CYBERNETICS AND PHYSICS, VOL. 3, NO.

1, 2014 , 28–37

MODELLING OF A CLAMPED-PINNED
PIPELINE CONVEYING FLUID
FOR VIBRATIONAL STABILITY ANALYSIS

Begoña Mediano-Valiente M. Isabel Garcı́a-Planas

Departament de Matemàtica Aplicada I Departament de Matemàtica Aplicada I
Universitat Politècnica de Catalunya Universitat Politècnica de Catalunya
Spain Spain
bego med@yahoo.es maria.isabel.garcia@upc.edu

Abstract 1 Introduction
Recent developments in materials and cost reduc- In the last decades, the dynamics and stability of pipes
tion have led the study of the vibrational stability of conveying fluid has been studied thoroughly with var-
pipelines conveying fluid to be an important issue. ious techniques of analysis, considering different end
Nowadays, this analysis is done both by means of sim- conditions and different models of the fluid-conveying
ulation with specialized softwares and by laboratory pipeline (see for example [Gregory and Paı̈doussis,
testing of the preferred materials. The former usually 1966; Kameswara and Simha, 2008; Kerboua and
requires of complex modelling of the pipeline and the Lakis, 2008; Kuiper and Metrikine, 2004; Mediano,
internal fluid to determine if the material will ensure vi- 2011; Misra, Wong, Paı̈doussis, 2001; Paı̈doussis,
brational stability; and in the latter case, each time there 1998; Paı̈doussis and Issid, 1974; Paı̈doussis, Tian,
is a mistake on the material selection is necessary to Misra, 1993; Stoicuta et al., 2010]. These authors an-
restart all the process making this option expensive. In alyze stability of pinned-pinned, clamped-clamped and
this paper, the classical mathematical description of the cantilevered fluid-conveying pipes, even in the pres-
dynamic behavior of a clamped-pinned pipeline con- ence of a tensile force and a harmonically perturbed
veying fluid is presented. Then, they are approximated flow field.
to a Hamiltonian system through Garlekin’s method be- The Interest in the problem of vibrational stability not
ing modelled as a simple linear system. The system only is in modeling pipes but in other fields as for
stability has been studied by means of the eigenvalues example analyzing the stability of fixed speed wind
of the linear system. From this analysis, characteristic turbine (see [Dominguez-Garcia]) or analyzing vibra-
expressions dependent on material constants has been tions of a thin stretched string, with an alternating elec-
developed as inequalities, which ensures the stability tric current passing through, in a nonuniform magnetic
of the material if it matches all expressions. This new field, (see [López Reyes and Kurmyshev]).
model provides a simplified dynamical approximation It is well known that the dynamical behavior of pipes
of the pipeline conveying fluid depending on material of a finite length depends strongly on the type of bound-
and fluid constants that is useful to determine if it is ary. The type of supports considered (fixed, one end
stable or not. It is worth to determine that the model fixed, etc.) and their position (horizontal, vertical) must
dynamics does not correspond with the real, but the be distinguished.
global behaviour is well represented. Finally, some The dynamics of the system can be described by a
simulations of specific materials have been use to val- partial differential equation [Seyranian and Mailybaev,
idate the results obtained from the Hamiltonian model 2003; Thompson, 1982]
and a more complex model done with finite element
software.
∂4y ∂2y ∂2y ∂2y
a4 + a3 2 + a2 + a1 2 = 0 (1)
∂x 4 ∂x ∂x∂t ∂t

Key words with boundary conditions at ends of a clamped-pinned

Stability, eigenvalues analysis, pipe conveying fluid, pipe. We find approximate solution of this equation
material selection, Garlekin’s method, Hamiltonian using Galerkin’s method obtaining as a result a linear
systems. gyroscopic system possessing the properties of linear
CYBERNETICS AND PHYSICS, VOL. 3, NO. 1 29

