Proceedings of the ASME 2013 Pressure Vessels & Piping Conference

PVP2013
July 14-18, 2013, Paris, France

PVP2013-97676

APPLICATION OF THE TRANSFER MATRIX METHOD TO THE ANALYSIS OF

HYDRO-MECHANICAL VIBRATION OF NPP PIPING
 

Igor Orynyak Sergii Radchenko Iaroslav Dubyk

“IPP-CENTRE” LTD “IPP-CENTRE” LTD “IPP-CENTRE” LTD
Kiev, Ukraine Kiev, Ukraine Kiev, Ukraine
[email protected] [email protected] [email protected]
 

ABSTRACT level and may cause significant mechanical vibrations in

The transfer matrix method (ТММ) was used for pipelines. These phenomena are called the coupled hydro-
description of harmonic vibrations of piping with transported mechanical vibrations (HMV). An interaction between the
medium. Apart from 12 well-known mechanical parameters dynamic behavior of the fluid (a liquid or gas) and the
which characterize the state of piping system in each cross mechanical structure occurs through superposition of three
section two additional parameters that characterize the vibration mechanisms [1, 2].
of the medium, namely its translation and pressure pulsations Junction Coupling. This effect prevails over the others and
were considered. The solution of these equations, which take is due to a pressure differential between specific points of the
into account the Poisson contraction of the pipe wall, in the piping system, such as tees, bends, tapered reducers, and end
form suitable for the transfer matrix method application was plugs. The pressure differential leads to a resulting hydraulic
derived. The biggest uncertainty in the analytical modeling is to force that makes the pipe points move, which in turn has an
adopt the boundary conditions for above mentioned 2 effect on the relative motion of the fluid and its pressure.
parameters for the considered piping section. To solve this Poisson Coupling. This effect may only occur due to a
problem of identification of the most probable induced non-zero value of Poisson’s ratio . Pressure waves give rise to
frequency we developed the technique of choosing such radial stresses and strains in the pipe wall, which are related to
boundary conditions at which the maximum of energy is axial strains via the Poisson effect. Furthermore, the axial
confined within the considered piping section. The validity of stresses in the pipe lead to hoop strains and result in a reduced
the approach was tested on some analytical examples. This cross-sectional area and thus in a changed deformations
method was used to analyze the forced vibration of the second (pressure) in a liquid (medium). Coupling equations for straight
circuit loop of unit 1 Zaporizhia Nuclear Power Plant (ZNPP) pipe runs were first derived in [3]. They essentially imply that
with VVER-1000 (from Russian: Vodo-Vodyanoi instead of two separate pairs of equations for the axial
Energetichesky Reactor; Water-Water Power Reactor) arising deformation of a straight pipe run (force and displacement) and
from turbulent eddies in the flow of steam. Natural frequencies equations for the deformation of a liquid (pressure and
and forms of mechanical, hydrodynamic, and related hydro- displacement) four interrelated equations are written.
mechanical vibration were found, a number of Friction Coupling. This effect takes place due to a bilateral
recommendations were given to reduce the vibration levels. force of friction of a liquid against a pipe wall. For industrial
piping systems the real pipe surface roughness parameters are
small and this effect can be disregarded.
INTRODUCTION The practical analysis of the hydro-mechanical interaction
comes down to three possible cases [4]. The first case is a
A gas or a liquid running in pipelines is an elastic medium
trivial one: there occurs no interaction and the liquid can be
and may experience, much like the pipe material, vibration
considered as an added mass only. The second case is an
processes which are usually called the hydrodynamic or
intermediate one: vibrations are excited mainly by fluid surges,
acoustic processes. As this takes place, the fluid vibrations may
and the analysis of behavior of the mechanical system is
propagate to longer distances because of a fairly low damping

