Simplified approach for the seismic analysis of Non-structural elements

© All Rights Reserved

0 Aufrufe

Simplified approach for the seismic analysis of Non-structural elements

© All Rights Reserved

- Bridge Structure Response Spectrum
- StaadPro Knowledge Base
- Silo Buckling
- Mathematical Methods for Physics
- REAL_TIME_FACE_RECOGNITION_SYSTEM_USING_2_article.pdf
- Complexe i Gs
- Dirac Matrices
- Lecture2 Matrix Algebra
- Termpaper-Applications of Eigen Values-math
- eigen
- Akira Takahashi and Shaul Mukamel- Anharmonic oscillator modeling of nonlinear susceptibilities and its application to conjugated polymers
- Modes in Linear Circuits
- Problem Set Final
- noise ppt
- Understanding Negative Eigenvalue Messages
- Prony method for exponential
- Thin walled Closed Section Beams
- 9
- Test 3 Solutions
- Order and Chaos

Sie sind auf Seite 1von 8

SPECTRUM METHOD

INTERA TOM, Internationale Atomreaktorbaugesellschaft mbH, D-5060 Bergisch Gladbach 1, Fed. Rep. Germany

In this paper a method to determine floor response sepctra is proposed which is based on the modal analysis of a support

structure with interaction-free, one-degree-of-freedomsystem attached. The time-consumingmethods using real or artificial

soil accelerations are avoided as well as some of the arbitrarinesses in the approaches of Biggs or Kapur-Shao.

usually are available only rather unexactly.

Whatever method is used to calculate the dynamic This paper presents a calculation method for floor

response of structures, practical considerations 6ften response spectra which solves the described problems

require a subdivision into support structure and sup- in the following manner.

ported structure or equipment. If the equipment af- (a) The maximum accelerations of the one-degree-

fects the dynamic behaviour of the support only to a of-freedom systems derived from calculations accord-

small extent, the response of the support structure is ing to the response spectrum method and the floor

the input for calculating the load on the equipment. response data calculated according to the method des-

Using the response spectrum method means develop- cribed below are identical outside the resonance range.

ing floor response spectra from defined soil response (b) In the case of resonance, at which any rule for

spectra, i.e. determining the acceleration of interaction- the superimposition of modular fractions is question-

free one-degree-of-freedom systems which are arranged able, the evaluation of the floor response spectra is

on the investigated 'floor' of the support structure. derived from the dynamic behaviour of certain two-

In the past, Biggs and Roesset [1], Biggs [2] and degree-of-freedom systems. The description of the dy-

Kapur and Shao [3] proposed methods to calculate namic properties of these systems is given in simple

floor response spectra which avoid time history meth- formulae and can be considered as empirically well

ods with spectrum compatible soil accelerations. The proven over the interesting range of structure and

fact that these methods are not used very often may equipment damping.

be due to the following reasons. (1) The necessary There is one and only one condition for the pre-

magnification factors for two-degree-of-freedom sys- sented method to be valid. We must assume the res-

tems are not given by means of simple equations, but ponse spectrum analysis to be a suitable tool to pre-

have to be determined (perhaps statistically) for each dict dynamic loads. If we do not so why should we

special case. (2) There is some arbitrariness in fixing evaluate floor response spectra?

the frequency boundaries where certain formulae are

considered valid. Furthermore, it is often overlooked

2. The dynamics of interaction free one-degree-of-free-

that time history methods may not yield reliable re-

dom systems attached to a support structure

suits. We may consider the choice of a special time

history as the choice of a special rule to superimpose 2.1. General assumptions and notations

modular effects. This rule is very sensitive to and in a

most unclear way dependent on structure properties Let ~n be a linear elastic support structure with n

256 K.A, Peters et al. / D e t e r m i n a t i o n o f floor response spectra

degrees of freedom. Let In this section, only the solution for (*) is given. For

the proof, see section 4.

K = [k/] ] be the stiffness matrix,

M = [mq] be the mass matrix,

2.2.1. Eigenva&es

=[t¢ii] be an eigenvector matrix of

There are scalar functions X~(m) ..... X*(m) contin-

the eigenvalue problem

uous in 0 satisfying the following statements:

K¢ = kM~p with ~TM~b = I.

