Experimental study of a controlled variable double-baffle distortion generator engine test rig

Aiguo XIA,Xudong HUANG,*,Wei TUO,Ming ZHOU

aSchool of Aerospace Engineering,Tsinghua University,Beijing 100084,China

bBeijing Aeronautical Technology Research Center,Beijing 100076,China

KEYWORDS Distortion generator;Engine stability;Inlet flow;Inlet total-pressure distortion;Turbofan engines

Abstract In order to explore the total-pressure distortion test assessment method for a turbofan engine,a Controlled Variable Double-Baffle Distortion Generator(CVDBDG)with a horizontal symmetry moving form was developed,which can adjust the steady-state and time–variant distortion separately in real time.The inlet total-pressure distortion test was conducted on an afterburner turbofan engine.The distortion parameters of CVDBDG and the instability characteristics of the engine were measured.The experimental data were modeled and analyzed by using back propagation artificial neural networks,and the work envelope of CVDBDG was obtained.Based on the analysis of the data on the engine’s instability,the properties of CVDBDG used for the stability assessment were preliminarily evaluated.The results show that CVDBDG can simulate both steady-state and time–variant distortions simultaneously in a range determined by three envelopes.Under the condition of symmetric double baffles,a critical depth of insertion exists,beyond which the symmetric baffles will generate an asymmetric flow field.In the case of double baffles,compared to a single baffle,the engine exhibited different instability characteristics.Based on CVDBDG,it is expected that more efficient engine stability and durability assessment methods can be developed.Ⓒ2018 Chinese Society of Aeronautics and Astronautics.Production and hosting by Elsevier Ltd.This is an open access article under the CCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1.Introduction

Inlet total-pressure distortion assessment is an important part of an aero-engine test.Inlet total-pressure distortion can cause stalling or surges of the fan/compressor,induces synchronous vibration response of the blades of the fan and compressor blades,and change the temperature distribution of the combustor and turbine,which can lead to higher thermal stress on the turbine blades.1–3Since the end of the 1960s,after the F-111 aircraft engine surge failure,the problem of inlet distortion has been the subject of continuous research and attention in various countries,and extensive research has been conducted on a large number of ground-based and aerial testing technologies.4–7In recent years,more and more mobility and stealth performance have been designed into advanced fighters.In addition,large mobile flight and the use of S-bend and other shapes of inlet stealth designs have made the problem of air flow distortion more prominent.8–10

Due to the adverse effects of inlet total-pressure distortion on engine performance,structural safety,and reliability,the assessment of inlet total-pressure distortion has become one of the most important aspects of aero-engine assessment.11Compared with the experiments of flight tests,propulsion wind tunnel tests,and free jet altitude tests,the use of the inlet total pressure distortion simulator in direct-connect altitude test facilities or ground test beds has better safety and economic performance.12–17

Inlet total-pressure distortion assessment includes steadystate distortion and the time–variant distortion test.18The steady-state distortion simulators that have been developed mainly include a distortion screen19and air jet distortion generators.20,21Time–variant distortion simulators mainly include a controllable plug and ramp random frequency generators,discrete frequency generators,22and a splitting airfoil transient total-pressure distortion generator.23The distortion simulators,which can produce steady-state distortion and time–variant distortion simultaneously,mainly include a simulation board and a baffle distortion generator.24Currently,the distortion screen,simulation board,and baffle distortion generator are the most extensively used types in the research and engineering application of inlet total-pressure distortion.The distortion screen is simple to use,which only requires limited steady-state measurement,and can be designed for a selected distortion map.The Society of Automotive Engineers(SAE)has published a series of standards on inlet total-pressure distortion,18,25–27the main content of which is based on the distortion screen.The main disadvantage of the distortion screen is that it can only simulate steady-state distortion and cannot do real-time simulation.Thus,the simulation of several distortion maps is time-consuming and laborious.Both the simulation board and the baffle distortion generator can simulate both steady-state and time–variant distortion of total pressure,but the simulation board is a fixed shape that only applies to specific assessment situations.The movable baffle distortion generator utilizes a movable mechanism to enable real-time state conversion,and the simulated distortion state has a good similarity to the actual distortion characteristics.It is considered to be the most advanced inlet total-pressure distortion simulator developed in the last 100 years.28The movable baffle distortion generator was first used by Russia;however,there are very few public literature references on the distortion characteristics of the baffle distortion generator and the instability of the engine’s performance with a baffle distortion generator.In China,there also have been some studies on baffle distortion generators,including tests of engine parts,29,30testing in a ground-level test facility,1,31,32testing in direct-connect altitude test facilities,33flight tests,34and other related research work.Most of these experimental studies were based on one movable baffle,in a baffle position and engine operating condition,which can only produce a specific corresponding steady-state distortion and time–variant distortion,and it can only produce one low-pressure zone,which limits the baffle distortion generator for simulating actual complex and variable flight conditions. In practical applications, the phenomenon often occurs that the surge caused by the generator is inconsistent with the actual features.This is a problem that cannot be solved by a baffle distortion generator,which only assist in additional assessment of the special simulation board.In this case,adding another movable baffle is a natural idea.However,there are several questions that must be considered,e.g.,what the distortion characteristics of the two baffles would be,whether it can be used for stability evaluation,and what the instability characteristics of the engine would be.In the public literature,there are few studies on multi-baffle distortion generators.In Ref.35,four kinds of baffle structure forms,including a single flat plate,two symmetrical flat plates,and a fan plate,and two flat plates with a certain angle were tested on a two-stage,low-speed,axial- flow compressor test rig.It was found that the complex total-pressure distortion descriptor of single flat plate baffle was the largest by the same blockage coefficient,and using it as the standard baffle of distortion assessment is conservative and strict.The literature drew this conclusion without considering the performance of compressor stability for different baffles,so whether this conclusion conforms to reality must be analyzed further and studied in an engine test.Ref.24gave a kind of work envelope of the double-baffle distortion generator when the flow function was 0.5,but it did not explain the meaning of the boundary line mean or how the line influenced the engine’s stability.