Hamiltonian systems. Then, the eigenvalues of this lin- tance x and time t and is based on the beam theory,
ear Hamiltonian system gives information about stabil- [Abid-Al-Sahib, Jameel, Abdlateef, 2010; Morand and
ity: a stable Hamiltonian system is characterized by Ohayon, 1995]:
pure imaginary eigenvalues. It is known that the sta-
bility of a linear Hamiltonian system is not asymptotic,
nevertheless the study provides the necessary stability ∂4y ∂2y
EI + mp 2 = fint (x, t) (2)
condition for the original non-linear system. ∂x4 ∂t
Different qualitative analysis of multiparameter linear
systems as well bifurcation theory of eigenvalues can where EI is the bending stiffness of the pipe (N m2 ),
be found in [Seyranian and Mailybaev, 2003; Galin, kg
mp is the pipe mass per unit length ( ) and fint is an
1982; Garcia-Planas, 2008; Garcia-Planas and Tar- m
ragona, 2012; Mediano, 2011; Mediano and Garcia- inside force acting on the pipe.
Planas, 2011; Mediano and Garcia-Planas, 2014].
The aim of the paper is by means of linear Hamil- The internal fluid flow is approximated as a plug flow,
tonian system to model the clamped-pinned pipeline so all points of the fluid have the same velocity U rel-
problem and to analyze the structural stability of the ative to the pipe. This is a reasonable approximation
proposed model. This paper refers to a one end fixed for a turbulent flow profile. Because of that the inside
horizontal pipeline. force can be written as:
The paper is structured as follows. Section 2 presents
a mathematical statement of the problem. Section 3 
is devoted to analyze the stability of linear gyroscopic d2 y 
fint = −mf 2  (3)
system obtained in subsection 2.1. Section 4 presents dt x=U t
and a simulation of the dynamic system using ANSYS
for some different materials used in real cases, such as kg
where mf is the fluid mass per unit length ( ) and U
PVC, Polyethylene and Concrete, in order to validate m
the results obtained analytically. m
is the fluid velocity ( ).
s

2 Preliminaries Total acceleration can be decomposed into local ac-

The system under consideration is a straight, tight and celeration, Coriolis and centrifugal.
of finite length pipeline, passing through it a fluid. The
following assumptions are taken into account in the  [ ( ) ]
analysis of the system: d2 y  d ∂y ∂y dx 
mf 2  = mf + =
i) Are ignored the effects of gravity, the coefficient dt x=U t dt ∂t ∂x dt x=U t
damping material, the shear strain and rotational [ ( ) ]
d ∂y ∂y 
inertia = mf +U =
ii) The pipeline is considered horizontal dt ∂t ∂x x=U t
iii) The pipe is inextensible [ 2 ]
∂ y ∂2y 2
2∂ y
iv) The lateral movement of y(x, t) is small, and with = mf + 2U + U
large length wave compared with the diameter of ∂t2 ∂x∂t ∂x2
(4)
the pipe, so that theory Euler-Bernoulli is applica-
The internal fluid causes an hydrostatic pressure on
ble for the description of vibration bending of the
the pipe wall.
pipe.
v) It ignores the velocity distribution in the cross sec-
tion of pipe.
T = −Ai Pi (5)

where Ai is the internal cross sectional area of the pipe

(measured in m2 ) and Pi is the hydrostatic pressure in-
side the pipe (measured in P a).
Finally if by considering that the total acceleration is
equal to the composition of local, coriolis and centrifu-
gal acceleration. The resulting equation describing the
oscillations of the pipe ([Mediano, 2011]):
Figure 1. Pipeline
∂4y ∂2y ∂2y
EI 4
+ (mf U 2 − T ) 2 + 2mf U +
∂x ∂x ∂x∂t
(6)
The equation for a single span prestressed pipeline ∂2y
where the fluid is transported is a function of the dis- (mp + mf ) 2 = 0
∂t
30 CYBERNETICS AND PHYSICS, VOL. 3, NO. 1

