Sei sulla pagina 1di 7

Proceedings of the ASME 2014 Pressure Vessels & Piping Conference

PVP2014
July 20-24, 2014, Anaheim, California, USA

PVP2014-28610

WAVE PROPAGATION OF WATER HAMMER IN COMPOUND TUBE

Kengo Takeda, Kazuaki Inaba, and Kosuke Takahashi


Tokyo institute of Technology
2-12-1, Ookayama, Meguro-ku, Tokyo, 152-8550, JAPAN
E-mail: takeda.k.ah@m.titech.ac.jp, inaba@mech.titech.ac.jp, ktakahashi@mech.titech.ac.jp

ABSTRACT INTRODUCTION
There are pipes with square or rectangular cross sectional A water hammer occurs by a sudden closing of a valve or
shape rather than more common circular shape. Moreover, a an accidental stopping of a pump [1-4]. The pressure of fluid in
part of the walls can be made of different materials or have a tube rises rapidly because kinetic energy of fluid is converted
different wall thicknesses. To estimate pressure wave into pressure by the occurrence of the water hammer. This
propagation in such a channel, we examined water hammer study is important in the fact that water hammer may cause a
phenomenon and measured wave speeds in rectangular tubes pipe rupture and leaking harmful material. There are many
assembled with different wall materials and thicknesses by studies of water hammer in circular tube but few studies of
strain gages and pressure transducers. We newly proposed water hammer in rectangular tube. However, there are many
theoretical wave speeds based on the classical theory and the rectangular tubes which have different thickness walls or
junction conditions of two walls: fixed, simply supported, and different material walls in industry.
mixed supported. Circumferential strains on the wall can be The present study reveals that water hammer wave speed
estimated from the different forms derived by the different in compound rectangular tube, which has different thickness
junction conditions. We confirmed that theoretical predictions walls or different material walls by experiment by building
with the simply supported conditions agreed with experimental compound tube made with different kinds of plates. By
wave speed in compound tubes, while wave speeds in ready- developing the existing water hammer theory for wave speeds
made square tubes with the same wall-thickness were close to in rectangular tubes, we will derive water hammer theory to
those by the fixed conditions. estimate wave speeds in compound rectangular tube with
considering aspect ratio of tube section and ratio of tube wall
NOMENCLATURE thickness. Comparing experiment wave speeds with theoretical
A Area of rectangular tube section [m2] speeds, we will examine the validity of the derived theory.
c Sound speed in elastic rectangular tube [m/s]
D Length of rectangular tube section [m]
E Young’s modulus of tube material [GPa] FORMULATION OF WAVE SPEEDS IN COMPOUND
e Thickness of tube wall [m] RECTANGULAR TUBE
G Modulus of transverse elasticity [GPa] The classical theory for wave speeds in uniform polygonal
I Second moment of area [m4] (square) tube [5, 6] is given by eq. (1) where ΔA / A is
K Bulk modulus of water [GPa] calculated by three terms as presented in eq. (2). The first term
L Length of rectangular tube section [m] indicates the influence of deformation by direct strain. The
P Pressure [Pa] second term and the third term are corresponding to bending
T Tensile load [N] deformation and shear deformation, respectively.
V Shear load [N]
u Buffer speed [m/s]
y Deflection of tube wall [m] 𝑲 ⁄𝝆
𝒄=√ (1)
ρ Density of water [kg/m3] ∆𝑨
𝟏+ 𝑲
𝑨∆𝑷

1 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


∆𝐴 ∆𝐴 ∆𝐴 ∆𝐴 𝑳
=( ) +( ) +( ) (2) 𝑽 = ∆𝑷 ( − 𝒙) (6)
𝐴 𝐴 𝐷 𝐴 𝐵 𝐴 𝑆 𝟐
1
Figure 1 indicates the cross-sectional dimensions of 𝑦= (𝐿𝑥 − 𝑥 2 )∆𝑃 (7)
2𝐺1 𝑒1
compound rectangular tube. In this study, we consider two pairs
of plates whose thickness and materials are the same while it is ∆𝐴 𝐿2 𝐷2
possible to expand our equations into the cases of three plates ( ) = + (8)
𝐴∆𝑃 𝑆 4𝐷𝐺1 𝑒1 4𝐿𝐺2 𝑒2
plus one plate or four different plates. In Fig. 2, equilibrium of
forces and deformations of the tube were considered after the
cross-sectional area in Fig. 1 was divided into 8 parts by Figure 3 illustrates the deformation of tube wall for the
considering symmetry. Tensile load T is given by eq. (3) and fixed supported due to pressure loading of water hammer.
cross-sectional area change ΔA due to the tensile load can be Bending moment M at displacement x is calculated as eq. (9).
calculated by eq. (4). The angle θ in eq. (4) is 90º for the Since the curvature of tube wall can be defined by the second
rectangular tube. According to eqs. (3) and (4), ΔA / A in eq. (5) derivative of y, the deflection y is calculated with the second
can be defined as the term of influence of the direct tensile load moment of area I in eq. (10). Cross-sectional area change ΔA
on tube wall lengths L and D. can be estimated by integrating y with respect to x from 0 to L
(from 0 to D). Bending deformation results in ΔA / A in eq.
(11).