Based on the above introduction,a baffle distortion generator,compared with other inlet total-pressure distortion simulators,can simultaneously simulate steady-state and time–variant total-pressure distortion in real time.It is easy to use,and the simulated distortion is very close to the actual distortion characteristics.Thus,the basic data and theoretical guidance required for engineering applications can be acquired by conducting the development and testing of a two-baffle distortion generator to obtain the distortion characteristics and its effect on the stability of the engine.It can also provide a large number of real and credible distortion test cases available for the CFD simulation of flow field distortion.Based on the CVDBDG and the engine test system developed in this work,a turbofan engine was used as a test object in the ground test bed to test the distortion characteristics of the double baffles at different combinations,engine operating responses,and surge characteristics.In this paper,we mainly analyze the distorted flow field characteristics and engine stability with double baffles.

2.Test equipment and engine

2.1.Engine test bed

The test was conducted on the indoor engine test bed at Beijing Aeronautical Technology Research Center.The test bed was a U-shaped structure,with an intake cross-section of 12 m×12 m.The maximum thrust test capacity was 250 kN,and the engine was suspended from the roof in the center of the intake air flow.

2.2.Test engine

The test engine was a two-spool after burning turbofan engine.The fan had an adjustable inlet guide vane,which is controlled according to the corrected rotational speed of the low-pressure spool.The inlet and first two stages of the high-pressure compressor had adjustable guide vanes,and the regulation law was determined according to the corrected rotational speed of the high-pressure spool.The nozzle was an adjustable Laval nozzle,and the regulation law of the nozzle of the throttling state was determined according to the corrected rotational speed of the high-pressure spool.The main parameters that were measured to determine the engine’s performance included the speed of the high pressure spool,the speed of the low pressure spool,thrust,fuel consumption,temperature and total pressure after the low pressure turbine,vibration of the inlet casing,vibration of the intermediate casing,total pressure after the fan,and total pressure and static pressure after the high pressure compressor.

2.3.Controlled variable double-baffle distortion generator and flow field measurement

Fig.1 shows the layout of the distortion test system and the engine.The distortion test system was in front of the engine,and it included the bell mouth,air meter,CVDBDG,Aerodynamic Interface Plane(AIP)measurement cross-section and air flow duct.

The designed CVDBDG was installed at 3D in front of the engine,where D is the diameter of engine’s inlet,which was 905 mm.The distortion generator had two horizontal baffles.Both baffles were rectangular plates that were driven by two servo cylinders moving horizontally in a sealed box.Both Baffle 1 and Baffle 2 could travel to the relative insertion depth,with the ratio of length L1or L2to the inlet diameter D being 0.5.Fig.2 shows the location layout of the two baffles.

