Sunteți pe pagina 1din 18

Bulletin of Earthquake Engineering 1: 303320, 2003. 2003 Kluwer Academic Publishers. Printed in the Netherlands.

303

Seismic Design of Symmetric Structures for Accidental Torsion


1 Central Laboratory for Seismic Mechanics and Earthquake Engineering, Bulgarian Academy of Sciences, P.O. Box 27, Soa 1582, Bulgaria; Currently Seconded National Expert at ELSA, JRC of the European Commission, TP480, I-21020, Ispra (VA), Italy; 2 University of Architecture, Civil engineering and Geodesy, 112 P.O. Box 18, Soa 1712, Bulgaria; Author for correspondence (Tel: +359 2 979 33 84; Fax: +359 2 971 24 07; E-mail: sdimova@geophys.bas.bg; silva.dimova@prc.it

SILVIA L. DIMOVA1, and ILIA ALASHKI2

Abstract. The paper presents an analytical estimation of the dynamic effects, caused by the shifting of the centre of mass with accidental eccentricity in symmetric structures. The approximate analytical solution proves, that even under small accidental eccentricities the symmetric structures exhibit irregular behaviour and the accidental torsional effects cannot be described properly by static application of torsional moments. The prescribed application rule by Eurocode 8 for multimodal analysis underestimates the accidental torsional effects up to 21% for 5% eccentricity for the structures considered in the paper. An expression for the correction of member responses is derived. It is proved by numerical simulations of the dynamic response of three-dimensional models of symmetric structures, that the proposed correction coefcient gives accurate results in cases of single-storey and multi-storey structures. It gives a convenient way for the design practice to estimate accurately the accidental torsional effects on planar and 3-D models of symmetric structures. Key words: accidental eccentricity, correction coefcient for accidental torsional effects, Eurocode 8, seismic design of symmetric structures, static application of torsional moments Abbreviations: CM centre of mass; CR centre of rigidity; R/C reinforced concrete; Principle of Eurocode 8 principle requirement of Eurocode 8 for estimation of accidental torsional effects; Application rule of Eurocode 8 application rule of the principle requirement of Eurocode 8 for estimation of accidental torsional effects.

1. Introduction It is a common practice in the seismic codes to include the effects of accidental torsion in the dynamic analysis of buildings by displacing the centre of mass (CM) of each oor from its nominal position by a distance equal to the accidental eccentricity. The rotational motion of the building base, the uncertainty in the location of CM and in some codes - the uncertainty in the stiffness of structural elements as well as the uncertainty in the yield force and ductility of the bearing elements are considered as sources of accidental torsion. Recently, Humar and Kumar (2000) reviewed and compared the torsion design provisions of UBC and Codes of Canada, New Zealand and Mexico. The results

304

SILVIA L. DIMOVA AND ILIA ALASHKI

show that the provisions of Canada and Mexico Codes are overly conservative for the design of the elements on the exible side of the building (i.e., the more distant from CR side). When the ratio of the translational over the torsional natural periods is less than 0.75, the torsional provisions of all the codes considered by the authors are unconservative for the design of elements on the stiff side of the building (i.e., the less distant from centre of rigidity (CR) side). Basing on parametric studies performed on non-orthogonal and torsionally irregular structures, Tezcan and Alhan (2001) have drawn recommendations about the amplication constant of the accidental eccentricity of the Turkish Code in view of the safety of the stiff-side elements. The accounting for the accidental eccentricity requires considerable computational efforts, since different positions of the CM have to be considered to nd the envelope of the resulting torsional effects. The principle requirement of Eurocode 8 (2002) for accidental torsional effects (referred further as the Principle of Eurocode 8), prescribes that the calculated centre of mass at each oor i shall be considered displaced from its nominal location in each direction by an additional accidental eccentricity eli = 0.05 Li , where Li is the oor-dimension perpendicular to the direction of the seismic action. In this way, when working with three dimensional (3-D) models of the structures, four additional calculations of the seismic action effects have to be performed to obtain the most unfavourable member forces for each structural element. In order to simplify this situation, the following application rule is supplied in Eurocode 8 (referred further as the Application rule of Eurocode 8), for the case of multimodal response spectrum analysis: whenever a spatial model is used for the analysis, the accidental torsional effects may be determined as the envelope of the effects resulting from an analysis for static loading consisting of torsional moments about the vertical axis of each storey, produced by the horizontal seismic force shifted in accidental eccentricity, derived by simplied modal spectrum analysis for all relevant directions. The above Application rule implies consideration of the accidental torsional effects as a separate loading case, when multimodal response spectrum analysis is applied. In such a case combination and not algebraic sum of the accidental torsional effects has to be applied with the effects of the horizontal components of the seismic excitation. A procedure to extend the static torsional provisions of Eurocode 8 to asymmetrical multi-storey buildings is presented by Moghadam and Tso (2000). They developed a rened method for determination of CM eccentricity and torsional radius for multi-storey buildings. Dempsey and Tso (1982) proposed a scheme for estimating the equivalent static seismic eccentricity in buildings with torsionally unbalanced distributions of mass and stiffness. The concept of effective eccentricity is introduced on the place of the concept of dynamic eccentricity, so that the edge lateral displacement is matched. De la Llera and Chopra (1995) proposed a fourstep procedure for including the effects of all sources of accidental torsion in the seismic design of buildings. The ratio of the static displacements of a system with

