Home Search Collections Journals About Contact us My IOPscience

Experimental study on the similarity of flow in pump sump models

This content has been downloaded from IOPscience. Please scroll down to see the full text.

2012 IOP Conf. Ser.: Earth Environ. Sci. 15 062047

(http://iopscience.iop.org/1755-1315/15/6/062047)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 201.173.238.112
This content was downloaded on 27/06/2016 at 04:01

Please note that terms and conditions apply.

26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062047 doi:10.1088/1755-1315/15/6/062047

Experimental study on the similarity of flow in pump sump

models

K Kawakita, J Matsui and H Isoda

Yokohama National University, Tokiwadai 79-5, Hodogaya, Yokohama, JAPAN

E-mail: jmat@ynu.ac.jp

Abstract.To study the similarity law in the flow of pump sump, especially on the critical
submergence for both air entraining vortex and submerged vortex, three model pump sumps
are constructed with geometrical similarity. Changing the flow rate and water height, the
condition when sump vortex occurs is studied in each model. On the air entraining vortex,
there are two critical lines of submergence, one is in proportional to the flow rate or velocity at
bell inlet, the other is almost independent from flow rate. The former may be the effect of
Froude number, but the later is not clear. On the submerged vortex, the line is not clear in the
smallest model. Comparing the coefficients of above lines, the kinetic similarity is got at the
same velocity condition, not Froude number similarityor Reynolds number similarity.

1. Introduction
In a pump sump, the flow is very complicated with three dimensional flow patterns, deformed free
surface, and unsteady strong vortices. In some cases air entraining (free surface) vortices or submerged
(sub-surface) vortices occur, and they cause acoustic noise and vibration. When these vortices develop
more, the performance of the pump system is reduced, and in some cases the pumps cannot be
operated. It is desired to verify the validity of the sump design by model sump testing.
Though the model test is very important in the design of the sumps, the similarity law of the model
testing is not clear yet, and there are different opinions for velocity setting in the model test[1]. The
similarity of the vortex generation is different according to the type of the vortex, since the mechanism
of the generation is different, and it is under the influence of various phenomena. For example, the
generation of the air entraining vortex is mainly controlled by the Froude number. But when the flow
speed is increased, or when the length scale is decreased, we must think on the effect of viscosity, so
Reynolds number or scale effect should be considered.
One of authors has been studied flows in a model sump whose diameter of suction bell is 150 mm,
and report a CFD benchmark on it[2]. Here we make a larger apparatus, which is twice the size of
previous one.The bell diameter in the new apparatus is 300mm. This larger experiment system has a
geometric similarity with the previous system. Also we made a smaller system with bell of 82mm
diameter. Comparing the occurrence limits of air entraining vortex and submerged vortex, we
investigate the similarity law in these three scale sumps.

2. Experimental method
The configuration of the experimental apparatus of the model sump is shown in figure 1. The test
section of the flow passage has a rectangular shape, and it is constructed by transparent acrylic resin
for visualization and optical measurements. The water in the flow passage flows into a suction pipe

Published under licence by IOP Publishing Ltd 1

26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062047 doi:10.1088/1755-1315/15/6/062047

with a suction bell. This suction pipe has a shape of a siphon. Then the water gets the head from the
pump, and it goes back to the flow passage through an exhaust area, which has punched plates,
honeycomb plates, and filters. There also are reductions of the section of flow passage both in
horizontal and vertical direction, so that the velocity distribution becomes flat and the turbulence is
small.
The sizes in figure 1 are of the apparatus whose outer diameter of bell mouse D=150mm.
Figure 2 is a photograph of D=300mm apparatus. At the upstream channel wall, windows are set to
measure the velocity distribution of inlet flow.

Figure 1.Experimental System (D=150 mm)

Figure 2. D=300mm system and LDV window Figure 3. Test section and symbols
Figure 3 shows the configuration in test section area and symbols. The diameter of bell mouse is D,
the width of sump passage is B, and the water height is H. The submergence is S, which is the distance
between water surface and bell inlet.
As the water surface is not horizontal because the flow near bell mouse goes down to the bell inlet,
the water height is measured about 2D upstream side from the center of bell mouse.
In this paper, the coordinate x is a direction of main flow in the flow passage. The y direction is a
horizontal and normal direction to the main flow. The z direction is vertical and upper direction.
The origin of the coordinate axes is at the corner of the flow passage, shown in figure 4.

