Vous êtes sur la page 1sur 66

AECL-5852

ATOMIC ENERGY

mj& &

L'ÉNERGIE ATOMIQUE

OFCANADA LIMITED

^Kj

r

DU CANADA LIMITÉE

"FLOW-INDUCED VIBRATIONOF

NUCLEAR POWER STATION COMPONENTS"

by

M.J. PETTIGREW

Presented at the 90th Annual Congress of the

Engineering

Institute of Canada, Halifax, Nova Scotia, October 4-8, 1976

Chalk River Nuclear Laboratories Chalk River, Ontario

September 1977

"FLOW-INDUCED VIBRATION OF NUCLEAR POWER STATION COMPONENTS"*

M.J.

by

Pettigrew,

H.C.S.M.E.

^Presented at the 90th Annual Congress of the Engineering Institute of Canada, Halifax, Nova Saotia, October 4-8, 1976.

Atomic Energy of Canada Limited Chalk River Nuclear Laboratories Chalk River, Ontario

KOJ 1J0

Se-pterriber 1977

AECL-5852

"VIBRATIONS ENGENDREES PAR L'ECOULEMENT DES F.LUIDES DANS LES COMPOSANTS DES CENTRALES ELECTRONUCLEAIRES"*

par

M.J. Pettigrew

Plusieurs composants des centrales électronucléaires CANDU** sont sujets à des vitesses d'écoulement des fluides relativement grandes en régime liquide ou biphasé (eau/vapeur). Le combustible nucléaire dans les canaux de combustible et les faisceaux de tubes dans les générateurs de vapeur sont des composants typiques. Souvent on augmente les vitesses d'écoule- ment pour améliorer le rendement des composants, par example, pour obtenir un meilleur échange calorifique dans les canaux de combustible. Pour des raisons économiques on préférerait spécifier des composants plus petits ou éliminer des éléments de structure, par example, on utilise des tubes de petits diamètres pour réduire l'inventaire d'eau lourde. De grandes vitesses d'écoulement et une réduction des éléments de structure peuvent causer des problèmes de vibrations. CetL? communication traite des problèmes et analyses de vibrations des composants des centrales électronucléaire engendrées par les écoulements.

L'usure par frottement, la fatigue, le bruit acoustique et les difficultées opérationelles sont les problèmes causés par les vibrations. On examine de récents problèmes comme l'usure de tubes de générateur de vapeur.

Les écoulements dans les composants nucléaires peuvent être parallèles ou trans- versaux . Dans les canaux à combustible l'écoulement est surtout par- allèle. L'écoulement est transversal et liquide au travers des faisceaux de tubes d'échangeurs de chaleur tandis qu'il est aussi transversal mais biphasé dans la région des générateurs de vapeur où les tubes sont coudés en U. On discute des mécanismes d'excitation dominants en écoulement parallèle et transversal. Eh écoulement parallèle on considère deux méchanismes principaux qui sont l'excitation aléatoire due à la turbulence de l'écoulement et l'instabilité fluidelastique. En écoulement transversal on considère en plus le détachement périodique des tourbillons. Notre méthode d'analyse des composants nucléaires est présentée. L'analyse des vibrations des générateurs de vapeur est donnée en example.

Nos études courrantes sur les vibrations engendrées par les écoulements sont décrites. Ceci inclus l'étude du comportement vibratoire des éléments de combustible nucléaire dans un réacteur expérimental.

On conclu que, même si le travail de recherches n'est pas encore terminé, la plupart des problèmes de vibrations peuvent être évités, pourvu que les composants nucléaires sont analyses au stage de la conception et que ces analyses sont appuyées par des études expérimentales au besoin. On n'a pas encore rencontré de situations où les vibrations ont sérieusement limité l'ingénieur au stage de la conception.

* Cormuniaation présentée au BOieme congrès annuel de l'Institut canadien des ingénieurs, Halifax, Nouvelle-Eaosse, octobre 4-8, 1976.

** CANDU - CANada Deuterium Uranium.

L'Energie

Atomique

du

Canada, Limitée

Chalk

River,

Ontario

Canada ,

KOJ 1J 0

Septembre

1977

AECL-5852

"FLOW-INDUCED

VIBRATION

STATION

by M.J. Pettigrew, M.C. S.M.E. Atomic Energy of Canada Limited Chalk River Nuslear Laboratories

OF NUCLEAR

POWER

ABSTRACT

COMPONENTS"*

Several components of CANDU** nuclear power stations are subjected to relatively high flow velocities in either liquid or two-phase (steam/water) flow. Typical of such components are the nuclear fuel in the fuel channels and tube bundles in the steam generators. Often higher component performance, requires higher flow velocities, for instance, to improve heat transfer in fuel channels. Economics sometimes dictates smaller components or minimum structural constraints, for example small diameter tubes are used in steam generators to minimize heavy water inventory. High flow velocities and decreased structural rigidity could lead to problems due to excessive flow-induced vibration. This paper generally treats the problems and the analyses related to flow-induced vibration of nuclear power station components.

Fretting-wear, fatigue, acoustic noise and operational difficulties are the problems caused by flow-induced vibration. Some recent problems such as fretting of Jteam generator tubes are reviewed.

Flow in nuclear components may be parallel or transverse. In fuel channels the

Liquid cross-flow exists in heat

flow is mainly parallel to the fuel elements.

exchanger tube bundles and U-bend tube regions of steam generators are sub- jected to two-phase cross-flow. The vibration excitation mechanisms predominant in parallel and transverse flow are discussed and formulated. In parallel flow two basic vibration excitation mechanisms are considered, namely random excitation due to flow turbulence and fluidelastic instability. The above and periodic wake shedding are considered in cross-flow.

Our approach to the vibration analysis of nuclear components is presented. This is illustrated by the vibration analysis of steam generator designs.

Current investigations related to flow-induced vibration are outlined. This includes the experimental study of the in-reactor vibration behaviour of fuel elements.

It is concluded that, although there are still areas of uncertainty, most flow- induced vibration problems can be avoided provided that nuclear components are properly analysed at the design stage and that the analyses are supported by adequate testing and development work when required. There has been no case yet where vibration considerations have seriously constrained the designer.

* Presented at the 90th Annual Congress of the Engineering Institute of Canada, Halifax, Nova Saotia, Ootobev 408, 1976.

** CANDU - CANada Deuterium Uranium.

Chalk River, Ontario KOJ 1J0 September 1977

AECL-5852

CONTENTS

 

Page

1. INTRODUCTION

2

2. FLOW-INDUCED VIBRATION PROBLEMS

 

3

3. FLOW CONSIDERATIONS IN NUCLEAR STATION COMPONENTS

5

4. VIBRATION EXCITATION MECHANISMS IN AXIAL FLOW

8

Fluidelastic Instability

8

Forced Vibration

.

10

5. VIBRATION EXCITATION MECHANISMS IN CROSS-FLOW

17

1) Forced Vibration

 

17

2) Fluidelastic Instability

18

Periodic Wake Shedding Resonance

20

6. VIBRATION ANALYSIS OF NUCLEAR COMPONENTS

21

7. CURRENT VIBRATION STUDIES Vibration Behaviour of Nuclear Fuel in Reactor

25

Vibration Damping and Support Dynamics of Heat Exchanger Tubes

26

Other Vibration and Related Studies Currently Underway

27

8. CONCLUDING REMARKS

28

REFERENCES

30

FIGURES

,

35

1.

-2-

INTRODUCTION Several components of CANDU* nuclear power stations are subjected to relatively high flow velocities. Typical of such components are the nuclear fuel bundles in the fuel channels and the tube bundles of steam generators and heat exchangers. Often higher component performance requires higher flow velocities, for instance, to improve heat transfer in fuel channels. Economics sometimes dictates smaller components or minimum structural con- straints, for example small diameter tubes are used in steam generators to minimize the inventory of expensive heavy water. High flow velocities and decreased struc- tural rigidity could lead to problems due to excessive flow-induced vibration. Such problems could seriously affect the performance and reliability of nuclear power stations.

The above is best illustrated by an example. Fretting- wear due to vibration of one of the many tubes in a steam generator could result in leakage of heavy water primary coolant into the secondary system. A station shut-down lasting a few days would be required for repairs, This is very undesirable in terms of lost production and of radiation exposure limitation of maintenance personnel. Although an effective tube plugging technique has been developed 1 ' 2 in preparation for the unlikely event of a tube failure, it is much preferable to avoid vibration problems altogether. This can be achieved by proper flow-induced vibration analysis of nuclear station com- ponents at the design stage.

