Académique Documents
Professionnel Documents
Culture Documents
14
R = 1, 2, 4, 6, 8 (thin → thick lines) 10 d = 1.41, ρf == 0.72
0.99
10 d = 0.85, ρf == 1.17
0.97
10 d = 2.59, fρ==1.23
0.90
12 IND spectra
DES spectrum
10 µ
8 0 0 0
0 50 0 4 0 25
µ
6
50 d = 1.41, ρf == 0.72
0.99
50 g = 1.56, ρh == 0.96 50 g = 0.26, ρh == 0.34
0.89
4 0.57
2 µp
0 0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 10 0 4 0 25
T (sec) 4 ρ = 0.97
4 ρ = 0.96
4 ρ = 0.95
d = 0.85, f = 1.17 g = 0.64, h = 1.75 g = 0.24, h = 0.70
µr
2
R = 1, 2, 4, 6, 8 (thin → thick lines) 0 0 0
0 10 0 50 0 25
COV(∆nlin )
25 ρ = 0.90
25 ρ = 0.89
25 ρ = 0.95
d = 2.59, f = 1.23 g = 3.92, h = 2.93 g = 4.10, h = 1.42
MIV n
Sˆ (To ) y
a
γ
MIV
Sˆ (To→T ) 0 0 0
0
a µ
0 10 0 50 0 4
0 T (sec) 2.0 data
regression
by
Kenneth T. Farrow
Former Graduate
Research Assistant
and
Yahya C. Kurama
Assistant Professor
Seismic design approaches in current U.S. building provisions advocate using several
static and dynamic analysis procedures. Among the static procedures, it is common to use linear
and nonlinear methods that depend on capacity-demand index relationships (e.g., the relationship
between the design lateral strength and the maximum lateral displacement). The benefit of using
these relationships comes from their simplicity and adaptability, however significant deficiencies
exist in their development.
This research proposes new capacity-demand index relationships and ground motion scal-
ing methods and shows that: (1) previous capacity-demand index relationships developed using
linear-elastic ground motion spectra can lead to unconservative designs, particularly for survival-
level, soft soil, and near-field conditions; (2) the correlation between the maximum displacement
ductility demand and other demand indices is relatively strong; and (3) scaling methods that work
well for ground motions recorded on stiff soil and far-field conditions lose their effectiveness for
soft soil and near-field conditions.
ii
3.5 Reference Response Spectra ..................................................................................43
3.6 Nonlinear Dynamic Time-History Analyses..........................................................47
3.6.1 SDOF analyses...........................................................................................47
3.6.2 MDOF analyses .........................................................................................48
3.7 Statistical Evaluation of the Results ......................................................................51
3.7.1 SDOF demand estimates............................................................................52
3.7.2 MDOF demand estimates ..........................................................................56
iii
CHAPTER 9 REGRESSION ANALYSES BETWEEN THE DEMAND INDICES...104
9.1 Regression Relationships Developed Based on IND Spectra ..............................104
9.2 Effect of Reference Response Spectra on the Relationships Between
the Demand Indices..............................................................................................123
REFERENCES ............................................................................................................................167
iv
LIST OF TABLES
Table 2.2: Demand indices introduced by Mahin and Lin (1983). .........................................13
Table 3.5: Site soil definitions (adapted from IBC 2000 (ICC, 2000)) ...................................35
Table 3.6: Values for a and b coefficients developed in this study .........................................43
Table 3.7: Seismic coefficients for the smooth design (DES) response spectra .....................45
Table 8.1: Regression coefficients a and b for the N&K ground motion ensemble,
EP hysteresis type, α = 0.10 ..................................................................................94
Table 8.2: Regression coefficients a and b for the SAC ground motion ensemble.................95
Table 9.2: Regression coefficients d and f: SD, BE, and BP hysteresis types,
IND spectra ..........................................................................................................106
Table 9.4: Regression coefficients g and h: SD, BE, and BP hysteresis types,
IND spectra ..........................................................................................................108
v
Table 9.5: Correlation coefficient, ρ: EP hysteresis type, IND spectra ................................109
Table 9.6: Correlation coefficient, ρ: SD, BE, and BP hysteresis types, IND
spectra ..................................................................................................................110
Table 10.1: Structure properties and results for the design example ......................................145
Table 10.2: µp, µr, and ny demands for the design example ...................................................146
Table A.1: University of Notre Dame (UND) very dense (SC) soil ensemble.......................174
Table A.2: University of Notre Dame (UND) stiff (SD) soil ensemble .................................175
Table A.3: University of Notre Dame (UND) soft (SE) soil ensemble ..................................176
Table A.4: Nassar and Krawinkler (N&K) 15s very dense (SC) soil ensemble.....................177
Table A.5: SAC Boston design-level stiff (SD) soil ensemble ...............................................178
Table A.8: SAC Boston survival-level soft (SE) soil ensemble .............................................181
Table A.9: SAC Los Angeles design-level stiff (SD) soil ensemble ......................................182
Table A.10: SAC Los Angeles survival-level stiff (SD) soil ensemble ....................................183
Table A.11: SAC Los Angeles design-level soft (SE) soil ensemble.......................................184
Table A.12: SAC Los Angeles survival-level soft (SE) soil ensemble.....................................185
vi
Table A.13: SAC Los Angeles near-field design-level stiff (SD) soil ensemble ......................186
vii
LIST OF FIGURES
Figure 2.1: R-µ-T relationships developed by Miranda (1993) for rock and soft soil sites. ....10
Figure 2.2: Hysteresis types used by Foutch and Shi (1998) ...................................................11
Figure 2.3: Bilinear elasto-plastic (EP) hysteresis type with definitions for demand indices
in Table 2.2.............................................................................................................13
Figure 3.2: Hysteresis types: (a) LE; (b) EP; (c) SD; (d) BE; (e) BP. ......................................21
Figure 3.5: BP hysteresis type with: (a) βs > βr, βr < 1; (b) βs = βr < 1. .................................24
Figure 3.6: EP hysteresis type versus BP hysteresis type: (a) βs > βr; (b) βs = βr (matching
yield point). ............................................................................................................25
Figure 3.7: Layout of structural system: (a) elevation; (b) four-story structure;
(c) eight-story structure..........................................................................................26
Figure 3.8: MDOF models: (a) four-story elevation; (b) eight-story elevation;
(c) close-up of analytical model.............................................................................29
Figure 3.9: Element models: (a) beam end rotational spring element; (b) column
base fiber element ..................................................................................................30
viii
Figure 3.11: Equivalent linear model for the site response analyses: (a) assumed nonlinear
hysteretic stress-strain behavior of the soil; (b) equivalent
linear model. ..........................................................................................................36
Figure 3.12: Soil properties for the site response analyses: (a) shear wave velocity;
(b) shear modulus reduction factor and damping ratio. .........................................38
Figure 3.13: Acceleration time-history, cumulative RMSA function (CRF), and derivative of
the CRF function (1940 El Centro (ELCN) x 2.0):
(a) forward CRF; (b) reverse CRF. ........................................................................39
Figure 3.14: Scaling based on the spectral acceleration (UND SC soil ground motion
ensemble): (a) at the structure period ( Sˆa ( T o ) method); (b) over a
range of structure periods ( Sˆa ( T o → T µ ) method). ..............................................41
Figure 3.15: Design spectra: (a) general shape; (b) design-level; (c) survival-level. .................44
Figure 3.16: Smooth response spectra: (a) Boston, design-level, SD soil; (b) Boston, survival-
level, SD soil...........................................................................................................45
Figure 3.17: Smooth response spectra: (a) Boston, design-level, SE soil; (b) Boston, survival-
level, SE soil. ..........................................................................................................46
Figure 3.18: Smooth response spectra: (a) Los Angeles, design-level, SD soil;
(b) Los Angeles, survival-level, SD soil. ................................................................46
Figure 3.19: Smooth response spectra: (a) Los Angeles, design-level, SE soil;
(b) Los Angeles, survival-level, SE soil. ................................................................46
Figure 3.20: Smooth response spectra: (a) Los Angeles, design-level, SC soil;
(b) Los Angeles, design-level, SD soil, NF. ...........................................................47
Figure 3.21: Flowchart describing parameters studied in the analytical procedure. ..................51
Figure 3.22: Average response spectra of the ground motions used in the MDOF analyses:
(a) MIV scaling method; (b) Sˆa ( T o → T µ ) scaling method. ................................52
Figure 3.24: Effect of a and b coefficients on c coefficient: (a) c1 term; (b) c2 term;
(c) c = c1 + c2. ........................................................................................................54
Figure 3.25: Effect of a and b coefficients on µ: (a) b = 0.1; (b) b = 1.0; (c) b = 2.0. ...............55
ix
Figure 4.1: Comparison between CDSPEC and DRAIN-2DX (EP hysteresis type,
α = 0.10, R = 8): (a-b) EQ09; (c-d) EQ15. ............................................................59
Figure 4.2: Comparison between CDSPEC and BISPEC (SD hysteresis type, α = 0.10, R =
8): (a-b) LPPR; (c-d) PACH; (e) different reloading rules.....................................60
Figure 4.4: Comparison between CDSPEC and BISPEC for the N&K ground
motion ensemble (α = 0.10): (a) EP hysteresis type; (b) SD hysteresis type. .......62
Figure 4.5: R-µ relationship for the EP hysteresis type with α = 0.00 and T = 0.20 sec.
under the 1983 Coalinga Parkfield Zone 16 ground motion (IND response
spectrum)................................................................................................................63
Figure 4.6: R-µ relationships for the EP hysteresis type with α = 0.00 (IND
reference spectra): (a) T = 0.20 sec.; (b) T = 0.92 sec. ..........................................64
Figure 4.7: R-µ spectra (N&K ensemble, EP hysteresis type, α = 0.10): (a) AVG versus
IND spectra; (b) DES versus IND spectra. ............................................................65
Figure 4.8: Smooth design (DES) response spectrum versus average (AVG)
ground motion response spectrum. ........................................................................66
Figure 4.9: R-µ relationships using different reference response spectra (EP
hysteresis type, α = 0.10, T = 3.0 sec.): (a) N&K ensemble; (b) N&K ensemble
without EQ14. ........................................................................................................66
Figure 5.1: Effect of post-yield stiffness ratio, α (EP hysteresis type, SAC Los Angeles,
survival-level, SD soil): (a) µ; (b) µp; (c) µr; (d) ny................................................68
x
Figure 6.1: Effect of site soil characteristics (EP hysteresis type, α = 0.10, SAC
Los Angeles, design-level): (a) µ; (b) µp; (c) µr; (d) ny.........................................75
Figure 6.2: Effect of site soil characteristics (SD, BE, and BP hysteresis types,
α = 0.10, SAC Los Angeles, design-level): (a) µ; (b) µp; (c) µr; (d) ny. ...............76
Figure 6.3: Response spectra: (a) SD soil ground motion spectrum versus SE soil ground
motion spectrum; (b) ground motion spectra versus smooth
design spectra.........................................................................................................77
Figure 6.4: Effect of seismic demand level (EP hysteresis type, α = 0.10, SAC
Los Angeles, SD and SE soil): (a) µ; (b) µp; (c) µr; (d) ny. ....................................78
Figure 6.5: Effect of seismic demand level (SD, BE, and BP hysteresis types, α = 0.10,
SAC Los Angeles, SD soil): (a) µ; (b) µp; (c) µr; (d) ny.........................................79
Figure 6.7: Effect of site seismicity (EP hysteresis type, α = 0.10, SAC
survival-level, SD and SE soil): (a) µ; (b) µp; (c) µr; (d) ny....................................81
Figure 6.8: Effect of site seismicity (EP hysteresis type, α = 0.10, SAC
design-level, SD and SE soil): (a) µ; (b) µp; (c) µr; (d) ny. .....................................83
Figure 6.9: Effect of site seismicity (SD, BE, and BP hysteresis types, α = 0.10,
SAC survival-level, SD soil): (a) µ; (b) µp; (c) µr; (d) ny. ......................................84
Figure 6.10: Effect of epicentral distance (EP hysteresis type, α = 0.10, SAC Los Angeles,
design-level, SD soil): (a) µ; (b) µp; (c) µr; (d) ny..................................................85
Figure 7.1: R-µ spectra for low seismicity (EP hysteresis type, α = 0.10):
(a-d) stiff soil profile, SD; (e-h) soft soil profile, SE...............................................87
Figure 7.2: R-µ spectra for high seismicity (EP hysteresis type, α = 0.10):
(a-d) stiff soil profile, SD; (e-h) soft soil profile, SE...............................................89
Figure 7.3: R-µ spectra for near-field (Los Angeles design-level SD soil, EP
hysteresis type, α = 0.10: (a) AVG versus IND spectra; (b) DES
versus IND spectra. ................................................................................................90
Figure 7.4: Effect of hysteresis type and reference response spectra on the µ
demand (Los Angeles survival-level SD soil, α = 0.10): (a-b) SD hysteresis
type; (c-d) BE hysteresis type; (e-f) BP hysteresis type (βs = βr = 1/3). ...............92
xi
Figure 8.1: Regression curves (EP hysteresis type, α = 0.10): (a) constant-µ
versus constant-R approaches; (b) IND spectra versus AVG and DES spectra. ....93
Figure 8.2: Comparison between mean R-µ spectra and regression curves (EP hysteresis
type, α = 0.10): (a-b) SAC Boston design-level; (c-d) SAC
Los Angeles SD soil; (e-f) SAC Los Angeles SE soil; (g) SAC Boston
survival-level SE soil; (h) SAC Los Angeles design-level SD soil, NF. .................96
Figure 8.3: Effect of hysteretic behavior on regression curves using IND spectra (SAC
Los Angeles, survival-level, SD soil): (a) post-yield stiffness
ratio, α; (b) hysteresis type; (c) βs = βr..................................................................97
Figure 8.4: Effect of site soil characteristics on regression curves using IND
spectra (SAC Los Angeles, design-level): (a) EP hysteresis type;
(b) SD hysteresis type; (c) BE hysteresis type; (d) BP hysteresis type. ................98
Figure 8.5: Effect of seismic demand level on regression curves using IND
spectra (SAC Los Angeles): (a-b) EP hysteresis type; (c) SD
hysteresis type; (d) BE hysteresis type; (e) BP hysteresis type. ............................99
Figure 8.6: Effect of site seismicity on regression curves using IND spectra:
(a-d) EP hysteresis type; (e) SD hysteresis type; (f) BE hysteresis
type; (g) BP hysteresis type. ................................................................................100
Figure 8.7: Effect of epicentral distance on regression curves using IND spectra
(SAC Los Angeles, design-level, SD soil)............................................................101
Figure 8.9: Effect of reference response spectra on regression curves (SAC Los Angeles,
EP hysteresis type, α = 0.10): (a-b) SD soil; (c-d) SE soil. ..................................102
Figure 8.10: Effect of reference spectra on regression curves (SAC Los Angeles,
design-level, SD soil, NF, EP hysteresis type, α = 0.10)......................................102
Figure 8.11: Effect of reference response spectra on regression curves (SAC Los Angeles,
survival-level, SD soil, α = 0.10): (a) SD hysteresis type;
(b) BE hysteresis type; (c) BP hysteresis type (βs = βr. = 1/3). ...........................103
Figure 9.1: Matrix plots of cross-correlations: effect of post-yield stiffness ratio, α (EP
hysteresis type, SAC Los Angeles, survival-level, SD soil): (a) α = 0.00;
(b) α = 0.05; (c) α = 0.10.....................................................................................111
xii
Figure 9.2: Matrix plots of cross-correlations: effect of hysteresis type (α = 0.10, SAC
Los Angeles, survival-level, SD soil): (a) SD hysteresis type;
(b) BE hysteresis type. .........................................................................................112
Figure 9.4: Matrix plots of cross-correlations: effect of site soil characteristics (EP
hysteresis type, α = 0.10, SAC Los Angeles, design-level): (a) SD soil;
(b) SE soil. ............................................................................................................114
Figure 9.5: Matrix plots of cross-correlations: effect of site soil characteristics (SD
hysteresis type, α = 0.10, SAC Los Angeles, design-level): (a) SD soil;
(b) SE soil. ............................................................................................................115
Figure 9.6: Matrix plots of cross-correlations: effect of site soil characteristics (BE
hysteresis type, α = 0.10, SAC Los Angeles, design-level): (a) SD soil;
(b) SE soil. ............................................................................................................115
Figure 9.7: Matrix plots of cross-correlations: effect of site soil characteristics (BP
hysteresis type, α = 0.10, βs = βr = 1/3, SAC Los Angeles,
design-level): (a) SD soil; (b) SE soil....................................................................116
Figure 9.8: Matrix plots of cross-correlations: effect of seismic demand level (EP
hysteresis type, α = 0.10, SAC Los Angeles, SD soil):
(a) survival-level; (b) design-level. ......................................................................116
Figure 9.9: Matrix plots of cross-correlations: effect of seismic demand level (EP
hysteresis type, α = 0.10, SAC Los Angeles, SE soil):
(a) survival-level; (b) design-level. ......................................................................117
Figure 9.10: Matrix plots of cross-correlations: effect of seismic demand level (SD
hysteresis type, α = 0.10, SAC Los Angeles, SD soil):
(a) survival-level; (b) design-level. ......................................................................117
Figure 9.11: Matrix plots of cross-correlations: effect of seismic demand level (BE
hysteresis type, α = 0.10, SAC Los Angeles, SD soil):
(a) survival-level; (b) design-level. ......................................................................118
Figure 9.12: Matrix plots of cross-correlations: effect of seismic demand level (BP
hysteresis type, α = 0.10, βs = βr = 1/3, SAC Los Angeles, SD soil):
(a) survival-level; (b) design-level. ......................................................................119
xiii
Figure 9.13: Matrix plots of cross-correlations: effect of site seismicity (EP
hysteresis type, α = 0.10, SAC survival-level, SD soil): (a) Los
Angeles; (b) Boston. ............................................................................................119
Figure 9.20: Matrix plots of cross-correlations: effect of epicentral distance (EP hysteresis
type, α = 0.10, SAC Los Angeles, design-level, SD soil):
(a) far-field; (b) near-field. ...................................................................................123
xiv
Figure 9.24: Matrix plots of cross-correlations: effect of reference response
spectrum (EP hysteresis type, α = 0.10, SAC Boston, survival-level, SE soil):
(a) AVG spectrum; (b) DES spectrum. ................................................................131
Figure 9.33: Λ-µ regression curves using IND, AVG, and DES spectra (EP
hysteresis type, α = 0.10, SAC Boston): (a-b) SD soil; (c-d) SE soil. ..................136
Figure 9.34: Λ-µ regression curves using IND, AVG, and DES spectra (EP
hysteresis type, α = 0.10, SAC Los Angeles): (a-b) SD soil; (c-d) SE
soil........................................................................................................................137
xv
Figure 9.35: Λ-µ regression curves using IND, AVG, and DES spectra (EP
hysteresis type, α = 0.10, SAC Los Angeles, design-level, SD soil,
NF). ......................................................................................................................138
Figure 9.36: Λ-µ regression curves using IND, AVG, and DES spectra (α = 0.10,
SAC Los Angeles, survival-level, SD soil): (a) SD hysteresis type;
(b) BE hysteresis type; (c) BP hysteresis type, βs = βr = 1/3. .............................139
Figure 10.1: Inelastic demand spectra based on DES response spectra (EP
hysteresis type, α = 0.10): (a-b) SAC Boston, SD soil; (c-d) SAC
Boston, SE soil. ....................................................................................................141
Figure 10.2: Inelastic demand spectra based on DES response spectra (EP
hysteresis type, α = 0.10): (a-b) SAC Los Angeles, SD soil; (c-d) SAC
Los Angeles, SE soil.............................................................................................141
Figure 10.3: Inelastic demand spectra based on DES response spectra (EP
hysteresis type, α = 0.10): SAC Los Angeles, SD soil, NF. .................................142
Figure 10.4: Inelastic demand spectra based on DES response spectra (SAC Los Angeles,
survival-level, SD soil, α = 0.10): (a) SD hysteresis type;
(b) BE hysteresis type; (c) BP hysteresis type, βs = βr = 1/3. .............................142
Figure 10.5: Capacity curve-demand spectra for design example: (a) using DES spectrum;
(b) using IND spectra...........................................................................................144
Figure 11.1: Scatter in ∆nlin for the UND SD soil ensemble (EP hysteresis type, α = 0.10) -
PGA method compared to: (a) EPA method; (b) EPV method;
(c) MIV method; (d) A95 method; (e) Sˆa ( T o ) method;
(f) Sˆa ( T o → T µ ) method.....................................................................................148
Figure 11.2: Scatter in ∆nlin using the Sˆa ( T o ) and Sˆa ( T o → T µ ) methods
compared to the MIV method (EP hysteresis type, α = 0.10): (a) UND
SD soil; (b) UND SC soil; (c) UND SE soil; (d) SAC SD soil, NF........................150
Figure 11.4: Effect of hysteresis type on the scatter in ∆nlin (UND SD soil ensemble,
α = 0.10): (a) SD type; (b) BE type; (c) BP type (βs = βr = 1/3).........................152
Figure 11.5: Scatter in ∆nlin for the UND SC soil ensemble (EP hysteresis type, α = 0.10) -
PGA method compared to: (a) EPA method; (b) EPV method;
xvi
(c) MIV method; (d) A95 method; (e) Sˆa ( T o ) method;
(f) Sˆa ( T o → T µ ) method.....................................................................................153
Figure 11.6: Scatter in ∆nlin for the UND SE soil ensemble (EP hysteresis type, α = 0.10) -
PGA method compared to: (a) EPA method; (b) EPV method;
(c) MIV method; (d) A95 method; (e) Sˆa ( T o ) method;
(f) Sˆa ( T o → T µ ) method.....................................................................................154
Figure 11.7: Scatter in ∆nlin for the SAC Los Angeles, design-level, SD soil, NF ensemble
(EP hysteresis type, α = 0.10) - PGA method compared to:
(a) EPA method; (b) EPV method; (c) MIV method; (d) A95 method;
(e) Sˆa ( T o ) method; (f) Sˆa ( T o → T µ ) method. ...................................................156
Figure 11.8: Scatter in MDOF demands for the UND SE soil ensemble using the
MIV method and the Sˆa ( T o → T µ ) method: (a) four-story structure;
(b) eight-story structure........................................................................................158
Figure 11.9: Covariance in the MDOF demands for the UND SE soil ensemble
using the MIV method and the Sˆa ( T o → T µ ) method: (a) four-story structure;
(b) eight-story structure........................................................................................159
Figure A.1: Ground motion records: University of Notre Dame (UND) very dense
(SC) soil ensemble................................................................................................187
Figure A.2: Ground motion records: University of Notre Dame (UND) stiff (SD)
soil ensemble........................................................................................................189
Figure A.3: Ground motion records: University of Notre Dame (UND) soft (SE)
soil ensemble........................................................................................................191
Figure A.4: Ground motion records: Nassar and Krawinkler 15s very dense (SC)
soil ensemble........................................................................................................193
Figure A.5: Ground motion records: SAC Boston design-level stiff (SD) soil
ensemble. .............................................................................................................195
Figure A.6: Ground motion records: SAC Boston survival-level stiff (SD) soil
ensemble ..............................................................................................................197
Figure A.7: Ground motion records: SAC Boston design-level soft (SE) soil
ensemble ..............................................................................................................199
Figure A.8: Ground motion records: SAC Boston survival-level soft (SE) soil ensemble
(generated using EERA site response analysis program,
Bardet et al., 2000)...............................................................................................201
xvii
Figure A.9: Ground motion records: SAC Los Angeles design-level stiff (SD) soil
ensemble. .............................................................................................................203
Figure A.10: Ground motion records: SAC Los Angeles survival-level stiff (SD) soil
ensemble. .............................................................................................................205
Figure A.11: Ground motion records: SAC Los Angeles design-level soft (SE) soil
ensemble. .............................................................................................................207
Figure A.12: Ground motion records: SAC Los Angeles survival-level soft (SE)
soil ensemble (generated using EERA site response analysis program, Bardet
et al., 2000). .........................................................................................................209
Figure A.13: Ground motion records: SAC Los Angeles design-level near-field
(NF) ensemble......................................................................................................211
Figure A.14: Response spectra: University of Notre Dame (UND) very dense (SC)
soil ensemble........................................................................................................213
Figure A.15: Response spectra: University of Notre Dame (UND) stiff (SD) soil ensemble. ..213
Figure A.16: Response spectra: University of Notre Dame (UND) soft (SE) soil ensemble. ...214
Figure A.17: Response spectra: Nassar and Krawinkler 15s very dense (SC) soil ensemble....214
Figure A.18: Response spectra: SAC Boston design-level stiff (SD) soil ensemble. ................215
Figure A.19: Response spectra: SAC Boston survival-level stiff (SD) soil ensemble. ..............215
Figure A.20: Response spectra: SAC Boston design-level soft (SE) soil ensemble..................216
Figure A.21: Response spectra: SAC Boston survival-level soft (SE) soil ensemble
(generated using EERA site response analysis program, Bardet et al., 2000).....216
Figure A.22: Response spectra: SAC Los Angeles design-level stiff (SD) soil ensemble. .......217
Figure A.23: Response spectra: SAC Los Angeles survival-level stiff (SD) soil ensemble. .....217
Figure A.24: Response spectra: SAC Los Angeles design-level soft (SE) soil ensemble. ........218
Figure A.25: Response spectra: SAC Los Angeles survival-level soft (SE) soil ensemble
(generated using EERA site response analysis program,
Bardet et al., 2000)...............................................................................................218
xviii
Figure A.26: Response spectra: SAC Los Angeles design-level stiff (SD) soil near-
field (NF) ensemble..............................................................................................219
xix
LIST OF SYMBOLS
AVG = average ground motion response spectrum based on the ground motion ensemble
C = sample covariance
Ca = seismic coefficient tabulated in UBC 1997 based on site seismicity and site soil characteris-
tics
xx
Cv = seismic coefficient tabulated in UBC 1997 based on site seismicity and site soil characteris-
tics
DES = smooth design response spectra from current U.S. seismic design provisions
Es = Arias Intensity
Fdes = design lateral strength based on first “significant yield” of the structure
xxi
Ft = portion of the design force concentrated at the top of the structure in addition to Fn
F yr ( µ ) = required nonlinear force capacity based on bedrock motion below the soft soil site
F yr, e = linear-elastic force demand based on bedrock motion below the soft soil site
F ys ( µ ) = required nonlinear (i.e., reduced) force capacity at the soft soil site
hc = column depth
xxii
Ks2 = moment-curvature post-yield stiffness of the zero-length rotational spring model
kµ = secant stiffness
m = structure mass
Na = near-field factor in UBC 1997 based on the closest distance from the site to a known seismic
source
Nv = near-field factor in UBC 1997 based on the closest distance from the site to a known seismic
source
NF = near-field
PI = plasticity index
xxiii
px = probability of having x occurrences during the design life, tdes
Rdes = response modification coefficient based on first “significant yield” of the structure
Sa = spectral acceleration
Sd = spectral displacement
xxiv
SM1 = seismic coefficient corresponding to the “maximum considered earthquake” in IBC 2000
SMS = seismic coefficient corresponding to the “maximum considered earthquake” in IBC 2000
t = time
T = period
Tr = return period
xxv
αbe = post-yield stiffness ratio of the BE component for the BP hysteresis type
αep = post-yield stiffness ratio of the EP component for the BP hysteresis type (equal to zero)
γ = shear strain
δ = dispersion
∆
˙˙ g = ground motion acceleration
∆max = maximum mean floor or roof lateral displacement over the height of the structure
∆*nlin = maximum nonlinear displacement demand of the idealized SDOF representation for the
structure
xxvi
∆r = residual displacement demand
∆*ult = displacement capacity of the idealized SDOF representation for the structure
η = viscosity
θmax = maximum mean interstory drift over the height of the structure
ρ = correlation coefficient
xxvii
Ρ = coefficient of determination
τ = shear stress
φp = curvature corresponding to θp
φs = curvatures corresponding to θs
ϕ = stiffness factor
xxviii
ACKNOWLEDGMENTS
The project was funded by the National Science Foundation (NSF) under Grant No. CMS
98-74872 as part of the 1999 CAREER Program. The support of the NSF Program Directors Dr. S.
C. Liu and Dr. P. Chang is gratefully acknowledged. The authors thank Professor R. Sause of
Lehigh University for his comments and suggestions. The opinions, findings, and conclusions
expressed in this report are those of the authors and do not necessarily reflect the views of the
individuals and organizations acknowledged above.
xxix
CHAPTER 1
INTRODUCTION
In current practice, the design lateral strength, Fdes, is often determined by dividing the
design lateral force required to keep the structure linear-elastic during an earthquake, Felas, by a
response modification coefficient, R = Rdes, as shown in Figure 1.1a. This force reduction is
allowed provided that the resulting maximum nonlinear displacement demand, ∆nlin, can be
accommodated. The maximum displacement, ∆nlin, depends on the R coefficient used in design
and can be estimated from simple capacity-demand index relationships based on an idealized
bilinear lateral force-displacement relationship of the structure as shown in Figure 1.1a. The ben-
efit of using these capacity-demand index relationships comes from their simplicity, however sig-
nificant deficiencies exist in their development, which are addressed by this research as follows.
1
Felas
CHANGE IN LATERAL
DISPLACEMENT
FORCE REDUCTION
2
lateral force, F
Fnlin
αK
Fy = Felas
R
3
Felas
Fdes=
R des
"significant yield" 1 Linear-Elastic Behavior
2 Nonlinear Behavior
0 ∆ ∆ ∆ ∆ nlin= µ∆ y
des y elas
lateral displacement, ∆
(a)
0.6
BOSTON DESIGN-LEVEL SOFT (SE ) SOIL
0.5 Ground motion response
spectrum (ξ = 5%)
Design response spectrum,
0.4 IBC 2000 (ξ = 5%)
W
1
.
R = 1 (linear-elastic)
0.3
F elas
R
0.2
0.1
unconservative inconsistency
in determining Felas / R R=2
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
period, T (sec)
(b)
Figure 1.1: (a) Lateral force-displacement relationships; (b) Ground motion response spec-
tra versus smooth design response spectra.
2
the characteristics of the ground motion spectra may be significantly different than the character-
istics of smooth design response spectra as shown in Figure 1.1b (where W is the total seismic
weight assigned to the lateral load-resisting system). As illustrated in Figure 1.1b, this inconsis-
tency may result in different values for the design lateral force, Felas/R, for the same value of the R
coefficient. In particular, the design force using the smooth spectrum may be lower than the
design force using the ground motion spectrum, leading to an unconservative design.
To the best of the authors’ knowledge, previous research on the development of capacity-
demand index relationships using smooth design response spectra instead of linear-elastic ground
motion response spectra as the basis for calculating Fy = Felas/R has not been published in the lit-
erature. This research shows that previous relationships developed using linear-elastic ground
motion response spectra can be significantly different than those developed using smooth design
response spectra and can lead to unconservative designs, particularly for survival-level, soft soil,
and near-field ground motions.
To the best of the authors’ knowledge, previous research on the development of closed-
form relationships for a comprehensive set of seismic demand indices has not been published in
the literature. This research develops such relationships and shows that the correlation between µ
and the other demand indices is usually relatively strong. In some cases, the cross-correlations
between the demand indices show weak to no correlation, indicating that these demand indices
can carry independent measures of seismic demand.
3
onstrated that some scaling methods result in an excessive scatter in the estimated seismic
demands, indicating that these scaling methods may not be able to adequately define the damage
potential (i.e., seismic intensity) for given site conditions and structural characteristics.
For example, nonlinear dynamic analysis procedures have been often performed using
ground motions scaled to constant peak ground motion characteristics (e.g., peak ground acceler-
ation, PGA, and peak ground velocity, PGV). However, it has been shown that this method of
ground motion scaling introduces a large scatter in the analysis results (Nau and Hall, 1984;
Miranda, 1993; Vidic et al., 1994; Shome and Cornell, 1998). This indicates that the seismic
demand estimates may be biased, leading to designs with significant uncertainty and unknown
margins of safety, unless a relatively large ensemble of ground motions are used (Shome and Cor-
nell, 1998). Therefore, it is imperative that nonlinear dynamic analysis procedures are performed
using ground motion records scaled based on methods that adequately define the damage potential
for given site conditions and structural characteristics, thus resulting in consistent prediction of
the demand estimates by minimizing the scatter.
