Turbulent Flow Over A Rough Backward

You might also like

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 10

Turbulent flow over a rough backward-facing step

1. Introduction
Improved understanding of the impacts of realistic roughness on turbulence is of
particular interest to engineers who work with practical wall-bounded turbulent flows
such as those over turbine blades, aircraft and ship hulls which can become rough
after some time of operation due to various damage mechanisms.Very recently, quite
a few studies have concentrated on characterizing the effects of realistic roughness on
various aspects of turbulent flows.Wu and Christensen performed PIV experiments in
the stream wise–wall-normal planes of a fully developed turbulent channel flow over
a short fetch of a model surface damaged by pitting/installation. They observed that
the Reynolds shear stress within the internal layer was significantly enhanced due to
the creation of more intense sweeps and ejections in the presence of the studied rough
surface.Wu and Christensen measured turbulence statistics in a self-similar boundary
layer over a surface replicating a land-based turbine blade roughened by fuel
deposition. They observed that, outside of the roughness sub layer, turbulence
statistics in the rough-wall boundary layer collapse with those in the smooth-wall
flow, providing support for Townsend’s wall similarity hypothesis.This hypothesis
was also supported by the results of Allen ET AL. who measured turbulent pipe flows
in the presence of a honed surface. The spatial characteristics of large-scale turbulent
structures in boundary layers were found by Wu and Christensen to be modified by
the studied realistic roughness while the small-scale structures were observed to be
much less sensitive to the surface conditions. The impacts of various roughness scales
of complex irregular roughness topography in turbulent channel flows and boundary
layers are studied by Johnson and Christensen and Mejia-Alvarez and Christensen .

2. Experiments

The current study utilized Particle Image Velocimetry (PIV) experiments conducted
in an Eiffel-type open circuit boundary layer wind tunnel with a free stream
turbulence intensity of 0.45%. The test section dimensions are 67 x 67 cm in cross-
section, and it is 3 m long, housing a 2.90 m long hydraulically smooth flat plate with
an elliptically shaped leading edge suspended 90 mm above the tunnel floor.

A cylindrical rod downstream of the plate tripped the boundary layer, while a 100
mm-long tail flap at the trailing edge, set at 5 degrees, aimed to prevent separation.
The boundary layer growth and the inclined tail flap created a slight favorable
pressure gradient. A rough block, printed with a 3D printer and painted black to
reduce laser reflection, was introduced into the span wise center of the boundary layer
at a distance of 2.50 m from the leading edge.

The rough block, exhibiting realistic roughness representative of foreign material


deposition, measured 90 mm in length, 169 mm in width, and had a mean height of
6.35 mm. Two smooth blocks flanked the rough block, covering the entire wind
tunnel width. For smooth boundary layer flow, the rough block was replaced. The
roughness topography showed irregularities on the top surface, mimicking the trailing
edge of land-based turbine blades.
Scaled from original topographical data, the roughness had an average peak-to-valley
height of 4.2 mm and a Root Mean Square (RMS) height of 1.0 mm. The realistic
roughness, as opposed to idealized ones like wire mesh or cylinder arrays,
demonstrated complexity with elliptical elements distributed randomly across various
scales and oriented roughly in the flow direction. Further details about this roughness
can be found in previous work by Wu and Christensen (2007, 2010).

Fig. 1.Contour plot of the fluctuating heights of the roughness on the top surface of the rough block. B1 and B2 mark the
locations of the PIV measurement planes.

Two-dimensional Particle Image Velocimetry (PIV) measurements were conducted in


the stream wise–wall-normal (x–y) planes at two span wise positions (B1 and B2) for
both smooth and rough backward-facing steps of blocks at a Reynolds number Reh =
U1h/m = 3450, where U1 is the free stream velocity, h is the mean step height, and m
is the kinematic viscosity of air.