This paper is a general outline of our work in the area of flow-induced vibration. Some recent vibration

* CANDU (CANada Deuterium Uranium)

-3-

problems are reviewed. Flow-induced vibration excitation mechanisms are discussed. The paper outlines our approach and techniques to analyse nuclear power station components from a flow-induced vibration point of view. The prevention of flow-induced vibration problems is emphasized. Some current vibration studies are described.

FLOW-INDUCED VIBRATION PROBLEMS The problems related to flow-induced vibration are gener- ally fretting-wear, fatigue, acoustic noise and operational difficulties. Figures la and b show a case of steam gene- rator tube fretting-wear which occurred in the Douglas Point nuclear power station 3 . The "U" bend tubes near the outlet are subjected to high velocity two-phase (steam/water) flow. In a few of the Douglas Point steam generators the "U" bend tubes were not supported at the top and vibrated with sufficient amplitude to contact each other resulting in the fretting-wear shown on Fig. la. Vibration of the "U" bend tubes also caused fretting at the location of nearby supports. In one tube the fretting was extensive enough to cause leakage as shown on Figure, lb.

In most of the steam generators the "U" bend tubes were supported at the top and no fretting problem occurred. This problem could have been prevented simply by providing for adequate tube supports.

A case of heat exchanger tube fretting-wear is shown in Figure 2. He.e the fretting-wear occurred at the location of lacing metal strips which were added to provide addi- tional support near the inlet where flow velocities are relatively high. The problem was attributed to the com- bination of excessively loose lacing of the metal strips and partial blockage of the inlet which resulted in much

-4-

hlgher than expected flow velocities in the region of

the damage". Avoidance of inlet blockage and the

replacement of the lacing strips by proper support plates

were the corrective actions taken in this case.

Fretting-wear was observed on the top fuel bundles in 40% of the high flow fuel channels of the Gentilly-1 nuclear power station. Figure 3 is a photograph of

typical fretting damage taken through an optical magnifier during "hot cell" examination. Figure 4 is a simplified flow diagram of the Gentilly-1 station which is of the CANDU-BLW* type. The fuel bundles (Fig. 5) are assembled in the form of a string held together with a central sup-

porting tube. The latter is

flux suppressor and at the bottom by a spring assembly. The strings are inserted in upward flow vertical fuel channels as shown on Figure 6. They are attached at the

terminated at the top by a

bottom and free at the top of the fuel channels.

flow gradually becomes two-phase as boiling occurs along

the fuel and reaches i 16% steam quality near the top. —2 —1

.s . The fretting

The

The mass flux is typically 4400 kg.m

problem was attributed to transverse flow-induced vibra- tion of the fuel strings. Unexpectedly some of the flux suppressors were assembled eccentrically. This caused the fuel strings to be bent and promoted fretting-wear. The corrective measures taken were to as-.ure the concentric assembly of the fuel and to increase fuel string flexural rigidity to reduce vibration.

We now consider an example where flow-induced vibration could have lead to operational difficulties. In the Gentilly-1 station, control absorber guide tubes are cantilevered and suspended vertically in the calandria as

* BLW -(Boiling Light Water)

- "5-

shown on Figure 7. They extend past the horizontal

booster fuel rods. The absorber guide tubes were directly exposed to the submerged jet flow emerging

from the booster rod

outlet.

During prototype testing

the absorber guide tubes vibrated severely. In the reactor core this would have resulted in local reactivity disturbances which could have caused operational problems. The designers avoided the problems altogether by providing a protective shroud attached to four adjacent calandria tubes as shown on Figure 7a.

We have encountered other problems such as excessive acoustic noise due to flow control valve dynamics and fatigue cracking due to noise-induced vibration of steam discharge nozzles. So far all our flow-induced vibration problems have been solved by simple design modifications or changes in operational conditions.

3. FLOW CONSIDERATIONS IN NUCLEAR STATION COMPONENTS Consider the simplified flow diagram of a typical CANDU- PHW* nuclear power station as shown on Figure 8. Most

stations in Canada are of that type.

Starting at the

primary pumps, the heavy water coolant flows in the headers, into the feeder pipes leading to each fuel

channel. The fuel channels are horizontal. The flow in the channels is essentially axial to the fuel bundles.

Flow velocities

The

bundles are held down in the channel by gravity forces. They are not held together by me.chanical means although they are pushed together against a downstream stop by hydraulic forces. This is different than the string type fuel bundle assembly of vertical CANDU-BLW fuel channels.

in the order of 9 m/s are typical.

* PHW - (Pressurized Heavy Water)

-6-

The fuel bundles may be partly subjected to cross-flow during refuelling operations when they .are moved past the inlet or outlet feeders. In Pickering and earlier stations, the flow remains liquid throughout the fuel channels. In post Bruce stations and to some extent in Bruce the coolant is allowed to boil and downstream fuel bundles and outlet feeders are subjected to some two- phase (steam/water) flow. For example in Gentilly-2 and Point Lepreau, the average channel outlet quality is expected to be around 4%. These stations are sometimes called CAFDU-BHW*.

The outlet feeders are coupled to main headers which lead to the steam generators. Figure 9 shows a typical recirculating type steam generator. All flow situations are possible in this component. Heavy water flows in the tubes at varying conditions from 5% steam quality to subcooled liquid. The tubes are subjected to liquid

cross—flow in the preheater section and in the recirculated water entrance region near the tubesheet. The saturated

water then flows

steam quality at the top. Thus liquid and two-phase

axial flow exists along the tubes. Two-phase cross- flow is predominant at the top of the "U" tube region

where the mass flux is typically 300 kg m~*.s

up and gradually boils, to reach 15 - 20%

.

There are many heat exchangers in a nuclear station, e.g. the moderator heat exchangers. The tubes of heat exchangers are mostly subjected to cross-flow particularly near inlets and outlets. The steam produced by the steam generators is

* CANDU-BHW - (Boiling Heavy Water)

our

ICro

-7-

condensed after going through the turbine. The condenser is an enormous heat exchanger whose tubes are exposed to high velocity steam flow. The immersion heaters located at the bottom of the pressurizer are another category of interesting components. The heater elements are exposed to incoming liquid or two-phase flow during station start-up and out going liquid flow during shutdown. Flow-induced vibration of the calandria tubes may also be possible. They are subjected to some moderator cross-flow and may be exposed to submerged jet for example near the effluent of booster fuel rods.

Thus from a flow-induced vibration point of view, nuclear station components are essentially cylindrical structures or bundles of cylinders subjected to axial or transverse

flow. "The flow may be internal or external to the cylinders and it may b e liquid, vapour or two-phase. This is outlined on Table 1. The first task in any flow-induced vibration

analysis is to define the flow nuclear component under study.

conditions prevailing in the

TABLE 1: Possible Flow Conditions in Nuclear Power Stations

«1

STATION COMPONENTS

% g

•3 «

Ë e

ë

01

s

41

(H

1 ii

sa tu

i 1 S

l/

1

a a

zz

J

Yes

/

Ho

/

/

Fuel

Channel

M

w

Feeder

Pipe

Fuel

(Normally

/

I During

Loading

Calandria

Tube

 

/

Control Rod

,/

Steam

Entrance

,'

Genera-

"U" tub e

/

tors

 

Ptehee'—r

/

Elseirhere

/

Heat Exchangers

 

/

Condenser

J

/ j

/ z

/ /

/

/ /

/ /

/ J

/

/

/ /

/ /

/

/

-8-

VIBRATION EXCITATION MECHANISMS IN AXIAL FLOW In axial flow we consider two flow-induced vibration excitation mechanisms, namely: fluidelastic instability and forced vibration response to random excitation due to flow turbulence. Other excitation mechanisms such as self-excited vibration 5 and parametric vibration 1 " 7 have been suggested. However we have not yet needed to consider them. For a comprehensive review of this topic, the reader is referred to Paldoussis 8 .