Compared with a traditional electric motor and an ordinary hydraulic cylinder drive,the servo cylinder drive improved the control of the speed of the baffles and the accuracy of their positioning.The maximum adjusting speed of the servo cylinder was 100 mm/s,and the control error can reach 0.05%for different control speeds.The distortion generator was supported by a fixed platform that was fixed on the ground.There was a clearance of 2 mm between the rear of the distortion generator and the inlet section in the front of the engine,and this clearance was filled and sealed with flexible materials to isolate the distortion generator and the engine.This ensured that the air flow channel was smooth and the engine’s net thrust could be obtained during the distorted air flow.

The inlet section between the engine and the distortion generator was provided with an engine inlet measuring section,which we referred to as the AIP,and was marked as Crosssection 2.The distance between the measuring section and the engine was 270 mm.There were a total of six pressure test probes and six wall static pressure measurement points uniformly distributed along the circumferential direction.The pressure test probes and the measurement points for the wall static pressure were interlaced,and the space was 12.5°.Each total-pressure test probe included six steady-state total pressure measuring points that were located in the centroids of equal areas,and one pulsating total-pressure measuring point,arranged at 0.9 times the radius of the flow channel.The average of the 36-point steady-state total-pressure measurements on the six probes was recorded as Pt2fav.The measured average of the 6-point wall static pressures was recorded as Ps2fav.

Fig.2 Location layout of two baffles(viewed from engine).

The air meter(named as Cross-section 1)and bell mouth sections were in the front of the distortion generator.The distance between the air meter measuring section and the distortion generator was 2D.Four total-pressure probes and four wall static pressure measurement points were distributed uniformly along the circumferential direction of the air meter with spaces of 45°.Each total-pressure probe included six steadystate total-pressure measuring points that were located in the centroids of equal areas.In total,there were 24 total pressure measurement points located in Cross-section 1,with the average value recorded as Pt1fav.The average of the 4-point wall static pressure was recorded as Ps1fav.The engine inlet air flow rate was calculated by the total pressure and static pressure of Cross-section 1.The bell mouth section had a standard,double twist line inlet,and the distance from the tangent point of the bell mouth to Cross-section 1 was 0.3D.The bell section,the air flow meter,and the distortion generator were connected rigidly and supported on the mounting platform.Four atmospheric temperature measurement points were distributed evenly outside the inlet port,giving the test parameters marked as Tt0and the average temperature marked as

Among the tested parameters,the six-point pulsating totalpressure parameter of the AIP section and the vibration parameter of the inlet casing were collected and processed by a Gen3i high-speed acquisition system,and the sampling frequency was 25000 Hz.The other parameters were collected and processed by a PXI(PCI eXtensions for Instrumentation)acquisition system configured for the test bed,and it had a sampling frequency of 50 Hz.

Fig.1 Diagram of engine and location of distortion test system.

3.Definition of parameters and calculation method

3.1.Distortion parameter

There were two main characteristic parameters of the double baffles,i.e.,the relative insertion depth H1/H2of the baffle,defined as the ratio of length L1/L2to the inlet diameter D,and the relative insertion area S of the baffle,defined as the ratio of the insertion occlusion area to the inlet cross-section area.To facilitate understanding,the following S was expressed as a percentage of the relative insertion area and H1/H2was expressed as the percentage of the relative insertion depth.

The main inlet total-pressure distortion evaluation indexes mainly included the AIP cross-section area-averaged totalpressure recovery coefficient σav,the steady-state circumferential distortion descriptorthe maximum ring steady-state radial distortion descriptorthe area-averaged turbulence εav,and the complex total-pressure distortion descriptor W.

The area-averaged total-pressure recovery coefficient of AIP was calculated by

The steady-state circumferential distortion descriptorwas calculated by

where σ0is the averaged total-pressure recovery coefficient of AIP low-pressure sector.

Area-averaged turbulence εav,was calculated by

where Ndis the measurement points of the pulsation total pressure,and ε(i)is the turbulence of the ith pulsation totalpressure measurement point,calculated according to the following equation:

where(ΔPt2)RMSis the root mean square of the pulsation total pressure,and Pt2is the time-averaged value of pulsation total pressure,with the averaging period of two seconds.

The complex total-pressure distortion descriptor,W,which represents the total characteristic of the total-pressure distortion,is composed of a steady-state circumferential distortion descriptor and the area-averaged turbulence.

The maximum ring steady-state radial distortion descriptoris the maximum steady-state radial distortion descriptor of the six total-pressure test rings of the AIP section.The ring steady-state radial distortion descriptorcan be calculated by

where(Pt2av)iis the average value of six steady-state total pressure in the ith ring.