FIG. 1 Cross-sectional dimensions of compound


rectangular tube. FIG. 3 Fixed wall.

𝟐
𝑳𝟐 𝟏 𝑳
𝑴 = −∆𝑷 + ( − 𝒙) ∆𝑷 (9)
𝟐𝟒 𝟐 𝟐
∆𝑃 1 1
𝑦= ( 𝑥 4 − 𝐿𝑥 3 + 𝐿2 𝑥 2 ) (10)
12𝐸1 𝐼1 2 2

∆𝐴 1 𝐿4 𝐷4
( ) = ( + ) (11)
𝐴∆𝑃 𝐵 30 𝐸1 𝑒13 𝐷 𝐸2 𝑒23 𝐿

In eqs. (9-11), we assumed fixed boundary condition for


FIG. 2 Equilibrium of forces and deformations of
considering the bending deflection of the tube wall. However, it
compound rectangular tube.
is also possible to consider different boundary conditions for
tube walls depending on the ratios of wall thickness in
𝟏 designing car body flame [7]; fixed supported or simply
𝑻= 𝑫∆𝑷 (3)
𝟐 supported. With this idea, we can consider the junction of a side
1 𝐷 𝐷 𝐷𝐿 plate is simply supported when the side plate and the vertical
∆𝐴 = ( + ∆𝐷) ( + ∆𝐷) tan 𝜃 − (4) plate are the same wall-thickness and same material, or the side
2 2 2 8
plate has thicker or larger Young’s modulus than that of the
∆𝐴 𝐷 𝐿 vertical plate. In the simply supported condition, deflection y
( ) = + (5)
𝐴∆𝑃 𝐷 2𝐸1 𝑒1 2𝐸2 𝑒2 becomes eq. (12). In the simply supported condition, the term
of bending moment is now calculated by eq. (13).
Equation (6) indicates the shear load V in Fig. 2.
Displacement of shear direction y can be estimated by eq. (7). ∆𝑷 𝟏 𝟏
Cross-sectional area change ΔA due to the shear load is 𝒚= ( 𝒙𝟒 − 𝑳𝒙𝟑 + 𝑳𝟑 𝒙) (12)
𝟏𝟐𝑬𝟏 𝑰𝟏 𝟐 𝟐
obtained by integrating y with respect to x from 0 to L (from 0
to D), and ΔA / A can be calculated by eq. (8).

2 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


∆𝐴 1 𝐿4 𝐷4
( ) = ( + ) (13)
𝐴∆𝑃 𝐵 5 𝐸1 𝑒13 𝐷 𝐸2 𝑒23 𝐿

FIG. 5 Experiment apparatus with ready-made rectangular tube.


FIG. 4 Simply supported wall. We also examined rectangular compound tubes by
assembling four plates, two pairs of side and vertical plates as
When the plate is jointed to thicker or stiffer plates, we presented in Fig. 6. Vertical plate is aluminum with thickness of
consider the joint as the fixed support. On the other hand, when 10 mm, and three types of plates are used for side plates;
the plate is jointed to thinner or softer plates, we consider the aluminum plates with thickness of 10 mm or 5 mm, or
joint as the simply supported. We can also consider the mixed polycarbonate (PC) plates with thickness of 5 mm. Table 1 lists
supported if one plate of the tube wall is supported by the fixed side plates materials and dimensions. Vertical plate and side
support and the other plate is supported as simply supported as plate are combined by M5 bolts and glued by silicon system
given in eq. (14).
adhesive. Inner cross-section of rectangular tube is 50 mm × 46
mm. Three pressure transducers are mounted at the interval of
∆𝑨
100 or 200 mm. Strain gages are glued in the same manner as
𝑨∆𝑷 the ready-made rectangular tubes. By changing side plates, we
𝑫 𝑳 𝑳𝟐 measure strain signals and obtained water hammer speeds from
= + + (14)
𝟐𝑬𝟏 𝒆𝟏 𝟐𝑬𝟐 𝒆𝟐 𝟒𝑫𝑮𝟏 𝒆𝟏 the arrival times at each gage location.
𝑫𝟐 𝟏 𝑳𝟒 𝟏 𝑫𝟒
+ + +
𝟒𝑳𝑮𝟐 𝒆𝟐 𝟑𝟎 𝑬𝟏 𝒆𝟑𝟏 𝑫 𝟓 𝑬𝟐 𝒆𝟑𝟐 𝑳

We derived three types of theoretical water hammer wave


speeds; all fixed, all simply supported, and mixed supported.