SEISMIC DESIGN OF SYMMETRIC STRUCTURES

305

Figure 1. Plan of the single-storey model.

and without accidental eccentricity is taken as a measure of the increase of the response due to accidental eccentricity. De la Llera and Chopra (1994a) proved that the discrepancies between the computed and actual stiffness values of the structural elements imply that a building with nominally symmetric plan is actually asymmetric to some unknown degree and will undergo torsional vibrations, when subjected to purely translational ground motion. The sensitivity analysis of the building response with respect to the uncertainty in the element stiffness shows that the response of nominally symmetricplan systems increases by a greater amount relative to asymmetric-plan systems. Using results from Monte Carlo simulations for estimation of the effects of the uncertainty of the structural stiffness, the design eccentricity is presented as a function of the ratio p of the uncoupled translational over the torsional natural periods. The numerical results show, that the design eccentricity increases to a maximum at p 0.9, decreases to a minimum at p 1, next increases to a maximum at p 1.2 and remains almost constant for p 1.2. In the context of the results obtained by De la Llera and Chopra (1994a), a considerable difference should be expected between the results obtained by consideration of the accidental torsional effects using dynamic analysis with shifted CM at the accidental eccentricity (for example, applying the Principle of Eurocode 8) and the results obtained by static application of torsional moments (for example, using the Application rule of Eurocode 8). The accounting of such difference in case of symmetric structures would be very important, since their seismic design is usually performed on two planar models and the accidental torsional effects are accounted by static application of the torsional moments. This paper deals with analytical estimation of the dynamic effects in symmetric structures, caused by the shifting CM with accidental eccentricity and the subsequent application of the results in the seismic design.

306

SILVIA L. DIMOVA AND ILIA ALASHKI

Figure 2. Displacements of points L and R caused by the translational and torsional modes.

2. Mathematical formulation The dynamic effects of shifting CM with accidental eccentricity are considered on the example of a single-storey structure. Mass and stiffness are symmetrically distributed. The roof diaphragm is assumed to be rigid in its plane, but it is exible in bending in the vertical direction, which is a reasonable representation of the true behaviour of widely used types of oor systems, e.g., cast-in-place one (Chopra, 2001). CM, as required by the Principle of Eurocode 8, is shifted with eccentricity e from its nominal location along the x axis (Figure 1). The considered system has three degrees of freedom displacements along the x and y-axes and rotation of the roof diaphragm around CM. When e = 0, the dynamic behaviour of the structure is described by three uncoupled equations, since both, the mass and the stiffness matrices are diagonal. When CM is displaced along x axis, the equations, describing the dynamic response about the y axis and the rotation around the CM are coupled due to the off-diagonal terms in the stiffness matrix, arising from the eccentricity. The seismic response of the undamped system to a seismic excitation represented by an accelerogram agy (t) acting in y direction could be presented as: m 0 0 Im y + eky ky eky k + e2 ky y = m 0 agy (t), (1)

where m is the mass of the system; I m is the mass moment of inertia; ky is the lateral stiffness of the system; k is the torsional stiffness of the system about CR. Exact solution of Eq. (1) is given by Dempsey and Irvine (1979), Tso and Dempsey (1980), Dempsey and Tso (1982). In this paper an approximate solution is

SEISMIC DESIGN OF SYMMETRIC STRUCTURES

307

Figure 3a. Scheme of the displaced by dynamic analysis structure.

