You are on page 1of 14

DM MS205

Élèves :
Vinícius SIMIÃO PITTA
Diogo SANTOS GIMENEZ
Responsable: Olivier Doaré

Palaiseau, March 31, 2024

1
Contents

1 Introduction 3

2 Modal Decomposition 3

3 Reducing to an order one differential equation 4

4 Harmonic Solution 4

5 Stability Analysis 5

5.1 Types of instability . . . . . . . . . . . . . . . . . . . . . . . . . . 5

5.2 Clamped-Free Case . . . . . . . . . . . . . . . . . . . . . . . . . . 6

5.3 Double Clamped Case . . . . . . . . . . . . . . . . . . . . . . . . 9

6 Effect of Dampening 11

6.1 The expression of c . . . . . . . . . . . . . . . . . . . . . . . . . . 14

7 Bibliography 14

2
1 Introduction

In this report, we will analyze the problem of the fluid-conveying pipe with one
side clamped and the other free. We know that the dynamics of this problem
are modeled by the following dimensionless equation:

∂4w ∂2w 2
2∂ w 1 ∂2w
+ + u + 2β 2u = 0. (1)
∂x4 ∂t2 ∂x2 ∂t∂x
µf √
Where β = µf +µ s
, u = Ur M RL0 , M = µf /µs and w is the adimensional
vertical displacement of the pipe.

The boundary conditions of this problem are:

∂w ∂3w ∂2w
w(x = 0) = = = = 0.
∂x x=0 ∂x3 x=1 ∂x2 x=1

2 Modal Decomposition

This equation has no analytical solution, thus, we will solve it numerically by


expanding the solution on N eigenmodes of the beam without gravity:

N
X
w= qj (t)ϕj (x)
j=1

where ϕj (x) are the beam eigenmodes, given by


ϕj (x) = cosh λj x − cos λj x − σj (sinh λj x − sin λj x) ,
λj and σj being:
sinh λj − sin λj
cos λj cosh λj + 1 = 0 , σj = .
cosh λj + cos λj

And we introduce this decomposition into the equation and project into a
mode ϕi obtaining:

X p X
q̈i + λ4i qi + u2 qj cij + 2 βu q̇j bij = 0,
j j

with Z 1 Z 1
bij = ϕi (x)ϕ′j (x) , cij = ϕi (x)ϕ′′j (x).
0 0

3
Or, in matrix form :

⃗q¨ + 2 βuB ⃗q˙ + K0 + u2 C ⃗q = 0,


h p i
(2)
 

where K0 is the stiffness matrix of the system without fluid with K0ij = λ4i δij ,
the coefficients of the matrices B and C are given above, and ⃗q writes:
 
q1
 q2 
⃗q = 
 
 . 

 . 
qN

3 Reducing to an order one differential equation

To solve this problem, we can write the order one differential equation satisfied
by the following vector:

 
⃗q
⃗z =
⃗q˙

Such that :

⃗z˙ = A⃗z

By manipulating the equation (2) we arrive at:

⃗q˙
    
O √I ⃗q
=
⃗q¨ −K0 − u2 C −2 βuB ⃗q˙

4 Harmonic Solution

We can suppose a solution in the form ⃗z = Ze


⃗ νt , reducing to an eigenvalue
problem:

⃗ = AZ
νZ ⃗

The eigenvalue of A has the form ν = ξ +ωi. Its imaginary part (ω) is related
to the oscillation of the pipe while the real part (ξ) is related to its amplitude,

4
if ξ is negative, the amplitude will decrease over time, however, if ξ is positive,
it will increase over time, thus, the system is unstable.

5 Stability Analysis

Right now we will be analysing how the different values of β influence the critical
value of the flow’s velocity. The idea is that β represents the ratio between the
density of the fluid with the sum of the fluid‘s and structure‘s density, that being
it reflects the relative influence of the fluid’s mass when compared to the total
mass of the system, influencing its dynamics. In our analysis we have 3 cases:

• β = 0: for this case we have that the fluid’s density is way smaller than
the structure’s (µs ≫ µf ), so we have that the dynamics of the system
can be understood as the dynamics of the structure. In this case we have
that the instability comes more from the structure’s dynamics itself than
the fluid-structure interaction

• 1 > β > 0: as the value of β increases we have that the fluid’s density
becomes greater in comparison with the structure’s density. So, the fluid-
structure interaction’s contribute more and more to the dynamics of the
system.
• β = 1: here we have that the fluid’s contribution is greater than the struc-
ture’s, so the system dynamics is mainly dictated by the fluid’s dynamics.