π 2π
Making the scalar product by sen ξ and sen ξ, re-
2.1 Linear Hamiltonian System L L
spectively, it can be obtain:
This subsection is devoted to prove that equation (6)
can be written as a linear dynamic system.
L 8
(mp + mf )q̈1 (t) − mf U q̇2 (t)+
2 3
Proposition 2.1. The equation ( )
π4 (mf U 2 − T )π 2
EI 3 − q1 (t) = 0
2L 2L
(8)
∂4y ∂2y ∂2y L 8
EI 4 + (mf U 2 − T ) 2 + 2mf U + (mp + mf )q̈2 (t) − mf U q̇1 (t)+
∂x ∂x ∂x∂t 2 3
∂2y ( )
(mp + mf ) =0 8π 4 4π 2
∂t2 EI 3 − (mf U 2 − T ) 2 q2 (t) = 0
L L

can be written as
The previous equation system can be written as matrix
form like:
Ẋ = AX

M q̈ + B q̇ + Cq = 0
for some specific Hamiltonian matrix.
that corresponds to gyroscopic lineal system:

To find approximate solution to equation (6), the

method used is the Galerkin’s method with two coor- ẍ + Gẋ + Kx = 0 (9)
dinate{function,}that is to say, taking n = 2 with re-
iπ with M −1/2 q = x (we write the variable as x if con-
spect sin x basis defined over a open set
L i=1,2,... fusion it is not possible),
∫L
Ω ⊂ Rn and the scalar product < f, g >= 0 f g the
approximate solution is: ( )
−1/2 −1/2 16mf 0 −1
G=M BM =
L(mf + mp ) 1 0

π 2π
y(x, t) = q1 (t)sen x + q2 (t)sen x ( )
L L 2 K1 0
K = M −1/2 CM −1/2 =
L(mf + mp ) 0 K2

Replacing the solution in the equation (6), it is ob-

tained that: and

π4 (mf U 2 − T )π 2
K1 = EI −
( ) 2L 3 2L
π4 π 16π 4 2π
EIq1 (t) sen x + q2 (t) sen x +
L4 L L4 L
( )
π2 π 4π 2 2π 8π 4 4π 2
(mf U 2 − T ) −q1 (t) 2 sen x − q2 (t) 2 sen x + K2 = EI − (mf U 2
− T )
L L L L L3 L2
( )
π π 2π 2π
2mf U q̇1 (t) cos x + q̇2 (t) cos x +
L L L L
Introducing the vector:
( )
π 2π
(mp + mf ) q̈1 (t)sen x + q̈2 (t)sen x =0 ( ) ( )
L L x x
(7) =
y ẋ + Gx/2
CYBERNETICS AND PHYSICS, VOL. 3, NO. 1 31

and calculating the derivatives of x and y it can be Removing the variable change it is known that:
found ẋ = y − Gx/2, ẏ = ẍ + Gẋ/2 and consider-
ing that ẍ = −Gẋ − Kx and linearizing the system a 8mf
a=
linear Hamiltonian equation is obtained: L(mf + mp )
−64m2f EIπ 4
( ) ( )( ) ( ) b= − +
ẋ −G/2 I2 x x L2 (mf + mp )2 L4 (mf + mp )
= =A .
ẏ G2 /4 − K −G/2 y y 2
(mf U + Ai Pi )π 2

L2 (mf + mp )
It is easy to prove that the matrix A is Hamiltonian −64m2f 16EIπ 4
c= 2 − +
because QA is symmetrical, where Q is the antisym- L (mf + mp )2 L4 (mf + mp )
2 2
metrical matrix: 8(mf U + Ai Pi )π
.
L3 (mf + mp )
( 0 0 1 0
)
0 0 0 1
Q= −1 0 0 0 . So, the pipeline is modeled as a Hamiltonian system.
0 −1 0 0

So, the linear gyroscopic system obtained in (9) pos- 3 Stability

sess the properties of linear Hamiltonian systems. In this section the stability properties of linear dy-
namic systems representing the pipeline is studied.
Also, a detailed explanation of the effect of the stabi-
In order to simplify, the following parameters consid- lization in terms of the bifurcation theory of eigenval-
ered ues is presented.
Concretely, this section is devoted to prove that the
EIπ 4 space of stability can be described in terms of param-
Λ=
L3 eters of the matrix A ([Mediano and Garcia-Planas,
π2 2011; Mediano and Garcia-Planas, 2014]) and to prove
δ = (mf U 2 − T ) that it is an open set whose boundary consists of alge-
L
1 braic varieties of smaller dimension
β= (10)
L(mf + mp ) Theorem 3.1. let (a, b, c) be the parameters of the ma-
trix A in the hamiltonian equation. The set of points
and the matrices G and K are written as: (a, b, c) of stability is determined by