Fluidelastio Instability Fluidelastic instabilities result from the interaction between hydrodynamic forces and the motion of structures. For cylinders in axial flow, the pertinent hydrodynamic forces 9 are the frictional forces, the fluid acceleration forces and in some cases the drag forces (e.g., cylinders with one free end). Instabilities appear in the form of either buckling or flutter-like oscillations. Figure 10 shows a flexible cylinder experiencing fourth mode buckling while being subjected to confined liquid flow. Fluidelastic instabilities are possible with both internal and external liquid flow. In spite of some experimental efforts 10 , we have not yet confirmed that instabilities are possible in two-phase axial flow. To be conservative we assume they exist in our analyses.

The fluidelastic behaviour of cylindrical structures in

axial flow has been formulated by Paldoussis 8 ' 9 . The dynamic response y at a time t of a uniform cylinder diameter D, length L, flexural rigidity £1, mass and hydrodynamic mass m and M respectively, subjected to an axial velocity U is governed by:

of

-9-

- î C T ^ Hl-f6)L-x} â_I _ {6To +

3x

2

3x

D

D

3t

c

i (1 _ ô)e ; M U 2 }

iîj =0 .

^

3x*

In this equation, x is a point along the cylinder, y is an internal damping coefficient, €„ is the axial frictional force coefficient, T is an externally imposed tension, C' is a downstream end base drag

coefficient, C is the normal frictional force coefficient and C D represents a viscous damping coefficient at zero flow velocity. Finally, 6= 0 corresponds to the case where the downstream end Is free to move axially and S «

1 when it is not.

For U^ 0, solution of this equation yields

the eigenvalues and eigenfunctions of the system, which are complex. By varying U one may determine the critical flow velocities for fluidelastic instabilities and the corres- ponding mode shapes associated with these instabilities.

In a very approximate way, critical velocities for fluid- elastic instability may be formulated in terms of the non-dimensional velocity

u

- OL /MT U

(2)

For a given mode it is desirable to keep u much lower than the critical value to avoid instability. In general

if u is lower than unity there should be no problem 8 ' 11 .

In a nuclear component where M

and V may be fixed for

-10-

other considerations, instability problems may be avoided by increasing the flexural rigidity El or increasing the number of support points (e.g., decreasing L) .

Fortunately, critical velocities for fluidelastic insta- bilities are much higher than the axial flow velocities normally encountered in nuclear components. For instance

the critical velocity of a typical steam generator tube is

in the order of 100 m/s. The relatively

flexible and heavy fuel strings of CANDU-BLW fuel channel are the exception for which the possibility of fluid-

elastic instabilities must be considered 11 .

long, very

Forced Vibration Nuclear components may respond to 1) excitation forces that are of mechanical origin and are structurally trans- mitted, or 2)boundary layer pressure fluctuations that are generated by the fluid. Structurally transmitted forces may be generated by rotating machinery such as pumps or the turbine-generator or by other components with moving parts such as control valves and fuelling machines. It is also possible that the flow-induced vibration response of other components such as the feeder pipes be structurally transmitted to for example the fuel bundles. It is very difficult to evaluate structur- ally transmitted forces as they are not characterized by the component under consideration. They depend on the overall system to which the component is integrated. Fortunately we have not experienced vibration problems due to structurally transmitted vibration.

Fluid-borne pressure fluctuations may be divided in two groups, namely: far field and near field. Far field disturbances are generated by upstream components such

-li-

as pumps, valves, elbows and headers and are transmitted by the fluid. Pressure fluctuations due to far field sources would generally be broadband in nature except for those generated by pumps. These would be at a fre-

quency related to the pump speed times the number of impeller

vanes.

as the fluid dynamic behaviour of the overall system needs to be understood. Far field disturbances are insignificant in two-phase flow as they are quickly attenuated by the inherently high damping of two-phase mixtures. This is

fortunate since two-phase flow induced vibrations are generally more severe.

Such forces are again very difficult to formulate

Near field disturbances are generated locally by the fluid as it flows around the component of interest. They may be generated in a number of ways such as general turbulence, swirl, cross-flow components, flow regime changes and nucleate boiling. The result is a broadband random

pressure

ponents. At a given time, the pressure is not uniform around the periphery of a component. This results in a net time varying force which excites the component to vibrate. It may be shown with the assistance of

References 12, 13 and 14, that the mean square response

~2

y (x) of a uni-dimensional continuous uniform cylindrical structure to distributed random forces g(x,t) may be expressed by:

field acting at the surface of cylindrical com-

-12-

y2(x) = EET^77V^ ) / |H r (f)||H s (f)|cos[e r (f)-e s (f)]

r

s

J J

J

J 4> r (x)

r

r

s

m

f> s (x') R(x,x',f) dx dx' df

(3)

where: 1) the spatial correlation density function R(x,x',f) is defined by

R(x,x',f) = 2 f T->a

|^

/" i(x,t) g(x',t+T) dt e " j(2irf) dT

(4)

2) the frequency response function is

a ' <f> -?—A

r

•• <5>

Ç r is the damping ratio at the r of H r (f).

mode and 6

r

is the argument

3) <f> (x) and <f>(x) represent the normal mode of

mode, and

r

s

i.

vibration of the structure for the r

.,

and s

4) x and x* are points on the structure and T is

a uifference in time t.

For the above derivation we assume that the damping is small

and that it does not introduce coupling between modes to

justify modal analysis. The natural modes are normalized

so that

m <j>_ 2 (x) dx = 1

(6 )

-13-

"lonsider now the fundamental mode only of a lightly

lamped lamped simply simply supported supported cy cylinder (i.e., <f> (x) = (2/£m)

sin (TTX/A).

If we assume:

1) that the random force field is homogeneous, the power

spectral density function of the force S(g) is independent

of location, i.e.,

R(x,x',f) =• R'(x.x') S(g)

-' (7)

and 2) that both S(g) and the spatial correlation R'(x,x') are fairly independent of frequency near the fundamental frequency of the cylinder, we can show that the space and frequency term» in Equation 3 may be separated and that:

|H g (f)j cos U(f ) - e s (f)| df =

Ïïf x /4Ç

(8)

substituting Equation 6, 7 and 8 in 3 we get for x = 1/2 (i.e., midspan):

7 (ft/2)=

L

16

2

m

<S>

f 1

ir Ç

where ip t is a ratio of

effective cylinder length over

actual length and is a measure of the spatial correlation

of the forcing function.

\p is defined as:

*L " lï f f

o

o

*1 <X) *1< X '> R'^^.x 1 ) dx dx'

(9)

(10)

-14-

Equation 9 is very similar to that derived by Gorman 15 ' 16 and Reavis . Similarly if we define a peripheral spatial

correlation ratio IJJ such that D

response may be expressed in terms of the power spectral density S(p) of the random pressure field p(x,t).

2

2

i^

S(p) = S(g), the

If we further assume that the random excitation forces are completely spatially correlated (i.e., R'(x,x') = 1) we obtain from Equation 6, 9 and 10:

-,

S(g)

y

(H/2) = —,

3

2

(11)

4TT f x J in

^

This equation may be useful to make a first outside guess at forced random vibration response. Obviously if the random forces are not well correlated, the vibration response would be much less.

The main difficulty here is to determine the statistical properties of the forcing function,that is its spatial correlation density function R(x,x',f) for all the configurations of interest. Gorman has measured values

of ty T , (f»_ and S(p) for some typical

figurations 15 ' 16 ' 18 . Using the

fuel element con-

above measured values,

he has had remarkable success in predicting the vibration response of a fuel element in axial two-phase flow simulated by air/water mixtures 16 . Predicted and measured vibration amplitudes are compared on Figure 11.

As shown on Figure 11 the vibration response is maximum at a simulated steam quality of approximately 15%. We have done further testing in steam/water flow where much higher qualities were achieved 19 . The results on Figure 12 show two maxima in the vibration response vs steam quality curves. This is explained in terms of two-phase flow

-15-

regime changes as follows: At low steam quality, bubbly/slug flow regime exists and the vibration amplitude increases with increasing quality until it reaches a maximum. This is reasonable since higher vibration amplitudes are expected at the higher velocities related to higher steam qualities. The maximum amplitude corresponds to the start of the transition between bubbly/slug and annular flow regime. As the quality is increased, annular flow regime is estab- lished. This flow regime is presumably less turbulent and the vibration amplitude reaches a minimum. As the quality is increased further, the flow velocity and con- sequently the vibration amplitude increases again to reach another maximum. This maximum corresponds to the transition between annular flow and dispersed/fog flow regime. The vibration amplitude decreases again to a minimum as fog flow regime is established.

