You are on page 1of 44

Experimental Study of the Flow in an

External Gear Pump by Time


Resolved Particle Image Velocimetry

by
Nihal Ertrk

Supervised by:
Anton Vernet and Josep A. Ferr

A Thesis Submitted to
Graduate Programme in Chemical and Process Engineering
University of Rovira I Virgili
In the fulfillment of the Requirements for
The Degree of Master of Science in Chemical and Process Engineering

June, 2008, Tarragona Spain

Contents

1 Introduction

2 Objectives

3 Experimental Procedure
4
3.1 Experimental set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.2 Flow Seeding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
4 Techniques for TRPIV Image Analysis
8
4.1 Preliminary Image Processing . . . . . . . . . . . . . . . . . . . . . 8
4.2 Interrogation Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
4.3 Triple Image Correlation. . . . . . . . . . . . . . . . . . . . . . . . . 10
4.4 Boundary Treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
4.5 Conditional Average . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
5 Results and Discussion
13
5.1 Velocity fields and streamlines . . . . . . . . . . . . . . . . . . . . 13
5.2 Velocity profiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
6 Conclusions

21

7 Future Work

22

References

23

Appendix

25

List of Figures
1.1

Scheme of an external gear pump

3.1
3.2

Schematic drawing of the test bench


Examples of the experimental image series of the external gear pump (a)
suction chamber (b) impulse chamber
Rise velocities dependence on bubble radius
For several bubble diameters, the ratio of vertical deviation and test
section length in function of mean horizontal velocity. The limit value, H/L
0.03 has been indicated as straight horizontal line.

3.3
3.4

4.1
4.2
4.3

4.4

Removing reflections by median estimator across the time series. (a)


Original Image, (b) Image with Clean-Up Mask process
Cross-correlation procedure.
Triple Image Correlation scheme and example of correlation of peak
improvement. Right and left correlation planes have been multiplied to
obtain an enhanced peak
Representation of the selected image. (a) Original image. (b) Selected
image

4.5

Correlation of the image frames to define a Specific Position of the Gear

5.1

Velocity fields results which are obtain in different frequency rates (a)
Inlet with 500fps (b) Inlet with 1000fps (c) Inlet with 2000fps (d) Outlet
with 1000fps
Streamlines results which are obtain in different frequency rates (a) Inlet
with 500fps (b) Inlet with 1000fps (c) Inlet with 2000fps (d) Outlet with
1000fps
Velocity fields of suction chamber with 1000fps for different positions of
gear teeth
Streamlines of suction chamber with 1000 fps for different positions of
gear teeth
Inlet flow in the suction chamber at 1000fps (a) Mean v (b) Mean u (c)
Magnitude of mean v velocity contours (d) Magnitude of mean u velocity
contours.
Outlet flow in the impulse chamber at 1000fps (a) Mean v (b) Mean u (c)
Magnitude of mean v velocity contours (d) Magnitude of mean u velocity
contours.
Inlet flow in the suction chamber at 1000fps rms of velocity (a) v
component (b) u component.
Outlet flow in the impulse chamber at 1000fps rms of velocity (a) v
component (b) u component.

5.2

5.3
5.4
5.5

5.6

5.7
5.8

Acknowledgments
I would like to thank my supervisors, Anton Vernet and Josep A. Ferre for their
support and help. Thanks to Robert Castilla and Esteve Codina for their
collaboration and support in the laboratory experimentation in LABSON at UPC.
Also, thanks to everybody at ECCoMFiT group at URV. This study was
financially supported by the Spanish Ministry of Science and Education and
FEDER under projects DPI2006-02477 and DPI2006-14476.

Abstract
Time Resolved Particle Image Velocimetry (TRPIV) has been used to
investigate the turbulent flow in an external gear pump. The fluid movement
through the pump is maintained by the rotation of the gears that carries the fluid
from the intake side to the discharge side of the system. Small air bubbles have
been used as flow seeding to obtain the images. For the range of velocities
used in this study the buoyancy effects have been found negligible. The time
sequences of TRPIV recordings images have been processed using domestic
PIV software. The software uses the Local Field Correction which is able to
resolve the flow structures smaller than interrogation window. Processing the
images is done by the usual cross-correlation PIV proceeding based on FFT
algorithm. In order to improve the correlation peak detection, Triple Image
Correlation is used in place of the usual cross-correlation. In addition, a method
to improve the accuracy of TRPIV image analysis near boundaries has been
applied. A weighting function is used to the interrogation windows for the
correction to estimate the actual placement of the velocity vector when the
interrogation area overlaps the image boundary. All of these give to the
technique advantages in terms of accuracy and robustness. Instantaneous and
average fluid motions in the suction and in the impulse chamber of the pump
have been analyzed. Conditional averages in the suction and impulse chamber
around gears have been obtained using a correlation procedure to catch the
flow field at a fixed position of the gears. Time evolution of the average motion
shows that the direction of the velocity patterns changes as a function of the
movement of the gearwheel. The results obtained can help to understand the
effect of the flow field in the pump performance and its efficiency.

Chapter 1
Introduction
Internal flow in systems which consists of the rotating passages is exceedingly
complex, involving rotation and turbulence effects. The flow is interesting from a
fluid mechanical perspective as it is often influenced by rotor-stator interaction
mechanisms. A variety of measurement techniques have been applied to
several industrial machines in the struggle for accurate quantitative flow
descriptions. This means that methods have provided much fundamental
knowledge of the flow phenomena occurring in rotating machines [1,2,3].
However, the quest that maintains high efficiencies and performances at a
broader range of operating conditions raises the need for a more detailed
knowledge of the local and instantaneous features of the rotating passages
flow.
A gear pump is used for transferring and metering of liquids and power transfer
in a process. In this study, the flow phenomena of an external gear pump
(Figure 1.1) have been investigated on the increase of its efficiency and
performance. The fluid is transferred around the interior of the casing in the
pockets by the meshing of two gears rotating against each other to pump the
fluid from the suction side to the discharge (impulsion) side under pressure. As
the gears rotate, the spaces between the gears teeth transport the fluid at
constant amount of fluid per revolution.

INFLOW

OUTFLOW
Suction
Chamber

Impulse
Chamber

Figure 1.1 Scheme of an external gear pump.

External gear pumps are capable of working against high differential pressures.
The pressure in the outlet side is higher than the inlet side. Accordingly, the fluid
will try to find the path of least resistance and slip-back through the pump. To
prevent this phenomenon, a dynamic sealing must be implemented [4]. There
1

are clearances for the dynamic seal parts to move and these clearances permit
fluid to slip-back through the pump and reduce its theoretical efficiency. The
degree of internal slippage in a gear pump determines its volumetric efficiency
[1] which is the relation between actual pumped fluid flow rate, Q to the losses
of flow, QL due to the leakage or slip-back of the fluid around the gear and
casing (eqn. 1.1). The mean flow rate of the pump is the result of the volumetric
capacity, C v and the rotational velocity, (eqn 1.2).

v =

Q
Q + QL

(1.1)

Q=

C v
2

(1.2)