Figure 3b. Correction criterion.

used, aiming at representation of the expressions in a suitable form for the design practice. Supposing that the accidental eccentricity does not change substantially the natural periods of the symmetric system, it is accepted that Ttr = 2 m ky and Ttor = 2 Im , k (2)

308

SILVIA L. DIMOVA AND ILIA ALASHKI

where T tr is the period of the mode of natural vibrations with predomination of translation (called further translational mode); T tor is the period of the mode of natural vibrations with predomination of torsion (called further torsional mode). This assumption is supported by the numerical results of the case studies discussed in Section 4. When shifting CM in 5% of the respective oor dimension of the studied systems, the highest changes in the rst three periods of natural vibrations do not exceed 3%. Under this assumption an approximate solution of the eigenvalue problem is obtained. It gives the following normalized eigenvectors: 1 and 2 els 2 2 ry ls { tor } = 1 , (3)

tr }

e 2 2 ry + e2 ls

where { tr } and { tor } are the eigenvectors of the translational and torsional 2 = k /ky is the square of the stiffness radius of gyration mode, respectively; ry 2 = Im /m is the about vertical axis passing through CR (torsional radius); ls square of radius of gyration of the system plan about vertical axis passing through CM. As seen from Eq. (3), for close values of ry and ls the two mode shapes change very signicantly even for small eccentricities. This effect conrms the considerable sensitivity of the symmetric structures to the accidental torsional effects for close values of ry and ls proved by De la Llera and Chopra (1994a). Thus, when ry ls , even under small accidental eccentricities, the symmetric structures exhibit irregular behaviour and their dynamic behaviour should be studied by 3-D models. Based on Eq. (3), the seismic forces for the two modes are calculated as: 2 2 2 Sa(Ttr )m(ry + e2 ls ) 2 + e 2 l 2 )2 + l 2 e 2 (r y s s Ftr = ; Mtr 2 2 2 2 Sa(T )ml e(r + e l ) tr s y s 2 + e 2 l 2 )2 + l 2 e 2 (ry s s 2 2 Sa(Ttor )mls e 2 l 2 )2 + l 2 e 2 (r y s s Ftor = , (4) Mtor 2 2 2 Sa(Ttor )mls e(r l ) y s (r 2 l 2 )2 + l 2 e2 y s s where Sa(T) is the acceleration response spectrum of agy (t); F tr and M tr are the force and moment induced by the translational mode; F tor and M tor are the force and moment induced by the torsional mode.

SEISMIC DESIGN OF SYMMETRIC STRUCTURES

309

The displacement ycr of CR and the rotation cr around vertical axis passing through CR under a moment M and a force F applied at CM, are: ycr = F ky M + eF cr = k

(5)

It should be noted, that cr = , since the roof diaphragm is rigid in its plane. By use of Eq. (5), the edge displacements of the system can be expressed under the seismic loading of the two modes. The displacements of the edge points R and L (shown in Figure 1) are:
R = ycr + ytr

Ftr cr .A (Mtr + eFtr )A = + 2 ky 2k Ftor cr .A (Mtor + eFtor )A R = ytor = ycr + + 2 ky 2k F (M .A + eF cr tr tr tr )A L = ytr = ycr 2 ky 2k Ftor cr .A (Mtor + eFtor )A L = ytor = ycr , 2 ky 2k

(6)

where A is the dimension of the building plan, orthogonal to y (see Figure 1); yR tr and yR are the displacements of point R under the seismic loading of the transtor L lational mode and torsional mode, respectively; yL tr and ytor are the displacements of point L under the seismic loading of the translational mode and torsional mode, respectively; F tr , M tr , F tor and M tor are dened by Eq. (4). The displacements yL and yR of points L and R, induced by the two modes are obtained by square root of the sum of the squares combination of the displacements caused by the separate modes. Since for ry ls the natural periods of the torsional and translational modes are closed, beating phenomenon appears in the seismic response of the structure leading to the summation of the effects of the two modes. The beating phenomenon is not taken into account in the solution, since for ry ls R L R the displacements yL tr and ytr are negligible in comparison with ytor and ytor , as depicted in Figure 2. The displacements shown in Figure 2 are calculated for the system and e = 5% presented in Section 3. Also, the good coincidence between the theoretical and numerical results, presented in Sections 3 and 4 veries this assumption. When no accidental eccentricity is applied to the system, the seismic force F 0 , acting at CR is: F0 = Sa(Ttr )m and the corresponding displacement y0 of CR is: y0 = F0 /ky (8) (7)

310

SILVIA L. DIMOVA AND ILIA ALASHKI

Figure 4. Plan of the single-storey test structure.

