Sie sind auf Seite 1von 12

Engineering Structures 22 (2000) 210221 www.elsevier.

com/locate/engstruct

Displacement-based seismic design of buildingsapplication


M.S. Medhekar, D.J.L. Kennedy
*

Department of Civil and Environmental Engineering, The University of Alberta, Edmonton, Alta., Canada T6G 2G7 Received 27 April 1998; received in revised form 7 August 1998; accepted 7 August 1998

Abstract The displacement-based seismic design method is applied to the design of a two-storey and an eight-storey building having concentrically braced steel frames as the lateral load resisting system. To the authors knowledge, this marks the rst application of this method to the seismic design of (steel) buildings. Displacement spectra for design are generated numerically from appropriate earthquake accelerograms. The buildings are designed to have both elastic and inelastic responses in the design earthquake. The inuence of torsion due to an asymmetric building layout, column shortening, and higher vibration modes, is addressed. Nonlinear static and dynamic time history analyses are used to assess the seismic response. The displacement-based design method is shown to be a viable alternative to the current spectral acceleration-based design approach. 1999 Elsevier Science Ltd. All rights reserved.
Keywords: Concentrically braced steel frames; Displacement-based seismic design

1. Introduction The objective of this paper is to apply the theory of displacement-based seismic design, presented in the accompanying paper [1], to the design of a two-storey and an eight-storey steel building with concentrically braced frames (CBFs) for the lateral load resisting system (LLRS). A total of seven design examples are considered, conclusions are drawn, and issues requiring further work are identied.

2. Two-storey building The two-storey building, adapted from Chien [2] and shown in Fig. 1, is located in Vancouver, BC, Canada. The built-up roong system is carried by a 38 mm metal deck supported by open web steel joists and steel girder framing. The oor system has a 75 mm metal deck topped with a 75 mm concrete cover slab. The vertical LLRS consists of two CBFs in each direction but more frames could be added to provide redundancy. The nonstructural components comprise precast concrete cladding panels supported by the exterior framing, and

Fig. 1.

Symmetric layout of building.

* Corresponding author. Tel: 1 403 492 1906; Fax: 1 403 492 0249.

interior partitions of gypsum wallboard. The dead load of the roof is 1.05 kPa and of the oor is 4.80 kPa. The dead load of the cladding is 11 kN/m and 8 kN/m for the oor and the roof, respectively. The oor live load

0141-0296/00/$ - see front matter 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 1 4 1 - 0 2 9 6 ( 9 8 ) 0 0 0 9 3 - 5

M.S. Medhekar, D.J.L. Kennedy / Engineering Structures 22 (2000) 210221

211

of 2.40 kPa is subject to the live load reduction factor specied in the National Building Code of Canada (NBCC) [3]. The snow load is 1.60 kPa. 2.1. Displacement spectrum for design Fig. 2 shows the displacement response spectrum (DRS) obtained from the specied NBCC [3] elastic acceleration spectrum for Vancouver. Also shown are the 5% damped DRS at the mean plus one standard deviation ( 1) level and the mean level, of 20 accelerograms scaled for use in Vancouver [4]. These spectra were developed by integrating numerically the equation of motion of a 5% damped oscillator. The 1 DRS of the accelerograms agrees well with the NBCC DRS, also based on 5% damping, for the range of periods shown. Thus, the NBCC DRS is used for design and the response from nonlinear dynamic time history analyses is evaluated at the 1 level of the scaled responses. 2.2. Design of the building The LLRS is designed for (a) a symmetric layout of the building in plan for both elastic and inelastic responses in the specied earthquake, and (b) an asymmetric layout (with inelastic action) so that the torsional response can be assessed. For simplicity, accidental torsion is neglected for the symmetric layout shown in Fig. 1, and the stiffness and strength contributions of the exterior cladding and interior partitions are ignored. The columns are assumed to be provided in single-storey lifts, although in practice two-storey lifts would probably be used. Thus any contribution of the bending stiffness of the columns is neglected in assessing the strength of the braced frames. At the start of the seismic design process, the known quantities are the approximate weight of the oor and the roof of 12 800 kN and 4070 kN, respectively, and the geometry and material properties of the CBFs. Columns and beams are made from CSA G40.20 grade 300W steel with a yield strength of 300 MPa, while the hollow structural section (HSS) braces are made from CSA G40.20 grade 350W steel with a yield strength of 350 MPa. The CBFs are assumed to have the same geometry in the northsouth and the eastwest direc-

tions, as shown in Fig. 1. Hence, the storey drift to yield the braces, if the axial deformations of the columns are neglected, is Y FYLb cos Ecos (1)

