FLOW SEPARATION OF AXIAL COMPRESSOR CASCADE BLADES

An experimental and theoretical investigation of the effect of flow separation on the performance of a cascade NACA 65_(12)10 axial compressor blade has been carried out. The experimental work includes the fabrication of three blades from wood, each having a chord (100mm) but one of these blades having a span of (90mm) for smoke tunnel testing and the other two blades having a span of (380mm) for wind tunnel testing.The two blades were connected by suitable mechanism in order to be fixed in the wind tunnel protractor and rotated in the required stagger angle. The cascade was tested in an open type low-speed subsonic (Mach number=0.117) wind tunnel, for Reynolds number (Re=239605) based on maximum velocity (35 m/s) and airfoil chord length. The total and static pressures were measured in selected points between the two blades for stagger angles of (4 0 , 0 0 ,- 4 0 ,-8 0 and -12 0 ) by using a multi-tube manometer and a pitot static tube. The small blade (90mm span) is tested in the smoke tunnel to visualize the real behavior of flow separation. The theoretical work includes using the software FLUENT (V6.2) to simulate the flow between the two blades. The study shows that the flow separation begins when the cascade are inclined at a stagger angle of (- 4 0 ) on the suction side of the lower blade at a position (96%chord experimentally and 98%chord theoretically). Then, the separation zone increases with increased stagger angle (in clockwise direction) and reach to the position (61%chord experimentally and 63%chord theoretically) at a stagger angle (-12 0 ).These results are validated by a smoke tunnel tests.This separation affects the performance of the compressor, where the static pressure ratio ( eS p


ABSTRACT
An experimental and theoretical investigation of the effect of flow separation on the performance of a cascade NACA 65_(12)10 axial compressor blade has been carried out.The experimental work includes the fabrication of three blades from wood, each having a chord (100mm) but one of these blades having a span of (90mm) for smoke tunnel testing and the other two blades having a span of (380mm) for wind tunnel testing.The two blades were connected by suitable mechanism in order to be fixed in the wind tunnel protractor and rotated in the required stagger angle.The cascade was tested in an open type low-speed subsonic (Mach number=0.117)wind tunnel, for Reynolds number (Re=239605) based on maximum velocity (35 m/s) and airfoil chord length.The total and static pressures were measured in selected points between the two blades for stagger angles of (4 0 , 0 0 ,-4 0 ,-8 0 and -12 0 ) by using a multi-tube manometer and a pitot static tube.The small blade (90mm span) is tested in the smoke tunnel to visualize the real behavior of flow separation.The theoretical work includes using the software FLUENT (V6.2) to simulate the flow between the two blades.The study shows that the flow separation begins when the cascade are inclined at a stagger angle of (-4 0 ) on the suction side of the lower blade at a position (96%chord experimentally and 98%chord theoretically).Then, the separation zone increases with increased stagger angle (in clockwise direction) and reach to the position (61%chord experimentally and 63%chord theoretically) at a stagger angle (-12 0 ).These results are validated by a smoke tunnel tests.This separation affects the performance of the compressor, where the static pressure ratio ( e S p / i S p ) decreases as the separation zone gets bigger.The range of working stagger angle is then calculated.It was found in the range (-18 0 to 36 0 ).The flow behavior between the two blades shows that the blade-to-blade configuration works as nozzle-diffuser.The theoretical results were compared with the experimental results and good agreement was obtained.