We have studied the effect of subcooled and bulk nucleate

boilin g at the surface of a cylinder on its vibration response 19 .

Nucleate boiling does not contribute significantly to

vibration excitation for typical cylindrical structures.

It should be apparent by now that predicting the vibration response of nuclear components subjected to random pressure fields is not always simple. Several researchers have developed semi-empirical expressions which may be useful for some axial liquid flow problems. The expression developed by Païdoussis 8 is relatively successful as shown on Figure 13 where it is compared to experimental data. Paldoussis' expression converted te dimen&ional form is:

Y

-A

^ - 5x10

K

-4

a

V 25 (l

.65-,

.2.2

M

1.47

/m , 0.67

M

1 + 4M/m

(12)

-16-

where Y is defined as the "maximum" amplitude, K is a proportionality constant, a- is the dimensionless first mode eigenvalue, V is the kinematic viscosity of the fluid and D, the hydraulic diameter. K = 1 for laboratory controlled conditions with minimum far field disturbances. For nuclear power station components K = 5 is more realistic,

In Equation 12 the velocity exponent is 1.85.

This is

reasonable since fluid forces are generally related to

velocity squared. The vibration amplitude is directly

related to L * /(El)' . Again this makes sense as it is

comparable to the deflection of beams under distributed

loading. The latter depends on L /El. We have found in

many practical cases in liquid flow that vibration is

roughly proportional to velocity squared. The above is

not much affected by the non-dimensional velocity term

u

seen earlier, the velocity in most nuclear components

is much lower than the critical velocity for instability,

thus U<< 1 and

2 = ML 2 U 2 /El in the denominator of Equation 12.

2

the term u is not significant.

As

The flow vs vibration response relationship is somewhat different in two-phase flow. We have found that for equivalent qualities, the vibration response is roughly proportional to flow velocities 18 ' 19 as shown on Figure 14, This unexpected behaviour is attributed to flow regime consideration. For instance the spatial correlation may be reduced at higher flows as the scale of turbulence becomes finer.

The simple relationships expressed above may be very useful to the designer. They should permit him to roughly compare different designs or to estimate the effect of changes in operational conditions.

-17-

5. VIBRATION EXCITATION MECHANISMS IN CROSS-FLOW Generally in cross-flow induced vibration problems, we consider three basic flow-induced vibration excitation mechanisms: 1) forced vibration, where the cylinders are forced to vibrate by random excitation due to flow turbulence; 2) fluidelastic instability, where the fluid forces are related to the relative motion between cylinders in such a way that coupling and instability results; and, 3) reso- nance, where the natural frequency of the cylinders coin- cides with the frequency of periodic wake shedding.

We have observed the first two excitation mechanisms in both liquid and two-phase cross-flow. Periodic wake shedding resonance is possible in liquid flow but has not been observed in two-phase flow. Either it does not exist or it is dominated by the response to random flow excitation. Hence we do not consider it.

1) Foraed Vibration We treat the problem of forced vibration response due to random cross-flow turbulence in the same manner as for axial flow. The vibration response may be estimated using Equation 3. Here again the difficulty is to deter- mine the statistical properties of the forcing function due to cross-flow turbulence. Gorman 20 has done this in

some typical configurations of heat exchanger tube bundles by direct measurements. We have also deduced the forcing

function from the vibration

response 21 ' 22 . Much more

information is needed in this area. However in cases where forcing function information is lacking, the analy- tical techniques to predict vibration response may still be used advantageously. For instance the vibration response of a prototype component may be compared to that

-18-

of an existing nuclear station component exposed to similar flow conditions.

Vibration response to random flow turbulence is usually not a serious problem in liquid cross-flow. Other flow- induced vibration excitation mechanisms are generally the limiting criteria. This is not true however in two- phase cross-flow where the turbulence is inherently much stronger.

As in axial flow, the random vibration response in cross- flow is approximately related to velocity in two-phase

flow

and to velocity squared in liquid flow 21 ' 22 .

2) Fluide las tic Instability

The fluidelastic instability phenomena is somewhat different in cross-flow than in axial flow. In a bundle of cylinders, the hydrodynamic forces on one cylinder are affected by the motion of neighbouring cylinders. This creates an interaction between hydrodynamic forces and the motion of cylinders. When the energy absorbed by the cylinders from the unsteady part of the hydrodynamic forces exceeds that dissipated by damping during one cycle of motion, fluidelastic instabilities occur. This situation is possible when the flow velocity is sufficiently high. Theoretically the motion of the cylinders should increase indefinitely. However in practice it is limited by non- linearities such as the presence of neighbouring cylinders. This results in severe rattling and possible damage. For an isolated circular cylinder there is no interaction and

hence no instability

a non-circular body as interaction between fluid forces and torsional motion may occur. We have found 22 that

fluidelastic instabilities are possible in both liquid and two-phase cross-flow.

of this type.

This is not true for

-19-

The critical velocity V rc at which

fluidelastic insta-

bility occurs in the s be expressed as:

vibration mode of a cylinder may

-

V = Ke f

I

S /(2pm

J f

2

X l

S c|> 2

->

(x)dx)

J

J

1/2

(13)

where c is the viscous damping coefficient, K, is a factor determined experimentally, p is the fluid density and

x 1 , x. define the length over which subjected to flow.

the cylinder is

Equation (13) is a generalized expression derived from Connor's formulation 23 of fluidelastic instability in a

single -array of cylinders .

velocity and is equal to V ac p/(p-D) in which V

V is the reference critical gap

is the

free stream velocity (i.e., velocity taken as if there were no cylinders), p is the pitch of the tube bundle and D the cylinder outside diameter. If the cylinders are exposed to cross-flow over their entire length, knowing that c = Airmfç, Equation 13 reduces to Connors' expression

V rc /fD = K(m6/pD 2 ) Js

(14)

in which the logarithmic decrement <5 = 2irç.

Connors obtained a value of K = 9.9 for a single array of

cylinders in air flow.

We found K = 6.6 for cylinder

bundles of p/D = 1.5 in liquid and two-phase cross-flow 22 (see Figure 15). More recently Gorman 2 * has found lower bound values around .K 3/ 5 for more closely packed bundles (p/D - 1.36). We consider good design practice to keep the actual velocity below the critical velocity by a reasonable margin to allow for uncertainties. Thus we use K » 3.3 for design purposes.

-20-

'Peviodio Wake Shedding Résonance The periodic formation of vortices downstream of an isolated cylinder in cross-flow is a Classical phenomena called Karman vortex shedding. The frequency of vortex formation is defined in terms of a Strouhal Number S ~ FD/V where S is usually ^ 0.2. What happens in closely packed bundles of cylinders is not so well understood. Three mechanisms which could lead to periodic forces may be postulated as shown on Figure 16, namely:

1) Vortex shedding 25 ; The formation of vortices should, however, be much affected by the close proximity of adjacent and particularly downstream cylinders. 2) Buffeting; Periodic forces may arise on a given cylinder a:, it is being

subjected to the vortices generated by the upstream cylinder. 3) Turbulent theory 26 ; The argument here is that the scale of turbulence is controlled by the geometry of the cylinder bundle configuration. For a given flow velocity, same

scale turbulence leads to narrow band turbulent to some degree of periodicity.

forces and

Whatever the mechanism, periodic wake shedding forces could result in a resonance problem if their frequencies coincide with one of the natural frequencies of the cylinders.

The peak

mode of vibration of a cylindrical structure is given by:

vibration amplitude Y

at resonance for the r

 

C T pDV 2 (x)

<>|

(x)

dx

(15

)

Airf r c

/

L

r

•/o

where: C T is the dynamic lift coefficient attributed to

Li

periodic wake shedding and V(x) is the flow velocity dis- tribution at any point along the cylinder.

-21-

We normally assume for conservatism that at resonance the periodic forces are spatially correlated.

In our experience we have not observed periodic wake shedding resonance for tubes inside a tube bundle. We

have mostly encountered it for upstream tubes 20 , that is

in the first and to a lesser extent row (see Figure 17). We have found

based on the free stream velocity V to be generally less

than unity.

shedding frequency in terms of a predictable criterion such

as the Strouhal No., fd/V.