1 Copyright © 2013 by ASME

performed by specifying the appropriate forces at the junctions. where F0 is the pipe bore cross-sectional area in the case of no
This approach has been substantiated for yielding (easily
compressible) fluids (gases) or a rigid mechanical system. Its fluid pressure action, F0  R 2 ; F1 is the pipe bore cross-
drawback is that the problem for the fluid has to be solved sectional area under the fluid pressure action, dx is the pipe
separately and the method of applying hydraulic forces is run length, and du f is the fluid displacement increment (Fig.
inconvenient (and has to be additionally substantiated). The
1).
third case implies a complete solution of the coupled problem,
which is not always easy to accomplish. Moreover, the Considering the radial displacement of pipe points w p and
equations that govern the solution for the Poisson coupling are the change of the pipe cross-sectional area we have
usually written in the form which leads to void results when Pp R 2
Poisson’s ratio is zero and the natural frequencies of F1  F0  2  w p R, w p  , (3)
mechanical and hydrodynamic vibrations are equal. hE
Our opinion is that in the case of numerical analysis of the where h is the pipe wall thickness, R is the inner radius, and
piping behavior it would be unpractical to split the partial or E is the Young modulus of the pipe material.
full HMV interaction or absence thereof. Clearly, the Substituting (3) into (2), in view of (1), we arrive at a
representation of the resulting equations in the form that allows differential expression for the fluid deformation. When
a continuous solution for any value of Poisson’s ratio is of combined with more trivial equations of force balance for the
practical importance. fluid and pipe and the physical law of strains for the pipe
In this paper, formed a general algorithm for solving the material, this expression forms the following set of Skalak
related hydro-mechanical vibrations (using matrix transfer equations [3]:
method) considering Poisson and junction coupling. Introduced
energy approach to the definition of the boundary conditions in uf Wp
the case of calculating the part of a complex branching pipeline.
These algorithms and calculation methods used to analyze the F1
related hydro-mechanical vibration of the secondary circuit
loop of unit 1 ZNPP with VVER-1000 arising from turbulent Pf F0 Pf+dPf
eddies in the flow of steam.
1. SOLUTION OF THE SKALAK EQUATIONS FOR
STRAIGHT PIPE RUN
According to the Skalak model [3], the friction force dx
between the transported liquid and the pipe inner surface can be
neglected. The model is based on the following postulates: the FIG. 1. ELEMENTARY VOLUME OF A PIPE UNDER
fluid has some bulk elasticity K f (its dimension coincides THE ACTION OF A FLUID PRESSURE
with that for the Young modulus, Pa); the fluid is continuous. du f ( x,  ) Pf  2 RK f 
Thus, two additional parameters – the fluid pressure pulsation  1    N t /( fE ) ; (4a)
dx K f  hE 
Pf and the fluid longitudinal displacement u f – are
dPf ( x, ) d 2u f
introduced into the mathematical model of vibrations of a bar  f ; (4b)
system. Let  f be the fluid volume strain. If v0 is the fluid dx d 2
dvt ( x, )
undeformed volume and v1 is the deformed volume, we can put   N t /( fE )  Pf R /( hE ) ; (4c)
dx
down the expression
v v P dN t ( x, ) d 2 vt
f  1 0  f K , (1) f m , (4d)
v0 f dx d 2
where f is the cross-sectional area of the pipe material,
which follows from the definition of K f . Here, the sign “-”
f  h2 R  h  ;  f / т is the density of the fluid or pipe
indicates that the positive pressures leads to a decrease in
volume, while the negative one to an increase in volume, and material, and  is time. The beam unknown mechanical
Pf is the parameter of pressure variation with respect to some parameters are as follows: N t is the axial force and vt is the
static magnitude and thus may have a negative value. From (1) pipe longitudinal displacement.
it follows that: We assume all the processes to be harmonic ones, i.e.,

f 
 
dx  du f F1  dx  F0
,
proportional to sin  , where  is some frequency of harmonic
vibrations. The same notation will be used for the new
(2)
dx  F0 introduced functions that depend on x only. Differentiation of

2 Copyright © 2013 by ASME

(4a) and (4c) and substitution of (4b) and (4d) into them give of the displacements are known, i.e., vt ( x  0)  vt 0 and
the following system of initial differential equations: u ( x  0)  u 0 , we have
d 2 vt
 vt  a  u f  d ; (5a) vt 0  Xu 0 u 0  Yvt 0
dx 2 A 2
, C . (14)
1   XY 1   2 XY
d 2u f For convenience we introduce the following notation:
 vt  a  u f  d , (5b)
dx 2 du f ~ dv ~
where   Pf ; t   N t , (15)
dx dx
  f  2 RK f  2 where
a  2 т , d  1   ; (6a)
E K f  hE  ~ Pf  2 RK f 
Pf  1    N t /( fE ) ,
RK f K f  hE  (16)
  1 /(1  21 )  1 , 1  . (6b) ~
hE N t  N t /( fE )  Pf R /( hE ) .
A general solution to the system of differential equation is ~ ~
sought for in the form Taking the initial conditions Pf ( x  0)  Pf 0 and
v t ( x)  A cos 1 x  B sin 1 x  CX cos  2 x  ~ ~
; (7a) N t ( x  0)  N t 0 to be known, we have
 DX sin  2 x ~ ~ ~ ~
1 N t 0  XPf 0 1 Pf 0  YN t 0
u f ( x)  AY cos 1 x  BY sin 1 x  B , D  . (17)
, (7b) 1 1   2 XY 2 1   2 XY
 C cos  2 x  D sin  2 x In view of the defined values of A, B, C , D , solutions (7a)
where the values of 1 and 2 are found from the solution of and (7b) become
the characteristic equation  ~
 ~
v t ( x)  v t 0 cos 1 x  F1 ( x)  u f 0  y F1 ( x) 
4  (a  d )2  ad (1   2 )  0 ; (8) ~
 ~ ~
 ~
 N t 0 sin 1 x 1  F2 ( x)  P f 0  y F2 ( x);
(18a)

 
2 2
1,2   (a  d )  (a 2d )  4ad  .
2 (9) ~ ~
u f ( x)  vt 0  x F1 ( x)  u f 0 cos  2 x  F1 ( x) 

Here, the sign “+” points to the first root that tends to a when
~ ~ ~
 ~
 N t 0  x F2 ( x)  Pf 0 sin  2 x  2  F2 ( x) , 
(18b)