where Y is the yield elongation of the brace, Lb is the brace length, FY is the yield stress, E is the elastic modulus, and is the angle of inclination of the brace with the horizontal. Interstorey drifts to yield the braces in the rst and second storeys are 14.7 and 13.8 mm, respectively. 2.3. Example 1, elastic response with a uniform ductility demand of 1.0 over height In this example, the CBF is designed for elastic response in the design earthquake. This is achieved by assuming an initial displaced shape which corresponds to yielding of the braces in both storeys. The brace ductility is dened as the maximum extension of the brace divided by its yield elongation. The interstorey drifts required to yield the braces are added together to give the desired displaced shape. The design procedure for a multi-degree-of-freedom system, outlined in section 5 of the accompanying paper [1], is implemented in Table 1, and the equation numbers in the following discussion refer to that paper [1]. The effective displacement of the single-degree-of-freedom system, eff, from equation (19), is 620/31 20 mm, the effective mass, meff, from equation (16) is 1.55 kN s2/mm and is about 90% of the total mass. The effective damping, eff, is taken as 5% of critical. From the NBCC displacement response spectrum in Fig. 2, corresponding to eff (20 mm) an effective period, Teff, of 0.40 s is obtained. The effective stiffness, Keff, from equation (3) is 382 kN/mm, and the base shear, Vb, from equation (20) is 7640 kN. The static lateral force at each level is obtained from equation (17). The P- effect is accounted for by adding equivalent storey shears, (P)/h, as given in Appendix J of CSA Standard S16.1-94 [5]. This force is added to the lateral force at each level and the resultant is reported in the last column of Table 1. Each CBF is designed for one half of the lateral force, Fi. Structural design of the CBF is carried out with the computer program soda [6]. The compression brace is designed to resist at least 30% of the storey shear. The members selected are given in Table 2. 2.3.1. Structural analyses and results A two-dimensional model of the CBF is used for free vibration analysis, nonlinear static analysis, and nonlinear dynamic time history analyses. The stiffness and strength contributions of the non-structural components

Fig. 2.

Displacement response spectra for Vancouver.

212

M.S. Medhekar, D.J.L. Kennedy / Engineering Structures 22 (2000) 210221

Table 1 Design parameters for example 1 Static lateral force Fi (kN) 2904 4737 7641 Static lateral force P- shear (incl. P-) Fi (kN) (kN) 14 45 2918 4782 7700

Height to Floor oor H (m) 2 1 8.1 4.2

Weight W (kN) 4069 12 767 16 836

Drift to Assumed yield brace shape yi i (mm) (mm) 13.8 14.7 28.5 14.7

mii (kN s2) 11.8 19.2 31

mi 2 i (kN s2 mm) 338 283 620

Prole of shape ci

mici (kN s2/mm) 0.59 0.96 1.55

1.43 0.74

Table 2 CBF details for example 1 Columns Beams Braces Exterior column Central column Floor beam Roof beam Storey 1 Storey 2 W200 W200 W610 W530 HSS 254 HSS 152 86 59 84 82 254 11 152 8

are neglected for simplicity. The exibility of the roof diaphragm is also neglected. This assumption is considered to be justied partly by two facts. First, only about 24% of the building mass is at the roof level and hence exibility of the roof deck may not inuence the dynamic behaviour signicantly. Second, all the columns of the building, which are likely to be provided in two-storey lifts, when bent in the second mode shape restrain the roof deck. The free vibration analysis, assuming that both the tension and compression brace are effective, gives periods of 0.38 and 0.23 s for the rst and second mode, respectively. When axial deformations of the columns are neglected, the periods of the rst and second modes decrease marginally to 0.37 and 0.21 s, respectively, indicating that column deformations have little inuence on the two-storey structure. A nonlinear static analysis under the lateral forces gives drifts of 13.2 mm at the oor and 25.5 mm at the roof. These drifts are slightly less than the assumed displaced shape as shown in Fig. 3(a).

Nonlinear dynamic time history analyses are performed with the 20 accelerograms selected for Vancouver. Braces of the CBF are modelled with the EL9 buckling element [7] of computer program drain-2D [8]. This element mimics the hysteresis behaviour of a pin-ended brace with a multilinear hysteresis rule as shown in Fig. 4. The yield capacity of the brace, Pyp, is set equal to its nominal tensile resistance. The rst cycle buckling load of the brace, Pyn, is set equal to its nominal compressive resistance. The second and subsequent cycle buckling load, Pync, is assumed equal to 0.5 Pyn, to account approximately for the reduction in compressive resistance due to cyclic buckling. Pinching behaviour of the hysteresis loop is dened by point F whose co-ordinates are an empirical function of the effective slenderness ratio. Rayleigh damping is used with 5% damping in the rst and last (second) modes to represent the nominal viscous damping. The hysteretic damping is accounted for by using the (inelastic) phenomenological model for the brace behaviour. (The sum of the nominal

Fig. 3.

Results for example 1.

Fig. 4.

Axial hysteresis behaviour of the EL9 element.