INTRODUCTION:
The compressor, which is the important part of gas turbine engines, has to be given special attention during operation.The main task of the axial-flow compressor is to increase the pressure of air by converting air kinetic energy through a series of rotating and stationary blades.One of the most important problems that affect performance is the flow separation.Separation starts by deviation of fluid particles away from blade surface in the boundary layer.This causes a drop in kinetic energy, and cause the flow to re circulate [You D. AND Moin, P., 2006].After stall region the fluid particles velocity reaches to zero in boundary layer near to blade surface and this deceleration causes increasing in boundary layer thickness and at a small distance after stall region the particles stop and reverse in direction due to positive pressure gradient.The low Reynolds number in conjunction with the local adverse pressure gradient makes it susceptible to flow separation [Meinhard T. Schobeiri etal, 2003].This study consists of two major parts: the experimental part a cascade tested in an open jet low speed wind tunnel.The total and static pressure between two axial compressor blades were measured using a Pitot -static tube and a manometer.To visualize the flow a smoke tunnel was used.The stagger angle was taken equals (4 0 , 0 0 ,-4 0 ,-8 0 , and-12 0 ).Secondary, in the theoretical part that depends on simulation the flow between the two blades by using a software FLUENT.The objectives of this paper are: 1. Study the effect of viscous flow separation (through the cascade of an axial compressor) on the flow variables (velocity, static pressure, total pressure…etc) by utilizing FLUENT (V6.2) software.2. Investigate the effect of stagger angle on flow separation.3. Comparison of the experimental results with the theoretical results.

* Governing Equations:
The domain for which the model is build is shown in fig.
(1) since the flow through a compressor is three-dimensional and quite complex, a simplified approach is adopted to analyze the fluid flow through cascade passage in two dimensions.Figs.
(1) and (2) show the cascade and the physical domain.The equations of continuity, momentum and turbulence model are [Hill, P. G., 1965]:  Continuity Equation:  X-Momentum Equation:  Y-Momentum Equation:  The turbulence kinetic energydissipation model at high Reynolds number is used [Moult, A. etal,1977].A-Turbulence kinetic energy: The quantity k G is the generation term for the kinetic energy of turbulence given by: turbulence model was extended by ref. [Jones, W. P. and Launder, B. E., 1972] to low-Reynolds number flow as follows:  Plotting the results.In the present work a cascade consists of two blades, each having a chord of (100mm) but one of these blades has a span of (90mm) and the other two blades have a span of (380mm) were made from wood fig.Low-speed subsonic (Mach number=0.117)open type wind tunnel is used in this work of cross section of (305mm* 305mm).Wind speed of (35m/s) is achievable allowing experiments on many aspects of incompressible air flow and subsonic aerodynamics to be performed at satisfactory Reynolds number.Reynolds number is (Re=239605) based on inlet velocity and blade chord.The tunnel has a smooth contraction fitted with the protective screen.The test section is constructed of clear Perspex, to see the blade to blade configuration clearly with a square cross section of (305mm 305mm) and a length of (610mm).The blade to blade configuration was put inside the test section parallel to flow direction and connect by protractor to limit stagger angle.The control of the stagger angle of the blade to blade configuration was made by a suitable mechanism, fig.( 9).The upper surface of the test section has a slot used to fix the pitot-static tube to measure the static pressure and total pressure.Fig.( 8) is a photo of the wind tunnel and the working section.Downstream of the test section is a diffuser which leads to an axial flow fan driven by a (5.6kW three phase A.C motor).The flow is controlled by a butterfly valve before exhaust to atmosphere through exhaust section.The air enters the tunnel through a carefully shaped diffuser.The test section gives a full visibility of flow field.-Pitot -Static Tube: The purpose of using pitotstatic tube is to measure air static pressure, total pressure and then velocity inside test section.The external dimensions are (5mm) diameter, (200mm) arm length.Tube reading were corrected according to reference [Omran, K. J.,2003] as following:  12).The air is drawn by a fan which is rotated by an electrical motor of variable velocity at the top of the tunnel.The air enters to the tunnel at the base.The test section has dimensions of (180mm) width, (240mm) height and (100mm) deep.The models installed in the back wall of the test section, the front wall of the test section are easy to remove.It is made of clear Perspex.Smoke generation is controlled at the bottom of the tunnel.The section has (23) holes from it smoke enter.The space between any two adjacent holes is (7mm).A high light source is put in the sides of the test section to see smoke clearly.The smoke generated by a smoke generator which evaporates kerosene in class evaporator carried on the front wall of the smoke generator.The smoke generated is dragged by the fan at the top of the tunnel through the test section.Flow Photo is taken using a digital Camera.