Thereby I is identity. X*(m) is a multitude 1 eigenvalue o f ( * ) , (la)

lim X*(m) = co~. (lb)

At the nth degree of freedom of Y'n a one-degree- m--+O

of-freedom system is attached which is characterized

2.2.2. Eigenvectors

by mass m, circular frequency Wo and degree of damp-

There are vectorial functions ff~(m) ..... ¢*(rn) and

ing Do. The thus defined structure will be referred to

ro(m ) continuous in 0 satisfying

as Zn+l(m, COo,Do).

We assume that the circular frequencies col, ..., Wn ~*(m) is eigenvector of (*) with respect to X*(m), (2a)

of 2;n (i.e. the eigenvalues co~ ..... con2 of K¢ = kMtp),

are of multitude 1. Furthermore, w o 4: wi for 1 ~< i

~< n. These conditions do not constitute essential lim-

lim ~*(m) = ~°ni

w~ , forl<i~<n, (2b)

itations with respect to the problems with which we

are concerned. Moreover, we assume that Zn is mod-

ularly damped with dampings D1, ..., Dn, respectively.

- m i:1 CO2o-W~ ~Oni~Oli

Throughout this section, coo and D o are assumed ~(m) = + ro(m ) ,

to be fixed. Stiffness matrix K*(m) and mass matrix n

(2c)

M*(m) of Zn+l(m, ~ o , D o ) are then as follows:

- m i=IG w ~ - w ~ ~°ni~°ni

kin 0 1

kq where ro(m ) of at least second order tends to 0 for

m-*O.

i = l ..... n - 1

]=1 ..... n - 1 2.2.3. Participation factors

K*(m) = Let F i be the participation factor of E n with res-

kn_l, n 0

pect to ~i, 1 ~< i ~< n. Let ~o~(m), ..., ~0*(m) be eigen-

knl ... kn, n-t II knn +rnw2 -mw~ vectors of (*) normalized to 1 according to M*(m).

Let P * ( m ) b e the corresponding participation factors.

0 ... 0 ', -mw 2 meg_ Then the following equations are valid:

lira F*(m) = Fi , for 1 ~< i ~< n , (3a)

I ol

t 0 m-*O

M I "

f lira P~(rn) ~on+l,o(m)

*

M*(m) = m -* O

-5-

n

...0 I

i=1 ¢-O~-ZL 2 l~i~Oni " (3b)

Using these notations the following class of eigenvalue

problems is solved: 2.3. Modular dampings

m small due to non-interaction. (b) damped, Zn+l(m, Wo, Do) in general will not be

K.A. Peters et aL / Determination or floor response spectra 257

modular dampings Di(m) are determined according to n 2

methods described e.g. by Sachs [4], it is easily de-

# = [ ~--~ rich' ( 1 - 000200°-

~;2 b) l)(/~3° +

rived from eqs. (1) and (2) that

lim Di(m) = Di , for 0 < i ~< n.

m"*0 + " 002 ri~ni(~i + b ) .

i=1

2. 4. The govern&g equation for the interaction-free,

Thus eqs. (6) mean that

one-degree-of-freedom system

ri~m (1 002-00H

00~

Now we are able to describe the governing equation

for the interaction-free, one-degree-of-freedom system is neglected for i 4=]. One can see that eq. (6b) is the

[i.e. for the (n + 1)th degree of freedom of Y'n+l(m, governing equation for an interaction-free, one-degree-

000, Do)] in a form suitable to be applied to the res- of freedom system connected to a one-degree-of-free-

ponse spectrum method. If ~ is the acceleration of the dom support structure.

one-degree-of-freedom system induced by an accelera-

tion b(t) acting on the fixed points of the support

structure, in accordance with sections 2.2 and 2.3, the 3. Determination of floor response spectra according

following equations result: to the response spectrum method

"

3.1. General assumptions and notations

"~ = ~i=1 03 2 -- 002 r ; ~Oni(~i + b) To evaluate floor response spectra F(00, D) we as-

sume that the accelerations acting upon the fixed

( . 002 ) (5a)

points of the support structure are given by a response

spectrum S(00, D). In addition, the response spectrum

+ I - ~i:, co2- 00] Pi~Pni (~o + b),

method is characterized by a superimposition rule for

~i + 2Di6°i~i + 00217i = -b, for 0 < i < n . (5b) the modal parts of the motion such as the root-sum-

square (rss) rule. It was already stated in the introduc-

In the case of resonance, i.e. in the case that there is tion that such rules are only meaningful for clearly

1 ~<j ~< n with kO2o- ~o~1 small, eq. (5a) becomes ap- separated natural frequencies. By determining floor

proximately: response spectra we therefore distinguish between the

non-resonance and the resonance case. The latter is

defined in section 2.4. In the following we use the no-

= r/~n/e(00o, 00/) tations of section 2.