To the best of the authors’ knowledge, an investigation of ground motion scaling methods
incorporating different site conditions and structure characteristics (e.g., site soil characteristics,
epicentral distance, hysteretic behavior), which is one of the main focuses of this research, has not
been previously published in the literature. The research shows that scaling methods that work
well for ground motion records on stiff soil profiles lose their effectiveness for ground motions on
other site conditions, particularly on soft soil profiles and for large R coefficients. A new scaling
method is proposed for use under these conditions.
The broad objective of this research is to address some of the research needs for the imple-
mentation of performance-based procedures in current U.S. seismic design provisions. The
research has four specific objectives: (1) to develop new nonlinear SDOF capacity-demand index
relationships based on linear-elastic smooth design response spectra; (2) to develop new relation-
ships that quantify cumulative damage, hysteretic energy, and residual displacement demands; (3)
to investigate the effects of the structure fundamental period of vibration, strength level, hysteretic
behavior, and site conditions on the demand estimates; and (4) to investigate the effect of the
ground motion scaling method on the scatter in the demand estimates.
Chapter 3 describes the research program in terms of the analytical models, ground
motion records, analysis procedures, demand indices, and statistical evaluation of results. Three
major suites of ground motion records are used in this research: (1) ground motions compiled by
4
the authors at the University of Notre Dame; (2) ground motions compiled by the SAC steel
project (SAC, 1997; Somerville et al., 1997); and (3) ground motions compiled by Nassar and
Krawinkler (1991).
The University of Notre Dame (UND) ground motions are used to investigate the effect of
the ground motion scaling method on the scatter in the demand estimates. The SAC steel project
(SAC) ground motions are primarily used to develop new capacity-demand index relationships,
including relationships based on linear-elastic smooth design response spectra. Finally, the Nassar
and Krawinkler (N&K) ground motions are used to validate the analytical procedure used in this
research by comparing the results with previous results. Within these major suites, the ground
motion records are further subdivided into ensembles with different site soil characteristics, site
seismicities, epicentral distances, and seismic demand levels (earthquake hazard level as defined
in FEMA 356 (ASCE, 2000)).
Chapter 4 presents comparisons between results obtained in this research program and
results presented by Nassar and Krawinkler (1991) as validation for the analytical procedure. In
addition, implications of some of the assumptions used by Nassar and Krawinkler are investi-
gated.
In Chapter 10, the results of the regression analyses in the previous two chapters are inter-
preted for the purpose of producing viable structural designs using performance-based engineer-
ing concepts (e.g., inelastic capacity-demand spectra methods).
5
Chapter 11 presents a comprehensive examination of the effect of several ground motion
scaling methods on the scatter in the demand estimates. The effects of variations in the following
major types of parameters on the effectiveness of the scaling methods are examined: (1) input
parameters such as site soil characteristics and epicentral distance; and (2) structural parameters
such as fundamental period of vibration, hysteresis type, and strength level (i.e., response modifi-
cation coefficient, R). MDOF analyses are conducted to validate the findings from the SDOF anal-
yses.
Chapter 12 summarizes the research and the major conclusions derived from this research.
It is emphasized that this research focuses almost solely on SDOF systems. A total of
300,000 SDOF dynamic time-history analyses were conducted. A limited number of MDOF anal-
yses (a total of 80 analyses) were performed to validate the SDOF analysis results on the scaling
of ground motion records, however this was not done for the full range of analysis parameters
(e.g., reference response spectra, demand indices).
For a wide range of structures, results derived from the SDOF analyses can be directly
applied to current seismic design provisions based on simplifying assumptions for mode shapes
and mass participation (SEAOC, 1996). Implicitly, these design provisions assume that the “glo-
bal” seismic displacement response of the structure (e.g., roof-displacement response) is governed
by the fundamental (first) mode of vibration, making a SDOF representation adequate.
Conversely, the results from this research may not be directly applicable to structures that
are expected to have significant multi-mode effects (e.g., flexible diaphragm structures (Fleis-
chman and Farrow, 2001)) and modifications to the capacity-demand index relationships may be
necessary. Furthermore, local force and deformation demands (i.e., demands in the structural
members and joints) are not addressed in this research.
Nonlinear static and dynamic procedures specified in current seismic design provisions
are popular because of their simplicity and cost-effectiveness. However, the use of these proce-
dures in design may be compromised with unknown levels of protection imposed by the present
code provisions. In some cases, these procedures may lead to significantly unconservative
designs. This research represents an attempt to address some of the present needs for seismic
design by developing capacity-demand index relationships and ground motion scaling methods
that are consistent with current code provisions.
6
CHAPTER 2
BACKGROUND
This chapter provides background information on the following topics: (1) previous
research on maximum displacement ductility demand relationships; (2) previous research on other
demand indices; (3) current seismic design procedures that use capacity-demand index relation-
ships; and (4) previous research on ground motion scaling methods.
R ( µ, T , α ) = [ c ( µ – 1 ) + 1 ] 1 / c 2.1
where,
F elas ∆ nlin
- , µ = ----------
R = ----------- - 2.2
Fy ∆y
Ta b
c ( T , α ) = --------------
a
- + --- 2.3
T +1 T
Values for the a and b coefficients in Equation 2.3 can be found in Table 2.1 for a given value of α
(Nassar and Krawinkler, 1991).
7
Table 2.1: Values for a and b coefficients
α (%) a b
0 1.00 0.42
2 1.01 0.37
10 0.80 0.29
It should be noted that the definition of the R coefficient in current code provisions (e.g.,
IBC 2000) is based on a point of first “significant yield” in the lateral load resisting system (i.e.,
Fdes, ∆des in Fig. 1.1a) which is different from the R coefficient defined in Equation 2.2 based on a
“global” yield point for the structure (i.e., Fy, ∆y). The R coefficient defined in Equation 2.2 is
used in the remainder of this research since it takes into account the overstrength in the structure
and is considered to result in a better measure of the maximum displacement ductility demand
through µ.
In the development of the above R-µ-T relationship, the lateral strength, Fy = Felas/R was
determined from the 5%-damped linear-elastic SDOF acceleration response spectrum for each
ground motion. As mentioned earlier, this is different from current code provisions where the
design lateral strength is calculated using a linear-elastic smooth design response spectrum.
It can be verified that Equation 2.1 satisfies the following well-established limits (Chopra,
1995): (1) as T → 0, R → 1 (linear-elastic response); and (2) as T → ∞, R → µ (equal displace-
ment assumption). The first limit recognizes that for very short-period (i.e., T < 0.25 sec.) struc-
tures, the maximum displacement ductility demand becomes extremely excessive and
unattainable for relatively small reductions in the lateral force capacity. Therefore, these struc-
tures should be designed to remain linear-elastic (i.e., R = 1). The second limit implies that for
long-period structures, the maximum nonlinear displacement demand, ∆nlin, approaches the max-
imum linear-elastic displacement demand, ∆elas. This result is expected since for very long-period
(i.e., flexible) structures, the maximum displacement demand is similar to the peak ground dis-
placement, independent of R.
It is noted that the R-µ-T relationship defined in Equations 2.1-2.3 was developed using
far-field ground motions recorded at sites with stiff soil, making it likely that the results may not
8
be directly applicable for sites with medium and soft soil or near-field conditions. This is
described next.
The effects of site soil characteristics on the R-µ-T relationship have been reported by
Elghadamsi and Morhaz (1987), Peng et al. (1988), Krawinkler and Rahnama (1992), Miranda
(1993), Rahnama and Krawinkler (1993), and Miranda and Bertero (1994). The studies by
Krawinkler and Rahnama (1992), Miranda (1993), and Miranda and Bertero (1994) suggest that
local site soil characteristics may have a significant effect on the R-µ-T relationship, especially for
sites with soft soil.
Krawinkler and Rahnama (1992) recommend a modification to be used for sites with soft
soil in the form:
F yr, e
F ys ( µ ) = F yr ( µ )S ( T s, µ ) = ------------ S ( T s, µ ) 2.4
R(µ)
where F ys ( µ ) is the required nonlinear (i.e., reduced) lateral force capacity at the soft soil site,
F yr ( µ ) is the required nonlinear force capacity based on the bedrock motion below the soft soil
site, F yr, e is the linear-elastic force demand based on the bedrock motion below the soft soil site,
S ( T s, µ ) is a soft soil modification factor, and Ts is the predominant period of vibration for the
soft soil site. Although an expression is not provided for the soft soil modification factor,
S ( T s, µ ) , it was observed that S ( T s, µ ) is approximately 5 for linear-elastic structures (R = 1)
and 3-4 for nonlinear structures with fundamental periods close to Ts.
µ–1
R̂ µ = ------------ + 1 ≥ 1 2.5
Φ
where Φ is a function of T, µ, and site soil characteristics. Expressions for Φ are given by Miranda
(1993) for rock, alluvium, and soft soil sites as:
1 1
Φ = 1 + ----------------------- – ------- exp ( – 1.5 ( ln ( T ) – 0.6 ) 2 ) for rock sites 2.6
10T – µT 2T
1 2
Φ = 1 + ----------------------- – ------- exp ( – 2 ( ln ( T ) – 0.2 ) 2 ) for alluvium sites 2.7
12T – µT 5T
T 3T s
Φ = 1 + ------s- – --------- exp – 3 ln ----- – 0.25
T 2
for soft sites 2.8
3T 4T T s
9
Figure 2.1 shows the R-µ-T relationships for rock and soft soil sites using Equations 2.5,
2.6, and 2.8. From this study it was determined that: (1) the use of R coefficients developed for
sites with rock or alluvium can lead to unconservative designs (due to larger R coefficients, and
thus, smaller lateral strengths) for short-period (i.e., shorter than two-thirds of the site soil pre-
dominant period, Ts) structures located on sites with soft soil; and (2) the R coefficients developed
for sites with soft soil are much larger than the displacement ductility ratio, µ, for structures with
periods near the site soil predominant period (0.75Ts ≤ T ≤ 1.5Ts). The second observation above,
which implies that smaller lateral strengths are needed for structures with periods near the site
predominant period, is probably because the lateral force capacities, Fy = Felas/R, of the systems
in Miranda’s study were determined based on the linear-elastic acceleration response spectrum for
each ground motion. As shown in Figure 1.1b and discussed in more detail in Chapter 6, this
results in increased Fy values for structures with periods close to the site predominant period,
leading to the reduced seismic demands for soft soil profiles.
The occurrence of recent earthquakes close to heavily urbanized areas (e.g., 1994
Nothridge, California; 1995 Kobe, Japan; 1999 Izmit, Turkey; 1999 Chi-Chi, Taiwan; 2001
Gujarat, India) has shown that severe damage can occur to the built environment due to near-field
ground motions. The effects of near-field ground motions on seismic demands have been exam-
ined by several researchers (e.g., Hall et al., 1995; Naeim, 1995; Makris, 1997; Attalla et al.,
1998; Bozorgnia and Mahin, 1998; Malthora, 1999). These studies suggest that near-field ground
motions may impose large ductility demands and residual displacements (Attalla et al., 1998),
particularly for flexible and base-isolated (i.e., long-period) structures (Hall et al., 1995; Makris,
1997). However, closed-form R-µ-T relationships are not available in the literature.
2/3Ts Ts
9
5
R
3
µ = 2 ,3, 4, 5, 6 (thin → thick lines)
2 rock site
soft soil site
1
0 1 2 3
T (sec)
Figure 2.1: R-µ-T relationships developed by Miranda (1993) for rock and soft soil sites.
10
2.1.3 Hysteretic lateral load-displacement behavior
Foutch and Shi performed a parametric study on MDOF models of three-, six-, and nine-
story structures which were designed for Los Angeles using the 1994 NEHRP provisions (BSSC,
1994). They concluded that the hysteretic behavior has a small effect on the maximum displace-
ment demand (defined as the maximum story displacement in the structure) with the exception
that the maximum displacement of systems with pinched (i.e., Types 5 and 6) and bilinear-elastic
(i.e., Type 8) hysteresis types can be up to 40 percent larger than the maximum displacement of
systems with a bilinear elasto-plastic hysteresis type (i.e., Type 1). Note that the results in Foutch
and Shi’s study are presented for an R value of 8 only.
Nassar and Krawinkler (1991) conducted SDOF nonlinear dynamic time-history analyses
using two hysteresis types, namely, the bilinear elasto-plastic (EP) hysteresis type and the stiff-
ness degrading (SD) hysteresis type (i.e., Types 1 and 3 in Fig. 2.2), however, the R-µ-T relation-
ship in Equation 2.1 was developed only for the EP hysteresis type. They observed that, except for
very short-period systems, the SD hysteresis type allows larger R coefficients (i.e., larger force
reduction) for a given µ than the EP hysteresis type for systems with no post-yield stiffness (i.e., α
= 0.0). This interesting behavior can be attributed to the larger self-centering capability (i.e., abil-
ity of the system to return toward zero displacement upon unloading from a nonlinear displace-
ment) of the SD hysteresis type as compared with the EP hysteresis type for α = 0.0. It is also
noted that even though the inelastic energy dissipation capacity of the EP hysteresis type is larger
than that of the SD hysteresis type during large displacement cycles, the SD hysteresis type con-
tinues to dissipate energy during small displacement cycles whereas the EP hysteresis type does
not dissipate any energy unless yielding occurs.
F F F F
10% 10%
∆ ∆ ∆ ∆
F F F F
10%
10% 40%
∆ ∆ ∆ ∆
10%
11
Similar studies by Gupta and Krawinkler (1998) and Gupta and Kunnath (1998) used
ground motions recorded on stiff soil sites from the western United States to examine the effect of
the hysteretic lateral load-displacement behavior on the maximum displacement demand of SDOF
structures. It was observed that for structure periods longer than 0.5 seconds, the effect of hyster-
etic behavior on the maximum displacement demand is small except for systems with a negative
post-yield stiffness ratio, α (which may occur due to P-∆ effects).
MacRae and Kawashima (1997) performed a study on the permanent (i.e., residual) dis-
placement demand, ∆r, for elasto-plastic SDOF systems with varying post-yield stiffness ratios, α
(see Fig. 1.1a). This demand index, which is shown in Figure 2.3, is important because even if a
structure survives a severe seismic event without collapse, large residual displacements may ren-
der the structure irreparable, leading to eventual demolition. MacRae and Kawashima (1997) con-
cluded that: (1) the residual displacement demand is almost completely independent of
earthquake magnitude, epicentral distance, and site soil conditions; and (2) the residual displace-
ment demand depends significantly on the post-yield stiffness ratio, α.
Current seismic design procedures that use capacity-demand index relationships include:
(1) the equivalent lateral force procedure; and (2) the capacity spectrum procedure. These proce-
dures are described below.
According to the IBC 2000 (ICC, 2000) design provisions, all “regular” structures up to 70
m high may be designed using the ELF procedure. The design structure base shear, V (which is
equal to Fdes in Fig. 1.1a), is specified as:
12
Table 2.2: Demand indices introduced by Mahin and Lin (1983).
Demand Index Description
∑ ∆ p, -i
µ p = ---------------- cumulative plastic deformation ductility
∆y
F i + F i + 1
-----------------------
∑ 2 - ( ∆i + 1 – ∆i ) normalized hysteretic energy
µE = ----------------------------------------------------------------- + 1
F y∆y
ny number of yield events
nyrev number of complete yield reversals
nzero number of zero crossings
F
∆ p,1
(F , ∆ ) F
yield event (F , ∆ )
y y (F ,∆ ) y y
max max
(F , ∆ ) unloading
i+1 i+1
(F , ∆ )
i i
∆ ∆
0 0
zero crossing
(F ,∆ ) complete yield
min min
∆ p,2 reversal
∆ max
∆ 0
∆r
∆ min
time, t
Figure 2.3: Bilinear elasto-plastic (EP) hysteresis type with definitions for demand indices
in Table 2.2.
13
V = CsW 2.9
where Cs is the seismic response coefficient determined using a smooth design response spectrum
and a response modification coefficient, R, and W is the seismic weight. The value of the R coeffi-
cient ranges from a low of 1.25 for structural systems with a limited amount of ductility and
redundancy (e.g., ordinary steel moment frames in inverted pendulum systems) to a high of 8 for
structural systems with a high level of ductility and redundancy (e.g., special reinforced-concrete
moment resisting frames) (ICC, 2000).
As noted earlier, the calculation of V is based on a point of significant yield in the lateral
load resisting system. The term “significant yield” is defined as that level causing complete plasti-
fication of at least the most critical region of the structure (ICC, 2000, e.g., formation of a first
plastic hinge in the structure). In the case of a reinforced concrete frame, the level of significant
yield is reached when at least one of the sections of its most highly stressed member reaches its
strength.
The base shear, V, is used to prescribe an equivalent static lateral force distribution along
the height of the structure. The ELF procedure is a “force-based” design procedure where a linear-
elastic analysis of the lateral load resisting system is conducted under the equivalent lateral forces.
The elastic deformations calculated at this reduced force level are then amplified to account for
the expected nonlinear behavior of the structure. The structural members are designed and
detailed for the reduced forces from the linear-elastic analysis and the amplified nonlinear defor-
mations.
The ELF procedure does not explicitly consider the inherent relationship between the R
coefficient and the µ demand. Recognizing that nonlinear-inelastic procedures may provide a bet-
ter basis for seismic design, the capacity spectrum procedure, which was first proposed by Free-
man et al. (1975), is specified in the ATC 40 (ATC, 1996) and FEMA 356 (ASCE, 2000)
provisions for the seismic evaluation and rehabilitation of existing structures.
By graphical implementation, the capacity spectrum procedure compares the lateral force
and displacement capacity of a structure with estimated demands from an earthquake ground
motion (Freeman, 1998; Reinhorn, 1997; Chopra and Goel, 1999). The design seismic demand
can be represented using two different approaches: (1) highly-damped linear-elastic response
spectra; and (2) inelastic response spectra. As described below, the first approach does not require
an R-µ-T relationship, whereas an R-µ-T relationship is necessary in the second approach.
(1) In the first approach (Fig. 2.4a), the ground motion demand spectra are constructed by
plotting the linear-elastic SDOF acceleration response spectra, Sa, versus the displacement
response spectra, Sd, for different values of viscous damping ratio, ξ. Radial lines (originating
from “0”) in Figure 2.4a represent constant values of Sa/Sd which are related to the period, T, as
follows:
14
A
Capacity curve Sa Capacity curve
acceleration, S a
acceleration, S a or a*
A’ A’’
a*y Sai
0 ∆* ∆* 0 S S S
y nlin de d di
displacement, S d or ∆ * displacement, S d
(a) (b)
Figure 2.4: Capacity spectrum procedures: (a) highly-damped linear-elastic demand spec-
tra; (b) inelastic demand spectra.
S 2
4π
----a- ≅ --------
- 2.10
Sd T 2
The lateral force and displacement capacity of the structure is represented using a global
force-displacement (F-∆) relationship obtained from a nonlinear-static pushover analysis (curve
“2” in Fig. 1.1a). Assuming that the global seismic displacement response of the structure is gov-
erned by the fundamental (first) mode of vibration, the pushover curve is converted into an ideal-
ized equivalent SDOF acceleration versus displacement (a*-∆*) relationship as follows:
* F * ∆
a = ---------- , ∆ = --- 2.11
m Γ
* Γ
where m* is the mass of the equivalent SDOF system and Γ is the modal participation factor
(Chopra, 1995).
The a*-∆* relationship (i.e., capacity curve) is plotted together with the Sa-Sd demand
spectra as shown in Fig. 2.4a. The design approach requires an iterative procedure based on an
assumed initial value of the maximum nonlinear displacement, ∆*nlin, as described by Freeman
(1998). The assumed displacement is used to estimate an equivalent viscous damping ratio, ξeq,
based on the expected inelastic energy dissipation of the structure. Then, a new value of ∆*nlin is
estimated from the intersection point of the capacity curve and the demand spectrum correspond-
ing to ξeq. The procedure is repeated iteratively until the ∆*nlin value from the intersection of the
capacity curve and the demand spectrum (for ξeq) matches the ∆*nlin value used to estimate ξeq.
15
The intersection point between the capacity curve and the demand spectrum represents the period
of an equivalent linear-elastic SDOF system, Teq, as shown in Figure 2.4a.
It is noted that the use of highly-damped linear-elastic response spectra to estimate ∆*nlin
poses some difficulties as follows: (1) there is no significant evidence that justifies the existence of
a stable relationship between ξeq and the inelastic energy dissipation corresponding to a maxi-
mum nonlinear displacement, ∆*nlin, especially for highly nonlinear systems; and (2) Teq may
have little to do with the dynamic response of the nonlinear system (Krawinkler, 1994).
(2) To overcome the difficulties associated with the approach above, a second approach
(Fig. 2.4b) was recommended by Reinhorn (1997) and Chopra and Goel (1999). This approach
uses inelastic demand spectra developed from linear-elastic smooth design response spectra with
established R-µ-T relationships. Essentially, the same procedure as above is used to develop the
capacity curve for the structure, however in contrast, the demand spectra are constructed from
inelastic SDOF acceleration and displacement spectra for different values of R or µ. In general,
inelastic demand spectra are expected to provide better design estimates than highly-damped lin-
ear-elastic demand spectra, especially for high levels of ductility demand, µ (Reinhorn, 1997).
The inelastic demand spectra are constructed by first dividing the linear-elastic accelera-
tion demand, Sa, with the R coefficient to determine the inelastic acceleration demand (see line
“1” in Fig. 2.4b) as:
S
S ai = ----a- 2.12
R
Then, the inelastic displacement demand, Sdi, is determined by multiplying the linear-elastic dis-
placement demand, Sde (corresponding to Sai), with the displacement ductility demand, µ (see line
“2” in Fig. 2.4b) as:
2
S di = µS de = µ ------ S ai
T
2.13
2π
Thus, point A on the linear-elastic demand spectrum in Figure 2.4b is transformed to point A” on
the inelastic demand spectrum through path A-A’-A”.
Either R or µ can be specified to construct the inelastic demand spectra for use within the
framework of a “force-based” or “displacement-based” design procedure. In the displacement-
based design procedure, a target displacement ductility, µ = µt, is specified for Equation 2.13 and
the corresponding R coefficient to be used in Equation 2.12 is calculated using an assumed R-µ-T
relationship with the linear-elastic structure period, T, and the specified target displacement ductil-
ity, µt. The inelastic demand spectrum shown in Figure 2.4b is constructed using this procedure.
Inversely, in the force-based procedure, an R coefficient is specified for Equation 2.12 and the cor-
responding displacement ductility demand, µ, for Equation 2.13 is calculated using an R-µ-T rela-
tionship based on the linear-elastic structure period, T, and the specified R coefficient.
16
The point at which the capacity curve and a demand spectrum intersect in Figure 2.4b
defines the seismic demand on the structure, as in the first approach. If the acceleration demand,
Sai, is less than or equal to the acceleration capacity, a*y, and the displacement demand, Sdi, is less
than or equal to the displacement capacity, ∆*ult, then the current design of the structure is ade-
quate. Otherwise, redesign is required. Note that in this approach neither an equivalent viscous
damping ratio, ξeq, nor an equivalent period, Teq, of a linear-elastic SDOF system is specified.
There are a number of papers that mention or explicitly address the scaling of ground
motion records for seismic design and analysis (Nassar and Krawinkler, 1991; Miranda, 1993;
Kennedy, 1984; Arias, 1969; Carballo and Cornell, 1998; Shome and Cornell, 1998; Shome et al.,
1998; Martinez-Rueda, 1998; Nau and Hall, 1984; Vidic et al., 1994). Scaling, in general, is per-
formed: (1) to match a parameter associated with the expected damage potential for a given site;
and/or (2) to match smooth design response spectra (Nassar and Krawinkler, 1991). In the study
by Nassar and Krawinkler (1991) to determine capacity-demand index relationships between the
response modification coefficient, R, the maximum displacement ductility demand, µ, the post-
yield stiffness ratio, α, and the period of vibration, T, the ground motions were scaled to a peak
ground acceleration (PGA) of 0.4g. A large scatter in the results was observed, particularly for
large values of the R coefficient. Miranda (1993) performed a similar study and observed that
using acceleration parameters (i.e., PGA and “effective” peak acceleration, EPA) to scale the
ground motions increases the scatter in the inelastic acceleration demand spectra (calculated for
constant values of µ) for longer-period structures.
Shome and Cornell (1998) and Shome et al. (1998) found that seismic demand estimates
are strongly correlated with the linear-elastic spectral response acceleration at the structure funda-
17
mental period, To, also called the spectral intensity, Sˆa ( T o ) . This observation was made for two
structures (To = 0.25 and 1.05 sec.) using ground motions representative of records on soil profile
S2 (stiff, per UBC 1994 (ICBO, 1994)) in California with similar magnitudes (measured as the
amount of strain energy released at the source of the earthquake) and similar epicentral distances
from the source of the earthquake. Shome and Cornell (1998) demonstrated that the scatter in the
demand estimates can be significantly reduced by scaling the ground motion records in an ensem-
ble up or down so that their spectral intensities at the structure linear-elastic fundamental period
are equal (i.e., Sˆa ( T o ) scaling method). Scatter can be further reduced by: (1) scaling the ground
motion records based on the average spectral intensity of the ensemble over a range of periods
(Kennedy et al., 1984; Shome and Cornell, 1998; Shome et al., 1998; Martinez-Rueda, 1998); and
(2) scaling at higher levels of damping (typically around 5% to 20% of critical damping)
(Kennedy et al., 1984; Shome and Cornell, 1998). These methods of scaling were found to be bet-
ter methods to define the damage potential (i.e., seismic intensity) for given site conditions and
structural characteristics by reducing the scatter in the demand estimates.
Additionally, for scaling methods based on peak ground motion characteristics (e.g.,
PGA), scatter was found to be significantly large (e.g., the scatter, δ, calculated as the standard
deviation of the log demand estimates, which is approximately equal to the coefficient of varia-
tion, COV, was as much as 92% of the mean demand) even for a relatively large sample size (20
ground motions), implying that the demand estimates are subject to significant uncertainty
(Shome et al., 1998). Shome et al. concluded that to obtain an estimate of the mean response, X,
within a certain range, X (e.g., X ± 0.10X or X = 0.10X), with 95% confidence, the number of
ground motions required, neq, can be determined using the following:
2
δ
n eq = 4 -----2- 2.14
X
Thus, the required number of ground motion records to obtain a reasonably good estimate of the
mean response can be significantly reduced by reducing the scatter in the demand estimates.
18
CHAPTER 3
This chapter describes the research program as follows: (1) analytical models; (2) seismic
demand levels; (3) earthquake ground motion ensembles; (4) ground motion scaling methods; (5)
reference response spectra; (6) nonlinear dynamic time-history analysis procedure; and (7) statis-
tical evaluation of the results.
˙˙ ( t ) + c ∆˙ ( t ) + k∆ ( t ) = – m∆
m∆ ˙˙ g ( t ) 3.1
d
2
˙˙ ( t ) + 4π
∆
4π
- ∆ ( t ) = –∆
------ξ∆˙ ( t ) + -------- ˙˙ g ( t ) 3.2
T 2
T
19
Felas
CHANGE IN LATERAL
Linear-Elastic DISPLACEMENT
Behavior
FORCE REDUCTION
1
Idealized
F (1 - )
elas R
Bilinear
lateral force, F
Behavior
Fnlin
Fy =
Felas αk
R
kµ
∆
k
m
0 ∆
y
∆
elas
∆ nlin= µ∆ y
lateral displacement, ∆
Five hysteresis types are considered as illustrated in Figure 3.2. The importance of each
hysteresis type is briefly described as follows:
(1) The linear-elastic (LE) hysteresis type is used to determine the force demand, Felas, for
the structure to remain linear-elastic under a ground motion. These analyses correspond to a
response modification coefficient, R, of 1;
(2) The bilinear elasto-plastic (EP) hysteresis type represents a structure with a high level
of inelastic energy dissipation and small self-centering capability (e.g., a steel structure);
(3) The stiffness-degrading (SD) hysteresis type represents a structure with a modest level
of inelastic energy dissipation and stiffness degradation due to cyclic damage (e.g., a monolithic
cast-in-place reinforced-concrete structure).
The SD hysteresis type is used to represent the behavior of a structure that has a memory
limited to the largest positive and negative displacements from past yield excursions and may be
representative of reinforced-concrete structures in which opening and closing of cracks lead to
pinching in the hysteretic behavior (Mahin and Lin, 1983; Hadidi-Tamjed, 1988; Nassar and
Krawinkler, 1991). Figure 3.3 shows a detailed schematic of the SD type.
The initial linear-elastic stiffness of the SD hysteresis type is equal to k. Yielding during
first loading in the positive direction occurs when Fy is reached (point a in Fig. 3.3). Upon follow-
ing a “yielding” branch with stiffness αk, unloading occurs with a stiffness equal to the initial lin-
ear-elastic stiffness, k, similar to the EP type (line bc). Loading in the reverse (i.e., negative)
direction from point c requires that cracks be closed before yielding can occur at point d. This
20
F F "yielding" "shooting" F "yielding"
branch branch branch
0 ∆ 0 ∆ 0 ∆
F F
0 ∆ 0 ∆
(4) The bilinear-elastic (BE) hysteresis type represents a structure with no inelastic energy
dissipation and a high level of self-centering capability (e.g., a post-tensioned precast concrete
structure as described by Kurama (2001)); and
The BP hysteresis type is constructed by placing the BE and EP hysteresis types in paral-
lel, as illustrated in Figure 3.4. The BE hysteresis type represents the primary system (e.g., precast
concrete structure) and the EP hysteresis type represents the secondary system (e.g., supplemental
energy dissipation system). The structural properties of the BP hysteresis type are defined by set-
21
F "yielding"
(F , ∆ ) branch
y y
αk b
a
k g (F , ∆ )
b b
F
b
ksh,fb=
∆ b- ∆f k
f j
c ∆
∆f o h ∆c
i
k "shooting"
branch
F
y
d ksh,cd =
∆y + ∆ c
e αk
(-F , -∆ )
y y
ting the combined linear-elastic stiffness, ktot, to 175,000 kN/m. The linear-elastic force demand,
Felas, is based on this linear-elastic stiffness. The reduced force, Fy, is likewise based on the result-
ing linear-elastic force demand, Felas (see Fig 3.4). Thus, the BP hysteresis type lateral stiffness
and strength are:
k tot = k be + k ep 3.3
F elas
F y = F be + F ep = -----------
- 3.4
R
where kbe and kep are the linear-elastic stiffnesses of the BE and EP hysteresis types, respectively,
and Fbe and Fep are the yield strengths of the BE and EP hysteresis types, respectively.
The relationships between the lateral stiffnesses and strengths of the BE and EP hysteresis
types are:
k ep = β s k be 3.5
F ep = β r F be 3.6
22
F
Bilinear-Elastic
Fbe α k
be be
k
be
+
Elasto-Plastic
α k =0
ep ep
Fep k
k be
m
ep
k
ep
Bilinear-Elastic/Elasto-Plastic
Felas =
Fy Felas
= α k
R be be 2Fep
k
be
k +k α k +k
be be ep
be ep
∆
∆ ∆ ∆
ep rmax be
where βs and βr are the BP type stiffness and strength ratios, respectively. The inelastic hysteretic
energy dissipation of the BP type can be increased by increasing βr .
From Equations 3.3 through 3.6, the BE type stiffness and strength are:
k tot
k be = ------------------
- 3.7
( 1 + βs )
F elas
F be = ----------------------
- 3.8
R ( 1 + βr )
The EP type post-yield stiffness ratio, αep, is set equal to zero and the BE type post-yield
stiffness ratio, αbe, is prescribed to result in an overall combined system post-yield stiffness ratio
of α, i.e.:
23
α be k be = αk tot = α ( 1 + β s )k be 3.9
thus,
α be = α ( 1 + β s ) 3.10
The increase in strength due to the supplemental energy dissipation system is typically
smaller than the yield strength of the primary system (i.e., Fep < Fbe ⇒ βr < 1). Assuming that βs
> βr, the load-displacement relationship for the BP hysteresis type can be seen in Figure 3.5a. The
behavior of a similar system with βs = βr < 1 is shown in Figure 3.5b.
