Nn CAO,Xing LUO,b,Zeyu WU,Jie WEN,b,*

aNational Key Laboratory of Science and Technology on Aero-Engine Aero-Thermodynamics,School of Energy and Power Engineering,Beihang University,Beijing 100083,China

bCollaborative Innovation Center for Advanced Aero-Engine,Beijing 100083,China

KEYWORDS Height;Protrusion;Radial position;Rim seal;Sealing performance

Abstract This paper presents an experimental investigation on the effect of protrusion radial position and height on the sealing performance and flow structure in the rotor-stator cavity.The rotor mounted protrusions are assembled at three radial positions and are set to three heights.The cavity is equipped with three rim seals:a radial seal,an axial seal and a seal with double fins on the stator.The annulus Reynolds number is set at 4.39×105and the rotational Reynolds number ranges from 7.51×105to 1.20×106.Heat and mass transfer analogy is applied.Pressure and CO2concentration are measured.The experimental results show that in cavities with different rim seals,radial distributions of the sealing efficiency,pressure and swirl ratio are basically the same.The sealing performance is improved by protrusions compared with the cavity without protrusion and improves with the increase of protrusion radial position and height.The effect of protrusion increases with the increase of the rotational Reynolds number.The windage loss and the flow resistance introduced by protrusions are investigated.It is found that induced windage loss and flow resistance decrease with the increase of protrusion radial position but increase with the protrusion height.

1.Introduction

The ingestion of hot mainstream gas is observed in turbines.The ingested gas raises the temperature of the rotor-stator cavity and leads to the increase of thermal stress,which decreases its working life and reliability.To avoid overheating of turbine discs,the ingestion of mainstream gas must be prevented.This can be achieved by designing effective rim seals and supplying cooling air.Many studies have been carried out on the mechanism of ingestion,rim seal geometry,minimal sealing air flow rate and interactions between ingested gas and sealing air.

Owen1,2modified and utilized an orifice model to explore the ingress of hot gas through turbine rim seals.The ingress of hot gas can be divided into three types:Externally Induced(EI)ingress,Rotationally Induced(RI)ingress and combined ingress.Models of ingestion were further improved and verified experimentally.3–6Bayley and Owen7examined a rotor-stator system without mainstream and obtained a correlation for the minimal sealing air flow rate,which was later improved so that it was appropriate for various rim seal geometries.8The effect of the blades and vanes on ingestion was studied numerically and experimentally and it was found that vanes and blades made the distribution of mainstream gas and annulus circumferential pressure uneven,which intensified ingestion.9,10The tracer particle concentration method was proposed to investigate the sealing efficiency and minimal sealing air.11,12With the development of the experimental technique,the temperature and temperature based sealing efficiency were also measured.13,14The interaction between the sealing air and ingested annular flow was also investigated.It was found that gap sealing flow enhanced the fluctuations and flow loss of the mainstream and the ingested mainstream also affected the flow structure in the cavity.15,16

The studies mentioned above are focused on the rotorstator cavity without protrusion.However,in an actual engine turbine,many components are connected by protrusions for the purpose of structural reliability.Many researchers have investigated the additional windage loss caused by protrusions.The windage loss was subdivided into three sources:form drag,boundary layer losses and pumping losses.17Zimmermann et al.17investigated the effect of various bolt designs on shaft torque and proposed that the shape of protrusions had an effect on rotor moment and that covering the protrusions could reduce the friction moment effectively.The results were verified by Luo18and Daniels et al.19Zimmermann et al.17also found that protrusions affected each other,which complicated the windage loss.Long et al.20investigated the effect of diameter and number of bolts on the windage experimentally and found that increasing the number of bolts increased the moment coefficient and windage heating.Empirical correlations of windage loss of the cavity with rotor-mounted protrusions,including the number,radial position and volume of protrusions,have been concluded.17,20,21Many studies on how protrusions affect the sealing performance and the flow structure have been developed.Liu et al.22proposed a modified orifice model and introduced a new decisive factor H(H=p/ρ +V2θ/2,p is the static pressure,ρ is the fluid density and Vθis the fluid tangential velocity)for the ingestion in the cavity with protrusions.Liu et al.23presented an experimental investigation of the effect of protrusion parameters,including the number and radial position,on the efficiency of converting additional windage loss for the alleviation of ingress.Miles24used the Particle Image Velocimetry(PIV)measurement to investigate the flow inside the cavity with protrusions and found that protrusions increase the tangential velocity of the core flow.