M.S. Medhekar, D.J.L. Kennedy / Engineering Structures 22 (2000) 210221

213

viscous damping and the hysteretic damping is, in effect, the effective damping used in the design procedure.) Fig. 3(a) shows that the 1 drift from dynamic analyses is marginally greater than the assumed value. The 1 value of the brace ductility demand, dened as the maximum extension of the brace divided by its yield elongation, plotted in Fig. 3(b), also exceeds the assumed value marginally. This arises because, at the fundamental period of 0.38 s, the 1 displacement response spectrum of the accelerograms in Fig. 2 has a greater spectral ordinate than the NBCC spectrum which was used for design. 2.4. Example 2, inelastic response with a uniform ductility demand of 2.0 over height In this example, the CBF is designed to respond inelastically by assuming an initial displaced shape wherein the braces at both storeys extend to 2.0 Y. Thus the amplitude of the displacements is doubled from that of example 1 in Table 1. The effective displacement from equation (19) of the accompanying theoretical paper [1] is 40 mm, and the effective mass from equation (16) is 1.55 kN s2/mm. This mass is the same as for example 1 because the displaced shapes for both examples are identical even though the displacements are doubled. From the NBCC spectrum in Fig. 2, for the target displacement of 40 mm, the effective period, Teff, is 0.66 s, as compared to 0.40 s when the ductility demand was 1.0 Y. The effective stiffness from equation (3) is 140 kN/mm and the base shear from equation (20) is 5600 kN. The effective displacement is twice that in example 1 and the base shear reduces from 7640 to 5600 kN. The lateral forces at the oor and the roof, including the P- effect, are 3560 and 2160 kN, respectively, and have the same distribution over the height as in example 1. The members selected for this CBF are given in Table 3. It is noted, as would be expected, that the columns and braces are smaller than for example 1. 2.4.1. Structural analyses and results The free vibration analysis gives periods of 0.44 and 0.26 s for the rst and second modes, respectively. The period of the rst mode is less than the effective period of 0.66 s because the former is based on the initial elastic stiffness of the CBF while the latter is based on the lesser
Table 3 CBF details for example 2 Columns Beams Braces Exterior column Central column Floor beam Roof beam Storey 1 Storey 2 W250 W200 W610 W530 HSS 254 HSS 178 67 52 84 82 254 8 178 5

Fig. 5.

Results for example 2.

secant stiffness corresponding to the effective displacement of 40 mm. Fig. 5(a) shows the oor drifts corresponding to an assumed brace ductility demand of 2.0, those from a linear static analysis, those from a pushover analysis, and those from nonlinear dynamic time history analyses. The linear elastic drifts of 13.4 and 22.2 mm at the oor and roof are, as expected, about one half of the assumed values. The pushover analysis, with the lateral force prole, indicates that the brace ductility demand is localized in storey 1. The braces in storey 2 remain elastic. The drifts from the nonlinear dynamic time history analyses agree with the assumed value in storey 1 but are signicantly less than the assumed value in storey 2 because the braces here respond elastically. Fig. 5(b) shows that the ductility demand from the time history analyses is slightly less than that assumed in storey 1 and much less than that in storey 2. 2.5. Example 3, inelastic response with an asymmetric building layout Consider the modied building plan shown in Fig. 6 wherein the locations of CBFs 2 and 3 are changed from those in Fig. 1, while the CBF geometry and material

Fig. 6.

Asymmetric layout of building.

214

M.S. Medhekar, D.J.L. Kennedy / Engineering Structures 22 (2000) 210221

properties are assumed to be the same. The design procedure for torsion, discussed in section 6 of the accompanying paper [1], is applied here. The effect of torsion is neglected initially and the displaced shape of example 2, with an implied brace ductility demand of 2.0 at each storey, is assumed at the centre of mass, CM, at each level. Therefore, the lateral forces at the oor and roof are the same as in example 2. Hence, the design given in Table 3 is adopted for the CBFs in both north south and eastwest directions. Once a preliminary design is available for analysis, the effect of torsion can be calculated. The centre of strength, CS, with all four CBFs having the same strength, lies 4500 mm to the east and 6000 mm to the south of the centre of mass, CM, as shown in Fig. 6. Pertinent calculations for the case of the earthquake force acting in the northsouth direction are summarized in Table 4 under the heading preliminary design. The torsional moment is due to the lateral force acting at an eccentricity of 4500 mm. The torsional stiffness is based on the assumption that CBFs 3 and 4 which are orthogonal to the direction of earthquake force remain elastic. CBFs 1 and 2 are assumed to have yielded in translation and therefore do not contribute any torsional stiffness. The rotation is multiplied by the distance of CBF 1 from the centre of strength (18 000 mm) to obtain the additive translation of this frame due to torsion. These displacements are added to those assumed at the start to obtain the revised drift reported in part (a) of Table 4. The corresponding brace ductility demands are 2.41 in storey 1 and 2.35 in storey 2. As the ductility demand desired is set at 2.0, the CBFs need to be strengthened to reduce the ductility demand. One alternative is to revise the initial displaced shape so that it corresponds to a reduced ductility demand in each storey. Selecting a ductility demand of 1.6 for
Table 4 Example 3, effect of torsion Item (a) Preliminary design Lateral force (kN) Torsional moment (kN mm 106) Torsional stiffness (kN mm 109) Rotation (rads 103) Translation along CBF 1 (mm) Revised drift estimate (mm) Revised brace ductility demand (b) Revised design Lateral force (kN) Torsional moment (kN mm 106) Torsional stiffness (kN mm 109) Rotation (rads 103) Translation along CBF 1 (mm) Revised drift estimate (mm) Revised brace ductility demand Floor Roof