Notice that for the system with βs > βr (Fig. 3.5a), the combined load-displacement curve
is trilinear. This behavior is typical of a structure with an “early-yield” supplemental energy dissi-
pation system (Kasai et al., 1998). Although the early yielding of the supplemental energy dissi-
pation system is not expected to significantly affect the overall dynamic response of the combined
(i.e., primary plus secondary) system, it does make comparison with the other hysteresis types dif-
ficult since their respective yield points do not match. This is illustrated in Fig. 3.6a. To make a
clearer comparison possible, the alternate system in Figure 3.5b which has a bilinear load-dis-
placement relationship is evaluated. In this case, ∆ep must equal ∆be (see Fig. 3.4) and, thus, it can
α(1 + βs)k
F α(1 + βs)k F F
be
be
β
(βr F , r ∆ ) ((1 + βr)F , ∆ )
be β be be be
s
(F , ∆ ) k
be be be
∆ + ∆ = ∆
0 0 0
k
be β k be (1 + βs)k
s be
Bilinear-Elastic/
Bilinear-Elastic (BE) Elasto-Plastic (EP) Elasto-Plastic (BP)
(β > β , β < 1)
s r r
(a)
α(1 + βs)k
F α(1 + βs)k F F be
be
(F , ∆ ) β
be be
(βr F , r ∆ ) ((1 + βr)F , ∆ )
be β be be be
s
∆ + ∆ = ∆
0 k 0 0
be
(1 + βs)k
β k be be
s
Bilinear-Elastic/
Bilinear-Elastic (BE) Elasto-Plastic (EP) Elasto-Plastic (BP)
(β = β < 1)
s r
(b)
Figure 3.5: BP hysteresis type with: (a) βs > βr, βr < 1; (b) βs = βr < 1.
24
F F
EP type EP type
∆ ∆
(a) (b)
Figure 3.6: EP hysteresis type versus BP hysteresis type: (a) βs > βr; (b) βs = βr (matching
yield point).
be shown that βs must equal βr . This constraint requires βs to be less than 1.0 since βr is assumed
to be less than 1.0. The resulting system is compared to the EP hysteresis type in Figure 3.6b.
The BP hysteresis type improves upon the BE hysteresis type by increasing the inelastic
hysteretic energy dissipation of the system. However, this increase in the inelastic energy dissipa-
tion is obtained at the expense of the system’s inherent self-centering capability. It can be shown
that the resulting maximum possible residual displacement, ∆rmax (see Fig. 3.4), of the BP hyster-
esis type is equal to βr∆be. As recommended by Kurama (2001), βr is limited to ≤ 0.5 in this
research to ensure that the residual displacements after a ground motion event are small.
Two MDOF models are considered as shown in Figure 3.7. These models represent four-
story and eight-story cast-in-place reinforced concrete special moment-resisting office building
frames, designed according to the UBC 1997 equivalent lateral force procedure for a region with
high seismicity (Los Angeles) and for the SE (soft) soil profile. Both structures have identical floor
plans. For each structure, an interior frame in the E-W direction is analyzed assuming that the
floor diaphragms are sufficiently rigid under in-plane forces (Fig. 3.7).
Beam and column member sizes are shown in Figure 3.7. The concrete unconfined com-
pressive strength is 41.4 MPa. The assumed yield strength of the reinforcing steel is 414 MPa.
Structural properties for the four-story and eight-story frames are provided in Table 3.1. Gravity
loads, including the weight of partitions, architectural finishes, utilities, structural members, and
live loads, are provided in Table 3.2. Note that the differences (due to different column tributary
25
roof
(level 8)
7 stories at 3.35m
4 stories at 3.35m = 13.4m
roof
(level 4)
4m
(level 1)
ELEVATION
(a)
N
PLAN
(b) (c)
Figure 3.7: Layout of structural system: (a) elevation; (b) four-story structure; (c) eight-
story structure.
26
heights) between the seismic weights, wx, of the intermediate floors and the seismic weights of the
2nd floor and roof are assumed to be negligible.
The required design base shear capacities of the frames, Fdes (denoted as V in UBC 1997),
defined at first “significant yield” (see Fig. 1.1a) were determined based on the UBC 1997 linear-
elastic smooth design response spectrum with R = 8.5 (ICBO, 1997) using:
2.5C a I CvI
----------------- W ≥ F des = --------
- W ≥ 0.11C a IW 3.11
R RT
where W is the total frame seismic weight (including tributary self-weight and partitions), I is the
seismic importance factor (taken as 1.0), and Cv and Ca are seismic coefficients tabulated in UBC
1997 based on site seismicity and site soil characteristics. For Los Angeles (seismic zone 4 in
UBC 1997) and the SE soil profile, the Cv and Ca coefficients are given as:
where Nv and Na are near-field factors based on the closest distance from the site to a
known seismic source. The structures are assumed to be farther than 10 km from a known seismic
source, thus Nv = Na = 1.0 (ICBO 1997).
Structure Height Length Frame seismic Total frame Fundamental Frame design
weight per floor or seismic weight, W period, T base shear, Fdes
(m) (m) roof, wx (kN) (kN) (sec.) (kN)
27
The design base shear force, Fdes, is distributed over the height of each structure as:
n
F des = F t + ∑ Fi 3.13
i=1
where n is the number of floor and roof levels, Ft is the portion of the design force concentrated at
the top of the structure in addition to Fn (for the roof), and Fi is the lateral force applied at level i.
The concentrated force at the roof, Ft, is calculated as:
The value of Ft need not exceed 0.25Fdes and may be considered zero when T is less than or equal
to 0.7 sec. (which is the case for the four-story structure). The remaining portion of Fdes is distrib-
uted over the height of the structure as follows:
( F des – F t )w x h x
F x = --------------------------------------
n
- 3.15
∑ wi hi
i=1
where wx and wi are equal to the seismic weights at level x and level i, respectively, and hx and hi
are the heights from ground to level x and level i, respectively.
The beams and columns were proportioned based on the UBC 1997 load combinations,
including both gravity and earthquake loads. The load combinations considered for this design
are:
and
where DL is the design dead load, LL is the design live load, f1 is the live load modification factor,
and EL is the design earthquake load. The live load modification factor, f1, depends on the type of
live load and was taken as 0.5 (for miscellaneous live loads (ICBO, 1997)). The load combina-
tions in Equations 3.16-3.17 were applied to each structure to determine the design member force
demands, as illustrated in Figures 3.8a-b. The gravity loads on the beams were assumed to act at a
distance of 1/3L from the column faces, where L is the beam clear span length.
A simple analytical model is developed to conduct nonlinear static and nonlinear dynamic
time-history analyses of each frame. The structures are assumed to be fixed at the column bases.
The initial flexural stiffnesses of the interior columns are assumed to be 100% of the gross-section
28
Figure 3.8c
(a) (b)
node
linear-elastic
beam-column element
rigid end zone
hc zero-length rotational
spring element
fiber beam-column
1.5h c element
(c)
Figure 3.8: MDOF models: (a) four-story elevation; (b) eight-story elevation; (c) close-up
of analytical model.
linear-elastic flexural stiffnesses. The initial “effective” flexural stiffnesses of the exterior columns
and floor and roof beams are assumed to be 80% and 30% of the gross-section linear-elastic flex-
ural stiffnesses, respectively, to account for cracking in the members. An effective T-beam flange
width of 625mm (1/8 the center-to-center span length of the beam (Paulay and Priestley, 1992)) is
included in the calculation of the gross-section flexural stiffness for the beams. The beams and
columns are modeled using linear-elastic beam-column elements in the structural analysis pro-
gram, DRAIN-2DX (Prakash et al., 1993). Shear deformations in the members are included. Full
depth rigid end zones are assumed at the beam-column joints (Fig. 3.8c).
Nonlinear behavior at the beam ends and in the beam-column joints is modeled as concen-
trated plastic hinges at the beam-column joints using zero-length rotational spring elements with
stiffness-degrading properties, as shown in Figures 3.8c and 3.9a (Wu, 1995). The moment-rota-
tion behavior of the springs under positive and negative bending are assumed to be the same. The
29
steel stress-strain
relationship
P 700
Stress (MPa)
M
M-θ
Ms, θs Mp, θp 0
0 0.06
Strain (mm/mm)
θ M
M k s2 fiber
unconfined concrete
stress-strain relationship
k s1
θ 45
0
Stress (MPa)
compression
crushing of
concrete not
tension modeled
0 −3
axis of bending −5 x 10
−4 0 4
Strain (mm/mm)
(a) (b)
Figure 3.9: Element models: (a) beam end rotational spring element; (b) column base fiber
element.
yield moment capacities of the springs, Ms, are set equal to the design beam moment demands,
ignoring any overstrength. It is assumed that the interior and exterior joint springs used in every
four levels of each structure are the same. The largest design beam moment demand is used as the
Ms value for each set of interior and exterior joint springs in the structure as shown in Table 3.3.
The initial stiffness of a beam rotational spring, ks1, is set greater than the flexural stiffness
of the adjacent linear-elastic beam element such that little deformation occurs in the spring before
yielding. This is achieved by assigning a linear-elastic stiffness to the spring as follows:
ϕ4EI
k s1 = ---------------e- 3.18
L
where ϕ is a stiffness factor, E is the concrete Young’s modulus, Ie is the assumed effective
(including the effect of cracking) moment of inertia of the adjacent beam element (i.e., 30% of the
gross-section T-beam moment of inertia), and L is the adjacent clear span length. For this study, ϕ
is taken as 5, resulting in a maximum spring rotation of 0.4% relative to the beam-column joint
(i.e., node) rotation before yielding occurs.
(M p – Ms)
k s2 = -------------------------
- 3.19
( θ p – θs )
where Mp, θp, and θs are defined in Figure 3.9a. The plastic rotation, (θp - θs), assuming an equiv-
alent plastic hinge length of lp (Paulay and Priestley, 1992), can be calculated as:
30
Table 3.3: Yield moment capacities for beam rotational springs
( θ p – θ s ) = ( φ p – φ s )l p 3.20
With the linear-elastic beam elements and the beam rotational springs in series, the follow-
ing expression can be written:
1 1 1
--------------- = -------- + -------- 3.21
α φ EI e EI e K s2
where αφ is the moment-curvature post-yield stiffness ratio and Ks2 is the moment-curvature post-
yield stiffness for the equivalent plastic hinge length as:
(M p – Ms)
K s2 = -------------------------
- 3.22
( φ p – φs )
α φ EI e
K s2 = -------------------
- 3.23
( 1 – αφ )
Thus, using Equations 3.19, 3.20, 3.22, and 3.23, the moment-rotation post-yield stiffness of the
springs, ks2, can be calculated as:
K s2 α φ EI e
k s2 = -------- = ------------------------
- 3.24
lp ( 1 – α φ )l p
The equivalent plastic hinge length, lp, is assumed to be 1.5 times the beam depth and the
moment-curvature post-yield stiffness ratio, αφ, is set at 2%.
31
Yielding in the columns other than at the bases is prevented. Nonlinear behavior at the col-
umn bases is modeled using fiber beam-column elements to account for axial-flexural interaction
(Prakash et al., 1993). The cross-section of each column at the base is modeled with an arrange-
ment of longitudinal “fibers”, as shown in Figure 3.9b. The length of the fiber beam-column ele-
ments is assumed to be 1.5 times the column depth, hc (see Fig. 3.8c). The stress-strain behavior
of each fiber is assigned based on the assumed material properties. The assumed stress-strain rela-
tionships of the reinforcing steel and unconfined concrete used in this study are provided in Figure
3.9b. The tensile strength of the concrete is included, considering the effect of tension stiffening.
Crushing of concrete in compression is not modeled assuming that a sufficient amount of confine-
ment reinforcement is used in the critical regions of the members to ensure ductile behavior. The
number of fibers and the amount of steel reinforcement used in the columns at the base are listed
in Table 3.4.
0.2 0.2
base shear/seismic weight (g)
0.1 0.1
0 0
−0.1 −0.1
−0.2 −0.2
−0.3 −0.3
−0.03 −0.02 −0.01 0 0.01 0.02 0.03 −0.03 −0.02 −0.01 0 0.01 0.02 0.03
roof drift (rad) roof drift (rad)
(a) (b)
Figure 3.10: Normalized cyclic base-shear-roof-drift behavior: (a) four-story frame; (b) eight-
story frame.
32
other than the analyses used in design, the gravity load is assumed to be 100% of the design dead
load, DL, plus 25% of the design live load, LL, with no live load reduction. The lateral load distri-
bution over the height of the frames follows the UBC 1997 equivalent lateral force pattern given in
Equation 3.15 (ICBO, 1997). P-∆ effects are considered in the analyses of the frames.
It is noted that it may be possible to develop more accurate analytical models for the struc-
tures, but this was not done to keep the computational time manageable for the large number of
ground motion records considered in the research.
The primary objective of the recent seismic design provisions in IBC 2000 is to present
criteria for the seismic design of building structures to minimize the risk to life (ICC, 2000;
BSSC, 1998). To satisfy this objective, two seismic demand levels are specified in these provi-
sions.
The seismic demand levels in IBC 2000 are defined by three characteristics (Hart and
Wong, 2000): (1) design life; (2) probability of exceedance; and (3) return period. Design life, or
exposure time, of the structure is a best estimate at the time of the design of the structure of how
long the building will be operational before demolition or seismic upgrading. Probability of
exceedance is the chance of a particular level of earthquake being exceeded during the design life
of the structure. Return period is the expected number of years between occurrences of a particu-
lar level of earthquake.
Predictions for possible future earthquake occurrences are often modeled with a Poisson
distribution (Ang and Tang, 1975):
t des
x – -------
T
t-------
des
- e r
Tr
p x = -------------------------------
- , x = 0, 1, 2, . . . 3.25
x!
where Tr is the return period, tdes is the design life, and px is the probability of having x occur-
rences during the design life, tdes (Hart and Wong, 2000). Using Equation 3.25, the probability of
non-occurrence can be found by setting x equal to 0:
t des
– -------
Tr
px = 0 = e 3.26
t des
– -------
Tr
px ≠ 0 = 1 – px = 0 = 1 – e 3.27
33
Thus, for a given expected design life and probability of exceedance, the return period of a partic-
ular earthquake can be calculated by rearranging Equation 3.27:
t des
T r = – -------------------------------
- 3.28
ln ( 1 – p x ≠ 0 )
The first seismic demand level in IBC 2000 is defined using a “maximum considered
earthquake” ground motion which has a 2 percent probability of being exceeded in 50 years (cor-
responding to a return period, Tr, of approximately 2500 years, by substituting tdes = 50 and px ≠ 0
= 0.02 in Eq. 3.28). The maximum considered earthquake ground motion is referred to as the
“survival-level” ground motion herein.
The second seismic demand level specified in the current IBC 2000 provisions is defined
using a “design-level” ground motion. The acceleration response spectrum for the design-level
ground motion is determined by multiplying the spectrum for the survival-level ground motion by
a factor of 2/3 (referred to as the “seismic margin” in the NEHRP provisions (BSSC, 1998)). This
demand level roughly corresponds to an earthquake with a 10 percent probability of being
exceeded in 50 years, or a return period of approximately 500 years, for coastal California and a
lower probability of occurrence (a return period of approximately 1400 years) for the eastern
United States (BSSC, 1998).
To derive clear conclusions and unbiased results from this research, it is necessary to
appropriately select and group ground motion records considering the following parameters: (1)
site soil characteristics; (2) seismic demand level; (3) site seismicity; and (4) epicentral distance.
The ground motion records used in this research are listed in Tables A.1 through A.13 and their
acceleration time histories are shown in Figures A.1 through A.13 of Appendix A. The ground
motion records include both historic and generated records. A large number of records are used in
order to generate a large statistical data set.
Three major ground motion suites are used in the research: (1) ground motions compiled
at the University of Notre Dame; (2) ground motions compiled by Nassar and Krawinkler (1991);
and (3) ground motions compiled by the SAC* steel project (Somerville et al., 1997).
* SAC is a joint venture funded by the Federal Emergency Management Agency (FEMA) comprised of three
34
Ground motion records for sites with very dense (SC), stiff (SD), and soft (SE) soil profiles
were collected at the University of Notre Dame (UND) to investigate the effect of ground motion
scaling method on the scatter in the seismic demand indices, including the effect of site soil char-
acteristics (Tables A.1 through A.3). The SC, SD, and SE soil profiles correspond to site classes C,
D, and E in IBC 2000 (ICC, 200), respectively. Twenty ground motion records were selected for
each soil profile. Detailed site soil properties (e.g., shear wave velocities) were not available for
many recording stations, thus the classification of the ground motion records was based on gen-
eral soil descriptions (e.g., rock, alluvium, soft clay, etc.), which are available for all stations.
Each ground motion record was classified by matching, as close as possible, the general soil
descriptions with the definitions for the site soil characteristics provided in Table 3.5 (ICC, 2000).
Table 3.5: Site soil definitions (adapted from IBC 2000 (ICC, 2000))
The ground motions compiled by Nassar and Krawinkler (N&K) are used as a baseline
ensemble to validate the results in this research and to provide additional results for the very dense
(SC) soil profile. These ground motions include fifteen records (Table A.4) representative of the S1
soil profile (equivalent to the SC soil profile in IBC 2000) from the western United States.
The ground motion records compiled by the SAC steel project (Tables A.5 through A.13)
are primarily used to investigate the effect of the reference response spectra on the R-µ-T relation-
ships and to generate new relationships for other seismic demand indices. These ground motions
are categorized by: (1) site soil characteristics (stiff soil and soft soil, i.e., SD and SE soil profiles
in IBC 2000); (2) seismic demand level (design-level and survival-level); (3) site seismicity (low
seismicity and high seismicity, i.e., Boston and Los Angeles); and (4) epicentral distance (near-
field, NF, and far-field).
The ground motion records in Tables A.5, A.6, A.9, and A.10 were scaled by the SAC
steel project to match (at selected periods of 0.3, 1, 2, and 4 sec.) linear-elastic smooth design
acceleration response spectra representative of each ground motion ensemble in the tables listed
above (BSSC, 1998; ICC, 2000; Somerville et al., 1997). The shapes of the acceleration response
spectra of the individual ground motion records were not modified in the scaling procedure.
Instead, this scaling was done by finding a single factor for each ground motion record that mini-
mized the weighted sum of the square error between the linear-elastic ground motion response
35
spectrum and the corresponding smooth design response spectrum for 5% damping. The weights
used were 0.1, 0.3, 0.3, and 0.3 for T = 0.3, 1, 2, and 4 sec., respectively.
The design-level soft (SE) soil ground motions in Tables A.7 and A.11 were generated by
the SAC steel project using site response analyses conducted on different soil profiles broadly rep-
resentative of SE soil conditions in Boston and Los Angeles (Somerville et al., 1997). The site
response analyses were conducted using the equivalent linear analysis computer program
SHAKE91 (Idriss and Sun, 1992). The design-level ground motions for soft (SE) soil were gener-
ated using the SAC Boston and Los Angeles design-level ensembles for stiff (SD) soil as the bed-
rock motions.
The ground motions in Table A.13 were selected by the SAC steel project to represent
near-field (NF) ground motions from earthquakes having a variety of faulting mechanisms (strike-
slip, oblique, and thrust) with magnitudes ranging from 6 3/4 to 7 1/2 and epicentral distances
ranging from 0 to 18 km (Somerville et al., 1997). The ground motions were not scaled to match a
smooth design response spectrum.
The SAC ground motion ensembles compiled by Somerville et al. (1997) do not include
survival-level ground motions on soft (SE) soil for sites with low and high seismicities. Thus,
these ground motions were augmented with two ground motion ensembles (Tables A.8 and A.12)
generated at the University of Notre Dame following the same procedure used by Somerville et al.
(1997). The additional ensembles are intended to be representative of survival-level ground
motions on SE soil for Boston and Los Angeles.
The site response analyses were conducted using the equivalent linear analysis computer
program EERA, which has been shown to compare well with the SHAKE91 program used by the
SAC steel project (Bardet et al., 2000). The equivalent linear approach consists of performing lin-
ear dynamic time-history analyses by approximating the expected nonlinear hysteretic shear
stress-strain behavior of the soil, shown in Figure 3.11a, as an equivalent linear system, shown in
γ γ° τ
G Gγ
G
0 shear strain, γ ηγ°
η
(a) (b)
Figure 3.11: Equivalent linear model for the site response analyses: (a) assumed nonlinear
hysteretic stress-strain behavior of the soil; (b) equivalent linear model.
36
Figure 3.11b. In this procedure, the shear stress, τ, depends on the shear strain, γ, and the shear
strain rate, γ° , as follows:
τ = Gγ + η γ° 3.29
where G is the equivalent linear shear modulus and η is the viscosity (Fig. 3.11b). The equivalent
linear shear modulus, G, is taken as the secant shear modulus as shown in Figure 3.11a. The vis-
cosity, η, is related to G and the equivalent linear damping ratio, ξ:
2ξG
η = ----------- 3.30
ω
where ω is the equivalent linear soil frequency (Bardet et al., 2000). Both the equivalent linear
shear modulus, G, and the equivalent linear damping ratio, ξ, are dependent on the shear strain
amplitude, γc (Fig. 3.11a). The shear strain amplitude, γc, is determined as the maximum shear
strain from equivalent linear site response analyses that are conducted iteratively as described in
Bardet et al. (2000).
The site response analyses were performed based on similar soil conditions as used by
Somerville et al. (1997) in the development of the SAC design-level SE soil ground motion
records for Boston and Los Angeles. A soft soil profile with a unit weight of 19.66 kN/m3, an
average shear wave velocity of 136 m/sec., and a total depth of 30.5 m was used. The shear wave
velocities as a function of depth (Fig. 3.12a) were taken from Somerville et al. (1997), and the
shear modulus reduction factor, G/Gmax, and damping ratio, ξ, as a function of the shear strain
amplitude, γc, for soil with a plasticity index, PI†, of 30 (Fig. 3.12b) were taken from empirical
curves proposed by Vucetic and Dobry (1991). This soil profile is broadly representative of soft
soil conditions encountered in Boston and Los Angeles. The survival-level ground motions for
soft (SE) soil were generated using the SAC Boston and Los Angeles survival-level ensembles for
stiff (SD) soil as the bedrock motions. Tables A.8 and A.12 list the ground motions generated from
the site response analyses using the EERA program.
In total, 255 ground motion records are used in this research. Apart from the ground
motions generated using the EERA program, the acceleration time-history records of the ground
motions were obtained from the National Geophysical Data Center (NGDC) data archive, the
Multidisciplinary Center for Earthquake Engineering Research (MCEER) database, ground
motion records collected by Kurama et al. (1997), and the SAC steel project (Somerville et al.,
1997).
† The plasticity index, PI, is a measure of the consistency, or degree of firmness, of the soil based on the
plastic state, which is between the semisolid and liquid states (Cernica, 1995). PI = 0 corresponds to grav-
els and sands (cohesionless) and PI ≅ 50 corresponds to high plasticity clays (Vucetic and Dobry, 1991).
37
0
1 30
)
−5
max
Shear Modulus Reduction Factor (G/G
G/Gmax
−10
−15
0.5 15
−20
ξ
−25
−30
0 −4 −3 −2 −1 0
01
100 120 140 160 180 10 10 10 10 10 10
Shear Wave Velocity (m/sec) Shear Strain, γ (%)
c
(a) (b)
Figure 3.12: Soil properties for the site response analyses: (a) shear wave velocity; (b)
shear modulus reduction factor and damping ratio.
Properties of the ground motion records, including peak ground acceleration, PGA, and
maximum incremental velocity, MIV, are given in Tables A.1 through A.13. Incremental velocity,
IV, is the area under the acceleration time-history of a ground motion between two consecutive
zero acceleration crossings. The impulsive loading during a ground motion can be calculated by
multiplying the mass, m, with the IV. The maximum IV (i.e., MIV, which is the maximum area
under the acceleration time-history of a ground motion between two consecutive zero acceleration
crossings) may be a better indicator of the damage potential of a ground motion than PGA since it
captures the impulsive characteristics of the ground motion (Kurama et al., 1997).
For a ground motion with a large PGA and a small MIV (short-duration pulse, called an
acceleration spike), most of the applied impulse is absorbed by the structural inertia and minor
deformation, thus damage, occurs in the structure. A ground motion with a moderate PGA and a
large MIV (long-duration pulse) can produce significant structural deformation especially if the
duration of the acceleration pulse is long compared to the fundamental period of the structure. As
can be seen in Tables A.3, A.7-A.9, and A.11-A.13, the ground motions with large acceleration
pulses (i.e., larger values of MIV) are more likely to occur on soft soil and near-field sites (Singh,
1985; Naeim and Anderson, 1993).
38
The periods are exponentially spaced because the variation in the seismic response of SDOF sys-
tems tend to be large at short periods and small at long periods.
The strong motion duration for the ground motions, Dsm, was determined by cutoff meth-
ods and by the root mean square acceleration (RMSA) method (McCann and Shah, 1979). The
cutoff methods define Dsm as the maximum duration for which the acceleration remains larger
than a specified value. In this research, ten percent of PGA and five percent of the gravitational
constant, g, were used as cutoff values. The RMSA method consistently predicted longer strong
motion durations for the ground motion records, and thus was adopted.
The RMSA method makes use of a cumulative root mean square function (CRF), defined
in discrete form as:
∑ a2( t j )
CRF = j=1
-----------------------
- n = 2, …, l 3.31
n–1
where a(tj) is the acceleration record, tj is the time at discretization point j, n is the current discret-
ization point, and l is the total number of discretization points. Figure 3.13 shows a plot of the
CRF function for the 1940 El Centro ground motion record (ELCN) scaled by a factor of 2.0. The
derivatives of the CRF and reverse CRF (where the acceleration time-history is reversed) are
taken as a function of t and the times at which these derivatives become and remain negative are
taken as the final and initial cutoff points, tend and tinit, respectively (see Fig. 3.13). The cutoff
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 10 20 30 40 50 60 0 10 20 30 40 50 60
300 100
CRF (cm/sec )
2
200
50
100
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
10 10
dCRF/dt (cm/sec )
3
5 5
0 0
−5 −5
−10 −10
0 10 20 30 40 50 60 0 10 20 30 40 50 60
t
end t
t (sec) Dsm= tend
- t init t (sec) init
(a) (b)
Figure 3.13: Acceleration time-history, cumulative RMSA function (CRF), and derivative
of the CRF function (1940 El Centro (ELCN) x 2.0): (a) forward CRF; (b) reverse CRF.
39
times for the strong motion duration are taken at these points because these points indicate that the
rate of input energy is decreasing (the slope of the CRF function remains negative), implying that
the remainder of the ground motion record does not contain any high energy pulses.
Seven different ground motion scaling methods are used to investigate the scatter in the
estimated demands from the dynamic analyses. These scaling methods are described below:
(1) Peak ground acceleration (PGA): Each ground motion record is scaled to the average
PGA of the ground motion ensemble. As described earlier, most of the previous research using
nonlinear dynamic analysis procedures is based on this scaling method.
(2) Effective peak acceleration (EPA): Each ground motion record is scaled to the average
EPA of the ground motion ensemble. According to the 1994 NEHRP provisions (BSSC, 1994),
EPA is calculated as the average linear-elastic 5%-damped spectral acceleration for the period
range of 0.1 and 0.5 seconds divided by 2.5. The 2.5 coefficient relates back to the formulation of
the smooth design response spectra in ATC 3-06 (ATC, 1978).
(3) Effective peak velocity (EPV): Each ground motion record is scaled to the average
EPV of the ground motion ensemble. According to the 1994 NEHRP provisions (BSSC, 1994),
EPV is equal to the linear-elastic 5%-damped spectral pseudo-velocity at a period of 1 second. In
this research, the EPV values of the ground motions are calculated as the average spectral pseudo-
velocity for periods between 0.8 and 1.2 seconds as recommended by Kurama et al. (1997).
(4) Maximum incremental velocity (MIV): Each ground motion record is scaled to the
average MIV of the ground motion ensemble.
(5) Arias intensity-based parameter (A95): Each ground motion record is scaled to the
average A95 parameter of the ground motion ensemble. The A95 parameter is defined as the level
of acceleration which contains up to 95 percent of the Arias Intensity, Es (Sarma and Yang, 1987),
which is defined as (Arias, 1969):
t
Es = ∫0 a 2 ( t ) dt 3.32
where a(t) is the acceleration time-history of the ground motion record. The A95 parameter, simi-
lar to MIV, has been shown to be a good measure of the damage potential of a ground motion
record (Sarma and Yang, 1987). Based on a linear regression study performed by Sarma and Yang
(1987), the relationship between the A95 parameter and Es can be approximated as:
or,
40
A 95 = 0.764E s0.438 3.34
(6) Spectral acceleration at the structure fundamental period ( Sˆa ( T o ) ): Each ground
motion record is scaled to the average linear-elastic 5%-damped spectral acceleration of the
ground motion ensemble at the linear-elastic fundamental period of the structure being analyzed,
To. Different from the scaling methods described above, the Sˆa ( T o ) method depends on the struc-
ture properties (i.e., To) as well as the ground motion characteristics. For example, if the structure
has a period of To = 0.18 sec., each ground motion is scaled such that the spectral acceleration at
0.18 sec. is equal to the average spectral acceleration, Sˆa ( T o ) , of the ground motion ensemble at
0.18 sec. as illustrated in Figure 3.14a. The Sˆa ( T o ) parameter is sometimes referred to as the
structure-specific ground motion spectral intensity (Shome and Cornell, 1998).
(7) Spectral acceleration over a range of structure periods ( Sˆa ( T o → T µ ) ): Each ground
motion record is scaled to the average linear-elastic 5%-damped spectral acceleration of the
ground motion ensemble over a range of structure periods. First, the average spectral acceleration,
2
To ξ = 0.05
Sˆ (To)
a
Scaled Acceleration Response Spectra
Sa (g)
1
Average Response Spectrum
0
0 0.5 1 1.5 2 2.5 3 3.5
T (sec)
(a)
2
To Tµ
ξ = 0.05
Sˆ (To→T )
a µ
Sa (g)
0
0 0.5 1 1.5 2 2.5 3 3.5
T (sec)
(b)
Figure 3.14: Scaling based on the spectral acceleration (UND SC soil ground motion
ensemble): (a) at the structure period ( Sˆa ( T o ) method); (b) over a range of structure
periods ( Sˆa ( T o → T µ ) method).
41
Sˆa ( T o → T µ ) , of the ground motion ensemble over the period range T o → T µ is calculated.
Then, the ground motions are scaled such that the average spectral acceleration of each ground
motion over the period range T o → T µ is equal to Sˆa ( T o → T µ ) , as illustrated in Figure 3.14b.
This scaling method takes into account the elongation of the structure period due to nonlinear
behavior. Thus, the period range depends on the amount of nonlinear deformation expected in the
structure. In this research, the elongated period, Tµ, is calculated based on the secant stiffness, kµ,
corresponding to the maximum displacement demand, ∆nlin, as (Fig. 3.1):
F nlin k∆ y + αk ( ∆ nlin – ∆ y )
k µ = ----------- = --------------------------------------------------
- 3.35
∆ nlin ∆ nlin
m
T = 2π ---- 3.36
k
T µ2 k ∆ nlin ∆ nlin µ
------2 = ----- = --------------------------------------------
- = --------------------------------------------------
- = -------------------------- 3.37
To kµ ∆ y + α ( ∆ nlin – ∆ y ) ∆ nlin αµ + 1 – α
∆ y 1 + α ---------- -–1
∆y
µ
T µ = T o -------------------------- 3.38
αµ + 1 – α
Note that the a and b coefficients in Equation 2.3 were developed by Nassar and Krawin-
kler (1991) using nonlinear regression analyses based on far-field ground motions recorded at
sites representative of the SC soil profile for the EP hysteresis type (Table 2.1). Thus, these coeffi-
cients may not be applicable for sites with SD and SE soil profiles and near-field conditions
(Miranda, 1993; Krawinkler and Rahnama, 1992) or for different hysteresis types. New a and b
coefficients were developed in this study for the UND SC, SD, and SE soil ground motions and the
SAC Los Angeles near-field (NF) ground motions (Tables A.1-A.3 and A.13), and for the EP, SD,
BE, and BP hysteresis types using regression analyses similar to those used by Nassar and
Krawinkler (1991). These new coefficients are given in Table 3.6.