  0 (which corresponds to the solution for a mechanical where the following functions are entered for convenience:
system), while the sign “-” indicates the second root that ~
F1 ( x)  cos 1 x  cos  2 x ; (19a)
approaches d when   0 (which corresponds to the solution
~
for hydrodynamic vibrations). It is evident that due to (8) both F2 ( x)  sin 1 x 1  sin  2 x  2 , (19b)
roots satisfy the equality expression which have the respective properties:
 i
2
 2
 2
 a i  d  ad  , i=1, 2. (10) ~ ~
F1 (0)  F2 (0)  0 ; (20a)
The value of X is determined by substituting (7b) into ~ ~
F2 ( x)  F1 ( x) ;
(5a) and the value of Y by substituting (7a) into (5b):
X
d
;Y
a
. (11)
~
2
 ~
F1( x)   1    2 2 sin 1 x 1  F2 ( x) 2 2 .
(20b)

2   a
2
1 2  d Differentiation of function (18) in view of notation (15) and
We introduce the following notation: properties (20) gives

x 
Y
 
X
 
 2 XY
~
 ~
N t ( x)  vt 0 a sin 1 x 1    2 2 F2 ( x)  
 
; y ; . (12) ~
1   2 XY 1   2 XY 1   2 XY u f 0  d sin 1 x 1   y  2 2 F2 ( x)  (21a)

form
In view of (11) and (10) expressions (12) are written in the ~
 ~ ~
 ~
 N t 0 cos 1 x  F1 ( x)  Pf 0  y F1 ( x);
x 
a
1 2  2 2
; y 
d
1 2  2 2
;
~
 ~
Pf ( x)  vt 0 a sin 1 x 1   x  2 2 F2 ( x)  
1 2  a  2 2  d .
(13)  ~
u 0 d sin  2 x  2   1 2 F2 ( x)   (21b)
 
1 2  2 2 1 2  2 2
~ ~ ~
 ~
 N t 0  x F1 ( x)  Pf 0 cos  2 x  F1 ( x) . 
The unknown coefficients A, B, C , D are found from initial The advantage of these expressions, in particular over the
conditions (7a) and (7b) with x  0 . Assuming that the values solutions provided in [5], is that for any real values of the pipe
material parameters and transported fluid parameters there is no
division by zero.

3 Copyright © 2013 by ASME

 ~

Expressions (18), (21) set a matrix A that relates the u f ( x)  vt 0 
a ~
F1 ( x)  u f 0 cos 1 x 
~ ~ d
unknown parameters [ vt , u f , N t , Pf ] at the beginning and , (24b)
~ a ~ ~
end of the straight pipe run. To involve these expressions in the  N t 0  F2 ( x)  P f 0 sin 1 x / 1 ;

~
TMM the matrix A should be transformed into a similar ~

d
~

N t ( x)  v t 0 a sin  2 x  2   2F2 ( x)  u f 0 d 
matrix A for the unknown parameters [ vt , u f , N t , Pf ]. Let ~ ~ ~ , (24c)
 sin 1 x 1   N t 0 cos  2 x  P f 0 F1 ( x);
the matrix B  transform [ vt , u f , N t , Pf ] into [ vt , u f ,
~ ~
N t , Pf ] {trivially results from expressions (16)}. Then, the
~ 
P f ( x)  v t 0  a sin 1 x 1  

a2
d
  ~ 
1   2 F2 ( x)  

  ,
expression for A will be as follows: (24d)
~ a ~ ~
  
A  B 1 A~ B . (22)
 u 0 d sin 1 x / 1  N t 0  F1 ( x)  Pf 0 cos 1 x.
d
There are two approaches to constructing the matrix A : Note that this solution can be obtained directly from (5) in

~
(i) compute the matrices A , B  and B 1 during the  the boundary transition with d  a .
Let us analyze expressions (24) for a straight pipe. The
execution of the program and multiply them by each other ~
hydrodynamic components u f and Pf are substituted into
numerically;
~
  