2
26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062047 doi:10.1088/1755-1315/15/6/062047

In the D=150[mm] model, the height of the suction bell from the bottom wall of the flow passage is
fixed to Z=100[mm]. The width of the flow passage in y direction is B=300[mm].
The diameter of suction pipe is 100[mm], whose center axis is set at x=-100[mm] and
y=140[mm].So the suction pipe has a bias in y direction.
In the D=350[mm] model, all the size is set twice of D=150 [mm] system, to get the geometric
similarity. Also in the D=82[mm] model, all geometric sizes new the suction pipe is similar.

Figure 4. Coordinates
In the D=150[mm] system, the velocity distribution at the upstream cross section at x=-450[mm] is
measured by LDV system. The result is shown in figure 5(a). These velocity distribution are measured
at flow quantity Q=1.1 [m3/min] and height of the water in the flow passage H=240[mm]. Though the
suction pipe is asymmetrically set, the velocity profile at the upstream is flat. The standard deviation
of velocity u is under 5 percent of the average velocity U=Q/(HB).
The velocity distributions in the D=300[mm] system are shown in figure 5(b,c,d,e). In this system,
the velocity distribution is flat, too. To check the flow pattern caused by the decrease of cross section
of flow passage, a simple CFD simulation is made. The passage shape is modeled as infigure 6. The
inlet flow has flat distribution at the upper inlet of the CFD model. The CFD results agree with the
LDV measurement, so the tendency of velocity distribution in Figure 5 is from the shape of flow
passage, not from the turbulence or some disturbances of the experimental system.

(a) At x= -450[mm] of D=150[mm] system (b) At x= -4800[mm] of D=300[mm] system

3
26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062047 doi:10.1088/1755-1315/15/6/062047

(c) At x=-4800[mm] of D=300[mm] system (d) At x=-4800[mm], in D=300[mm] system

(e)In vertical direction in D=300[mm] system

Figure 5. Velocity distribution at upstream section

Figure 6. CFD configuration to check the velocity distribution in D=300[mm] system

3. Results and discussions

3.1. Vortex map

Changing the flow rate Q and submergence S, the occurring conditions of air entraining vortex and
submerged vortex are observed. When some surface dimple grows to large air entraining vortex, with
continuous air suctioning, we judge that the ‘Continuous vortex’ occurred. In some experiments, the
time length while the continuous vortex survived is measured and ratio T = (time while vortex is
surviving)/(time of observation) is measured.Figure 7 shows an example of continuous vortex.

4
26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062047 doi:10.1088/1755-1315/15/6/062047

When some visible air cores are observed from wall of flow passage to the bell, we judge that the
'submerged vortex' occurred. Usually at this time the continuous vortex occurs. Figure 8 shows an
example of submerged vortex from bottom wall.

Figure 7. Photograph of continuous vortex

Figure 8. Photograph of submerged vortex

Figure 9 (a,b,c) are 'vortex map', which show the relation of vortex occurring and parameters Q and
S.Figure 9(a) is the result of D=300[mm] system. Figure 9(b) is of D=150[mm], and Figure 9(c) is of
D=82 [mm].When Q becomes larger or S/D becomes smaller, the air entraining vortex occurred, and it
becomes stronger. We can see the smallest S value with no vortex, that is the critical submergence, Sc,
for each Q.The American National Standard for Pump Intake Design [3] recommends the formula of
Hecker’s [4],

√
where: FD is a Froude number and V is the average velocity at the inlet of bell, with Q= VD2/4.This
formula is also plotted in the Figure 9(a) and (b). The plotted S/D is much higher than the
experimental Sc/D, so the American standard has enough room from the critical submergence, even
though the flow in our experiments has some circulation caused by the shifted center of suction pipe.

5
26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062047 doi:10.1088/1755-1315/15/6/062047

The formula of Hecker’s means that Sc/D is in proportional to Q, and we can see the similar
relation in Figure 9.