The volumetric efficiency has to be improved by minimizing the mechanical


tolerances of manufacturing [4].
Gear pumps can produce a high frequency pressure pulsation and thus
increase of fluctuations of delivery flow flow rate ripples in suction and
impulsion chambers, which tends to damage pressure gauges. To reduce the
ripples, tooth profile, gear shape and pump body plates are needed to be
improved. Investigations show that it is not possible to get external gear pumps
with no delivery fluctuation [5]. The efficiency of the pump is directly related with
the relationship between the moving parts and clearances factors. In addition,
the viscosity of the flow will effect for a thin fluid (such as water) or a moderately
viscous fluid (such as particular oil). Increasing the performance of an external
gear pump can be achieved by reducing the size of the pump, increasing the
pressure as well as the rotational velocity [6,7].
In the last decades, Digital Particle Image Velocimetry (DPIV) technique had
been developed and applied to various flow fields. To allow the Time Resolved
Particle Image Velocimetry (TRPIV) the images have to be captured using high
speed digital cameras which make possible to increase the time resolution.
DPIV needs tracing particles to follow the flow movement. In general these are
small solid or liquid particles that reflect the laser light. In the case of the
external gear pump analyzed here, small air bubbles have been used efficiently
as particle seeding since solid particles and water drops can seriously damage
the pump model. In order to show the potential of the TRPIV technique as an
efficient analysis tool in the design of industrial gear pumps, the main objective
of the present study is to provide detailed instantaneous and mean data of the
internal flow field.
Chapter 2 describes the purpose and
explains the experimental procedure and
on the methodology for analyzing the
presents the results and discussion,
averaged PIV velocity data followed by
work in Chapter 7.

objectives of this study. Chapter 3


PIV setup and Chapter 4 gives details
images by using TRPIV. Chapter 5
including instantaneous ensembleconclusions in Chapter 6 and future

Chapter 2
Objectives
The purpose of this paper is to clarify the role of the suction and impulse
chamber and analyze the flow occurring in it. In addition, these results can help
to decide modifications of the geometry of the pump in order to increase its
performance. For this purpose, the Time Resolved Particle Image Velocimetry
(TRPIV) has been applied to the analysis of the turbulent flow inside an external
gear pump. The TRPIV is a non-invasive technique and is a powerful instrument
for the analysis of complex instantaneous flow structures allowing the study of
fast changing systems.
In order to demonstrate the potential of the TRPIV technique as an efficient
analysis tool in the design of industrial gear pumps, the main objectives of the
present study for the technique are,

To provide detailed instantaneous data of the internal flow field in the


rotating passages of a pump gear by using air bubbles as flow seeding,
Implementation of local field correction to cross-correlation PIV proceeding
based on FFT algorithm,
Improving the analysis techniques to obtain more accurate results.

Chapter 3
Experimental Procedure
3.1 Experimental Set-up
The pump system analyzed is from the LABSON group of the Universitat
Politecnica de Catalunya (UPC). The pump is an external gear pump (Figure
1.1) where each cogwheel has a diameter of 54 mm and a height of 36 mm.
The number of teeth in each wheel is 11, the volumetric capacity of this model
is 44 cm3 /rev and the rotational velocity of the gear was 200 rpm. The cover of
test pump has been completely made of methacrylate in order to allow the
image acquisition.
The test bench (Figure 3.1) is composed by two hydraulic circuits. The upper
circuit is the primary or driven one, contains the test pump that takes the moving
fluid from the tank and impulses it through pressure fall back to the tank again.
The pump is driven by an oleohydraulic motor and it is a component of the
secondary circuit which is placed under the pump system. The motor is in turn
driven by a hydraulic power-pack. This scheme allows modifying very easily the
rotational velocity of the test pump acting on the flow rate of the driver circuit,
but has the disadvantage that it is not possible to select a certain velocity with
precision.

High velocity
Digital camera

Oil tank

Test pump

Computer

Oil tank
and
power

Laser generator

Laser sheet

Figure 3.1 Schematic drawing of the test bench.

The light source was a pulsed Monocrom Infrared laser with a wavelength of
800nm. A high velocity digital camera (Photron Ultima APX-RS) with resolution
of 10241072 pixel has been used. Digital images have been obtained with an
acquisition frequency of 500 fps, 1000 fps and 2000 fps. The buffer memory of
the digital camera allows to record up to 2048 images per experiment,
equivalent to 4, 2 or 0.5 seconds depending of the sampling rate used. All the
images obtained (Figure 3.2) were stored in a digital support for later
processing. The data and post processing was done using a domestic PIV
software developed by ECCoMFiT group of Universitat Rovira i Virgili (URV).

(a)

(b)

Figure 3.2 Examples of the experimental image series of the external gear pump (a) suction
chamber (b) impulse chamber

3.2 Flow seeding


Most PIV experiments have been reported to use small solid particles for flow
seeding. However, for this gear pump system, the use of solid materials can
produce material erosion and damage the transparent surface of methacrylate
and can also cause problems in the gear system because of metal-metal
contact between the teeth. The use of water drops as particle seeding could be
considered but it can produce problems of oxidation of the steel gears. Finally,
small air bubbles have been used in spite of some disadvantages: (i) the size of
the bubbles is not easily controllable and a large variability in the its size can
make difficulties to estimate the velocity lag [8], (ii) the density ratio is very large
and (iii) the presence of gas in a liquid can reduce the velocity of sound and
hence it can make the flow becoming compressible at relatively low velocity [9].
In the present case, the size of the bubbles is controlled by using pressurized
air flowing through a porous media that avoids the generation of large size
bubbles, the control of the air flow also allow to control the density of particles in
the measurement area. Drag and buoyancy forces associated with acceleration
are the main forces that act on bubbles for their motion in fluid than the force of
the fluid flowing. These forces can be optimized to allow bubbles to quickly
relocate to a desired area [10]. By combining the drag force and the buoyancy
force, Stokes Law given in eqn 3.1 can be formed based on gravity acceleration
(g), bubble radius (r) and kinematic fluid viscosity () to estimate the bubble rise
velocity ( v rise ).

v rise

2r g
=
9

(3.1)

From this equation, Figure 3.3 has been illustrated to show that the rise
velocities have a strong dependence on the bubble radius for the mineral oil (oil;
= 885 kg/m3, = 0.028 Pa.s) used in this study.
0.07

Rise velocity (vs ) [ m / s ]

0.06
0.05
0.04
0.03
0.02
0.01
0
0

0.2

0.4
0.6
Bubble Radius (r) [ mm ]

0.8

Figure 3.3 Rise velocities dependence on bubble radius.

If the flow has a horizontal mean velocity ( v y ) and when the particle reaches the
end of the test section, it has gone out off its path with an amount (eqn 3.2)

H = v rise

L
vy

(3.2)

where the length of the test section is ( L ), Using equations (3.1) and (3.2), the
ratio of vertical deviation and horizontal length of the test section can be defined
as in equation 3.3 in order to find the ratio and keep the bubbles in the laser
sheet.
2
H 2r g
=
L 9 vy

(3.3)

In the experiments, the laser sheet has a 1 mm thickness and the test section
has a length of 30 mm. In order to keep the bubbles in the laser sheet, H/L ratio
needs to be approximately 0.03. In Figure 3.4 the ratio of the vertical deviation
and the horizontal length of the test section is plotted against mean velocity for

various bubble diameters for the characteristics of the mineral oil used. The
mean velocity of the flow is function of the rotational speed of the pump. For
200 rpm, the mean velocity in the suction chamber has been obtained about
0.25 m/s and for this mean velocity the limit diameter size of the bubble is 0.7
mm. Then we can optimize the bubble size with a negligible value for the
particles move in vertical direction. It has been found the optimum diameter size
of the air bubbles 0.1 mm which is also supported by the analysis of Bolinder
[11].