The increase of the system response due to application of accidental eccentricity can be accessed by means of a correction coefcient C, dened by: y0 C = max(y L , y R ) (9)

Since the element forces R in the individual lateral load resisting elements are linearly proportional to the displacements, the element forces Rc of a structural member located in distance x from CR corrected for the accidental torsional effects are: R c = [1 + (C 1) 2x ]R A (10)

The correction condition given by Eq. (9) is derived from the displacements of the edge points of the oor slab. As seen from Figures 3a and 3b, such assumption gives a conservative estimation of the member responses of the interior resisting elements. When an exact estimation of the increase of the response of an element located in distance x from CR is needed, the displacements in Eq. (6) should be recalculated for the points R and L, positioned symmetrically in distance x from CR, as shown in Figures 3a and 3b. Also, in the case of the outermost lateral load resisting elements located in xs < A/2, the displacements in Eq. (6) should be calculated for the points R and L, corresponding to their position xs , as illustrated in Section 4. 3. Numerical verication The proposed method in Section 2 for consideration of accidental torsional effects is veried with numerical results for the single-storey symmetric structure shown in Figure 4. The dimensions in plan are 42 18 m, the storey high is 6 m. The lateral load resisting elements are frames and shear walls. The shear walls are 0.3 m thick and 6 m long, the columns cross-section is 0.5 0.5 m, and the

SEISMIC DESIGN OF SYMMETRIC STRUCTURES

311

thickness of the oor slab is 0.25 m. The mass of the structure is 100.103 kg. The design seismic loading is calculated using the Eurocode 8 response spectrum for subsoil category C and design ground acceleration for Soa area ag = 0.27 g. The behaviour factor is considered 2.5 for all structural modications. The shear walls are shifted symmetrically along x axis to obtain different values of p. Twelve structural modications are considered with p varying between 0.358 and 1.592. For each p the seismic response of the symmetric system to the above-specied response spectrum is calculated. The effect of the accidental eccentricity is estimated as the envelope from the dynamic analysis of two new models with CM shifted in x = e. Accidental eccentricities of 2.5%, 5% and 10% are considered. The eccentricity of 2.5% is considered having in mind two reasons. First, there are Codes, which prescribe eccentricities of this order for symmetric buildings (for example the Bulgarian Code, 1987). The second reason is connected with the possibility given by the concept of the effective seismic wave to incorporate the effects of the spatial variation of the ground motion directly in the expressions for the design seismic loading (Tzenov et al. 2001; Dimova et al., 2000). In this case the accidental eccentricity is connected mainly with the uncertainties in the location of the masses and it could be smaller, than the value of 5%, prescribed by Eurocode 8. In Figures 5a, b, c the calculated correction coefcients C are shown for the column in point L (see Figure 4) for accidental eccentricity of 2.5%, 5% and 10%, respectively. The results from the exact dynamic analysis are marked by dots. In the same gure the theoretical curves are plotted for points R and L, obtained by Eq. (9) and with bold line is marked their envelope. In Figures 5a, b, c are shown also the correction coefcients, obtained by static application of torsional moment dened according to the Application rule of Eurocode 8. As seen from Figure 5a, for 2.5% accidental eccentricity there is very good coincidence between the proposed theoretical curve and the numerical results, obtained by dynamic analysis. On the contrary, the correction coefcients calculated by the static application of torsional moment deviate substantially from those, calculated by dynamic analysis. For p < 0.5 it overestimates the structural dynamic response (differences between the correction coefcients up to 24%), for p 0.5 the static application of torsional moment underestimates the structural dynamic response (differences up to 25%). For 5% accidental eccentricity (Figure 5b) there is also very good coincidence between the proposed theoretical curve and the numerical results, obtained by dynamic analysis of the system with accidental eccentricity. Only the numerical result for p = 1.07 exhibit 11% difference from the theoretically predicted value. As in the case of e = 2.5%, the correction coefcients calculated by static application of torsional moment deviate from those, calculated by dynamic analysis, but the differences are smaller up to 21% for p = 1.07. In the case of simplied modal response analysis or multimodal analysis performed on planar models, Eurocode 8 allows accounting for the accidental torsional effects by amplifying the seismic action effects in the individual structural mem-

312

SILVIA L. DIMOVA AND ILIA ALASHKI

Figure 5a. Correction coefcient for 2.5% eccentricity.