42
Table 3.6: Values for a and b coefficients developed in this study
UND SD soil 1.49 0.46 1.60 0.50 2.74 0.76 2.33 0.66
In order to investigate the effect of reference response spectra on the R-µ-T relationships,
three types of linear-elastic acceleration response spectra were used to calculate Fy = Felas/R,
namely: (1) response spectra based on the individual ground motion records (IND); (2) average
ground motion response spectra based on the ground motion ensembles (AVG); and (3) smooth
design response spectra from current seismic design provisions (DES).
In the case of the IND spectra, Fy is calculated as the mass, m, times the linear-elastic
spectral acceleration (at the structure period) of the individual ground motion record divided by R.
The IND spectra for the ground motion records are shown by the thin solid lines in Figures A.14-
A.26.
Similarly, in the case of the AVG spectra, Fy is calculated as the mass, m, times the aver-
age linear-elastic spectral acceleration (at the structure period) of the ground motion ensemble
divided by R. The AVG spectra for the ground motion ensembles are shown by the thick dashed
lines in Figures A.14-A.26.
In the case of the DES spectra, Fy is calculated as the mass, m, times the spectral accelera-
tion (at the structure period) from the design response spectrum divided by R. The general shape
of the design response spectra in IBC 2000 and UBC 1997 is used as shown in Figure 3.15a. The
seismic coefficients Sd1, Sds, Cv, and Ca are mapped or tabulated based on site seismicity, seismic
demand level, site soil characteristics, and epicentral distance. Note that Sd1 and Sds for the sur-
vival-level ground motion correspond to SM1 and SMS for the “maximum considered earthquake”
in IBC 2000, and similarly, Sd1 and Sds for the design-level ground motion correspond to SD1 and
SDS for the “design earthquake” in IBC 2000. Additionally, it is noted that the seismic coefficients
Cv and Ca in UBC 1997 are specified for the design-level ground motion.
Figures 3.15b and c show the smooth design (DES) response spectra used in this research
for sites with low seismicity and high seismicity (i.e., Boston and Los Angeles, respectively) and
43
ξ = 5%
Sds (IBC 2000)
2.5Ca (UBC 1997)
Sa
Cv /T (UBC 1997)
(a)
2.5 2.5
ξ = 5% SD ξ = 5% SD
Boston S Boston
SE E
2 2
SD
SC Los Angeles
SE
SD Los Angeles
1.5 SE 1.5
Sa (g)
Sa (g)
1 1
0.5 0.5
0 0
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
T (sec) T (sec)
(b) (c)
Figure 3.15: Design spectra: (a) general shape; (b) design-level; (c) survival-level.
with design-level and survival-level seismic demands, respectively. The seismic coefficients that
define these response spectra are obtained from the IBC 2000 provisions and are given in Table
3.7. It is noted that the IBC 2000 provisions do not provide design spectra for soft (SE) soil sites in
regions with high seismicity. Thus, the SE soil design spectra for Los Angeles are based on the
UBC 1997 provisions (ICBO, 1997) instead as shown in Figure 3.15 and Table 3.7. Also, the IBC
2000 provisions do not take near-field conditions into account. Thus, the Los Angeles design-
level SD soil DES spectrum in Figure 3.15a and Table 3.7 is used for both far-field and near-field
(NF) conditions.
Comparisons between the average (AVG) response spectra and the design (DES) response
spectra used in this research are provided in Figures 3.16-3.20.
44
Table 3.7: Seismic coefficients for the smooth design (DES) response spectra
Site Seismicity Demand Level Site Soil Sds Sd1
SD 0.31 0.13
Design
SE 0.47 0.19
Boston
SD 0.47 0.19
Survival
SE 0.70 0.28
SC 1.37 0.70
Design SD 1.37a 0.81a
Los Angeles SE 0.90b 0.96b
SD 2.05 1.22
Survival
SE 1.35c 1.44c
a These coefficients are used for both far-field and near-field (NF) conditions.
b 2.5C
a and Cv from UBC 1997 (ICBO, 1997) are used for the SE soil design-level
response spectrum for Los Angeles
c The seismic coefficients for survival-level are determined by multiplying the seis-
3 3
Boston, SD soil, design−level Boston, SD soil, survival−level
AVG spectrum (SAC) AVG spectrum (SAC)
DES spectrum (IBC 2000) DES spectrum (IBC 2000)
2 2
ξ = 0.05 ξ = 0.05
Sa (g)
Sa (g)
1 1
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a) (b)
Figure 3.16: Smooth response spectra: (a) Boston, design-level, SD soil; (b) Boston, sur-
vival-level, SD soil.
45
3 3
Boston, S soil, design−level Boston, S soil, survival−level
E E
AVG spectrum (SAC) AVG spectrum (SAC)
DES spectrum (IBC 2000) DES spectrum (IBC 2000)
2 2
ξ = 0.05 ξ = 0.05
Sa (g)
Sa (g)
1 1
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a) (b)
Figure 3.17: Smooth response spectra: (a) Boston, design-level, SE soil; (b) Boston, sur-
vival-level, SE soil.
3 3
Los Angeles, SD soil, design−level Los Angeles, SD soil, survival−level
AVG spectrum (SAC) AVG spectrum (SAC)
DES spectrum (IBC 2000) DES spectrum (IBC 2000)
2 2
ξ = 0.05 ξ = 0.05
Sa (g)
Sa (g)
1 1
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a) (b)
Figure 3.18: Smooth response spectra: (a) Los Angeles, design-level, SD soil; (b) Los
Angeles, survival-level, SD soil.
3 3
Los Angeles, SE soil, design−level Los Angeles, SE soil, survival−level
AVG spectrum (SAC) AVG spectrum (SAC)
DES spectrum (UBC 1997) DES spectrum (UBC 1997)
2 2
ξ = 0.05 ξ = 0.05
Sa (g)
Sa (g)
1 1
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a) (b)
Figure 3.19: Smooth response spectra: (a) Los Angeles, design-level, SE soil; (b) Los
Angeles, survival-level, SE soil.
46
3 3
Los Angeles, SC soil, design−level Los Angeles, S soil, design−level, NF
D
AVG spectrum (N&K) AVG spectrum (SAC)
DES spectrum (IBC 2000) DES spectrum (IBC 2000)
2 2
ξ = 0.05 ξ = 0.05
Sa (g)
Sa (g)
1 1
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a) (b)
Figure 3.20: Smooth response spectra: (a) Los Angeles, design-level, SC soil; (b) Los
Angeles, design-level, SD soil, NF.
The SDOF and MDOF dynamic time-history analyses conducted in this research are
described in the following sections.
(3) Five hysteresis types described earlier (LE, EP, SD, BE, and BP);
(5) Three BP stiffness and strength ratios, βs = βr = 1/6, 1/3, and 1/2;
(6) Thirty structure periods, exponentially spaced, ranging from T = 0.1 to 3.0 seconds;
47
(7) Seven ground motion scaling methods as described earlier; and
In all, 300,000 SDOF nonlinear dynamic time-history analyses were conducted from com-
binations of the parameters described above. The parameters that are included in the analytical
procedure are provided in Tables 3.8 and 3.9 and shown in Figure 3.21. As shown in Table 3.8, the
SAC ground motion records were primarily used to investigate the effect of the reference response
spectra on the R-µ-T relationships. The N&K ground motion records were used to validate the
CDSPEC program and provide results for the very dense (SC) soil profile. Similarly, as shown in
Table 3.9, the UND ground motion records were used to investigate the effect of ground motion
scaling method on the scatter in the demand indices, including the effect of site soil characteris-
tics. In addition, the SAC near field (NF) ground motion records were used to investigate the
effect of epicentral distance on the scatter in the demand indices. The IND reference response
spectra were used for the analyses listed in Table 3.9.
The MDOF nonlinear dynamic time-history analyses were conducted using the DRAIN-
2DX program (Prakash et al., 1993). A total of 80 analyses were conducted to reinforce the find-
ings from the SDOF analyses to investigate the effect of the scaling method on the scatter in the
demand indices. The UND soft (SE) soil ground motion ensemble was used to excite the MDOF
structures shown in Figure 3.7. These ground motions were first scaled using the MIV method or
the Sˆa ( T o → T µ ) method. Then, the entire ground motion ensemble was scaled so that the aver-
age linear-elastic acceleration response spectrum of the ensemble is not less than 1.4 times the
5%-damped UBC 1997 Los Angeles design-level SE soil design response spectrum for periods
between 0.2To and 1.5To, where To is the structure fundamental period (see Table 3.1), as shown in
Figure 3.22 and required by UBC 1997 (ICBO, 1997). As described in Section 3.1.2, the MDOF
frame structures were designed using this design response spectrum. In the scaling of the entire
ground motion ensemble, the scaling factors required for the eight-story structure are larger than
the scaling factors required for the four-story structure. For simplicity, the scaling factors for the
eight-story structure were used for both structures (scaling factors of 3.19 and 4.11 for the MIV-
based and Sˆa ( T o → T µ ) -based ensembles, respectively).
Note that the shapes of the acceleration response spectra of the individual ground motion
records were not modified in the scaling procedure described above. The average response spec-
trum for the UND soft (SE) soil ground motion ensemble shown in Figure 3.22a is different from
the average response spectrum shown in Figure 3.22b since the Sˆa ( T o → T µ ) method was used in
the latter case. While Figure 3.22a shows the average response spectrum for the ground motion
ensemble scaled to one constant factor using the MIV method, Figure 3.22b shows the average
response spectrum for the ground motion ensemble scaled by a constant factor at each structure
period using the Sˆa ( T o → T µ ) method, resulting in average response spectra that are different
(see definition of the Sˆa ( T o → T µ ) scaling method in Section 3.4).
48
Table 3.8: SAC/N&K ground motion ensembles: parameters studied (shaded areas indicate the N&K ensemble)
Variable Classification
Ref. Response Spectrum Individual Ground Motion Spectra (IND) Average Spectrum (AVG) Smooth Design Spectrum (DES)
Hysteresis Type EP SD BE BP EP SD BE BP EP SD BE BP
α (%) 0 5 10 10 10 10 10 10 10 10 10 10 10 10
βs = βr (%) - - - - - 17 33 50 - - - 33 - - - 33
Site Seismicitya L L L B L B L B L L B L L B L L L L B L L L
Epicentral Distanced F F F F N F F F F F F F F F F F F F F F F F F F F F F F N F F F F F F F F F F F F N F F F F F F F F F F F
Hysteresis Type EP SD BE BP EP SD BE BP EP SD BE BP EP SD BE BP EP SD BE BP EP SD BE BP EP SD BE BP
α (%) 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10
βs = βr (%) 33 33 33 33 33 33 33 33 33 33 33 33 33 33 33 33 33 33 33 33 33 33 33 33 33 33 33 33
Site Seismicitya - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Epicentral Distanced F N F F F F F F N F F F F F F N F F F F F F N F F F F F F N F F F F F F N F F F F F F N F F F F F
CDSPEC
Demand Indices
The seismic mass of each frame was lumped at the beam-column joint nodal degrees-of-
freedom in the horizontal direction based on tributary areas. Gravity loading of 100% of the
design dead load, DL, plus 25% of the design live load, LL, was applied. P-∆ effects were consid-
ered.
Mass and stiffness proportional Rayleigh damping of ξ = 5% was specified for the 1st and
2nd vibrational modes for the four-story structure and the 1st and 3rd modes for the eight-story
structure. As described earlier in Section 3.1.2, the linear-elastic stiffnesses of the beam rotational
springs were set equal to 5 times the linear-elastic stiffnesses of the adjacent beam elements (i.e.,
stiffness factor, ϕ = 5). This introduced a significant increase in the stiffness proportional damping
for the beam rotational springs since the stiffness proportional damping of an element in DRAIN-
2DX is calculated based on the linear-elastic stiffness of the element even after yielding (Prakash
et al., 1993). To reduce this effect, the stiffness proportional damping coefficient for the beam
rotational springs was multiplied by the inverse of the stiffness factor, i.e., 1/ϕ = 0.2.
The statistical evaluation of the demand estimates from the SDOF and MDOF analyses are
described as follows.
51
2.5 2.5
ξ = 5% ξ = 5%
UBC 1997 Los Angeles design-level UBC 1997 Los Angeles design-level
SE soil design response spectrum SE soil design response spectrum
1.4x design response spectrum 1.4x design response spectrum
UND SE soil ground motion ensemble UND SE soil ground motion ensemble
avg. response spectrum (MIV) avg. response spectrum (Sa(To→ Tµ))
3.19x average response spectrum 4.11x average response spectrum
S (g)
S (g)
0.2To,4stry ≤ T ≤ 1.5To,4stry 0.2To,4stry ≤ T ≤ 1.5To,4stry
0.2To,8stry ≤ T ≤ 1.5To,8stry 0.2To,8stry ≤ T ≤ 1.5To,8stry
a
a
0 0
0 To,4stry To,8stry 3.5 0 To,4stry To,8stry 3.5
T (sec) T (sec)
(a) (b)
Figure 3.22: Average response spectra of the ground motions used in the MDOF analyses:
(a) MIV scaling method; (b) Sˆa ( T o → T µ ) scaling method.
Results are presented as mean demand spectra of the maximum displacement ductility, µ,
cumulative plastic deformation ductility, µp, residual displacement ductility, µr = ∆r/∆y, and the
number of yield events, ny, corresponding to different R coefficients. These demand indices are
illustrated in Figure 2.3.
A two-step nonlinear regression analysis, similar to the procedure used by Nassar and
Krawinkler (1991), is performed on the results to develop relationships between R, µ, and T as
given by Equations 2.1-2.3. The first step regression is carried out in the R-µ domain, using the
constant-R procedure as described later in Chapter 4, to relate the period and other parameter
dependencies (i.e., reference response spectra, post-yield stiffness, hysteresis type, and site condi-
tions) to the c coefficient (e.g., see Fig. 3.23a). The second step regression is carried out in the c-T
domain to relate the parameter dependencies to the a and b coefficients (e.g., see Fig. 3.23b).
To better understand the sensitivity of the regression equations to the variation in the a and
b coefficients, c (Eq. 2.3) and µ (Eq. 2.1) are plotted in Figures 3.24 and 3.25, respectively, for a
and b coefficients of 0.1, 1.0, and 2.0.
In Figure 3.24, the c coefficient is plotted as the sum of two separate terms as given in
Equation 2.3:
a
T
c 1 = --------------
a
- 3.39
T +1
52
10 4.5
9 T = 0.92 sec, IND spectra, EP type, α = 0.10 4 IND spectra, EP type, α = 0.10
8 data point
3.5 regression
7
3 a = 1.35, b = 0.44
6
2.5
R
c
c = 0.94 2
4
T = 0.92 sec, c = 0.94
data point, constant-R 1.5
3
mean
2 regression 1
1 0.5
N&K Ground Motion Ensemble N&K Ground Motion Ensemble
0 0
0 5 10 15 0 0.5 1 1.5 2 2.5 3 3.5
µ T (sec)
(a) (b)
Figure 3.23: Regression analysis (IND spectra, EP hysteresis type, α = 0.10): (a) first step,
R-µ domain (T = 0.92 sec.); (b) second step, c-T domain.
b
c 2 = --- 3.40
T
a
T b
c = c 1 + c 2 = --------------
a
- + --- 3.41
T +1 T
c
R –1
µ = -------------- + 1 3.42
c
The values for the c1 term in Figure 3.24a are bounded and have lower and upper limits.
The limit value for c1 as T → ∞ is 1.0, regardless of a. As T → 0, c1 → 0, regardless of a. T = 1
sec. represents a transition point in the equation: for T < 1 sec., c1 is inversely proportional to a;
for T > 1 sec., c1 is directly proportional to a. The coefficient a controls the rate at which c1
approaches 1.0 for T → ∞ and the rate at which c1 approaches 0 for T → 0.
The coefficient c2 in Figure 3.24b has no upper bound as T → 0 and has a lower bound of
0 as T → ∞. Relative to the relationship between c1 and a, c2 is significantly more sensitive to
changes in b especially as T decreases. When the two terms are combined, as in Figure 3.24c, it
can be seen that c is highly sensitive to small changes in b and slightly sensitive to changes in a.
In terms of µ, increasing a while keeping b constant decreases the ductility demand for T <
1 sec. and increases the ductility demand for T >1 sec.; µ is more sensitive to changes in a for
larger values of the R coefficient (Fig. 3.25). Increasing b while keeping a constant results in sig-
nificantly large increases in µ for the entire period range. It can be inferred that b controls the rate
53
1
0.8
c = T /(T + 1)
0.6
a
a
0.4
a = 0.1
1
0.2 a = 1.0
a = 2.0
0
0 1 2 3
T (sec)
(a)
10
b = 0.1
8 b = 1.0
b = 2.0
c = b/T
4
2
0
0 1 2 3
T (sec)
(b)
4
3 a = 2.0
b = 2.0
2
a = 1.0
c=c +c
a = 0.1
2
1
b = 1.0
1 b = 0.1
0
0 1 2 3
T (sec)
(c)
Figure 3.24: Effect of a and b coefficients on c coefficient: (a) c1 term; (b) c2 term; (c)
c = c1 + c2.
at which µ → ∞ for shorter periods. From these observations, intuitive predictions based solely on
comparing a and b values for different parameters can be achieved.
1⁄ f
Λ = d(µ – 1) 3.43
where d and f are the regression coefficients. The regression for cross-correlations between the
demand indices other than µ (i.e., Λ = µp, µr, and ny) is performed using an equation of similar
form:
1⁄h
Λ j = gΛ i 3.44
54
10
R=2
8 R=4
a = 2.0
6 a = 1.0
a = 0.1
µ
4
2
b = 0.1
0
0 1 2 3
T (sec)
(a)
10
R=2
8 a = 2.0 R=4
a = 1.0
6 a = 0.1
µ
2
b = 1.0
0
0 1 2 3
T (sec)
(b)
10
R=2
8 R=4
6
µ
a = 2.0
4 a = 1.0 a = 0.1
2
b = 2.0
0
0 1 2 3
T (sec)
(c)
Figure 3.25: Effect of a and b coefficients on µ: (a) b = 0.1; (b) b = 1.0; (c) b = 2.0.
The regression analyses relate the parameter dependencies (e.g., reference response spec-
tra, post-yield stiffness, hysteresis type, and site conditions) to the a, b, d, f, g, and h coefficients.
The correlation between the demand indices is measured by calculating the correlation coeffi-
cient, ρ (Bendat and Piersol, 1986), defined as:
C ( Λ, µ )
ρ Λµ = ------------------------- 3.45
σ ( Λ )σ ( µ )
for the correlation between µ and the other demand indices, Λ, and:
C ( Λ i, Λ j )
ρ Λ, ij = ----------------------------
- 3.46
σ ( Λ i )σ ( Λ j )
for the cross-correlations between the demand indices, Λ, other than µ. In these equations, σ is the
standard deviation and C is the covariance. This normalized quantity lies between -1 and +1,
55
inclusively. A correlation coefficient magnitude of 1 implies full correlation and a magnitude of 0
implies no correlation.
It is noted that the square of the correlation coefficient, ρ, is commonly referred to as the
coefficient of determination, Ρ, which is used as a measure for the adequacy of the assumed
regression relationship to represent the variability in the data (Montgomery and Peck, 1991).
Since ρ has a range of -1 to +1, inclusively, it follows that Ρ ranges from 0 to 1, inclusively. For
example, values of Ρ that are close to 1 imply that most of the variability in the data is explained
by the regression relationship.
Note that the correlation coefficient, ρ, is only suitable for measuring the amount of the
linear correlation between two variables. Since the relationships between the demand indices are
assumed to be exponential in nature (i.e., Eqs. 3.43 and 3.44), they can be transformed into linear
relationships by taking the log of the equations (Bates and Watts, 1988; Montgomery and Peck,
1991):
1
log Λ = log d + --- log ( µ – 1 ) 3.47
f
1
log Λ j = log g + --- log Λ i 3.48
h
Thus, the correlation coefficient, ρ, is calculated on the log of the data, i.e., µ’ = log(µ-1) and Λ’ =
log Λ. As shown in Equations 3.47-3.48, the correlation coefficient, ρ, depends on the assumed
regression relationship. Using more accurate regression relationships than the relationships used
in this research (which are of simple exponential form) may increase ρ, implying better correla-
tion.
The effect of ground motion scaling method on the scatter in the demand estimates from
the SDOF analyses is investigated using the dimensional (i.e., unit dependent) maximum dis-
placement demand, ∆nlin, and the non-dimensional maximum displacement ductility demand, µ =
∆nlin/∆y. The results are presented as dispersion spectra corresponding to different R coefficients
by calculating the coefficient of variation, COV, defined as the ratio between the sample standard
deviation, σ, and the sample mean. This measure is used to assess the effectiveness of the different
scaling methods in reducing the scatter in the demand estimates.
The evaluation of scatter in the demand estimates from the MDOF analyses is presented
as response profiles of: (1) mean(∆i/∆max) ± σ(∆i/∆max), where ∆i is the lateral displacement at
floor or roof level i, and ∆max is the maximum mean floor or roof displacement, calculated by tak-
ing the mean lateral displacement at each floor or roof level and then taking the maximum value
of the mean lateral displacements over the height of the structure; and (2) mean(θi/θmax) ± σ(θi/
θmax), where θi is the interstory drift at story i:
56
∆i – ∆i – 1
θ i = ----------------------
- , hi = height at level i from ground level 3.49
hi – hi – 1
and θmax is the maximum mean interstory drift, calculated by taking the mean interstory drift at
each story and then taking the maximum value of the mean interstory drifts over the height of the
structure.
Additionally, results are presented as dispersions by calculating the COV of ∆i/∆max and
θi/θmax for the MIV and Sˆa ( T o → T µ ) scaling methods.
57
CHAPTER 4
This chapter provides a verification of the CDSPEC program, which was developed to
conduct nonlinear dynamic time-history analyses as well as pre-processing of the variables and
post-processing of the results. The results from CDSPEC are compared with results using two
analysis packages as follows: (1) nonlinear-dynamic time-history analyses using DRAIN-2DX
(Prakash et al., 1993); and (2) spectral analyses using BISPEC (Hachem, 2000). Furthermore,
comparisons are made with results from Nassar and Krawinkler (1991). Unless otherwise noted,
the lateral force capacities, Fy = Felas/R, of the systems described in this chapter are determined
based on the linear-elastic acceleration response spectrum for each ground motion (i.e., IND spec-
tra).
Figure 4.1 compares the nonlinear dynamic time-history response of two SDOF systems
obtained using CDSPEC and DRAIN-2DX. As examples of representative behavior, the results
are presented for the EP hysteresis type with α = 0.10, R = 8, and T = 0.92 and 3.0 sec.
Figures 4.1a-b compare the responses under the 1965 Puget Sound, Olympia Washington
Highway ground motion (EQ09) and Figures 4.1c-d compare the responses under the 1979 Impe-
rial Valley, Calexico Fire Station ground motion (EQ15). Both ground motions are from the N&K
ensemble. The responses obtained using the two analysis packages are nearly identical, validating
the nonlinear dynamic time-history analysis algorithm used in CDSPEC.
Similarly, Figure 4.2 compares the nonlinear dynamic time-history response for the SD
hysteresis type with α = 0.10, R = 8, and T = 0.50 and 3.0 sec. obtained using CDSPEC and
BISPEC. Figures 4.2a-b compare the responses under the 1989 Loma Prieta, San Francisco-Pre-
sidio ground motion (LPPR) and Figures 4.2c-d compare the responses under the 1966 Parkfield,
Parkfield-Cholame Shandon #2 ground motion (PACH). Both ground motions are from the UND
ensemble. The responses obtained using the two analysis packages are nearly identical, except for
the responses of the T = 3.0 sec. structure subjected to the PACH ground motion after t = 8 sec.
(Fig. 4.2d). The reason for the differences in the two responses is described below.
58
12.5 12.5
EP TYPE T = 0.92 sec EP TYPE T =3.0 sec
∆ (cm)
∆ (cm)
0 0
DRAIN-2DX DRAIN-2DX
CDSPEC N&K: EQ09 CDSPEC N&K: EQ09
−12.5 −12.5
0 5 10 15 20 25 0 5 10 15 20 25
time, t (sec) time, t (sec)
(a) (b)
12.5 12.5
EP TYPE T = 0.92 sec EP TYPE T = 3.0 sec
∆ (cm)
0 ∆ (cm) 0
DRAIN-2DX DRAIN-2DX
CDSPEC N&K: EQ15 CDSPEC N&K: EQ15
−12.5 −12.5
0 5 10 15 20 25 0 5 10 15 20 25
time, t (sec) time, t (sec)
(c) (d)
Figure 4.1: Comparison between CDSPEC and DRAIN-2DX (EP hysteresis type, α =
0.10, R = 8): (a-b) EQ09; (c-d) EQ15.
The SD type hysteretic reloading rules in the BISPEC and CDSPEC analysis packages are
slightly different, as illustrated by the force-displacement responses in Figure 4.2e. After unload-
ing with the linear-elastic stiffness from the “shooting” branch at point a, the structure reloads
back in the negative direction from point b. While the SD type in the CDSPEC program reloads
along the linear-elastic branch towards point a, and then towards the largest displacement in the
negative direction along the shooting branch (line ac), the SD type in the BISPEC program
reloads directly towards the largest displacement along line bc, resulting in a less realistic loop
(Mahin and Lin, 1983).
Comparisons between the DRAIN-2DX and CDSPEC analysis packages for the BE hys-
teresis type with α = 0.10 and the BP hysteresis type with α = 0.10 and βs = βr = 1/3 are shown in
Figure 4.3. Again, the responses obtained using the two analysis packages are nearly identical,
validating the nonlinear dynamic time-history analysis algorithm used in CDSPEC.
59
25 25
SD TYPE T = 0.50 sec SD TYPE T =3.0 sec
∆ (cm)
∆ (cm)
0 0
BISPEC BISPEC
CDSPEC UND: LPPR CDSPEC UND: LPPR
-25 -25
0 5 10 15 20 25 0 5 10 15 20 25
time, t (sec) time, t (sec)
(a) (b)
25 25
SD TYPE T = 0.50 sec SD TYPE T = 3.0 sec
∆ (cm)
0 ∆ (cm) 0
BISPEC BISPEC
CDSPEC UND: PACH CDSPEC UND: PACH
-25 -25
0 5 10 15 20 25 0 5 10 15 20 25
time, t (sec) time, t (sec)
(c) (d)
12000 BISPEC
SD TYPE
CDSPEC
T = 3.0 sec
UND: PACH
F (kN)
0 b
a
different
c reloading rules
-12000
−25 0 25
∆ (cm)
(e)
Figure 4.2: Comparison between CDSPEC and BISPEC (SD hysteresis type, α = 0.10,
R = 8): (a-b) LPPR; (c-d) PACH; (e) different reloading rules.
60
30 30
BE TYPE T = 0.50 sec BE TYPE T =3.0 sec
∆ (cm)
∆ (cm)
0 0
DRAIN-2DX DRAIN-2DX
CDSPEC UND: LPPR CDSPEC UND: LPPR
-30 -30
0 5 10 15 20 25 0 5 10 15 20 25
time, t (sec) time, t (sec)
(a) (b)
30 30
BE TYPE T = 0.50 sec BE TYPE T = 3.0 sec
∆ (cm)
∆ (cm)
0 0
DRAIN-2DX DRAIN-2DX
CDSPEC UND: PACH CDSPEC UND: PACH
-30 -30
0 5 10 15 20 25 0 5 10 15 20 25
time, t (sec) time, t (sec)
(c) (d)
30 30
BP TYPE T = 0.50 sec BP TYPE T =3.0 sec
∆ (cm)
∆ (cm)
0 0
DRAIN-2DX DRAIN-2DX
CDSPEC UND: LPPR CDSPEC UND: LPPR
-30 -30
0 5 10 15 20 25 0 5 10 15 20 25
time, t (sec) time, t (sec)
(e) (f)
30 30
BP TYPE T = 0.50 sec BP TYPE T = 3.0 sec
∆ (cm)
∆ (cm)
0 0
DRAIN-2DX DRAIN-2DX
CDSPEC UND: PACH CDSPEC UND: PACH
-30 -30
0 5 10 15 20 25 0 5 10 15 20 25
time, t (sec) time, t (sec)
(g) (h)
Figure 4.3: Comparison between CDSPEC and DRAIN-2DX (α = 0.10, R = 8): (a-d)
BE hysteresis type; (e-h) BP hysteresis type, βs = βr = 1/3.
61
4.2 Spectral Analyses
The previous section provides a validation of CDSPEC for a limited number of parameters
only. To this end, results from a spectral analysis using CDSPEC under the entire N&K ground
motion ensemble are compared with results obtained using another nonlinear spectral analysis
program, BISPEC (Hachem, 2000).
The results are compared in terms of mean R-µ spectra for the EP hysteresis type with α =
0.10 as shown in Figure 4.4a. The thin to thick lines represent increasing values of R = 1, 2, 4, 6,
and 8. The R-µ spectra generated by CDSPEC are almost identical to the spectra generated by
BISPEC. Similar results are shown for the SD hysteresis type with α = 0.10 in Figure 4.4b. Thus,
CDSPEC is satisfactorily validated.
14 14
R = 1,2,4,6,8 (thin → thick lines) EP type R = 1,2,4,6,8 (thin → thick lines) SD type
12 12
BISPEC BISPEC
CDSPEC CDSPEC
10 10
8 8
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a) (b)
Figure 4.4: Comparison between CDSPEC and BISPEC for the N&K ground motion
ensemble (α = 0.10): (a) EP hysteresis type; (b) SD hysteresis type.
The results are compared with results previously obtained by Nassar and Krawinkler
(1991). Comparisons are made in terms of: (1) constant-R versus constant-µ approaches; and (2)
IND spectra versus smooth response spectra.
Two approaches can be used to obtain relationships between R and µ. In the first approach,
the R coefficient is set to predetermined constant values and the µ demand for each structure
model is determined from nonlinear dynamic time-history analyses. In the second approach, µ is
set to predetermined constant values and the dynamic analyses are conducted to calculate the
62
required R coefficient by using iteration. These two approaches are referred to as “constant-R”
and “constant-µ” approaches, respectively. The constant-R approach is adopted in this research
for reasons described below.
Figure 4.5 shows the R-µ relationship for T = 0.20 sec. under the 1983 Coalinga, Parkfield
Zone 16 ground motion (EQ20). The relationship is determined by using the constant-R proce-
dure for 25 linearly-spaced R coefficients ranging from 1 to 6 (represented by the solid circles).
For this record, it can be seen that the µ demand does not always increase monotonically as the R
coefficient increases. In particular, there is a range where three R coefficients can be found for
each µ (shaded region in Fig. 4.5).
This phenomenon has been reported by several researchers (Newmark and Hall, 1973;
Nassar and Krawinkler, 1991; Miranda, 1993; Chopra, 1995) and must be considered when devel-
oping constant-µ spectra. The smallest R coefficient is taken in these instances, since it corre-
sponds to a larger strength, and thus, is more conservative (e.g., point a, instead of points b or c,
for µ = 5 in Fig. 4.5). However, it can be shown that more conservative results are obtained using
the constant-R approach as follows.
The thick lines in Figures 4.6a and b show the mean R-µ relationships under the N&K
ground motion ensemble (15 ground motions as represented by the thin lines) for T = 0.2 sec. and
0.92 sec., respectively. The thick dashed lines represent the constant-µ relationships with iteration
to determine the smallest required R coefficient, as is done by Nassar and Krawinkler (1991). The
empty circles show the mean R values corresponding to µ = 1, 2, 3, 4, 5, 6, and 8. Upon compari-
6
c
5 constant-R
constant-µ b
4
T = 0.20 sec.
a
R
Figure 4.5: R-µ relationship for the EP hysteresis type with α = 0.00 and T = 0.20 sec.
under the 1983 Coalinga Parkfield Zone 16 ground motion (IND response spectrum).
63
6 6
mean, constant-µ
T = 0.92 sec.
mean, constant-R
5 N&K 1991 5
EQ20
4 4
T = 0.20 sec. 2
1
R
R
3 3
mean, constant-µ
2 2 mean, constant-R
N&K 1991
1 1
(a) (b)
Figure 4.6: R-µ relationships for the EP hysteresis type with α = 0.00 (IND reference
spectra): (a) T = 0.20 sec.; (b) T = 0.92 sec.
son with Nassar and Krawinkler’s results (Nassar and Krawinkler, 1991), represented by the dia-
mond markers in Figure 4.6, a good match is observed.