3.2.Engine parameters

The inlet air flow rate Wais calculated by the total pressure,the static pressure of Cross-section 1,and the ambient air temperature,based on the following equation:

where KWis the boundary layer correction factor for the airflow meter,and K is an aerodynamic constant(for the conditions used in this paper,K is a constant with the value of 0.0404.);A is the area of Cross-section 1;λ1is the velocity coefficient of the air flow meter;q(λ1)is the flow function of the airflow meter.

Corrected air flow rate is calculated by

The air flow rate corrected to Cross-section 1 is identified as Wac1,and when it is corrected to AIP,it is identified as Wac2.In order to facilitate a relative comparison,the air flow rate parameters are represented by the ratio of the corrected air flow rate to the design air flow rate,and the corresponding parameters were represented with a line over the original name,e.g.,Wac2and

The total-pressure ratio of the fan was calculated by

where Pt21is the total pressure downstream from the fan.

In this paper,stability index,KS,is used to reflect the situation of stability margin.It is defined as the ratio of fan totalpressure ratio on the stable boundary of the engine and the relative corrected air flow rate of the AIP section of the engine:

4.Test method

The main purpose of the experiment was to analyze the flow field characteristics and the engine’s response to the two baffles at different locations,so this paper takes two relative insertion depths of the baffle as the names of the test cases,i.e.,Case H1/H2.In order to facilitate the expression,the percentage symbol of H1/H2are removed.The comparison case of the benchmark is Case 0/0.In the experiment,we tested the quasisteady-state response of the engine from minimum to maximum flow conditions under different combinations of baffles.The performance of the 90%corrected high-pressure spool speed engine was tested by changing the positions of the baffles until the surge of the engine was tested without changing the engine’s throttle.

5.Test results and analysis

5.1.Matching work of engine and air inlet with CVDBDG

The engine and the air inlet with CVDBDG are analyzed as a system.The rotating engine sucks air in from the air inlet with CVDBDG;at the same time,the change in the position of the adjustable baffle will also cause a change in the inlet flow field of the engine,resulting in a change of the state of the engine.In this paper,we select three parameters to jointly analyze the combining work of the engine and the distortion generator,i.e.,the corrected low-pressure spool rotational speed Nlc,the corrected air flow rateof Cross-section 1 at the entrance to the air inlet,and the area-averaged total-pressure recovery coefficient of AIP σav.Fig.3 shows the variation of Wac1and σavfor 18 different combinations of baffle positions in the test in the case thatchanges from 50%to 80%(analysis interval of 5%).The baffle position,H1/H2,and the corresponding percentage of relative insertion area,S,in the 18 test cases are included in the graphical representation.In order to facilitate comparison,the same color and logo are used to reflect the state of the single-baffle and the double-baffle with similar relative insertion areas.

Fig.3 shows that both Wac1and σavtend to decrease gradually with the insertion of the baffle while the state of the engine is kept unchanged(i.e.,Nlcremains unchanged).As the position of the baffle remains unchanged,Wac1gradually increases,while σavgradually decreases as the condition of the engine improves.In the case of a low air flow rate and/or a low insertion area,the rate of the reduction of σavis close to a gentle linear decrease.Later,as the flow/insertion depth increases,there is a transition zone of the rate at which σavdeclines,and after going beyond this zone,σavdeclines faster.

5.2.Results and analysis of distorted flow field

Fig.4 shows the total-pressure recovery coefficient(σ)maps of the AIP section at 18 different baffle positions with an engine corrected high-pressure spool rotational speed Nhcof 88%.The relative insertion depth of the baffle for the corresponding location of the map is listed in the left and right positions of each map.According to the AIP section test,the 36-point total-pressure values are interpolated by the inverse-distance interpolation method.The maximum radius of these maps corresponds to the total-pressure measurement point in the maximum radial position.

In this paper,the distorted flow field affected by the baffles is divided into three regions for analysis.The high-pressure region,where the total-pressure recovery coefficient is higher than the average total-pressure recovery coefficient,is either less affected or not affected by the distortion.The low pressure zone is where total-pressure recovery coefficient is lower than the average total-pressure recovery coefficient.In addition,the transition area is the junction area between the high-and low-pressure areas.Fig.5 shows the difference between the total-pressure recovery coefficient and the average total-pressure recovery coefficient of the AIP section(Δσ).All of the areas below zero are low-pressure areas.The range and strength of the low-pressure area can be seen more intuitively on the difference map.