(a) D=300mm result

(b) D=150mm result

6
26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062047 doi:10.1088/1755-1315/15/6/062047

(c) D=82mm result

Figure 9. Vortex map

3.2. Similarity on air-entraining vortex

To consider the similarity of occurring condition of air entraining vortex, the flow rate Q is converted
to the averaged velocity at bell inlet, V = Q/( D2/4). The converted vortex maps are shown in Figure
10.
We can see that there are two lines of critical submergence on air entraining vortex.Line A1 in the
Figure 10 is similar to the Hecker’s formulation, in which S/D is proportional to V. The other line A2
is horizontal, no dependency on the V or Q.
The A1 line is thought to be under the control of Froude number, and A2 may be the effect of wave
disturbance.Becausevery strong wave will break the surface vortex.

(a) D=300[mm] system

7
26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062047 doi:10.1088/1755-1315/15/6/062047

(b) D=150mm system

(c) D=82mm system

Figure 10. Vortex occurring map with bell inlet velocity V
Though the A1 lines in Figure 10 are plotted very roughly, the formula is modeled as
S/D = C V
witha coefficient C.In D=300[mm] case, the coefficient CLis about 1.8, and in D=150[mm] case, the
coefficient CM = 1.5.
If the Froude number is only the dominant parameter, the velocity in the middle size
systemVMshould be set as
VM/VL = (DM/DL)0.5
Here VLis the velocity in the large system. DM and DL are the diameter of bell in each case.
When Reynolds number effect becomes large, the relation becomes
VM/VL = (DM/DL)n (1)
When only Reynolds number is the dominant parameter, nis-1. In TSJ[1], n= 0.2 is recommended
for air entraining vortex.

8
26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062047 doi:10.1088/1755-1315/15/6/062047

Now we assume that in the D=300[mm] caseSc/D = CLVL,and that in the D=150[mm] caseS c /D =
CMVM.If the kinetic similarity is got in our experiments, the critical S c /D value in both cases should be
same. So
(CL VL) / (CM VM) = 1
is settled. Using formula (1), this relation is
CL/ CM = (DM/ DL)n
As we see in Figure 10, CL = 1.8 and CM = 1.5, and using DM / DL = 0.5, n is calculated as
n = -0.26
When n= 0, VM becomes same as VL. The above value of n says that when the velocity is almost
same, the kinetic similarity is got in our experiments.
Between D=300[mm] and D=82[mm] cases, CS = 1.5, so n = -0.15. This similarity condition is
almost same as the D=150[mm] case.
On the A2 line, the similarity is not clear. We should measure the wave height and velocity
distribution near the bell mouse in the future work.

3.3. Similarity on submerged vortex

On the critical submergence of submerged vortex, the same process can be adopted.In the D=300[mm]
case,CL is about 0.7, and D=150[mm] case CM= 0.45.In D=82[mm] case the relation is a little different
from the model formula S/D= CV, butCS is almost 0.65.
Between CL and CM, n = -0.64, and between CLand CS, n = -0.06. In TSJ[1], the submerged vortex
test should be done in n=0 condition. Our result agrees with this recommendation.

4. Conclusions
Using three sump models that are geometrically similar, the occurrence of air entraining vortex and
submerged vortex is studied. Even though the scale ratio between models are 0.5 and 0.3, the kinetic
similarity is established when velocity is same in both on air entraining vortex and on submerged
vortex.
There are other limitations of air entraining vortex, which is almost independent to the Q. This may
be the effect of surface waves, that may break the air entraining vortex, but the phenomenon is not
clear yet.

Acknowledgment
This work was supported by KAKENHI 22560158.

References
[1] The Turbomachinery Society of Japan 2005 Standard Method for Model Testing the
Performance of a Pump Sump (TSJ S002)
[2] Matsui J, Kamemoto K and Okamura T 2006 Proc. the 23th IAHR Symp.
[3] Hydraulic Institute 1998 American National Standard for Pump Intake Design
[4] Hecker G E 1987 IAHR Hydraulic Structures Desgin Manual 1