Professional Documents
Culture Documents
FEDSM2012
July 8-12, 2012, Rio Grande, Puerto Rico
FEDSM2012-72094
ABSTRACT INTRODUCTION
Active and passive techniques have been used in the past, Low pressure turbine plays a vital role in increasing the
to control flow separation. Numerous studies were published on effectiveness of engine. Many researches were carried out to
controlling and delaying the flow separation on low pressure investigate the boundary layer separation in LP turbine. This
turbine. In this study, a single dimple (i.e. passive device) is separation occurs at any place on the wings, airfoils, or in
engraved on the suction side of LP turbine cascade T106A. The engine inlets where the pressure gradient is high. When the
main aim of this research is to find out the optimum parameters Reynolds number is low the velocity of fluid particles near the
of dimple i.e. diameter (D) and depth(h) which can produce wall surface decreased so much that reversal of the flow is
strong enough vortex that can control the flow either in being done on that point which we generally called the
transition or fully turbulent phase. Furthermore, this optimal separation. This separation directly affects the efficiency of low
dimple is engraved to suppress the boundary layer separation at pressure turbine and increase the fuel consumption. It has been
different Reynolds number (based on the chord length and inlet perceived that 1% improve in efficiency will reduce the fuel
velocity). The dimple of different depth and diameter are used cost of 0.5 to 1% [1].
to find the optimal depth to diameter ratio. Computational Turbine is designed to operate efficiently at high Reynolds
results show that the optimal ratio of depth to diameter (h/D) number but at higher altitude, this Reynolds number decrease to
for dimple is 0.0845 and depth to grid boundary layer (h/δ) is 25000 due to the low air density [2]. Sharma [3] observed that
0.5152. This optimized dimple efficiently reduces the 300% increase in loss coefficient if the gas turbine Reynolds
normalized loss coefficient and it is found that the negative number reduces to 95000 or below. This decrease Reynolds
values of shear stresses found in uncontrolled case are being number result in higher pressure gradients. Due to these
removed by the dimple. After that, dimple of optimized pressure gradients, shear wall stress at wall approaches to zero
parameters are used to suppress the laminar separation bubble and flow starts to separate with a formation of laminar
at different Re~25000, 50000 and 91000. It was noticed that the separation bubble (LSB) on the suction side of LP turbine that
dimple did not reduce the losses at Re~25000. But at may or may not attach with the blade surface as reported by
Re~50000, it produced such a strong vortex that reduced the Luo et al [4]. The size of separation bubble has direct relation
normalized loss coefficient to 25%, while 5% losses were with the aerodynamics performance formulated by Mayle [5].
reduced at Re~91000. It can be concluded that the optimized Turbulent flow is more energetic flow and is more resistant
dimple effectively controlled flow separation and reduced over pressure gradient in comparison with the laminar flow.
normalized loss coefficient from Re 25000 to 91000. As the Due to this advantage, different flow control devices are to use
losses are decreased, this will increase the low pressure turbine to suppress the boundary layer separation or trip the incoming
efficiency and reduce its fuel consumption. laminar flow into turbulent. The parameters which directly
affect the transition are free stream turbulence intensity,
turbulent length scale, pressure gradient and roughness of blade
GRID GENERATION
Figure 3 shows 2D mesh around the T106A cascade in
which the seal view of leading and trailing edge are also shown.
Laminar boundary layer is created around the cascade by using
the formula [20] while the unstructured meshing is used in the
rest of the flow domain. Grid points are converged close to
cascade surface which guarantees the y+ remains below 1.
Velocity inlet is used in each case while the pressure outlet at
the exit. In the Fig. 1., graphical smooth edge shallow dimple is
shown while the mesh around this shallow dimple is shown in
the Fig. 4.
ANSYS Fluent® finite vloume based CFD commerical FIGURE 4: MESH AROUND THE DIMPLE
code is used for computations. The pressure based solver is
used because Re is low and due to locally un-steadiness in the model explain the better flow history on the location of
flow, transient case is run. In viscous model, the SST transition transition onset by correcting of the model coefficients [21].