(ii) analytically multiply the matrices A , B  , and B 1 and (24a) and (24c) for pipe components. The peak values of vt
~
use the final expression for A , which is similar to (18), (21), and N t are put into (24b) and (24d), and we have that the
during the TMM programming. hydrodynamic components vary insignificantly (about
Note that the first approach requires some modification of  2 a 1 ) because of the influence of the pipe components
the process of filling the matrix, while the second one calls for
through the Poisson effect. A similar order of error takes place
deriving rather cumbersome expressions for vt , u f , N t and
for the pipe components too. Thus, the solution can be split into
Pf . Using mathematical software packages one can derive, two parts: equations for the fluid vibrations (24b) and (24d) and
equations for mechanical longitudinal vibration of the pipeline
quite fast, explicit expressions for vt , u f , N t , Pf in the form
(24a) and (24c). In this case, the hydrodynamic vibration
suitable for direct TMM programming. frequencies remain unchanged, while the frequencies of the
2. DERIVING THE RELATIONSSHIPS FOR A mechanical longitudinal vibrations vary according to (23b).
STRAIGHT PIPE RUN IN EXTREAM CASES However, for solving real problems it is advisable to use
equations (18), (21), or (24), which take into account the
One of alternative representations of Eqs. (5) is an extreme mutual influence of mechanical and hydrodynamic parameters
case where the values of a and d differ significantly. The case via the Poisson effect. For fittings and bends these parameters
of a  d is unlikely for industrial pipelines; therefore, we will have an essential influence on each other. So, the Poisson effect
restrict our consideration to the case of d  a , which is of cannot be neglected for straight pipes.
practical interest. For this case the following expressions for Note that the parameter c  d  is the inverse of a water
1, 2 ,  x ,  y and  are valid: hammer wave velocity in a fluid and is described by the known
expression derived by Zhukovskii. In this case, if the pipe
1 2  d  ad 2 /(d  a)  d  a 2 ; (23a)
cross-section deformation W p due to the fluid pressure can be
2 2  a  ad 2 /(d  a)  a1   2  ; (23b) disregarded (e.g., for a pipe with high-rigidity walls or for a
fluid of a relatively small volume elasticity K f ), the water
x  ad ; (23c)
hammer velocity will be equal to the sound velocity in the fluid
 a and, therefore, the expression for d will be simplified. For c we
 y   1     ; (23d)
 d have the following trivial expression:
 a 2  1 f
  1   2  2   1 . (23e) c  . (25)
 d d  Vf Kf
Then, instead of expressions (18) and (21) we can put down
~ 3. BOUNDARY CONDITIONS FOR HYDRODYMANIC
v t ( x)  vt 0 cos  2 x  u f 0 F1 ( x )  VIBRATION
~ ~ ~ (24a)
 N t 0 sin  2 x /  2  P f 0 F2 ( x); The governing equations in the form (18), (21), or (24)
should be supplemented with boundary conditions and

4 Copyright © 2013 by ASME

conditions of continuity between element sections for straight Let us now look at the fluid–pipe interaction at bend
pipes and bend elements. elements. Since the fluid is only characterized by
The continuity condition for pipe displacements at a pressure (force) and displacement, the equations for
junction of several piping elements is evident and well-known turning angles and moments will remain unchanged, i.e., in the
and has been described in numerous publications. form (6) [9]. First, we will consider the equations for the

The equation for pressure at a junction point is a trivial one balance of forces. So, the increment of mechanical forces Q p
and based on the condition for pressure continuity at the point:
and the pressure-induced force Q is given by
Pf1  Pf2  Pf3  ... . (26a) 
  dQ p  
Note that condition (26a) is valid both for the junction of Q    P f F0 t ( s )  R  0 (28)
straight pipes and for the bend element as a place where the  ds 
 
local coordinate axes turn [6-9].
Since the pressure is coordinate-independent (the bend
We consider a model where a transported fluid vibrates
element is assumed to have an infinitesimal length), system
longitudinally (along the pipeline axis) with an amplitude u f . (28) is reduced to the form of (5) [9], except for the second
It is assumed that the fluid fully takes the form of the pipe. The equation:
fluid transverse displacements, i.e., in the directions normal to dQn / d  Qt  Qk  P f F0  0 . (29)
the pipe axis the fluid is allowed for as an added mass only.
Based on the condition for the fluid continuity at a point, in where the notation   R / B a ratio of normal radius to
view of the mutual longitudinal displacement of the pipe and binormal radius.
fluid, we put down the following expression: A solution to system (5) [9] will be represented, in view of
(30), by the modified solution (6) [9]:
 F u   F u 
i j

i
i
f  vt i  j
j
f  vt j , (26b) Qk ,0  Qt ,0  Pf F0
вх вых Qn ( )  Qn,0 cos m  sin m ;
m
where i and j are the respective parameters of the element ends,
“incoming” to and “outgoing” from the point during the  2 Qt ,0  Qk ,0  Pf F0 Qn,0
tracking of the elements according to TMM [8]. This Qt ( )   sin m 
m2 m
expression is valid for the junction of several piping elements
Qt ,0  Qk ,0  Pf F0
(e.g., at tees and nozzles) as well as for the junction point of  cos m ; (30)
two pipes. In the last-mentioned case, with Fi  F j expression m2
(26b) is simplified to the form Qn,0  Qt ,0  Qk ,0  Pf F0 
Qk ( )   sin m  
u if  u fj . (26c) m m2
The equation of continuity for mechanical forces at a Qt ,0  Qk ,0  Pf F0
 cos m  ;
junction of several elements is well known and based on the m2
equilibrium condition for the junction. To allow for the pressure
pulsation Pср , we should put down the total axial force
where the notation m  1   2 is used for brevity sake.
    In particular, for a two-dimensional curvilinear element
N п = N x + F0  P f  t , (where t is the tangent vector in the   0 we have
section and F0  Pf represents a force in the bore due to the Q y1  Q y 0 cos   N 0 sin   P f  F0 sin  ; (31a)

inner pressure) in place of the axial force N x in the respective N1  N 0 cos  Q y 0 sin   Pf  F0 1  cos  . (31b)
equations. Thus, the equilibrium equation for the Expressions (30) and (31) have been written in the form

