation clarifies there exists over-cooled accumulation coolant in downstream area,on which inlet swirl has just slight effect.

Fig.16 is the film-cooling effectiveness distribution on‘Type B’endwall with the inlet swirl and M equals to 0.7,corresponding to Fig.7.The comparison indicates that the inlet swirl also has a damage impact on the cooling effectiveness on ‘Type B’endwall.Although the whole pattern of coolant film is not changed,the coolant traces behind the cooling holes almost disappears,instead of long wakes in the Fig.7.Fig.17 is the film-cooling effectiveness distribution on ‘Type B’endwall with the inlet swirl of M equal to 0.3,corresponding to Fig.9.The difference between Figs.9 and 17 suggests the swirl impact on cooling effectiveness with low M.It is observed that four holes in the first row inside the blind area cannot flow coolant,and the film near pressure-side is nothing.Under strong inlet swirl,the general film cooling effectiveness distribution is somewhat terrible.Laterally averaged film-cooling effectiveness distribution along the axial direction on the‘Type B’endwall with different M values without inlet swirl is shown in Fig.18.In the figure the trend of averaged cooling effectiveness is as low as the M,monotonically.The same information with inlet swirl is shown in Fig.18,too.

Aaccording to Fig.18,it is claimed that the swirl has considerable effects on the endwall film cooling effectiveness.However,due to uniform coolant allocation of‘Type B’endwall,the changing trends under one specific M are approximately similar.It is found that for the ‘Type B’endwalls on which the holes are along isobars,only under the low M and with swirl disturbed,the averaged cooling effectiveness is poor(lower than 0.2).The film effectiveness of other cases is still satisfying.

According to the experimental results and film lift-off theory,the mechanism could be modelled as below.Fig.19 shows the lift-off physical process that normally happens several D distance far from the ejection hole.The mainstream with momentum entrainment always conducts reciprocal momentum transfer with coolant kidney vortex pair, finally the vortex pair grow up and get lift-off,then no coolant exists between endwall and freestream hot gas,which means cooling film blows away.

Taking swirl perturbation into consideration,it is clear that the mutual effect between coolant and hot gas will be more intense.Due to powerful swirl flow,the secondary flow intensity and momentum entrainment of freestream will be more.Freestream has an extremely strong trend to merge vortex pair of coolant,and mixes with coolant vortices to conduct momentum and energy transfer, finally,one side of vortex pair of coolant will get enough energy and momentum to detach from endwall earlier,thus the balance of coolant vortex pair is broken.As the result, film coverage also cannot be maintained immediately.The final outcome is freestream and coolant vortices get blended together,like swirl flow sweeps coolant film away from endwall.It is the sweeping and blending effect of swirl perturbation that destroy the film coverage on the endwall.The sweeping increases the mixing process between mainstream and coolant ejection.During the interaction,swirl transfers more momentum to cooling ejection,these movements have a negative effect on film attachment and could damage cooling effectiveness.In the current paper,it could be found that ‘Type B’endwall has more resistance against swirl flow than traditional‘Type A’endwall,less sensitivities to inlet swirl,because of the uniform coolant allocation and holes laying on isobars.

4.3.Cooling effect difference between two endwalls with inlet swirl

Based on the above measurement data,the experiment results with consideration of both swirl and different endwall showed that under the same intensity of disturbance,‘Type B’endwall has stronger resistance against disturbance and the film coverage effectiveness is relatively high even after swirl disturbance.For both two endwalls,it is difficult for the coolant to flow outside the upstream area when M is low.However,there also is severe excessive cooling downstream on ‘Type B’.This is because the ‘Type B’endwall has less holes downstream where pressure is fairly low.The cost of coolant near downstream is less,so the upstream endwall gets more coolant allocation,the distribution of coolant can be more averaged overall.In upstream part,the coolant trail from the first row can cover the next row due to the fact that the holes are distributed in a high density.As a result,this superposition effect makes the better cooling effect.

Figs.13 and 16 show the film-cooling effectiveness distribution on ‘Type A’and ‘Type B’endwall with inlet swirl of M equal to 0.7,respectively.The comparison suggests that both the endwalls could be protected by successive coolant film.For ‘Type A’endwall the situation is more adverse.For instance,the cooling blind area around pressure-side and the area between adjacent rows of holes is hardly covered with film due to short of coolant.The momentum of coolant injection will join the convection and exchange momentum with swirl flow and mainstream.As a result,the cooling film would be swept by secondary flow in the passage.Figs.20 and 21 compare the film-cooling effectiveness distribution on ‘Type B’and‘Type A’endwall with inlet swirl of a mild M value,which is 0.5.