( ) 
0 −1 2a2 − b − c > 0 
G = 16mf β (a2 + c)(a2 + b) > 0 . (14)
1 0 
  (2a2 − b − c)2 > 4(a2 + c)(a2 + b)
1 1
 Λ − 2δ 0 
K = 2β  2 4  (11)
0 8Λ − δ A stable hamiltonian system is characterized by eigen-
L values lying on the imaginary axis. The characteristic
equation of the matrix is:

Therefore, matrix A is: λ4 + (2a2 − b − c)λ2 + (a2 + c)(a2 + b) = 0 (15)

and the eigenvalues

 
0 a 1 0
−a 0 0 1 √ √
A=
 b 0
 (12) λ = ± −2a +b+c± −8a 2b−8a c−2bc+b +c
2 2 2 2 2
(16)
0 a
0 c −a 0

where: that in terms of Λ, δ, β are

a = 8mf β √
√
λ1 ± β λ2
b = −64m2f β 2 − βΛ + βδ λ=± (17)
(13) 2

8
c = −64m2f β 2 − 16βΛ + βδ. with
L
32 CYBERNETICS AND PHYSICS, VOL. 3, NO. 1

if a2 + b ̸= 0, and
( )
8  
λ1 = −256m2f β 2 − 17βΛ + 1 + βδ 0a 0 a(a2 + c)
L 1 0 c − a 2 0 
S2 = 
0 0 a(b + c)


0
λ2 = 65536m4f β 2 + 8704m2f βΛ−
0c 0 −b(a2 + c)
( ) (18)
4096
512 + m2f βδ + 225Λ2 +
L if a2 + c ̸= 0. (Observe that a2 + b and a2 + c can not
( ) ( ) be zero simultaneously because a ̸= 0).
64 16 240 If a = 0 the Jordan form of A is
1+ 2 − δ + 30 −
2
Λδ
L L L
 
00 0 0
1 0 0 0
As we says, the system is stable in Lyapunov’s sense,  
0 0 0 0
if the eigenvalues lie on the imaginary axe and they are
smple or semi-simple. 00 1 0
Taking into account that the values in the system are
know only approximately, the matrix A in the system In this case, a Jordan basis is
can be considered as a family of matrices depending on
parameters a, b, c in a neighborhood of a fixed point  
p0 , that permit us to study the stability border. The 00 0 1
0 1 0 0
point p0 , in which correspond only simple pure imagi- S=
0 0

nary eigenvalues, is always an interior point of the sta- 1 0
bility domain, while the points on the boundary of the 10 0 0
stability domain are characterized by the existence of
multiple pure imaginary or zero eigenvalues, (when the At the points (19) the system have singularities of the
other eigenvalues are simple and pure imaginary). type 04 and the more degenerate 02 02 on the stability
Stability conditions√requires that the roots obtained in boundary. In both cases eigenvalues lie in imaginary
λ1 ± β λ2 axis but they are not semisimple.
(17), λ2 = are real and negative. Impos-
2 Near of these singularities it is possible to find the lest
ing these conditions the stability zone in the parameter
degenerate matrices, for example
space can be determined.
It can be observed that the points p = (a, b, c) such
that  
0 ε 1 0
} −ε 0 0 1
2a2 − b − c = 0 A(ε) =  
(a2 + c)(a2 + b) = 0
, (19)  0 0 0 ε
0 0 −ε 0
the characteristic polynomial is λ4 ,
The set (19) corresponds to the union of parame- where the eigenvalues are ±εi, and the stable case:
terized curves φ(α) = (α, 3α2 , −α2 ) and φ(α) =
(α, −α2 , 3α2 ). In the intersection can be found the  
most degenerate case, with respect the algebraic struc- 0 −0.1 1 0
 0 1 
ture of the system as it can be seen below. e =  0.1
A
0 
−0.0001 0 0 −0.1
If a ̸= 0 the Jordan form of A is
0 −0.0001 0.1 0
 