The experimental setup, illustrated in Fig. 2, involved seeding the flow with 1 µm
olive oil droplets generated by a La skin nozzle. A 200 µm-thick laser sheet, produced
by Nd:YAG lasers with cylindrical and spherical lenses, illuminated the flow field
through a transparent glass ceiling. PIV measurements were made with an 8-bit
frame-straddle CCD camera and a 105 mm lens, providing a field of view of 6h x 3.8h
(stream wise x wall-normal) and an imaging resolution of about 50 pixels/mm.

Measurements were initially taken in the upstream region of the steps, and the camera
was then moved to capture flow fields downstream of the steps. More than a thousand
pairs of particle images were acquired for each measurement. A two-step recursive
two-frame cross-correlation method was employed, resulting in velocity vector fields
with a grid spacing of 0.2 mm or h/32.
The vector fields underwent validation using statistical methods, achieving 97–99%
validity for velocity vectors. Low-pass filtering was applied to remove noise
associated with frequencies higher than the interrogation's sampling frequency. The
random error in velocity vectors was approximately 0.1 pixels, considered negligible
for turbulence statistics. Bias errors due to image pair loss and peak-locking effect
were minimized using the recursive interrogation method, estimating bias errors to be
about 1% of the full-scale velocity.

PIV measurements were also conducted on the upstream smooth-wall turbulent


boundary layer 50h ahead of both smooth and rough steps, and the turbulence
statistics of the measured upstream boundary layer were not altered by the presence of
the steps. The Reynolds number based on the momentum thickness of the
approaching boundary layer was Reh = 3130, with a ratio of upstream boundary layer
thickness (d) to mean step height (h) of d/h = 8.
3. Results and discussion
3.1. Mean flow structures
Fig. 3.Mean velocity fields over (a) the smooth BFS, (b) the rough BFS at position B1, and (c) the rough BFS at position B2. The flow
direction is from left to right.Every other velocity vectors are shown for clarity. Only those velocity vectors whose stream-wise velocities,
U , are less than 0.2 U are shown in order to illustrate the mean flow in the re circulation region. The more uniform flows above the shear
layer and upstream of the steps are represented by streamlines. The dashed lines mark the locations where U=0.158 Y. Wu et
al./International Journal of Heat and Fluid Flow 44 (2013) 155–169

Two-dimensional PIV measurements were employed to investigate the mean flow


characteristics over smooth and rough backward-facing steps (BFS) in a boundary
layer wind tunnel. The reattachment length (Xr), a crucial parameter for
characterizing BFS flow, was determined to be 4.25 ± 0.15h for the smooth BFS,
slightly lower than values reported in previous studies. The study also examined the
effects of surface roughness on the mean flow structures, revealing a significant
decrease in reattachment length to 3.3 ± 0.15h for the rough BFS at one position (B1),
and a slight increase to 4.4 ± 0.15h at another position (B2). The modifications in
mean flow structures downstream of the rough step were attributed to the
irregularities in the roughness topography and the three-dimensional nature of the
roughness.

Mean velocity profiles near the reattachment point were compared across various
studies by re normalizing the stream wise coordinate to eliminate differences in
velocity profiles caused by variations in reattachment lengths. Eaton and Johnston
(1981) observed excellent agreement in mean velocity profiles for turbulent
backward-facing step (BFS) flows with d/h up to 2, challenging the suggested criteria
by Bradshaw and Wong (1972). The current study, with a d/h value of 8,
demonstrated a deviation from previous findings, suggesting that the perturbation of
the step at this higher d/h may be considered strong. Despite this, the effects of
surface roughness were found to mitigate the step's perturbation of the boundary layer
in comparison to a smooth BFS flow.