Thus we assume resonance in the

analysis whenever this mechanism appears possible.

in the second tube the lift coefficient C^

We have not yet been able to correlate the wake

6. VIBRATION ANALYSIS OF NUCLEAR COMPONENTS The first step in the vibration analysis of a nuclear component is to define its dynamic parameters, that is:

stiffness or flexural rigidity, mass including the hydro- dynamic mass and both structural and viscous damping. Oncfc these are known the natural frequencies f . and mode shapes <(>.(x) may be calculated. Then the vibration response may be predicted.

Take for example the case of a nuclear steam generator.

A typical steam generator tube and the flow conditions

to which it is subjected is shown on Figure

18.

From

a mechanical dynamics point of view "U" tubes are simply multi-span beams clamped at the tubesheet and held at the baffle-supports with varying degree of constraint. The latter is dependent on support geometry and parti- cularly tube-to-support clearance. To be conservative we assume the intermediate supports to be essentially

hinged.

between tube and support to keep our analysis linear.

We do not yet take into account the clearance

-22-

The tube dynamics is completely defined by knowing m, c, El, H and the boundary conditions (i.e., the support locations). We assume that either the damping coefficient c or the damping ratio Ç be independent of frequency. Typical values for c are 0.04-0.07 kg rad/cm.s and 0.25- 0.5 kg rad/cm«s in liquid and two-phase flow respectively. These correspond roughly to Ç = 0.02 and Ç = 0.08 at typical tube frequencies.

Based on Equations 3, 13 and 15 we have developed a computer program called "PIPEAU" to predict the vibration response of multispan tube bundles. The program first calculates the mode shapes <$>.(x.) and the natural frequencies f. (i.e., eigenvalue solution) using a method similar to that suggested by Darnley 27 . Then the response and the critical velocities for fluidelastic instability are estimated.

For the example shown on Figure 18 the threshold velocity for instability in the U-bend region (between supports 6

and

be 3.3 times the actual velocity. A similar calculation

for the inlet region (between supports 0 and 4) where m =

0.53 kg/m and

entered from a high mode oscillation, at a threshold velocity more than 5 times the actual velocity.

12) where m = 0.42 kg/m and Ç = 0.08 is calcualted to

t, = 0.028 shows that instability would be

Calculations of tube response to random excitation are shown on Figure 19. Two sections of the tube described on Figure 18 are subjected to different flow conditions and thus to forcing functions of different power spectral densities and spatial correlations. The first few modes are considered in the response.

-23-

Ideally our approach to the vibration analysis of heat exchanger and steam generator designs should be that out- lined on Figure 20. That is: starting from an initial

design, given the flow conditions (1); the flow distribution and velocities are calculated (2) ; this indicates the excitation mechanisms and permits the formulation of the

forcing function g(x,t) (3); the

the system (tube bundle) which needs to be defined in

terms of f , <j>.(x), ç (4); then the

in the form of y(x,f), the dynamic stresses cr(x,f) and

the forces at

is to predict fatigue and fretting damage

latter is the input to

response is calculated

the supports F(x,f) (5); the next step

(6) ; this

leads to the last step which is to either accept (7) or modify the design depending on whether or not there are problems. The response calculation technique described earlier essentially links (3) to (5). We are not yet at this ideal stage. It is sometimes difficult to determine flow velocities in complex three-dimensional flow path particularly in two-phase flow. We do not yet have enough information to formulate the forcing function in all cases. We would like more tube damping numbers. It is desirable to express the tube-to-tube support dynamics in terms of the statistical properties of the impact forces. This will likely prove to be the criterion governing the vibration- fretting relationship. Finally we need to understand better the vibration-fretting relationship for different materials in various relevant environments.

Our practical approach to design analysis is as follows:

1) avoid fluidelastic instabilities; 2) make sure the tube response to random excitation is low enough to avoid fretting or fatigue problems; and 3) avoid periodic wake shedding resonance or demonstrate it is not a problem.

-24-

When our response calculation technique is not sufficient to satisfy the above specifications we can use it to compare the design under study to that of an existing satisfactory design. The calculation technique is then used as a normalization tool. Alternately we can test a model of the region in doubt 21 . It is also possible to conduct a fretting endurance test on a single tube sub- jected to the vibration response we estimate using our response calculation technique. If the heat exchanger component is easily accessible after installation in the reactor system, we can measure its vibration behaviour and take corrective action subsequently if necessary. For this purpose we have developed in collaboration with a manufacturer a very sensitive biaxial accelerometer probe that can be inserted in the tubes during operation (see Figure 21).

A similar approach may be used to analyse other nuclear station components. For instance we have developed a comprehensive computer model for the dynamics of CANDU-BLW type fuel strings in collaboration with Paidoussis 18 . The model analyses the stability of a fuel string and predicts its forced vibration response. The model is based on a matrix type formulation analogous to Equation 1 to suit the system of discrete fuel bundles. Currently a dynamic model for CANDU-PHW fuel bundles in horizontal fuel channels is being developed at the Whiteshell Nuclear Research Establishment 28 ' 29 .

-25-

CURRENT VIBRATION STUDIES Vibration Behaviour of Nuclear Fuel in Reactor

Vibration studies of nuclear fuels are usually conducted in out of reactor adiabatic test facilities. Under actual reactor conditions, the mechanical characteristics of the fuel are affected by thermal expansion of, in particular, the U0_ fuel pellets. Also, under diabatic conditions, additional flow-induced vibration excitation sources are possible, e.g. enhanced cross-flow between fuel bundle subchannels due to possible enthalpy imbalance. We have studied the effect of in-reactor conditions on typical CANDU-BLW fuel bundles in the experimental reactor NRU at the Chalk River Nuclear Laboratories 38 .

A string of five fuel bundles was inserted in a two- phase test loop simulating a CANDU-BLW fuel channel as shown on Figure 22. Fuel vibrations were measured with in- tegral lead weldable strain gauges installed on seven typically located fuel elements. Figure 23 shows a typical strain gauge installation on a fuel bundle. Measurements were taken over a wide range of flow conditions, i.e., from 0 to 100% fuel power (0 to 100 W/cm 2 heat flux),

from 70 C of subcooling to 25% steam quality, at pressures of 28 to 90 bars, and at mass fluxes up to 4600kg-m — 2—1 ^s .

Steam was generated by the fuel and/or added at the inlet of the test section from external boilers.

We investigated in particular the effect of fuel power. The natural frequency of fuel elements increases rapidly by roughly 50% during the first reactor start-up as shown on Figure 24. During the first shut-down it decreases quickly down to 75% power and remains essentially constant at lower power. The second start-up and serond shut-down.

-26-

are somewhat similar to the first shut-down. This behaviour is explained in terms of fuel rigidity increase due to fuel pellet expansion with power. During the first shut- down and subsequent cycles the frequency vs power relation- ship is different than during the first start-up because then the fuel sheath has already been deformed plastically by the first start-up. Then it takes a higher reactor

power for the fuel to expand increase its rigidity.

firmly in the sheath and to

The fuel element vibration behaviour is much dependent on fuel history. This is attributed to change in rigidity, internal damping and boundary conditions due to UO. pellets expansion inside the -Fuel sheath, element bowing and other geometrical changes. This is shown on Figure 25 where vibration spectra taken at different times under essentially similar conditions are compared for a typical fuel element.

We have found that fuel element vibration amplitudes were generally small being less than 10 vim RMS under normal CANDU-BLW operating conditions.

Vibration Damping and Support Dynamias of Heat Exchanger Tubes We are currently studying the damping behaviour of heat

exchanger tubes. The experiments are done on tubes of different diameters ranging from 0.75 to 2.5 mm. The tubes are installed in the trough shown on Figure 26 where they

in any other fluids to study

the effect of viscosity. Single and multispan tubes are tested with both idealized or realistic heat exchanger supports. The effect of frequency is explored by varying span length. To obtain the damping values, we use both the simple logarithmic decrement technique and the frequency response method.

can be immersed in water or

-27-

Typical vibration damping results are shown on Figure 27 for a simply supported 12.7 mm diameter heat exchanger tube. The net viscous damping due to water decreases with frequency.