translational effects only, the drift is 23.5 mm at the oor and 45.6 mm at the roof, the effective displacement, eff, is 32 mm, and the effective period from the displacement spectrum is 0.57 s. The effective stiffness is 188 kN/mm and the base shear is 6020 kN. When the P- effect is included, the lateral force is 3790 kN at the oor and 2320 kN at the roof, as reported in part (b) of Table 4. The HSS 178 178 5 brace provided in storey 2 (Table 3) is adequate but the HSS 254 254 8 brace provided in storey 1 is inadequate and is revised to a HSS 254 254 10 brace. The revised estimate of brace ductility demand is 1.97 in storey 1 and 1.98 in storey 2. As these demands are less than the original target value of 2.0, the design is satisfactory. 2.5.1. Structural analyses and results In order to assess realistically the effect of torsion on the inelastic response of the two-storey building, threedimensional inelastic dynamic time history analyses are carried out with the program drain-3DX [9]. The symmetric layout of the LLRS shown in Fig. 1 and the asymmetric layout shown in Fig. 6 are investigated. The oor and roof are modelled as rigid diaphragms with three degrees of freedom each: two translations and a rotation. Axial deformation of the columns is ignored. The brace members are taken to be as given in Table 3. They are assumed to have an elasto-plastic behaviour in tension and compression. The yield capacity in tension is set equal to the nominal tensile resistance, while the yield capacity in compression is set equal to 0.5 times the nominal compressive resistance. Earthquake ground acceleration is specied in the northsouth direction only. The symmetric model has periods of 0.39 and 0.20 s for the translational modes in each direction, and 0.36 and 0.19 s for the torsional modes. The asymmetric model has the following periods: 0.46, 0.39, 0.32, 0.24, 0.20, and 0.17 s. The longer period of the rst mode of the asymmetric model will reduce its seismic demand. The responses from the time history dynamic analyses at the 1 level are reported in Table 5. The asymmetric model has lower drifts but marginally greater ducTable 5 Results for example 3 Item Symm. Floor drift (mm) Roof drift (mm) Brace ductility demand (N-S) Storey 1 Storey 2 Brace ductility demand (E-W) Storey 1 Storey 2 18.6 28.1 1.26 0.86 0 0 Model Asym. 16.2 25.4 1.27 0.92 0.55 0.50

3560 25.7 76.6 0.336 6.05 35.4 2.41 3790 27.5 91.0 0.302 5.44 28.9 1.97

2160 9.72 36.0 0.270 4.86 67.9 2.35 2320 10.4 36.0 0.289 5.20 56.2 1.98

M.S. Medhekar, D.J.L. Kennedy / Engineering Structures 22 (2000) 210221

215

tility demands than the symmetric model. CBFs 3 and 4, which are orthogonal to the direction of excitation, carry no force in the symmetric model and respond elastically in the asymmetric model. The brace ductility demands of the symmetric three-dimensional model differ from those of the two-dimensional model of example 2, shown in Fig. 5(b), due to the fact that in the former, axial deformation of the columns is ignored and that the elasto-plastic model of the braces overestimates the stiffness in the compression region of the hysteresis loop as compared to the EL9 buckling element [7].

velocities scaled to the peak horizontal velocity. Fig. 8 shows the mean plus one standard deviation ( 1) displacement response spectra (DRS) of these two sets of accelerograms and that obtained from the specied NBCC [3] elastic acceleration spectrum. The average of the two periods obtained from the 1 DRS of the acceleration and velocity sets is used for design. The response from nonlinear dynamic time history analyses is evaluated at the 1 level of the scaled responses. The fundamental period of this eight-storey building is expected to be less than 1.5 s. 3.2. Design of the building

3. Eight-storey building The eight-storey building, adapted from Chien [10] and shown in Fig. 7, has been used by Redwood and Channagiri [11] for seismic design using the spectral acceleration-based procedure of the NBCC [3]. It is assumed to be located in Victoria, BC, earthquake zone 5 in Canada with a peak horizontal acceleration of 0.3 g and a peak horizontal velocity of 0.3 m/s. Although the lateral load resistance in the eastwest direction is provided by only two CBFs that lie within the building core, redundancy could be provided by having more frames. The weight of the building, including 25% of the roof snow, is about 54 200 kN, with a load of 7030 kN at the roof and 6740 kN at each oor. 3.1. Displacement spectrum for design Two sets of earthquake accelerograms are selected for use in Victoria, BC, Canada [4]. The acceleration set has 12 accelerograms with peak ground accelerations in each scaled to the peak horizontal acceleration and the velocity set has seven accelerograms with peak ground The LLRS is designed for (a) elastic response, (b) inelastic response with an arbitrarily increasing ductility demand over height. In addition, (c) the effect of column shortening, and (d) the effect of the higher modes of vibration on the ductility demand, are addressed. The CBFs are designed for earthquake loads acting in the eastwest direction. For simplicity, the CBFs are lumped together, designed and analysed as a single frame, and accidental torsion is neglected. Columns and braces are grade 350W steel with a yield strength of 350 MPa. The storey drift to yield the braces, neglecting column defor-

Fig. 8.

Displacement response spectra for Victoria.

Fig. 7. Layout of eight-storey building.