 
i   
 that enables them to be directly used in filling the resolving
( N xi  Fi  Pfi  t i )  Q yi  Q zi  matrix for a rotary element.
in The pipe–fluid interaction shows up where the fluid passes
. (27)
 
j   
 the rotary elements. Equation (26b) describes the influence of
( N xj  F j  Pfj  t j )  Q yj  Q zj the pipeline displacement on the fluid displacement at
out junctions. To derive the appropriate relations we will consider a
Analysis of Eq. (27) in view of (26a) shows that additional generalized vectorial displacement of a liquid, which is written
forces due to pressure arise where the pipe cross section is as follows:
changed at the junctions, including the case of a pipe blind end,    
  f  u f t  wn , f n  wk , f k , (32)
and where the tangent vector t changes its direction, which
occurs in tees as well as in bend elements. It is assumed that the differential of this equation along the
Note that this expression is valid both for the harmonic length of the rotary element is zero, as well. Moreover, the
vibrations of the fluid and pipe and for their static state. normal displacements of the fluid are equal to those of the pipe.

5 Copyright © 2013 by ASME

Taking this into account, we differentiate expression (32) with air is fed to and discharged from the system through long
respect to arc element ds of the rotary element [9]. Applying piping runs fitted with valves, choke valves, vibration dampers,
the differentiation rules for expressions of the form (28) [9] we and other equipment, i.e., the hydrodynamic edge conditions
have for this subsystem are and complex and ambiguous – it is not
du f  wn d  0 , (33) clear whether they should be described by expression (35) or
(36) or a linear combination thereof.
where wn ( ) - are the pipe displacements. By the example of the subsystem shown in Fig. 2 let us
Upon integrating this expression at the ends of the rotary state a method of searching for frequencies of forced
element using the relation between wn and vt [9], we arrive at hydrodynamic vibrations for piping systems with undefined
the following expression for the fluid displacement: edge conditions. Here, sections 1 and 2 can be arbitrarily
considered the “inlet” and “outlet” of the subsystem.
u f 1  u f 0  v t 1  vt 0 , (34)
Energy approached is believed to be the optimal method of
where the subscripts 0 and 1 imply that the value of vt is used searching for vibration frequencies. It essentially implies
respectively for the start and end sections of the rotary element. searching for the vibration modes and respective frequencies
Hydrodynamic vibrations are modeled by supplementing whereby minimum amount of energy leaves the subsystem and
the mechanical-vibration TMM with two unknown parameters the maximum amount remains inside.
for each section, two coupling equations for each element, two
parameters-continuity equations for each boundary between l=0.14m
elements and for the edge conditions. The design model and the
method of constructing the mathematical model remain L=1m d=0.051m
unchanged. Since all of the 14 parameters of the stress-strain
state of the fluid and pipe (12 mechanical parameters and two
hydrodynamic ones) are related to each other, we should model D=0.089m
the coupled HMVs “as a whole” rather than split the vibration
analysis into subtasks.
In some cases, a separate analysis of hydrodynamic
vibrations without any influence of the pipe displacements on 2
the fluid should be carried out. This is easy to accomplish by 1
using equations (24), (26), and (35) as written without beam
components. Such a “simplified” analysis makes it possible to FIG. 2. TEST SUBSYSTEM OF PIPELINE: 1 AND 2 -
find natural frequencies of hydrodynamic vibrations of a fluid BOUNDARIES OF MODEL FOR TRANSFER MATRIX
in a pipeline of a preset configuration, disregarding the pipe METHOD
displacements during vibrations; this is important, for example,
for TMM verification and for comparing the results to those Specific energy of vibrations (per pipe unit length) in an
published elsewhere. arbitrary pipe section consists of two components – kinetic and
potential energy, respectively:
4. ENERGY APPROACH TO THE DETERMINATION OF V f 2  f F0 u f 2 2  f F0
HYDRODYNAMIC VIBRATION FREQUENCIES Wkin   ; (37a)
2 2
The stress-strain state (SSS) analysis for both static and
Pf 2 F0
dynamic bar systems can be based on the use of different W potent  . (38b)
variants of boundary conditions that correspond to a free bar 2K
end, a hinged end, rigid fixing, etc. The following boundary The expression for the total specific energy in the starting
conditions can be also used in the hydrodynamic vibration pipe section (with superscript 0) is as follows:
analysis:
Pf  0 (35)
F 
 
W 0 f total  0  P f 0
2 
2 1

K
 
2 
 u 0f ω 2 ρ f .