+~i=, 002-~o2 l-','~oni(~i+b), (6a)

i4:/ In this case - according to the response spectrum

2 method - maxt~(t)l can be estimated by means of the

rss-rule. Observing eqs. (5) the following equality re-

suits:

002

+~

000 --(~0~

(~/+b). (6b)

F(00 o, Do) = {[(1 i)

~i=1 002 _ 002 ri~°n S(000, Do

= I is valid because of #~TM= q~-l. Then it follows that + ,=,L00o-00, r,,n,S(00,,D,)] I

2 2 (7)

258 K.A. Peters et al. / Determination o f floor response spectra

~

I D1 I 02 I

2

+

is~

c o 2 7 ~ PieniS(coi'Di

1/2

(11

L. . . . ~e [ . x1 t ~ X2

Fig. 1. Resonance system R(coO, D l, D2), where k 1/m 1 (7) yields too high values for coo in the vicinity of co/.

= k2/m 2 = tog, D 1, D2 are degrees o f damping, and m 2 < < rn I- Therefore, we define the resonance range with respect

co/to be the neighbourhood of co/where

F(co/, Do) is smaller than the value according to eq. (7).

3.3. F(Wo, Do)for coo in the resonance range In the resonance rage we set

The vibrating system shown in fig. 1 will be de- If the dynamically relevant data of the support struc-

noted as resonance system R(coo, DI, Dz). Let ture are known, the-floor response values for all pairs

bi(e, We, D1, D2) be the acceleration of the mass m i (coo, Do) can be determined by means of the rather

due to e. The following inequality is the result: simple eqs. (7) and (12), respectively.

' bl(e, coo,D1 ,D2) 4. The Zn+ l-problem

Xlbl(e, Coo,Dx,D2) (8) Let K,M, 0, coo be choosen as in section 2.1. The

category of eigenvalue problems

<~ V(e, coo,D1 ,D2)S(coo,D1)

K*(m) if* : X*M*(m)ff*, (13a)

with the 'magnification factor'

m>0, (13b)

b2(e, coo, DI ,D2) (9)

V(e, Coo,D1, D2 ) = i-~, (e--~,co-~o~ , ~ 2 ) will be referred to as the £n+ 1-problem with respect

i

to K, M, coo. K*(m) and M*(m) are defined in section

In the special case of e being a soil motion induced by 2.2. We will solve the Zn+ 1-problem for small m. The

an earthquake, the authors consider the following esti- following statements are readily shown. For ~ v~ 6o1

mation to be meaningful: and ~ eigenvalue of the Zn+ l-problem with eigenvec-

1 tor ~the following equations are satisfied:

V(e, coo,D1 ,D2) <,

2x/D2(D1 +D2)

0 K " ~M(m,~,) , (14a)

= V(D1,D2). (10)

The reasons for eq. (10) are given in section 5. In this cog

section the factor V(D1, D2)will also be compared (14b)

with the magnification factors proposed by Biggs and :%+, cog_ 7.,

Newmark et al. [5].

mi j m In

3.3.2. The resonance range i,j=l ..... n-1 !

If coo ~ coj for an appropriate j, analogously to the = mn-l,.

response spectrum method and according to eqs. (6)

and (1 O) we may estimate I L m cog

toni ... mn, n-1 I mnn L1 + 60I_~_

I mnn

max I~(t) l ~< F(co/, Do)

t

(14c)

= [[ r::n/ S(co/, On the other hand, if the pair ~,, ~'satisfies eq. (14a),

( L2x/Do~/+ Do) D/)I2

K.A. Peters et al. / Determination of floor response spectra 259

(14b) and (14c) for a certain m, then it is a solution lyrically on m, because ~b~(m) and X*(m) are analyt-

of eqs. (13). ical functions• Therefore all functions involved are

In section 2.1 we assumed the eigenvalues 602 of analytical and we conclude from eqs. (1 5) and from

K¢ = kM~pto be of multitude 1. Therefore it follows the equation (qffM) -1 = ~b:

by means of the implicit function theorem from eq.

