Earthquake and Electrical Equipment

8/12/2019 Earthquake and Electrical Equipment

1/28

n180

earthquakesand electrical

equipment

E/CT 180(e)first issued, April 1997

Eric MELMOUX

Graduated in mechanicalengineering from the INSA School,Lyon in 1981, and obtained a DEA(equivalent to 1st year of Phd) invibrations the same year.After ten years spent with acompany which specialises in sound

and vibrations, he joinedMerlin Gerin.He is, at present, head of theshocks and vibrations group atSchneider Electric.


2/28

Cahier Technique Merlin Gerin n180 / p.2

glossary

Epicenter point at ground level, vertically above the hypocenter.

Frequency appropriation excitation frequency corresponds with resonance frequency of the structure.

Hypocenter or seismic focus position of the earthquake in the earths crust.

Intensity measures the force of the earthquake in terms of the effects produced (MERCALLI

scale).LOVE Waves correspond to the horizontal component of the surface waves.

Magnitude measures the force of the earthquake in terms energy released at the seismic focus(RICHTER scale).

Meshing action whereby a complex structure is broken down into its beam gantrys - plates -volume.

MHEL Maximum Historical Earthquake Likelihood for a site.

Modal shape oscillating deformation adopted by an elastic structure during excitation on one of itsresonance frequencies.

Qualification process which consists in establishing the appropriate withstand capabilities forequipment with required or normal stresses.

RAYLEIGH Waves correspond to the vertical component of the surface waves.

Response Spectrum device which enables a characterization of the earthquake to be effectuated in terms ofits effects on a simple structure.

Seismic Activity violent movement of tectonic plates which produces an earthquake.

Single DOF mechanical system single degree of freedom structure characterized by mass, spring and damper.

Space appropriation excitation forces apply pressure on the antinodes of the modal shape.

SSE Safe Shutdown Earthquake (MHEL plus one degree on the MERCALLI scale).

Strong part of response spectrum corresponds to the frequencies which cause the structure to amplify groundmovements.

Time-history recording of ground acceleration during the earthquake.

Zero Period Acceleration ZPA part corresponds, on a response spectrum, to the frequencies which cause the structure totrace ground accelerations without amplification of motion.


3/28


earthquakes and electrical equipment

contents

1. Earthquakes Causes - location p. 4

Propagation of seismic waves p. 4

Characteristics of vibrations p. 4generated at ground level

Intensity and magnitude p. 4Seism characterization p. 6

Defining the seismic severity p. 8of a site

Reading the response spectrum p. 9applicable to a piece of equipment

2. Dynamic behaviour of structures Brief summary of single degree p. 12of freedom oscillator

Elastic structures (with N p. 12degrees of freedom)

3. Equipment design Defining objectives p. 16

Design principles p. 16

Simulation by analysis at p. 18

design stage4. Qualification by simulation or test Introduction p. 20

Combined qualification p. 20(numerical analysis andexperimental tuning)

Qualification by real size p. 21tests preceded by numericalanalysis

Qualification by test p. 22

5. Conclusion p. 25

6. Bibliography p. 26

In all countries there is either a zone ofsignificant seismic activity orinstallations which require high securityin order to operate (e.g. nuclear power

stations, which generally have lowseismic activity). In both cases theelectrical and control and monitoringequipment must assure their safetyfunctions correctly.

This technical paper aims to facilitatedialogue between operators andspecialists.

After briefly summarizing theearthquake phenomena and the wayin which they are specified, the authorpresents the theoretic approachrequired for seismic withstand

capabilities to be taken into account atthe design stage.Both design and qualification are,today, increasingly requiring numericalanalysis and, as a result, powerfulscientific and technical data processingmethods.


4/28


1. earthquakes

causes - locationThe majority of earthquakes occur onthe fault lines which demarcate thetectonic plates from the earths crust.Tension accumulates when the platesslowly shift away from each other. Thesudden release of distorting energywhich has thus accumulated inside theearths crust, or in the underlying layercalled the mantle, provokes a localagitation. Some of this energy is thentransformed into seismic waves at the

surface of the ground.It is the creation of a fault line, or morefrequently the slide along an existingfault line, which constitutes thegenerating mechanism of anearthquake. The place where it occursis called the seismic focus or thehypocenter and the projection fromthis point to the ground is called theepicenter(see fig. 1).The depth of the hypocenter variesgreatly: from a few kilometres to upto 100 km.

propagation of seismicwavesEarthquakes propagate in waveswhich, taking the heterogeneity of theground into account, provoke acomplex vibratory movement at thesurface which is difficult to predict for agiven site.

A distinction is made between twotypes of wave: bulk waves and surfacewaves.

Bulk waves

They originate at the seismic focus andpropagate inside the earths mantle intwo different forms:clongitudinal waves characterised byalternating compressions and dilationswhich propagate at a speedof 7 to 8 km/s,ctransversal waves characterised by aplane distortion perpendicular to the

direction of the propagation, whichprovoke shear and propagate at aspeed of 4 to 5 km/s (see fig. 2).

Remark:

It is the difference in speed betweenthe longitudinal and transversal wavesand recordings taken from severalseismographs, which enables an earth-quakes seismic focus to be located.

Surface waves

These are generated by the bulk waveswhich reach the surface and propagateat a speed of 1.5 to 5 km/s.

A distinction is made between:cRAYLEIGH waves which cause theground points to describe ellipses in thevertical plane; they engendercompression and shear in the ground,cLOVE waves which cause the groundpoints to shift at a tangent to thesurface, perpendicular to thepropagation direction; they engendershear only (see fig. 2).

characteristics ofvibrations generated atground levelIn reality things are far more complex;the propagation of a seismic wave in aheterogeneous environment provokes acomplex system of refracted andreflected waves for each discontinuity,so that the seismic movement iscompletely random at ground level.

However, vibratory movements broughtabout at ground level by earthquakesdo produce common characteristics,and a certain number of parameters are

generally employed to describe them.