(39)

in modeling a vessel of infinite volume, and: From the coupling equations for the fluid vibration
parameters (24) one can easily derive that for an arbitrary pipe
u f  vt . (36) section 1 (with superscript 1) the energy value with be the
for a closed-end pipe. same. Consequently, for all pipe sections with a constant inner
For model and test problems the use of these conditions is diameter, with no branches, the total specific energy is constant:
self-evident, while for real piping the edge conditions are
complex and ambiguous. For instance, Ziada and Buehlmann
F 
 
W f total  0  Pf 0
2 
2 1

K
 
2 
 u 0f ω 2 ρ f  

[10] studied hydrodynamic vibrations induced by turbulent . (40)
vortices in a subsystem with two straight branches installed at
2 
 
F0  1 2 1
 Pf
K
1
 
2 2 
 u f ω ρ f   const

different angles with respect to each other (Fig. 2). In that case,

6 Copyright © 2013 by ASME

 Thus, we can formalize the problem of finding the are shown on Fig.3. The environment is an air under normal
vibration frequencies and modes for a piping subsystem, which conditions. For each frequency the system is solved using
are physically governed according to the energy approach. Let TMM with two kinds of boundary conditions, (42a) and (42b)
W fb1 W fb 2 and W fb i total energies at the inlet to each branch of respectively. Then according to (44) energies ratio was written:
W fb1    W fb 2  
the subsystem (1st, 2nd, and ith), and W f1 W f2 are energies at     . (45)
W f1    W f2  
the inlet and outlet of the subsystem at hand. Then, the most
physically governed frequencies are those for which the Maximum of (45) can be fond using equation:
following functional takes on the maximum value:   
 0, (46)
W fb1  W fb 2  ...  W fbi  ... 
 . (41) which leads to the cubic equation for , and can be easily
W f1  W f2
solved for each frequency.
For practical purposes, an algorithm of maximization of Figure 3 show a ratio of energies  for this subsystem,
this functional can be proposed. Vibration modes (fluid natural frequencies found experimentally under boundary
displacement values u f and pressure pulsations Pf for every conditions (35) and (36) and those calculated by the
section) are plotted for various boundary conditions and for approximation formula
every frequency in the range under consideration. In our ( 2n  1 )V f
ωn  2π , (47)
example, we can restrict ourselves to two modes which are 4(L  l 2 )
specified by the following boundary conditions: where n is the natural frequency number.
u1f  1, u 2f  0; (42a) 6

u 1f  0 , u 2f  1; (42b)
-2 -3
5
where superscripts 1 and 2 denote respectively the fluid
displacements at the inlet (Fig. 2) and outlet of the subsystem. -5 -6
Thus, any natural mode will be represented by a linear 4
Energy ratio

combination of modes (42a) and (42b). For an arbitrary its

section we introduce a notion of “parameterized energy”: 3
 1
 

2
 Pi    Pi 2 1  
W f    F0i 
i K , (43) 2

 2 2
  ui f    ui f ω ρ f 

1 2

  1
where  is the parameter that defines a resulting mode which is
a superposition of (42a) and (42b). 0
1 4

Substitution of (43) into (41) will yield, instead of (40), the 0 500 1000 1500 2000 2500 3000
following expression:  , rad/s
W fb1    W fb 2    ...  W fbi    ... FIG. 3. RESULTS FOR TEST EXAMPLE:
    . (44)
W 1f    W f2   1 - energies' ratio ;
2 – natural frequencies of vibration for boundary conditions
For each frequency ω the functional (44) is maximized (35); 3 – for boundary conditions (36);
with respect to , and the resulting data are used for plotting a 4 – frequencies based on energy approach;
curve of optimal  vs. frequency  . Local peaks on this 5 – experimental data;
curve correspond to the most physically governed frequencies 6 – theoretical calculations(47).
from the range under consideration.
This algorithm is not the only possible method for The frequencies determined using the proposed approach is
maximizing (44). In general, the energy approach can be close to the experimentally found ones [10], while the values
efficiently used in the cases where a source of vibrations is obtained by theoretical approximation (47) are too low. By all
located inside the piping subsystem which vibrates with the appearances, formula (47) allows for only the total length 2L+l
largest amplitude. of the branches and a pipe piece in between, and disregards the
5. COMPARISION BETWEEN CALCULATED RESULTS difference in cross section and the possibility of vibrations
AND EXPERIMENTAL DATA spreading beyond the subsystem. Thus, (47) can be used for
approximate estimate of vibration frequencies for relatively
The proposed approach has been verified by comparing the simple piping subsystems (as shown, for example, in Fig. 2).
calculated data and those given in [10]. The system dimensions