0 000
1 0 0 0 where the eigenvalues are 0 + 0.1100i, 0 − 0.1100i,
 
0 1 0 0 0 + 0.0900i, 0 − 0.0900i.
0 010 Following the analysis of eigenvalues, also it is ob-
tained the eigenvalue 0 at the points such that (a, b, c)
A Jordan basis transforming the matrix in its reduced
}
form is (a2 + c)(a2 + b) = 0
(20)
  2a2 − b − c ̸= 0
1 0 b−a 2
0
0 −a 0 −a(a 2
+ b)
S1 = 
0 b

At the points (a, b, −a2 ) there are two possibilities de-
0 c(a − b) 
2

0 0 −a(b + c) 0 pending on b if it is equal or not to −a2

CYBERNETICS AND PHYSICS, VOL. 3, NO. 1 33

For b ̸= −a2 the Jordan form is The last case

  
01 0 0 2a2 − b − c > 0 
0 0  (a2 + c)(a2 + b) > 0 (22)
 √ 0 0  (2a2 − b − c)2 > 4(a2 + c)(a2 + b)

0 0 −3a2 + b 0 
√
00 0 − −3a + b
2
determined the stability points (a, b, c) remaining
within the area bounded by the above singularities.
For b = −a2 the Jordan form is
 
00 0 0 4 Stability of the Pipes: Case Study
0 0 0 0 
 .
0 0 2ai 0  Taking as a constant parameters L = 1000mm, I =
00 0 −2ai 2, 185 · 106 , Ai = 2500π due to the geometry of the
Tn
pipe and mf = 2, 5π · 10−6 assuming the fluid is
This case correspond to a stability point because of mm
water. It is also supposed that the study is applied to
all eigenvalues are semisimple and lie in the imagi-
the inside wall of the pipe so U at these points are zero.
nary axe. It is important to note that in this case the
reduced form is not structurally stable (a small pertur- Therefore the values a, b y c are:
bation makes that the double eigenvalue bifurcates into
two distinct eigenvalues or into a double nonderogatory 2π · 10−8
eigenvalue of type 02 . a=
(2, 5π · 10−6 + mp )
By symmetry, at the points (a, −a2 , c) there are two
cases depending on c be equal or not to −a2 −4 · 10−16 π 2 2, 185 · 10−6 Eπ 4
b = − +
For c ̸= −a2 the Jordan form is (2, 5π · 10−6 + mp )2 (2, 5π · 10−6 + mp )
2, 5 · 10−3 Pi π 3
  (2, 5π · 10−6 + mp )
0 1 0 0
0 0√ 0 0  −4 · 10−16 π 2 34, 96 · 10−6 Eπ 4
  c = − +
0 0 −3a + c √ 0
2  −6
(2, 5π · 10 + mp ) 2 (2, 5π · 10−6 + mp )
0 0 0 − −3a2 + c 2 · 10−5 Pi π 3
(2, 5π · 10−6 + mp )
For c = −a2 the Jordan form coincides with the case
b = −a2 . That permit to obtain the following relation depending
Analogously, the case c = a2 is out of the stability only on mp , E and Pi :
space
For the case b ̸= −a2 and c ̸= −a2 the system have
singularities of the type 02 in the boundary of stability. 16 · 10−13
+ 37.145 · 10−3 π 2 E − 2.52πPi > 0
It remains to study the case that no eigenvalue is zero 2.5 · 10−6 π + mp
The roots of µ2 +(2a2 −b−c)µ+(a2 +c)(a2 +b) = 0,
are real and negative when 15.27752 · 10−4 π 2 E 2 + Pi2 − 17.48874 · 10−1 πEPi > 0

 ( )2
16 · 10−13
2a2 − b − c > 0  + 37.145 · 10−3 πE − 2.52Pi >
2.5 · 10−6 π + mp
(a2 + c)(a2 + b) > 0 (21)

(2a2 − b − c)2 ≥ 4(a2 + c)(a2 + b) 4π 2 (76.3877 · 10−6 π 2 E 2 − 87.4437 · 10−3 πEPi +