5.1 Types of instability

We know the system becomes unstable with a certain value ucrit , let’s analyse
the origin of this instability. There are two types, buckling instability, which
originates when the stiffness of the system becomes zero, and flutter instability,
which appears when there is a negative damping in the system.

Looking closer to our system, its total stiffness is:

Kt = K0 + u2 C

We can see that the fluid flow interfere with the total rigidity, and once it
becomes negative we have buckling instability.

The total damping is:

Ct = 2 βuB
p

5
Here, the damping of the system is only due to the presence of flow, if this
damping is negative, the fluid gives energy to the structure instead of dissipate
it, the flutter instability occurs.

Also, flutter instability is associated with oscillation, while buckling isn’t,


so, a direct way of identifying it is by looking at the imaginary part of the
eigenvalue whose real part became positive. If the imaginary part is zero, there
is no oscillation associated with this mode for this velocity, so we have buckling
instability, otherwise, we have flutter instability.

5.2 Clamped-Free Case

So, for our case with one end clamped and the other free, by using the Jupyter
notebook we have the following results:

Figure 1: Real and imaginary part of the eigenvalues in function of the fluid’s
velocity for the clamped-free case for β = 0

6
Figure 2: Real and imaginary part of the eigenvalues in function of the fluid’s
velocity for the clamped-free case for β = 0.1

Figure 3: Real and imaginary part of the eigenvalues in function of the fluid’s
velocity for the clamped-free case for β = 0.5

7
Figure 4: Real and imaginary part of the eigenvalues in function of the fluid’s
velocity for the clamped-free case for β = 0.9

Figure 5: Real and imaginary part of the eigenvalues in function of the fluid’s
velocity for the clamped-free case for β = 1

By analysing these graphics we obtain the following values for the critical ve-
locity (the value when the real part of a mode is positive) and type of instability
(if there is a imaginary part associated with this mode for this velocity).

8
β ucrit (m/s) Type of instability
0 4.47 flutter
0 82 Buckling
0.1 4.75 flutter
0.5 9.32 flutter
0.9 [14.3, 15.4] flutter
0.9 [17, 1, ∞[ flutter
1 [74.2, 75.4] flutter
1 [77.3, ∞[ flutter

Table 1: ucrit for different β values

Analysing the data from Table (2) we have that as the influence of the
fluid increases, the critical velocity of the fluid also increases. When looking
extreme cases ( β = 0 and β = 1) we can see that the jump between them
and the coupled case (where both solid an fluid dynamics influence the system’s
dynamics) is different for them. From β = 0 to β = 0.1 we have only a small
increase in the critical velocity, but when we look at the jump from β = 0.9
to β = 1 the value of the critical velocity increased five fold. This indicates
that the biggest source of instability comes from the structure and not from
the fluid. Besides that, from analysing the imaginary part of eigenvalue we can
identify that we have a flutter instability since it is not zero, which indicates
its oscillatory nature. It is interesting that for the case of β = 0, we have that
the imaginary component of the eigenvalue converges to 0. This means that as
the velocity of the fluid increases the instability goes from a flutter regime to
a buckling regime. It is also worth noting that in the cases where β = 0.9 and
beta = 1 we have that if we increase the velocity after entering the instability
we go into a zone of stability, before going back into instability. This might
alter the fluid-structure interaction dynamics in such a way that the system
finds a new equilibrium or damping mechanism. For instance, changes in the
flow pattern around the structure might reduce the lift coefficient or introduce
a damping effect not present at lower velocities, leading to stabilization and all
of this might be responsible for this phenomena.