It can be seen that many studies on protrusions have been carried out and most of them investigated the windage loss introduced by protrusions.But the effect of protrusion parameter on the sealing performance and flow structure is still unknown.In this paper,the effect of protrusion radial position and height on the sealing performance and flow structure in the rotor-stator cavity is investigated experimentally.The annulus Reynolds number is set at 4.39×105and the rotational Reynolds number ranges from 7.51×105to 1.20×106.Protrusions are mounted at three radial positions and are set to three heights.The cavity is equipped with three different rim seals.The sealing efficiency,minimal sealing air flow rate and pressure of all models are obtained.Additionally,the additional windage loss and flow resistance are investigated.By integrating the sealing performance and the windage loss,the proper protrusion parameter is proposed.

2.Experimental apparatus and measurement technique

Experiments were conducted at the National Key Laboratory of Science and Technology on Aero-Engine Aero-Thermodynamics at Beihang University.A schematic diagram of the experimental system is shown in Fig.1.The system mainly consists of a dynamic system,air supply system,test section and measurement system.

Fig.1 Schematic diagram of experimental system.

2.1.Dynamic system and air supply system

The rotor is driven by a 22 kW Direct Current(DC)electric motor(model Z4-132-2)and the annulus centrifugal compressor(model KF3-95 No6.3E)is driven by a 55 kW DC motor(model Z4-160-32).Both of the motors allow the rotational speed to vary up to 3000 r/min.A conveyor belt with a transmission ratio of around 2.5 is utilized to connect the belt pulley on the rotating axis and the 22 kW DC motor axis.The motor speed is controlled by two DC power distribution cabinets with manual adjustment.A photoelectric sensor(model P+F OBT200-18GM60-E5)connected with a tachometer(model YK-23)is applied to measure the rotation speed of the rotor.The uncertainty of the rotation speed is±7 r/min.As shown in Fig.1,the experimental investigation of ingestion comprises two kinds of flows:annulus flow and sealing air.The annulus flow simulating mainstream is drawn from the atmosphere by a centrifugal compressor.The maximum flow rate reaches up to 13525 m3/h and the corresponding static pressure is 8130 Pa.The sealing air is supplied by a 75 kW compressor and flows through a stabilizing box and mixes uniformly with the elevated level of CO2.After shunting and cooling,the sealing air is divided into six channels in a splitter.It is axially superimposed into the cavity at low radius and then flows through the rim seals into the annulus.

2.2.Test section and rim seal configurations

A cross-section of the test section of cavity with an axial rim seal is shown in Fig.2.There are 32 guide vanes installed around the stator disk and 59 rotor blades around the rotor disk.In the paper,b is the radius of the cavity(inner radius of the test section),a is the outer radius of the test section,s is the rotor-stator gap value,h is the height of protrusion,r is the radial mounting position of protrusion,D is the diameter of protrusion,r1is the radius of sealing air inlet and n is the number of protrusion.In cavities with three different rim seals,b=267 mm, a=287 mm, s=18 mm, D=10 mm,r1=20 mm and n=12.Protrusions are hexagonal and made of aluminum alloy.Protrusions are mounted uniformly around the circumference of the rotor.The orientation of protrusions relative to the rotation is shown below.

Fig.2 Cross-section of test section.

Fig.3 Rim seal configurations and structure of models.

Three rim seal configurations that are adopted in actual engines were examined,as shown in Fig.3.Rim Seal A is an axial seal with a fin on the stator.Rim Seal B is a radial seal that has an inner fin on the rotor and an outer fin on the stator.Rim Seal C has two fins on the stator and a fin between them on the rotor.Gap value between the rotor and the fin,Sc,is 2 mm,the fin thickness,St,is 3 mm and the length of seal fins,l,is 16 mm.For models with Seals B and C,the gap value between two fins,Sf,is 2 mm and the overlap between two fins,So,is 2 mm.

Fig.4 Distribution of measurement taps.