216

M.S. Medhekar, D.J.L. Kennedy / Engineering Structures 22 (2000) 210221

mations, is 15.9 mm for rst storey and 12.7 mm for all other storeys. 3.3. Example 4, elastic response with a uniform ductility demand of 1.0 over height The initial displaced shape corresponds to yielding of the braces in all storeys. The design procedure for a multi-degree-of-freedom (MDOF) system, outlined in section 5 of the accompanying paper [1], is implemented in Table 6. The effective displacement of the singledegree-of-freedom (SDOF) system, eff, from equation (19) of the accompanying paper [1] is 75 mm (24 875/334) and the effective mass, meff, from equation (16) is 4.48 kN s2/mm. The latter is about 81% of the total mass. The effective damping, eff, is taken as 5% of critical. The period corresponding to eff obtained from the displacement response spectrum of the acceleration set is 0.75 s and from the velocity set is 0.60 s. The mean of the two periods, 0.68 s, is used for design. The effective stiffness, Keff, from equation (3) is 389 kN/mm, and the base shear, Vb, from equation (20) is 28 940 kN. The lateral force at each oor, Fi, from equation (17), given in Table 6, does not include the P effect. 3.3.1. Structural design For ease of analysis and in the interpretation of results, the following simplications are made: (a) the braces are assumed to have the same capacity in tension and compression to avoid the complexity of overstrong braces; (b) the columns are assumed to be in single-storey lifts and hence do not contribute lateral stiffness by bending in the higher modes; and (c) the cross-sectional area of the columns, designed to resist the gravity load and the earthquake induced forces, is increased arbitrarily by 1.5 times to ensure elastic response in the nonlinear analyses. Alternatively, well-established capacity design principles [11] could be adopted to ensure that
Table 6 Design parameters for example 4 Floor Height to oor H (m) 29.7 26.1 22.5 18.9 15.3 11.7 8.1 4.5 Drift to yield brace yi (mm) 12.7 12.7 12.7 12.7 12.7 12.7 12.7 15.9 Assumed shape i (mm) 104.5 91.9 79.2 66.5 53.9 41.2 28.5 15.9

inelastic action is restricted to the braces only. The member areas, for the pair of CBFs lumped together, are given in Table 7. 3.3.2. Structural models and assumptions A two-dimensional truss idealization is used. The oors are assumed to be rigid and their rotational inertia is neglected. To investigate the effect of column deformations, this deformation is rst constrained and then allowed. The P- effects are not modelled in this example but are considered subsequently. In the nonlinear static analysis, the columns and braces are modelled as truss elements having a bi-linear behaviour in tension and compression. A strain-hardening modulus equal to 1% of the initial elastic modulus is specied for the braces only. For the nonlinear dynamic analyses, Rayleigh damping is assumed with the coefcients specied to give 5% damping in the rst and the last (eighth) mode to represent the nominal viscous damping. Braces are assumed to have an elasto-plastic behaviour with no strain-hardening and thus contribute hysteretic damping. Again, the effective damping is the sum of the two. 3.3.3. Structural analyses and results Fig. 9(a) shows the assumed displaced shape and that from nonlinear static analysis when column deformations are neglected. The two shapes match, as expected, because the displaced shape is obtained from the brace extensions at yielding as assumed. Fig. 10(a) shows the drifts when column deformations are considered. The structure has signicantly greater lateral exibility due to the nite axial stiffness of the columns. As the two shapes differ appreciably, the member stiffness (or the assumed displaced shape) would be revised in a normal design situation. In this example, however, no revisions are made and the brace and column areas given in Table 7, are used for dynamic analysis. When column deformations are neglected, the period

Weight W (kN) 7028 6740 6740 6740 6740 6740 6740 6740 54 208

mii (kN s2) 75 63 54 46 37 28 20 11 334

mi 2 i (kN s2 mm) 7831 5800 4311 3042 1994 1166 559 173 24 875

Prole of shape ci

Static lateral force mici Fi (kN s2/mm) (kN) 1.01 0.85 0.73 0.61 0.50 0.38 0.26 0.15 4.48 6490 5470 4715 3961 3207 2453 1698 944 28 938

8 7 6 5 4 3 2 1

1.40 1.23 1.06 0.89 0.72 0.55 0.38 0.21

M.S. Medhekar, D.J.L. Kennedy / Engineering Structures 22 (2000) 210221

217

Table 7 Brace and column areas provided in example 4 Storey 1 2 3 4 5 6 7 8

Cross-sectional area provided (mm2) Brace Column 62 225 342 220 53 805 270 700 50 540 214 955 45 825 162 480 39 660 114 740 32 050 73 185 22 985 39 270 12 475 14 450

value of the ductility demand on the braces, shown in Fig. 9(b), is obtained from the program [8] output for the brace elements. The ductility demand for storeys 1 to 6 compares well with the assumed demand of 1.0 at each storey. The dynamic values exceed the assumed ones at storeys 7 and 8 with a demand of 1.9 in storey 8, and is due to the higher modes of vibration. Fig. 10(a) shows the lateral drift from the dynamic analysis with column deformations considered. The ductility demand is less than that assumed for storeys 1 to 6 and greater for storeys 7 and 8, as shown in Fig. 10(b). The demand at storey 8 is 2.7. The increased ductility demand in the upper storeys is again due to the higher modes of vibration.
Fig. 9. Results for example 4 (column deformations neglected).