The contour plot of mean stream wise velocity revealed thicker shear layers
immediately downstream of the rough step, and at one position, the reversed flow in
the re circulation region was weaker than in the smooth-step case. One-dimensional
mean stream wise velocity profiles showed significant effects of the roughness
topography upstream of the step, impacting the stream wise velocity at different
locations. The presence of surface roughness influenced the development of the
separated shear layer, resulting in differences in mean stream wise velocity profiles
between smooth- and rough-step BFS flows, particularly in the region with strong
mean shear and within the re-circulation region.
3.2. Reynolds stresses

The normalized stream wise and wall-normal Reynolds stresses in turbulent flows
over smooth and rough backward-facing steps (BFS). For the smooth BFS, the
stream-wise Reynolds stress distribution downstream of the step exhibited similarities
with other studies, with a plateau of relatively constant values observed in a specific
region. In the case of the rough BFS, distinct differences were noted, particularly in
the wider spreading of elevated stream-wise Reynolds stresses in the wall-normal
direction. The shear layers for the rough-step cases appeared to originate from
locations upstream of the step, possibly indicating earlier flow separation due to
adverse pressure gradients induced by the roughness. The study also highlighted the
significant reduction of wall-normal stress along the shear layer over the rough steps,
particularly at specific roughness positions, emphasizing the influence of surface
topography on turbulent characteristics in boundary layer flows.
The normalized wall-normal Reynolds stresses (hv02=U21i) downstream of a
smooth backward-facing step (BFS) and two rough BFS configurations (positions B1
and B2). The peak value of hv02 for the smooth step is observed approximately 1h
before reattachment at y/h ≈ 0.7, consistent with previous studies, while the rough-
step cases exhibit a significant reduction in wall-normal stress along the shear layer,
particularly at position B1. The normalized Reynolds shear stresses (hu0v0=U21i)
follow similar patterns, with the laminar-like flow region at the corner of the rough
step slightly smaller than that of the smooth BFS, and the peak values of shear stress
significantly reduced at position B1 for the rough-step case.
3.3. Quadrant analysis
Quadrant analysis is employed to investigate the impact of roughness topography on
the dominant contributors to Reynolds shear stress in the backward-facing step (BFS)
flow, decomposing the two-dimensional mean Reynolds shear stress at each grid point
into contributions from outward interactions (Q1 events), ejections (Q2 events),
inward interactions (Q3 events), and sweeps (Q4 events), excluding a hyperbolic hole
of size H.

Hole size H= 0.1 is used here. Of particular interest in this study are the Q2 (ejections)
and Q4 (sweeps) events which contribute the most to the Reynolds shear stress.

In Figure 15, the contributions from ejections (Q2) and sweeps (Q4) are analyzed for
both smooth and rough backward-facing step (BFS) configurations. The peak values
of Q2-contributions follow the general trend of mean Reynolds stress, dipping toward
the wall after the step for the smooth BFS but maintaining a constant wall-normal
location for Q4-contributions. In Figure 16, space fractions occupied by ejections and
sweeps reveal that the weak mean Reynolds shear stresses in the rough BFS flow are
attributed to equally weak contributions from Q2 and Q4 events, with lower
contributions observed at measurement position B1 compared to the smooth-step
case.
3.4. Mean span-wise vorticity
In Figure 17, the mean span-wise vorticity (xh/U1) is analyzed for both smooth and
rough backward-facing step (BFS) configurations. For the smooth step, a strip of
strong negative vorticity aligns with the shear layer separated from the step, gradually
weakening and expanding downstream. The introduction of roughness at positions B1
and B2 significantly weakens the vorticity, with B1 experiencing a more pronounced
reduction and altered spatial extent compared to B2, highlighting the distinctive
effects of roughness on the mean span-wise vorticity in the BFS flow.

4. Summary and conclusions


This study employs high spatial resolution PIV experiments in the x–y planes at two
different span-wise positions to investigate turbulent flows over smooth and rough
backward-facing steps (BFSs). Unlike past canonical backward-facing step flows, the
self-similar upstream turbulent boundary layer in this study flows over a block before
separation at the step, simulating non-equilibrium boundary layer conditions in
engineering flows. The rough step alters the mean flow structures, reducing the mean
reattachment length and influencing the primary and secondary re-circulation bubbles.
The roughness weakens turbulence in the separated shear layer, with effects more
pronounced at position B1 due to the downward-sloping roughness profile. Quadrant
analysis reveals a reduction in contributions to Reynolds shear stress from ejections
and sweeps due to the studied roughness, and the mean span-wise vorticity shows
significant alterations, eliminating concentrated negative vorticity and indicating early
separation of the shear layer with reduced magnitudes.

