BOUNDARY LAYER SEPARATION CONTROL
Introduction: Conical diffusers are common in engineering design when the flow velocity needs to be decreased and the static pressure needs to be increased. For example, in a combined cycle power plant, the diffuser between the power turbine and the steam generator fulfills such requirements. The function of the diffuser is to lower the outlet pressure for the last stage turbine and reduce the inlet velocity for the steam generator. Flow separation is inevitable in a large area ratio.
Diffuser because of the sustained adverse pressure gradient. The presence of flow separation is disadvantageous because it reduces the pressure recovery of the diffuser. Flow stability and uniformity also suffer, causing detrimental effects on downstream components. In the power plant example, the boiler effectiveness will be reduced by a non uniform flow while the components will be fatigued by an unstable flow.
The center body of the annular inlet creates a wake that can significantly affect the flow downstream in the diffuser. Our previous work (Lo et al. 2011) showed specifically that the boundary layer development along the diffuser outer wall is highly dependent on the flow in the center region. In the presence of an adverse pressure gradient, the separated wake of the center body extends over the entire diffuser length. The blockage effect of this separation bubble relieves the adverse pressure gradient and results in a relatively thin boundary layer along the outer diffuser wall. This central separation bubble can be dramatically shortened by using Coanda blowing to entrain high momentum main flow directions for the main experiments.
EXPERIMENTAL SETUP: The MRV experiments were performed at the Richard M. Lucas Center for Imaging, using a General Electric 1.5T model S3 whole body scanner. Three-component velocity measurements were obtained on a uniform Cartesian mesh following the phase contrast MRV techniques described by Elkins et al. (2003). The velocity encoding (VENC) values control the maximum measurable velocity that will be free of aliasing. The VENC values were 5 m/s for all three directions for the main experiments. A flip angle of 15 degrees was used. The X-axis is the stream wise direction, while the Y- and Z-axes are the vertical and span wise directions, respectively. The XZ plane field of view has a nominal dimension of 280 mm by 280 mm, and an effective dimension of 280 mm by 126 mm due to a phase FOV factor of 0.45. The spatial resolution in the X, Y, and Z directions are 1.09 mm, 1 mm, and 1.09 mm, respectively. The velocity data are measured on a uniform grid with 256 points in the X direction, 106 points in the Y direction, and 116 points in the Z direction.
RESULTS AND DISCUSSIONS: The present study focuses on controlling the outer wall boundary layer separation using various step-wall diffusers. The first step-wall diffuser had an axisymmetric step with a step length l=9.5h. This configuration was tested with the Coanda tail piece in place but no flow through the Coanda jet (BR=0). The contour values are azimuthally averaged and normalized by Ubulk. Azimuthally averaged velocity vectors are overlaid to visualize velocity profile development. A separation bubble is present in the immediate wake of the center body. The straight section defers the onset of the adverse pressure gradient and allows the central separation bubble to close without Coanda blowing. However, the flow still enters the diffuser with a strong velocity deficit at the center. The flow separates at the step edge and forms a recirculating region downstream of the step. When the total reverse flow area is less than 1% of the total flow area, the flow is considered to have reattached. Using this criterion, the flow reattaches at around 9.2h downstream of the step and the separation bubble is considered to be closed. The boundary layer grows along the outer wall but remains attached all the way to the diffuser exit. The center region of the flow remains slow throughout the diffuser as the adverse pressure gradient continues to decelerate the center body wake.
CONCLUSION: Three-component mean velocity field measurements were acquired for a set of conical diffusers with an annular inlet. The center body of the annular inlet creates a central separation bubble that extends over the entire diffuser length. Coanda blowing can mitigate this central separation, but at the same time the boundary layer along the diffuser wall becomes thick and in some cases separates in an unpredictable manner. A backward-facing step incorporated into the outer wall of the conical diffuser acts to stabilize the boundary layer separation. Results show that the behavior of the step separation bubble is predictable, stable, and independent of the Coanda blowing ratio.