10

H/L ratio

10

10

10

10

D = 0.1 mm
D = 0.2 mm
D = 0.3 mm
D = 0.4 mm
D = 0.5 mm
D = 0.7 mm
H/L = 0.03

-1

-2

-3

-4

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

mean vy [m/s]

Figure 3.4 For several bubble diameters, the ratio of vertical deviation and test section length in
function of mean horizontal velocity. The limit value, H/L 0.03 has been indicated as straight
horizontal line.

The effect of gas-liquid mixture on the sonic speed of the flow has also been
considered. A sufficiently high volume fraction of air can reduce the sonic speed
down to 20 m/s [9]. In the present case, the gas maintains its temperature
constant and the pressure of the pump system is quite low. When the size of
the interrogation area (64x64 pixel) and usual density of particles (suggested
around 10-15 particles per interrogation area [8] are used for low velocities, the
flow shows reasonably far away from compressibility characteristics. In the lest
desirable situation which is the sonic speed is approximately 20 m/s, the
rotational speed of the pump should be around 1000 rpm in order to have a
Mach number. In the present configuration of the experimental setup, the
rotational velocity of the gear was working at 200 rpm.

Chapter 4
Techniques for TRPIV Image Analysis
Specific aspects of the time resolved PIV technique have been applied to
analyze the turbulent flow in the external gear pump. The aim is to take
advantages for the use of time resolved PIV series of images to overcome
some issues that can effect PIV measurements and to improve the performance
and the accuracy of the technique.

4.1 Preliminary Image Processing


First consideration on the processing of time series of PIV images can be
introduced apart from the characteristics of the flow that is being measured.
Since the time history of the illumination at each image location is available,
statistical properties of the series can be analyzed.
Hence, a Clean-up Mask process has been applied to improve on the
processing of time series of the experimental images. This procedure allows
removing and/or reducing the spurious permanent reflections of the light from
the illumination process of the laser. The median value across the image time
series is estimated to clean these reflections from the original images. Figure
4.1a presents an outlet region of the field of view for a single instantaneous
image, while Fig 4.1b displays the differences between the original image and
the median image from a time series of 400 images. The median value of the
illumination at each point provides information that affects the detection of the
actual displacement of the particles [16].

50

50

100

100

150

150

200

200

250

250

300

300

100

200

300

400

500

600

700

100

(a)

200

300

400

500

600

700

(b)

Figure 4.1 Removing reflections by median estimator across the time series. (a) Original Image,
(b) Image with Clean-Up Mask process

4.2 Interrogation Area


Instantaneous images have been analyzed using local field correction (LFC)
[12] and TRPIV. LFC is a correlation PIV method able to accurately resolve flow
structures smaller than interrogation window [13]. The technique used here is a
cross-correlation method that provides a remarkable capability for accurately
resolving small scale structures in the flow. Typical dimensions of an
interrogation area are given in the literature for PIV between 16x16 to 128x128
pixels. In order to obtain a reliable estimator of the particle image displacement,
about 10 to 15 particles in an interrogation area have to be present [8]. In the
present work, 64x64 pixels interrogation area has been used by considering the
adequate particles intensity in each interrogation area. In order to get an
estimated displacement, the usual cross-correlation PIV processing is
performed for each interrogation area. Figure 4.2 illustrates the digital PIV
process. The cross-correlation shifts the second window across the first and
sums the matching values (eqn 4.1) [8].

RII ( x, y ) =

i = K

j = L

I (i, j ) I ' (i + x, j + y )

(4.1)

At the point where images match best, the correlation is at its peak value. This
peak is located and provides the best estimate for the displacement of the
particles in the window.

cross-correlation
t

I
Image 1

peak search
v

t+t
I
Image 2

Figure 4.2 Cross-correlation procedure.

To calculate the cross-correlation between two corresponding interrogation


windows from successive images, fast-Fourier transforms (FFTs) are used.
Digital recording and computer analysis led to the application of a FFT in PIV
image processing, which significantly decreased the time required for the
necessary operations to produce a velocity measurement [13].

4.3 Triple Image Correlation


An image can be paired with the next or previous image in the time series. Thus,
a correlation algorithm involving the three images should prove more robust to
out of plane motion than the usual single pair correlation algorithm. A similar
approach was proposed in another background [14,15]. The algorithm used
here implements this strategy by multiplying both correlation planes in order to
improve the peak detection [16]. This leads to the reduction of the spurious
correlation peaks appearing in only one of the correlation planes.
Triple Image Correlation scheme is illustrated in Figure 4.3 on behalf of an
example. The interrogation area has been selected from the time series where
the gear movement is appearing. The interrogation area is coordinated as 345
pixel horizontal and 81 pixel vertical of the images. Two correlation planes are
obtained from the image pairs (ti-1, ti) and (ti, ti+1) for the image corresponding to
time ti. After that, these correlation planes are combined into the correlation
plane and this plane verifies an enhanced correlation peak for the case of
appearing one tooth one of the interrogation windows. The spurious peaks that
appear in only one of the two correlation peaks can lead to the erroneous
estimation of the displacement of the particles. But as shown in the Figure 4.3,
those spurious peaks have been cancelled by this algorithm. The peak obtained
from the triple image correlation is more obvious than the corresponding peaks
on the standard cross correlation planes.

ti

ti+1
10

10

20

20

30

30

40

40

50

50

60

60

10

20

30

40

50

60

ti-1
10

20

30

40

50

60

10

20

30

40

50

60

10

20

30

40

50

60

Figure 4.3 Triple Image Correlation scheme and example of correlation of peak improvement.
Right and left correlation planes have been multiplied to obtain an enhanced peak.

10

4.4 Boundary Treatment


Since, iterative standard algorithms introduce significant errors when the
interrogation location is closer to the image boundary, a special treatment of the
interrogation area near the image boundaries has been introduced to obtain the
same level of accuracy available at inner locations [16]. The boundary treatment
is applied to the images with weighting function which is responsible computing
the corrected position of the velocity displacement relative to the boundary. A
weighting function is needed to avoid instabilities in the iterative process of
compensation of the particle pattern or changing the frequency response of a
moving average [17]. The use distorts the grey level of the original images,
introducing error. The LFC-TRPIV method works accurately even if it has this
error built in, but with double weighting, which consists of averaging the result
that calculated by its symmetrical counterpart, the error can be reduced [18].

4.5 Conditional Average


The flow structure in the inlet/outlet chamber depends on the position of the
gear. Thus, a full time mean will give non real flow structures in these
chambers. A conditional mean based on the location of the gear is obtained.
The gears are continuously rotating in a specific time interval. Analyzing this
specific area, we have introduced a conditional average function which provides
the gears a stable position in a specific time interval with velocity field of the
flow at this time. The image series in time are correlated with a selected image
(Figure 4.4b) which consists of one tooth of the gear to define the specific
position of the gear. This allows locating the instantaneous fields that has the
gear in the selected position. Those instantaneous velocity vectors are
averaged to obtain a conditional velocity field representing the characteristic
flow structure at this gear position (Figure 4.4).

100

20

40

200

60

80

100

300
120

140

160

180
20

400

40

60

80

100

120

140

160

180

(b)
500

600

50

100

150

200

250

300

350

400

450

500

550

(a)
Figure 4.4 Representation of the selected image. (a) Original image. (b) Selected image.

11

72 images with the same position which has been shown in the Figure 4.5 as
the peaks of crossing lines of the correlation plane have been found for the
suction chamber and they have been used for the analyzing of the rotating
parts. The same process has been applied to the suction chamber. The
conclusion that can be drawn is that the turbulence effect of the rotating gears
to the system has been investigated by the conditional average function with a
reliable velocity field and accurate results.