5. Reference
Allen, J.J., Shockling, M.A., Kunkel, G.J., Smits, A.J., 2007. Turbulent flow in
smoothand rough pipes. Philos. Trans. Roy. Soc. A 365, 699–714.
Avancha, R., Pletcher, R., 2002. Large eddy simulation of the turbulent flow past a
backward-facing step with heat transfer and property variations. Int. J. Heat
Fluid Flow 23, 601–614.
Barber, B., Jordan, E., Gell, M., Geary, A., 1999. Assessment of damage
accumulationin thermal barrier coatings using a fluorescent dye infiltration technique.
J.Therm. Spray Technol. 8, 79–86.
Bons, J., 2010. A review of surface roughness effects in gas turbines. J. Turbomach.
132, 021004.
Bons, J.P., Taylor, R.P., McClain, S.T., Rivir., R.B., 2001. The many faces of turbine
surface roughness. J. Turbomach. 123, 739–748.
Bons, J., Wammack, J., Crosby, J., Fletcher, D., Fletcher, T., 2008. Evolution of
surfacedeposits on a high-pressure turbine blade, Part I: Physical characteristics. J.
Turbomach. 130, 021020.
Bradshaw, P., Wong, F., 1972. The reattachment and relaxation of a turbulent shear
layer. J. Fluid Mech. 52, 113–135.
Christensen, K.T., 2004. The influence of peak-locking errors on turbulence statistics
computed from PIV ensembles. Exp. Fluids 36, 484–497.
Davis, J.R., 1999. Corrosion of Aluminum and Aluminum Alloys. ASM
International.,
Materials Park, OH.
Dejoan, A., Leschziner, M., 2004. Large eddy simulation of periodically perturbed
separated flow over a backward-facing step. Int. J. Heat Fluid Flow 25, 581–592.
Dubief, Y., Delcayre, F., 2000. On coherent-vortex identification in turbulence. J.
Turbul. 1, 1–22.
Eaton, J., Johnston, J., 1981. A review of research on subsonic turbulent flow
reattachment. AIAA J. 19, 1093–1100.
Hutchings, I.M., 1979. Mechanisms of the Erosion of Metals by Solid Particles.
American Soc. of Testing and Materials, West Conshohocken, PA.
Isomoto, K., Honami, S., 1989. The effect of inlet turbulence intensity on the
reattachment process over a backward-facing step. J. Fluids Eng. 111, 87–92.
Johnson, B., Christensen, K., 2009. Turbulent flow over low-order models of highly
irregular surface roughness. AIAA J. 47, 1288–1299.
Jovic, S., Driver, D., 1994. Backward-Facing Step Measurement at Low Reynolds
Number, Reh = 5000. Technical Report 108807. NASA Tech. Mem.
Kostas, J., Soria, J., Chong, M., 2002. Particle image velocimetry measurements of a
backward-facing step flow. Exp. Fluids 33, 838–853.
Le, H., Moin, P., Kim, J., 1997. Direct numerical simulation of turbulent flow over a
backward-facing step. J. Fluid Mech. 330, 349–374.
Licari, A., Christensen, K., 2011. Modeling cumulative surface damage and assessing
its impact on wall turbulence. AIAA J. 49, 2305–2320.
McClain, S., Collins, S., Hodge, B., Bons, J., 2006. The importance of the mean
elevation in predicting skin friction for flow over closely packed surface
roughness. J. Fluids Eng. 128, 579–586.
Mejia-Alvarez, R., Christensen, K., 2010. Low-order representations of irregular
surface roughness and their impact on a turbulent boundary layer. Phys. Fluids
22, 015106.
Neto, A., Grand, D., Metais, O., Lesieur, M., 1993. A numerical investigation of the
coherent vortices in turbulence behind a backward-facing step. J. Fluid Mech.
256, 1–25.

You might also like