The thick solid lines in Figure 4.6 show the mean constant-R relationships using 25 lin-
early-spaced R coefficients ranging from 1 to 6. For a given R coefficient, the mean constant-R
relationship consistently predicts larger µ values than the mean constant-µ relationship. The
results indicate that the mean constant-R relationship is highly nonlinear, especially for short-
period structures. In other words, small incremental increases in R can lead to significantly large
increases in µ, especially for large values of R. This large (and non-constant) variance weighs the
mean distribution towards larger values of µ (Lam et al., 1998). On the other hand, the mean con-
stant-µ relationship has relatively less variance, resulting in little or no shift in the distribution.
Therefore, comparison of the mean values with respect to each “variable” (either R or µ) leads to
the constant-R approach as being more conservative.
As an example, point 1 in Figure 4.6a shows that the required R coefficient to limit µ to a
target value of 7 for a structure with T = 0.20 sec. is approximately 3.4. However, if a structure
designed using R = 3.4 is subjected to the same ensemble of ground motions, an average value of
approximately µ = 9 would be observed (point 2). Thus, the constant-R approach is more conser-
vative than the constant-µ approach even when an iteration procedure that selects the smallest R
coefficient is used.
The constant-R approach is adopted in this study since: (1) current design provisions are
based on constant R coefficients; (2) developing constant-R spectra is relatively simple and
requires less computational effort since there is no need for an iterative procedure; and (3) the
constant-R approach is more conservative than the constant-µ approach.
64
4.3.2 IND spectra versus smooth response spectra
Figure 4.7 shows the mean R-µ spectra for the N&K ground motion ensemble (with very
dense, SC, soil profile) using the constant-R approach with the IND, AVG, and DES spectra to
determine Fy = Felas/R. There is a non-negligible increase in µ when the results from the IND
spectra are compared with the results from the two smooth response spectra, especially the AVG
spectrum for the ensemble at long periods (Fig. 4.7a). This increase in µ, together with the
increase in µ due to the constant-R approach, indicates that the use of previously-developed R-µ-T
relationships in design may be unconservative.
Figure 4.7b shows that the increase in µ is not as significant when the DES spectrum for
the ensemble (i.e., IBC 2000 Los Angeles, design-level, SC soil) is used, particularly for T > 1.25
sec. This is because the AVG spectrum has spectral accelerations (thus, results in Fy values) lower
than the DES spectrum for this period range, as shown in Figure 4.8. It seems that a slight
decrease in Fy results in a large increase in the mean µ demand. This is particularly true for the
1979 Imperial Valley Holtville P.O. ground motion (EQ14) of the N&K ground motion ensemble,
as shown by the R-µ relationships in Figure 4.9a. While the DES spectrum (represented by the
dashed line with diamond markers) results in a large increase in µ for EQ14 as compared to using
IND spectra (represented by the dashed line with circle markers), a significantly larger increase in
µ occurs when using the AVG spectrum (represented by the dashed line with square markers).
Thus, the mean µ demand using the AVG spectrum (represented by the solid line with square
markers) is larger than the mean µ demand using the DES spectrum (represented by the solid line
with diamond markers). Once the µ demand for EQ14 is removed from the ensemble, as shown in
Figure 4.9b, a slight decrease in the mean µ demand is observed when using the DES spectrum
(represented by the solid line with diamond markers) and a significant decrease in the mean µ
demand is observed when using the AVG spectrum (represented by the solid line with square
markers). Notice that the mean µ demand for the AVG spectrum is now smaller than the mean µ
demand for the DES spectrum. Furthermore, the differences between the mean µ demands for the
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 IND spectra 12 IND spectra
AVG spectrum DES spectrum
10 10
8 8
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a) (b)
Figure 4.7: R-µ spectra (N&K ensemble, EP hysteresis type, α = 0.10): (a) AVG versus
IND spectra; (b) DES versus IND spectra.
65
2.0
LOS ANGELES DESIGN-LEVEL SC SOIL
R = 1 (linear-elastic)
DES spectrum, ξ = 5%
Sa (g)
AVG spectrum results in lower Fy values
than DES spectrum
0
0 3.5
T (sec)
Figure 4.8: Smooth design (DES) response spectrum versus average (AVG) ground
motion response spectrum.
8 8
IND
7 DES 7
EQ14
6 6 IND
AVG
5 5 AVG DES
R
(a) (b)
Figure 4.9: R-µ relationships using different reference response spectra (EP hysteresis
type, α = 0.10, T = 3.0 sec.): (a) N&K ensemble; (b) N&K ensemble without EQ14.
AVG and IND spectra are smaller than the differences between the mean µ demands for the DES
and IND spectra. This finding, in general, is expected since the AVG spectrum represents the
mean response spectrum of the ground motion ensemble, and thus, is considered to provide a bet-
ter representation of the IND spectra.
66
CHAPTER 5
In this chapter, the effect of hysteretic behavior on the capacity-demand index relation-
ships is investigated in terms of mean spectral values of the demand indices µ, µp, µr, and ny. The
effect of the post-yield stiffness ratio, α, is discussed first. Then, comparisons are given between
the EP hysteresis type and the SD, BE, and BP types. The results are presented for the SAC Los
Angeles survival-level SD (stiff) soil ground motion ensemble. The lateral force capacities, Fy =
Felas/R, of the systems described in this chapter are determined based on the linear-elastic acceler-
ation response spectrum for each ground motion (i.e., IND spectra).
Figure 5.1 compares the mean demand spectra for the EP hysteresis type with α = 0.10
(solid lines) and α = 0.05 and 0.00 (dashed lines). The thin to thick lines represent results for
increasing R coefficients. Both µ and µr demands increase with R. Figure 5.1a indicates that there
is a slight decrease in the µ demand as α increases, more for medium- to short-period structures (T
< 1.0 sec.). Unlike the µ demand, Figure 5.1b shows that the µp demand increases as α increases
for nearly the entire range of periods (T > 0.25 sec). This increase is possibly due to the ability of
the system with a larger α to yield and accumulate plastic deformation in the opposite direction
upon unloading (as a result of the larger amount of elastic energy stored during loading).
In general, the residual displacement ductility demand, µr, significantly decreases with a
small increase in α (Fig. 5.1c). The µr demand is highly erratic at shorter periods (T < 0.5 sec.),
tends to be more uniform at longer periods, and seems to increase as R increases. Figure 5.1c
shows that the dependency of the µr demand on the R coefficient increases as α decreases; this
dependency is smaller at higher levels of R. The number of yield events, ny, increases with R and
is independent of α, as shown in Figure 5.1d.
Figure 5.2 compares the mean demand spectra for the EP (solid lines) and SD (dashed
lines) hysteresis types with α = 0.10. Figure 5.2a indicates that there is a small increase in µ for
the SD hysteresis type. As shown in Figure 3.2, the SD hysteresis type dissipates less energy than
the EP type during large displacement cycles. The increase in µ due to the smaller hysteretic
67
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 EP type, α = 0.10 12 EP type, α = 0.10
EP type, α = 0.05 EP type, α = 0.00
10 10
8 8
µ
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a)
100 100
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10 EP type, α = 0.10
80 EP type, α = 0.05 80 EP type, α = 0.00
60 60
p
µp
µ
40 40
20 20
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(b)
4 4
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10 EP type, α = 0.10
EP type, α = 0.05 EP type, α = 0.00
3 3
r
2 2
µ
1 1
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(c)
50 50
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10 EP type, α = 0.10
40 EP type, α = 0.05 40 EP type, α = 0.00
30 30
y
ny
n
20 20
10 10
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(d)
Figure 5.1: Effect of post-yield stiffness ratio, α (EP hysteresis type, SAC Los Angeles,
survival-level, SD soil): (a) µ; (b) µp; (c) µr; (d) ny.
68
14 100
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 EP type, α = 0.10 EP type, α = 0.10
SD type, α = 0.10 80 SD type, α = 0.10
10
8 60
p
µ
µ
6 40
4
20
2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a) (b)
4 50
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10 EP type, α = 0.10
SD type, α = 0.10 40 SD type, α = 0.10
3
30
µr
ny
2
20
1
10
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(c) (d)
Figure 5.2: EP versus SD hysteresis types (α = 0.10, SAC Los Angeles, survival-level, SD
soil): (a) µ; (b) µp; (c) µr; (d) ny.
energy dissipation of the SD type during large displacement cycles, however, is mitigated by the
additional energy dissipation which occurs in the SD type during small displacement cycles (e.g.,
see cycle g-h-i-j-b in Figure 3.3). As a result, the differences between the µ demands of the EP
and SD hysteresis types are not large.
The µp-spectra (Fig. 5.2b) exhibit considerably large increases for the SD hysteresis type,
especially for smaller values of the R coefficient. This increase in µp occurs as a result of the accu-
mulation of “damage” during the shooting branch of the SD type (Fig. 3.2c). The ratio between
the µp demands for the SD and EP types decreases as the R coefficient increases, in particular, for
systems with large values of α. This is explained as follows. As shown in Figure 5.3a, the behav-
ior of the SD type under large displacement cycles approaches the behavior of the EP type for
large values of α and R, and thus, the SD type spends “less time” in the shooting branch and more
time in the yielding branch during an earthquake. Figure 5.3b shows that the differences between
the SD and EP types are more prominent for smaller values of α. Thus, for small values of α, the
differences between the µp demands for the two hysteresis types are expected to be larger.
The differences in the µr-spectra for the SD and EP hysteresis types are quite erratic but
the two systems follow a similar general trend as shown in Fig. 5.2c. The number of yield events,
ny, for the SD type is smaller and seems to be slightly less dependent on R as compared to the EP
69
17800 17800
yielding branch
EP Type EP Type
SD Type SD Type
yielding branch
shooting branch
F (kN)
F (kN)
0 0
α = 0.10 α = 0.00
T = 0.92 sec T = 0.92 sec
R=8 R=8
SAC Los Angeles, survival-level, S soil: Palos Verdes, LA40 SAC Los Angeles, survival-level, S soil: Palos Verdes, LA40
D D
−17800 −17800
−100 0 100 −100 0 100
∆ (cm) ∆ (cm)
(a) (b)
Figure 5.4 compares the mean demand spectra for the EP and BE hysteresis types with α =
0.10. For the BE hysteresis type, all demand indices, except µr, are significantly larger than those
for the EP type: the µ demand can be 1.75 times as large (Fig. 5.4a), the µp demand can be more
than 4 times as large (Fig. 5.4b), and the ny demand can be 2.25 times as large (Fig. 5.4d). The dif-
ferences in the µ demands for the two hysteresis types remain relatively constant while the differ-
ences in the µp and ny demands decrease as R increases. These results demonstrate the effect of
hysteretic energy dissipation in reducing the seismic demands. The BE type, however, is superior
to the EP type in the residual displacement ductility demand, µr, since the system always returns
to the zero displacement, or plumb, position at the end of the earthquake event, regardless of the R
coefficient or structure period (Fig. 5.4c).
Figure 5.5 compares the mean demand spectra for the EP hysteresis type (solid lines) and
BP hysteresis type with βs = βr = 1/6, 1/3, and 1/2 (dashed lines). As compared to the BE type dis-
cussed in the previous section (Figs. 5.4a and b), the µ and µp demands for the BP type (Figs. 5.5a
and b) are smaller as a result of the increased hysteretic energy dissipation. There is a small
increase in the µr demand (Fig. 5.5c) as compared to the BE type (Fig. 5.4c), however the BP type
is still far more effective in reducing the residual displacements to a negligible level than the EP
type (especially for short periods, i.e., T < 0.5 sec.). The ny demand is slightly decreased for the
BP type (Figure 5.5d), more for smaller R values, as compared to the BE type (Figure 5.4d).
70
14 100
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 EP type, α = 0.10 EP type, α = 0.10
BE type, α = 0.10 80 BE type, α = 0.10
10
8 60
µp
µ
6 40
4
20
2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a) (b)
4 50
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10 EP type, α = 0.10
BE type, α = 0.10 40 BE type, α = 0.10
3
30
r
y
2
µ
n
20
1
10
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(c) (d)
Figure 5.4: EP versus BE hysteresis types (α = 0.10, SAC Los Angeles, survival-level, SD
soil): (a) µ; (b) µp; (c) µr; (d) ny.
As expected, the decrease in the µ, µp, and ny demands for the BP hysteresis type is larger
for larger values of βr due to the increased hysteretic energy dissipation as βr is increased (Fig.
5.6). The amount of hysteretic energy dissipated by the BP type normalized with the amount of
hysteretic energy dissipated by the EP type during a displacement cycle to ± ∆nlin = ± µ∆y can be
expressed as follows:
µ – ---- β r 1 + β s
- --------------
βr β s 1 + β r
EH = ------------------------------------- ----------------------------------------- 5.1
( 1 + βs ) ( 1 – α ) µ–1
βr
EH ( β s = β r ) = ------------------------------------
- 5.2
( 1 + βr ) ( 1 – α )
71
14 14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 EP type, α = 0.10 12 EP type, α = 0.10 12 EP type, α = 0.10
BP type, α = 0.10, βs = βr = 1/6 BP type, α = 0.10, βs = βr = 1/3 BP type, α = 0.10, βs = βr = 1/2
10 10 10
8 8 8
µ
µ
6 6 6
4 4 4
2 2 2
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec) T (sec)
(a)
100 100 100
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10 EP type, α = 0.10 EP type, α = 0.10
80 BP type, α = 0.10, βs = βr = 1/6 80 BP type, α = 0.10, βs = βr = 1/3 80 BP type, α = 0.10, βs = βr = 1/2
60 60 60
p
µp
µ
µ
40 40 40
20 20 20
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec) T (sec)
(b)
4 4 4
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10 EP type, α = 0.10 EP type, α = 0.10
3 BP type, α = 0.10, βs = βr = 1/6 3 BP type, α = 0.10, βs = βr = 1/3 3 BP type, α = 0.10, βs = βr = 1/2
r
µr
2 2 2
µ
1 1 1
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec) T (sec)
(c)
50 50 50
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10 EP type, α = 0.10 EP type, α = 0.10
40 BP type, α = 0.10, βs = βr = 1/6 40 BP type, α = 0.10, βs = βr = 1/3 40 BP type, α = 0.10, βs = βr = 1/2
30 30 30
y
ny
y
n
20 20 20
10 10 10
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec) T (sec)
(d)
Figure 5.5: EP versus BP hysteresis types (α = 0.10, SAC Los Angeles, survival-level, SD
soil): (a) µ; (b) µp; (c) µr; (d) ny.
The normalized energy dissipation, EH, of the BP hysteresis type with different βs = βr values
used in this research are provided in Figure 5.6. For example, the BP type with βs = βr = 1/2 dissi-
pates 37% of the energy dissipated by the EP type during the same displacement cycle.
For βr = 1/2, the µ and ny demands are moderately decreased (by about 14 to 25 percent,
Figs. 5.5a and d) and the µp demands are significantly decreased (by about 50 percent, Fig. 5.5b)
as compared to the BE type, while the µr demands still remain negligible (µr ≅ 0.10, on average,
Fig. 5.5c). Thus, the objective of reducing the µ, µp, and ny demands while keeping the µr demand
within allowable limits can be reasonably achieved with the BP hysteresis type.
72
EP type
BP type, βs = βr = 1/6, EH = 0.16 F
BP type, β = β = 1/3, EH = 0.28
s r
BP type, β = β = 1/2, EH = 0.37
s r
-∆ nlin ∆
∆y ∆nlin
As described in Section 3.1.1, the maximum possible residual displacement of the BP hys-
teresis type is limited to ∆rmax = βr∆be = βr∆y. Thus, the maximum possible residual displacement
ductility demand is µrmax = ∆rmax/∆y = βr. The maximum µr values for the BP hysteresis type in
Figures 5.5a-c are smaller than the corresponding βr values, especially for βr = 1/2. These results
indicate that it may be possible to decrease the µ, µp, and ny demands further without significantly
increasing the µr demand by using βr values larger than 1/2, however this is not investigated by
the research.
73
CHAPTER 6
In this chapter, the effect of site conditions on the capacity-demand index relationships is
investigated. The effects of site soil characteristics are discussed first. Then, the effects of seismic
demand level, site seismicity, and epicentral distance are described. The lateral force capacities,
Fy = Felas/R, of the systems described in this chapter are determined based on the linear-elastic
acceleration response spectrum for each ground motion (i.e., IND spectra). The results are pre-
sented for the SAC ground motion ensemble.
Figure 6.1 compares the mean demand spectra for the EP hysteresis type with α = 0.10 for
stiff (SD) and soft (SE) soil profiles (solid and dashed lines, respectively). The results are presented
for the SAC Los Angeles design-level ground motion ensemble. For structures with T < ~1.25-
1.50 sec., the µ and µp spectra for the SE soil are on average larger than the spectra for the SD soil
(up to 4 times as large, Figs. 6.1a and b). The SE soil profile tends to result in larger µr demands
than the SD soil profile for T < 1 sec. and smaller values of R (Fig. 6.1c). For the same period
range, the ny demands are on average slightly less for the SE soil profile than for the SD soil profile
(Fig. 6.1d). Similar observations can be made for the mean demand spectra of the SD, BE, and BP
(with βs = βr = 1/3) hysteresis types in Figure 6.2, except that µr for the BE type is zero and µr for
the BP type is negligible regardless of soil profile, and ny for the SD type are more or less the
same for the two soil profiles.
For longer-period structures (T > ~1.0-1.5 sec.), the µ, µp, and µr demands are smaller and
ny is virtually unchanged for the SE soil profile as compared to the SD soil profile. Similar obser-
vations can be made for the mean µ and µp demand spectra of the SD, BE, and BP hysteresis
types and for the mean µr demand spectra of the SD and BP hysteresis types. This phenomenon
was previously reported by Miranda (1993). Demands at shorter periods (relative to the predomi-
nant ground motion period) are in general larger for soft soil profiles since period elongation due
to yielding runs the structure into energy-rich regions of the ground motion spectra as shown in
Figures A.22 and A.24 (e.g., To,s to Teff,s in Fig. 6.3a, dashed versus solid lines). In contrast, yield-
ing in longer-period structures with initial linear-elastic periods around the predominant ground
motion period of the soft soil shifts the “effective” structure period, Teff, to the regions of the
ground motion spectra with significantly less energy, reducing the seismic demands (e.g., To,l to
Teff,l in Fig. 6.3a). As shown in Figure 6.1, the period at which the demands for the SE soil profile
74
14 100
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 EP type, α = 0.10, SD soil EP type, α = 0.10, SD soil
EP type, α = 0.10, SE soil 80 EP type, α = 0.10, SE soil
10
8 60
µp
µ
6 40
4
20
2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a) (b)
4 50
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10
0.10, SD soil EP type, α = 0.10,
0.10 SD soil
EP type, α = 0.10, β
BP βs = 1/3
SEr =soil 40 BP type, α = 0.10, S
EP βEr =soil
βs = 1/3
3
30
r
2 ny
µ
20
1
10
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(c) (d)
Figure 6.1: Effect of site soil characteristics (EP hysteresis type, α = 0.10, SAC Los Ange-
les, design-level): (a) µ; (b) µp; (c) µr; (d) ny.
become smaller than the demands for the SD soil profile decreases as R increases. This is expected
since the amount of nonlinear behavior increases as R increases.
It is noted that the lateral force capacities, Fy = Felas/R, of the systems described in this
chapter are determined based on the linear-elastic acceleration response spectrum for each ground
motion, referred to as IND reference spectra (e.g., thin solid line in Fig. 6.3b). This results in
increased Fy values for structures with linear-elastic periods near the predominant ground motion
period, particularly for soft soil profiles, leading to the reduced µ, µp, and µr demands in Figures
6.1 and 6.2. However, as previously discussed in Chapters 1 and 4, smooth design response spec-
tra (referred to as DES reference spectra) are used in current design practice (ICC, 2000; BSSC,
1998; ICBO, 1997), resulting in significantly smaller Fy values for structures near the predomi-
nant ground motion period (e.g., thin dashed line in Fig. 6.3b). Thus, the reduced µ, µp, and µr
demands for SE soil in Figures 6.1a-c and 6.2a-c for longer periods may be unconservative for
structures where smooth design response spectra are used to determine Fy (see shaded region in
Fig. 6.3b for R = 2). This is discussed in more detail later in Chapter 7.
75
14 14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 SD type, α = 0.10, SD soil 12 BE type, α = 0.10, SD soil 12 BP type, α = 0.10, βs = βr = 1/3, SD soil
SD type, α = 0.10, SE soil BE type, α = 0.10, SE soil BP type, α = 0.10, βs = βr = 1/3, SE soil
10 10 10
8 8 8
µ
µ
6 6 6
4 4 4
2 2 2
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec) T (sec)
(a)
100 100 100
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
SD type, α = 0.10, SD soil BE type, α = 0.10, SD soil BP type, α = 0.10, βs = βr = 1/3, SD soil
80 SD type, α = 0.10, SE soil 80 BE type, α = 0.10, SE soil 80 BP type, α = 0.10, βs = βr = 1/3, SE soil
60 60 60
p
p
µ
µ
40 40 40
20 20 20
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec) T (sec)
(b)
4 4 4
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
SD type, α = 0.10, SD soil BE type, α = 0.10, SD soil BP type, α = 0.10, βs = βr = 1/3, SD soil
3 SD type, α = 0.10, SE soil 3 BE type, α = 0.10, SE soil 3 BP type, α = 0.10, βs = βr = 1/3, SE soil
r
µr
2 2 2
µ
1 1 1
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec) T (sec)
(c)
50 50 50
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
SD type, α = 0.10, SD soil BE type, α = 0.10, SD soil BP type, α = 0.10, βs = βr = 1/3, SD soil
40 SD type, α = 0.10, SE soil 40 BE type, α = 0.10, SE soil 40 BP type, α = 0.10, βs = βr = 1/3, SE soil
30 30 30
y
y
n
20 20 20
10 10 10
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec) T (sec)
(d)
Figure 6.2: Effect of site soil characteristics (SD, BE, and BP hysteresis types, α = 0.10,
SAC Los Angeles, design-level): (a) µ; (b) µp; (c) µr; (d) ny.
Figure 6.4 compares the mean demand spectra using the EP hysteresis type with α = 0.10
under the SAC Los Angeles design-level and survival-level ground motion ensembles for sites
with SD and SE soil profiles. For T < ~1.25 sec., the µ and µp demands are smaller (by as much as
60%) for the SD and SE soil design-level ensembles as compared to the SD and SE soil survival-
level ensembles, respectively (Figs. 6.4a and b). For T > ~1.25 sec., the µ and µp demands for the
SD and SE soil design-level ensembles are larger than the demands for the SD and SE soil survival-
level ensembles, respectively (the design-level demands are as much as 1.4 times the µ demands
and 1.6 times the µp demands for the survival-level ensemble).
76
2.0 2.0
LOS ANGELES DESIGN-LEVEL LOS ANGELES DESIGN-LEVEL SOFT SOIL
LS02 response spectrum
LA02 response spectrum
S soil, ξ = 5%
SD soil, ξ = 5% R = 1 (linear-elastic) E
LS02 response spectrum Design response spectrum
S soil, ξ = 5% UBC 1997, S soil, ξ = 5%
E E
Sa (g)
S (g)
a
R=2
unconservative inconsistency
shift of "effective"
structure period
in determining F.y
0 0
0 To,s Teff,s To,l Teff,l 3 0 3
T (sec) T (sec)
(a) (b)
Figure 6.3: Response spectra: (a) SD soil ground motion spectrum versus SE soil ground
motion spectrum; (b) ground motion spectra versus smooth design spectra.
For the SD soil profile, on average, the design-level ensemble tends to result in larger µr
demands for T < 0.75-1.0 sec. and T > 1.75-3.0 sec., depending on R (Fig. 6.4c). For 0.75-1.0 < T
< 1.75-3.0 sec., µr for the SD soil design-level ensemble ranges from 1 to 0.5 times the µr for the
SD soil survival-level ensemble, depending on R. For the SE soil profile, the design-level µr
demands are on average slightly larger than the survival-level µr demands for almost the entire
period range, especially for larger R. The differences in the ny demands for the SD soil design-
level and survival-level ensembles are small for T < 1 sec. (Fig. 6.4d). For T > 1 sec., the ny
demands for the design-level ensemble can be as much as 1.5 times the ny demands for the sur-
vival-level ensemble. For the SE soil profile, the ny demands for the design-level ensemble are
larger than the ny demands for the survival-level ensemble for almost the entire period range (as
much as 50% larger).
Similar observations can be made for the mean demand spectra of the SD, BE, and BP
hysteresis types in Figure 6.5.
The increase in the µ, µp, µr, and ny demands of long period structures from survival-level
to design-level ground motions is explained as follows. Comparing the average linear-elastic
response spectra for the survival-level and the design-level ground motion ensembles (solid and
dashed lines, respectively, Fig. 6.6a), it can be seen that the intensity of the survival-level ground
motion ensemble in the longer period range decreases at a faster rate as compared to the intensity
of the design-level ensemble. Since the lateral force capacities, Fy, of the structures are based on
the IND response spectra, the survival-level ground motion ensemble results in larger R coeffi-
cients to obtain µ, µp, µr, and ny demands similar to the demands for the design-level ensemble.
Thus, the reduced survival-level demands in Figures 6.4 and 6.5 for longer periods may be uncon-
servative for structures where smooth design (DES) response spectra are used to determine Fy as
described in more detail in Chapter 7. This finding is also evident when comparing the average
response spectra for different site seismicities (Fig. 6.6b) and different epicentral distances (Fig.
6.6c), which is discussed in further detail in Sections 6.3 and 6.4, respectively.
77
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 EP type, α = 0.10, survival-level 12 EP type, α = 0.10, survival-level
EP type, α = 0.10, design-level EP type, α = 0.10, design-level
10 10
8 8
µ
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a)
100 100
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10, survival-level EP type, α = 0.10, survival-level
80 EP type, α = 0.10, design-level 80 EP type, α = 0.10, design-level
60 60
µp
40 µp 40
20 20
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(b)
4 4
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10, survival-level EP type, α = 0.10, survival-level
EP type, α = 0.10, design-level EP type, α = 0.10, design-level
3 3
r
2 2
µ
1 1
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(c)
50 50
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10, survival-level EP type, α = 0.10, survival-level
40 EP type, α = 0.10, design-level 40 EP type, α = 0.10, design-level
30 30
ny
ny
20 20
10 10
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
µ
6 6 6
4 4 4
2 2 2
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec) T (sec)
(a)
100 100 100
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
SD type, α = 0.10, survival-level BE type, α = 0.10, survival-level BP type, α = 0.10, βs = βr = 1/3, survival-level
80 SD type, α = 0.10, design-level 80 BE type, α = 0.10, design-level 80 BP type, α = 0.10, βs = βr = 1/3, design-level
60 60 60
p
p
µ
µ
40 40 40
20 20 20
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec) T (sec)
(b)
4 4 4
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
SD type, α = 0.10, survival-level BE type, α = 0.10, survival-level BP type, α = 0.10, βs = βr = 1/3, survival-level
3 SD type, α = 0.10, design-level 3 BE type, α = 0.10, design-level 3 BP type, α = 0.10, βs = βr = 1/3, design-level
r
µr
2 2 2
µ
1 1 1
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec) T (sec)
(c)
50 50 50
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
SD type, α = 0.10, survival-level BE type, α = 0.10, survival-level BP type, α = 0.10, βs = βr = 1/3, survival-level
40 SD type, α = 0.10, design-level 40 BE type, α = 0.10, design-level 40 BP type, α = 0.10, βs = βr = 1/3, design-level
30 30 30
y
y
n
20 20 20
10 10 10
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec) T (sec)
(d)
Figure 6.5: Effect of seismic demand level (SD, BE, and BP hysteresis types, α = 0.10,
SAC Los Angeles, SD soil): (a) µ; (b) µp; (c) µr; (d) ny.
Figure 6.7 compares the mean demand spectra using the EP hysteresis type with α = 0.10
under the Los Angeles and Boston survival-level ground motion ensembles for sites with SD and
SE soil profiles. The µ demands are smaller for the Boston ensemble as compared to the Los
Angeles ensemble for almost the entire period range and, on average, the differences decrease as
T increases (Fig. 6.7a). The same effect that was observed for the seismic demand level in the pre-
vious section begin to appear for T > ~2.2 sec. for the SD soil profile and for T > ~1.75 sec. for the
SE soil profile. This critical structure period increases as R increases. At these longer periods, the
79
2.0 2.5
SAC LOS ANGELES S SOIL SURVIVAL-LEVEL S SOIL
D E
Los Angeles average
survival-level average response spectrum
response spectrum, ξ = 5% ξ = 5%
design-level average
Boston average
response spectrum, ξ = 5% response spectrum
ξ = 5%
S (g)
Sa (g)
a
shift of "effective"
structure period
shift of "effective"
structure period
0 0
0 To Teff 3.5 0 To Teff 3.5
T (sec) T (sec)
(a) (b)
2.0
SAC LOS ANGELES DESIGN-LEVEL S SOIL
D
far-field average
response spectrum, ξ = 5%
near-field average
response spectrum, ξ = 5%
Sa (g)
shift of "effective"
structure period
0
0 To Teff 3.5
T (sec)
(c)
Figure 6.6: Average response spectra: (a) design-level versus survival-level; (b) Boston
versus Los Angeles; (c) far-field versus near-field (NF).
Boston µ demands are larger than the Los Angeles µ demands in the same fashion that the design-
level µ demands are larger than the survival-level µ demands in the previous section.
The increase in seismic demands at longer periods is more evident in the µp and ny
demands. For the SD soil profile, the Boston µp and ny demands are larger than the Los Angeles
demands for T > ~1.75 sec. and T > ~0.5-0.75 sec., respectively (Figs. 6.7b and d). The µp
demands for the SE soil profile are larger for the Boston ensemble as compared to the Los Angeles
ensemble for T > ~1.75 sec. The ny demands for the SE soil profile are always larger for the Bos-
ton ensemble as compared to the Los Angeles ensemble, except for T ≅ 0.3 and 0.6 sec. and R = 2.
As shown in Figure 6.6b, the intensity of the Los Angeles ground motion ensemble in the
longer period range decreases at a faster rate as compared to the intensity of the Boston ensemble,
in the same fashion that the intensity of the survival-level ensemble in the longer period range
decreases at a faster rate as compared to the intensity of the design-level ensemble (Fig. 6.6a).
This results in smaller long-period seismic demands for Los Angeles, which may be unconserva-
80
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 EP type, α = 0.10, Los Angeles 12 EP type, α = 0.10, Los Angeles
EP type, α = 0.10, Boston EP type, α = 0.10, Boston
10 10
8 8
µ
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a)
100 100
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10, SLos Angeles EP type, α = 0.10, SLos Angeles
80 EP type, α = 0.10, SBoston 80 EP type, α = 0.10, SBoston
60 60
µp
40 µp 40
20 20
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(b)
4 4
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10,
0.10 Los Angeles EP type, α = 0.10,
0.10 Los Angeles
BP type, α = 0.10, Boston
EP βr = βs = 1/3 BP type, α = 0.10, Boston
EP βr = βs = 1/3
3 3
r
2 2
µ
1 1
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(c)
50 50
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10,
0.10 Los Angeles EP type, α = 0.10,
0.10 Los Angeles
40 BP type, α = 0.10, Boston
EP βr = βs = 1/3 40 BP type, α = 0.10, Boston
EP βr = βs = 1/3
30 30
y
y
n
20 20
10 10
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
Figure 6.7c shows that the differences between the µr demands for the two seismicity lev-
els are erratic but relatively small.
Similar observations can be made for the mean demand spectra under the design-level
ground motion ensemble in Figure 6.8 and for the mean demand spectra of the SD, BE, and BP
hysteresis types in Figure 6.9, except that the differences in the µp demands between the Boston
and Los Angeles ensembles are relatively small when using the BE and BP hysteresis types.
Figure 6.10 compares the mean demand spectra using the EP hysteresis type with α = 0.10
for the Los Angeles design-level SD soil far-field and near-field ground motion ensembles. For T <
~1.0-1.5 sec., the µ and µp demands are larger for the near-field ensemble as compared to the far-
field ensemble (up to 3.25 times as large for µ and 4 times as large for µp, Figs. 6.10a and b). On
average, the near-field ensemble has smaller or similar µr demands for T < ~0.5-1.10 sec. and T >
~1.75-2.75 sec. (Fig. 6.10c), depending on R. For ~0.5-1.10 < T < ~1.75-2.75 sec., the µr demand
for the near-field ensemble can be as much as 2.5 times the µr demand for the far-field ensemble.