3.4. Effect of column deformations Column deformations result in an increased lateral drift, a lengthening of the periods, and increased ductility demand in the upper storeys. The lateral drift of the nth oor, nc, due to the elastic axial deformations of all columns below it is nc

i1

Cih2 i AiEDs

n1

i1

Tih2 i AiEDs

(2)

Fig. 10. Results for example 4 (column deformations considered).

of the rst mode, 0.68 s, is equal to the design period, Teff. This period increases to 0.91 s corresponding to a softer structure when column deformations are considered. The periods of the second and higher modes are also greater than the corresponding ones obtained when column deformations are neglected. Fig. 9(a) also shows the 1 value of the lateral drift from the nonlinear dynamic time history analyses with column deformations neglected. The displaced shape is in excellent agreement with the assumed shape but it is noted that the drift of the different oors do not occur simultaneously in the time history. The 1

where Ds is the width of the concentrically braced frame, E is the modulus of elasticity, hi is the storey height, Ai is the area of cross-section of the column, and Ci and Ti are the compressive and tensile axial forces in the columns of the ith storey. The axial stress, Ci/Ai and Ti/Ai, may be assumed equal to 0.7 FY to obtain an initial estimate of the oor drift, nc. Alternatively, the effect of column deformations can be accounted for approximately in the preliminary design stage by selecting an appropriate period which is less than that of the equivalent SDOF system and thus providing extra stiffness in the braces that counter-balances the effect of column deformations. 3.5. Example 5, inelastic response with nonuniform ductility demand over height This example models the frame more realistically in that: (a) the P- effect is considered by adding equival-

218

M.S. Medhekar, D.J.L. Kennedy / Engineering Structures 22 (2000) 210221

ent storey shears, P/h, as given in Appendix J of CSA Standard S16.1-94 [5]; (b) column deformations are considered; (c) the cross-sectional areas of the columns are not increased arbitrarily as in example 4; and (d) strainhardening in the braces is considered. The initial assumed displaced shape corresponds to increasing brace ductility demand over the height, as given in Table 8. From equation (17) of the accompanying paper [1], a greater displacement within a storey results in a greater lateral force. Thus, the assumed shape results in increased lateral forces in the upper storeys. Following the design procedure of example 4, the properties of the equivalent SDOF system are: mass, meff, of 3.85 kN s2/mm; effective displacement, eff, of 128 mm; period, Teff, of 1.31 s; stiffness, Keff, of 88 kN/mm; and base shear, Vb, of 11 260 kN. The effective displacement, eff, is greater than that in example 4 and results in a decrease in base shear from about 28 940 to 11 260 kN. The base shear of 11 450 kN reported in Table 8 includes the addition for the P effects. Although the base shear is reduced due to the greater effective displacement, the distribution of lateral forces is biased towards the upper storeys where the ductility demands were assumed to be greater. Fig. 11(a) shows the oor drifts. The drifts from the nonlinear static analysis are greater than those assumed due to the column deformations. No iterations are performed to make the two shapes match. An elastic free vibration analysis, neglecting column deformations, gives a fundamental period of 1.04 s, which is lengthened to 1.51 s when column deformations are considered. The design period, Teff, assuming inelastic behaviour but neglecting column deformations is 1.31 s. About 71% of the total mass participates in the rst mode. For the dynamic analyses, the P- effect is modelled by adding in parallel a stack of dummy columns with appropriate gravity loads. In the rst (dynamic 1) of the two dynamic analyses, the braces are assumed to be elaTable 8 Design parameters for example 5 Assumed ductility shape demand

Fig. 11.

Results for example 5.

sto-plastic, and in the second (dynamic 2) to have a strain-hardening modulus of 0.05 E. (In retrospect, this assumption may be unconservative. However, the increase in the yield strength due to the strain rate under dynamic loading has not been considered). Fig. 11(a) shows the drifts from both dynamic analyses which agree with the assumed shape. Strain-hardening has no noticeable effect. It is recognized that the drifts shown do not occur simultaneously and therefore may not correlate with the brace ductility demands. Fig. 11(b) gives the brace ductility demand. Dynamic (analysis) 1 gives ductility demands that exceed those assumed only in storey 8, while dynamic (analysis) 2 gives demands that are signicantly lower. It is seen that strain-hardening reduces the ductility demand where signicant yielding occurs. 3.6. Example 6, realistic design of the CBF This example assumes the same displaced shape as in example 5; thus the base shear and the lateral forces are the same. The ductility demand assumed in storeys 6, 7,