The radius of the cavity is267 mm in all models(b=267 mm).Radial mounting positions of protrusions are set at 0.846b,0.884b and 0.922b.Heights of protrusions are set to 6,10 and 14 mm and relative heights are 0.022b,0.037b and 0.052b.Models without protrusion are tested as reference.For convenience,the models are named using the notation ‘Model letter number”,where letter is the type of rim seal and the number represents the radial mounting position and height of protrusions.Radial mounting positions 0.846b,0.884b and 0.922b are named Models 2,3 and 4.Heights 0.022b,0.037b and 0.052b are named Models 5,6 and 7.For example,the model with Seal A and protrusion height set to 0.022b can be named Model A5.The structure of models with Seal A is shown in Fig.3.Protrusion height is set to 0.037b in models(Models 2,3 and 4)when the effect of protrusion radial mounting position is investigated.Protrusion radial mounting position is set at 0.884b in models(Models 5,6 and 7)when the effect of protrusion height is investigated.Therefore,Model 3 and Model 6 are the same.

2.3.Measurement system

2.3.1.Pressure measurement

The distribution of measurement taps of the cavity with Seal C is shown in Fig.4.The stator is instrumented with 30 taps.The taps are located in three rows to measure the static pressure,total pressure and CO2concentration,respectively.The tangential angle between concentration taps and total taps is 10°(θ =10°).The radial position of the taps,r/b,ranges from 0.84 to 0.98.Total pressure measurement taps are set towards the tangential direction and the combination of the static pressure and the tangential dynamic pressure （p/ρ +V2θ/2）is obtained.The study showed that the tangential velocity is increased by protrusions and is considerably higher than the radial velocity.24Therefore,p/ρ +V2θ/2 is considered roughly equal to the total pressure.The differential pressure transducer(Rosemount 3051S)with the calibration range of 0–6.216 kPa is applied to measure the pressure.The uncertainty of the pressure measurement is less than±2%.

Fig.5 Radial efficiency of models with different protrusion radial positions(ReW=4.39×105，ReΦ =1.05 × 106).

2.3.2.Concentration measurement and mass flow rate measurement

The heat and mass transfer analogy is applied in the experiment.The gas containing CO2with a volumetric concentration of 3%is selected as a tracer particle and is combined with the sealing air to calculate the sealing efficiency.The radial sealing efficiency εris calculated by the following fraction:

Fig.6 Variation of εcwith Cwof models with different protrusion radial positions(ReW=4.39×105).

where c denotes the CO2concentration at the measurement positions,c∞denotes the concentration of annulus flow and c0denotes the concentration of sealing air combined with CO2at the inlet of the cavity.The radial sealing efficiency εrmeasured in the second measurement tap(r/b=0.96)represents the sealing efficiency of the cavity εc.When the sealing efficiency of the cavity εcreaches 0.98,it is considered that annulus flow ingestion does not occur inside the cavity and the corresponding sealing air flow rate coefficient Cwis the minimal sealing air flow rate coefficient Cw，min.The volumetric concentration is tested by an infrared analyzer(model GXH-3010E).The measurement scope is 0–5%and the linear error is less than±2%.The total pressure of the annulus flow is measured by a pitot tube set towards the incoming annulus flow and the static pressure is measured by the static tube.Both of the tubes utilize the differential pressure transducer(Rosemount 3051S)with a calibration range 0–722.5 Pa.The velocity in the annulus is obtained through measurement of the pressure difference and the mass flow rate of the annulus flow can be calculated.The sealing air is adjusted by valves and its mass flow rate is measured by thermal flow meter.

3.Results and discussion

In Section 3.1,the experimental results of the pressure and sealing efficiency are shown and the effect of the protrusion radial position and height is analyzed.The windage loss and the flow resistance of models are discussed in Section 3.2.In Section 3,ReΦis the rotational Reynolds number(ReΦ= ρΩb2/μ,Ω is rotating velocity of disk,μ is dynamic viscosity),ReWis the annulus Reynolds number(ReW= ρWb/μ,W is axial velocity in external annulus),Cwis the flow rate coefficient of the sealing air(Cw=˙mc/μb,˙mcis mass flow rate of the sealing air),p*is the total pressure,p0is the atmospheric pressure.