Figure 4.5 Correlation of the image frames to define a Specific Position of the Gear.

12

Chapter 5
Results and Discussion
5.1 Velocity fields and streamlines
Figures 5.1 and 5.2 show the conditional averages obtained for the inlet
chamber at different sampling rate (Figures 5.1a-5.2a, 5.1b-5.2b and 5.1c-5.2c
at 500 fps, 1000 fps and 2000fps respectively) and for the outlet chamber
(Figures 5.1d-5.2d) at 1000 fps. All the inlet measurements were taken in a
horizontal (x-y) plane at vertical location coincident with the flow entrance, while
the outlet measurements were taken only in a horizontal plane in the upper
plane of the chamber. Hence, it is not possible to see in Figure 5.1d that the
vectors are leaving from the chamber.
For the suction chamber, it could be seen that the fluid flows through the gears
from the two sides of the chamber symmetrically and produces two vortices on
the right and left side of the chamber. The small vortices also appear in the end
points of gear teeth. For the impulse chamber is clearer to observe the flow with
two vortices which are closer to the center side of the pipe and small vortices
are not obtained in the end points of gear teeth. It is clear that the flow in the
inlet chamber is more complex that the flow at the outlet.

13

INFLOW

u velocity
x - direction

Suction
Chamber

v velocity
y - direction

Impulse
Chamber

OUTFLOW
x [mm]

x [mm]

y
[mm]

y
[pixel]

y
[pxel]

y
[mm]

x [pxel]

x [pixel]

(b)

(a)
x [mm]

x [mm]

y
[pixel]

y
[pixel]

y
[mm]

y
[mm]

x [pixel]

x [pixel]

(c)

(d)

Figure 5.1 Velocity fields results which are obtain in different frequency rates (a) Inlet with
500fps (b) Inlet with 1000fps (c) Inlet with 2000fps (d) Outlet with 1000fps

14

x [mm]

x [mm]

y
[mm]

y
[pixel]

x [pixel]

y
[pxel]

y
[mm]

x [pxel]

(a)

(b)

x [mm]

x [mm]

y
[pixel]

y
[pxel]

y
[mm]

y
[mm]

x [pixel]

x [pixel]

(d)

(c)

Figure 5.2 Streamlines results which are obtain in different frequency rates (a) Inlet with 500fps
(b) Inlet with 1000fps (c) Inlet with 2000fps (d) Outlet with 1000fps

It has been shown that the sampling rate do not have an important effect on the
flow structure obtained, at least at the rotation frequency used here. To find the
flow evolution inside the suction chamber the instantaneous data obtained for a
sampling rate of 1000 fps has been used since it gives a rather better resolution
the other two cases (500 fps and 2000 fps). Figure 5.3 and Figure 5.4 show the
velocity vectors and streamlines at six consecutive times which corresponds to
different position of the gear teeth. It can be observed that the centre of the
large vortices do not change their position with the rotation of gear, while the
small vortices could appear, disappear or join to large ones.

15

ti

ti+3

ti+6

ti+9

ti+12

ti+15

Figure 5.3 Velocity fields of suction chamber with 1000fps for different positions of gear teeth.

16

ti

ti+3

ti+6

ti+9

ti+12

ti+15

Figure 5.4 Streamlines of suction chamber with 1000 fps for different positions of gear teeth.

17

5.2 Velocity profiles


The suction chamber velocity profiles (Figure 5.5) show that the v velocity
component is considerably increasing when the fluid is flowing through the gear
teeth and the maximum negative velocity of mean v has been found in the
center of the inlet side of the gear pump. The mean u velocity component
reaches its maximum on the right middle and on the left middle side of the gear
pump. Figure 5.5 shows that the profiles become less symmetric as they move
away from the inlet section. This lack of symmetry could be generated by the
model performance and needs a more detailed analysis with a different rotation
velocity and image acquisition at more than one horizontal plane.

Mean v vs x

Mean u vs x

100
50
0

y=3.48
y=5.4
y=7.5
y=9.2
y=10.8
y=12.9
y=15.4

300

200
u [ mm / s ]

-50
v [ mm / s ]

400

y=3.48
y=5.4
y=7.5
y=9.2
y=10.8
y=12.9
y=15.4

-100
-150

100

-200

-250
-100
-300
-350
0

10

15
x [ mm ]

20

25

-200
0

30

10

(a)

Magnitude of mean v velocity [ mm / s ]

y [ mm ]

y [ mm ]

-100
-150

-200

-250

-300

25

(c)

0
-100

0
0

30

100

10

20

200

12

-50

10

300

14

12

15
x [ mm ]

30

Magnitude of mean u velocity [ mm / s ]

14

10

25

16

50

20

(b)

16

0
0

15
x [ mm ]

-200
5

10

15
x [ mm ]

20

25

(d)

Figure 5.5 Inlet flow in the suction chamber at 1000fps (a) Mean v (b) Mean u (c) Magnitude of
mean v velocity contours (d) Magnitude of mean u velocity contours.

18

30

Mean v vs x

Mean u vs x

100
50

-50
-100

y=1.8
y=3.7
y=6.0
y=7.9
y=9.8
y=11.7
y=13.7

80
60
40
u [ mm / s ]

0
v [ mm / s ]

100

y=1.8
y=3.7
y=6.0
y=7.9
y=9.8
y=11.7
y=13.7

20
0
-20

-150

-40
-200

-60

-250
0

10

15
x [ mm ]

20

25

-80
0

30

10

15
x [ mm ]

20

25

30

(b)

(a)

Magnitude of mean u velocity [ mm / s ]

Magnitude of mean v velocity [ mm / s ]


50

15

80

15

60

0
-50
-100
5

y [ mm ]

y [ mm ]

40
10

10

20
0
-20

-150

-40
-200
0
0

10

15
x [ mm ]

20

25

-60
0
0

30

(c)

10

15
x [ mm ]

20

25

30

(d)

Figure 5.6 Outlet flow in the impulse chamber at 1000fps (a) Mean v (b) Mean u (c) Magnitude
of mean v velocity contours (d) Magnitude of mean u velocity contours.

Velocity profiles at the impulse chamber (Figure 5.6), shows that the v velocity
component is increasing when the fluid is passing through the gear teeth and
the maximum negative velocity of mean v has been found in the center of the
outlet side of the gear pump. The mean u velocity component reaches to
maximum on the right middle and on the left middle side of the gear pump.
Results show that the flow in the suction chamber is much more complex than
the flow in the impulse chamber. Therefore the inlet chamber is the one that
needs more detailed an extended study.
Figures 5.7 and 5.8 show the results of rms of velocities (u and v components)
of the flow at suction and impulse chamber of the external gear pump system. It
is essentially to observe that rms value is increasing when the fluid is close to
the rotating gears and rms value reaches its maximum at the gear zone.