In the case (2a2 − b −√c)2 = 4(a2 + c)(a2 + b) the 5 · 10−2 Pi2 ).

eigenvalues are λ = ±i 2a2 − b − c = ±iω double. (23)
It is easy to observe that rank (A − (±iω)I) = 3 so
the Jordan form is This study is done to show the stability of pipes with
different materials assuming in all of them that the fluid
  transported is water and causes a constant pressure on
iω 1 0 0
0 iω 0 0  its walls of 4 bar. The geometrical conditions of the
 
0 0 −iω 1  pipe are the inside diameter equal to 50 mm and the
0 0 0 −iω thickness of the pipe which is 6 mm. The materials
chosen are PVC, Polyethylene and Concrete.
The values of E and mp of the PVC pipe are:
At the points (a, b, c) with 2a2 − b − c > 0, (a2 +
c)(a2 + b) > 0 and (2a2 − b − c)2 = 4(a2 + c)(a2 + b) N
the system have singularities of the type ±iω 2 . E = 30, 581
mm2
34 CYBERNETICS AND PHYSICS, VOL. 3, NO. 1

Tn
mp = 2, 76 · 10−6
mm

Applying the inequalities (23) it is found that the so- 1

−8
x 10

lution is unstable. 0.8

0.6

The values of E and mp of the PE pipe are: 0.4

250
0.2

0 200

0.2
150
0.25

N 0.3
100

E = 9, 174 0.35

0.4

mm2
50
0.45

0.5 0

Figure 3. Representation for values of E less than 250 N/mm2

Tn
mp = 1, 91 · 10−6
mm

Applying the inequalities (23) it is found that the so-

To perform the analysis we have seen that we had to
lution is unstable.
find the Jordan reduced form of the matrix for each
The values of E and mp of the Concrete pipe are:
particular case because as it is well known the obten-
tion of its continuous invariants is not stable when they
N are not simple and that these values are inevitable for
E = 221, 203
mm2 small disturbances. One way to address these kind of
problems is analyzing families of elements dependent
on parameters, defined in a neighborhood of the given
element. In this approach versal deformations can be
Tn
mp = 4, 40 · 10−6 constructed. A versal deformation represents the most
mm general family possessing (in some sense) the proper-
ties of all deformations of the element ([?]).
Applying the inequalities (23) it is found that the so- Let us consider the Lie algebra sp(2k, ( C) =) {A ∈
lution is stable. M4 (C) | QA + At Q = 0} with Q = −I0 2 I02 of the
So, the eigenvalues obtained for this materials are Hamiltonian matrices under the action α of Lie group
G = {P ∈ Gl(n; C) | P t QP = Q} defined in the
form α(P, A) = P −1 AP . A miniversal deformation
Material Eigenvalues is given A + TA O(A)⊥ where O(A) is the orbit of A
PVC 48.042 48.042 98.936i 98.936i with respect the action and ⊥ is the orthogonal with re-
PE −54.549 54.549 56.341i 56.341i spect the standard scalar product. It is easy to compute
Concrete 0.362i 0.362i 2.479i 2.479i T O(A)⊥ = {X | X t A − AX t = 0}.
Although the stability analysis for the Hamiltonian
Figure 2 shows the distribution of these values in the
matrix have been made without having a miniversal
complex plane.
deformation, this method can be used for higher-order
Observe that the case of PVC pipe is the furthest away
matrix symplectic dependent on an arbitrary number of
from stability zone.
parameters.
In fact, if it is considered the miniversal deformation
in a neighborhood of a most degenerate case we ob-
6000
Valores Propios de los Materiales

PVC
tain the following four parameter familiy of Hamilto-
PE
Hormigón
Acero
Aluminio
nian matrices
4000

 
2000
0 −p1 1 0
p1 1 
Parte Imaginaria

0 0
A(p) = 
p2

−p1 
0

p3 0
p3 p4 p1 0
−2000

−4000

−6000
−60 −40 −20 0 20 40 60
that restricted to the particular setup of systems describ-
Parte Real
ing the pipe, is p3 = 0 and A(p) coincides with the
matrix A in (12).
Figure 2. Values of the different pipes