Characteristics of random vibrationsprovoked at ground level by anearthquake:

cdirectionThe movement is made up ofsimultaneous independent vertical andhorizontal components;

cduration

It is usually between 15and 30 s(anintense earthquake can last between60 and 120 seconds).

cfrequencyBroad band random movementproduces preponderant energybetween 1 and 35 Hz, and provokesthe most destructive effects atbetween 1 and 10 Hz.

clevel of acceleration

There is no correlation between thewaves observed in the two differentdirections: at any given moment theamplitudes and frequencies areindependent.

Horizontal ground acceleration isgenerally lower than 0.5 g(exceptionally higher than 1g,or 10 m/s2).Vertical acceleration has a loweramplitude. Observations show that therelationship between the maximumvertical and horizontal amplitudes is in

the order of 2

3(for frequencies higher

than 3.5 Hz).

intensity and magnitude

Intensity

The scale of an earthquake is generallymeasured in terms of its intensity at theobservation site. This subjectiveevaluation is established in terms of theeffects felt by the population and thedamage incurred.

Different intensity scales have beendefined, which class the seismic effectsin order of increasing size, with the helpof some conventional descriptions:

cthe MERCALLIscale describescommonly observed effects on theenvironment, buildings and man afteran earthquake,

cthe MSKscale (or modified Mercalliscale), more precise than the original,includes an evaluation of the damage,


5/28


fig.1: geosismic vocabulary and characteristic features.

fig.2: seismic bulk and surface waves.

the type of building and percentage ofbuildings affected.

These estimations are useful forassessing the size of earthquakes ifthere are no time-histories or spectraavailable. They do not enable theseismic load of a site to be specified.

Magnitude

Another method of determining thescale of an earthquake is to establishits magnitude, the intrinsiccharacteristic of an earthquake whichmeasures the total energy released.Magnitude, defined in1935 byRICHTER, and the scale which ensued,are used universally.Magnitude is, in practice, determined

according to ground movementrecordings taken at a number ofobservation points at various distancesfrom the epicenter.From these observations seismologistscalculate the energy E (expressed inergs) of the earthquake, from which themagnitude M is deduced.The simplified empirical equation:log E = 9.9 + 1.9 M + 0.024 M2givesan approximate, but rapid calculation.

Intensity / maximum groundacceleration / seismic zonecorrelation

The table in figure 3 (see overleaf)establishes a correlation between thevarious subjective levels of intensity inthe modified MERCALLI scale and themaximum ground acceleration levelresponsible for the damage observed.This table also indicates the type ofseismic activity zones prone to suchseismic intensity (see fig. 4, page 7 forthe division of the worlds seismicactivity zones).

Intensity / magnitude correlation

Theoretically, no relationship can exist

between intensity and magnitude;intensity is dependent on the distancefrom the site in question to the seismicfocus, on soil type, the type offoundations used, the type of buildingand the duration of the earthquake.However, an approximate correlation isproposed by the experts (see table infigure 5, page 7).

, ,

, ,

, ,

distance from epicenter

focal distance

rock

fault line

hypocenter or seismic focus

epicenter

earth

P: longitudinal wavesbulk

ground level

hypocenter

vertical waves

horizontal waves

R

S: transversal waves

Q


6/28


intensity modified Mercalli scale approximate seismic

horizontal zone

acceleration

values in m/s2

1 Tremor not felt. zone 0

2 Tremor felt by those resting or situated on top floors.

3 Hanging objects move. Slight vibrations.

4 Vibrations similar to those provoked by a heavy lorry passing 2are felt. Window panes and crockery rattle.

Stationary vehicles sway.

5 Tremor felt outside buildings. Those sleeping are woken up. zone 1

Small objects fall down. Pictures are tipped sideways.

6 Tremor felt by everyone. Furniture is moved.

Damage: broken glass, objects thrown from shelves,

slight cracks appear in plastering.

7 Tremor felt in moving vehicles. zone 2

Those standing fall off balance, church bells ring. Damage: chimneysand other external architecture break away, plaster falls down,

broken furniture, extended cracks in plastering and masonry,

some houses in brick collapse.

8 Drivers in moving vehicles experience difficulties. 3 zones 3 and 4

Branches fall from trees. Fissures appear in waterlogged ground.

Destruction: water towers, monuments, brick houses.

Slight to subtantial damage: brick buildings, prefabricated houses,

irrigation works, causeways.

9 Sand craters in sandy and waterlogged ground in urban areas.

Landslides. Fissuring in ground. Destruction: non-reinforced brick

masonry. Slight to substantial damage: insufficiently

reinforced concrete structures, underground piping.

10 Landslides ans major ground destruction. 5

Destruction: bridges, tunnels, certain reinforced concrete

strutures. Slight to substantial damage: the majority of buildings,

dams, railway lines.

11 Permanent ground deformation.

12 Almost total destruction.

seism characterizationIntensity, magnitude or maximumground accelerations do not sufficewhen estimating the risks to a buildingor a piece of equipment. In fact, inorder to estimate the response of agiven structure, a more detailedknowledge of the duration andfrequency of the ground movement isrequired.

There are two methods forcharacterizing ground movement:ctime-history: = f(t);

cresponse spectrum whichcharacterizes the effectsproduced bythe seism on an elementary structure(1st order linear mechanical system).

Time-history

Ground acceleration evolution as afunction of time (see fig. 6). This type ofinformation, recorded by seismographsaccording to the three spatialdirections, is used to estimate theseismic risk incurred by equipment,when determining the withstand eitherby test or analysis.

The time-history is the only possible

option for determining the chronology of

a structures response to seismic

excitation, which is required when

ascertaining the evolution of thedifferent com-ponents relative

displacement over time.

However, this rarely figures in

specifications sheets either because it

is not available or because it does not

lend itself to the seismic severity

calculations of a site.

fig. 3: Mercalli scale.

Seismic zones correspond to the anticipated level of intensity according to observations carried out over a period of 200 years.


7/28


Mercalli 1 2 3 4 5 6 7 8 9 10 - 11 - 12

(intensity)

Richter 0-2 1-2 2-3 3-4 4-5 5-6 5-7 6-8 7-9 > 8

(magnitude)

fig. 4: seismic activity zones in the world.

fig. 5: indicative correlation between Mercalli and Richter scales.