The ny demands for the near-field ensemble are similar to the ny demands for the far-field ensem-
ble for T < ~0.5 sec. and can be as small as 0.5 times the demands for the far-field ensemble for T
> ~0.5 sec. (Fig. 6.10d).
The differences between the near-field and far-field demands for µ and µp increase as R
increases for T < ~1.0-1.5 sec. and the differences remain relatively constant for T > ~1.0-1.5 sec.
(Figs. 6.10a and b). The differences in µr are highly dependent on the R coefficient (Fig. 6.10c).
The differences in ny are somewhat independent of the R coefficient for the entire period range
(Fig. 6.10d).
Similar to what was observed for the design-level ensemble versus survival-level ensem-
ble discussed previously, there is an increase in the µ, µp, and ny demands from near-field to far-
field ground motions for longer period structures (T > ~0.5-1.5 sec.) since these results are based
on the IND spectra. Figure 6.6c shows that, in the longer period range, the intensity of the near-
field ensemble decreases at a faster rate as compared to the intensity of the far-field ensemble. As
noted earlier and discussed in detail in Chapter 7, the smaller long-period seismic demands for
near-field ground motion records in Figure 6.10 may be unconservative for structures where
smooth design (DES) response spectra are used to determine Fy.
82
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 EP type, α = 0.10, Los Angeles 12 EP type, α = 0.10, Los Angeles
EP type, α = 0.10, Boston EP type, α = 0.10, Boston
10 10
8 8
µ
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a)
100 100
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10, SLos Angeles EP type, α = 0.10, SLos Angeles
80 EP type, α = 0.10, SBoston 80 EP type, α = 0.10, SBoston
60 60
µp
40 µp 40
20 20
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(b)
4 4
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10,
0.10 Los Angeles EP type, α = 0.10,
0.10 Los Angeles
BP type, α = 0.10, Boston
EP βr = βs = 1/3 BP type, α = 0.10, Boston
EP βr = βs = 1/3
3 3
r
2 2
µ
1 1
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(c)
50 50
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10,
0.10 Los Angeles EP type, α = 0.10,
0.10 Los Angeles
40 BP type, α = 0.10, Boston
EP βr = βs = 1/3 40 BP type, α = 0.10, Boston
EP βr = βs = 1/3
30 30
y
y
n
20 20
10 10
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
µ
6 6 6
4 4 4
2 2 2
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec) T (sec)
(a)
100 100 100
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
SD type, α = 0.10, Los Angeles BE type, α = 0.10, Los Angeles BP type, α = 0.10, βs = βr = 1/3, Los Angeles
80 SD type, α = 0.10, Boston 80 BE type, α = 0.10, Boston 80 BP type, α = 0.10, βs = βr = 1/3, Boston
60 60 60
p
µp
µ
40 µ 40 40
20 20 20
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec) T (sec)
(b)
4 4 4
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
SD type, α = 0.10, Los Angeles BE type, α = 0.10, Los Angeles BP type, α = 0.10, βs = βr = 1/3, Los Angeles
3 SD type, α = 0.10, Boston 3 BE type, α = 0.10, Boston 3 BP type, α = 0.10, βs = βr = 1/3, Boston
r
µr
2 2 2
µ
1 1 1
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec) T (sec)
(c)
50 50 50
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
SD type, α = 0.10, Los Angeles BE type, α = 0.10, Los Angeles BP type, α = 0.10, βs = βr = 1/3, Los Angeles
40 SD type, α = 0.10, Boston 40 BE type, α = 0.10, Boston 40 BP type, α = 0.10, βs = βr = 1/3, Boston
30 30 30
y
y
n
20 20 20
10 10 10
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec) T (sec)
(d)
Figure 6.9: Effect of site seismicity (SD, BE, and BP hysteresis types, α = 0.10, SAC sur-
vival-level, SD soil): (a) µ; (b) µp; (c) µr; (d) ny.
84
14 100
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 EP type, α = 0.10, far-field EP type, α = 0.10, far-field
EP type, α = 0.10, near-field 80 EP type, α = 0.10, near-field
10
8 60
µp
µ
6 40
4
20
2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a) (b)
4 50
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.10, far-field EP type, α = 0.10, far-field
EP type, α = 0.10, near-field 40 EP type, α = 0.10, near-field
3
30
µr
ny
2
20
1
10
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(c) (d)
Figure 6.10: Effect of epicentral distance (EP hysteresis type, α = 0.10, SAC Los Angeles,
design-level, SD soil): (a) µ; (b) µp; (c) µr; (d) ny.
85
CHAPTER 7
As described earlier, the results shown in the previous chapters are based on Fy = Felas/R
values determined using the linear-elastic acceleration response spectrum for each ground motion
(i.e., IND spectra). In this chapter, the effect of the reference response spectrum (i.e., smooth ver-
sus IND spectra) on the R-µ-T relationships is investigated. The results are presented as mean R-µ
spectra for different hysteresis types, site seismicities, seismic demand levels, site soil conditions,
and epicentral distances using the SAC ground motion ensemble.
Figures 7.1a-b show the mean R-µ spectra using the EP hysteresis type with α = 0.10 for
sites with a stiff soil profile in a region with low seismicity under design-level and survival-level
ground motions, respectively. The solid and dashed lines represent results obtained using the IND
and AVG spectra, respectively. The differences between the results obtained using the two spectra
are moderate. For long period structures (i.e., T > 1 sec.), the µ demands estimated using the AVG
spectrum are consistently larger than the demands estimated using the IND spectra.
Similarly, Figures 7.1c-d show comparisons between the R-µ spectra obtained using the
IND and DES spectra. Under design-level ground motions, the differences between the DES and
IND spectra (Fig. 7.1c) are similar to the differences between the AVG and IND spectra (Fig.
7.1a). For survival-level ground motions, the increase in the µ demands when using the DES spec-
trum is significantly larger for T < 2 sec. From these observations, it can be concluded that R-µ-T
relationships developed using IND spectra can lead to underestimated µ demands, and thus,
unconservative designs.
Figures 7.1e-h show the mean R-µ spectra using the EP hysteresis type with α = 0.10 for
sites with a soft soil profile in a region with low seismicity under design-level and survival-level
ground motions. The solid lines represent results obtained using the IND spectra and the dashed
lines represent results obtained using the AVG and DES spectra. Similar to the stiff soil profile,
the increase in the µ demands is moderate when using the AVG spectrum and is significant for the
survival-level ground motions when using the DES spectrum.
86
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 IND spectra 12 IND spectra
AVG spectrum AVG spectrum
10 10
8 8
µ
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 IND spectra 12 IND spectra
DES spectrum DES spectrum
10 10
8 8
µ
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 IND spectra 12 IND spectra
AVG spectrum AVG spectrum
10 10
8 8
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 IND spectra 12 IND spectra
DES spectrum DES spectrum
10 10
8 8
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
87
It is noted that, for the survival-level ground motions, the increase in the µ demand due to
the use of the DES spectrum is much larger for sites with a soft soil profile (Fig. 7.1h) than for
sites with a stiff soil profile (Fig. 7.1d: e.g., the µ demand from the DES spectrum can be about 2
times the µ demand from the IND spectra as shown in Fig. 7.1h). These large increases occur for
almost the entire period range (i.e., T > 0.25 sec.) and are especially severe for 1.0 < T < 1.5 sec.
where the predominant period of the soil resides (see Fig. A.21). This period range corresponds to
a “dip” in the R-µ spectra (especially for large R coefficients) when using the IND spectra since
the lateral strength, Fy, is increased, as mentioned earlier in Section 6.1. The R-µ spectra for the
AVG spectrum follow this dip (Fig. 7.1f). However, the R-µ spectra for the DES spectrum do not
show a dip since, in this case, Fy is not based on the linear-elastic acceleration response spectrum
of each ground motion. It is obvious from these results that using IND spectra instead of DES
spectra to develop R-µ-T relationships can lead to significantly unconservative designs, especially
for medium- to long-period structures under survival-level, soft soil ground motions.
Figures 7.2a-b show the mean R-µ spectra using the EP hysteresis type with α = 0.10 for
sites with a stiff soil profile in a region with high seismicity under design-level and survival-level
ground motions. The solid and dashed lines represent results obtained using the IND and AVG
spectra, respectively. On average, the differences between the results obtained using the IND and
AVG spectra are small, particularly for the design-level ground motions. For long period struc-
tures (i.e., T > 1 sec.) under survival-level ground motions, the µ demands estimated using the
AVG spectrum are consistently larger than the demands estimated using the IND spectra.
Similarly, Figures 7.2c-d show comparisons between the R-µ spectra obtained using the
IND and DES spectra. Contrary to the AVG spectrum, the µ demands under design-level ground
motions decrease when using the DES spectrum. Thus, R-µ-T relationships developed using the
IND spectra would result in more conservative designs for design-level ground motions than rela-
tionships developed using the DES spectrum.
Under survival-level ground motions, the µ demands obtained using the DES spectrum
can be significantly larger than the demands obtained using the IND spectra, particularly for large
R coefficients and T > 1 sec. From this important observation, it can be concluded that R-µ-T rela-
tionships developed using IND spectra can lead to underestimated seismic demands and, thus,
unconservative designs, especially under survival-level ground motions.
Figures 7.2e-h show the mean R-µ spectra using the EP hysteresis type with α = 0.10 for
sites with a soft soil profile in a region with high seismicity under design-level and survival-level
ground motions, respectively. Similar to the stiff soil profile, the differences between the results
obtained using the IND and AVG spectra are small, particularly for the design-level ground
motions. However, there is a significant increase in the demands when using the DES spectrum,
especially for the survival-level ground motions (e.g., the µ demand from the DES spectrum can
88
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 IND spectra 12 IND spectra
AVG spectrum AVG spectrum
10 10
8 8
µ
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 IND spectra 12 IND spectra
DES spectrum DES spectrum
10 10
8 8
µ
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 IND spectra 12 IND spectra
AVG spectrum AVG spectrum
10 10
8 8
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 IND spectra 12 IND spectra
DES spectrum DES spectrum
10 10
8 8
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
89
be about 2.5 times the µ demand from the IND spectra as shown in Fig. 7.2h). These large
increases occur for almost the entire period range (i.e., T > 0.25 sec.) and are especially severe for
1.0 < T < 2.5 sec. where the predominant period of the soil resides (see Fig. A.25). It is obvious
from these results that using IND spectra instead of DES spectra to develop R-µ-T relationships
can lead to significantly unconservative designs, especially for medium- to long-period structures
under survival-level, soft soil ground motions in regions with high seismicity. Furthermore, for
the DES spectrum, the µ demands are extreme and possibly uncontrollable at higher values of R,
especially for survival-level, soft soil ground motions in regions with high seismicity. Thus, the R
coefficients specified in current seismic design provisions may be very unconservative for these
conditions. Either the smooth design response spectra in the current provisions need to be modi-
fied (e.g., using AVG spectra instead) or the R coefficients recommended in the provisions should
be reduced under these conditions.
7.5 High Seismicity (Los Angeles), Near-Field (NF), Stiff Soil Profile (SD)
Figure 7.3 shows the mean R-µ spectra using the EP hysteresis type with α = 0.10 for sites
with a stiff soil profile in a region with high seismicity under near-field ground motions. The dif-
ferences between the results obtained using the IND and AVG spectra are small (Fig. 7.3a). How-
ever, the increase in the µ demands when using the DES spectrum is dramatic (the µ demand from
the DES spectrum can be almost 3 times the demand from the IND spectra as shown in Fig. 7.3b).
The large amplification factors using the DES spectrum occur for almost the entire period
range (i.e., T > 0.25 sec.) and are especially severe for T > 0.5 sec. The pulse-like (large ampli-
tude, long period) characteristics (see Fig. A.26) of the NF ground motions result in large µ
demands for both medium-period and long-period structures. From these observations, it is con-
cluded that R-µ-T relationships developed using IND spectra can lead to significantly unconserva-
tive designs for structures within close proximity of an active fault. Furthermore, for the DES
spectrum, the near-field µ demands are extreme and possibly uncontrollable at higher values of R.
Thus, the R coefficients specified in current seismic design provisions may be very unconservative
for these conditions. Either the smooth design response spectra in the current provisions need to
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines)
12 IND spectra 12
AVG spectrum
10 10
8 8
µ
6 6
4 4
R = 1, 2, 4, 6, 8 (thin → thick lines)
2 2 IND spectra
DES spectrum
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a) (b)
Figure 7.3: R-µ spectra for near-field (Los Angeles design-level SD soil, EP hysteresis
type, α = 0.10: (a) AVG versus IND spectra; (b) DES versus IND spectra.
90
be modified (e.g., using AVG spectra instead) or the R coefficients recommended in the provisions
should be reduced under these conditions.
This section investigates the effect of hysteresis type on the R-µ-T relationships developed
using different reference response spectra. The results are presented as mean R-µ spectra for the
SAC Los Angeles survival-level SD soil ground motion ensemble.
Figure 7.4 shows the mean R-µ spectra for the SD, BE, and BP (βs = βr = 1/3) hysteresis
types (α = 0.10). The solid lines represent the results obtained using the IND spectra and the
dashed lines represent the results obtained using the AVG and DES spectra. Figures 7.2 b and d
provide results for the EP hysteresis type (α = 0.10) for similar site conditions. The differences
between the µ demands for the IND spectra, and the AVG and DES spectra are not significantly
affected by the hysteresis type. Thus, it is concluded that the effect of the reference response spec-
tra on the mean µ demand is not significantly dependent on the hysteresis type.
91
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 IND spectra 12 IND spectra
AVG spectrum DES spectrum
10 10
8 8
µ
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a) (b)
14 14
12 12
10 10
8 8
µ
µ
6 6
4 4
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
2 IND spectra 2 IND spectra
AVG spectrum DES spectrum
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(c) (d)
14 14
12 12
10 10
8 8
µ
6 6
4 4
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
2 IND spectra 2 IND spectra
AVG spectrum DES spectrum
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(e) (f)
Figure 7.4: Effect of hysteresis type and reference response spectra on the µ demand
(Los Angeles survival-level SD soil, α = 0.10): (a-b) SD hysteresis type; (c-d) BE hyster-
esis type; (e-f) BP hysteresis type (βs = βr = 1/3).
92
CHAPTER 8
To achieve these objectives, the R-µ-T relationships are developed using the two-step non-
linear regression analysis scheme and the form of the regression equations developed by Nassar
and Krawinkler (Eqs. 2.1-2.3) as described previously in Section 3.7.1.
Figure 8.1 compares the R-µ-T relationships developed in this study with relationships
developed by Nassar and Krawinkler (1991) in terms of R-µ spectra. The results are presented for
the EP hysteresis type with α = 0.10 based on the N&K ground motion ensemble.
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
constant−µ, IND spectra: a = 0.80, b = 0.29 IND spectra, constant−R: a = 1.35, b = 0.44
12 constant−R, IND spectra: a = 1.35, b = 0.44 12 AVG spectrum, constant−R: a = 2.74, b = 0.57
DES spectrum, constant−R: a = 1.74, b = 0.55
10 10
8 8
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a) (b)
Figure 8.1: Regression curves (EP hysteresis type, α = 0.10): (a) constant-µ versus constant-
R approaches; (b) IND spectra versus AVG and DES spectra.
93
The regression coefficients, a and b, corresponding to the R-µ spectra in Figure 8.1 are
given in Table 8.1. As explained in Section 4.3, there are two differences between the new coeffi-
cients and previous research: (1) the use of the constant-R approach (this research) instead of the
constant-µ approach (previous research); and (2) the use of AVG and DES spectra (this research)
instead of IND spectra (previous research). These differences are illustrated in Figures 8.1a and b,
respectively, and quantified in Table 8.1. The results indicate that the previous R-µ-T relationships
can result in underestimated seismic demands for design.
Table 8.1: Regression coefficients a and b for the N&K ground motion
ensemble, EP hysteresis type, α = 0.10
Regression Coefficient
Approach a b
Table 8.2 shows the a and b regression coefficients developed in this study for the SAC
ground motion ensemble using the analyses presented earlier. Regression coefficients are given
for all three reference response spectra, where applicable. Figure 8.2 provides a sampling of com-
parisons between the mean R-µ spectra and the regression curves developed through this research.
A reasonably good fit is observed, except for the SAC Los Angeles, design-level, SD soil, NF
ground motion ensemble (Fig. 8.2h). Thus, care should be taken when using the regression rela-
tionship developed for this case. The regression relationships developed for the other cases inves-
tigated by the research (as shown in Table 8.2) provide a reasonably good representation of the
mean demands from the dynamic analyses.
The R-µ spectra based on the regression coefficients listed in Table 8.2 are provided in
Figures 8.3 through 8.11 to reinforce the previous findings from the mean R-µ spectra. The figures
are organized in the same order that the results are presented in Chapters 5-7: (1) effect of hyster-
esis type (Fig. 8.3); (2) effect of site conditions (Figs. 8.4-8.7); and (3) effect of reference
response spectra (Figs. 8.8-8.11).
The effect of the post-yield stiffness ratio, α, can be clearly seen in Figure 8.3a, especially
between α = 0.00 and α = 0.10. For hysteresis type (Fig. 8.3b), the significant difference between
94
Table 8.2: Regression coefficients a and b for the SAC ground motion ensemble
Hysteresis Demand Site Soil/ IND AVG DES
α, βs, βr Site Seismicity
Type Level Distance a b a b a b
α = 0.00 Los Angeles Survival SD 0.53 0.82 -- -- -- --
α = 0.05 Los Angeles Survival SD 0.37 0.74 -- -- -- --
SD 0.48 0.45 1.08 0.49 0.65 0.62
Survival
SE -0.63 0.52 0.11 0.57 0.41 0.84
Boston
SD 0.56 0.45 1.21 0.49 1.08 0.42
Design
EP SE 0.12 0.41 0.82 0.48 0.87 0.43
α = 0.10 SD 0.40 0.72 1.01 0.73 2.04 0.79
Survival
SE -0.71 0.94 -0.07 1.00 2.32 1.31
Los Angeles SD 1.89 0.68 1.92 0.66 1.71 0.59
Design SE 0.15 0.85 0.41 0.86 0.80 0.88
NF 0.35 0.89 0.67 0.92 6.23 1.34
Boston Survival SD 0.54 0.48 -- -- -- --
Survival SD 0.52 0.75 1.14 0.76 2.20 0.82
SD α = 0.10
Los Angeles SD 2.09 0.70 -- -- -- --
Design
SE 0.22 0.86 -- -- -- --
Boston Survival SD 1.30 0.72 -- -- -- --
Survival SD 1.53 0.97 2.20 1.00 3.68 1.06
BE α = 0.10
Los Angeles SD 4.74 0.94 -- -- -- --
Design
SE 0.94 1.09 -- -- -- --
α = 0.10
Los Angeles Survival SD 1.25 0.93 -- -- -- --
βs = βr = 1/6
Boston Survival SD 0.93 0.62 -- -- -- --
α = 0.10 Survival SD 1.08 0.89 1.65 0.91 3.00 0.97
BP βs = βr = 1/3 Los Angeles SD 3.83 0.87 -- -- -- --
Design
SE 0.65 1.02 -- -- -- --
α = 0.10
Los Angeles Survival SD 0.96 0.87 -- -- -- --
βs = βr = 1/2
the µ demands for the EP and BE types is demonstrated. For the BP hysteresis type with varying
βs = βr (Fig. 8.3c), the µ demands decrease as βr increases. The dependence of the µ demands on
site conditions (i.e., site soil characteristics, seismic demand level, site seismicity, and epicentral
distance) is demonstrated in Figures 8.4-8.7.
The dramatic effect of using smooth design response spectra to determine the design lat-
eral force capacity, Fy, can be clearly seen, especially for structures subjected to survival-level
ground motions in Figures 8.8d, 8.9b and d. For the DES spectrum, the µ demands are extreme
and possibly uncontrollable at higher values of R, especially for soft soil and near-field ground
motions in regions with high seismicity (Figs. 8.9d and 8.10). Thus, the R coefficients specified in
current seismic design provisions may be very unconservative for these conditions. Either the
smooth design response spectra in the current provisions need to be modified (e.g., using AVG
spectra instead) or the R coefficients recommended in the provisions should be reduced under
these conditions.
The lack of dependence of the effect of reference response spectra on hysteresis type can
be seen in Figure 8.11.
95
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12 mean 12 mean
regression: a = 0.56, b =0.45 regression: a = 0.12, b = 0.41
10 10
µ 8 8
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
8 8
µ
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
8 8
µ
6 6
4 4
2 2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
8 8
µ
6 6
4 4
R = 1, 2, 4, 6, 8 (thin → thick lines)
2 2 mean
regression: a = 6.23, b = 1.34
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
96
14 14
12 12
10 10
8 8
µ
µ
6 6
4 4
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
EP type, α = 0.00: a = 0.53, b = 0.82 EP type, α = 0.10: a = 0.40, b = 0.72
2 2
EP type, α = 0.05: a = 0.37, b = 0.74 SD type, α = 0.10: a = 0.52, b = 0.75
EP type, α = 0.10: a = 0.40, b = 0.72 BE type, α = 0.10: a = 1.53, b = 0.97
(a) 0 (b) 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
14
12
10
8
µ
4
R = 1, 2, 4, 6, 8 (thin → thick lines)
BP type, α = 0.10, βs = βr = 1/6: a = 1.25, b = 0.93
2 BP type, α = 0.10, βs = βr = 1/3: a = 1.08, b = 0.89
BP type, α = 0.10, β = β = 1/2: a = 0.96, b = 0.87
(c) 0
0 0.5 1
s
1.5
r
2 2.5 3 3.5
T (sec)
Figure 8.3: Effect of hysteretic behavior on regression curves using IND spectra (SAC Los
Angeles, survival-level, SD soil): (a) post-yield stiffness ratio, α; (b) hysteresis type; (c) βs
= βr.
97
14 14
12 12
10 10
8 8
µ
µ
6 6
4 4
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
2 EP type, α = 0.10, S soil: a = 1.89, b = 0.68 2 SD type, α = 0.10, S soil: a = 2.09, b = 0.70
D D
EP type, α = 0.10, S soil: a = 0.15, b = 0.85 SD type, α = 0.10, SE soil: a = 0.22, b = 0.86
(a) 0
E
(b) 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
14 14
12 12
10 10
8 8
µ
µ
6 6
4 4
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
2 BE type, α = 0.10, S soil: a = 4.74, b = 0.94 2 BP type, α = 0.10, β = β = 1/3, S soil: a = 3.83, b = 0.87
D s r D
BE type, α = 0.10, SE soil: a = 0.94, b = 1.09 BP type, α = 0.10, βs = βr = 1/3, SE soil: a = 0.65, b = 1.02
(c) 0 (d) 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
Figure 8.4: Effect of site soil characteristics on regression curves using IND spectra (SAC
Los Angeles, design-level): (a) EP hysteresis type; (b) SD hysteresis type; (c) BE hystere-
sis type; (d) BP hysteresis type.
98
14 14
12 12
10 10
8 8
µ
6 6
(a) (b)
4 4
14 14
12 12
10 10
8 8
µ
µ
6 6
(c) (d) SAC Los Angeles
4 4
14
12
10
8
µ
(e) 4
R = 1, 2, 4, 6, 8 (thin → thick lines)
2 BP type, α = 0.10, β = β = 1/3, survivallevel: a = 1.08, b = 0.89
s r
BP type, α = 0.10, βs = βr = 1/3, design level: a = 3.83, b = 0.87
SD soil 0
0 0.5 1 1.5 2 2.5 3 3.5
T (sec)
Figure 8.5: Effect of seismic demand level on regression curves using IND spectra (SAC
Los Angeles): (a-b) EP hysteresis type; (c) SD hysteresis type; (d) BE hysteresis type; (e)
BP hysteresis type.
99
14 14
12 12
10 10
8 8
µ
6 6
(a) (b)
4 4
14 14
12 12
10 10
8 8
µ
µ
6 6
(c) (d)
4 4
14 14
12 12
10 10
8 8
µ
6 6
(e) (f)
4 4
14
12
10
8
µ
(g) 4
R = 1, 2, 4, 6, 8 (thin → thick lines)
2 BP type, α = 0.10, β = β = 1/3, Los Angeles: a = 1.08, b = 0.89
s r
BP type, α = 0.10, βs = βr = 1/3, Boston: a = 0.93, b = 0.62
0
survival-level 0 0.5 1 1.5 2 2.5 3 3.5
T (sec)
Figure 8.6: Effect of site seismicity on regression curves using IND spectra: (a-d) EP hys-
teresis type; (e) SD hysteresis type; (f) BE hysteresis type; (g) BP hysteresis type.
100
14
12
10
µ
6
Figure 8.7: Effect of epicentral distance on regression curves using IND spectra (SAC Los
Angeles, design-level, SD soil).
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
IND spectra: a = 0.56, b = 0.45 IND spectra: a = 0.48, b = 0.45
12 AVG spectrum: a = 1.21, b = 0.49 12 AVG spectrum: a = 1.08, b = 0.49
DES spectrum: a = 1.08, b = 0.42 DES spectrum: a = 0.65, b = 0.62
10 10
8 8
µ
µ
6 6
(a) (b)
4 4
2 2
design-level 0 survival-level 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
IND spectra: a = 0.12, b = 0.41 IND spectra: a = −0.63, b = 0.52
12 AVG spectrum: a = 0.82, b = 0.48 12 AVG spectrum: a = 0.11, b = 0.57
DES spectrum: a = 0.87, b = 0.43 DES spectrum: a = 0.41, b = 0.84
10 10
8 8
µ
6 6
(c) (d)
4 4
2 2
design-level 0 survival-level 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
Figure 8.8: Effect of reference response spectra on regression curves (SAC Boston, EP hys-
teresis type, α = 0.10): (a-b) SD soil; (c-d) SE soil.
101
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
IND spectra: a = 1.89, b = 0.68 IND spectra: a = 0.40, b = 0.72
12 AVG spectrum: a = 1.92, b = 0.66 12 AVG spectrum: a = 1.01, b = 0.73
DES spectrum: a = 1.71, b = 0.59 DES spectrum: a = 2.04, b = 0.79
10 10
8 8
µ
µ
6 6
(a) (b)
4 4
2 2
design-level 0 survival-level 0
0 0.5 1 1.5 2 2.5 3 3.5
0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
14 14
R = 1, 2, 4, 6, 8 (thin → thick lines)
IND spectra: a = 0.15, b = 0.85
12 AVG spectrum: a = 0.41, b = 0.86 12
DES spectrum: a = 0.80, b = 0.88
10 10
8 8
µ
µ
(c) 6 (d) 6
4 4
R = 1, 2, 4, 6, 8 (thin → thick lines)
2 2 IND spectra: a = −0.71, b = 0.94
design-level survival-level AVG spectrum: a = −0.07, b = 1.00
DES spectrum: a = 2.32, b = 1.31
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
Figure 8.9: Effect of reference response spectra on regression curves (SAC Los Angeles, EP
hysteresis type, α = 0.10): (a-b) SD soil; (c-d) SE soil.
14
12
10
8
µ
4
R = 1, 2, 4, 6, 8 (thin → thick lines)
2 IND spectra: a = 0.35, b = 0.89
AVG spectrum: a = 0.67, b = 0.92
DES spectrum: a = 6.23, b = 1.34
0
0 0.5 1 1.5 2 2.5 3 3.5
T (sec)
Figure 8.10: Effect of reference spectra on regression curves (SAC Los Angeles, design-
level, SD soil, NF, EP hysteresis type, α = 0.10).
102
14 14
12 12
10 10
8 8
µ
µ
6 6
4 4
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
2 IND spectra: a = 0.52, b = 0.75 2 IND spectra: a = 1.53, b = 0.97
AVG spectrum: a = 1.14, b = 0.76 AVG spectrum: a = 2.20, b = 1.00
DES spectrum: a = 2.20, b = 0.82 DES spectrum: a = 3.68, b = 1.06
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a) (b)
14
12
10
8
µ
4
R = 1, 2, 4, 6, 8 (thin → thick lines)
2 IND spectra: a = 1.08, b = 0.89
AVG spectrum: a = 1.65, b = 0.91
DES spectrum: a = 3.00, b = 0.97
0
0 0.5 1 1.5 2 2.5 3 3.5
T (sec)
(c)
Figure 8.11: Effect of reference response spectra on regression curves (SAC Los Angeles,
survival-level, SD soil, α = 0.10): (a) SD hysteresis type; (b) BE hysteresis type; (c) BP
hysteresis type (βs = βr = 1/3).
103
CHAPTER 9
The previous chapter provided regression relationships between R, µ, and T. Similarly, this
chapter provides regression relationships between the demand indices, µ, µp, µr, and ny, in the
form of Equations 3.43 and 3.44. A simple one-step nonlinear regression analysis is performed for
each relationship. Regression relationships developed using IND spectra are given in Section 9.1.
Similarly, Section 9.2 presents regression relationships developed using AVG and DES spectra,
including comparisons with relationships developed using IND spectra. All regression relation-
ships presented in this chapter are based on the SAC ground motion ensembles.
Tables 9.1 and 9.2 show the d and f regression coefficients developed for the relationship
between µ and the other demand indices, Λ = µp, µr, and ny, based on Equation 3.43. Similarly,
Tables 9.3 and 9.4 show the g and h regression coefficients developed for the relationship between
the demand indices other than µ (i.e., Λ = µp, µr, and ny) based on Equation 3.44. In Tables 9.3-
9.4, the columns represent the independent demand indices, Λj, and the rows represent the depen-
dent demand indices, Λi, in Equation 3.44. The lateral strength, Fy = Felas/R, of the systems used
in this section are based on the IND reference response spectra.
The correlation coefficients, ρ, between the demand indices as defined in Section 3.7.1, are
listed in Tables 9.5 and 9.6. Figures 9.1-9.20 provide matrix plots for the cross-correlations
between the demand indices. The figures are organized in the same order that the results are pre-
sented in Chapters 5 and 6: (1) effect of hysteresis type (Figs. 9.1-9.3); and (2) effect of site con-
ditions (Figs. 9.4-9.20). The dots represent the data from the dynamic analyses and the lines
represent the regression relationships using Equations 3.43 and 3.44.
In Figures 9.1-9.20, the correlation between µ and the other demand indices is relatively
strong, especially for µp (correlation coefficient, ρ, is as high as 0.99, close to full correlation). In
some cases, the cross-correlations between the demand indices show weak correlation, indicating
that these demand indices can carry independent measures of seismic demand (e.g., µ versus ny
and µp versus ny in Fig. 9.3). Designers should be careful in using the regression relationships
developed for these cases. Note that ρ values that are very close to zero indicate that one of the
demand indices shows close to no variation as another demand index is varied (e.g., µ versus µr,
µp versus µr, and ny versus µr for the BE hysteresis type since µr for the BE type is always zero,
Fig. 9.2b).
104
Using the relationships between µ and the other demand indices, a more extensive perfor-
mance-based seismic design procedure can be realized. For example, for a given “target-µ” as part
of a displacement-based design approach, it is possible to design (or redesign) structures with
enhanced objectives in mind (e.g., limiting residual displacement and/or cumulative damage).
This is outlined later in Chapter 10.