(14) that there are analytical functions XT(m) ..... k~o(m)

X*(m) and ~O~(m)..... ff*(m) defined on a neighbour- 60g_

hood of 0 with • -~ - - m ~ ) + r(m) (16)

60g

X*(m), ff*(m) is a solution ofeq. (13) k*o(m) 60g_

form > 0 , (15a)

where r(m) of at least second order tends to 0 for

X*(m) is of multitude 1, (15b) m~O.

Solving eq. (13a) for ~ ( m ) , eq. (16) yields the fol-

lim X*(m) = 60], (15c)

m---~ 0 lowing result for X~(m):

lim qJ~(m) = ~i], (15d) lim Xg(m) -- 602. (17)

m---~ O m----~O

60~

0*+ 1,i(m) - 602 __X*(m) ~*i(m) . (15e)

of X*(m) normalized to 1 with respect to M*(m). Let

['*(m) be the participation factor• From the choice of

Eqs. (15a)-(15e) are valid for 1 ~< i, j ~< n.

4~and eq. (15) follows

Owing to definiteness and symmetry the Zn+ 1

problem has another solution ~b~(m), X~(m). O~(m) lim ¢~.(m) = ~il, for 1 ~< i, ] ~< n

is characterized by being orthogonal with respect to m.--~ O

M*(m) and O*(m) for all 1 ~< i ~< n. It is easily shown and therefore

that the vector

lim W(m)= re, for 1 ~< i ~< n. (18)

[~b~o(m)

qJ~(m) = [~*o(m) The following equation is an easily proved consequence

of the orthogonality condition and normalization of

¢*(m):

fulfils this condition of orthogonality, if the follow- n

ing equation is satisfied: ~FT(m)~p~l(m)=l, forl<.i<.n+l.

[

j=o

~b~.(m)

TM[qj~°(m)" Now eqs. (1 5) and (1 8) give the following result:

lim r~(m)~*+l,o(m)

m--'~ O

i,]=l ..... n lqJ*o(m) " 602

=1- ~ 2 2 ri~ni " (19)

Jo i = 1 O 3 0 - - 60i

60g-x~(m) ~n*l(m)

With eqs. (15)-(19) the statements in sections 2 . 2 . 1 -

=-m

2.2.3 are proved.

602 •

2 • ~bn,n(m)

60o - Xn(m)

Due to continuity and eq. (15d), the matrix [~k~.(m)] 5. Magnification for resonance systems

can be considered as regular• Thus, the equation for

Og(m) can be solved (on a whole neighbourhood of In the following, the notations of section 3.3.1 are

0) in an unequivocal way. The solution depends ana- used. The governing equations for the resonance sys-

260 K.A. Peters et al. /Determination of floor response spectra

tern R(coo, D1, D 2) due to excitation e at the base are (a) the magnification V(D1, D2) according to eq. (10);

(b) the magnification according to Newmark et al.

21 + 2Dlcoo21 + CO2X 1 = 2DlCOOd + COge,

= 1/(D1 +D2);

22 + 2D2COo5C2 + COgx2 = 2D2COo~:1 + CO2oxl. (20) (c) for all nine earthquake-like accelerations 0 the max-

According to eqs. (20) the transfer functions H i with imum of V(e, COo,D1, D2) for 0.27r ~< c~o ~< 407r

[according to eq. (9)]; and

respect to acceleration input at the base and accelera-

(d) the sum of average and variance of V(e, Coo,D1,

tion output for the masses m i are the following:

D2). Average and variance are taken over all nine

CO~ + 2iDl COCOo accelerations 0 and 0.27r ~< coo ~< 287r.

HI (CO,COo,Dl ) = CO2

° _ CO2+ 2iDl COCOo ' (21) [ - magnification (a), - - - magnification (b),

• magnification (c), • magnification (d)].