The correlations between the Mercalli and Richter scales are completely indicative because they depend on soil type, distance from the seismic

focus (from 5 to 100 km), and earthquake duration.

Response spectrum

The response spectrum allows theearthquake to be characterized interms of the effects it produces onequipment. For this, the effect of thetime-history (in seismic waves) iscalculated on standardized equipment,that is to say, on an array of single

degree of freedom oscillators,caracterised by their reso-nancefrequency and damping values.

A single DOF system is characterizedby: a mass M, spring K and damper ;its resonance frequency is:

FrK

M= .

1

2

fig. 6: sample time history = f(t) record of horizontal north-south ground acceleration,El Centro - California, Mai 18, 1940.

4

0

m/s2(max)

6 8 10 12 14 16 18 20 22 24 t (s)

seismic zone0 - 1234


8/28


The maximum response of this systemto the seismic wave (maximumacceleration of mass) gives one pointon the response spectrum (see fig. 7).

By varying the resonance frequency(K/M), the curve obtained:max= f (Fr) (see fig. 8) is the responsespectrum which characterizes theseverity of the seismic wave for a givendamping.Figure 9 shows the family of curvesobtained when the damping is modified.The response spectrum generallyfigures in specifications sheets, so thatit can be applied in horizontaldirections. The vertical responsespectrum is deduced by using acoefficient.The response spectrum is the mostwidely used tool today for determiningthe seismic severity of a site, because itlends itself naturally to:cseverity comparison,celaboration of severity envelopes forseveral sites,csimple severity increases,

capproximate estimates of seismiceffects on equipment (damagepotential).

Remark:

The response spectrum must not be

confused with the FOURIER seriesexpansion of a periodic phenomenon,or the FOURIER transform of anaperiodic phenomenon, which are notused in seismic studies.

defining the seismicseverity of a site

Safe Shutdown Earthquake (SSE) -Maximum Historical EarthquakeLikelihood (MHEL)

Defining the seismic severity of a sitegenerally requires the sites geologicaldata and seismic history.In France for example, data from theexceptionally well documented seismichistory (100 years) enables the seismicrisk of a site to be established.This allows the Maximum HistoricalEarthquake Likelihood to be defined

fig. 7: application of seismic excitation (time-history, see fig. 6) to 1st single DOF, induces

accelerations. The maximum value (max) is by definition one point on the response spectrumof the seismic system.

fig.8: construction of seismic response spectrum (various K/M with constant).

Fr = 1

2.

K

M

M

t

K K

M

Fr

max

single DOF systems

time-history spectrum

M: mass

K: stiffness: damping

Mn

M1Mi

Fr

response spectrum

Fr1 Fri Frj Frn

K1 Ki Kn


9/28


which is likely to provoke the maximumeffect on a given site. For thedimensioning of works or equipment itis the SSE which is taken into account:

the SSE is equivalent to the MHEL plusone degree on the MSK scale (modifiedMercalli scale).

Basic response spectrum

Macro seismic data which correspondto the above definitions are notsufficient for the engineer who has todesign a building or an equipment. Hewill also require the representativeresponse spectrum of the siteconcerned, which is established byusing instrumental seismic data.

A seismotheque has been created

(readings taken in regions ofconsiderable seismic activity), whichcorresponds to a scale of magnitudes,seismic focus depths and epicentraldistances for very diverse geologicalcontexts. This seismotheque allows theform of the response spectrum, or basicresponse spectrum as it is called, to beestablished, for a given region, with itsamplitude depending on the chosenSSE.This response spectrum definesseismic severity at ground level. Theseismic severity for the storey where

the equipment will be installed still hasto be evaluated.

Dimensioning spectrum

Seismic withstand specifications arewidely presented in the form of a familyof response spectra for each storey.These are calculated by taking thebuildings transfer impedance intoaccount. An example is given infigure 10.

reading the response

spectrum applicable to apiece of equipmentThe benefit of the response spectrum isthat it visualises the extremeacceleration effects (or displacementeffects) provoked by excitation on asingle DOF system.

Acceleration / velocity /displacement conversion

Response spectra are oftenrepresented in an acceleration/

frequency system of coordinates butare sometimes represented in thevelocity/frequency system ofcoordinates. For low damping ofequipment studied (i10%), theresponse spectra measured in terms ofvelocity and the relative displacementcan both be deduced from accelerationspectra by applying the followingequations to each frequency:

Vf

max max=2

; Df

max( )

.max=

2 2

fig. 9: family of response spectra obtained

for different dampings during the same

earthquake.

fig. 10: dimensioning spectrum, according to floor levels (in metres) for an industrial site. This is

a spectrum for a damping of 2 %.

0.1

1

0.2

0.3

0.4

0.5

0.6

0.8

1

(g)

2 3 5 10 20 30 50

2

3

4

5

6

Fr (Hz)

0

+ 7

+ 27

Fr

1: 2: 3:

1>2>3


10/28


In fact, everything occurs as ifsinusoidal quantities were involved,with

V t dt et d t dt ( ) ( )= = With log/log coordinates the responsespectrum can be read alongacceleration, velocity or displacementaxes, (see fig. 11).

Maximum floor acceleration anddisplacement

Since the energy from the seismicexcitation is limited to a frequencyof 35 Hz, the points on the spectrumsituated above this frequencyrepresent the behaviour of a rigidoscillator (very high K/M), whichremains dimensionally stable under

seismic excitation.The relative displacement of the massin relation to the support is thereforezero and its acceleration is equivalentto the supports acceleration(see fig. 12 a).The high frequency asymptotic curveon the response spectrum(Fr u 35 Hz) corresponds therefore tothe maximum floor acceleration (seeright-hand part of figure 13).

Remark:

For the right hand part of the spectrum(which corresponds to theinfiniterelative frequency), expertsuse the abbreviation ZPA (Zero PeriodAcceleration) to establish theacceleration level.