19

rms of velocity (v component) versus x

rms of velocity (u component) versus x


y=3.48
y=5.4
y=7.5
y=9.2
y=10.8
y=12.9
y=15.4

3.5

rms of v [ mm / s ]

3
2.5
2
1.5

3
2.5
2
1.5

0.5

0.5

0
0

10

15
x [ mm ]

20

25

y=3.48
y=5.4
y=7.5
y=9.2
y=10.8
y=12.9
y=15.4

3.5

rms of u [ mm / s ]

0
0

30

10

15
x [ mm ]

20

25

30

Figure 5.7 Inlet flow in the suction chamber at 1000fps rms of velocity (a) v component (b) u
component.

rms of velocity (v component) versus x

rms of velocity (u component) versus x


y=1.8
y=3.7
y=6.0
y=7.9
y=9.8
y=11.7
y=13.7

1.2

rms of v [ mm / s ]

1
0.8
0.6

0.7

0.5
0.4
0.3

0.4

0.2

0.2

0.1

0
0

10

15
x [ mm ]

20

25

0
0

30

y=1.8
y=3.7
y=6.0
y=7.9
y=9.8
y=11.7
y=13.7

0.6

rms of u [ mm / s ]

1.4

10

15
x [ mm ]

20

25

30

Figure 5.8 Outlet flow in the impulse chamber at 1000fps rms of velocity (a) v component (b) u
component.

20

Chapter 6
Conclusions
The use of air bubbles as tracing particles for PIV has been proved as a fine
alternative to the use of solid or liquid particles. TRPIV has been applied to the
study of the flow structures in the suction and impulse chamber of an external
gear pump. Results show the possibility that the analysis technique presented
can be used to obtain detailed information of the instantaneous velocity fields, in
systems with moving elements, which are not part of the fluid flow. The ability to
separate particles from the reflection and to clean/remove the spots allows
improving the peak detection for the direction of the velocity. A simple triple
image correlation algorithm can improve the peak correlation in rather moving
parts appear in the image. The technique for boundary treatment developed by
Usera et.al [15] has been applied with the use of weighting function to obtain
the same level of accuracy available at inner locations of the system. Corrected
positions of the velocity displacement relative to the boundary have been
computed. A conditional average velocity field has been obtained for specific
gear position allowing an average time evolution of the flow structures in the
suction chamber. The results obtained show that a detailed analysis of the
suction chamber is needed for a better understanding of the dynamic behavior
of the flow. The results of this study will be presented in 14th International
Symposium on Applications of Laser Techniques to Fluid Mechanics in
Portugal, July, 2008 and will be published in the proceedings of the conference
(see Appendix).

21

Chapter 7
Future Work
The aim of the work that has been presented here was the possibility of using
air bubbles to investigate with TRPIV the flow inside the suction chamber and
impulse chamber of an external gear pump. For future work, it is been decided
to study with different rotational velocities of the gear to increase the efficiency
and performance of the system.
In addition to this, experimental study of cavity flow analysis in channel will be
investigated by simultaneously measuring the velocity, the temperature and
concentration fields with a combined Time-Resolved Particle Image Velocimetry
(TRPIV) and Planar Laser-Induced Fluorescence (PLIF) system. PLIF
technique is an optical measurement tool which will provide to have quantitative
information that can be obtained on heat transfer phenomena and temperature
distribution of the fluid in the cavity channel. In addition, PLIF system can be
also used to obtain whole-field concentration data for mixing performance in this
channel.

22

References
[1] Wernet M. P., (2000) Application of DPIV to study both steady state and
transient turbomachinery flows, Optics & Laser Technology 32, 497-525.
[2] Pedersen N., Larsen P.S., Jacobsen C. B., (2003) Flow in a Centrifugal
Pump Impeller at Design and Off-design Conditions. Part 1: PIV and LDV
measurements, Journal of Fluids Engineering, vol: 125, pages: 61-72.
[3] Day S.W., McDaniel J.C., (2005) PIV Measurements of Flow in a Centrifugal
Blood Pump: Steady Flow Journal of Biomechanical Engineering, Volume 127,
Issue 2, pp. 244-253.
[4] Dearn R, B.Sc.(Hons), (June 2001) European Marketing Manager, The fine
art of gear pump selection and operation, World pumps, Volume 2001, Issue
417, pp 38-40.
[5] Iyoi H and Ishimura S (1983) -Theory in gear geometry, Transaction of
ASME Journal of Mechanisms, Transmissions, and Automation in Design 105,
pp 286290.
[6] Codina E, Kamashta M, (1999) ECOPUMP project, Enhanced design of high
pressure gear pumps using environmentally acceptable hydraulic fluids, BRITE
Contract n BRPRCt95-0094, Tech. rep., LABSON-UPC.
[7] Castilla R, Gamez-Montero P, Huguet D, Codina E, (2007) Turbulence in
Internal Flows in Minihydraulic Components, CIMNE, pp 241-251.
[8] Raffel M, Willert C, Kompenhans J (1998) Particle Image Velocimetry: A
Practical guide, Springer.
[9] Brennen C E (2005) Fundamentals of Multiphase Flow, Cambridge
University Press.
[10] Moore J (2007) Dry sump pump bubble elimination for hydraulic hybrid
vehicle systems, Master thesis in the department of Mechanical engineering,
The University of Michigan.
[11] Bolinder J (1999) On the accuracy of a digital particle image velocimetry
system, Tech. rep., Lund Institute of Technology.
[12] Usera G, Vernet A, Ferr JA (2004) Consideration and Improvements of
the Analysis Algorithms Used for Time Resolved PIV of Wall bounded Flows,
12th International Symposium on Applications of Laser Techniques to Fluid

23

Mechanics, Lisbon.
[13] Nogueria J, Lecuona A, and Rodriguez PA, (1997) Data validation, false
vectors correction and derived magnitudes calculations on PIV data, Meas. Sci.
Technol. (8), 1493-501.
[14] Willert CE and Gharib M, (1991) Digital particle image velocimetry, Exp.
Fluids (10), 181-193.
[15] Hart DP (1998) The Elimination of Correlation Errors in PIV Processing, 9th
International Symposium on Applications of Laser Techniques to Fluid
Mechanics, Lisbon.
[16] Hart DP (1999) Super-Resolution PIV by Recursive Local-Correlation,
Journal of Visualization (10).
[17] Nogueria J, Lecuona A, and Rodriguez PA (2001) Local field correction
PIV, implemented by means of simple algorithms, and multigrid versions, Meas.
Sci. Technol, 12, 1911-1921.
[18] Nogueria J, Lecuona A and Rodriguez PA (2002) Accuracy and time
performance of different schemes of the local field correction PIV technique,
(33), 743-751.

24

APPENDIX

25

14th Int Symp on Applications of Laser Techniques to Fluid Mechanics


Lisbon, Portugal, 07-10 July, 2008
1037

Analysis of the Turbulent Flow of an External Gear Pump by Time Resolved


Particle Image Velocimetry
Nihal Ertrk1, Anton Vernet1, Josep A. Ferr1, Robert Castilla2, Esteve Codina2

1: Department of Mechanical Engineering, University of Rovira I Virgili, Tarragona, Spain, nihal.erturk@urv.cat,


anton.vernet@urv.cat, josep.a.ferre@urv.cat
2: Department of Fluid Mechanics, Technical University of Catalonia, Terassa, Spain,
castilla@mf.upc.edu, ecodina@mf.upc.edu

Abstract
Time Resolved Particle Image Velocimetry (TRPIV) has been used to investigate the turbulent flow in an
external gear pump. The fluid movement through the pump is maintained by the rotation of the gears that
carries the fluid from the intake side to the discharge side of the system. Small air bubbles have been used as
flow seeding to obtain the images. For the range of velocities used in this study the buoyancy effects have
been found negligible. The time sequences of TRPIV recordings images have been processed using domestic
PIV software. The software uses the Local Field Correction which is able to resolve the flow structures
smaller than interrogation window. Processing the images is done by the usual cross-correlation PIV
proceeding based on FFT algorithm. In order to improve the correlation peak detection, Triple Image
Correlation is used in place of the usual cross-correlation. In addition, a method to improve the accuracy of
TRPIV image analysis near boundaries has been applied. A weighting function is used to the interrogation
windows for the correction to estimate the actual placement of the velocity vector when the interrogation
area overlaps the image boundary. All of these give to the technique advantages in terms of accuracy and
robustness. Instantaneous and average fluid motions in the suction and in the impulse chamber of the pump
have been analyzed. Conditional averages in the suction and impulse chamber around gears have been
obtained using a correlation procedure to catch the flow field at a fixed position of the gears. Time evolution
of the average motion shows that the direction of the velocity patterns changes as a function of the
movement of the gearwheel. The results obtained can help to understand the effect of the flow field in the
pump performance and its efficiency.