The map shows that,as the depth of a single-baffle increases,the total-pressure recovery coefficient of the affected area is decreasing.The low-pressure region extends gradually from the edge to the center of the circle,the range of the low-pressure region has its largest value near the edge,and the range decreases gradually from the edge to the center.The hump-type low-pressure region forms gradually on the side of the baffle.The variable trend of the high-pressure region is opposite that of the low-pressure area.With the insertion of the baffle,the range that is not influenced by distortion in the high-pressure region is reduced gradually.The totalpressure recovery coefficient in the high-pressure region also decreases gradually,and the saddle-shaped high-pressure concentration zone is formed gradually on the non-baffle side.There is an obvious transition area between the high-and lowpressure regions.With the gradual insertion of the baffle,the transition zone is gradually stabilized near the longitudinal center line of the section,forming a strip that has a totalpressure recovery coefficient that is basically equivalent to that of the average total pressure of the surface.

Fig.4 Total-pressure recovery coefficient maps of AIP section.

Fig.5 Total-pressure recovery coefficient difference maps.

When double baffles are inserted simultaneously,the AIP section’s total-pressure recovery coefficient atlas also has three types of flow field characteristics,i.e.,(A)an asymmetric flow field under the asymmetric baffles,(B)a symmetric flow field under symmetric baffles,and(C)an asymmetric flow field under the symmetric baffles.When the baffles are inserted asymmetrically,two low-pressure areas are formed at the positions of the two baffle occlusions,and the low-pressure area has a hump shape.The high-pressure area is sandwiched between two low-pressure areas.When the baffles on both sides are inserted to a lower depth(in this case,the insertion depth is less than 15%),the high-pressure area is a symmetrical saddle around the longitudinal centerline of the cross section.When either side is inserted deeper than a certain position(in this case,the insertion depth is greater than 20%),the high-pressure area deviates the longitudinal centerline of the cross section and is offset to the side with lower insertion depth;the saddle-shaped high-pressure zone is also separated into two parts that are essentially symmetrical up and down.In the case of a symmetric baffle,there is a critical baffle depth.The flow field is basically symmetrical when the insertion depth of the bilateral baffle is below the critical depth(in this case,the insertion depth is below 15%),beyond which the flow field of the high-pressure district begins to favor one side,and it is separated into two parts with up and down symmetry.Similar to the single baffle,the double baffles also create a strip region in the longitudinal center of the cross-section where the total-pressure recovery coefficient is roughly equivalent to the coefficient of the area-average total-pressure recovery of the section.The mechanism of the asymmetric flow field under symmetric double baffles requires further study.

5.3.Results of distorted parameters

In this part,we tested the distortion parameters under different baffle positions from the engine idle state to the maximum air flow rate state.Because some baffle locations will surge before they reach the maximum engine air flow rate,only part of the baffle states can work on the maximum air flow state of the engine.

Figs.6–9 show the variation of,W andwithfor 18 cases.In order to facilitate comparison,the same color and logo are used to reflect the single-baffle and double-baffle states with similar relative insertion areas.The fourth-order polynomial fitting method is used to fit W,and εavof the test points with different baffle positions.The fitted curves well reflect the trends of the variations of the parameters and their tendencies.Compared to Δσ0and W,and εavhave some jump fluctuations in several test points,which also reflects that these two parameters have fluctuations in the test.In general,the figures show that the four distortion indexes increase gradually with.As the depth of the baffle increases,the rate of variation also increases.

Fig.6 compares the variation ofwithin the singlebaffle and double-baffle states.The figure shows that,with the same baffle position,gradually increases asincreases;with the same,increases as the relative insertion area increases.From the comparison of the states that the relative insertion areas are approximate in the figure,it is apparent that,for the same relative insertion area and,of the veneer status is higher than the status of the doublebaffle.For example,the relative insertion area of veneer status Case 25/0 is 19.6%,and thevalue for the correspondingexceeds both status Case 10/20(S=19.4%)and status Case 15/20(S=23.6%).This can be explained based on the effect of the baffle on σav,as explained above.As the relative insertion area increases,the σavnonlinearity decreases rapidly,i.e.,the total pressure decreases rapidly in a nonlinear fashion.In this way,for the same relative insertion area,the state of the veneer will produce a lower local low-pressure state,resulting in an increase in the degree of circumferential distortion.