We are now preparing tests to study the dynamics of the tube-to—tube support interaction. This is particularly important when the tube-to-support clearance is significant. Our intentions are to measure the statistical properties

of the impact

supports when realistic vibration amplitudes are simulated. We plan to use this information to correlate

forces generated by the tubes at the tube

vibration response and fretting-wear data.

Other Vibration and Related Studies Currently Underway We have discussed above two typical vibration studies related to nuclear components. Other experimental and analytical investigations are underway, such as:

1) Vibration vs Fretting Relationship: This is the subject of an extensive program for both nuclear fuel

and heat exchanger materials 31 ' 32 . The effects of

parameters such as frequency, clearance, amplitude and impact forces are investigated in both laboratory and realistic environments.

several

2) Analytical Modelling of the Dynamics of Tube-to- Support Interaction: An analytical model is being deve- loped to treat the problem of tube-to-tube support impacting in heat exchangers 33 . It takes into considera- tion the non-linearity due to tube-to-tube support clea- rance.

3) Vibration of Heat Exchanger Tube in Liquid Cross-Flow:

This work 21 * is continuing. Several different triangular and square heat exchanger tube bundle geometries have

-28-

been studied. We are now investigating the effect of irregularities such as the presence of sealing strips, sealing rods and tube free lanes on neighbouring tube vibration response.

4) Vibration of Tube Bundles in Two-Phase Cross-Flow:

We are preparing further experiments in support of steam generator designs. Air-water mixtures will be used to simulate steam/water two-phase flow.

5) Dynamics of Flexible Cylinder in Confined Flow:

Further experiments are underway particularly to explore the dynamics and stability of flexible cylinders subjected to two-phase axial flow.

6) Nuclear Fuel Dynamic Parameters: Tests have been done to determine fuel bundle and fuel string dynamic parameters such as dynamic stiffness, viscous damping, hydrodynamic mass and structural damping. We are preparing further tests particularly to study hydrodynamic mass and damping in two-phase flow.

8. CONCLUDING REMARKS

It is concluded that, although there are still areas of uncertainty, most flow-induced vibration problems can be avoided. This requires that nuclear components be properly analysed at the design stage and that the analyses bd supported by adequate testing and development work. There has been no case yet where vibration considerations have seriously constrained the designer. Although some- times difficult to analyse, vibration problems usually require simple solutions.

-29-

ACKNOWLEDGEMENT: Many people have contributed to the work discussed in this paper. Among those are R.I. Hodge, R.B. Turner, A.O. Campagna, P. Tiley, Y. Sylvestre, J. Platten and P.L. Ko of the Chalk River Nuclear Laboratories; I. Oldaker of the Whiteshell Nuclear Research Establishment; M.P. Paldoussis of McGill University; D.G. Gorman of the University of Ottawa and C.F. Forrest and N.L. Carlucci of Westinghouse Canada Ltd. The author is very grateful to all.

REFERENCES

-30-

1. R.I .

Hodge,

J.E .

LeSurf,

J.W.

Hilborn ,

"Steam

Generator

Reliability ,

The Canadian

Approach",

Presente d

a t

th e

XIX Nuclea r

Congress

o f

Rome,

March

1974 ,

als o

Atomic

Energy

of

Canada

Limited

Report

AECL-4471

(1974) .

2. D.G.

Dalrymple,

"Current

Canadian

Use

o f

Explosive

Welding

fo r

Repair

and Manufacture

o f

Nuclear

Steam

Generators",

AECL

Research

and Development

i n

Engineering ,

Atomic

Energy

of

Canada

Limited

Report

AECL-4427,

Winter

1972 .

3. R.T. Hartlen , "Recent Fiel d

of

Heat

Exchanger

Experience

Tubes",

with

Flow-induced No. 611 ,

Vibration

Internationa l Symposium on Vibratio n Problems i n Industry ,

Paper

Keswick,

U.K.

1973 .

4 R.I .

.

Hodge,

P.L .

Ko,

and A.O.

Campagna,

 

Personal

Communication,

Aug. 1976 .

 

5 E.P .

.

Quinn,

"Vibratio n

o f

Fuel

Rods

i n

Paralle l

Flow",

U.S .

Atomic

Energy

Commission

Report

GEAP-4059

(1962) .

6 Y.N .

.

Chen,

"Flow-induce d

Vibration s

i n

Tube

Bundle

Heat

Exchangers

with

Cross

and

Paralle l

Flow.

Part

1:

Paralle l

Flow",

Symposium

on

Flow-induced

i n

Heat

Exchangers ,

New York:

ASME 5 7-6 6

Vibratio n (1 9 70) .

 

7.

M.P. Pal'doussis , "Stabilit y of Flexibl e Slende r Cylinder s

i n

Pulsatil e

Axia l

Flow",

J .

of

Sound

and

Vibration ,

42

(1) ,

1-11

(1975).

 

8.

M.P. Païdoussis , "The Dynamical

Behaviou r of Cylindrica l

Structure s

i n

Axia l

Flow",

Annals

o f

Nuclear

Scienc e

and

Engineering,

Vol.

1,

No.

2,

pp

83-106

(1974).

9.

M.P. ?aïd?y»sif» , "Dynamics of Cylindrica l Structure s

Subjecte d t o

Axia l

Flow:,

J .

of

Sound

and

Vibration ,

Vol.

29,

No.

3,

pp.

365-385

(1973).

 

10.

M.J.

Pettigrew , M.P. Païdoussis , "Dynamics and Stabilit y

of

Flexibl e

Cylinder s

Subjecte d

t o

Liquid

and

Two-Phase

Axia l

Flow

i n

Confined

Annul!",

Paper

D2/6 ,

3rd

Interna -

tiona l

Conference

on

Structural

Mechanics

in

Reactor

Technology ,

London,

U.K.

Sept ,

1-5 ,

1975 ,

als o

Atomic

Energy

o f

Canada

Limited

Report

AECL-5502

(1975) .

-Si -

11 M.P. Païdoussis , "Mathematical Model fo r

.

the

Dynamics

of an

Articulate d Strin g of

Fuel

Bundles

in

Axial

Flow",

Paper D2/5 presented at the

3rd

Internationa l

Conference

on Structura l Mechanics U.K., Sept. 1-5, 1975.

i n

Reactor

Technology

in

London,

12 W.T.

.

Thomson,

"Vibratio n

Theory

and

Applications" ,

Prentice-Hall ,

Englewood

Cliffes ,

N.J. ,

1965.

 

13 L.

.

Meirovitch ,

 

"Analytica l

Methods

in

Vibration" ,

Macraillan

Company,

N.Y. ,

1967 .

 

14 S.H .

.

Crandall ,

and

W.D.

Mark,

"Random

Vibratio n

i n

Mechanical

Systems",

Academic

Press ,

N.Y.,

1963.

15 D.J . Gorman, "The Role o f Turbulence

.

i n

the

Vibratio n

of

Reactor

Fuel

Elements

in

Liquid

Flo"",

Atomic

 

Energy

of

Canada

Limited

Report

AiîCL-3371

(1969) .

16 D.J .

.

Gorman,

"An

Analytica l

an d

Experimenta l

Investigatio n

of

the

Vibration

of

Cylindrical

Reactor

Fuel

Elements

in

Two-phase

Paralle l

Flow",

J.

Nuclear

Science

Engineering

44.

277-290

(1971).

 

17 J.R . Reavis , "Vibratio n

.

Correlatio n

fo r

Maximum

Fuel -

element

Displacement

in

Paralle l

Turbulent

 

Flow",

J.

Nuclear

Science

Engineering

38,

63-69

(1969).

 

18 D.J .

.

Gorman,

"Experimenta l

and

Analytica l

Study

o f

Liquid

and

Two-Phase

Flow-Induced

Vibratio n

i n

Reactor

 

Fuel Bundles" ,

ASME Paper

75-PVP-52,

2nd

Nationa l

Congress

on

Pressur e

Vessel s

and

Piping ,

San

Francisco ,

June

23-27 ,

19 75.

 

19 .

M.J.

Pettigre w

and

D.J .

Gorman,

"Experimenta l

Studie s

 

on

Flow

Induced

Vibratio n

to

Support

Steam

Generator

Design ,

Part

1:

Vibration

of

a

Heated

Cylinder

in

Two-

Phase

Axial

Flow",

Paper

No.

424 ,

Internationa l

Symposium

on

Vibratio n

Problem s

i n

Industry ,

Keswick ,

U.K.