1. Introduction
The internal flow that develops in a system which consists of the rotating passages is exceedingly
complex, involving streamline curvature, rotation and turbulence effects. The flow is interesting
from a fluid mechanical perspective as it is often influenced by rotor-stator interaction mechanisms.
A variety of measurement techniques have been applied to several industrial machines in the
struggle for accurate quantitative flow descriptions which mean that methods have provided much
fundamental knowledge of the flow phenomena occurring in rotating machines. However, the quest
that maintains high efficiencies and performances at a broader range of operating conditions raises
the need for a more detailed knowledge of the local and instantaneous features of the rotating
passages flow. A gear pump is used for transferring and metering of liquids and power transfer in a
process. In this study, the flow phenomena of an external gear pump (Fig 1) have been investigated
on the increase of its efficiency and performance. The fluid is transferred around the interior of the
casing in the pockets by the meshing of two gears rotating against each other to pump the fluid from
the suction side to the discharge (impulsion) side under pressure. As the gears rotate, the spaces
between the gears teeth transport the fluid at constant amount of fluid per revolution.

-1-

14th Int Symp on Applications of Laser Techniques to Fluid Mechanics


Lisbon, Portugal, 07-10 July, 2008
1037

v velocity
y - direction

INFLOW

OUTFLOW
Suction
Chamber

u velocity
x - direction

Impulse
Chamber

Fig 1. Scheme of an external gear pump.


The mean flow rate of the pump is the result of the volumetric capacity and the rotational velocity.
The volumetric efficiency has to be improved by minimizing the mechanical tolerances of
manufacturing (Dearn 2001). Gear pumps can produce a high frequency pressure pulsation and thus
increase of fluctuations of delivery flow flow rate ripples in suction and impulsion chambers,
which tends to damage pressure gauges. To reduce the ripples, tooth profile, gear shape and pump
body plates are needed to be improved. Investigations show that it is not possible to get external
gear pumps with no delivery fluctuation (Iyoi and Ishimura, 1983). The efficiency of the pump is
directly related with the relationship between the moving parts and clearances factors. Increasing
the performance of an external gear pump can be achieved by reducing the size of the pump,
increasing the pressure as well as the rotational velocity (Codina and Kamashata, 1999, and Castilla
et al, 2007).
The purpose of this paper is to clarify the role of the suction chamber and analyze the flow
occurring in it. In addition, these results can help to decide modifications of the geometry of the
pump in order to increase its performance. For this purpose, the use of a Time Resolved Particle
Image Velocimetry (TRPIV) has been applied into the analysis of the turbulent flow inside an
external gear pump. The TRPIV is a non-invasive technique and is a powerful instrument for the
analysis of complex instantaneous flow structures allowing the study of fast changing systems.
In the last decades, Digital Particle Image Velocimetry (DPIV) technique had been developed and
applied to various flow fields. To allow the TRPIV the images have to be captured using high speed
digital cameras which make possible to increase the time resolution. DPIV needs tracing particles to
follow the flow movement. In general these are small solid or liquid particles that reflect the laser
light. In the case of the external gear pump analyzed here, small air bubbles have been used
efficiently as particle seeding since solid particles and water drops can seriously damage the pump
model. In order to show the potential of the TRPIV technique as an efficient analysis tool in the
design of industrial gear pumps, the main objective of the present study is to provide detailed
instantaneous and mean data of the internal flow field.

2. Experimental Procedure
The pump system analyzed is from the LABSON group of the Universitat Politecnica de Catalunya
(UPC). Each cogwheel has a diameter of 54 mm and a height of 36 mm. The number of teeth in
each wheel is 11, the volumetric capacity of this model is 44cm3/rev and the rotational velocity of
the gear was 200 rpm. The cover of test pump has been completely made of methacrylate in order to
allow the image acquisition. The test bench (Fig 2) is composed by two hydraulic circuits. The
upper circuit is the primary or driven one, contains the test pump that takes the moving fluid (oil;
=885 kg/m3, =0.028 Pa.s) from the tank and impulses it through pressure fall back to the tank
again. The pump is driven by an oleohydraulic motor, which is a component of the secondary
circuit which is placed under the pump system.

-2-

14th Int Symp on Applications of Laser Techniques to Fluid Mechanics


Lisbon, Portugal, 07-10 July, 2008
1037

Oil tank

High velocity
Digital camera

Test

Computer

Oil tank and


power pack

Laser sheet
Laser generator

Fig 2. Schematic drawing of the test bench.


The light source was a pulsed Monocrom Infrared laser (800nm). A high velocity digital camera
(Photron Ultima APX-RS) with resolution of 10241024 pixel has been used. Digital images have
been obtained with an acquisition frequency of 500 fps, 1000 fps and 2000 fps. The buffer memory
of the digital camera allows to record up to 2048 images per experiment, equivalent to 4, 2 or 0.5
seconds depending of the sampling rate used. All the images obtained were stored in a digital
support for later processing. The data and post processing was done using a domestic PIV software
developed by ECCoMFiT group of Universitat Rovira i Virgili (URV).
Most PIV experiments have been reported to use small solid particles for flow seeding. However,
for this gear pump system, the use of solid materials can produce material erosion and damage the
transparent surface of methacrylate and also problems in the gear system because of metal-metal
contact between the teeth. The use of water drops as particle seeding could be considered but it can
produce problems of oxidation of the steel gears. Finally, small air bubbles have been used in spite
of some disadvantages: (i) the size of the bubbles is not easily controllable and a large variability in
the its size can make difficulties to estimate the velocity lag (Raffel et al, 1998), (ii) the density ratio
is very large and (iii) the presence of gas in a liquid can reduce the velocity of sound and hence it
can make the flow becoming compressible at relatively low velocity (Brennen, 2005). In the present
case, the size of the bubbles is controlled by using pressurized air flowing through a porous media
that avoids the generation of large size bubble, the control of the air flow also allow to control the
density of particles in the measurement area. Drag and buoyancy forces associated with acceleration
are the main forces that act on bubbles for their motion in fluid than the force of the fluid flowing.
These forces can be optimized to allow bubbles to quickly relocate to a desired area (Moore, 2007).
By combining the drag force and the buoyancy force, Stokes Law given in Equation 1 can be
formed based on gravity acceleration (g), bubble radius (r) and kinematic fluid viscosity () to
estimate the bubble rise velocity ( v rise ).
2

v rise

2r g
=
9

(1)

v
If the flow has a horizontal mean velocity ( y ) and when the particle reaches the end of the test
section, it has gone out off its path with an amount
L
H = v rise
(2)
vy
where the length of the test section is ( L ), Using equations (1) and (2), the ratio of vertical
-3-