7 Copyright © 2013 by ASME

 It can be seen from Fig. 3 that for this symmetrical system (ii) a mathematical model has been constructed and static
configuration the “energy-based” frequencies are close to the stress analysis performed for the piping;
odd natural frequencies which are determined under boundary (iii) natural mechanical and hydrodynamic frequencies and
conditions (36). modes have been determined for the piping;
The application of these methods of modeling harmonic (iv) coupled hydro-mechanical vibrations have been
vibrations in piping will be illustrated below by the example of modeled and amplitudes of pressure and force pulsations
vibration analysis of NPP steam piping. in the piping have been determined;
(v) based on the above activities, some recommendations
6. VIBRATIONS SECOND CIRCUIT LOOP OF VVER-
on reduction of the vibration level have been proposed.
1000
Vibration amplitudes were measured using a special
The steam piping (630×25) of a nuclear power plant developed system. The system consists of acceleration sensors,
(NPP) with VVER-1000 type reactor is located in room A-820 a signal amplifier, an analog-to-digital converter, and a
and includes four lines (TX50, TX60, TX70, and TX80) from computer for recording the measurement data. Vibration
steam generators (SG-1, SG-2, SG-3, and SG-4) to turbine (Fig. accelerations in all coordinate axes can be measured by
4). Each line has a piping run with three vertical toroidal simultaneously receiving and recording data from three sensors.
branches for emergency steam discharge. There are massive Also, special software has been developed for
safety valves installed at the top of each torus. determination of vibration velocities and displacements by
The steam piping is fixed by means of spring hangers and using vibration acceleration data and for estimation of a
sliding supports and the safety valves are rigidly fixed to frequency spectrum. This software is based on the Fourier
supporting structural steel. From mechanical standpoint, this expansion of the acceleration–time function with subsequent
system is isolated, since its inlet and outlet are rigidly fixed in analytical integration of the series in order to plot velocity–time
massive reinforced concrete walls. and displacement–time functions.
Nevertheless, at some power units considerable vibrations The measurements were carried out on all four steam
in steam piping are observed, especially in three-torus systems. piping lines during the power unit operation at 90 and 100%
In this context, at one of the VVER-1000 power units the duty and gave the following results.
following set of activities was completed: The largest vibration amplitudes (27 mm/s) were found on
(i) vibration accelerations at certain points of steam piping the first torus of the TX60 line at a frequency of 40.33 Hz in the
have been measured, frequency spectrum and vibration 90% duty mode; in the 100% duty mode the values went down
velocities have been determined for the points; to 12.3 mm/s, the frequencies were recorded to be 44.7 and 32
Hz.

FIG. 4 STEAM PIPELINES OF VVER-1000

8 Copyright © 2013 by ASME

 The basic frequency for the TX50 line is at a level of 64 frequencies of steam for the TX60 and TX50 lines; the
Hz, while that for the TX70 and TX80 lines is about 44 Hz; the calculations were performed using the edge conditions (35).
level of vibration in the TX80 line (25 mm/s) is much higher The calculated results have demonstrated that in the range
than that in the TX50 and TX70 lines (5 mm/s). between 0 and 60 Hz there is a great number of natural
The basic frequencies for the structural steel, including the frequencies of both mechanical and hydrodynamic vibrations
attaching points of safety valves, are 44 and 40 Hz. and, as evident from the analysis of natural modes, all the
The data obtained show that the vibrations travel from line frequencies are physically substantiated. This suggests that
to line via steel structures, which also undergo vibrations at detuning from resonance according to the rules [11] is almost
frequencies peculiar to the toroidal runs. The change in unfeasible.
vibration intensity and frequency with increasing power unit
duty suggests that the source of vibrations in the steam piping is
of hydrodynamic nature.
The static calculations were performed by modification of
the transfer matrix method as implemented in 3D PipeMaster ТХ 60
software package [6–8]. The static stress analysis allowing for
TX 50
internal pressure, dead weight, and thermal expansion has given
the values which satisfy the strength conditions as per [11]:
 165 MPa in the first torus and 170 MPa at the rigid
support (at the entry to the turbine house) for the TX50
line;
 203 MPa in the first torus and 156 MPa at the rigid 0 10 20 30 40 50 60
support for the TX60 line; Frequency, Hz
 188 MPa in the first torus and 131 MPa at the rigid
support for the TX70 line; FIG. 6 HYDRO-DYNAMICAL FREQUENCIES OF
 224 MPa in the first torus and 193 MPa at the rigid VIBRATION
support for the TX80 line.
It has been found out that the piping central point undergo Considering that the vibration amplitudes for “three-torus”
no displacement under thermal expansion; this makes it sub runs greatly exceed those for the main pipe runs, it is
possible to replace the spring hanger 20 (Fig. 4) with a rigid (in advisable to apply the proposed energy approach to these sub
displacement) support where necessary. Figure 5 shows the runs. A function of  vs. vibration frequencies upon the
results of the search for natural frequencies of mechanical  optimization is shown in Fig. 7 for the TX60 line. It can be
vibrations for all steam piping lines and the measured seen that the energy governed frequencies are 40.3, 58.7, 66.1,
frequencies. and 69.7 Hz; to the frequency 40.3 Hz there corresponds the
Measured frequency 44,7 Hz highest peak in the diagram, which is consistent with the
measurements.
4 1.8

3
40,3 1.6

1.4

1.2
44,7
energies' ratio

2
1.0

0.8

1 0.6

0.4

0.2
0 10 20 30 40 50
Frequency, Hz 0.0
  35 40 45 50 55 60 65 70
FIG. 5 NATURAL MECHANICAL FREQUENCIES OF frequency, Hz
STEAM PIPELINES:  
1 – ТХ50, 2 – ТХ80, 3 – ТХ60, 4 – ТХ70. FIG. 7 ENERGIES' RATIO  DEPENDING ON
FREQUENCY FOR TX60
For the determination of natural frequencies of steam
vibrations, the following steam parameters were used: Considering that the hydrodynamic vibration frequency
ρ f =34.5 kg/m3 and K= 8.32 MPa; in this case the sound 40.3 Hz is physically substantiated for the TX60 line according
to the energy approach and agrees with the measured data, it is
velocity is 491 m/s. Figure 6 gives the natural hydrodynamic preferable to carry out modeling of forced coupled hydro-