4.1 Simulation for Some Specific Materials

In this section the equation (6) is solved and the struc-
tural stabilities found in the previous section with the
CYBERNETICS AND PHYSICS, VOL. 3, NO. 1 35

stability of the solution is compared. Moreover, vibra-

tion characteristics of a pipe conveying fluid is calcu-
lated using a Finite Element package called ANSYS.
To determinate the vibration characteristics modal
analysis have been used, with this analysis you find
natural frequencies and mode shapes which are impor-
tant parameters in the design of a structure for dynamic
studies.
The simulation of the problem varies depending on the
boundary conditions. On the left side of the pipe is a
rigid support, while the side right lateral displacement
can not but have a time proportional to the angle of
rotation of the pipe. So, the boundary conditions at
ends of a clamped-pinned pipe are given as:

Figure 5. First shape of PVC

y(0, t) = 0, y(L, t) = 0
∂ 2 y(0, t) ∂ 2 y(L, t) ∂y(L, t)
= 0, EI 2
= −Krs
∂t ∂x ∂x
(24)

where Krs is the stiffness of the rotational spring at the

right end.

Figure 6. First shape of Polyethylene

Figure 4. Boundary conditions

because is zero) and the biggest one is the Polyethylene

pipe (19, 435 Hz) but the greater displacement of x axis
The following table lists the values of natural fre-
is the PVC pipe. This combination result in instabil-
quency and displacement in both cases:
ity of Polyethylene and PVC pipe whereas in Concrete
pipe is stable.
First shape Second shape
f dx f dx
Concrete 0 0.034258 0.023795 0.022697 As seen in picture 6, 7 and 8 the lowest natural fre-
PE 19.435 36.609 19.668 38.279 quency is the concrete pipe (0,0238 Hz) and the biggest
PVC 9.806 50.513 24.043 31.546 one is the PVC pipe (24, 043 Hz) but the greater dis-
placement of x axis is the PE pipe. This combina-
Remark that the frequencies obtained solving linear tion result in instability of Polyethylene and PVC pipe
Hamiltonian system does not coincides with frequen- whereas in Concrete pipe is stable.
cies that can be obtained solving the second order dif-
ferential equation (6) with initial conditions (23), in
fact with Hamiltonian equation is studied structural sta- 5 Conclusion
bility giving information about the qualitative changes In this paper the classic non-linear dynamic model for
that can be in the behavior of systems when the systems a pipe conveying fluid have presented. Moreover, a lin-
are known only approximately. earization method have been done by approximation of
In the following pictures it is shown the performance the non-linear system to the linear gyroscopic system
of the first and the second shapes and the natural fre- (Garlekin’s method) providing a Hamiltonian system.
quencies of them. From the linear system, the stability of the pipeline has
As seen in picture 3, 4 and 5 the lowest natural fre- been analyzed in a general form by means of the first
quency is the concrete pipe (it is not write on the picture Lyapunov’s methods. The stability generalization of
36 CYBERNETICS AND PHYSICS, VOL. 3, NO. 1