In the same way, the lowestfrequencies on the spectrumrepresent the behaviour of anoscillator which remains infinitelysupple under seismic excitation.The relative displacement of this typeof oscillator equates to thedisplacement of the support(see fig. 12 b).At low frequency the asymptotic curveon the response spectrum, whenrepresented in log/log scales,corresponds to the grounddisplacement zone (see left-hand partof figure 13).

Maximum acceleration anddisplacement of oscillator

Between 1 and 35 Hz (central part offigure 13) the oscillator accelerationsand displacements are generallyhigher than the floor equivalents.

fig. 11: sample response spectrum which can be read along acceleration, velocity and

displacement axes.

fig. 12: response on a single DOF system for maximum values of its resonance frequency.

a)with a very highK/M ratio thesystemdoes not deform (the massadopts the ground displacement)

= sol

b)with a very lowK/M ratio thesystemdeforms (the mass remainsimmobile)

= 0

0.1

1000

500

100

50

20

10

5

2

1

0.5

0.2

0.1

0.05

0.02

0.01

200

100

50

20

10

5

2

1

0.5

velocity (cm/s)

0.2 0.5 1 2 5 10 20 50 100

frequency (Hz)

displacement(cm

)0.00

10.00

20.00

5 0.01

0.02

0.05

0.1

0.2

0.5

1

2

5

10

20

50

100

200

accele

ratio

n(g)

0.005

damping (%)

2

5

10

20

M

d

Fr = "infinite"

support

M

d

Fr = 0


11/28


Maximum acceleration and displa-cement, as well as the correspondingresonance frequencies are readdirectly off the spectrum (see fig. 13 -

readings along axes andd).Where the dynamic characteristics ofan equipment are not known, it isassumed that the equipmentcomprises a certain number of singleDOF oscillators.The majorant values of the maximumdisplacement and maximumacceleration it will be subjected toduring the earthquake, can beascertained from the responsespectrum.

Choice of damping for equipment

The damping factor taken into accountin the response spectrum analysis issupposed to represent the globaldamping for the equipmentconcerned.If the equipment comprisescomponents with different dampings,then it is normal practice to work withthe smallest, with regard to the choiceof response spectrum; this leads to anincrease in the stresses.The table in figure 14 shows,indicatively, the values commonlyagreed upon in terms of thepercentage of the equipmentsmaximum stress. Since seismicspecification is generally expressed interms of a family of responsespectrum which corresponds to thedifferent damping (2 %, 5 %, 10 %,etc.); the equipment designer can theneffectuate aninterpolation.

Benefits of the response spectrum

The information provided on anexamination of the response spectrumis much more useful to an equipmentdesigner than that provided by thetemporal representation of an earth-

quake. In fact, while the time-history provides the maximum flooracceleration, the response spectrumprovides a maximum amount ofinformation, noteably, the followingmaximum values:cmaximum floor acceleration,cmaximum floor displacement,cmaximum acceleration for a part ofthe equipment,cmaximum displacement for a part ofthe equipment.

fig. 13: response spectrum reading (in log/log scale it is possible to read the values whichcharacterize the acceleration and displacement).

type of structure damping in %

for 50 % of for 100 % of

the maximum yield the maximum yield

welded steel structures 2 4

bolted steel structures 4 7

reinforced concrete structures 4 7

cabinet 2 5

rack 2 5

fig. 14: commonly agreed upon dampings for diverse structures according to the yield of the

stress (deflection or traction/compression).

Fr35 Hz1 Hz

maxof the "system"

= maxof the ground

d = dmax

of the"system"

d = dmaxof the ground

d

strong part of spectrum


12/28


elastic structures (with Ndegrees of freedom)

Resonance frequencies and modalshapes

Generally speaking an elastic structure(for example a whip antenna, a beamgantry), is characterized by a multitudeof resonance frequencies (theoreticallyinfinite in number) which correspond tothe resonance modes or characteristicmodes, (these structures have N

degrees of freedom), (see fig. 15).Each of these resonances isaccompanied by a specific deformationof the structure termed the modalshape. For each resonance frequency,the structure deforms and oscillates onboth sides from its rest position (thepoints of the structure evolve in phaseand in antiphase and produce nodesand antinodes in the modal shape).The higher the mode, the morecomplex the corresponding modal

2. dynamic behaviour of structures

Today, the study of the dynamicbehaviour of structures consitutes anessential stage in the design of allindustrial equipment. At this stage, it isappropriate to present the mainconcepts which govern a structuresresponse to seismic excitation.Refer to the bibliography for furtherreading.

brief summary of single

degree of freedomoscillatorThe single DOF oscillator, constitutesone of the basic principles in dynamicanalysis of structures. In fact, thedynamic behaviour of an elasticstructure amounts to the behaviour of acertain number of basic oscillators.Furthermore, it is often the case thatthe inclusion of the first mode ofresonance for a given structure sufficesfor dimensioning; this equates tostudying a basic equivalent oscillator(two types are shown in figure 7).

The basic oscillator is characterized byits resonance frequency or naturalfrequency, and by its damping. Theresonance frequency corresponds tothe free movement of the oscillator,with no external force. In other words, itconcerns the frequency of theoscillators displacement when it isdistanced from its rest position (freeoscillation test) or on impulse. Whenthe oscillator is excited to thisfrequency, resonance is produced,that is to say, the movement isamplified. This amplification is inverselyproportional to the damping of theoscillator. Resonance frequency anddamping are sufficient for calculatingthis systems response under anyexcitation, and in particular, excitationby support displacement whichconstitutes the earthquake.

fig. 15: modal base: primary resonance modes of two basic structures.

first mode

second mode

third mode

modal shape

rest position

modal shape

modal shapenodes

antinodes


13/28


shape becomes, with an increasingnumber of nodes and antinodes.The dynamic behaviour of structures,comprising N basic structures with one

degree of freedom, (see fig. 16), isusually determined by using what isknown as the modal analysis of thestucture, which consists of tracing theresonance frequencies and modalshapes on the frequency band thatcorresponds to the earthquake. Thisestablishes a work basis comprising thestructures N primary modes, called themodal base, in which the initial problemwill be reduced to the study andcombination of N single DOF systems(see fig. 16 right).