105
Table 9.2: Regression coefficients d and f: SD, BE, and BP hysteresis types, IND spectra
Hysteresis Site Demand Site Soil/ Regression Coefficientsa
α, βs, βr
Type Seismicity Level Distance Demand Index, Λ d f
µp 7.88 0.76
Boston Survival SD µr 0.36 2.08
ny 8.32 3.89
µp 1.80 0.48
Survival SD µr 0.37 1.82
ny 3.86 3.02
SD α = 0.10
µp 3.61 0.58
Los Angeles SD µr 0.28 1.30
ny 5.21 4.69
Design
µp 7.16 0.88
SE µr 0.32 1.80
ny 3.20 3.44
µp 6.70 0.76
Boston Survival SD µr 0 0
ny 1.41 0.63
µp 4.31 0.68
Survival SD µr 0 0
ny 0.19 0.36
BE α = 0.10
µp 5.41 0.73
Los Angeles SD µr 0 0
ny 0.61 0.45
Design
µp 6.18 0.84
SE µr 0 0
ny 2.59 1.00
µp 5.04 0.83
α = 0.10
Los Angeles Survival SD µr 0.02 1.07
βs = βr = 1/6
ny 0.13 0.32
µp 4.49 0.72
Boston Survival SD µr 0.05 2.80
ny 4.56 1.09
µp 4.71 0.86
Survival SD µr 0.03 1.87
α = 0.10 ny 0.20 0.35
BP βs = βr = 1/3 µp 3.46 0.70
Los Angeles SD µr 0.06 1.49
ny 1.91 0.74
Design
µp 3.50 0.74
SE µr 0.05 1.71
ny 6.50 2.78
µp 3.99 0.82
α = 0.10
Los Angeles Survival SD µr 0.05 2.69
βs = βr = 1/2
ny 0.32 0.40
106
Table 9.3: Regression coefficients g and h: EP hysteresis type, IND spectra
Regression Coefficientsa
g h g h g h
µp -- -- 1.56 0.57 0.26 0.34
α = 0.00 Los Angeles Survival SD µr 0.64 1.75 -- -- 0.24 0.70
ny 3.92 2.93 4.10 1.42 -- --
µp -- -- 4.63 0.31 0.27 0.38
α = 0.05 Los Angeles Survival SD µr 0.22 3.23 -- -- 0.18 1.75
ny 3.73 2.65 5.63 0.57 -- --
µp -- -- 60.7 0.24 0.23 0.43
SD µr 0.02 4.17 -- -- 0.04 1.89
ny 4.42 2.31 26.7 0.53 -- --
Survival
µp -- -- 364 0.18 0.20 0.32
SE µr 0.003 5.56 -- -- 0.01 3.13
ny 5.00 3.13 73.7 0.32 -- --
Boston
µp -- -- 58.2 0.21 0.22 0.43
SD µr 0.02 4.76 -- -- 0.03 2.50
ny 4.59 2.34 28.9 0.40 -- --
Design
µp -- -- 91.0 0.17 0.25 0.42
EP SE µr 0.01 5.88 -- -- 0.04 2.38
ny 4.05 2.36 26.4 0.42 -- --
µp -- -- 45.7 0.23 0.28 0.40
α = 0.10 SD µr 0.02 4.35 -- -- 0.05 2.63
ny 3.59 2.49 21.0 0.38 -- --
Survival
µp -- -- 189 0.20 0.27 0.28
SE µr 0.005 5.00 -- -- 0.06 1.49
ny 3.68 3.63 17.0 0.67 -- --
µp -- -- 24.1 0.30 0.25 0.40
Los Angeles SD µr 0.04 3.33 -- -- 0.06 2.38
ny 4.06 2.51 16.9 0.42 -- --
µp -- -- 90.7 0.19 0.26 0.33
Design SE µr 0.01 5.26 -- -- 0.06 1.85
ny 3.79 3.04 17.5 0.54 -- --
µp -- -- 38.6 0.18 0.40 0.45
NF µr 0.03 5.56 -- -- 21.9 0.21
ny 2.53 2.23 0.05 4.76 -- --
a Λ = gΛ 1 ⁄ h
j i (Eq. 3.44 repeated)
107
Table 9.4: Regression coefficients g and h: SD, BE, and BP hysteresis types, IND spectra
Regression Coefficientsa
a Λ = gΛ 1 ⁄ h
j i (Eq. 3.44 repeated)
108
Table 9.5: Correlation coefficient, ρ: EP hysteresis type, IND spectra
Hysteresis Demand Site Soil/ Correlation Coefficient, ρ
α Site Seismicity
Type Level Distance µ µp µr ny
109
Table 9.6: Correlation coefficient, ρ: SD, BE, and BP hysteresis types, IND spectra
Hysteresis Site Demand Site Soil/ Correlation Coefficient, ρ
α, βs, βr
Type Seismicity Level Distance µ µp µr ny
µ 1.00 0.85 0.86 0.84
µp 0.85 1.00 0.64 0.93
Boston Survival SD
µr 0.86 0.64 1.00 0.72
ny 0.84 0.93 0.72 1.00
µ 1.00 0.78 0.94 0.62
µp 0.78 1.00 0.80 0.93
Survival SD
µr 0.94 0.80 1.00 0.71
ny 0.62 0.93 0.71 1.00
SD α = 0.10
µ 1.00 0.73 0.89 0.53
µp 0.73 1.00 0.53 0.87
Los Angeles SD
µr 0.89 0.53 1.00 0.47
ny 0.53 0.87 0.47 1.00
Design
µ 1.00 0.92 0.90 0.58
µp 0.92 1.00 0.80 0.77
SE
µr 0.90 0.80 1.00 0.64
ny 0.58 0.77 0.64 1.00
µ 1.00 0.96 0.00 0.60
µp 0.96 1.00 0.00 0.79
Boston Survival SD
µr 0.00 0.00 1.00 0.00
ny 0.60 0.79 0.00 1.00
µ 1.00 0.92 0.00 0.50
µp 0.92 1.00 0.00 0.78
Survival SD
µr 0.00 0.00 1.00 0.00
ny 0.50 0.78 0.00 1.00
BE α = 0.10
µ 1.00 0.92 0.00 0.44
µp 0.92 1.00 0.00 0.74
Los Angeles SD
µr 0.00 0.00 1.00 0.00
ny 0.44 0.74 0.00 1.00
Design
µ 1.00 0.95 0.00 0.64
µp 0.95 1.00 0.00 0.82
SE
µr 0.00 0.00 1.00 0.00
ny 0.64 0.82 0.00 1.00
µ 1.00 0.94 0.69 0.45
α = 0.10 µp 0.94 1.00 0.54 0.69
Los Angeles Survival SD
βs = βr = 1/6 µr 0.69 0.54 1.00 0.01
ny 0.45 0.69 0.01 1.00
µ 1.00 0.98 0.52 0.79
µp 0.98 1.00 0.44 0.88
Boston Survival SD
µr 0.52 0.44 1.00 0.34
ny 0.79 0.88 0.34 1.00
µ 1.00 0.96 0.62 0.47
µp 0.96 1.00 0.51 0.68
Survival SD
µr 0.62 0.51 1.00 0.10
α = 0.10 ny 0.47 0.68 0.10 1.00
BP βs = βr = 1/3 µ 1.00 0.97 0.49 0.61
µp 0.97 1.00 0.34 0.78
Los Angeles SD
µr 0.49 0.34 1.00 0.04
ny 0.61 0.78 0.04 1.00
Design
µ 1.00 0.98 0.54 0.90
µp 0.98 1.00 0.45 0.91
SE
µr 0.54 0.45 1.00 0.26
ny 0.90 0.91 0.26 1.00
µ 1.00 0.97 0.53 0.51
α = 0.10 µp 0.97 1.00 0.42 0.69
Los Angeles Survival SD
βs = βr = 1/2 µr 0.53 0.42 1.00 0.03
ny 0.51 0.69 0.03 1.00
110
10 d = 1.41, ρf == 0.72
0.99
10 d = 0.85, ρf == 1.17
0.97
10 d = 2.59, fρ==1.23
0.90
10 d = 1.69, ρf == 0.71
0.99
10 d = 0.71, ρf == 2.16
0.79
10 d = 2.13, fρ==0.97
0.86
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 d = 1.41, ρf == 0.72
0.99
50 g = 1.56, ρh == 0.96
0.57
50 g = 0.26, ρh == 0.34
0.89
50 d = 1.69, ρf == 0.71
0.99
50 g = 4.63, ρh == 0.77
0.31
50 g = 0.27, ρh == 0.38
0.84
µp µp
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.97 ρ = 0.96 ρ = 0.95 ρ = 0.79 ρ = 0.77 ρ = 0.70
4 4 4 4 4 4
d = 0.85, f = 1.17 g = 0.64, h = 1.75 g = 0.24, h = 0.70 d = 0.71, f = 2.16 g = 0.22, h = 3.23 g = 0.18, h = 1.75
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.90 ρ = 0.89 ρ = 0.95 ρ = 0.86 ρ = 0.84 ρ = 0.70
25 25 25 25 25 25
d = 2.59, f = 1.23 g = 3.92, h = 2.93 g = 4.10, h = 1.42 d = 2.13, f = 0.97 g = 3.73, h = 2.65 g = 5.63, h = 0.57
ny n
y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
10 d = 1.91, ρf == 0.69
0.99
10 d = 0.47, ρf == 2.84
0.67
10 d = 1.77, fρ==0.81
0.78
0 0 0
0 50 0 4 0 25
ρ = 0.99 ρ = 0.64 ρ = 0.81
50 d = 1.91, f = 0.69
50 g = 45.7, h = 0.23
50 g = 0.28, h = 0.40
µ
p
0 0 0
0 10 0 4 0 25
ρ = 0.67 ρ = 0.64 ρ = 0.48
4 4 4
d =0.47, f = 2.84 g = 0.02, h = 4.35 g = 0.05, h = 2.63
µ
r
0 0 0
0 10 0 50 0 25
ρ = 0.78 ρ = 0.81 ρ = 0.48
25 25 25
d = 1.77, f = 0.81 g = 3.59, h = 2.49 g = 21.0, h = 0.38
ny
0 0 0
0 10 0 50 0 4
data
(c) regression
Figure 9.1: Matrix plots of cross-correlations: effect of post-yield stiffness ratio, α (EP
hysteresis type, SAC Los Angeles, survival-level, SD soil): (a) α = 0.00; (b) α = 0.05; (c)
α = 0.10.
111
10 d = 1.80, ρf == 0.48
0.78
10 d = 0.37, ρf == 1.82
0.94
10 d = 3.86, fρ==3.02
0.62 10 d = 4.31, ρf == 0.68
0.92
10 d = 0.00, ρf == 0.00
0.00
10 d = 0.19, fρ==0.36
0.50
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 ρ = 0.78
50 ρ = 0.80
50 ρ = 0.93 50 d = 4.31, ρf == 0.68
0.92
50 g = 0.00, ρh == 0.00
0.00
50 g = 1.18, ρh == 0.74
0.78
d = 1.80, f = 0.48 g = 67.8, h = 0.38 g = 1.85, h = 0.66
µp
µ
p
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.94 ρ = 0.80 ρ = 0.71 ρ = 0.00 ρ = 0.00 ρ = 0.00
4 4 4 4 4 4
d =0.37, f = 1.82 g = 0.01, h = 2.63 g = 0.11, h = 1.82 d =0.00, f = 0.00 g = 0.00, h = 0.00 g = 0.00, h = 0.00
µ µr
r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.62 ρ = 0.93 ρ = 0.71 ρ = 0.50 ρ = 0.78 ρ = 0.00
25 25 25 25 25 25
d = 3.86, f = 3.02 g = 0.54, h = 1.51 g = 9.17, h = 0.55 d = 0.19, f = 0.36 g = 0.85, h = 1.35 g = 0.00, h = 0.00
ny n
y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
(a) regression (b) regression
Figure 9.2: Matrix plots of cross-correlations: effect of hysteresis type (α = 0.10, SAC Los
Angeles, survival-level, SD soil): (a) SD hysteresis type; (b) BE hysteresis type.
112
10 d = 5.04, ρf == 0.83
0.94
10 d = 0.02, ρf == 1.07
0.69
10 d = 0.13, fρ==0.32
0.45
10 d = 4.71, ρf == 0.86
0.96
10 d = 0.03, ρf == 1.87
0.62
10 d = 0.20, fρ==0.35
0.47
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 d = 5.04, ρf == 0.83
0.94
50 g = 307, hρ == 1.25
0.54
50 g = 0.63, ρh == 0.49
0.69
50 d = 4.71, ρf == 0.86
0.96
50 g = 297, hρ == 1.19
0.51
50 g = 0.32, ρh == 0.43
0.68
µ µp
p
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.69 ρ = 0.54 ρ = 0.01 ρ = 0.62 ρ = 0.51 ρ = 0.10
4 4 4 4 4 4
d =0.02, f = 1.07 g = 0.01 g = 0.003 d =0.03, f = 1.87 g = 0.003 g = 0.004
h = 0.80 h = 1.04 h = 0.84 h = 1.06
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.45 ρ = 0.69 ρ = 0.01 ρ = 0.47 ρ = 0.68 ρ = 0.10
25 25 25 25 25 25
d = 0.13, f = 0.32 g = 2.54, h = 2.03 g = 272, h = 0.96 d = 0.20, f = 0.35 g = 3.17, h = 2.35 g = 250, h = 0.94
ny n
y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
(a) regression (b) regression
10 d = 3.99, ρf == 0.82
0.97
10 d = 0.05, ρf == 2.69
0.53
10 d = 0.32, fρ==0.40
0.51
0 0 0
0 50 0 4 0 25
ρ = 0.97 ρ = 0.42 ρ = 0.69
50 d = 3.99, f = 0.82
50 g = 332, h = 1.05
50 g = 0.29, h = 0.40
µ
p
0 0 0
0 10 0 4 0 25
ρ = 0.53 ρ = 0.42 ρ = 0.03
4 4 4
d = 0.05, f = 2.69 g = 0.003 g = 0.004
h = 0.95 h = 1.15
µ
r
0 0 0
0 10 0 50 0 25
ρ = 0.51 ρ = 0.69 ρ = 0.03
25 25 25
d = 0.32, f = 0.40 g = 3.49, h = 2.53 g = 263, h = 0.87
ny
0 0 0
0 10 0 50 0 4
data
(c) regression
Figure 9.3: Matrix plots of cross-correlations: effect of βs = βr (BP hysteresis type, α = 0.10,
SAC Los Angeles, survival-level, SD soil): (a) βs = βr = 1/6; (b) βs = βr = 1/3; (c) βs = βr = 1/
2.
113
10 d = 2.30, ρf == 0.71
0.98
10 d = 0.46, ρf == 2.10
0.72
10 d = 2.44, fρ==0.88
0.78
10 d = 2.13, ρf == 0.70
0.99
10 d = 0.45, ρf == 2.98
0.58
10 d = 4.71, fρ==2.04
0.96
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
ρ = 0.98 ρ = 0.67 ρ = 0.86 ρ = 0.99 ρ = 0.49 ρ = 0.98
50 d = 2.30, f = 0.71
50 g = 24.1, h = 0.30
50 g = 0.25, h = 0.40
50 d = 2.13, f = 0.70
50 g = 90.7, h = 0.19
50 g = 0.26, h = 0.33
µ µ
p p
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.72 ρ = 0.67 ρ = 0.54 ρ = 0.58 ρ = 0.49 ρ = 0.46
4 4 4 4 4 4
d =0.46, f = 2.10 g = 0.04, h = 3.33 g = 0.06, h = 2.38 d =0.45, f = 2.98 g = 0.01, h = 5.26 g = 0.06, h = 1.85
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.78 ρ = 0.86 ρ = 0.54 ρ = 0.96 ρ = 0.98 ρ = 0.46
25 25 25 25 25 25
d = 2.44, f = 0.88 g = 4.06, h = 2.51 g = 16.9, h = 0.42 d = 4.71, f = 2.04 g = 3.79, h = 3.04 g = 17.5, h = 0.54
ny ny
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.4: Matrix plots of cross-correlations: effect of site soil characteristics (EP hystere-
sis type, α = 0.10, SAC Los Angeles, design-level): (a) SD soil; (b) SE soil.
114
10 d = 3.61, ρf == 0.58
0.73
10 d = 0.28, ρf == 1.30
0.89
10 d = 5.21, fρ==4.69
0.53
10 d = 7.16, ρf == 0.88
0.92
10 d = 0.32, ρf == 1.80
0.90
10 d = 3.20, fρ==3.44
0.58
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 d = 3.61, ρf == 0.58
0.73
50 g = 79.8, ρh == 0.53
0.33
50 g = 1.23, ρh == 0.57
0.87
50 d = 7.16, ρf == 0.88
0.92
50 g = 74.6, ρh == 0.80
0.53
50 g = 1.39, ρh == 0.53
0.77
µp µp
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.89 ρ = 0.53 ρ = 0.47 ρ = 0.90 ρ = 0.80 ρ = 0.64
4 4 4 4 4 4
d =0.28, f = 1.30 g = 0.01, h = 3.03 g = 0.09, h = 2.38 d =0.32, f = 1.80 g = 0.01, h = 1.89 g = 0.12, h =1.45
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.53 ρ = 0.87 ρ = 0.47 ρ = 0.58 ρ = 0.77 ρ = 0.64
25 25 25 25 25 25
d = 5.21, f = 4.69 g = 0.81, h = 1.75 g = 11.3, h = 0.42 d = 3.20, f = 3.44 g = 0.72, h = 1.89 g = 8.43, h = 0.69
n n
y y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
(a) regression (b) regression
Figure 9.5: Matrix plots of cross-correlations: effect of site soil characteristics (SD hystere-
sis type, α = 0.10, SAC Los Angeles, design-level): (a) SD soil; (b) SE soil.
10 d = 5.41, ρf == 0.73
0.92
10 d = 0.00, ρf == 0.00
0.00
10 d = 0.63, fρ==0.45
0.44
10 d = 6.18, ρf == 0.84
0.95
10 d = 0.00, ρf == 0.00
0.00
10 d = 2.59, fρ==1.00
0.64
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
ρ = 0.92 ρ = 0.00 ρ = 0.74 ρ = 0.95 ρ = 0.00 ρ = 0.82
50 d = 5.41, f = 0.73
50 g = 0.00, h = 0.00
50 g = 0.37, h = 0.46
50 d = 6.18, f = 0.84
50 g = 0.00, h = 0.00
50 g = 0.30, h = 0.34
µp µp
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.00 ρ = 0.00 ρ = 0.00 ρ = 0.00 ρ = 0.00 ρ = 0.00
4 4 4 4 4 4
d =0.00, f = 0.00 g = 0.00, h = 0.00 g = 0.00, h = 0.00 d =0.00, f = 0.00 g = 0.00, h = 0.00 g = 0.00, h = 0.00
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.44 ρ = 0.74 ρ = 0.00 ρ = 0.64 ρ = 0.82 ρ = 0.00
25 25 25 25 25 25
d = 0.61, f = 0.45 g = 2.73, h = 2.19 g = 0.00, h = 0.00 d = 2.59, f = 1.00 g = 3.35, h = 2.91 g = 0.00, h = 0.00
ny ny
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.6: Matrix plots of cross-correlations: effect of site soil characteristics (BE hystere-
sis type, α = 0.10, SAC Los Angeles, design-level): (a) SD soil; (b) SE soil.
115
10 d = 3.46, ρf == 0.70
0.97
10 d = 0.06, ρf == 1.49
0.49
10 d = 1.91, fρ==0.74
0.61
10 d = 3.50, ρf == 0.74
0.98
10 d = 0.05, ρf == 1.71
0.54
10 d = 6.50, fρ==2.78
0.90
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 d =3.46, fρ==0.70
0.97
50 g = 578, hρ == 0.67
0.34
50 g = 0.23, ρh == 0.37
0.78
50 d =3.50, fρ==0.74
0.98
50 g = 433, hρ == 0.79
0.45
50 g = 0.64, ρh == 0.31
0.91
µp µp
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.49 ρ = 0.34 ρ = 0.04 ρ = 0.54 ρ = 0.45 ρ = 0.26
4 4 4 4 4 4
d =0.06, f = 1.49 g = 0.002 g = 0.003 d =0.05, f = 1.71 g = 0.00 g = 0.002
h = 1.49 h = 1.64 h =1.27 h = 1.27
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.61 ρ = 0.78 ρ = 0.04 ρ = 0.90 ρ = 0.91 ρ = 0.26
25 25 25 25 25 25
d =1.91, f = 0.74 g = 4.27, h = 2.73 g = 376, h = 0.61 d =6.50, f = 2.78 g = 4.27, h = 3.26 g = 161, h = 0.79
n n
y y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.7: Matrix plots of cross-correlations: effect of site soil characteristics (BP hystere-
sis type, α = 0.10, βs = βr = 1/3, SAC Los Angeles, design-level): (a) SD soil; (b) SE soil.
10 d = 1.91, ρf == 0.69
0.99
10 d = 0.47, ρf == 2.84
0.67
10 d = 1.77, fρ==0.81
0.78
10 d = 2.30, ρf == 0.71
0.98
10 d = 0.46, ρf == 2.10
0.72
10 d = 2.44, fρ==0.88
0.78
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
ρ = 0.99 ρ = 0.64 ρ = 0.81 ρ = 0.98 ρ = 0.67 ρ = 0.86
50 d = 1.91, f = 0.69
50 g = 45.7, h = 0.23
50 g = 0.28, h = 0.40
50 d = 2.30, f = 0.71
50 g = 24.1, h = 0.30
50 g = 0.25, h = 0.40
µ µ
p p
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.67 ρ = 0.64 ρ = 0.48 ρ = 0.72 ρ = 0.67 ρ = 0.54
4 4 4 4 4 4
d =0.47, f = 2.84 g = 0.02, h = 4.35 g = 0.05, h = 2.63 d =0.46, f = 2.10 g = 0.04, h = 3.33 g = 0.06, h = 2.38
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.78 ρ = 0.81 ρ = 0.48 ρ = 0.78 ρ = 0.86 ρ = 0.54
25 25 25 25 25 25
d = 1.77, f = 0.81 g = 3.59, h = 2.49 g = 21.0, h = 0.38 d = 2.44, f = 0.88 g = 4.06, h = 2.51 g = 16.9, h = 0.42
ny ny
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
(a) regression (b) regression
Figure 9.8: Matrix plots of cross-correlations: effect of seismic demand level (EP hysteresis
type, α = 0.10, SAC Los Angeles, SD soil): (a) survival-level; (b) design-level.
116
10 d = 2.46, ρf == 0.77
0.99
10 d = 0.40, ρf == 3.35
0.56
10 d = 4.77, fρ==2.83
0.94 10 d = 2.13, ρf == 0.70
0.99
10 d = 0.45, ρf == 2.98
0.58
10 d = 4.71, fρ==2.04
0.96
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
ρ = 0.99 ρ = 0.49 ρ = 0.98
50 d = 2.46, ρf == 0.77
0.99
50 g = 189, hρ == 0.20
0.51
50 g = 0.27, ρh == 0.28
0.94 50 d = 2.13, f = 0.70
50 g = 90.7, h = 0.19
50 g = 0.26, h = 0.33
µp µ
p
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.56 ρ = 0.51 ρ = 0.47 ρ = 0.58 ρ = 0.49 ρ = 0.46
4 4 4 4 4 4
d =0.40, f = 3.35 g = 0.005 g = 0.06, h = 1.49 d =0.45, f = 2.98 g = 0.01, h = 5.26 g = 0.06, h = 1.85
h = 5.00
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.94 ρ = 0.94 ρ = 0.47 ρ = 0.96 ρ = 0.98 ρ = 0.46
25 25 25 25 25 25
d = 4.77, f = 2.83 g = 3.68, h = 3.63 g = 17.0, h = 0.67 d = 4.71, f = 2.04 g = 3.79, h = 3.04 g = 17.5, h = 0.54
n ny
y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.9: Matrix plots of cross-correlations: effect of seismic demand level (EP hysteresis
type, α = 0.10, SAC Los Angeles, SE soil): (a) survival-level; (b) design-level.
10 d = 1.80, ρf == 0.48
0.78
10 d = 0.37, ρf == 1.82
0.94
10 d = 3.86, fρ==3.02
0.62 10 d = 3.61, ρf == 0.58
0.73
10 d = 0.28, ρf == 1.30
0.89
10 d = 5.21, fρ==4.69
0.53
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
ρ = 0.73 ρ = 0.53 ρ = 0.87
50 d = 1.80, ρf == 0.48
0.78
50 g = 67.8, ρh == 0.80
0.38
50 g = 1.85, ρh == 0.66
0.93 50 d = 3.61, f = 0.58
50 g = 79.8, h = 0.33
50 g = 1.23, h = 0.57
µp
µ
p
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.94 ρ = 0.80 ρ = 0.71 ρ = 0.89 ρ = 0.53 ρ = 0.47
4 4 4 4 4 4
d =0.37, f = 1.82 g = 0.01, h = 2.63 g = 0.11, h = 1.82 d =0.28, f = 1.30 g = 0.01, h = 3.03 g = 0.09, h = 2.38
µ µ
r
r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.62 ρ = 0.93 ρ = 0.71 ρ = 0.53 ρ = 0.87 ρ = 0.47
25 25 25 25 25 25
d = 3.86, f = 3.02 g = 0.54, h = 1.51 g = 9.17, h = 0.55 d = 5.21, f = 4.69 g = 0.81, h = 1.75 g = 11.3, h = 0.42
ny ny
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
(a) regression (b) regression
Figure 9.10: Matrix plots of cross-correlations: effect of seismic demand level (SD hystere-
sis type, α = 0.10, SAC Los Angeles, SD soil): (a) survival-level; (b) design-level.
117
10 d = 4.31, ρf == 0.68
0.92
10 d = 0.00, ρf == 0.00
0.00
10 d = 0.19, fρ==0.36
0.50
10 d = 5.41, ρf == 0.73
0.92
10 d = 0.00, ρf == 0.00
0.00
10 d = 0.63, fρ==0.45
0.44
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 d = 4.31, ρf == 0.68
0.92
50 g = 0.00, ρh == 0.00
0.00
50 g = 1.18, ρh == 0.74
0.78
50 ρ = 0.92
50 ρ = 0.00
50 ρ = 0.74
d = 5.41, f = 0.73 g = 0.00, h = 0.00 g = 0.37, h = 0.46
µp µp
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.00 ρ = 0.00 ρ = 0.00 ρ = 0.00 ρ = 0.00 ρ = 0.00
4 4 4 4 4 4
d =0.00, f = 0.00 g = 0.00, h = 0.00 g = 0.00, h = 0.00 d =0.00, f = 0.00 g = 0.00, h = 0.00 g = 0.00, h = 0.00
µr µ
r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.50 ρ = 0.78 ρ = 0.00 ρ = 0.44 ρ = 0.74 ρ = 0.00
25 25 25 25 25 25
d = 0.19, f = 0.36 g = 0.85, h = 1.35 g = 0.00, h = 0.00 d = 0.61, f = 0.45 g = 2.73, h = 2.19 g = 0.00, h = 0.00
n n
y y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.11: Matrix plots of cross-correlations: effect of seismic demand level (BE hystere-
sis type, α = 0.10, SAC Los Angeles, SD soil): (a) survival-level; (b) design-level.
118
10 d = 3.46, ρf == 0.70
0.97
10 d = 0.06, ρf == 1.49
0.49
10 d = 1.91, fρ==0.74
0.61
10 d = 4.71, ρf == 0.86
0.96
10 d = 0.03, ρf == 1.87
0.62
10 d = 0.20, fρ==0.35
0.47
µ
µ
0 0 0
0 0 0 0 50 0 4 0 25
0 50 0 4 0 25
ρ = 0.51 ρ = 0.68 50 d =3.46, fρ==0.70
0.97
50 g = 578, hρ == 0.67
0.34
50 g = 0.23, ρh == 0.37
0.78
50 d = 4.71, ρf == 0.86
0.96
50 g = 297, h = 1.19
50 g = 0.32, h = 0.43
µp
µp
0 0 0 0 0 0
0 10 0 4 0 25
0 10 0 4 0 25
ρ = 0.49 ρ = 0.34 ρ = 0.04
ρ = 0.62 ρ = 0.51 ρ = 0.10 4 4 4
4 4 4 d =0.06, f = 1.49 g = 0.002 g = 0.003
d =0.03, f = 1.87 g = 0.003 g = 0.004 h = 1.49 h = 1.64
h = 0.84 h = 1.06
µ
µr r
0 0 0
0 0 0 0 10 0 50 0 25
0 10 0 50 0 25 ρ = 0.61 ρ = 0.78 ρ = 0.04
ρ = 0.47 ρ = 0.68 ρ = 0.10 25 25 25
25 25 25 d =1.91, f = 0.74 g = 4.27, h = 2.73 g = 376, h = 0.61
d = 0.20, f = 0.35 g = 3.17, h = 2.35 g = 250, h = 0.94
n
n y
y
0 0 0
0 0 0 0 10 0 50 0 4
0 10 0 50 0 4
data
data regression
(a) regression (b)
Figure 9.12: Matrix plots of cross-correlations: effect of seismic demand level (BP hystere-
sis type, α = 0.10, βs = βr = 1/3, SAC Los Angeles, SD soil): (a) survival-level; (b) design-
level.
10 d = 1.91, ρf == 0.69
0.99
10 d = 0.47, ρf == 2.84
0.67
10 d = 1.77, fρ==0.81
0.78
10 d = 2.78, ρf == 0.69
0.99
10 d = 0.46, ρf == 2.62
0.75
10 d = 4.62, fρ==1.12
0.92
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
ρ = 0.99 ρ = 0.64 ρ = 0.81 ρ = 0.99 ρ = 0.70 ρ = 0.95
50 d = 1.91, f = 0.69
50 g = 45.7, h = 0.23
50 g = 0.28, h = 0.40
50 d = 2.78, f = 0.69
50 g = 60.7, h = 0.42
50 g = 0.23, h = 0.43
µ µ
p p
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.67 ρ = 0.64 ρ = 0.48 ρ = 0.75 ρ = 0.70 ρ = 0.67
4 4 4 4 4 4
d =0.47, f = 2.84 g = 0.02, h = 4.35 g = 0.05, h = 2.63 d = 0.46, f = 2.62 g = 0.02, h = 4.17 g = 0.04, h = 1.89
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.78 ρ = 0.81 ρ = 0.48 ρ = 0.92 ρ = 0.95 ρ = 0.67
25 25 25 25 25 25
d = 1.77, f = 0.81 g = 3.59, h = 2.49 g = 21.0, h = 0.38 d = 4.62, f = 1.12 g = 4.42, h = 2.31 g = 26.7, h = 0.53
ny ny
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
(a) regression (b) regression
Figure 9.13: Matrix plots of cross-correlations: effect of site seismicity (EP hysteresis type,
α = 0.10, SAC survival-level, SD soil): (a) Los Angeles; (b) Boston.
119
10 d = 2.46, ρf == 0.77
0.99
10 d = 0.40, ρf == 3.35
0.56
10 d = 4.77, fρ==2.83
0.94
10 ρ = 0.97
d = 3.07, f = 0.68
10 d = 0.37, ρf == 2.63
0.48
10 d = 6.69, fρ==1.91
0.90
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 d = 2.46, ρf == 0.77
0.99
50 g = 189, hρ == 0.20
0.51
50 g = 0.27, ρh == 0.28
0.94
50 d = 3.07, ρf == 0.68
0.97
50 g = 3.64, ρh == 0.36
0.18
50 g = 0.20, ρh == 0.32
0.95
µp µp
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.56 ρ = 0.51 ρ = 0.47 ρ = 0.48 ρ = 0.36 ρ = 0.25
4 4 4 4 4 4
d =0.40, f = 3.35 g = 0.005 g = 0.06, h = 1.49 d =0.37, f = 2.63 g = 0.003 g = 0.01, h = 3.13
h = 5.00 h = 5.56
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.94 ρ = 0.94 ρ = 0.47 ρ = 0.90 ρ = 0.95 ρ = 0.25
25 25 25 25 25 25
d = 4.77, f = 2.83 g = 3.68, h = 3.63 g = 17.0, h = 0.67 d = 6.69, f = 1.91 g = 5.00, h = 3.13 g = 73.7, h = 0.32
n n
y y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.14: Matrix plots of cross-correlations: effect of site seismicity (EP hysteresis type,
α = 0.10, SAC survival-level, SE soil): (a) Los Angeles; (b) Boston.
10 d = 2.30, ρf == 0.71
0.98
10 d = 0.46, ρf == 2.10
0.72
10 d = 2.44, fρ==0.88
0.78
10 d = 2.76, ρf == 0.68
0.99
10 d = 0.51, ρf == 2.91
0.73
10 d = 5.79, fρ==1.30
0.93
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 d = 2.30, ρf == 0.71
0.98
50 g = 24.1, ρh == 0.67
0.30
50 g = 0.25, ρh == 0.40
0.86
50 d = 2.76, ρf == 0.68
0.99
50 g = 58.2, ρh == 0.68 50 g = 0.22, ρh == 0.43
0.96
0.21
µ µp
p
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.72 ρ = 0.67 ρ = 0.54
4 4 4 4 ρ = 0.73
4 ρ = 0.68
4 ρ = 0.61
d =0.46, f = 2.10 g = 0.04, h = 3.33 g = 0.06, h = 2.38
d =0.51, f = 2.91 g = 0.02, h = 4.76 g = 0.03, h = 2.50
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.78 ρ = 0.86 ρ = 0.54
25 25 25 25 ρ = 0.93
25 ρ = 0.96
25 ρ = 0.61
d = 2.44, f = 0.88 g = 4.06, h = 2.51 g = 16.9, h = 0.42 d = 5.79, f = 1.301 g = 4.59, h = 2.34 g = 28.9, h = 0.40
ny ny
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.15: Matrix plots of cross-correlations: effect of site seismicity (EP hysteresis type,
α = 0.10, SAC design-level, SD soil): (a) Los Angeles; (b) Boston.