14th Int Symp on Applications of Laser Techniques to Fluid Mechanics


Lisbon, Portugal, 07-10 July, 2008
1037

deviation and horizontal length of the test section can be defined as in equation 3 in order to find the
ratio and keep the bubbles in the laser sheet.
2
H 2r g
=
(3)
L 9 vy
In the experiments, the laser sheet has a 1 mm thickness and the test section has a length of 30 mm.
In order to keep the bubbles in the laser sheet, H/L ratio needs to be 0.025. The mean velocity of the
flow is function of the rotational speed of the pump. Then we can optimize the bubble size with a
negligible value for the particles move in vertical direction. It has been found the optimum diameter
size of the air bubbles 100 m which is also supported by the analysis of Bolinder 1999.
The effect of gas-liquid mixture on the sonic speed of the flow has also been considered. A
sufficiently high volume fraction of air can reduce the sonic speed down to 20 m/s (Brennen 2005).
In the present case, the gas maintains its temperature constant and the pressure of the pump system
is quite low. When the size of the interrogation area (64x64 pixel) and usual density of particles
(suggested around 10-15 particles per interrogation area (Raffel et al, 1998) are used for low
velocities, the flow shows reasonably far away from compressibility characteristics. In the lest
desirable situation which is the sonic speed is approximately 20 m/s, the rotational speed of the
pump should be around 1000 rpm in order to have a Mach number. In the present configuration of
the experimental setup, the rotational velocity of the gear was working at 200 rpm.

3. Techniques For TRPIV Image Analysis


Instantaneous images were analyzed using local field correction (LFC) (Nogueria et al, 1997) and
TRPIV. LFC is a correlation PIV method able to accurately resolve flow structures smaller than
interrogation window (Willert and Gharib, 1991). The technique used here is a cross-correlation
method that provides a remarkable capability for accurately resolving small scale structures in the
flow. Typical dimensions of an interrogation area are given in the literature for PIV between 16x16
to 128x128 pixels. In order to obtain a reliable estimator of the particle image displacement, about
10 to 15 particles in an interrogation area have to be present (Raffel et al, 1998). In the present
work, we have used 64x64 pixels for the interrogation area by considering the adequate particles
intensity in each interrogation area.
50

50

100

100

150

150

200

200

250

250

300

300

100

200

300

400

500

600

700

100

200

300

400

500

600

700

(a)
(b)
Fig 3. Removing reflections by median estimator across the time series. (a) Original Image, (b)
Image with Clean-Up Mask process
An improvement on the processing of time series of the experimental images is the use of a Cleanup Mask process to remove and/or reduce the spurious permanent reflections of the light from the
illumination process of the laser. The median value across the image time series is estimated to
clean these reflections from the original images. Fig 3a presents an outlet region of the field of view
for a single instantaneous image, while Fig 3b displays the differences between the original image
-4-

14th Int Symp on Applications of Laser Techniques to Fluid Mechanics


Lisbon, Portugal, 07-10 July, 2008
1037

and the median image from a time series of 400 images. The median value of the illumination at
each point provides information that adversely affects the detection of the actual displacement of
the particles; they tend to lock the correlation to null displacement.
In order to get an estimated displacement, the usual cross-correlation PIV processing is performed
for each interrogation area. To calculate the cross-correlation between two corresponding
interrogation windows from successive images, fast-Fourier transforms (FFTs) are used. Digital
recording and computer analysis led to the application of a FFT in PIV image processing, which
significantly decreased the time required for the necessary operations to produce a velocity
measurement (Willert and Gharib, 1991). An image can be paired in principle with the next or
previous image in the time series. Thus, a correlation algorithm involving the three images should
prove more robust to out of plane motion than the usual single pair correlation algorithm. A similar
approach was proposed in another background (Hart 1998 and Hart 1999). The algorithm used here
implements this strategy by multiplying both correlation planes in order to improve the peak
detection (Usera et al, 2004). This leads to the reduction of the spurious correlation peaks appearing
in only one of the correlation planes. Since, iterative standard algorithms introduce significant
errors when the interrogation location is closer to the image boundary, a special treatment of the
interrogation area near the image boundaries has been introduced to obtain the same level of
accuracy available at inner locations (Usera et al, 2004). The boundary treatment is applied to the
images with weighting function which is responsible computing the corrected position of the
velocity displacement relative to the boundary. A weighting function is needed to avoid instabilities
in the iterative process of compensation of the particle pattern or changing the frequency response
of a moving average (Nogueria et al, 2001 and Nogueria et al, 2002).

4. Results
The flow structure in the inlet/outlet chamber depends on the position of the gear. Thus, a full time
mean will give non real flow structures in these chambers. A conditional mean based on the
location of the gear is obtained. The gears are continuously rotating in a specific time interval.
Analyzing this specific area, we have introduced a conditional average function which provides the
gears a stable position in a specific time interval with velocity field of the flow at this time. The
image series in time are correlated with a selected image (Fig 4b) which consists of one tooth of the
gear to define the specific position of the gear. This allows locating the instantaneous fields that has
the gear in the selected position. Those instantaneous velocity vectors are averaged to obtain a
conditional velocity field representing the characteristic flow structure at this gear position (Fig 4).

100

20

40

200

60

80

100

300
120

140

160

180

400

20

40

60

80

100

120

140

160

180

(b)
500

600

50

100

150

200

250

300

350

400

450

500

550

(a)

Fig 4. Representation of the selected image. (a) Original image. (b) Selected image.

-5-

14th Int Symp on Applications of Laser Techniques to Fluid Mechanics


Lisbon, Portugal, 07-10 July, 2008
1037

Fig 5 and 6 show the conditional averages obtained for the inlet chamber at different sampling rate
(Fig 5a-6a, 5b-6b and 5c-6c at 500 fps, 1000 fps and 2000 fps respectively) and for the outlet
chamber (Fig 5d-6d) at 1000 fps. All the inlet measurements were taken in a horizontal (x-y) plane
at vertical location coincident with the flow entrance, while the outlet measurements were taken
only in a horizontal plane in the upper plane of the chamber. Hence, it is not possible to see in Fig
5d that the vectors are leaving from the chamber. For the suction chamber, it could be seen that the
fluid flows through the gears from the two sides of the chamber symmetrically and produces two
vortices on the right and left side of the chamber. The small vortices also appear in the end points of
gear teeth. For the impulse chamber is clearer to observe the flow with two vortices which are
closer to the center side of the pipe and small vortices are not obtained in the end points of gear
teeth. It is clear that the flow in the inlet chamber is more complex that the flow at the outlet.
x [mm]

x [mm]

y
[mm]

y
[pixel]

y
[pxel]

y
[mm]

x [pxel]

x [pixel]
x [mm]

(b)

(a)

x [mm]

y
[pixel]
y
[mm]

y
[pixel]

y
[mm]

x [pixel]

x [pixel]

(c)

(d)

Fig 5. Velocity fields results which are obtain in different frequency rates (a) Inlet with 500fps (b)
Inlet with 1000fps (c) Inlet with 2000fps (d) Outlet with 1000fps
-6-

14th Int Symp on Applications of Laser Techniques to Fluid Mechanics


Lisbon, Portugal, 07-10 July, 2008
1037
x [mm]

x [mm]

y
[mm]

y
[pixel]

y
[pxel]

x [pixel]

y
[mm]

x [pxel]

(a)

(b)

x [mm]

x [mm]

y
[pixel]

y
[pxel]

y
[mm]

y
[mm]

x [pixel]

x [pixel]

(d)

(c)

Fig 6. Streamlines results which are obtain in different frequency rates (a) Inlet with 500fps (b) Inlet
with 1000fps (c) Inlet with 2000fps (d) Outlet with 1000fps
It has been shown that the sampling rate do not have an important effect on the flow structure
obtained, at least at the rotation frequency used here. To find the flow evolution inside the suction
chamber the instantaneous data obtained for a sampling rate of 1000 fps has been used since it
gives a rather better resolution the other two cases (500 fps and 2000 fps). Fig 7 and Fig 8 show the
velocity vectors and streamlines at six consecutive times which corresponds to different position of
the gear teeth. It can be observed that the centre of the large vortices do not change their position
with the rotation of gear, while the small vortices could appear, disappear or join to large ones.