Frequential and spatial adaptationAn elastic structures resonance isobtained on two conditions:cthat the excitation frequencycoincides with the structuresresonance frequency. Here it is amatter of frequential appropriation(only condition required for systemswith 1 DOF),cthat the excitation direction as well asits localization are coherent with thecorresponding modal shape. If pontual,the excitation must not act on one ofthe structures nodes, and is all the

more efficient when it acts on anantinode in a direction parallel to theantinode displacement.

Where multiple excitations occur, theyalso have to respect the phase relation-ships of the modal shape (see fig. 17).

fig. 16: complex structure with N degrees of freedom.

fig. 17: examples of spatial appropriation for punctual excitations.

excitations

in antiphase

excitations in phaseexcitation

or or

"appropriated" excitations "non-appropriated" excitations

excitation

excitation

excitation

excitation

excitation

structure with N degrees of freedom in the modal base

Mn

M1 M2

mn

m1m2


14/28


15/28


fig. 19: to obtain the deformation for all the points of a structure, when the appropriation ratios and the corresponding max are known, ageometric sum suffices.

t1= 100 % t2= 60 % t3= 30 %

D1

Fr1

D2

Fr2

D3

Fr3

3 D32 D21 D1

t11 D1

primary modes

(frequencies Fri, modal shapes Di)

appropriation ratio

(ti)

resulting deformation of the structure: D = t1 1 D1 + t2 2 D2 + t33 D3

FrFr3Fr2Fr1

12

3

ground horizontal acceleration

response spectrum

t22 D2 t33 D3 resulting modal shape D


16/28


3. equipment design

(level of acceleration at which theequipment no longer performs itsfunction) has to be determinedbeforehand.

Functional withstand capability

The vibratory stresses to which thefunctional devices will be subjectedmust be evaluated, and there must beassurance that they would functioncorrectly when put under thesestresses, or that their availability isunaffected.

There are two possibilities:cthe functional device is a protectiveor monitoring device produced inseries: the equipment generallyundergoes a vibratory environmentqualification, the results of which canbe exploited in order for the seismicwithstand capabilities to be evaluated.Otherwise, the equipments behaviouron the seismic excitation range(0-40 Hz) must be studied,cthe functional device is a specialdevice, in which case an evaluation bytest is necessary.

In certain cases, an analysis of thetests carried out on an analogousequipment, can provide the technicalelements which will reveal thefunctional withstand capabilities of anapparatus.

design principlesIt is essential that the transfer notionfigures in the design. In fact, electricalequipment generally comprise a frame

or skeleton (structure) which, in theeventuality of an earthquake,transmits more or less faithfully thefloor vibrations towards the functionaldevices.

Response spectrum and dynamiccharacteristics of the structure

The response spectrum, which repre-sents the floor movements, allows:

con the one hand, the excitationseverity the structure is likely toundergo to be determined (in terms ofacceleration and displacement(see end of first chapter)),

defining objectivesIf the design is to take the seismicstresses fully into account, then thedegree of safety it has to assureduring and after the earthquake mustbe known.

The demands made on the behaviourof equipment exposed to earthquakesare:cstability (equipment must notbecome projectile),cintegrity (equipment must conserveits initial geometry),cfunctioning (equipment must assureeither complete, partial or degradedfunctioning).

While the designer is only concernedwith the equipments mechanicalwithstand capability as far as the firsttwo demands are concerned, theprocedure is more complex for thethird demand, since the differentfunctional aspects have to be takeninto account. This is usually the casefor electrotechnical equipment for

which operating conditions, in theeventuality of an earthquake, areusually similar to nominal operatingconditions. The characteristics of thestructure which transfer the excitationsmust therefore be defined according tothe functional cells fragility threshold.

See fig. 20for vocabulary.

Mechanical withstand capability

In order to check an equipmentsmechanical withstand capability, itmust be ensured that there is gooddimensioning of the ground supportsand that the structures safe stresses

are not exceeded. Thesedimensioning criteria are entirelydependant on the position of the firstresonance frequencies envisaged forthe seismic sollicitations.

As far as electrotechnical equipment isconcerned, the nature of the transferof ground seismic sollicitation tofunctional cells, must be taken intoaccount at the design stage. Thefragility threshold of functional devices

fig. 20: definition of the terms used for an equipment (LV cabinet, HV circuit-breaker).

functional device

functional cell

frame

groundfloor


17/28


cand on the other hand, to determinewhether or not the structure will amplifythe seism, in view of the position of theresonance frequencies with respect to

the strong part of the responsespectrum.

It is therefore imperative that thedesigner knows the structures primaryresonance frequencies; these can beestimated by analysis, tests or byanalogy, remembering that a spectrumconsists of two zones (see fig. 21):cthe right hand side of the spectrum,during which the equipment has thesame accelerations as the groundwithout amplification.When the equipments resonance

frequencies are in this zone, theequipments mechanical behaviour istermed static equivalent or pseudostatic. Evaluation of maximumstresses is obtained by successivelyapplying the maximum groundacceleration (ZPA or Zero PeriodAccele-ration) to the masses concernedaccording to the three spatialdirections.cthe strong part of the spectrum,during which the structure amplifies theaccelerations by its resonances, whichleads to increased forces and stresses

than was previously the case. In thiszone, the equipments mechanicalbehaviour is dynamic and, as a result, itis necessary to know the frequenciesand modal shapes and to combinethem in order to evaluate the maximumpossible damage that the equipmentmight incur.

The process therefore consists in:vcharacterizing the natural vibrationalmodes (Fr, Di),vdetermining the modal responses,vsuperimposing the modal responses,

vdeducing the forces and stresses

induced.State of the art seismic design

Applying the above procedure willobviously avoid the equipment havingresonance frequencies in the strongpart of the response spectrum. Wherepossible, the designer shouldendeavour to limit the equipmentsdimensions and in particular its height(small dimensions favour increasedfrequencies).