H2(CO, COo,D2) =Hi(CO, COo,Dl)Hl(CO, COo,D2) •

In fig. 7, the magnifications according to Biggs-

For 0 a stationary random process with spectral den- Roesset and (d) are compared.

sity P we get the following equation for the time From figs. 2 - 6 it is evident that V(D1, D2) may

average of the acceleration squares: be considered a rather good estimation of the highest

magnification we have to expect in the case of an

earthquake. The factor according to Biggs-Roesset

(2~) = ; Ini(CO, COo,Oi)l 2 P(CO) dCO. (22)

gives results comparable to (d) and therefore may be

0

considered as not exceeded in the average. The New-

Especially for white noise we get an interpretation of mark et al. factor does not describe correctly the de-

the proposed magnification factor V(D1, D2): pendence of magnification from D l . In the case of

large dampings it is exceeded essentially by (d).

/; dg2 (co, COo,D2)[ 2 dCO

20-

/o ___

\\ o

g j 114t(CO,COo,D1)t dCO

0

1

-~ = V(D1, D2). (23)

2x/D2(Dt + D2)

" '~\\\\

Of course, these considerations - due to our limited

knowledge about the statistical characteristics of

earthquakes - permit only rough conclusions on the I0 ¸

and the magnification factors according to Newmark

and Biggs-Roesset have been counterchecked on the

•

determined from published acceleration diagrams of

the earthquakes listed below: ¥

E1 Centro 1940; NS

Taft 1952; N12E

Parkfield 1966

Pacoima Dam 1971; S16E, $74W, vert.

Golden Gate 1957; NIOE, S80E, vert. a'ol o'o2 oh3 o~, obs o'o6 , " D2

Q07

K.A. Peters et al. / Determination o f floor response spectra 261

1'0.

~7

- ? -

V

o01 abe ob~ ob~ ob5 0'06

Fig. 5. For legend see text (D 1 = 0.15).

0'0~ oe

5

¥

a#; oe

abl abz o~3 ab~ o.b5 o.~

Fig. 3. For legend see text ( O 1 = 0.07).

o.bl ab2 ab3 ab= ab5 o,b6 oh; °2 abl 0'02 oh3 a'o~ abs oh6 o'o~ °2

Fig. 4. For legend see text (D 1 = 0.10). Fig. 6. For legend see text (D 1 = 0.30).

262 K.A. Peters et al. /Determination of floor response spectra

Power Plants, R.J. Hansen, ed. (MIT Press, Cambridge,

Mass., 1970).

DI -" 0.0? [2] Biggs, Seismic response spectra for equipment design

15 + ~ Dt ~ O'04 in nuclear power plants, 1st SMIRT Conf., Berlin, 1971

[3] Kaput and Shao, Generation of seismic floor res-

ence on Structural Design of Nuclear Plant Facilities,

ASCE, Chicago, I11., 1973.

[4] Sachs, MODAL-Programmbeschreibung, KWU-Bericht

R2-3338 (1975).

[5 ] Newmark, Walker, Veletos and Mosborg,

m \ \ \ \ " ~ Report RTD TDR-63-3096, Vol. IV, Contract AF 29

(601)-6253 (1965).

Fig. 7. Comparison of the magnifications according to Biggs-

Roesset ( - - - ) and the sum of average and variance of V(e,

too, D I , D 2 ) : ( - - - - ) .

- Bridge Structure Response SpectrumHochgeladen vonsanusi69
- StaadPro Knowledge BaseHochgeladen voningsabrina
- Silo BucklingHochgeladen vonrahilml
- Mathematical Methods for PhysicsHochgeladen vonArmando Vicentini
- REAL_TIME_FACE_RECOGNITION_SYSTEM_USING_2_article.pdfHochgeladen vonfarfarfifi3
- Complexe i GsHochgeladen vonmido_u3
- Dirac MatricesHochgeladen vonAnton Gribovskiy
- Lecture2 Matrix AlgebraHochgeladen vonDenise Tan
- Termpaper-Applications of Eigen Values-mathHochgeladen vonOmkar Kumar Jha
- eigenHochgeladen vonajbioinfo
- Akira Takahashi and Shaul Mukamel- Anharmonic oscillator modeling of nonlinear susceptibilities and its application to conjugated polymersHochgeladen vonGlade680
- Modes in Linear CircuitsHochgeladen vonmtichy
- Problem Set FinalHochgeladen vonsamin_laghaee
- noise pptHochgeladen vonpriyankar
- Understanding Negative Eigenvalue MessagesHochgeladen vonGuilherme Sabino Barbom
- Prony method for exponentialHochgeladen vonsoumya
- Thin walled Closed Section BeamsHochgeladen vonrs0004
- 9Hochgeladen vonponjove
- Test 3 SolutionsHochgeladen vonFrantz Clermont
- Order and ChaosHochgeladen vonAbhishek Kulkarni
- ExpHochgeladen vonEvanora Java
- NotesHochgeladen vonKevin Racso
- Final Review Am s 211 SolHochgeladen vonmriboost
- MATLab Lecture 01Hochgeladen vonajamufam
- 2. First Order SystemsHochgeladen vonAhmet A. Yurttadur
- Paul C. Bressloff and S. Coombes- Dynamics of Strongly Coupled Spiking NeuronsHochgeladen vonNeerFam
- 01673431.pdfHochgeladen vonJoyce George
- Chapter4(4.1~4.3)Hochgeladen vonRezif Sugandi
- Parabotlic EquationHochgeladen vonanonguy404