Fig.7 Radial distribution of static pressure of models with different protrusion radial positions(ReW=4.39×105).

3.1.Effect of protrusion on sealing performance and flow structure

3.1.1.Effect of protrusion radial position on sealing performance

The radial sealing efficiency εrof models with different protrusion radial mounting positions is shown in Fig.5.It can be seen that in models with the same rim seal,the radial sealing efficiency of models with protrusions increases with the radial position of protrusions.In models with Seals A and B,the radial sealing efficiency of models with protrusions is higher than that of the model without protrusion.In models with Seal C,the radial sealing efficiency of Model C2 is lower than that of Model C1.The variation of the radial sealing efficiency of models with Seals A and B is basically consistent.The radial sealing efficiency decreases gradually with the increase of radius.The radial sealing efficiency of models with Seal C decreases slightly inside the cavity and decreases sharply between the fins.The cavity is divided into the inner cavity and the outer cavity by fins.Ingestion mainly occurs in the outer cavity.

Fig.8 Radial distribution of total pressure of models with different protrusion radial positions(ReW=4.39×105).

The sealing efficiency of the cavity εcis shown in Fig.6.The sealing efficiency of the cavity increases with the increase of the flow rate of the sealing air Cwof all models.Models with protrusions(except for Model C2)achieve higher sealing efficiency compared with models without protrusion.

3.1.2.Effect of protrusion radial position on pressure

The radial distribution of the static pressure of models with different protrusion radial positions is shown in Fig.7.In models with Seals A and B,the static pressure decreases with the increase of protrusion radial position.The static pressure of models with protrusions is lower than that of models without protrusion. ‘Pressure inversion” can be observed in high radius cases.This means that the static pressure decreases and then increases.What is different in models with Seal C is that‘pressure inversion” is not observed and the static pressure of Model C2 is the lowest.

The radial distribution of the total pressure is presented in Fig.8.It can be seen that the total pressure of models with protrusions is higher than that of models without protrusion.It is noted that in the same radial position,the total pressure does not increase with the increase of the protrusion radial position.In low radius cases,the total pressure of Model A2 is the highest.In middle radius cases,Model A3 has the highest total pressure and in high radius cases,Model A4 has the highest.In models with Seal C,the total pressure of Model C2 is lower than that of Model C1.The decrease of the total pressure at the highest measurement tap(r/b=0.98)is obvious,and this tap is located between fins and is affected by the mainstream.

Fig.9 Radial sealing efficiency of models with different protrusion heights(ReW=4.39×105，ReΦ =1.05 ×106).

3.1.3.Effect of protrusion height on sealing performance

The radial sealing efficiency εrand the sealing efficiency of the cavity εcof models with different protrusion heights are shown in Figs.9 and 10 respectively.It can be seen that the sealing efficiencies(εrand εc)of models with protrusions increase with the increase of the protrusion height and they are higher than those of models without protrusion.

3.1.4.Effect of protrusion height on pressure

Radial distributions of the static pressure and the total pressure are presented in Figs.11 and 12 respectively.The radial distribution of the pressure of models with different protrusion heights is similar to that of models with different protrusion radial positions.It is noted that with the increase of the protrusion height,the static pressure decreases while the total pressure increases.‘Pressure inversion” can also be observed.

Fig.10 Variation of εcwith Cwof models with different protrusion heights(ReW=4.39×105).

Fig.11 Radial distribution of static pressure of models with different protrusion heights(ReW=4.39×105).

3.1.5.Effect of protrusion on flow structure

The flow structure in the cavity with protrusions is analyzed in this section.The radial distribution of swirl ratios β(β =Vθ/bΩ,b is disk radius,Ω is rotating velocity of disk)is shown in Fig.13.It can be seen that the swirl ratio of models with protrusions is higher than that of models without protrusion.The tangential velocity of the fluid increases with the introduction of protrusions.The radial distributions of the swirl ratio and the total pressure are consistent.By comparing models with different protrusion radial positions(Models 2,3 and 4),it can be seen that in the same radial position,the swirl ratio of models with protrusions mounted in this radial position is the highest.In models with different protrusion heights(Models 5,6 and 7),the swirl ratio increases with the increase of protrusion height.Protrusions work on the fluid and increase the tangential velocity of the fluid near the mounting position.The overall tangential velocity increases with the height of protrusion.Besides,previous studies also verified that protrusions increase tangential velocity of the fluid considerably.24Therefore,in models with protrusions,the total pressure is higher than that of models without protrusion even if the static pressure decreases.The increase of the total pressure due to the protrusions depends on whether work converted into fluid outweighs the flow loss.The total pressure of Model C2 is lower than that of Model C1 because the static pressure of Model C2 is too low.