This being the case, the most widelyused solution is the stiffeningof thestructure to discard the first resonancefrequencies higher than the seismic

excitation range, or at the very leasthigher than the strong part of theresponse spectrum. In any case, it issensible to avoid resonance modessituated in the 0-10 Hz, orindeed 0-15 Hz frequency band, inorder to increase safety.

Putting these basic concepts intopractice does however have to fall inline with restrictions imposed by theequipments cost, dimensions andfunctioning.

When an equipment is obviously too

fragile and can not be sufficientlystiffened, it is possible toisolate theequipment from the ground byinterposing a suspension stage.To beeffective, the suspension must howeverpossess the characteristics required byvery high-performance damping plugs(suppleness and deflection). So as toachieve accelerations for theequipment of a lower amplitude thanthe ground accelerations, thesuspension should confer very lowresonance (suspension) frequencies,(in the order of 1 Hz) to the suspended

equipment, and should acceptdisplacements greater than 40 cm.

Such characteristics cannot of coursebe obtained with classical plugs and theresulting displacements are not freefrom inconveniencies for supported

equipment (respecting positioning,external connections, dielectricdistances). This method is not thereforewidely used.

On switchgear in cabinets

When for example a cabinet housesdiverse equipment, it is necessary todo carry out both a mechanicaltransfer study and an evaluation of therobustness of the equipment inquestion, with a view to establishingthe compatibility between the aptitudesof the cabinet and the limits of the

switchgears functional withstandcapability to the vibratory environment.To limit the amplifications of thecabinets movements, and as a resultthe transfer, it is necessary to haverigid cabinet frames (reinforced orbraced). The required degree ofrigidity is in accordance with therobustness of the switchgear.

The usual recommendations include:

ccabinet construction: assemblieswhich are bolted or welded together arepreferable to those which are rivetedtogether which can work loose and

generate impacts which are harmful tothe switchgear,

fig. 21: resonance frequencies in the strong part of the response spectrum are to be avoided.

FrFAPN

amplification

strong part

(dynamic modal

behaviour)

zero period

acceleration part

(static behaviour)

acceleration of the structure

= ground acceleration

ZPA level


18/28


cmounting of the cabinet: the idealsolution consists in bolting the cabinetto the ground and the wall, with fixingswhich must be dimensioned so as to

resist the loads resulting from seismicacceleration,cdisposition of the cabinet: if thecabinet is only mounted to the ground,it is better for the heavy masses to bearranged at the bottom of the cabinet;the same applies to fragile devices,cmounting of devices: rigid fixings arepreferable; otherwise, it is wise to bewary of local resonance modes, and tothe different movements during theseism,ccircuit boards: avoid boards whichare too large or too full, heavy compo-

nents; provide stiffeners if necessary,ccabling: so as to avoid inertia loads,flange the cable layer as close aspossible to the connectors.

simulation by analysis atdesign stageNumerical analysis for a structuresdynamic behaviour generally uses thefinite elements method. This numericaltechnique allows the mechanicalbehaviour of a structure, subjected todynamic sollicitations, similar to those

generated by an earthquake, to bepredicted. This technique is particularlywell suited to the design stage, whenthe structure only exists in the form ofdefinition or utilisation plan, and can stillbe modified. It provides essential datafor the designers with regard tostresses, anchorage loads and thedeformations produced by the seismicexcitation.

Principle

The principle behind this method is toconstruct a simplified model of theequipment, by using a certain numberof finite elements ( beams, plates,volumes) which represent the structure,as well as concentrated massesrepresenting the functional devices.Meshing uses the structuresgeometric data (profile, thickness,profile intertia) and the equipmentsphysical characteristics (Youngsmodulus, density) (see fig. 22 for anexample).

The detail of the meshing is notessential in order to access fundamentalmodes, but care must be taken torespect the distribution of the principle

elementsof stiffness and mass.However, when the stresses are beingcalculated the meshing must besufficient.

The computing program thendetermines, on a seismic excitationrange (0-40 Hz), the resonance

frequencies and modal shapesassociated to the model, as well as themodal participation factors: this ismodal analysis.

The table in figure 23 gives theelements of modal analysis for a trans-former; these reveal the LV insulator tobe a sensitive element, but that theconservator is even more so, becauseits second resonance mode has afrequency of 11 Hz (which is likely to

fig. 22: meshing of finite elements of a EHV transformer (2,000 elements and 1,500 nodes).

X

Y

Z


19/28


appear in the strong part of the seismicexcitation response spectrum).

The next stage consists of simulatingthe equipments response underseismic excitation and thus obtainingthe displacements, stresses and loadson the supports.

Two methods can be used accordingto the position of the structuresresonance frequencies with regard tothe strong part of the responsespectrum:

cpseudostatic analysis(The resonance frequencies appearabove the strong part of the responsespectrum).If this is the case, then the analysis is

static, and the model is subjected tomaximum ground acceleration forevery direction.

csuperposition method analysisanalysis(Certain resonance frequencies are inthe strong part of the responsespectrum).In this case the models response fora given direction is obtained bymultiplying the modal shape for eachmode (Di) by its modal participationfactor and by the acceleration read atthis modes frequency on the

response spectrum. The responses ofthe different modes are thencombined in order to obtain a globalresponse for a given direction.

It must be recalled that the modelsresponse spectrum does, by definition,include damping: it is thereforenecessary to use a response spectrumwhich corresponds to the damping ofthe structure studied, even a lessfavourable one, that is to say theweaker one.

The last stage involves ensuring thatthe maximum values ofdisplacements, stresses and loads onconnections are acceptable, that is tosay compatible with the structuresmechanical characteristics. There arecombination rules for seismic andstatic stresses (own weight, snow,wind, pressure, traction loads).Moreover, the spatial distribution ofaccelerations will make it possible tocheck the functional withstand

capability of the switchgear and theconnected instruments, either by usingspecific tests or by makingcomparisons with tests which have

already been carried out.