- Kristek Skaloud - Advanced Analysis and Design of Plated StructuresHochgeladen vonpvecci
- Westergaard 1942-Stress Concentrations in Plates Loaded Over Small AreasHochgeladen vonpvecci
- sackman Kelly-1979- seismic analysis of internal equipment and components.pdfHochgeladen vonpvecci
- Davies -1973- The thermal admittance of layered walls.pdfHochgeladen vonpvecci
- Robson Dodds Macvean 1971- Random VibrationHochgeladen vonpvecci
- Haensel -1975- Effects of Creep and Shrinkage in Composite ConstructionHochgeladen vonpvecci
- Thomasson_ Thin-walled C-shaped Panels in Axial CompressionHochgeladen vonpvecci
- Beams on Elastic Foundation by HetenyiHochgeladen vonpvecci

- 1 - Anisotropic dynamic damage and fragmentation of rock materials under explosive loading.pdfHochgeladen vonpaulocouceiro
- STPM 2007-2008 Experiment ListHochgeladen vonCalance Opfor
- Explantion Plate StaadHochgeladen vonAshish Loya
- Research on the Failure of the Induced Draft Fan's Shaft in a Power BoilerHochgeladen vonnaderbahrami
- 31X Problem Set 1 WEM-CTC-FINAL (POST).pdfHochgeladen vonJared Hector
- RM-sol-probHochgeladen vonSankar Kumarasamy
- 12603.pdfHochgeladen vonThulasirajan Krishnan
- Magnetic Levitation - WorkbookHochgeladen vontusharmali99
- MAE243 Lecture 3 Fall 2006Hochgeladen vonAustin Martin
- Mechanical TestingHochgeladen vonshrikantajit
- Nucl.Phys.B v.668Hochgeladen vonbuddy72
- Aerospace WebHochgeladen vonCrashCool
- Advanced Fibers MRSHochgeladen vonpraba_karanc4916
- Chapter Four: Comparing Lorentz’s understanding of Lorentz invariance to Einstein’sHochgeladen vonJuiomSDF
- Bending MomentHochgeladen vonGregory Miller
- Keynote Speaker 02Hochgeladen vonMangisi Haryanto Parapat
- Ansys ManualHochgeladen vonpriyadharshini natarajan
- Various Stresses in Ships StructureHochgeladen vonmeghbrishty
- Time-Energy Uncertainty RelationHochgeladen vonAdith Sai
- Yusoff et al. 2011.pdfHochgeladen vonAbhinay Kumar
- HC VERMA SOLUTIONS Electric Field and PotentialHochgeladen vonPawan Meena
- Lecture on Limit State Design by Manish BhutaniHochgeladen vonTarun
- Kelvin Helmholtz InstabilityHochgeladen vonAdrianio Lozhada
- Martini L1 IntroductionHochgeladen vonphdcao
- Turbulence Nutshell SlidesHochgeladen vonSelva Kumaran
- FEMA 451B Topic15-5a - Advanced Analysis Part1 Notes.pdfHochgeladen vonNivan Rolls
- Lab Report on Measuring Gravitational AccelerationHochgeladen vonkpasindu
- Forced and Natural ConvectionHochgeladen vonFélix Gallo Cruz
- 3-3 - FrictionHochgeladen voncarlos51952
- Module 6Hochgeladen vonrsrinath91