Fig.12 Radial distribution of total pressure of models with different protrusion heights(ReW=4.39×105).

As shown in Fig.14(a),protrusions impact the fluid,which forms a high pressure zone and a low pressure zone.The fluid rolls up,generating horseshoe vortexes and legs of vortexes interact with each other,which intensifies flow loss and decreases the static pressure.As a result,the static pressure of models with protrusions is lower than that of models without protrusion.When protrusions are mounted in higher radius cases,horseshoe vortexes interact with gap vortexes,which enhances the flow loss.Therefore,the static pressure decreases as the radial mounting position increases.In addition,the flow loss increases with the increase of the protrusion height,which results in the decrease of the static pressure.According to the experimental results,it can be concluded that in the cavity with protrusions and the smooth cavity,the radial distributions of flow parameters are similar.It can be inferred that for the gap ratio and operating conditions of the current experiment,the flow structure inside the cavity with protrusions is ‘Batchelor flow”25,26with a complex vortex system.There is a rotating core inside the cavity.The swirl velocity of the rotating core causes the uneven distribution of the axial velocity and the total pressure and results in the ingress and egress.The sealing air enters the cavity and mixes with the ingested mainstream in high radius cases and then the mixed air flows along the stator and blends continuously with the sealing air.The fluid flows from rotor to stator in high radius cases because of the axial pressure and this causes ‘pressure inversion”.In models with Seal C,the rotating core is compressed to the low radius position by fins and ‘pressure inversion”is not measured.The flow structure inside the cavity is shown in Fig.14(b).

Fig.13 Radial distribution of swirl ratio(ReW=4.39×105).

3.2.Further analysis of experimental results

3.2.1.Mechanism of ingress and egress in cavity with protrusions

Fig.14 Flow structure inside cavity with rotor-mounted protrusions.

Liu et al.proposed a decisive factor H of ingress and egress in the cavity with protrusions.22In the current experiment,the combination of the static pressure and tangential dynamic pressure is measured and is considered equal to the total pressure.The experimental results of the current study show that the sealing efficiency and total pressure of models with protrusions are higher than those of models without protrusion,which verifies Liu’s study22experimentally.The average total pressure of models is shown in Fig.15.It reveals an interesting phenomenon that in models with different protrusion heights,the average total pressure increases with the increase of the protrusion height,but in models with different protrusion radial positions,the average total pressure of models with protrusions(except for Model C2)is almost the same.The sealing efficiency increases with the increase of protrusion radial position.It is because protrusions mounted in high radial position work on fluid and increase the tangential velocity in the mounting area,which forms a high pressure area in high radius cases.The ingestion occurs in high radius cases,so it can be effectively relieved with protrusions mounted in high radius cases.Additionally,the increase of the protrusion height can increase the average total pressure,which can also relieve the gas ingestion.

The total pressure of the fluid near the rotor increases because of the work from protrusions,which makes the fluid flow out of the cavity.The total pressure of the fluid near the stator decreases because of the friction effect and the annulus flow enters the cavity.This is the rotationally induced ingress.Blades and vanes make the circular total pressure of mainstream uneven.The annulus flow ingestion occurs when the partial total pressure is higher than that of the cavity.This is the externally induced ingress.Therefore,in the cavity with protrusions,the sealing efficiency can be increased by injecting the sealing air(to increase the static pressure),mounting protrusions(to increase the dynamic pressure)and designing a complicated rim seal(to increase the flow resistance).