EXPERIMENTAL APPARATUS
First, water hammer experiments are conducted with two
ready-made rectangular tubes filled with water as shown in Fig. FIG. 6 Experimental apparatus with compound rectangular
5. The ready-made rectangular tubes are 1000 mm axial length tube
with square section whose outer length of the side is 50 mm.
One rectangular tube is made of aluminum with thickness of TABLE 1 Side plates materials and dimensions.
2.3 mm. The other is steel with thickness of 2 mm. The free-fall Thickness Young's Poisson's
projectile impacts a polycarbonate buffer placed on the top of Material
[mm] modulus [GPa] ratio
the water surface [8]. By this impact test facility, pressure wave
and elastic stress wave propagate through water and tube wall (1) Al 10 70 0.345
as a water hammer wave. Total ten strain gages are glued on the (2) Al 5 70 0.345
rectangular specimen tube in the circumferential direction,
perpendicular to the axial direction. Five gages are glued at the (3) PC 5 2.5 0.380
axial center of the tube and the others are glued at near the
corner of the tube with 5 mm offset from the corner by the
interval of 100 mm. RESULTS AND DISCUSSION
Figure 7 indicates strain histories measured at (a) g1-g5
and (b) g6-g10 in the ready-made rectangular tube (Al with
thickness 2 mm). Here, baselines of each strain signal are
shifted by offsets proportional to the distance from gages g1
and g6. As shown in Fig. 7, strains at the center of the tube
increases (tension on outer surface) while strains near the
corner of the tube decreases (compression on outer surface) in
both ready-made tubes. Wave speeds obtained from strain

3 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


signals with Al 2 mm tube and Steel 2.3 mm tube were 266 m/s side plate are given in (a) and strain signals near the corner are
and 536 m/s. plotted in (b). In Figs. 9 and 10, strain signals at the center are
Figure 8 shows pressure histories measured at p1 to p5 in larger than that at corner of side plate and both strains are
compound rectangular tubes. It is possible to mount pressure generally positive (tension loading on the surface). However, in
transducer on the vertical plates because they are enough thick. Fig. 11, strains at the center of side plate increase (tension
In Fig. 8, baselines of pressure signals are shifted by offsets loading) while strains near the corner slightly decrease
proportional to the distance from the transducer p1 to the other (compression loading). Strain signals with PC side plates
transducers. The wave speeds of water hammer are calculated indicated different behavior from those with Al side plates and
by the distance between transducers and arrival times of waves. are similar as ready-made rectangular tubes.
Since strain signals are unstable for these compound tubes, we
used pressure signals to determine the wave speeds rather than
strain signals. Averaged wave speed for Al 10 mm side plate
was 837 m/s and was higher than 575 m/s with Al 5 mm side
plate. This is because the strength of the fluid-structure
interaction is changed by the thickness of the side plates. This
tendency becomes clear when we used PC tube as the side
plates. The wave speed with PC 5 mm side plates was 145 m/s
and much slower than those with Al side plates.

(a) Side plate: Al with thickness of 10 mm

(a) At the center of the plate

(b) Side plate: Al with thickness of 5 mm

(b) Near the corner of the plate

FIG. 7 Strain histories (Al with thickness of 2 mm).

Figures 9-11 indicate strain histories measured at g1 to g5


and g6 to g10 in compound rectangular tubes. In Fig. 9, side
plate is Al with thickness of 10 mm, Fig. 10 is the case with Al (c) Side plate: PC with thickness of 5 mm
with thickness of 5 mm, and Fig. 11 is the case with PC with
thickness of 5 mm. In each figure, strain signals at the center of FIG. 8 Pressure histories.

4 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


(a) At the center (a) At the center

(b) Near the corner (b) At the corner


FIG. 9 Strain histories (Al with thickness of 10 mm). FIG. 11 Strain histories (PC with thickness of 5 mm).