Limitations of the finite elementsmethod

When test models with finite elementsare used, divergence is obviously goingto occur between the calculated andactual resonance modes of the installedequipment. This is mainly due to theapproximations made as far as the limitconditions (how structure is mounted tothe ground), internal connections, nonlinearities, as well as the differentsimplifications inherent in test modelsare concerned. Generally speaking onlythe primary vibrational mode calculationwhich is acceptable.

However, given the broad band natureof seismic excitation, the evaluation ofstresses calculated by responsespectrum tolerates an error as to the

resonance frequencies exact position.This explains why it still makes senseto use this method at the design stage.

Modal tuning

As soon as the equipment prototype isready, it is possible to correct the finiteelements model, by carrying out atuning according to the measurementdata. Different experimental techniquesenable the strucures actual dynamiccharacteristics to be obtained, by usingthe tuning software, and to make thenecessary modifications to the finiteelements model, so that it mightprovide a closer representation ofreality.

fig. 23 : rsultat du calcul modal pour les divers lments du transformateur de la fig. 22.

mode No frequency modal element concerned

(Hz) participation

factor

1 8.7 4 conservator

2 11 232 conservator

3 12.7 14 radiator

4 13.2 34 all auxiliary parts

5 13.8 5 heat exchanger

6 15.9 24 conservator7 17.2 11 heat exchanger

8 19 105 all auxiliary parts

9 19.3 51 all auxiliary parts


11 22.9 18 conservator


13 24.1 4 surge arrester




17 24.9 0,2 radiator


19 26 96 radiator20 26.5 6 heat exchanger

21 26.6 25 radiator

22 29.3 115 LV insulator





20/28


4. qualification by simulation or test

csecond stage: experimental modalanalysis (see fig. 25 and 26).

This analysis is carried out on theprototype. It consists in acquiring thetransfer functions between a point ofexcitation (generated force) and theresponse points (measuredaccelerations), then identifying theactual modes of the structure(resonance frequencies and associatedmodal shapes),

cthird stage: tuning of finite elementmodel.Here the parameters of the finiteelement model are readjusted (finenessof meshing, physical parameters:YOUNGs modulus, density, limitconditions) so that the models dynamiccharacteristics can be made to reflectreality as closely as possible,

cfourth stage: measuring the dampingcoefficient.

introductionTo qualify means proving theequipments withstand capability underidentified or normalized stresses.There are two ways of realizing seismicqualification:cthe first involves effectuating realsize tests on the equipment;cthe second uses finite element testmodels which can be combined with acertain amount of experimental data.The latter is becoming more and moreimportant in the qualification process,particularly as far as mechanicalwithstand capability is concerned. Buttoday it is still tricky to take account ofthe functional aspect by using a testmodel.

Qualification by test is used:

cfor equipment with dimensions whichlend themselves to the vibration testingmachine,cfor specific equipment (unitary, smallseries),cif the functional aspect is determining

(complex or high level of safety).Qualification by numerical analysisis used if:

cthe dimensions of the equipment areincompatible with the testing machines(as is the case for large transformers),ca device has already been testedunder other seismic conditions,cthe device is a modified version of aqualified device,cthe functioning of the equipment isnot requisite during the earthquake.

Combined qualification by numericalanalysis and experimental modal

tuning is used:cfor large series equipment,cwhen standards or operators permitthis kind of justification (knowledge offunctional data).

In fact, numerical analysis oftenprecede real size tests. Thismaximizes chances of correctly

effectuating the qualification testssuccessfully first time round.

We will now:cillustrate by means of two examples:combined qualification and qualificationby real size tests preceded by adesign test model,cdevelop the methodology ofqualification by test.

combined qualification

(numerical analysis andexperimental tuning)The method which combines bothanalysis and tests involves:ccreating a mathematical model,cgathering in the data from partialtests (modal experimental analysis)carried out on the prototype,concerning the devices dynamicbehaviour (damping, resonancefrequencies, modal shapes),ctuning the mathematical model withthe preceding data.

The analysis model then allows themechanical withstand capability to beevaluated under accumulated seismicsollicitations and service stresses. Thefunctional withstand capability involveschecking that the equipment is notbadly affected by the deformations andaccelerations delivered by the analysis.

The following example shows themethod which combines both tests andcalculation used to define seismicresistance of HV circuit-breaker.

Seismic qualification of a HV circuit-breaker

(see fig. 24)cfirst stage: numerical analysis ofcircuit-breaker.The model is made out of finiteelements: (beam-gantries, plates andshells for the insulators), the modelcomprises 2,670 elementsand 3,200 nodes,

fig. 24: Merlin Gerin circuit-breaker designed

for use in HV switchgear equipment.


21/28


In order to calculate the circuit-breakers response using theresponse spectrum method, it isnecessary to know which damping to

apply to the model. This is obtained bysubjecting the prototype to asubstantial mechanical deformation(free oscillation test) ; damping is thendeduced by observing the decline inoscillations,

cfifth stage: analysis of loads,stresses, and displacements underseismic charge.The analysis of the response iscarried out by using the responsespectrum method, it makes it possibleto take different earthquakes intoaccount,

csixth stage: verification of thedevices integrity and functioningunder seismic loads.This verification consists:von the one hand, in verifying themechanical withstand capability of thestructure in terms of loads onconnections and stresses in the

materials when the circuit-breaker issubjected to accumulated seismicstresses and service stresses (weight,internal pressure, static loads on the

terminals, wind) (see fig. 27 page 22,stresses in the HV circuit-breaker),vand on the other hand, in assuringthat the deformations which resultfrom the seismic sollicitation do notinterfere with the functioning of thedevice; this final verification is carriedout statically by imposing thedeformation obtained from theanalysis on the equipment, and byexecuting the different operations forwhich it is intended.

qualification by real sizetests preceded bynumerical analysisEven if the material is to undergoqualification tests, it is still of interest,in order to save time and money, toprecede building and testing by a testmodel effectuated from plans.

Qualification of control/monitoringcabinets

Control/monitoring cabinets intendedfor nuclear power stations are

considered in the following example.This equipment is subject to strictfunctioning safety regulations, and, tothis end, undergo real size tests forresistance to seismic sollicitations (seefig. 28, p. 23).