Figure 5b. Correction coefcient for 4% eccentricity.

bers with a correction factor, proportional to the distance of the load resisting elements from CR. This correction factor, as shown in Figure 5b, corresponds to C = 1.3. It should be mentioned, that for 0.7 < p < 1.3 the correction factor of Eurocode 8 underestimates up to 20% the accidental torsional effects calculated by dynamic analysis. It is near the numerical results obtained by static analysis of the system with accidental eccentricity for 0.7 < p < 1.3 (differences up to 10%). These results show the need of implementation of higher value of the correction factor of Eurocode 8 in order to give a correct estimate of the dynamic effects due to accidental eccentricity. As shown in Figure 5b, a value of C = 1.6 could be proposed as coinciding well with theoretical and numerical results for 0.6 < p < 1.1 (differences up to 7%) and giving a conservative estimate of the accidental torsional effects for p 1.1 (overestimation up to 28%). As seen from Figure 5c, for e = 10% there is relatively good coincidence between the proposed theoretical curve and the numerical results, obtained by dynamic analysis of the system with accidental eccentricity. The highest difference

SEISMIC DESIGN OF SYMMETRIC STRUCTURES

313

is observed for 0.78 < p < 0.92, where the theoretical curve overestimates the effects of accidental eccentricity up to 20%. It should be noted, that in this range of p the theoretical results give a conservative estimate of the accidental eccentricity effects. For e = 10% the correction coefcients calculated by static application of torsional moment coincide well with those, calculated by dynamic analysis (differences up to 13% for p = 1.15). 4. Implementation of the proposed correction coefcient in seismic design of multi-storey structures It is shown by De la Llera and Chopra (1994b), that the ratio of the edge displacements considering the effect of accidental torsion and the ratio of the edge displacements when neglecting accidental torsion, are essentially the same for a single-storey system and a multi-storey system belonging to the special class if both have the same frequency ratio p, uncoupled lateral period T y and static eccentricity e. In the same work it is outlined, that an important consequence of this result is the fact, that the correction coefcient obtained for a given resisting plane applies to all the stories of the multi-storey building and hence to the forces in all structural elements in the same resisting plane. This statement gives reasons to extend the theoretical estimate of the correction coefcient obtained in Section 2 for the accidental eccentricity to the case of symmetric multi-storey buildings. Four symmetric multi-storey buildings with different torsional rigidity are considered. Design seismic loading is calculated using the Eurocode 8 response spectrum for subsoil class B. For all the structures the design ground acceleration is taken for Soa area as ag = 0.27 g and the importance factor is accepted equal to 1. The behaviour factors are estimated according to Eurocode 8 for ductility class L. The denition of the eigen periods and vectors of the structures and the calculation of the member responses is performed by SAP 2000 (1998). The torsional rigidity of the structures is estimated by the ratios: px = Tx /Ttor and py = Ty /Ttor , (11)

where T x and T y are the periods of the translational modes in the directions x and y, respectively. The effect of the accidental eccentricity is estimated as the envelope from the dynamic analysis of four new models with CM shifted in x = e and y = e. The considered symmetric multi-storey buildings are described in Table I. The distance of the outermost lateral load resisting elements from CR is marked by xs in the x direction. Plans of the considered structures are shown in Figure 6a, b, c, d. In Table I the member responses obtained by exact dynamic analysis in the considered structures for accidental eccentricities of 2.5%, 5% and 10% and without consideration of accidental eccentricity are also shown. Basing on them, the exact correction coefcients are calculated and compared with those, obtained

314

SILVIA L. DIMOVA AND ILIA ALASHKI

Table I. Correction coefcients for multi-storey buildings. Structure 1 Plan dimensions: AB, m Levels Story high, m T 1 , s and mode shape T 2 , s and mode shape T 3 , s and mode shape xs ,m ls ,m px py element bending moment without accidental eccentricity, kNm with accidental eccentricity, kNm e = 2.5% correction coefcient from calculation correction coefcient from Eq. (9) difference with accidental eccentricity, kNm correction coefcient from calculation correction coefcient from Eq. (9) difference 6030 5 3.6 0.535 Ux 0.533 Uy 0.305 tors. 30 19.365 1.751 1.744 WR1 37129 39308 1.059 1.057 0% 41352 1.111 1.111 0% 44977 1.211 1.207 0.3% Structure 2 Structure 3 6030 5 3.6 0.331 Ux 0.215 Uy 0.15 tors. 30 19.365 2.208 1.426 WR1 6172 6860 1.11 1.11 0% 7449 1.207 1.206 0% 8322 1.348 1.348 0% 4224 4 3.6 0.2225 Uy 0.2222 tors. 0.2217 Ux 10 13.964 1 1 WL 50931 56708 1.113 1.056 5.1% 56876 1.117 1.115 1.8% 57532 1.130 1.251 10.7% Structure 4 6030 3 3.1 0.286 tors. 0.176 Uy 0.174 Ux 30 18.083 0.606 0.616 CLD 126 168 1.333 1.331 % 219 1.738 1.775 2.12% 318 2.524 2.551 1.07%

e = 5%

with accidental eccentricity, kNm e = 10% correction coefcient from calculation correction coefcient from Eq. (9) difference