To examine strain responses in detail, we calculated ratios


of strains at the center and near the corner. Figure 12 indicates
strain ratios. In Fig. 12, estimated strain ratios are plotted as
lines by assuming the simply supported condition and the fixed
condition. In Figs. 12(a, b), stain ratios are close to theoretical
ratios obtained by the simply supported condition. In Figs.
12(c-e), strain ratios are close to these with the fixed support
condition. According to these observations, we consider that the
deformations of the side plates with Al 10 mm and 5 mm are
close to the beam with the simply supported and the other
deformations are similar to the beam with the fixed boundary.
Table 2 summarizes experimental and theoretical wave
(a) At the center speeds. In the ready-made rectangular tubes, the experimental
wave speed agrees with theory wave speeds with the fixed
supported condition. In cases of Al side plates with thickness of
10 mm and 5 mm, experimental wave speeds are close to the
theoretical wave speeds with the simply supported condition.
With PC 5 mm side plates, wave speeds are around the middle
of the fixed supported condition and the simply supported
condition. We consider that deviations of wave speeds from
theories are caused by the bolt joints and incomplete joints.
Further investigation is required to verify these effects, by
clamping tubes at the space between bolts, and conducting
numerical analysis. However, we confirmed that our simple
theories can roughly predict wave speeds and that the fixed and
(b) Near the corner simply supported condition work better than the mixed support
FIG. 10 Strain histories (Al with thickness of 5 mm). generally used in the car industry for water hammer problems.

5 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


TABLE 2 Wave speeds for rectangular tubes in experiments
and theories.
Tube type Compound (Al 10 mm) Ready-made
Plate material and wall Al PC Al Steel
thickness (mm) 10 5 5 2 2.3
837 575 145 266 536
Experiment (m/s)
±77 ±81 ±19 ±5 ±18
Theory
881 495 103 119 253
(simply supported) (m/s)
Theory (fixed) (m/s) 1152 881 222 280 560

Theory (mixed) (m/s) 1018 816 221 155 326


(a) Compound tube (Al 10 mm)

We also evaluated pressure jump by the Joukowski


pressure with estimated wave speeds by eq. (15)

∆𝑷 = 𝝆𝒄∆𝒖 (15)

where ∆𝒖 is estimated by the averaged buffer speeds. We can


(b) Compound tube (Al 5 mm) confirm that theoretical pressure has similar tendency as
experimental results while pressure loss due to the imperfect
joints may occur and relatively low pressure was observed in
experiments. Further study will be necessary to predict pressure
jump across the water hammer in rectangular compound tubes.

TABLE 3 Comparison between averaged experimental


pressure and theoretical pressure for compound tubes.
Al PC
Wall thickness (mm) 10 5 5
(c) Compound tube (PC 5 mm)
Experiments (kPa) 483 381 191
Simply supported
994 558 116
(kPa)
Fixed supported
1300 994 250
(kPa)
Mixed supported
1148 921 249
(kPa)

(d) Ready-made tube (Al 2 mm)


CONCLUSION
We conducted water hammer experiments with ready-made
rectangular tubes and compound rectangular tubes whose wall
plate thickness and wall materials are different. We also
developed the classical theory for water hammer wave speeds
to predict wave speeds in rectangular compound tubes by
considering three boundary conditions; fixed, simply supported,
and mixed. In ready-made rectangular tube, experiment wave
propagation speed agreed with the theory by the fixed
(e) Ready-made tube (Steel 2.3 mm) supported condition. In the case of same material, experiment
wave speed was close to the theory by the simply supported
FIG. 12 Ratios of strain signals behind wave fronts condition while in the case of different material with PC and Al,

6 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


experimental speeds are middle of the theories by the fixed
condition and the simply supported condition.

ACKNOWLEDGMENTS
We thank Prof. Kikuo Kishimoto, Tokyo Institute of
Technology, for discussion and comments on this study.

REFERENCES
[1] G. Z. Watters, Analysis and Control of Unsteady Flow in
Pipelines, Butterworth Publishers, MA, (1984).
[2] E. B. Wylie and V. L. Streeter, Fluid Transients in
Systems, Prentice-Hall, Inc., NJ, (1993).
[3] A. S. Tijsseling, Fluid-structure interaction in liquid-filled
pipe systems: A review, Journal of Fluids and Structures,
10, (1996), pp. 109–146.
[4] D. C. Wiggert and A. S. Tijsseling, Fluid transients and
fluid-structure interaction in flexible liquid-filled piping,
Applied Mechanics Reviews, 54(5), (2001), pp. 455–481.
[5] A. R. D. Thorley and G. L. J. Buttigieg, Surge
Propagation Speeds in Ducts of Regular Polygonal Cross-
Section, Proceedings of 6th Australasian Hydraulics and
Fluid Mechanics Conference, (1997), pp. 561-565.
[6] A. R. D. Thorley and C. Guymer, Fundamental equations
governing pressure surge phenomena in rectangular cross
section, Proceedings of the Institution of Mechanical
Engineers, 189, (1975), pp. 325-332.
[7] D. E. Malen, Fundamentals of Automobile Body Structure
Design, SAE International, (2011), p. 85.
[8]. Inaba, K. and Shepherd, J. E., 2010. “Flexural waves in
fluid-filled tubes subject to axial impact”. ASME J.
Pressure Vessel Technol. 132, 021302.

7 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use