Fig.7 shows that,under the same baffle position,εavalso increases gradually with Wac1;with the sameWac1,εavincreases gradually as the relative insertion area increases.For the same relative insertion area andWac1,the εavvalue corresponding the single baffle is similar to or slightly higher than the status of the double-baffle,while the degree of excess is not as much as that of.Since W is the sum of and εav,W of a singlebaffle will be higher than that of a double-baffle under the same Wac1and relative insertion area.Fig.8 also shows this characteristic.

The characteristic features ofare different from the previous three distortion parameters.Fig.9 shows thatshows a gradually increasing trend as the relative insertion area and Wac1increase for either a single-or a double-baffle.The variation ofis small when the relative insertion area is between 10%and 30%,and the variation ofis severe when the relative insertion area exceeds 30%.In a state in which the relative insertion area is less than about 10%,for the same Wac1,in the double-baffle state is larger than what it is in the single-baffle state.The two are close to each other between 10%and 30%,andof the single-baffle state is larger than that of the double-baffle state when the relative insertion area is greater than about 30%.

Selecting some states with similar relative insertion areas as the object of the research,we studied the weight of the circumferential distortion descriptor and turbulence in the complex total-pressure distortion descriptor with the variation of the air flow rate.Fig.10 shows that the weight of the steadystate circumferential distortion descriptor in the complex total-pressure distortion descriptor increases gradually as the air flow rate increases.For the same relative insertion area,the single-baffle has a larger steady-state weight coefficient than the double-baffle.In the double-baffle states,the double-baffle state with a larger one-side baffle depth has a trend of having a larger steady-state weight coefficient.From another perspective,the weight of turbulence in the complex total-pressure distortion descriptor produced by the doublebaffle is higher than that produced by the single-baffle.

5.4.Modeling and analysis of distortion test data

In order to research the relationship and rules between the distortion parameters and the location of the baffle,we performed some data modeling and analysis using the experimental data.

Due to the excellent data fitting ability of Back-Propagation Artificial Neural Networks(BPANN),we selected BPANN with a 3-layer structure to model the distortion test data, consisting of three input nodes, 10 hidden layer nodes, and three output nodes. The input parameters were Wac1,H1andH2,and the out put parameters were εav,and W.The input and output parameters were normalized from 0 to 1.We selected 2341 testing points from the test data,and we obtained an additional 2341 testing points by exchanging the data of H1and H2,so the total modeling data consisted of4682points.70%of the data were selected as training data,15%were the verification data,and the remaining 15% were the test data.Using the

Fig.6 Variation of Δ

Fig.7 Variation of εavwith Wac1.

σ0with Wac1.Levenberg-Marquardt training algorithm,the network error performance obtained after training is shown in Fig.11.The abscissa is the difference between the model output data and the experimental data, and the ordinate is the number of samples for the corresponding error. The figure shows that the distortion network model based on BPANN has high enough precision for both the training data and the test data.The model can be used for the analysis of the test data.

Fig.8 Variation of W with

Wac1.

Fig.9 Variation of Δ

σrmwith Wac1.

Fig.10 Weight of steady-state circumferential distortion descriptor in complex total-pressure distortion descriptor varying with air flow rate.

Fig.11 Distortion network model based on BPANN error histogram.

The BPANN model presented above was used to conduct a comparative analysis of the single and double baffles.We selected the state in whichwas 0.9 and H1and H2were changing in the range of 0–20%to study the working characteristics of double baffles.Fig.12 shows the variation of εavandwith different baffle combinations.The pink scatter marks in the figure are the distortion parameter combination range formed by the two baffles working in the range of 0–20%relative insertion depths.Such an area is an enclosed area surrounded by three border lines.The analysis shows that the lower line marked with a solid red line is a work envelope with a baffle position of 0 and another baffle changing from 0 to 20%.This line reflects the ability of a single-baffle,which is defined as the single-baffle working line in this article.The right blue boundary line is the one-sided baffle relative insertion depth of 20%,and the other side is from 0 to 20%;in this paper,we define this line as the single largest baffle working line.The left boundary line of cyan is the working line of a symmetric baffle that changes simultaneously from 0 to 20%,which is defined as the symmetrical baffle working line.

Fig.12 Example of variation of εavand with different combinations of baffles(Wac1=0.9).