-7-

14th Int Symp on Applications of Laser Techniques to Fluid Mechanics


Lisbon, Portugal, 07-10 July, 2008
1037

ti

ti+3

ti+6

ti+9

ti+12

ti+15

Fig 7. Velocity fields of suction chamber with 1000 fps for different positions of gear teeth.
-8-

14th Int Symp on Applications of Laser Techniques to Fluid Mechanics


Lisbon, Portugal, 07-10 July, 2008
1037

ti

ti+3

ti+6

ti+9

ti+12

ti+15

Fig 8. Streamlines of suction chamber with 1000 fps for different positions of gear teeth.
-9-

14th Int Symp on Applications of Laser Techniques to Fluid Mechanics


Lisbon, Portugal, 07-10 July, 2008
1037

The suction chamber velocity profiles (Fig 9) show that the v velocity component is considerably
increasing when the fluid is flowing through the gear teeth and the maximum negative velocity of
mean v has been found in the center of the inlet side of the gear pump. The mean u velocity
component reaches its maximum on the right middle and on the left middle side of the gear pump.
Fig 9 shows that the profiles become less symmetric as they move away from the inlet section. This
lack of symmetry could be generated by the model performance and needs a more detailed analysis
with a different rotation velocity and image acquisition at more than one horizontal plane.
Mean v vs x

Mean u vs x

100
50
0

y=3.48
y=5.4
y=7.5
y=9.2
y=10.8
y=12.9
y=15.4

300

200
u [ mm / s ]

-50
v [ mm / s ]

400

y=3.48
y=5.4
y=7.5
y=9.2
y=10.8
y=12.9
y=15.4

-100
-150

100

-200

-250
-100

-300
-350
0

10

15
x [ mm ]

20

25

-200
0

30

10

(a)

15
x [ mm ]

20

25

30

(b)

Fig 9. Inlet flow in the suction chamber at 1000fps (a) Mean v (b) Mean u
Mean v vs x

Mean u vs x

100
50

-50
-100

y=1.8
y=3.7
y=6.0
y=7.9
y=9.8
y=11.7
y=13.7

80
60
40
u [ mm / s ]

0
v [ mm / s ]

100

y=1.8
y=3.7
y=6.0
y=7.9
y=9.8
y=11.7
y=13.7

20
0
-20

-150

-40
-200
-250
0

-60
5

10

15
x [ mm ]

20

25

30

-80
0

10

15
x [ mm ]

20

25

30

(b)

(a)

Fig 10. Outlet flow in the impulse chamber at 1000fps (a) Mean v (b) Mean u
Velocity profiles at the impulse chamber (Fig 10), shows that the v velocity component is
increasing when the fluid is passing through the gear teeth and the maximum negative velocity of
mean v has been found in the center of the outlet side of the gear pump. The mean u velocity
component reaches to maximum on the right middle and on the left middle side of the gear pump.
Results show that the flow in the suction chamber is much more complex than the flow in the
impulse chamber. Therefore the inlet chamber is the one that needs more detailed an extended
study.

- 10 -

14th Int Symp on Applications of Laser Techniques to Fluid Mechanics


Lisbon, Portugal, 07-10 July, 2008
1037

5. Conclusion
The use of air bubbles as tracing particles for PIV has been proved as a fine alternative to the use of
solid or liquid particles. TRPIV has been applied to the study of the flow structures in the suction
and impulse chamber of an external gear pump. Results show the possibility that the analysis
technique presented can be used to obtain detailed information of the instantaneous velocity fields,
in systems with moving elements, which are not part of the fluid flow. The technique for boundary
treatment developed by Usera et al. (2004) has been applied with the use of weighting function to
obtain the same level of accuracy available at inner locations of the system. Corrected positions of
the velocity displacement relative to the boundary have been computed. A conditional average
velocity field has been obtained for specific gear position allowing an average time evolution of the
flow structures in the suction chamber. The results obtained show that a detailed analysis of the
suction chamber is needed for a better understanding of the dynamic behavior of the flow.

Acknowledgments
This study was financially supported by the Spanish Ministry of Science and Education and FEDER
under projects DPI2006-02477 and DPI2006-14476.

References
Brennen C E (2005) Fundamentals of Multiphase Flow, Cambridge University Press.
Bolinder J (1999) On the accuracy of a digital particle image velocimetry system, Tech. rep., Lund
Institute of Technology.
Castilla R, Gamez-Montero P, Huguet D, Codina E, (2007) Turbulence in Internal Flows in
Minihydraulic Components, CIMNE, pp 241-251.
Codina E, Kamashta M, (1999) ECOPUMP project, Enhanced design of high pressure gear pumps
using environmentally acceptable hydraulic fluids, BRITE Contract n BRPRCt95-0094, Tech. rep.,
LABSON-UPC.
Dearn R, B.Sc.(Hons), (June 2001) European Marketing Manager, The fine art of gear pump
selection and operation, World pumps, Volume 2001, Issue 417, pp 38-40.
Hart DP (1998) The Elimination of Correlation Errors in PIV Processing, 9th International
Symposium on Applications of Laser Techniques to Fluid Mechanics, Lisbon.
Hart DP (1999) Super-Resolution PIV by Recursive Local-Correlation, Journal of Visualization
(10).
Iyoi H and Ishimura S (1983) -Theory in gear geometry, Transaction of ASME Journal of
Mechanisms, Transmissions, and Automation in Design 105, pp 286290.
Moore J (2007) Dry sump pump bubble elimination for hydraulic hybrid vehicle systems, Master
thesis in the department of Mechanical engineering, The university of Michigan.
Nogueria J, Lecuona A, and Rodriguez PA, (1997) Data validation, false vectors correction and
derived magnitudes calculations on PIV data, Meas. Sci. Technol. (8), 1493-501.
- 11 -

14th Int Symp on Applications of Laser Techniques to Fluid Mechanics


Lisbon, Portugal, 07-10 July, 2008
1037

Nogueria J, Lecuona A, and Rodriguez PA (2001) Local field correction PIV, implemented by
means of simple algorithms, and multigrid versions, Meas. Sci. Technol, 12, 1911-1921.
Nogueria J, Lecuona A and Rodriguez PA (2002) Accuracy and time performance of different
schemes of the local field correction PIV technique, (33), 743-751.
Raffel M, Willert C, Kompenhans J (1998) Particle Image Velocimetry: A Practical guide,
Springer.
Usera G, Vernet A, Ferr JA (2004) Consideration and Improvements of the Analysis Algorithms
Used for Time Resolved PIV of Wall bounded Flows, 12th International Symposium on Applications
of Laser Techniques to Fluid Mechanics, Lisbon.
Willert CE and Gharib M, (1991) Digital particle image velocimetry, Exp. Fluids (10), 181-193.

- 12 -

You might also like