SEISMIC DESIGN OF SYMMETRIC STRUCTURES

315

by the proposed approximate analytic solution. The denotations of the elements are shown in Figure 6 a,b,c,d. The seismic resistant elements of Structure 1 are R/C shear walls, as shown in Figure 6a. The structure is torsionally stiff, since px = 1.751 and py = 1.744. As seen from Table I, the calculated correction coefcient C fully coincides with the theoretically predicted by Eq. (9) one for accidental eccentricities of 2.5% and 5%. For e = 10%, the correction coefcient predicted by Eq. (9) underestimates with 0.3% the calculated one. The seismic resistant elements of Structure 2 are R/C shear walls and cores, as shown in Figure 6b. The structure is torsionally stiff, since px = 2.208 and py = 1.426, as shown in Table I. Similarly to the case of Structure 1, the calculated correction coefcient C practically coincides with the theoretically predicted by Eq. (9) one for the three considered eccentricities. The seismic resistant elements of Structure 3 are the R/C shear walls, as shown in Figure 6c. The structure exhibits px = 1 and py = 1 and its dynamic behaviour with shifted CM exhibits singularities. When CM is shifted in x = e the rst and third mode became predominantly torsional with contribution of translation in y direction. The second mode becomes translational in x direction. Thus the design seismic loading of the lateral loading bearing elements in y direction is contributed mainly by the torsional rst and third modes (cumulative sum of the modal participating mass ratios 77%). In comparison with the case of e = 0 these elements are unloaded, but the lateral loading bearing elements in x direction are overloaded. Similar phenomenon appears when shifting the CM in y = e. In this way the accidental torsional effects for the lateral loading bearing elements in y direction are obtained by shifting CM in y = e and those in x direction by shifting CM in x = e. As seen from Table I, for e = 2.5% the predicted correction coefcient by Eq. (9) underestimates the calculated one with 5.1% and for e = 5% by 1.8%. For e = 10% the predicted C overestimates the calculated one with 10.7%. It should be noted, that in this structure the outermost lateral load bearing elements are located inside the contour of the oor slabs in 10 m from CR in the both, x and y directions (Figure 6c). The correction coefcients presented in Table I are estimated from Eq. (9) for xs = 10 m. When working with A/2 instead of xs , the correction coefcients are C = 1.20 for e = 2.5%, C = 1.25 for e = 5% and C = 1.55 for e = 10%, thus leading to overestimation of the accidental torsional effects. The lateral load bearing elements of Structure 4 are R/C space frames and a core, as shown in Figure 6d. The structure is torsionally exible, since px = 0.606 and py = 0.616. Similarly to the Structure 1 and Structure 2, for the three considered eccentricities the calculated correction coefcient C almost fully coincides with the theoretically predicted by Eq. (9). There is no difference for e = 2.5%, the differences for e = 5% and e = 10% are 2.12% and 1.07%, respectively.

316

SILVIA L. DIMOVA AND ILIA ALASHKI

Figure 6a. Plan of Structure 1.

Figure 6b. Plan of Structure 2.

Figure 6c. Plan of Structure 3.

Figure 6d. Plan of Structure 4.

SEISMIC DESIGN OF SYMMETRIC STRUCTURES

317

5. Implementation of the correction coefcient in seismic design practice Aiming at an easier implementation of the proposed correction coefcient in the seismic design practice, the expressions presented in Section 2 could be modied in more convenient form. Since all the three, y0 , yL and yR in Eq. (9) are proportional to w = m/ky , one may work with the following expressions of the displacements, given in Eq. (6) and Eq. (8): yw R tr =
Rk ytr y Mtr tr =F + 2A 2[ m + m m ry R k ytor y Ftor A Mtor yw R 2[ m tor = m = m + 2ry L ytr ky Ftr A Mtr yw L 2[ m + tr = m = m 2ry L ytor ky Ftor A Mtor yw L 2[ m tor = m = m 2ry y ky yw0 = 0 = Sa(Ttr ) m

eFtr ] m

= G1 + = G1

A 2 G2 2ry A 2 H2 2ry

+ +

eFtor ] m

= H1 + = H1

eFtr ] m

A 2 G2 2ry A 2 H2 2ry

(12)

eFtor ] m

where: G1 = G2 =
2 2 2 Sa(Ttr )(ry + e2 ls ) 2 + e 2 l 2 )2 + l 2 e 2 (ry s s 2 2 2 Sa(Ttr )e(ry + e2 )(ry + e2 ls )