Furthermore,Figs.14 and 17 compare the film-cooling effectiveness distribution on ‘Type B’and ‘Type A’endwall respectively with inlet swirl of a low M value,which is 0.3.The tendency is clear that the film cooling effectiveness is getting lower as M gets lower,and the effectiveness distribution of‘Type B’endwall is more favourable than ‘Type A’endwall.When the M is extremely low at the value of 0.2,as shown in Figs.22 and 23,both distributions of cooling effectiveness are terrible under inlet swirl interference.For ‘Type B’endwall,the large area around the cooling blind area(dimensionless Pitch from 0.6 to 0.9,Caxfrom-0.2 to 0.6)almost has nothing in the coverage,and the situation is more dreadful for ‘Type A’endwall.On the ‘Type A’endwall,only three holes have the ability of supplying coolant out flow normally in the first row,and half of the holes in the second row cannot work as designed.Fig.24 shows pitchwise distribution of filmcooling effectiveness at 20%Caxwith inlet swirl when M is 0.5.Fig.25 is the same distribution of film-cooling effectiveness at 75%Caxwith inlet swirl when M is 0.5.Except the two peaks in the curves of‘Type B’endwall shown in Figs.24 and 25,for‘TypeB’endwalltheaveraged cooling effectiveness at upstream is approximately 0.3,and the value at downstream is nearly 0.2.In contrast,for the ‘Type A’endwall the averaged effectivenss at upstream is lower than 0.1,and the value at downstream is still about 0.25.Although the downstream of‘Type A’endwall is higher than ‘Type B’endwall,with the consideration of the accumulation of the coolant downstream,the effectiveness(0.25 level)is not necessary,and the effectiveness of‘Type B’endwall(equal to 0.2)is enough to protect the endwall.As for the paramount of cooling,the upstream of‘Type A’endwall has poor cooling effectiveness,but the situation of‘Type B’endwall is fairly promising.

To sum up the charcteristic of‘Type B’endwall and the reasons why the cooling is better,the keypoints are the uniform coolant allocation and overlay priciple.Inlet swirl does have a huge impact on film-cooling effectiveness,however,under high M values,the film coverage is not fragile and on endwall surfaces still form continuous coolant film.It means changes of cooling effect is not very sensitive to swirl.But when M is extremely low,the film coverage will be susceptible to swirl effect,coolant will blow away and be lift-off,the changes of cooling effect is sensitive to swirl disturbance at that time.

5.Conclusions

This research studied the effect of inlet swirl on the film effectiveness of‘Type B’endwall and ‘Type A’endwall under different M values by a test cascade tunnel rig.The difference of test conditions with varying M values and inlet uniformities indicates the reasons why film-cooling effectiveness is low somewhere.Film effectiveness is measured by pressure sensitive paint technology.In the experiments,the complete endwallsurfaceismainly compared with contentsunder different cooling structure protection,with the M varying from 1.3 to 0.2.In general,the inlet swirl has a huge in fluence on the endwall film-cooling.The coolant ejection and mainstream with strong secondary flow system get mixing together near the platform surface,therefore,sweeping the coolant around the mixture fluid vortex.

From this work,the following conclusions can be drawn:

(1)Hole arrangements of the endwall with holes along isobars have the ability to resist the inlet swirl disturbance and to protect the endwall surface under strong swirl conditions.

(2)The sensitivity of distribution of film cooling effectiveness on the endwall is low when the M is high.In contrast,the sensitivity is high when M is fairly low.

(3)The film cooling effectiveness of endwall with cooling holes along isobars is better than endwall with holes along straight lines vertical to axial,with other conditions remaining the same.

In the further step,more non-uniform features in the actual gas turbine will be investigated in experimental method,and the affected objects will be included but not limited to blade surface and tip plane.

Acknowledgements

This study was supported by National Natural Science Foundation of China(Nos.51476082 and 51506107).The language writing assistance from Mr.Ross Boyd and Mr.Zach Kossow was very helpful to this paper.

References

1.Han J.Fundamental gas turbine heat transfer.J Therm Sci Eng Applicat 2013;5(2):021007–21015.

2.Zhang M,Fu ZB,Lin YZ,Li JB.CFD study of NOxemissions in a model commercial aircraft engine combustor.Chin J Aeronaut 2012;25(6):854–63.

3.Yang CF,Zhang JZ.In fluence of multi-hole arrangement on cooling film development.Chin J Aeronaut 2012;25(2):182–8.

4.Shih TI,Lin YL.Controlling secondary- flow structure by leadingedge airfoil fillet and inlet swirl to reduce aerodynamic loss and surface heat transfer.J Turbomach 2003;125(1):48–56.

5.Rosic B,Denton JD,Horlock JH,Uchida S.Integrated combustor and vane concept in gas turbines.J Turbomach 2011;134(3):031005–31010.

6.Jacobi S,Rosic B.Development and aerothermal investigation of integrated combustor vane concept.J Turbomachin 2015;138(1):011002–11010.

7.Rahim A,Khanal B,He L,Romero E.Effect of nozzle guide vane lean under in fluence of inlet temperature traverse.J Turbomach 2013;136(7):071001–71012.

8.Khanal B,He L,Northall J,Adami P.Analysis of radial migration of hot-streak in swirling flow through high-pressure turbine stage.J Turbomach 2013;135(4):041005–41011.

9.Qureshi I,Beretta A,Chana K,Povey T.Effect of aggressive inlet swirl on heat transfer and aerodynamics in an unshrouded transonic HP turbine.J Turbomach 2012;134(6):061023-11.