Figure 7. First shape of Concrete Figure 9. Second shape of Polyethylene

Figure 8. Second shape of PVC Figure 10. Second shape of Concrete

the system have been done obtaining the stability lim- tems representing fixed speed wind turbine. Cybernet-
its as function of the material and fluid parameters. ics and Physics Journal, 2(2), pp. 77–83.
It have been shown that the dynamics and stability of Galin, D.M. (1982). Versal deformations of linear
pipes conveying fluid not only depends on the boundary Hamiltonian systems. AMS Transl. 118, (2), pp. 1–
conditions but it is also strongly important the material 12.
of the pipe and the pressure produced by the fluid. Garcia-Planas, M.I. (2008). Bifurcation diagrams of
The proposed Hamiltonian model of a pipeline con- generic families of singular systems under propor-
veying fluid provides a simple case to evaluate if the tional and derivative feedback. WSEAS Transactions
selected materials will remain stables. on Mathematics. 7(2), pp. 48–58.
In this paper the calculations of the proposed model Garcia-Planas, M.I. and Tarragona, S. (2012). Anal-
and the simulation of a complex model of typical ma- ysis of behavior of the eigenvalues and eigenvectors
terials for a pipe used in public works have been com- of singular linear systems. WSEAS Transactions on
pared to verify the results obtained. Mathematics. 11, pp. 957–965.
Gregory, R.W. and Paı̈doussis, M.P. (1966). Unsta-
ble oscillations of tubular cantilevers conveying fluid-
References parts I & II. Proc Royal Soc. 293, pp. 512–542.
Abid-Al-Sahib, N.K., Jameel, A.N. and Abdla- Kameswara, R.C. and Simha, H.S. (2008). Vibrations
teef, O.F. (2010). Investigation into the vibration of Fluid-Conveying Pipes Resting on Two-parameter
Characteristics and Stability of a Welded Pipe Con- Foundation, The Open Acoustics Journal, 1, pp. 24–
veying Fluid. Jordan Journal of Mechanical and In- 33.
dustrial Engineering. 4(3), pp. 378–387. Kerboua, Y. and Lakis, A.A. (2008). Dynamic Be-
Dominguez-Garcia, J.L. (2013). Computing bounds haviour of Plates Subjected to a Flowing Fluid.
for the distance of functional output-controllable sys- WSEAS Transactions on Fluid Mechanics. 3(2),
CYBERNETICS AND PHYSICS, VOL. 3, NO. 1 37

pp. 101–115. Morand, H.J.P. and Ohayon, R. (1995). Fluid struc-

Kuiper, G.L. and Metrikine, A.V. (2004). On stabil- ture interaction: applied numercial methods. Hard-
ity of a clamped-pinned pipe conveying fluid. Heron, bound, John Wiley & Sons.
49(3), pp. 211–232. Paı̈doussis, M.P. (1998). Fluid-structure interactions:
López Reyes, L.J. and Kurmyshev, R. (2013). Reso- slender structures and axial flow, Academic Pr.
nance instability of nonlinear vibrations of a string Paı̈doussis, M.P. and Issid, N.T. (1974). Dynamic sta-
under harmonic heating. Cybernetics and Physics bility of pipes conveying fluid. J Sound Vibrat, 33,
Journal, 2(3), pp. 165–168. pp. 267–294.
Mailybaev, A.A. (2001) Transformation to versal de- Paı̈doussis, M.P., Tian, B. and Misra, A.K. (1993).
formations of matrices. Linear Algebra and its Appli- The dynamic and stability of pinned-clamped coax-
cations. 337 pp. 87–108. ial cylindrical shells conveying viscous flow. Proceed-
Mediano, B. (2011). Análisis y simulación del com- ings of the Canadian Congress of Applied Mechanics,
portamiento de una tuberı́a mediante MEF. Master CANCAM 93, pp. 259–260.
Thesis. UPC. Seyranian, A.P. and Mailybaev, A.A. 2003. Multipa-
Mediano, B. and Garcia-Planas, M.I. (2001). Stability rameter Stability Theory with Mechanical Applica-
of a pipeline hydraulic fluid with one end fixed. IPACS tions. World Scientific, Singapore.
Electronic library. Proc. 5th International Conference Stoicuta, O., Silviu Nan, M., Dimirache, G., Buda, N.,
on Physics and Control (Physcon 2011). and Liviu Dandea, D. (2010). Research regard-
Mediano, B. and Garcia-Planas, M.I. (2004). Stabil- ing the use of new systems to control fluid flow in
ity Analysis of a Clamped-Pinned Pipeline Conveying pipelines WSEAS Transactions on Fluid Mechanics.
Fluid, WSEAS Trans. on Systems, 13, pp. 54–64 5(3), pp. 144–154.
Misra, A. K. Wong, S. S. T. and Paı̈doussis, M. Thompson, J.M.T. (1982). Instabilities and Catastro-
P. (2001). Dynamics and stability of pinned- phes in Science and Engineering. Wiley, New York.
clamped and clamped-pinned cylindrical shells con-
veying fluid. Journal of Fluids and Structures 15(8),
pp. 1153–1166.