3.2.2.Minimal sealing air flow rate

The minimal flow rate of the sealing air Cw，minis calculated by using the linear interpolation method and variations of Cw，minwith ReΦis shown in Fig.16.It can be seen that Cw，minof models with protrusions(except for Model C2)is lower than that of models without protrusion and decreases as protrusion radial position and height rise.As ReΦincreases,Cw，minof models without protrusion increases while Cw，minof models with protrusions(except for Model C2)decreases.This is because as ReΦincreases,the sealing performance improves because of protrusions and the rotation-induced ingression intensifies.By comparing the experimental results,it can be concluded that the effect of protrusion increases.The effect of protrusions on the sealing performance is the strongest in models with Seal A and the weakest in models with Seal C.

3.2.3.Windage loss and flow resistance

Many experiments have been carried out to investigate the effect of protrusion parameters,including the number,volume and radial mounting position,on the windage loss.Empirical correlations of the moment coefficient of the cavity with protrusions are proposed by Long et al.20and Coren,21as shown in Eqs.(2)and(3):

where Cmis the moment coefficient,r is the radial position of protrusions,P is the circumferential pitch between protrusions and C is the empirical coefficient.

The windage loss(Lu)can be expressed as follows:

where M is the rotor moment.Both of the correlations indicate that with a certain number and volume of protrusions,an increase in the radial position of protrusions results in less windage loss.This may be because,with a fixed number of protrusions,the circumferential pitch between protrusions increases as the radial mounting position increases,which causes interactions between protrusions to decrease.According to the experimental results,the sealing efficiency increases with the increase of the radial position of protrusions.Therefore,protrusions should be mounted in the highest radial position,which can achieve the best sealing performance and induces the least additional windage loss.Additionally,Zimmermann et al found that the increase of protrusion height could result in more form drag and windage loss.17Although the sealing performance improves with the increase of the protrusion height,the proper height should be set to obtain the high sealing efficiency and the low windage loss at the same time.

Based on the experimental results,it can be seen that in the same working condition,models with Seal C achieve the best sealing performance while those with Seal A show the worst performance.The more complex the rim seal is,the better sealing performance it achieves.However,when the rim seal is too complex and the arrangement of protrusions is improper,the flow resistance of radial out flow increases.How protrusions affect the flow resistance is still unknown.The flow resistance is evaluated by the overall pressure loss.The flow resistance coefficient is defined as follows:

Fig.15 Variation of average total pressure with ReΦ(Cw=4640,ReW=4.39×105).

Fig.16 Variation of Cw，minwith ReΦ (ReW=4.39 × 105).

Fig.17 Variation of ξratiowith ReΦ (Cw=12000).

It can be seen that in the same working condition,ξ of models with Seal B is far more than that of models with Seal A because of more complex rim seal configuration.In models with the same rim seal,ξ of models with protrusions is higher than that of models without protrusion,and ξ decreases with the increase of the radial mounting position but increases with the protrusion height.It is because that with the increase of protrusion radial position,the circumferential pitch between protrusions and the centrifugal effect increase,which decreases the flow resistance of protrusions on radial out flow.Additionally,the radial flow resistance increases with the increase of protrusion height.

4.Conclusions

The experimental investigation on the effect of protrusion radial position and height is carried out,which provides new insight into the analysis of the rotor-stator cavity with protrusions.The conclusions are shown below:

(1)For models with three different rim seals,radial distributions of flow parameters including the sealing efficiency,pressure and swirl ratio are consistent.Compared with models without protrusion,the static pressure decreases while the sealing efficiency,total pressure and swirl ratio increase for models with protrusions.The sealing efficiency increases with the increase of the protrusion radial mounting position and height.With the increase of the rotational Reynolds number,the minimal sealing air flow rate of models with protrusions decreases,but that of models without protrusion increases.

(2)The total pressure is the decisive factor of ingress and egress in the cavity with protrusions.The average total pressure increases with the increase of protrusion height while it does not change with radial mounting position.In the same radial position,the local total pressure and swirl ratio of models with protrusions mounted in this radial position is the highest.The effect of protrusion increases with the increase of the rotational Reynolds number.

(3)In the case with a certain number of protrusions,the

windage loss and flow resistance decrease with the increase of the radial mounting position but increase with the protrusion height.The proper arrangement of protrusion parameters(radial mounting position and height)is proposed to obtain the high sealing performance and low windage loss.