Floor

Height to oor H (m) 8 7 6 5 4 3 2 1 29.7 26.1 22.5 18.9 15.3 11.7 8.1 4.5

Weight W (kN) 7028 6740 6740 6740 6740 6740 6740 6740 54 208

Drift to yield brace y i (mm) 12.7 12.7 12.7 12.7 12.7 12.7 12.7 15.9

i (mm)
200 149 111 86 67 48 32 16

mii (kN s2) 143 102 76 59 46 33 22 11 491

mi 2 i (kN s2 mm) 28 535 15 232 8448 5028 3042 1552 690 173 62 699

Prole of shape ci

Static lateral force (incl. P-) mici Fi (kN s2/mm) (kN) 1.12 0.80 0.60 0.46 0.36 0.26 0.17 0.09 3.85 3376 2391 1745 1346 1083 748 529 232 11 450

4.00 3.00 2.00 1.50 1.50 1.25 1.25 1.00

1.56 1.17 0.87 0.67 0.52 0.37 0.25 0.12

M.S. Medhekar, D.J.L. Kennedy / Engineering Structures 22 (2000) 210221

219

and 8 is relatively large because the system is designed to respond inelastically in the specied earthquake, especially in these storeys. Inelastic action is likely to occur in the upper storeys due to the effect of the higher modes of vibration and hence is accommodated here. The program SODA [6] is used to design each CBF and select commercially available sections for the columns and braces. The beam members are not designed explicitly because the oors are modelled as rigid diaphragms in the dynamic analyses. The columns are assumed to be pin-ended in each storey for analysis and design although in practice they would probably be provided in two-storey lifts. The members selected are given in Table 9. The W310 202 and W250 73 column sections were revised from W310 158 and W250 58 sections, respectively, after the latter, selected in the preliminary design, were found to be inadequate by the capacity design procedure [11]. The braces are typically square hollow structural sections (HSS) except for the W shape provided in storey 1. The effective period for design neglecting column deformations is 1.31 s. The period of the rst mode is 0.87 s when column deformations are neglected and 1.29 s when column deformations are considered. In the nonlinear dynamic time history analyses, the EL9 buckling element [7] of drain-2D [8] is used to model the braces. The ratio of the cyclic buckling capacity to the rst cycle buckling capacity of the braces is taken as 0.6 for effective slenderness ratios less than 55, and 0.5 for effective slenderness ratios between 55 and 75. The assumed deected shape agrees well with that from nonlinear dynamic time history analyses, as shown in Fig. 12(a). The brace ductility demands in tension from the dynamic analyses are less than those assumed in storeys 1 to 7, but in storey 8 the assumed demand is marginally exceeded. 3.7. Example 7, control of ductility demand in the upper storeys The results of example 4 in Fig. 9(b) and Fig. 10(b) show that the ductility demand from dynamic analysis
Table 9 Members provided in example 6 Storey Section 1 2 3 4 5 6 7 8 WWF500 456 WWF500 456 WWF500 276 WWF500 276 W310 202 W310 202 W250 73 W250 73 Columns Area (mm2) 58 200 58 200 35 200 35 200 25 800 25 800 9280 9280

Fig. 12.

Results for example 6.

signicantly exceeds that assumed in storeys 7 and 8. This is due to the inuence of the higher modes of vibration which were not considered in the design procedure. In examples 5 and 6, the displaced shape is such that the distribution of lateral force is biased towards the upper oors. This increases the relative strength of the upper storeys and reduces the ductility demand there. In example 7, an alternative approach is used to control the ductility demand in the upper storeys. The lateral force distribution of example 4, as given in Table 6, is modied by applying 15% of the base shear at the roof with the remainder distributed over the height. (In example 4, a linear displaced shape with a ductility demand of 1.0 throughout was assumed.) This lateral force distribution was determined iteratively in order to reduce the ductility demand in the upper storeys to the target value. This approach happens to parallel that of the NBCC [3], where a fraction of the base shear is applied at the roof. Fig. 13(a) compares the lateral force prole, i.e. the lateral force normalized by the base shear, of example 4 (with column deformations considered) with that of example 7. The ductility demands from dynamic analyses are compared in Fig. 13(b). In example 7, the duc-

Braces Section W310 202 HSS 305 305 HSS 305 305 HSS 305 305 HSS 305 305 HSS 254 254 HSS 254 254 HSS 203 203 Area (mm2) 25 800 14 400 12 800 12 800 11 000 10 500 7660 4900

KL/r

13 11 11 10 11 8 6

75 46 45 45 45 55 54 67

Note: KL/r refers to the effective slenderness ratio.