* Procedure of Experiments:
The flow between two axial flow compressor blades was tested by wind tunnel as described in the following steps:  Measure the atmospheric pressure and temperature before carrying out the experiments to calculate the air density accurately. Fix the blade to blade configuration in the test section with the required stagger angle. Prepare the multi tube manometer, controlling speed valve, and pitot static tube.
 Operating the wind tunnel for (15min) to reach steady state conditions. Reading the dynamic head by using pitot static-tube which is connected to multi tube manometer by suitable connection tubes. Repeating the previous procedure for other stagger angles ranges from( 0 4 to -0 12 ). The experiments carried out in smoke tunnel were as following:  Fill the bottle with Kerosene to the required limit. Put the model inside smoke tunnel test section and fixed at a certain angle indicated by protractor. 3. Turn on the smoke generator. 4.After about (3minutes) the smoke begins to formulate. 5.After smoke formulation the fan is turned on and controlled by the speed controller. 6.A high resolution digital camera was then used to photograph the models and flow.

RESULTS AND DISCUSION: * Experimental Results:
The experimental results are presented for three different curve lines located at (0.125,0.5 and 0.875) of (S) as shown in fig.( 13).The operating and boundary conditions of the blades and flow passage are listed in table(1).The velocity and static pressure distribution (for the three sections) of flow between the two blades for stagger angle (4 0 , 0 0 ,-4 0 , -8 0 and -12 0 ) are presented in fig.( 14)and( 15).These figures show that the flow is accelerated along section 1 (0.125S) up to a certain position,( refer to third column of table(2).Then the flow decelerates to the exit for stagger angles (4 0 and 0 0 ) up to a certain position, see the fourth column of table (2) for stagger angles (-4 0 , -8 0 and-12 0 ) where the flow is separated.The flow decelerates along section 3 (0.875S) from the inlet to the exit for all taken stagger angles.The velocity and static pressure remains constant along the section 2(0.5S) from the inlet to the exit for stagger angles (-8 0 and -12 0 ) and to a certain position for stagger angles (4 0 , 0 0 and -4 0 ), the fifth column of table (2).Then the flow decelerates to the exit.The pressure ratio is the exit static pressure from blade to blade passage divided by inlet static pressure.The pressure ratio for all cases is shown in table (2).
The reason behind such behavior is the effect of stagger angle.Within a certain range of stagger angles (from 4 0 to -4 0 ) the flow remains in contact with the blade surface.As the stagger angle is increased, the blade profile forces the flow a way from its surface, hence, the flow starts to separate.Separation starts close to the trailing edge and progresses upwards as the stagger angle is changed (from -4 0 to -12 0 ).This effect is not clear in the middle section of the passage as the profile effects on flow behavior diminishes.The behavior of static pressure distribution is complimentary to the behavior of velocity.The flow separate from the suction side of the lower blade of cascade for stagger angle (from -4 0 to -12 0 ) because of adverse pressure gradient.The total pressure distribution of flow between two axial compressor blades for different stagger angle is presented in figure ( 16) .This figure show that the fluid total pressure decreases as it passage from the inlet to the exit due to friction losses.The total pressure losses increases when the stagger angle increases (in clockwise direction) and the maximum total pressure losses occurs for stagger angle (-12 0 ) in separation region.