2 + e 2 l 2 )2 + l 2 e 2 (ry s s 2 2 Sa(Ttor )ls e H1 = 2 2 )2 + l 2 e 2 (ry ls s 2 2 2 Sa(Ttor )ls e(ls ry + e2 ) H2 = 2 l 2 )2 + l 2 e 2 (ry s s

(13)

The following steps are required to correct the member forces for accidental eccentricity e by the proposed coefcient: 1. Preparation of data: 1.1 When using spatial model of the structure: calculate py = Ty /Ttor , Sa(Ty ), 2 2 = (A2 + B 2 )/12 for rectangular oor slabs or ls = Im /m for oor Sa(Ttor ), ls slabs with re-entrant corners or recesses and ry = py ls . 1.2 When using planar model of the structure and T y is known: calculate k and 2 2 kym xm + kxm1 ym kym , ky from the structural plan as k = 1 and ky =
m m1 m

where kym is the stiffness of the m-th resisting plane in y direction, kxm1 is the stiffness of the m1-th resisting plane in x direction, xm is the distance of the mth resisting plane to CR and ym1 is the distance of the m1-th resisting plane to 2 2 = k /ky . Calculate ls = (A2 + B 2 )/12 for rectangular oor CR. Determine ry 2 = Im /m for oor slabs with re-entrant corners or recesses. Determine slabs or ls py = ry / ls , T tor = Ty /py , Sa(Ty ) and Sa(Ttor ). More rened method for estimation of T tor is described by De la Llera and Chopra (1995). 2. Calculation of corrected member forces for accidental eccentricity:

318

SILVIA L. DIMOVA AND ILIA ALASHKI

2.1. Evaluate G1 , G2 , H 1 and H 2 from Eq. (13). R L L 2.2. Determine yw R tr , yw tor , yw tr , yw tor and yw0 from Eq. (12). R L R L 2 2 2 2 (yw L 2.3. Estimate yw = (yw tr ) + (yw R tor ) and yw = tr ) + (yw tor ) . R L max(yw , yw ) 2.4. Calculate C = yw0 2.5. Find the corrected element forces Rc in the individual lateral load resisting elements for the accidental torsional effects using Equation (10). As seen from the above procedure, the calculation of the correction coefcient C for planar or space models of structures is not complicated and could be easily implemented in the seismic design practice. 6. Implication in codes for seismic design of structures As seen from the above presented results, the effect of accidental eccentricity on symmetric buildings cannot be correctly assessed using static application of torsional moments M li about the vertical axis of each storey i: Mli = eli Fi (14)

where e1i is the accidental eccentricity of storey mass i from its nominal location, applied in the same direction at all oors; F i is the horizontal force acting on storey i derived by simplied modal spectrum analysis for all relevant directions. When comparing Eq. (14) with Eq. (4) and Eq. (5) it is seen, that the torsional moment arising from the change of the dynamic properties of the structure is not taken into account in Eq. (14). The smaller the accidental eccentricity, the smaller the value of M li and hence for small values of the accidental eccentricity, the static application of M li implies accidental torsional effects noticeably lower than those obtained by dynamic analysis (differences up to 25% in the considered in section 3 case studies). Only in case of p < 0.6 (torsionally very exible structures more theoretical, than practical case) the static application of M li gives conservative estimate of accidental torsional effects. Normally, the seismic design of symmetric structures is performed on two planar models. To account properly for the dynamic effects of the accidental eccentricity four 3-D models of a symmetric structure should be built and calculated and the envelope of the member responses should be estimated. The implementation of the proposed procedure for evaluation of the correction coefcient for accidental torsional effects gives a convenient way for the design practice to account for accidental torsional effects on planar models of the structures. Viewing at the forthcoming application of Eurocode 8, it should be noticed, that the amplication factor of member responses for accounting of accidental eccentricity underestimates up to 20% the accidental torsional effects calculated by dynamic analysis. In this connection a value of C = 1.6 could be proposed as coinciding well with theoretical and numerical results for 0.6 < p < 1.1 and