5.3 Double Clamped Case

Using the same methods to realize our analysis we have the following behaviour:

9
ucrit (m/s) Type of instability
[6.25, 8.99] Buckling
]8.99, 9.29[ Stability
[9.29, 11.73[ Flutter
[11.73, 21.14[ Buckling
[21.14, 65.49[ Flutter
[65.49, ∞[ Buckling

Table 2: The regimes of the clamped clamped beam with β = 0.5

Figure 6: Real and imaginary part of the eigenvalues in function of the fluid’s
velocity for the double clamped case for β = 0.5

In comparison to the |Figure X, which show the behaviour for the same β
value for the original case, we have the appearance of a new instability zone
for the velocities between 6.25m/s and 8.99m/s. This means that our system
will be unstable at lower velocities than at our original case, also, the type
of instability is different, we have alternating regimes of buckling and flutter
instability.

Another thing that is important noting is that when the dynamics of the
double clamped system consisted of small and fast vibrations, this can lead to
fatigue in the pipe, which can demand more maintaining and in critical cause
structural failure.

Here we can see that at the first instability regime we have a eigenvalue with
a null imaginary part, this means that between the velocities of 6.25m/s and
8.99m/s we’ll have a case of buckling instability and after m/s we have a case
of flutter instability.

10
6 Effect of Dampening

Now we’ll repeat our stability analysis but thus time adding the dampening
effect to our system, in this case we used a dampening coefficient of c =
25Kg/m2 .s.This gives us the following:

Figure 7: Real and imaginary part of the eigenvalues in function of the fluid’s
velocity for the clamped-free case with dampening, for β = 0

Figure 8: Real and imaginary part of the eigenvalues in function of the fluid’s
velocity for the clamped-free case with dampening, for β = 0.1

11
Figure 9: Real and imaginary part of the eigenvalues in function of the fluid’s
velocity for the clamped-free case with dampening, for β = 0.5

Figure 10: Real and imaginary part of the eigenvalues in function of the fluid’s
velocity for the clamped-free case with dampening, for β = 0.9

12
Figure 11: Real and imaginary part of the eigenvalues in function of the fluid’s
velocity for the clamped-free case with dampening, for β = 1

β ucrit (m/s) Type of instability


0 [4.1,82[ Flutter
0 [82, ∞[ Buckling
0.1 6.8 Flutter
0.5 11.49 Flutter
0.9 18.35 Flutter
1 [27.8, 28.9] Flutter
1 [30.9, 32.2] Flutter
1 [33.7, 34.9] Flutter
1 [36.8, 38.18] Flutter
1 [39.9, 40.95] Flutter
1 [43.23, 44.07] Flutter
1 [77.46, 80.9] Flutter
1 [81.64, 85.8] Flutter
1 [85.9, ∞[ Flutter

Table 3: ucrit for different β values in the case with c = 25Kg/m2

Here we can see that the damping actually lower the critical velocities for
the instability to occur. This indicates that the damping effect actually has
a destabilizing effect in our system, which reduces the range of the velocities
that we can use. We can also see that in the case of β = 0, as we increase the
velocity, the instability transitions from a flutter regime into a buckling regime.
With this, we have that it is a bad idea

13
6.1 The expression of c

We can deduce the expression of the dimensionless damping by looking back at


the dimensional equation:

∂2W ∂4W ∂W
µs 2
+ EI 4
+C = f ey
∂t ∂X ∂t

Where µs is the mass per unit length, E is the Young’s modulus and I is
the moment of inertia of the section of the pipe, f is the force exerted by the
fluid on the pipe.

We know the form of the dimensionless equation:

∂2w ∂4w ∂2w 1 ∂2w ∂w


2
+ 4
+ u2 2 + 2β 2 u +c =0
∂t ∂x ∂x ∂x∂t ∂t

So, by defining W = R0 w, where R0 is the radius of the pipe, and T = τ t,


with τ = Ro2 µs /EI, we can rewrite the first expression as:
p

µs Ro ∂ 2 w EI ∂ 4 w CRo ∂w
+ 3 + = f ey
τ 2 ∂t2 Ro ∂x4 τ ∂t

Thus, we can state that

CRo τ 2 CR0
c= =√
τ µs R0 µs EI

Conclusion

We can conclude that the pipe conveying fluid prov

7 Bibliography

14

You might also like