In addition,the difference between the single-baffle and double-baffle was analyzed from the view of the relative insertion area.The black°-mark in the lower right corner is a single-baffle status with one side baffle at 0 and the other side at 20%,and the relative insertion area of the state is 14.2%.Based on the relative insertion area of 14.2%,we can look for some double-baffle states for which the total insertion area is 14.2%.The black solid line in Fig.12 shows the characteristic line that consists of various double-baffle states with a relative insertion area of 14.2%.It can be seen that the equal relative insertion area states formed by the double baffles are a working line that extends from the working point of the single baffle to the left boundary.In this process, εavanddecline gradually,anddeclines faster.This means that a single-baffle state will correspond to a symmetrical doublebaffle state,with minimum εavand.Similarly,we can analyze the relationship between single-baffle and double-baffle in the case of the other insert areas.The equal relative insertion area working lines of Cases 0/18 and 0/16 are also shown in Fig.12,and it is apparent that they are similar to Case 0/20.The difference is that,as the insertion area decreases,the degree of reduction of the turbulence decreases.

Further analysis was performed on the influence of the status of the engine on the work envelope.Fig.13 shows the change of the working envelope of Wac1from 0.5 to 0.9 when the relative insertion depth of the baffle is kept at 20%or less.The figure shows that,with the same baffle relative insertion depth,as the engine state increases,the double-baffle simulating envelope expands gradually.

The analysis above shows that,after adding another baffle,the simulation capability of the baffle distortion generator is extended from only one line of a single baffle to a working area surrounded by three working lines.In the example of Fig.12,the maximumsimulation capability of the double-baffle increases by 198%compared with that of the single-baffle;εavhas an increase of 120%,and the simulation range of the distortion is increased significantly.

5.5.Results and analysis of engine instability test

In the test,we also conducted a comparative study of engine instability with different baffle forms.We chose the condition of Nhcaround 90%in this study,and we conducted engine surge test with different forms,i.e.,single-baffle,symmetrical double-baffle,and asymmetric double-baffle.Table 1 lists the main parameters of the engine and the distortion at the last operating point before the engine became unstable under the four conditions.Based on the data in the table,we conclude the following:

Fig.13 Change of working envelope of Wac1from 0.5 to 0.9(for a relative insertion depth not more than 20%).

(1)The operating conditions of the engines in the table are close to each other.The difference between the maximum and minimum values of Nlcis 2.7%,and the difference between the maximum and minimum corresponding critical complex total-pressure distortion descriptor(the value at the last operating point before the engine becomes unstable)is 11%.A complex totalpressure distortion descriptor is used to evaluate the stability of the engine with a single baffle.The experimental results show that the critical complex total-pressure distortion descriptor also has a certain degree of engineering accuracy if the double-baffle is used for the evaluation of the stability of the engine.

(2)By comparing the two cases of the single-baffle state,i.e.,Case 50/0 and Case 43.5/0,with the two doublebaffle cases,i.e.,Case 31.5/31.5 and Case 42.9/10,it is apparent that,irrespective of whether the states are single-baffle or double-baffle states,when the engine is unstable,the operating conditions and the corresponding critical complex total-pressure distortion descriptor are inversely proportional to each other.As the engine’s operating speed increases,its critical complex totalpressure distortion descriptor decreases.

(3)By comparing single-baffle Case 43.5/0 with two doublebaffle cases,i.e.,Case 31.5/31.5 and Case 42.9/10,it is apparent that the critical complex total-pressure distortion descriptor is comparatively higher when the operating speed is slightly higher.As a result,it can be inferred that,for the same engine operating condition,the critical complex total-pressure distortion descriptor in the double-baffle state will be smaller than that in the single-baffle state when the engine loses stability.

(4)No obvious correlation is found between the obtainedand stability.

(5)According to the stability index KS,Case 31.5/31.5 is about 3%lower than the other states.Although there is no further support for the test data,based on this example,we can infer that,if double-baffle is used for the stability assessment for the same engine operating condition,the estimated engine stability margin will tend to decrease gradually as the baffle symmetry increases.