SEISMIC DESIGN OF SYMMETRIC STRUCTURES

319

giving a conservative estimate of the accidental torsional effects for p > 1.1. Although, as seen from Figure 5b, the presentation of the amplication factor as a constant value in the whole interval of values of p gives a rough estimation of the accidental torsional effects. In this context the implementation of the proposed correction coefcient would allow to increase the accuracy of the simplied modal response analysis and the multimodal analysis performed on planar models. In case of spatial models of the structures, the prescribed static application of the torsional moments by the Application rule in Eurocode 8 underestimates the accidental torsional effects up to 21% for 5% eccentricity. The implementation of the proposed correction coefcient allows avoiding the prescribed static analysis by Eurocode 8 for estimation of the envelope member forces and, on the other hand, increases the reliability of the results. 7. Conclusions On the basis of the carried out analytical and numerical investigations of the accidental torsional effects on symmetric structures, the following conclusions could be drown up: 1. Even under small accidental eccentricities the symmetric structures exhibit irregular behaviour and the accidental torsional effects cannot be described properly by static application of torsional moments. For the considered structures, the prescribed application rule by Eurocode 8 for multimodal analysis underestimates the accidental torsional effects up to 21% for 5% eccentricity. To account properly for the dynamic effects of the accidental eccentricity four 3-D models of a symmetric structure should be built and calculated and the envelope of the member responses should be estimated. 2. The correction coefcient derived on the basis of analytical approximate solution accounts for the change of the dynamic properties of the symmetric structures under accidental eccentricity. It gives a convenient way for the design practice to estimate accurately the accidental torsional effects on planar and 3-D models of the symmetric structures. Acknowledgements The authors gratefully acknowledge Prof. Dr Konstantin Meskouris for the valuable consultations and support of the work on the project Seismic resistant design of irregular reinforced concrete structures in RWTH Aachen. The nancial support of Alexander von Humboldt Foundation for the project Seismic resistant design of irregular reinforced concrete structures is gratefully acknowledged. The rst author gratefully acknowledges the nancial support from the Bulgarian National Scientic Fund for granting the project NS1104 Seismic risk mitigation of building structures.

320 References

SILVIA L. DIMOVA AND ILIA ALASHKI

Chopra, A.K. (2001) Dynamics of Structures. Theory and Applications to Earthquake Engineering, Prentice-Hall, Upper Saddle River. De la Llera, J.C. and Chopra, A.K. (1994a) Accidental torsion in buildings due to stiffness uncertainty. Earthquake Engineering and Structural Dynamics, 23(2), 117136. De la Llera, J.C. and Chopra, A.K. (1994b) Using accidental eccentricity in code-specied static and dynamic analysis of buildings. Earthquake Engineering and Structural Dynamics, 23(7), 947967. De la Llera, J.C. and Chopra, A.K. (1995) Estimation of Accidental Torsion Effects for Seismic Design of Buildings. Journal of Structural Engineering, 21(1), 102114. Dempsey, K.M. and Irvine, H.M. (1979) Envelopes of maximum seismic response for a partially symmetric single storey building model. Earthquake Engineering and Structural Dynamics, 7(2), 161180. Dempsey K.M. and Tso, W.K. (1982) An alternative path to seismic torsional provisions. Soil Dynamics and Earthquake Engineering, 1(2), 310. Dimova, S.L, Petrov, P. and Bonev, Z. (2000) Precising of the design seismic loading of irregular structures. In Hasan Boduroglu et al. (eds), Proc. 3rd Japan-Turkish Workshop on Earthquake Engineering, Istanbul, 6576. Dimova, S.L. and Alashki, I. (2002) Seismic resistant design of irregular reinforced concrete structures. Final project report. RWTH-Aachen, Germany. Eurocode 8: DRAFT No 5 (May 2002) Design of Structures for Earthquake Resistance. Part 1: General Rules, Seismic Actions and Rules for Buildings. PrEH 1998-1, Commission of the European Communities, European Committee for Standardization. Regulations for Design of Buildings and Structures in Seismic Regions (1987). KTSU, Soa. Humar, J. and Kumar, P. (2000) A new look at the torsion design provisions in seismic building codes. Proc. 12th World Conference on Earthquake Engineering, New Zealand Society for Earthquake Engineering, Upper Hut, New Zealand. Paper No. 1707. Moghadam, A.S.and Tso, WK. (2000) Extension of Eurocode 8 torsional provisions to multi-storey buildings. Journal of Earthquake Engineering, 4(1), 2541. SAP2000 (1998) Integrated Finite Element Analysis and Design of Structures. Computers & Structures, Inc., Berkeley. Tezcan, S.S. and Alhan, C. (2001) Parametric analysis of irregular structures under seismic loading according to the new Turkish earthquake code. Engineering Structures, 23, 600609. Tso, W.K. and Dempsey, K.M. (1980) Seismic torsional provisions for dynamic eccentricity. Earthquake Engineering and Structural Dynamics, 8(3), 275289. Tzenov, L., Dimova, S.L., Bonev, Z. and Petrov, P. (2001) Seismic resistant design of irregular structures. Generalized method for determination of design seismic loading. Technical report TDV/TR 037-62, Turkish Earthquake Foundation, Istanbul.

S-ar putea să vă placă și