COMPUTATIONAL RESULTS:
The operating and boundary conditions for the flow passage between two axial compressor blades are listed in Table (3).The velocity and static pressure contours of flow between two axial compressor blades for stagger angles (4 0 , 0 0 ,-4 0 , -8 0 and -12 0 ) are presented in fig.( 19) and ( 20).These figures show that the fluid flow being accelerated near and along suction side (upper surface of lower blade) up to a certain position, see the third column of tables (4).Then the flow decelerates to the exit for stagger angles (4 0 and 0 0 ) and to a certain position for stagger angles (-4 0 , -8 0 and -12 0 ), see the fourth column of table (4), where the flow separates from the blade surface.The flow decelerates near and along the pressure side (lower surface of upper blade) from the leading edge to the tailing edge along the chord line for all cases.The velocity and static pressure remains constant in the midspace between the two blades from the inlet to the exit for stagger angles (-8 0 and -12 0 ) and to the certain position for stagger angles (4 0 , 0 0 and -4 0 ), see the fifth column of tables (4).Then the flow decelerates to the exit.The fluid flow being accelerated as it passes through the contracting flow area after that the flow being decelerated to the trailing edge.These phenomena prove that the suction side works as a nozzle-diffuser, the pressure side works as a diffuser and the blade to blade configuration works as a nozzle-diffuser.The total pressure contours of flow between the two blades for stagger angle (4 0 , 0 0 ,-4 0 , -8 0 and -12 0 ) are presented in fig.( 21) .This figure shows that the total pressure decreases as it passes from the inlet to the exit due to friction losses.The total pressure losses increases when the stagger angle increases (in clockwise direction) and the maximum total pressure losses occurs when stagger angle (-12 0 ) in separation region due to increased friction losses and adverse pressure gradient.
The flow separation affects the performance of a cascade and hence, affects the compressor performance where the pressure ratio decreases when the separation zone increases.

* Accuracy of Solution:
The FLUENT (V6.2) code is considered as a high accuracy computational fluid dynamics software package.Fig.( 22) presents the distribution of residual and number of iterations for continuity, momentum and k-ε model equations and for stagger angle (0 0 and -8 0 ).As the stagger angle increases, the number of iterations to convergence increases because the problem gets complicated when the flow separation occurs, see table (3).

THE COMPARISON BETWEEN THE EXPERIMENTAL AND THEORICAL RESULTS:
The theoretical results obtained in this study by using FLUENT (V6.2) were compared with those obtained experimentally using the wind tunnel.

CONCLUSIONS:
 1-The area between two axial compressor blades for NACA 65_ (12)10 works as a nozzlediffuser. 2-Smoke tunnel was successfully used to show separation of flow from upper blade surface. 3-The flow separation was seen to start at a stagger angle of (-4 0 ) experimentally and theoretically. 4-The mathematical relationship between the static pressure ratio and stagger angle for NACA 65_ (12)10 axial compressor cascade is concluded by using curve fitting method for polynomial distribution.The range of stagger angle for NACA 65_ (12) 10 axial compressor blade aerofoil is calculated from this relationship.It was found in the range (-18 0 to 36 0 ).
 5-Total pressure was seen to reduce through the blade passage.It drops sharply for the separation zone.
 6-FLUENT (V6.2) was used successfully to predict separation as compared to the experimental measurements in the wind tunnel and as compared with other works.
(6).The two blades of (380mm in span) are connected by suitable mechanism, as shown in fig.(7) in order to be fixed in wind tunnel protractor and achieved on required stagger angle.The two blates are fixed by aluminum plates, fig.(7).
Fig.(10): Multi Tube Manometer.Fig.(11): Correction of Pitot -static tube Distance.[Omran, K. J., 2003].Smoke Tunnel: To illustrate a real view of flow development and separation, a smoke tunnel was used as shown in fig.(12).The air is drawn by a fan which is rotated by an electrical motor of variable velocity at the top of the tunnel.The air enters to the tunnel at the base.The test section has dimensions of (180mm) width, (240mm) height and (100mm) deep.The models installed in the back wall of the test section, the front wall of the test section are easy to remove.It is made of clear Perspex.Smoke generation is controlled at the bottom of the tunnel.The section has (23) holes from it smoke enter.The space between any two adjacent holes is (7mm).A high light source is put in the sides of the test section to see smoke clearly.The smoke generated by a smoke generator which evaporates kerosene in class evaporator carried on the front wall of the smoke generator.The smoke generated is dragged by the fan at the top of the tunnel through the test section.Flow Photo is taken using a digital Camera.

FigX
Fig.(13): Blade to blade configuration with taken three sections in flow passage.

Fig
Fig.(23): The comparison of velocity distribution between the experimental results and the theoretical results.