or Gamma-Theta [21] is used for the transient case which is The transport equations of Gamma Theta Model are:
accurate for the prediction of transition. PISO algorithm is used
in the pressure-velocity coupling scheme as it is recommended 𝜕(𝜌𝛾) 𝜕(𝜌𝑈𝑖 𝛾) 𝜕 𝜇𝑡 𝜕𝛾
for transient flow calculations, especially for large time step- + = 𝑃𝛾 − 𝐸𝛾 + 𝜇+ (1)
𝜕𝑡 𝜕𝑥𝑖 𝜕𝑥𝑖 𝜎𝑓 𝜕𝑥𝑖
size. Least Squares Cell-Based Gradient Evaluation is used
instead of Node-Based Gradient, as it is computationally less 𝜕(𝜌ReΘt ) 𝜕(𝜌𝑈𝑖 ReΘt ) 𝜕 𝜕ReΘt
intensive and the accuracy of both is comparable in + = 𝑃Θt + 𝜎 𝜇 + 𝜇𝑡 (2)
𝜕𝑡 𝜕𝑥𝑖 𝜕𝑥𝑖 Θt 𝜕𝑥𝑖
unstructured meshing. Second order upwind scheme is used for The first equation is used to solve intermittency γ and second
the computations. Net mass flow rate is taken as a convergence one is designed for momentum thickness Reynolds number
criteria and it is computed up to 10-7kg/s. ReΘ. At the inlet intermittency (γ) is fixed to 1 and due to the
normal flux γ is taken as zero at walls. For the conservation of
1.06
1.05
1.04
1.03
1.02
1.01
1
0.99
0.98
0.07 0.12 0.17 0.22
h/D
1.3
In the second study we aim to find out the finest depth
1.2
(h) of the dimple where separation losses are minimum and
flow is better controlled. Diameter (D) of the dimple is found in 1.1
early section which is 20mm, so after this all geometries are 1
built with this diameter. In this study h/D ratio is varied by
keeping the D constant i.e. 20mm from where we find the 0.9
FIGURE 13(c): 85%Cax DIMPLE AND BASELINE CASE FIGURE 13: BOUNDARY LAYER PROFILES AND VELOCITY
GRADIENT PROFILES FOR MODIFIED T016A BLADE USING
DIMPLE
91000 5%
FIGURE 13(d): 95%Cax DIMPLE AND BASELINE CASE
50000 25%
Figure 13 (e) shows the plot of normalized velocity and 25000 1%
pressure gradients plot. The boundary layer is thick in dimple
case at 100%Cax in comparison with baseline case which shows Figure 14 represents the comparison of Cp plots of
that flow is turbulent. The values of velocity gradient are baseline case with the dimple case at Re 50,000. In the baseline
ACKNOWLEDGMENTS
Authors would like to thank Principal RCMS Engr.
Sikandar Hayat Mirza for providing excellent research
atmosphere in the center. We are also grateful to Mr.
SijalAhmed, Aamer Shehzad, Amaar Mushtaq and Muhammad
Nafees Mumtaz Qadarifor their valuable inputs and discussions
during this research work.
REFERENCES
[1] Chishty, M.A., Hamdani,H. R., Parvez, K., “Comparison
of Passive Devices for Flow Controlling on Low Pressure
FIGURE 15(d): 95%Cax DIMPLE AND BASELINE CASE Turbine Cascade,” Proceedings of ICAMS2011, Pakistan, Nov
28-30, 2011.
CONCLUSION [2] Huang, J., “Separation Control over Low Pressure Turbine
In this study, computations were performed on the T106A Blades using Plasma Actuators,” PhD Thesis, University of
LP turbine cascade and the smooth shallow dimple was used to Notre Dame, Indiana, July, 2005.
control the flow separation. For this, the dimple of different [3] Sharma, “Impact of Reynolds Number on Low Pressure
depths and diameters were engraved on the suction of the Turbine Performance,” NASA CP-1998-206958, 65-70, 1998.
cascade. It was found after computation that the dimple at 65% [4] Luo, H., Qiao, W., Xu, K., “Passive Control of Laminar
Cax with depth to grid boundary layer (h/δ) of 0.5152 and depth Separation Bubble with Spanwise Groove on a Low-speed
to diameter (h/D) of 0.0845 reduced the normalized loss Highly Loaded Low-Pressure Turbine Blade,” J. ThermSci Vol.
coefficient to 5% at Re~91,000. After that, the dimple of above 18, No. 3, pp. 193-201, 2009.
specification was used to control the flow separation at [5] Mayle, R. E., “The Role of Laminar-Turbulent Transition
different Re (i.e. 25,000 and 50,000). It was noticed that the in Gas Turbine Engines,” ASME Journal of Turbo machinery,
dimple did not reduce the losses at Re~25000. But at Vol. 113, Issue 4, pp. 509-537, 1991.
Re~50000, it produced such a strong vortex that reduced the [6] Schlichting, H., “Boundary Layer Theory,” 7 th Ed,
normalized loss coefficient to 25%, while 5% losses were McGraw-Hill, New York, USA, 1979.
reduced at Re~91000. It can be concluded that the optimized