10.Beard PF,Smith AD,Povey T.Effect of combustor swirl on transonic high pressure turbine efficiency.J Turbomach 2013;136(1):011002–12.

11.Yang H,Gao Z,Chen HC,Han J,Schobeiri MT.Prediction of film cooling and heat transfer on a rotating blade platform with stator-rotor purge and discrete film-hole flows in a 1–1/2 turbine stage.J Turbomach 2009;131(4):041003–12.

12.Dong P,Wang Q,Guo Z,Huang H,Feng G.Conjugate calculation of gas turbine vanes cooled with leading edge films.Chin J Aeronaut 2009;22(2):145–52.

13.Laskowski GM,Bunker RS,Bailey JC,Ledezma G,Kapetanovic GM,Itzel GM,et al.An investigation of turbine wheelspace cooling flow interactions with a transonic hot gas path—part II:CFD simulations.J Turbomach 2011;133(4):041020,041020-12.

14.Cha CM,Hong S,Ireland PT,Denman P,Savarianandam V.Experimental and numerical investigation of combustor-turbine interaction using an isothermal,nonreacting tracer.J Eng Gas Turb Power 2012;134(8):081501–18.

15.Koupper C,Caciolli G,Gicquel L,Duchaine F,Bonneau G,Tarchi L,et al.Development of an engine representative combustor simulator dedicated to hot streak generation.J Turbomach 2014;136(11):111007–10.

16.Ames FE,Barbot PA,Wang C.Effects of catalytic and dry low NOxcombustor turbulence on endwall heat transfer distributions.J Heat Transf 2005;127(4):414–24.

17.Hedlund CR,Ligrani PM,Moon HK,Glezer B.Heat transfer and flow phenomena in a swirl chamber simulating turbine blade internal cooling.J Turbomach 1999;121(4):804–13.

18.Papa M,Srinivasan V,Goldstein RJ.Film cooling effect of rotorstator purge flow on endwall heat/mass transfer.J Turbomach 2011;134(4):041014–8.

19.Barringer MD,Thole KA,Polanka MD.Effects of combustor exit pro files on vane aerodynamic loading and heat transfer in a high pressure turbine.J Turbomach 2009;131(2):021008–10.

20.Luque S,Povey T.A novel technique for assessing turbine cooling system performance.J Turbomach 2010;133(3):031013–31019.

21.Luque S,Kanjirakkad V,Aslanidou I,Lubbock R,Rosic B,Uchida S.A new experimental facility to investigate combustorturbine interactions in gas turbines with multiple can combustors.J Eng Gas Turb Power 2015;137(5):051503–9.

22.Krichbaum A,Werschnik H,Wilhelm M,Schiffer HP,Lehmann.A large scale turbine test rig for the investigation of high pressure turbine aerodynamics and heat transfer with variable in flow conditions.ASME turbo eExpo 2015:Turbine technical conference and exposition;2015 June 15–19;Montreal,Quebec,Canada.New York:ASME;2015.

23.Zhang LJ,Jaiswal RS.Turbine nozzle endwall film cooling study using pressure-sensitive paint.J Turbomach 2001;123(4):730–8.

24.Li SJ,Lee J,Han JC,Zhang LZ,Moon HK.Turbine platform cooling and blade suction surface phantom cooling from simulated swirl purge flow.J Turbomach 2016;138(8):081004–11.

25.Suryanarayanan A,Ozturk B,Schobeiri MT,Han JC.Filmcooling effectiveness on a rotating turbine platform using pressure sensitive paint technique.J Turbomach 2010;132(4):041001–13.

26.Suryanarayanan A,Mhetras SP,Schobeiri MT,Han JC.Filmcooling effectiveness on a rotating blade platform.J Turbomach 2008;131(1):011014-12.

27.Chung H,Hong CW,Kim SH,Cho HH,Moon HK.Heat transfer measurement near endwall region of first stage gas turbine nozzle having platform misalignment at combustor-turbine interface.Int Commun Heat Mass Transf 2016;78(11):101–11.

28.Zhang Y,Yuan X.Experimental investigation of nonperiodic endwall film cooling in neighboring passages with upstream rotating flow.ASME turbo expo 2014:Turbine technical conference and exposition;2014 June 16–20;Düsseldorf,Germany.New York:ASME;2014.

29.Zhang Y,Yuan X.Endwall film cooling using the staggered combustor-turbine gap leakage flow.ASME turbo expo 2014:Turbine technical conference and exposition;2014 June 16–20;Düsseldorf,Germany.New York:ASME;2014.

30.Zhang Y,Li Y,Yuan X.Effects of inlet swirl on suction side phantom cooling.ASME turbo expo 2016:Turbomachinery technical conference and exposition;2016 June 13–17;Seoul,South Korea.New York:ASME;2016.

31.Shi HH,Kiriyama K,Itoh M.Effect of diffusively shaped holes on the turbulent flow field of a film cooling plate.Chin J Aeronaut 2002;15(1):6–11.