The following is a further analysis of the characteristics of air flow when the engine is unstable.According to wellknown researchers,the surge process can be divided into classic surge and deep surge,depending on whether there is back flow during the surge.In order to investigate back flow at the inlet to the engine,we set the ratio of the area-averaged total pressure and static pressure on the AIP as a basis for judging the back flow.A ratio of less than one indicates that back flow has occurred.Fig.14 shows the changing state of Pt2fav/Ps2favin the four cases in Table 1 during the unstable operation of the engine.Fig.14 shows that all of the cases except Case 42.9/10 have back flow,which should have occurred in deep surge states.Fig.15 shows the changing process of total pressure on AIP(take the outermost measure point of No.2 steady-state total-pressure probe as a representativeexample,labeled as Pt2_62),the total pressure after the fan Pt21,and the total pressure after the high-pressure compressor(labeled as Pt3)during engine instability.The four samples in Fig.15 show three different characteristics.Cases 50/0 and 43.5/0 show that Pt21and Pt3continue to decrease until the surge disappears(note:All of the experiments have an automatic anti-surge system,and surges will automatically exit just after it occurs).In Case 31.5/31.5,Pt3continues to decline until the surge disappears,while Pt21has a process of decline,i.e.,it declines,then rises,and then declines again.Case 42.9/10 shows a synchronous oscillation of Pt2_62,Pt21and Pt3.In addition,from Case 50/0 in Fig.15,it is apparent that Pt21declines before Pt3,and its decline is accompanied by the increase of Pt2_62;thus,it can be concluded that the surge occurs first at the fan.In Cases 43.5/0 and 42.9/10,the total pressure on AIP increases earlier than the decline of Pt21begins.We also infer that the surge occurs first at the fan.In Case 31.5/31.5,Pt2_62increases,and Pt21and Pt3decrease almost simultaneously.It is difficult to distinguish whether the surge occurs at the fan or the high-pressure compressor from the existing data that are sample at 50 Hz.Further assessment of this issue will require the analysis of dynamic pressure data.According to Fig.14 and Fig.15,it can be assumed that Cases 50/0 and 43.5/0 indicate a deep surge that occurs first at the fan before it occurs at the high-pressure compressor.Case 42.9/10 shows classic surge characteristics with rotational stall.Case 31.5/31.5 shows a deep surge,but it differs from the performance of Cases 50/0 and 43.5/0.

Table 1 Main parameters of engine and distortion at the last operating point before engine becoming unstable.

Fig.14 Change of Pt2fav/Ps2favduring unstable state of engine.

6.Conclusions and outlook

In this paper,using a CVDBDG and an after burning turbofan engine,we conducted research on the inlet total-pressure distortion on an indoor engine test bed,and we measured the distortion simulation characteristics of the CVDBDG and the instability characteristics of the engine.The method of back propagation artificial neural networks was used to model the distortion test data and to analyze the working envelope of CVDBDG.Through the analysis of the engine instability data,we obtained guidance concerning the use of CVDBDG to evaluate the stability of an engine.Through the above experimental and modeling analyses,we can draw the following conclusions and prospects for future engineering applications and research of the CVDBDG:

(1)CVDBDG can simulate steady-state and time–variant total-pressure distortion in real time.The process of changing two baffles in the same maximum insertion depth range can form a simulated work area of steady state and time–variant total-pressure variations that consist of a single-baffle working line,a symmetrical double-baffle working line,and a single maximum baffle working line.

Fig.15 Change of Pt2_62,Pt21and Pt3during unstable state of engine.

(2)Compared with the movable single-baffle distortion generator,both the capability and range of the distortion simulation of steady-state and time–varianttotal pressure of CVDBDG are extended significantly.When another baffle is added,a simulated working point of the movable single-baffle distortion generator is expanded to a simulated working line while the relative insertion area is kept unchanged.When the relative insertion depth is kept constant,an analog working line of a movable single-baffle distortion generator is expanded into a simulated working area.

(3)When the engine becomes unstable during different combination styles of the double-baffle simulation conditions,the characteristics are different from the single-baffle.Based on this feature,double-baffle is expected to be used in evaluating some kinds of engine surges to verify whether the engine has some potential destabilization hazards through different kinds of baffle combinations.This is a direction worth further studying.

(4)Under the same engine operating conditions,the critical complex total-pressure distortion descriptor that led to instability of the engine decreases with the increase of the symmetry of the double-baffle.In addition,the estimated surge margin also shows a downward trend.From this perspective,if the CVDBDG is used to assess stability,the evaluation results tend to become more stringent as the symmetry increases.

(5)Under the condition of a symmetric double-baffle,there exists a critical depth of insertion beyond which the symmetric baffle will generate an asymmetric flow field.Further study of the mechanism of the critical insertion depth can achieve more accurate simulation of analog control using the double-baffle approach.

(6)CVDBDG can produce more complex and diverse flow fields,which also means that excitation air flows with a wide range of frequencies can be generated in this way.Perhaps it can be a new way to assess the durability of engines.

(7)The CVDBDG greatly extends its capabilities compared with a single movable baffle distortion generator,and it is expected to become a new and more effective means of assessing the stability and durability of engines.

Acknowledgements

The engine test was supported by the Beijing Aeronautical Technology Research Center.The authors express their appreciation for everyone who was involved in and contributed to the experiment.