In order to be able to present anequipment with higher guarantees ofgood resistance for testing, a certainnumber of simulations and investi-gations are carried out at the designstage. The procedure is as follows:

crobustness evaluation of maindevices installed in the cabinet.

For equipment which does not haveany historical data, a maximumwithstand capability test is carried outon the seismic excitation frequencyrange. This involves establishing theequipments fragility threshold (ifnecessary live). This data is then usefulfor defining the desired limitation of thecabinet transfer,

fig. 25: modal experimental analysis.

fig. 26: resulting experimental modal shape

(f = 3.8 Hz).

X

Y

Z

X

Y

Z


22/28


cevaluation of transfer for the cabinetsstructure.The response spectrum form, as wellas the fragility threshold of the

embarked equipments, indicate thedesirable characteristics of the cabinettransfer. Making a numerical analysisfor the cabinet aims to identify the mainresonance modes in the seismicexcitation frequency range (0-40 Hz).The cabinet is modelled in beam-gantryand plate elements and the equipmentis represented by punctual masses andinertia.Finite element analysis is carried out inorder to estimate the levels of vibrationthe cabinet will communicate to theequipment situated in the interior.

According to the position of the primaryresonance frequencies, modificationsare made (on a plan) to reduce theamplification of ground acceleration.

cexperimental verification of thecabinets characteristics (figure 29). Anexperimental measurement of theprimary resonance frequencies iscarried out on the equipped, wiredcabinet in order to ensure that thestructures actual characteristics do notdiffer from those provided by theanalysis,

cimpact of cabinets actualcharacteristics on the equipment.The impact, on the equipmentsresistance during seismic tests, of thediscrepancy between the cabinetscalculated dynamic characteristics andthose actually measured, is evaluatedby referring to the seismic excitationresponse spectrum.Modifications are made if thediscrepancies engender vibrationalamplifications which are incompatiblewith the equipments characteristics.For example, the mounting devices forfixing the equipment to the ground and/

or fixing other devices will bereinforced.

qualification by testQualification by test is not alwaysplausible (equipment is too heavy orthe dimensions are too big) and it isoften tricky to set up. Powerful testequipment is required (vibrating tableswith large displacements powered byhydraulic jacks, sophisticated controlsystems), and only some specialisedlaboratories are capable of carrying out

these tests. In addition to the laboratoryperformance, the following expenditurehas to be taken into account:ccost of transport,

ccost of mounting the test equipment,ccost of replacing material if it turnsout to be non operational after the test.

The procedure for carrying outqualification tests on an equipmentusually features in the specifications

sheet (or in a test program), andconforms with the current standards orrecommendations (IEC 68-3-3/UTEC 20420, ANSI, ENDESA, IEEE, etc.).

Several variations of the qualificationprocedure are possible, and they runaccording to:cthe information on the geoseismiccontext of the equipments location,cthe equipments complexity,

fig. 27: analysis of stresses.

15

25

40

45

50

55


23/28


cthe data on the dynamic behaviour,cthe representativeness of theequipment in tests with regard to the

series,cthe degree of safety the equipmenthas to ensure during an earthquake.Without going into too much detail, wenow propose an insight into the criteriawhich influence the choice of modalitiesof a qualification by test, according tothe IEC standard 68-3-3.

Configuration of material beingtested

Before proceding with seismicqualification test a certain number ofmodalities have to be fixed. These arenotified in a particular specificationwhich mentions, amongst other things,the arrangements to be taken intoaccount concerning:

cchoice of test specimen. Preliminaryanalyses are sometimes necessary inorder to ensure that the chosenspecimen does actually represent theworst case scenario,

cfixings and mounting. They have tobe identical to those used on site,

cservice conditions which have to betaken into account (mechanical orelectrical),

fig. 28: LV cabinet for a nuclear power station during qualification tests.

fig. 29: modal deformation of the cabinet's frame derived from the experimentation.

X

Z

Y


24/28


25/28


5. conclusion

The high degree of continuity ofservice required for electricaldistribution and control/monitoringdemands that all types ofenvironmental restrictions(mechanical, climatic,electromagnetic) are taken intoaccount. Seismic stresses,which areparticularly destructive are includedhere and they must be specified at thedesign stage of the equipment.

In order to do this it is necessary to

know the severity of the maximumhistorical earthquake likelihood in theform of a dimensioning spectrum, or,for mass-produced equipment, to

choose the level of severity of thenormalised seismic class.Today, an equipments mechanicalwithstand capability can beunderstood with a high degree ofprecision thanks to numerical analysisand finite element calcuation of thestresses

Proving that the equipment remainedoperational during or after theearthquake, is more difficult andgenerally requires the numerical

analysis to be combined with tests onthe operational elements. Theexamples of qualification by numericalanalysis and/or tests presented in

chapters 3 and 4, reveal the know-how of a company which, for manyyears has been providing countriesexposed to earthquakes withequipment for nuclear power stationsand other electrical equipment.

As for quality or electromagneticcompatibility, seismic withstand mustbe mastered at the design stage; ifthis is neglected then it is often difficultand more expensive to correctproblems at a later stage. As a result,

numerical analysis and powerfulcalculation methods are used widely inthe anti-seismic design of electricaland electronic equipment.


26/28


6. bibliography

Standards

cIEC 68-3-3 (UTE C20 420):Guidance. Seismic test methods forequipment.

cIEC 1166: HV circuit-breakers: guideto seismic qualification.

Merlin Gerlin Cahiers Techniques

cCT 85 (1977): Seismic withstandcapabilities of electrical equipment byP. PY, J.-Y. BERTHONNIER(presents technological solutions for HVcircuit-breakers).

Miscellaneous publications

cAFPS90 recommendations fordrafting rules relative to works andinstallations to be realised in regionssubject to seismic volumes 1 and 2.

cDynamic calculations of structures ina seismic zone by Alain CAPRA andVictor DAVIDOVICI.

cRevue des laboratoires dessais,ASTE publication, n9, 31, 36, 39.


27/28



28/28

Earthquake and Electrical Equipment

Documents