1973 ,

als o

Atomic

Energy

 

of

Canada

Limited

Report

AECL-4514

(1973) .

 

20 .

S. Mirza and D.J . Gorman, "Experimental and Analytica l

Correlatio n of Local Drivin g Force s and

Tube

Response

Liquid

Paper F6/5 , 2nd Conference

Berlin

Reactor

Flow

Induced

Vibratio n

on

of

Structura l

Heat

Exchangers" ,

Mechanics

Technology,

1973.

i n

i n

-32-

21. M.J. Pettigrew, J.L. Platten, Y. Sylvestre, "Experimental Studies on Flow Induced Vibration to Support Steam Generator Design, Part II: Tube Vibration Induced by Liquid Cross-flow in the Entrance Region of a Steam Generator". Paper No. 424, International Symposium on Vibration Problems in Industry, Keswick, U.K. 1973, also Atomic Energy of Canada Limited Report AECL-4515 (1973).

22. M.J. Pettigrew, D.J. Gorman, "Experimental Studies on Flow Induced Vibration to Support Steam Generator Design, Part iii: Vibration of Small Tube Bundles in Liquid and Two-phase Cross-flow", Paper No . 424 , International Symposium on Vibration Problems in Industry, Keswick, U.K. 1973, also Atomic Energy of Canada Limited Report AECL-5804 (1977).

23. H.J. Connors, Jr., "Fluidelastic Vibration of Tube Arrays Excited by Cross Flow", Proceedings of the Symposium on Flow Induced Vibration in Heat Exchangers, ASME Winter Annual Meeting, New York, Dec. 1, 1970, pp. 42-56.

24. D.J. Gorman, "Experimental Development of Design Criteria to Limit Liquid Cross-Flow Induced Vibration in Nuclear Reactor Heat Exchange Equipment", J. Nuclear Science and Engineering 61, 324-336 (1976).

25. Y.N. Chen, "Fluctuating Lift Forces of the Karman Vortex Streets on Single Circular Cylinders and in Tube Bundles, Part 1: The Vortex Street Geometry of the Single Circular Cylinder, Part 2: Lift Forces of Single Cylinders, Part 3: Lift Forces in Tube Bundles", ASME Transactions, Series B, J. of Engineering Industry, Vol. 94 (2), 603-628 May 1972.

26. P.R. Owen, "Buffeting Excitation of Boiler Tube Vibration", J. Mech. Eng. Sci. 7 (4), 431-439, 1965.

27. E.R. Darnley, "The Transverse Vibration of Beams and the Whirling of Shafts Supported at Intermediate Points", Phil. Mag. Vol. 41 (241), 56 Jan. 1921.

28. I.E. Oldaker, A.D. Lane, M.P. Pal'doussis and CF . Forrest, "An Overview of the Canadian Program to Investigate Vibration and Fretting in Nuclear Fuel Assemblies", May 1974. 73-CSME-89, EIC-74-Th; Nuc. 2 Engineering Journal, Fall, 19 74.

-33-

29 D.J . Jagannath, "A Model fe r Vibratio n of Nuclear

.

Fuel

Bundles"

(to

be

published).

 

30 M.J.

.

Pettigre w

and

R.B .

Turner,

"The

In-reacto r

 

Vibration

Behaviour

of

Nuclear

Fuel",

Paper

D3/7,

Inter -

nationa l

Conference

on

Structural

Mechanics

in

Reactor

Technology,

Berlin ,

Sept.

1973.

 

31.

P.L .

Ko,

"Impact

Frettin g

of

Heat

Exchanger

Tubes",

 

Atomic

Energy

of

Canada

Limited

Report

AECL-4653

(1973) .

32

.

P.L .

Ko,

"Fundamental

Studie s

of

Steam

Generator

and

 

Heat

Exchanger

Tube

Fretting" ,

publishe d

i n AECL

Research

and

Development

i n

Engineering ,

Winter

1975 ,

Atomic

Energy

of

Canada

Limited

Report

AECL-5310

(1975) .

33 .

R.J. Rogers, R.J . Pick , "On the

Dynamic

Spatia l

Response

o f a Heat Exchanger Tube with Intermitten t Baffl e Contacts" ,

Nucl.

Engrg.

and

Design ,

36 ,

81-9 0

(1976) .

-35-

FIGURE la: Fretting-Wear of Steam Generator Tubes:

Fretting Damage at Midspan.

-36-

Figure lb: Frettlng-Wear of Steam Generator Tubes; Fretting Damage and Hole at Support Location.

-37-

FIGURE 2: Typical Example of Heat Exchanger Tube Fretting- Wear,

-38-

FIGURE 3: Fretting Damage on Gentilly-1 Fuel Bundle.

GENTILLY

Nuclear Power Station

ORDINARY WATER

I.

•••.••! STEAM

RIVER WATER

liiiiii

HELfUM GAS

HEAVY WATER MODERATOR

TURBINE-GENERATOR BUILDING

ELECTRICITY

RIVER WATER INTAKE BAY

FIGURE

4 :

Simplified

Flow Diagram of CANDU-BLW Station.

RIVER WATER OUTLET

i

VO

I

G6NTILLY 1

CENTRALIZING PADS SPACERS BEARING PADS UO2 FUEL PELLETS ZIR.CALOY 4 SHEATH DELINEATING DISC END CAP END PLATE

PATTES DE CENTRAGE CALES D'ECARTEMENT

PATTES

PASTILLES

GAINE

DISQUES

Of

D'APPUI

DE

UO j

EN ZIRÇALOY

4

SEPARATION

BOUCHON D'EXTRÉMITÉ

PLAQUE D'EXTREMITE

SECTION THROUGH CENTRE OF FUEL BUNDLE

COUPE TRANSVERSALE

GRAPPE

PE

LA

COMBUSTIBLE

FIGURE 5: Gentilly-1 Fuel Bundle.

O

I

FIGURE

6:

Sketc h

of

-41-

OUTLET

FUEL BUNDLE (^ Z7 kg x 1OJ

10.4

cm

CENTRAL

SUPPORTING

TUBE

PRESSURE

TUBE

SPRING

ISSEMBLV

SHIELD

y—PLUG

INLET

CANDU-BLW Fue l

Channel ,

BOOSTER ROD

OUTLET NOZZLE

 

-42-

GENTILLY-Ï

CALANDRIA

ABSORBER

SUIDE TUBE

D

2 0

D

2 0

BOOSTER ROD

 
 

NOZZLE

 

TOP VIEW

CALANDRIA TUBE

PROTECTIVE SHROUD

GUIDE TUBE

FIGURE 7a: Modification with Protective Shroud.

REACTOR

BUILDING

I

I

MODERATOR

J

|

|

ORDINARY

WATER

HEAVY

WATER

HELIUM

GAS

LAKE

WATER

TURBINE.GENERATOR

BUILDING

FIGURE 8: Simplified Flow Diagram of CANDU-PHW Station.

MANWAY I N WATER BOX|

(ALSO

-44 -

 

BEND RADIUS

OF TUBES

TUBE BUNDLE

 

BAFFLE OR LATTICE BAR TUBE SUPPORTS

PRIMARY (IN

TUBES)

SECONDARY (IN

SHELL)

 
 

PREHEAT SECTION

 

(OR

IE G

I N U SHELL UNITSI

DOWNCOMER

OR

FEED WATER

NOZZLE

PRIMARY CHANNEL COVER

 

TUBE SHEET CLADDING

DIVIDE* PLATE

WATER BOX

-45-

FIGURE 10:

Clamped-Free Cylinder with Bullet-Shaped Downstream End Experiencing 4th Mode Buckling In Liquid Flow.

 

30 H

 

LEGEND

PREDICTED rms DISP

 

25

 

MEASURED rms OISP

E

 

20

o

TOTAL MASS FLOW

CI

RATE = 0.8fc kg/ s

to

o

a

15

o

IE

I

 

10

10

20

I

30

40

ON

I

SIMULATED QUALITY {%) FIGURE 11 MEASURED AND PREDICTED VIBRATION AMPLITUDE vs SIMULATED STEAM QUALITY IN TWO-PHASE AXIAL FLOW'•

00

0.5

3.Û

0.3

-47 -

Û.2

MASS FLUX;

47

g/(s-cm 2 )

PRESSURE;

D 2.86

MN/m 2

 

A

3.55

MN/m 2

O

4.23

MN/m 2

0. Î

O5.6 1

MN/m 2

 

STEAM QUALITY;

TO 65% MAX.