9 Copyright © 2013 by ASME

mechanical vibrations at this particular frequency, with hydro-mechanical vibrations have been used to analyze
excitation by the steam displacement velocity. Based on the vibrations in steam piping of VVER-1000 power unit.
measured amplitudes of vibration displacements of the main Few recommendations have been developed and
pipe (0.024 mm), we have obtained the following values of the implemented, which reduced the level of vibration.
steam pressure pulsation amplitudes:
 p1f  27 kPa (for toroidal runs); REFERENCES
1. C. S. Lavooij and A. S. Tijsseling, “Fluid-structure
 p 2f  12 kPa (for the main pipe). interaction in liquid-filled piping systems,” J. Fluids
The pressure pulsation amplitudes determined by the above Struct., 5, 573–595 (1991).
procedure correspond to the force at the bend centers 2. J. Kratz, W. Munch, and K. Ungar, “The influence of
N  2 Pf πR 2 sin 45 which is N 1  460 kg for toroidal runs fluid-structure interaction on pipe system loads,” in:
SMiRT-17 (2003), pp. 5–10.
and N 2  180 kg for the main pipe. 3. R. Skalak, “An extension of the theory of
The calculated real amplitudes of steam displacements are waterhammer,” Trans. ASME, 78, 105–116 (1956).
dozens of times larger than the pipe displacements: 4. A. S. Tijsseling, “Fluid-structure interaction in liquid-
u 1f  7,6 mm for toroidal runs and u 2f  3 mm for the main filled pipe systems: A review,” J. Fluids Struct., 10,
109–146 (1996).
pipe. 5. C. A. F. de Jong, Analysis of Pulsations and Vibrations
Thus, for these steam piping lines one can use a partial in Fluid-Filled Pipe Systems, Ph.D. Thesis, Eindhoven
model that assumes the pipe displacement to have no effect on University of Technology, Department of Mechanical
the fluid vibrations rather than the complete model of coupled Engineering (1994).
vibrations, where the fluid and the pipe are taken to have a 6. I. V. Orynyak, S. A. Radchenko, and A. S. Batura,
mutual influence on each other. “Calculation of natural and forced vibrations of a
This analysis has verified the hypothesis of turbulent piping system. Part 1. Analysis of vibrations of a 3D
vortices as the main factor responsible for piping vibrations. On beam system,” Strength Mater., 39, No. 1, 53–63
this basis, the following recommendations of how to reduce the (2007).
vibration level have been elaborated: 7. I. V. Orynyak, S. A. Radchenko, and A. S. Batura,
1. Subject to confirmation by seismic design calculations, “Calculation of natural and forced vibrations of a
replace spring hanger 20 with a guiding support. piping system. Part 2. Dynamic stiffness of a pipe
2. Increase spring rate of the hangers. bend,” Strength Mater., 39, No. 2, 144–158 (2007).
3. Increase rigidity of fixing points for safety valves and all 8. I. V. Orynyak, V. M. Torop, V. A. Romashchenko, and
steel structures. S. V. Zhurakhovskii, “Numerical analysis of a three-
At present, the last-mentioned recommendation has been dimensional branched pipeline using special software
implemented at the VVER-1000 power unit studied: the designed for estimation of the strength of the
supporting steel structures have been reinforced using girder equipment of nuclear power plants,” Strength Mater.,
type structures, thus reducing the vibration velocities by a 30, No. 2, 169–179 (1998).
factor of three and more. 9. I. V. Orynyak, A. S. Batura, and S. A. Radchenko,
“Application of the method of initial parameters to
analysis of coupled hydromechanical vibrations in
7. CONCLUSIONS piping systems. Part 1. Vibrodiagnostics of piping
1. Differential equations of coupled hydro-mechanical under mechanical vibrations,” Strength Mater., 43, No.
vibration of pipe and fluid have been obtained. These 6, 628–637 (2011).
equations take into account Poisson and junction coupling. 10. Ziada S. and Buehlmann E.T. Self-excited resonances
Solutions to these equations have been obtained in the of two-side branches in close proximity // J. Fluids and
form suitable for programming in transfer matrix Struct. – 1992. – 6. – P.583-601.
method. The advantage of such solutions over similar 11. PNAE G-7-002-86. Strength Analysis Code for
ones is that the set of equations does not degenerate for Nuclear Power Plant Equipment and Piping [in
any fluid and pipe parameters and any Poisson’s ratio . Russian], Valid since July 1, 1987.
Furthermore, these expressions are easy to use in
practice.
2. An original energy approach has been put forward for
the analysis of coupled hydro-mechanical vibrations
induced by turbulent vortices in a fluid flow.
3. Developed methods of calculations for the analysis of
harmonic mechanical, hydrodynamic, and coupled

10 Copyright © 2013 by ASME