220

M.S. Medhekar, D.J.L. Kennedy / Engineering Structures 22 (2000) 210221

with appropriate phenomenological models, the hysteretic damping is accounted for properly. Empirical data are needed that relate the effective damping to the ductility demand for various types of steel structures. 4.2. Initial displaced shape The initial displaced shape affects the design base shear, the static lateral force prole, and the member design requirements. Both linear and nonlinear initial displaced shapes were considered. The effective displacement of the assumed shape, eff, from equation (19), determines the base shear equation (20). A greater effective displacement results in a lower base shear. The reduction in base shear is a function of the shape of the displacement response spectrum. The static lateral force prole on the structure also depends on the initial displaced shape. A greater amplitude of displacement assumed at a particular level of the structure will increase the lateral force there. In example 5, a combined effect is evident as the base shear is reduced due to a relatively large effective displacement (equal to 128 mm) and the lateral force prole is biased towards the top due to the large displacement assumed at that level. MDOF systems will typically respond elastically in the rst mode before developing inelastic action. Hence, an initial displaced shape with a prole that matches the rst mode shape is a good choice. 4.3. Allowance for other effects The displacement-based seismic design method has been applied to the design of a two-storey and an eightstorey building with concentrically braced frames. Based on the design examples considered, the following conclusions are drawn. 4.1. Design displacement spectra Design displacement spectra may be generated by integrating numerically the response of a SDOF system subjected to appropriate ground motions. The mean plus one standard deviation displacement spectra considered in this study agree well with the NBCC [3] spectrum for periods up to 1.0 s for the same level of damping. For longer periods, the generated spectra are lower because the NBCC spectrum includes a correction for the effect of higher modes. In all the design examples considered, an effective damping ratio of 5% was used whether or not inelastic action occurred. Because the 5% corresponds to essentially elastic response (nominal viscous damping), the effective damping should be increased when hysteretic behaviour occurs. Having used the lower value, the effective period obtained from the displacement spectrum is less and the base shear used in design is greater. Nevertheless, when time history analyses are performed Effects due to column deformations and the P- effect may be accounted for approximately in the preliminary design stage itself. The lateral drift due to column deformations may be estimated from Eq. (2) of this paper and the P- effect may be calculated based on the assumed displaced shape and the gravity loads. 4.4. Results of dynamic analyses The variation of oor drift and brace ductility demand over the building height depends on the relative stiffness and relative yield strengths of the storeys. The displaced shape from dynamic analyses agrees fairly well with the assumed shape for most of the design examples. In example 4, the ductility demand in the upper two storeys exceeds the assumed demand due to the higher modes of vibration. When the assumed shape has large displacements in the upper oors, the distribution of lateral force is biased towards these oors. In example 7, the lateral force distribution was adjusted without changing the base shear so that 15% of the base shear was applied at the roof and the remainder distributed over the height. Both these procedures increase the relative strength of the upper storeys and reduce the ductility demand there. In assessing the dynamic response, brace ductility

Fig. 13.

Comparison of results, example 4 vs. example 7.

tility demand at storeys 7 and 8 is signicantly less than that in example 4 and the demand at all storeys is close to the target value of 1.0. The adjustment in the lateral force prole increases the strength and stiffness of the upper storeys and thereby decreases the ductility demand signicantly. The demand increases slightly at the bottom storey. Further study is required to determine the adequacy of the 15% base shear modication for tall CBFs.

4. Conclusions

M.S. Medhekar, D.J.L. Kennedy / Engineering Structures 22 (2000) 210221

221

demand is more important than the target oor drift because damage to structural and non-structural elements is directly related to the interstorey drift. 5. Further work This study indicates that further work is required in a number of areas to address the following issues: 1. Effective damping values for steel structures responding inelastically; 2. damage thresholds for common non-structural elements; 3. constitutive relationships under dynamic loading for steel braces and appropriate phenomenological models for dynamic analyses; 4. appropriate displacement spectra for design purposes; and 5. application of the method to the design of other lateral load resisting systems, such as moment resisting frames, eccentrically braced frames, and steel plate shear walls.

References
[1] Medhekar MS, Kennedy DJL. Displacement-based seismic design of buildingstheory. Engineering Structures, 2000;22(3):2019.

[2] Chien E. Low rise ofce building design aid. Canadian Institute of Steel Construction, Willowdale, Ontario, 1991. [3] Canadian Commission on Building and Fire Codes. National Building Code of Canada, National Research Council of Canada, Ottawa, Ontario, 1995. [4] Medhekar MS, Kennedy DJL. Seismic evaluation of steel buildings with concentrically braced frames. Structural Engineering Report 219. Department of Civil and Environmental Engineering, The University of Alberta, Edmonton, Alberta, Canada, October 1997. [5] CSA, Limit states design of steel structures, CSA Standard CAN/CSA-S16.1-94. Canadian Standards Association, Rexdale, Ontario, 1994. [6] SODA, Structural Optimization, Design, and Analysis, Version 3.2.5a. Acronym Software Inc., Waterloo, Ontario, 1994. [7] Jain AK, Goel SC. Hysteresis models for steel members subjected to cyclic buckling or cyclic end moments and buckling. Users guide for drain-2D: EL9 and EL10. Report no. UMEE 78R6, Department of Civil Engineering, University of Michigan, Ann Arbor, Michigan, 1978. [8] Kannan AE, Powell GH. drain-2D: a general purpose computer program for dynamic analysis of inelastic plane structures. Report no. EERC 73-6, University of California, Berkeley, California, 1973. [9] Prakash V, Powell GH, Campbell S. drain-3DX: base program description and user guide. Report no. UCB/SEMM-94/07, Department of Civil Engineering, University of California, Berkeley, California, August 1994. [10] Chien E. Multi-storey steel building design aid. Canadian Institute of Steel Construction, Willowdale, Ontario, 1987. [11] Redwood RG, Channagiri VS. Earthquake resistant design of concentrically braced steel frames. Canadian Journal of Civil Engineering 1991;18(5):83950.

Das könnte Ihnen auch gefallen