FIGURE 12 :

b

1

9

FLOW VELOCITY

12

(m/ s )

Effec t

of

Steam

Qualit y

and

15

Pressure .

Vr

10- 1

io- 2

lu" 3

4

10- 4

1

- 43 -

1

W 5

Ai è

BURGREEN

QUINN

e t

SOGREAH

al

ROSTRÔM & ANDERSON

ca

PAIDOUSSIS

i

ID" 3

10" 2 Ui.85 L 3 -V(EI)-8

5x10- 4 K a"*

+ML 2 U 2 /(EI))J|D 2 - 2

1

10- 1

1 + 4M/m

FIGURE 13 AGREEMENT BETWEEN MEASURED AND PREDICTED VIBRATION RESPONSE INAXIAL FLOW USING PAIOOUSSIS SEMI - EMPIRICAL EXPRESSION

a.

en

t—

C£. "=C CO 2 :

= 3

SO

x

LLJ

o

1 .5

1 .0

-49 -

0.5

PRESSURE:

Q

4.23

MN/m

2

 

O

5.61

MN/m

2

 

I

I

I

 

50

100

150

200

MASS

FLUX

(g/(s-cm 2 ))

FIGURE

14 :

Effect

of

Mass Flux and Pressure.

 

moo

100

10

0.1

P/d

f(Hz)

S

C

CONNORS

1.41

11.8

- 40

0.008 - 0.16

G

GORMAN 8 MIRZA 1.33

38

0.112

PI

PETTIGREW

1.5

30

0.156

P2

PETTIGREW

1.6

17

0.168

A

••LIQUI D

FLOW

^

^

3

^

TWO PHASE:

 

OAt>pO AIR

 

X ^

 

S

INSTABILITY NOT

 
 

OSINGL E

ROW

ANORMAL

TRIANGULAR

>PARALLEL

TRIANGULAR

DNORMAL

SQUARE

 

O

ROTATED

SQUARE

FIGURE

15 :

1.0

10

V „:

Approach

velocity

normal-

ized

for

uniform

flow

velocity-

 

100

1000

Non-Dimensional Presentation of Experimental Thresholds for Fluidelastic Instabilities.

O

I

v

-51 -

fr fry 5

VORTEX SHEDDING S = fD/V

2)V

0

BUFFETING

OO

oo

TURBULENT

EDDIES

CONTROLLED BY GEOMETRY

FIGURE 16: Postulated Mechanisms for Periodic Excitation.

T

UJ

«a

i

a

z:

1.00

1.00

0.75

0.50

0.25

0

••• •

FIGURE 17:

1

1

0 . 2

//

O

MEAN

1

1

fi / ^

cT«L

6

/

rtO

\ j O

WATER

1

0. 4

VELOCITY

(m/s )

1

0. 6

1

o

i

'

1

0.8

Vibration Response of First Upstream Tube In Liquid Cross-Flow 20 .

to

I

11

1.0m

-53-

OUTLET

CROSS FLOW

(20% QUALITY)

12

r :

o

CO

-I-

333 kg/5n*s>

CO

\ CLEARANCE

HOLES

(PINNED

,

SUPPORTS)

1 FIXED SUPPORTS

1

•*

PARALLEL

FLOU

210

kg/m 2 s

(SATURATED

TO 20»

QUALITY)

INLET

CROSS FLOU

(SATURATED)

359

kg/4n z .s>
P

FIGUR E 18 : Typical Steam Generator Tube and Flow Conditions.

-54-

M M m

CROSS FU I EICI m i CM

SUPPORT *

«MS

«EWTION

0

.00 9

.»0

FOilCES

.114

(H)

•<»

TOTAL

.190

TOTAL

RMS

AMPLITUDE

1

.016

.019

.25!

012 .007

OSS

.111

164 .215

195 .106

HINDOU PARALLEL FLO! EXCITATION

ODE

1

2

J

FREQUENCY

«I

37.2

47.7

49.6

MOOES 1 TO3

SUPPORT 1

RMS

REACTION

FORCES

H

TOTAL

I0KL

«MS

(HPLITUDE

15

1 0

>

.mo

.119

.095

.174

_

4

MODE

FREQUENCY

 

Hi

.022

1

34.7

.0»

3

«t.S

.060

I

59.7

.026

4

73.0

.068

MODES 1 TO4

 

0

.090

.209

.200

.113

.352

.352

.095

.«05

LENGTH (m)

FIGURE 19:

Example of Tube Response Calculation.

-55-

STAR T

11

DESIGN: GEOMETRY, FLOW CONDITIONS Calculation of Flow Distribution and Velocities

EXCITATION MECHANISMS Excitation Forcing Function

SYSTEM: TUBE DYNAMICS Damping £, Modes <L(x) and Frequencies f.

RESPONSE CALCULATION y(x,f), O(x,f), F(x,f)

FRETTING & FATIGUE - VIBRATION RELATIONSHIP Damage Estimate

FIGURE

20 :

Ideal

YES

Approach

NO

I7

ACCEPTANCE

to

Design Analyses

of

Heat

Exchangers.

I

FIGUR E 21 : Biaxir?. Accelerometer Probe for Heat Exchanger Tube Vibration Measurements.

-57 -

147.40m

— DECK PLATE

146.79m

TOP CLOSURE

140.51

Q-

139.27m —

138.81m —

136.89m —

136.78m —

136.40m-=*

'—? »

OUTLET

HANGER ROD

 
 

-SG3

SG2

 

U-118-IKX-X' )

 
 

STRAIN GAUGE

 

LEADS

 

SG191

SGI 93

 

TOP

INSTRUMENTED

SG151

SG153

 

G192

X

1

BUNDLE

 
 

SG152

 

FUEL

STRING

 

SGI

I I

 

SGI 13

   

SGI12

 

PRESSURE TUBE (104 mm I.D.)

 

U-118-HX-X' )

 
 

'SPACER FOR LEADS •BOTTOM INSTRUMENTED BUNDLE

SGI 5

STRAIN GAUGE

SPRING

 

CENTRAL

SUPPORT

U-118-IKY-Y' )

 
 

STRAIN GAUGE

LOCATION

 

MIXER

 
 

STEAM

INLET

 

FIGURE

22:

Strain Gauge

Instrumented

 

WATER

INLET

Fuel

String

Installed

in

 

U-l

LOOD of

Reactor

NRU.

FIGURE

23 :

Details

of Weldable Strain Gauge

Installatio n

on a Fuel Bundle.

00

~Z

u

§

£

70 r -

-59 -

60

55

5 0

45

40

35

 

50

 

SG 1&3 U-118-I I SG 5

OM

SG 12&14 SG 15S17 SG 18S20

*

S3 151 U-118-I

X

SG 191 U-118-I

-VOADO* X

: START-UP

: SHUT-DOWN

 

50

FIGURE

24 :

Fuel

Element

Natural

75

100

POWER(3!)

75

100

POWER(«)

Frequency vs Reactor

Power

7

u

rttcucNcr

i

en

o

FIGURE 25: Effect of Fuel History on Fuel Element ViDration Behaviour [SG No. 15, Liquid Flow, U-118-Il]

- 61 -

FIGURE 26: Heat Exchanger Tube Immersed in Trough to Study Vibration Damping.

1/1

LU

_ l

z

o

to

z

LLJ

o

o

o

a.

o.ior

0.01

0.001

10

FIGURE

27 :

-62-

y = 0.756.X-

TUBE

12.7 mm O.D. 304 S.S. TEMP.

19.4-C

O AIR MEDIUM

• WATER MEDIUM NET VISCOUS DAMPING

100

fn(FREQUENCY) Hz

Typical Tube Vibration Damping Result s Showing

the

Effect

of

Frequency.

The International Standard Serial Number

ISSN 0067-0367

has been assigned to this series of reports.

To identify individual documents in the series we have assigned an AECL-number.

Please refer to the AECL-number when requesting additional copies of this document from

Scientific Document Distribution Office Atomic Energy of Canada Limited Chalk River, Ontario, Canada

KOJ1J0

Price $5.00 per copy

1638-77