120
10 d = 2.13, ρf == 0.70
0.99
10 d = 0.45, ρf == 2.98
0.58
10 d = 4.71, fρ==2.04
0.96
10 d = 2.87, ρf == 0.70
0.99
10 d = 0.53, ρf == 3.61
0.68
10 d = 5.50, fρ==1.42
0.96
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 d = 2.13, ρf == 0.70
0.99
50 g = 90.7, ρh == 0.49
0.19
50 g = 0.26, ρh == 0.33
0.98
50 d = 2.87, ρf == 0.70
0.99
50 g = 91.0, ρh == 0.65
0.17
50 g = 0.25, ρh == 0.42
0.98
µ µp
p
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.58 ρ = 0.49 ρ = 0.46 ρ = 0.68 ρ = 0.65 ρ = 0.65
4 4 4 4 4 4
d =0.45, f = 2.98 g = 0.01, h = 5.26 g = 0.06, h = 1.85 d =0.53, f = 3.61 g = 0.01, h = 5.88 g = 0.04, h =2.38
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.96 ρ = 0.98 ρ = 0.46 ρ = 0.96 ρ = 0.98 ρ = 0.65
25 25 25 25 25 25
d = 4.71, f = 2.04 g = 3.79, h = 3.04 g = 17.5, h = 0.54 d = 5.50, f = 1.42 g = 4.05, h = 2.36 g = 26.4, h = 0.42
ny n
y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
(a) regression (b) regression
Figure 9.16: Matrix plots of cross-correlations: effect of site seismicity (EP hysteresis type,
α = 0.10, SAC design-level, SE soil): (a) Los Angeles; (b) Boston.
10 d = 1.80, ρf == 0.48
0.78
10 d = 0.37, ρf == 1.82
0.94
10 d = 3.86, fρ==3.02
0.62 10 d = 7.88, ρf == 0.76
0.85
10 d = 0.36, ρf == 2.08
0.86
10 d = 8.32, fρ==3.89
0.84
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
ρ = 0.85 ρ = 0.64 ρ = 0.93
50 d = 1.80, ρf == 0.48
0.78
50 g = 67.8, ρh == 0.80
0.38
50 g = 1.85, ρh == 0.66
0.93 50 d = 7.88, f = 0.76
50 g = 170, h = 0.32
50 g = 0.29, h = 0.31
µp
µ
p
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.94 ρ = 0.80 ρ = 0.71 ρ = 0.86 ρ = 0.64 ρ = 0.72
4 4 4 4 4 4
d =0.37, f = 1.82 g = 0.01, h = 2.63 g = 0.11, h = 1.82 d =0.36, f = 2.08 g = 0.006 g = 0.06, h = 0.96
h = 3.13
µ µ
r
r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.62 ρ = 0.93 ρ = 0.71 ρ = 0.84 ρ = 0.93 ρ = 0.72
25 25 25 25 25 25
d = 3.86, f = 3.02 g = 0.54, h = 1.51 g = 9.17, h = 0.55 d = 8.32, f = 3.89 g = 3.44, h = 3.19 g = 17.1, h = 1.04
ny ny
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
(a) regression (b) regression
Figure 9.17: Matrix plots of cross-correlations: effect of site seismicity (SD hysteresis type,
α = 0.10, SAC survival-level, SD soil): (a) Los Angeles; (b) Boston.
121
10 d = 4.31, ρf == 0.68
0.92
10 d = 0.00, ρf == 0.00
0.00
10 d = 0.19, fρ==0.36
0.50
10 d = 6.70, ρf == 0.76
0.96
10 d = 0.00, ρf == 0.00
0.00
10 d = 1.41, fρ==0.63
0.60
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 d = 4.31, ρf == 0.68
0.92
50 g = 0.00, ρh == 0.00
0.00
50 g = 1.18, ρh == 0.74
0.78
50 d = 6.70, ρf == 0.76
0.96
50 g = 0.00, ρh == 0.00 50 g = 0.23, ρh == 0.34
0.79
0.00
µp µp
0 0 0
0 10 0 4 0 25 0 0 0
0 10 0 4 0 25
ρ = 0.00 ρ = 0.00 ρ = 0.00
4 4 4 4 ρ = 0.00
4 ρ = 0.00
4 ρ = 0.00
d =0.00, f = 0.00 g = 0.00, h = 0.00 g = 0.00, h = 0.00
d =0.00, f = 0.00 g = 0.00, h = 0.00 g = 0.00, h = 0.00
µr µ
r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.50 ρ = 0.78 ρ = 0.00
25 25 25 25 ρ = 0.60
25 ρ = 0.79
25 ρ = 0.00
d = 0.19, f = 0.36 g = 0.85, h = 1.35 g = 0.00, h = 0.00
d = 1.41, f = 0.63 g = 4.33, h = 2.95 g = 0.00, h = 0.00
n n
y y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression
(a) (b) regression
Figure 9.18: Matrix plots of cross-correlations: effect of site seismicity (BE hysteresis type,
α = 0.10, SAC survival-level, SD soil): (a) Los Angeles; (b) Boston.
10 d = 4.71, ρf == 0.86
0.96
10 d = 0.03, ρf == 1.87
0.62
10 d = 0.20, fρ==0.35
0.47
10 d = 4.49, ρf == 0.72
0.98
10 d = 0.05, ρf == 2.80
0.52
10 d = 4.56, fρ==1.09
0.79
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
ρ = 0.96 ρ = 0.51 ρ = 0.68 ρ = 0.98 ρ = 0.44 ρ = 0.88
50 d = 4.71, f = 0.86
50 g = 297, h = 1.19
50 g = 0.32, h = 0.43
50 d = 4.49, f = 0.72
50 g = 446, h = 1.12
50 g = 0.18, h = 0.33
µp µp
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.62 ρ = 0.51 ρ = 0.10 ρ = 0.52 ρ = 0.44 ρ = 0.34
4 4 4 4 4 4
d =0.03, f = 1.87 g = 0.003 g = 0.004 d =0.05, f = 2.80 g = 0.002 g = 0.004
h = 0.84 h = 1.06 h = 0.89 h = 0.99
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.47 ρ = 0.68 ρ = 0.10 ρ = 0.79 ρ = 0.88 ρ = 0.34
25 25 25 25 25 25
d = 0.20, f = 0.35 g = 3.17, h = 2.35 g = 250, h = 0.94 d = 4.56, f = 1.09 g = 5.43, h = 3.01 g = 227, h = 1.01
n ny
y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
(a) regression (b) regression
Figure 9.19: Matrix plots of cross-correlations: effect of site seismicity (BP hysteresis type,
α = 0.10, βs = βr = 1/3, SAC survival-level, SD soil): (a) Los Angeles; (b) Boston.
122
10 d = 2.30, ρf == 0.71
0.98
10 d = 0.46, ρf == 2.10
0.72
10 d = 2.44, fρ==0.88
0.78
10 d = 1.70, ρf == 0.71
0.99
10 d = 0.56, ρf == 3.80
0.62
10 d = 1.39, fρ==0.85
0.80
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
ρ = 0.98 ρ = 0.67 ρ = 0.86
50 d = 2.30, f = 0.71
50 g = 24.1, h = 0.30
50 g = 0.25, h = 0.40 50 ρ = 0.99
50 ρ = 0.59
50 ρ = 0.86
d = 1.70, f = 0.71 g = 38.6, h = 0.18 g = 0.40, h = 0.45
µ µp
p
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.72 ρ = 0.67 ρ = 0.54
4 4 4 4 ρ = 0.62
4 ρ = 0.59
4 ρ = 0.33
d =0.46, f = 2.10 g = 0.04, h = 3.33 g = 0.06, h = 2.38
d =0.56, f = 3.80 g = 0.03, h = 5.56 g = 21.9, h = 0.21
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.78 ρ = 0.86 ρ = 0.54
25 25 25 25 ρ = 0.80
25 ρ = 0.86
25 ρ = 0.33
d = 2.44, f = 0.88 g = 4.06, h = 2.51 g = 16.9, h = 0.42
d = 1.39, f = 0.85 g = 2.53, h = 2.23 g = 0.05, h =4.76
ny n
y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.20: Matrix plots of cross-correlations: effect of epicentral distance (EP hysteresis
type, α = 0.10, SAC Los Angeles, design-level, SD soil): (a) far-field; (b) near-field.
9.2 Effect of Reference Response Spectra on the Relationships Between the Demand Indices
As discussed earlier in Section 3.5, several different reference response spectra can be
used to develop capacity-demand index relationships (i.e., IND, AVG, and DES response spectra).
The effect of using these different reference spectra on the R-µ-T relationships is discussed in
Chapter 7. This section investigates the effect of reference response spectra on the relationships
between the demand indices µ, µp, µr, and ny. For this purpose, Tables 9.7 through 9.10 provide
the d, f, g, and h regression coefficients based on the AVG and DES reference spectra. The corre-
lation coefficients, ρ, are listed in Tables 9.11 and 9.12 and the corresponding matrix plots are pro-
vided in Figures 9.21-9.32. The figures are organized in the same order that the results are
presented in Chapter 7: (1) regression relationships for regions with low-seismicity, Boston (Fig.
9.21-9.24); (2) regression relationships for regions with high seismicity, Los Angeles (Fig. 9.25-
9.28); (3) regression relationships for near-field conditions, NF (Fig. 9.29); and (4) regression
relationships for different hysteresis types, SD, BE, and BP (Fig. 9.30-9.32). In similar order, Fig-
ures 9.33-9.36 compare the Λ-µ regression relationships developed based on the AVG and DES
reference spectra with the relationships presented in Section 9.1 based on the IND spectra to
determine Fy = Felas/R.
It was demonstrated earlier in Chapter 7 that reference response spectra significantly affect
the R-µ-T relationships. Figures 9.33-9.36 show that the relationships between µ and the other
demand indices Λ = µp, µr, and ny are similarly dependent on the reference response spectra. The
123
Λ-µ relationships are significantly affected by reference response spectra for: (1) µ versus µp and
µ versus ny relationships for the Boston design-level SD and SE soil ground motions using the EP
hysteresis type (Figs. 9.33a,c); (2) µ versus ny relationship for the Los Angeles design-level SD
soil NF ground motions using the EP hysteresis type (Fig. 9.35); and (3) µ versus ny relationship
for the Los Angeles survival-level SD soil ground motions using the BE and BP hysteresis types
(Figs. 9.36b,c). In general, the differences between the regression relationships developed based
on the IND and DES spectra are larger than the differences between the relationships developed
based on the IND and AVG spectra.
It is concluded that the µp, µr, and ny demands should be estimated using regression rela-
tionships developed based on the reference response spectrum used in the estimation of µ.
124
Table 9.8: Regression coefficients d and f: DES spectra
Hysteresis Site Demand Site Soil/ Regression Coefficientsa
α, βs, βr
Type Seismicity Level Distance Demand Index, Λ d f
µp 1.09 0.52
SD µr 0.30 1.48
ny 1.32 0.65
Survival
µp 2.57 0.68
SE µr 0.44 4.48
ny 4.32 1.37
Boston
µp 0.38 0.41
SD µr 0.33 1.46
ny 0.46 0.47
Design
µp 0.63 0.49
SE µr 0.30 1.32
ny 1.20 0.71
µp 1.73 0.68
EP α = 0.10 SD µr 0.43 2.37
ny 1.80 0.88
Survival
µp 2.72 0.82
SE µr 0.28 1.78
ny 4.45 2.93
µp 1.49 0.63
Los Angeles SD µr 0.40 1.96
ny 1.56 0.78
µp 1.45 0.63
Design SE µr 0.45 2.66
ny 2.74 1.30
µp 2.42 0.81
NF µr 0.33 1.41
ny 0.02 0.25
µp 3.93 0.63
SD α = 0.10 Los Angeles Survival SD µr 0.27 1.32
ny 4.16 5.41
µp 6.06 0.84
BE α = 0.10 Los Angeles Survival SD µr 0 0
ny 1.17 0.66
µp 3.46 0.76
α = 0.10
BP Los Angeles Survival SD µr 0.03 1.67
βs = βr = 1/3
ny 1.09 0.62
125
Table 9.9: Regression coefficients g and h: AVG spectra
Regression Coefficientsa
g h g h g h
µp -- -- 25.4 0.37 0.39 0.58
SD
µr 0.04 2.70 -- -- 0.06 1.56
Survival ny 2.55 1.73 16.5 0.64 -- --
µp -- -- 86.7 0.29 0.29 0.41
SE µr 0.01 3.45 -- -- 0.04 1.79
Boston ny 3.47 2.41 26.4 0.56 -- --
µp -- -- 25.0 0.29 0.41 0.60
SD µr 0.04 3.45 -- -- 0.06 2.04
Design ny 2.43 1.68 17.8 0.49 -- --
µp -- -- 25.0 0.39 0.37 0.51
SE µr 0.04 2.56 -- -- 0.07 1.43
ny 2.73 1.95 14.0 0.70 -- --
EP α = 0.10
µp -- -- 23.7 0.34 0.36 0.48
SD µr 0.04 2.94 -- -- 0.07 2.33
Survival ny 2.74 2.09 14.6 0.43 -- --
µp -- -- 43.3 0.36 0.36 0.33
SE µr 0.02 2.78 -- -- 0.09 1.23
ny 2.78 3.00 11.2 0.81 -- --
Los Angeles
µp -- -- 34.3 0.24 0.35 0.49
SD µr 0.03 4.17 -- -- 0.04 3.45
ny 2.87 2.05 22.4 0.29 -- --
µp -- -- 52.2 0.26 0.30 0.36
Design SE µr 0.02 3.85 -- -- 0.07 1.43
ny 3.34 2.80 14.0 0.70 -- --
µp -- -- 13.4 0.49 0.60 0.58
NF µr 0.07 2.04 -- -- 0.13 2.22
ny 1.66 1.73 7.70 0.45 -- --
µp -- -- 62.5 0.41 1.64 0.60
SD α = 0.10 Los Angeles Survival SD µr 0.02 2.44 -- -- 0.12 2.00
ny 0.61 1.67 8.04 0.50 -- --
µp -- -- 0 0 0.89 0.65
BE α = 0.10 Los Angeles Survival SD µr 0 0 -- -- 0 0
ny 1.12 1.55 0 0 -- --
µp -- -- 327 1.13 0.49 0.52
BP α = 0.10 Los Angeles Survival SD µr 0.003 0.88 -- -- 0.004 1.10
βs = βr = 1/3
ny 2.03 1.91 226 0.91 -- --
a Λ = gΛ 1 ⁄ h
j i (Eq. 3.44 repeated)
126
Table 9.10: Regression coefficients g and h: DES spectra
Regression Coefficientsa
g h g h g h
µp -- -- 33.3 0.34 0.55 0.68
SD
µr 0.03 2.94 -- -- 0.05 2.17
Survival ny 1.82 1.47 20.7 0.46 -- --
µp -- -- 470 0.16 0.30 0.43
SE µr 0.002 6.25 -- -- 0.003 6.25
Boston ny 3.38 2.35 289 0.16 -- --
µp -- -- 22.2 0.22 0.51 0.66
SD µr 0.05 4.55 -- -- 0.05 4.17
Design ny 1.95 1.51 18.2 0.24 -- --
µp -- -- 16.3 0.33 0.45 0.56
SE µr 0.06 3.03 -- -- 0.09 2.17
ny 2.23 1.77 11.3 0.46 -- --
EP α = 0.10
µp -- -- 33.3 0.30 0.33 0.43
SD µr 0.03 3.33 -- -- 0.05 2.50
Survival ny 3.07 2.31 19.1 0.40 -- --
µp -- -- 58.6 0.34 0.31 0.29
SE µr 0.02 2.94 -- -- 0.13 0.003
ny 3.21 3.40 7.74 357 -- --
Los Angeles
µp -- -- 27.7 0.31 0.40 0.53
SD µr 0.04 3.23 -- -- 0.06 2.33
ny 2.49 1.87 16.8 0.43 -- --
µp -- -- 71.4 0.17 0.35 0.40
Design SE µr 0.01 5.88 -- -- 0.008 8.33
ny 2.89 2.53 119 0.12 -- --
µp -- -- 17.0 0.71 0.60 0.62
NF µr 0.06 1.41 -- -- 0.11 1.64
ny 1.67 1.62 9.36 0.61 -- --
µp -- -- 65.0 0.48 0.87 0.43
SD α = 0.10 Los Angeles Survival SD µr 0.02 2.08 -- -- 0.11 1.75
ny 1.15 2.32 8.95 0.57 -- --
µp -- -- 0 0 0.33 0.36
BE α = 0.10 Los Angeles Survival SD µr 0 0 -- -- 0 0
ny 3.07 2.79 0 0 -- --
µp -- -- 314 1.15 0.26 0.32
BP α = 0.10 Los Angeles Survival SD µr 0.003 0.87 -- -- 0.006 1.01
βs = βr = 1/3
ny 3.80 3.14 176 0.99 -- --
a Λ = gΛ 1 ⁄ h
j i (Eq. 3.44 repeated)
127
Table 9.11: Correlation coefficient, ρ: AVG spectra
Hysteresis Site Demand Site Soil/ Correlation Coefficient, ρ
α, βs, βr
Type Seismicity Level Distance µ µp µr ny
128
Table 9.12: Correlation coefficient, ρ: DES spectra
Hysteresis Site Demand Site Soil/ Correlation Coefficient, ρ
α, βs, βr
Type Seismicity Level Distance µ µp µr ny
129
10 d = 1.53, ρf == 0.59
0.97
10 d = 0.42, ρf == 1.94
0.66
10 d = 2.34, fρ==0.84
0.91 10 d = 0.38, ρf == 0.41
0.82
10 d = 0.33, ρf == 1.46
0.78
10 d = 0.46, fρ==0.47
0.72
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 d = 1.53, ρf == 0.59
0.97
50 g = 25.0, ρh == 0.55
0.29
50 g = 0.41, ρh == 0.60
0.97 50 d = 0.38, ρf == 0.41
0.82
50 g = 22.2, ρh == 0.75
0.22
50 g = 0.51, ρh == 0.66
0.98
µp µp
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.66 ρ = 0.55 ρ = 0.45 ρ = 0.78 ρ = 0.75 ρ = 0.65
4 4 4 4 4 4
d = 0.42, f = 1.94 g = 0.04, h = 3.45 g = 0.06, h = 2.04 d = 0.33, f = 1.46 g = 0.05, h = 4.55 g = 0.05, h = 4.17
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.91 ρ = 0.97 ρ = 0.45 ρ = 0.72 ρ = 0.98 ρ = 0.65
25 25 25 25 25 25
d = 2.34, f = 0.84 g = 2.43, h = 1.68 g = 17.8, h = 0.49 d = 0.46, f = 0.47 g = 1.95, h = 1.51 g = 18.2, h = 0.24
n n
y y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.21: Matrix plots of cross-correlations: effect of reference response spectrum
(EP hysteresis type, α = 0.10, SAC Boston, design-level, SD soil): (a) AVG spectrum; (b)
DES spectrum.
10 d = 1.75, ρf == 0.63
0.97
10 d = 0.32, ρf == 1.45
0.66
10 d = 2.30, fρ==0.85
0.92 10 d = 1.09, ρf == 0.52
0.95
10 d = 0.30, ρf == 1.48
0.70
10 d = 1.32, fρ==0.65
0.86
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 ρ = 0.97
50 ρ = 0.58
50 ρ = 0.97 50 d = 1.09, ρf == 0.52
0.95
50 g = 33.3, ρh == 0.62
0.34
50 g = 0.55, ρh == 0.68
0.96
d = 1.75, f = 0.63 g = 25.4, h = 0.37 g = 0.39, h = 0.58
µp µp
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.66 ρ = 0.58 ρ = 0.54 ρ = 0.70 ρ = 0.62 ρ = 0.53
4 4 4 4 4 4
d = 0.32, f = 1.45 g = 0.04, h = 2.70 g = 0.06, h = 1.56 d = 0.30, f = 1.48 g = 0.03, h = 2.94 g = 0.05, h = 2.17
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.92 ρ = 0.97 ρ = 0.54 ρ = 0.86 ρ = 0.96 ρ = 0.53
25 25 25 25 25 25
d = 2.30, f = 0.85 g = 2.55, h = 1.73 g = 16.5, h = 0.64 d = 1.32, f = 0.65 g = 1.82, h = 1.47 g = 20.7, h = 0.46
n n
y y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.22: Matrix plots of cross-correlations: effect of reference response spectrum
(EP hysteresis type, α = 0.10, SAC Boston, survival-level, SD soil): (a) AVG spectrum;
(b) DES spectrum.
130
10 d = 1.90, ρf == 0.65
0.98
10 d = 0.34, ρf == 1.49
0.76
10 d = 3.08, fρ==1.08
0.93
10 d = 0.63, ρf == 0.49
0.81
10 d = 0.30, ρf == 1.32
0.82
10 d = 1.20, fρ==0.71
0.68
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 d = 1.90, ρf == 0.65
0.98
50 g = 25.0, ρh == 0.70
0.39
50 g = 0.37, ρh == 0.51
0.97
50 d = 0.63, ρf == 0.49
0.81
50 g = 16.3, ρh == 0.89
0.33
50 g = 0.45, ρh == 0.56
0.96
µp µp
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.76 ρ = 0.70 ρ = 0.64 ρ = 0.82 ρ = 0.89 ρ = 0.79
4 4 4 4 4 4
d = 0.34, f = 1.49 g = 0.04, h = 2.56 g = 0.07, h = 1.43 d = 0.30, f = 1.32 g = 0.06, h = 3.03 g = 0.09, h = 2.17
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.93 ρ = 0.97 ρ = 0.64 ρ = 0.68 ρ = 0.96 ρ = 0.79
25 25 25 25 25 25
d = 3.08, f = 1.08 g = 2.73, h = 1.95 g = 14.0, h = 0.70 d = 1.20, f = 0.71 g = 2.23, h = 1.77 g = 11.3, h = 0.46
n n
y y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.23: Matrix plots of cross-correlations: effect of reference response spectrum
(EP hysteresis type, α = 0.10, SAC Boston, design-level, SE soil): (a) AVG spectrum; (b)
DES spectrum.
10 d = 2.42, ρf == 0.68
0.98
10 d = 0.30, ρf == 1.82
0.60
10 d = 4.24, fρ==1.39
0.94 10 d = 2.57, ρf == 0.68
0.96
10 d = 0.44, ρf == 4.48
0.26
10 d = 4.32, fρ==1.37
0.85
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 ρ = 0.98
50 ρ = 0.53
50 ρ = 0.97 50 d = 2.57, ρf == 0.68
0.96
50 g = 470, hρ == 0.16
0.15
50 g = 0.30, ρh == 0.43
0.95
d = 2.42, f = 0.68 g = 86.7, h = 0.29 g = 0.29, h = 0.41
µp µp
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.26 ρ = 0.15 ρ = 0.02
4 ρ = 0.60
4 ρ = 0.53
4 ρ = 0.45 4 4 4
d = 0.44, f = 4.48 g = 0.002 g = 0.003
d = 0.30, f = 1.82 g = 0.01, h = 3.45 g = 0.04, h = 1.79 h = 6.25 h = 6.25
µ µ
r
r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.85 ρ = 0.95 ρ = 0.02
25 ρ = 0.94
25 ρ = 0.97
25 ρ = 0.45 25 25 25
d = 4.32, f = 1.37 g = 3.38, h = 2.35 g = 289, h = 0.16
d = 4.24, f = 1.39 g = 3.47, h = 2.41 g = 16.5, h = 0.64
ny n
y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.24: Matrix plots of cross-correlations: effect of reference response spectrum
(EP hysteresis type, α = 0.10, SAC Boston, survival-level, SE soil): (a) AVG spectrum;
(b) DES spectrum.
131
10 d = 1.49, ρf == 0.63
0.97
10 d = 0.40, ρf == 1.96
0.85
10 d = 1.56, fρ==0.78
0.84
10 d = 1.69, ρf == 0.65
0.98
10 d = 0.45, ρf == 2.30
0.73
10 d = 1.87, fρ==0.84
0.84
µ
µ
0 0 0
0 0 0 0 50 0 4 0 25
0 50 0 4 0 25
ρ = 0.68 ρ = 0.91 50 d = 1.49, ρf == 0.63
0.97
50 g = 27.7, ρh == 0.84 50 g = 0.40, ρh == 0.53
0.93
50 d = 1.69, ρf == 0.65
0.98
50 g = 34.3, h = 0.24
50 g = 0.35, h = 0.49
0.31
µp
µp
0 0 0
0 0 0 0 10 0 4 0 25
0 10 0 4 0 25
ρ = 0.85 ρ = 0.84 ρ = 0.72
ρ = 0.73 ρ = 0.68 ρ = 0.52 4 4 4
4 4 4 d = 0.40, f = 1.96 g = 0.04, h = 3.23 g = 0.06, h = 2.33
d = 0.45, f = 2.30 g = 0.03, h = 4.17 g = 0.04, h = 3.45
µ
µ r
r
0 0 0
0 0 0 0 10 0 50 0 25
0 10 0 50 0 25 ρ = 0.84 ρ = 0.93 ρ = 0.72
ρ = 0.84 ρ = 0.91 ρ = 0.52 25 25 25
25 25 25 d = 1.56, f = 0.78 g = 2.49, h = 1.87 g = 16.8, h = 0.43
d = 1.87, f = 0.84 g = 2.87, h = 2.05 g = 22.4, h = 0.29
n
n y
y
0 0 0
0 0 0 0 10 0 50 0 4
0 10 0 50 0 4
data
data
regression
(a) regression (b)
Figure 9.25: Matrix plots of cross-correlations: effect of reference response spectrum
(EP hysteresis type, α = 0.10, SAC Los Angeles, design-level, SD soil): (a) AVG spec-
trum; (b) DES spectrum.
10 d = 1.64, ρf == 0.67
0.97
10 d = 0.43, ρf == 2.10
0.82
10 d = 0.98, fρ==0.65
0.70
10 d = 1.73, ρf == 0.68
0.98
10 d = 0.43, ρf == 2.37
0.82
10 d = 1.80, fρ==0.88
0.72
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 d = 1.64, ρf == 0.67
0.97
50 g = 23.7, ρh == 0.79
0.34
50 g = 0.36, ρh == 0.48
0.84
50 d = 1.73, ρf == 0.68
0.98
50 g = 33.3, ρh == 0.78 50 g = 0.33, ρh == 0.43
0.81
0.30
µp µ
p
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.82 ρ = 0.79 ρ = 0.55 ρ = 0.82 ρ = 0.78 ρ = 0.46
4 4 4 4 4 4
d = 0.43, f = 2.10 g = 0.04, h = 2.94 g = 0.07, h = 2.33 d = 0.43, f = 2.37 g = 0.03, h = 3.33 g = 0.05, h = 2.50
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.70 ρ = 0.84 ρ = 0.55 ρ = 0.72 ρ = 0.81 ρ = 0.46
25 25 25 25 25 25
d = 0.98, f = 0.65 g = 2.74, h = 2.09 g = 14.6, h = 0.43 d = 1.80, f = 0.88 g = 3.07, h = 2.31 g = 19.1, h = 0.40
n n
y y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.26: Matrix plots of cross-correlations: effect of reference response spectrum
(EP hysteresis type, α = 0.10, SAC Los Angeles, survival-level, SD soil): (a) AVG spec-
trum; (b) DES spectrum.
132
10 d = 1.69, ρf == 0.66
0.98
10 d = 0.38, ρf == 2.25
0.75
10 d = 3.62, fρ==1.64
0.93 10 d = 1.45, ρf == 0.63
0.96
10 d = 0.45, ρf == 2.66
0.65
10 d = 2.74, fρ==1.30
0.74
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
ρ = 0.96 ρ = 0.47 ρ = 0.87
50 d = 1.69, ρf == 0.66
0.98
50 g = 52.2, ρh == 0.68
0.26
50 g = 0.30, ρh == 0.36
0.96 50 d = 1.45, f = 0.63
50 g = 71.4, h = 0.17
50 g = 0.35, h = 0.40
µp µ
p
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.75 ρ = 0.68 ρ = 0.67 ρ = 0.65 ρ = 0.47 ρ = 0.17
4 4 4 4 4 4
d = 0.38, f = 2.25 g = 0.02, h = 3.85 g = 0.07, h = 1.43 d = 0.45, f = 2.66 g = 0.01, h = 5.88 g = 0.008
h = 8.33
µ µr
r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.93 ρ = 0.96 ρ = 0.67 ρ = 0.74 ρ = 0.87 ρ = 0.17
25 25 25 25 25 25
d = 3.62, f = 1.64 g = 3.34, h = 2.80 g = 14.0, h = 0.70 d = 2.74, f = 1.30 g = 2.89, h = 2.53 g = 119, h = 0.12
n n
y y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.27: Matrix plots of cross-correlations: effect of reference response spectrum
(EP hysteresis type, α = 0.10, SAC Los Angeles, design-level, SE soil): (a) AVG spec-
trum; (b) DES spectrum.
10 d = 1.73, ρf == 0.70
0.99
10 d = 0.32, ρf == 2.00
0.76
10 d = 3.38, fρ==2.13
0.94 10 d = 2.72, ρf == 0.82
0.98
10 d = 0.28, ρf == 1.78
0.41
10 d = 4.45, fρ==2.93
0.75
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 ρ = 0.99
50 ρ = 0.78
50 ρ = 0.95 50 d = 2.72, ρf == 0.82
0.98
50 g = 58.6, ρh == 0.36
0.34
50 g = 0.31, ρh == 0.29
0.80
d = 1.73, f = 0.70 g = 43.3, h = 0.36 g = 0.36, h = 0.33
µp µp
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.41 ρ = 0.36 ρ = 0.07
ρ = 0.76 ρ = 0.78 ρ = 0.70 4 4 4
4 4 4 d = 0.28, f = 1.78 g = 0.02, h = 2.94 g = 0.13
d = 0.32, f = 2.00 g = 0.02, h = 2.78 g = 0.09, h = 1.23
h = 0.003
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.75 ρ = 0.80 ρ = 0.07
25 ρ = 0.94
25 ρ = 0.95
25 ρ = 0.70 25 25 25
d = 4.45, f = 2.93 g = 3.21, h = 3.40 g = 7.74, h =357
d = 3.38, f = 2.13 g = 2.78, h = 3.00 g = 11.2, h = 0.81
ny n
y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.28: Matrix plots of cross-correlations: effect of reference response spectrum
(EP hysteresis type, α = 0.10, SAC Los Angeles, survival-level, SE soil): (a) AVG spec-
trum; (b) DES spectrum.
133
10 d = 1.33, ρf == 0.66
0.96
10 d = 0.32, ρf == 1.32
0.88
10 d = 0.57, fρ==0.60
0.70 10 d = 2.42, ρf == 0.81
0.92
10 d = 0.33, ρf == 1.41
0.62
10 d = 0.02, fρ==0.25
0.36
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 d = 1.33, ρf == 0.66
0.96
50 g = 13.4, ρh == 0.78 50 g = 0.60, ρh == 0.58
0.85 50 d = 2.42, ρf == 0.81
0.92
50 g = 17.0, ρh == 0.59
0.71
50 g = 0.60, ρh == 0.62
0.66
0.49
µp µp
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.88 ρ = 0.78 ρ = 0.46 ρ = 0.62 ρ = 0.59 ρ = 0.37
4 4 4 4 4 4
d = 0.32, f = 1.32 g = 0.07, h = 2.04 g = 0.13, h = 2.22 d = 0.33, f = 1.41 g = 0.06, h = 1.41 g = 0.11, h = 1.64
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.70 ρ = 0.85 ρ = 0.46 ρ = 0.36 ρ = 0.66 ρ = 0.37
25 25 25 25 25 25
d = 0.57, f = 0.60 g = 1.66, h = 1.73 g = 7.70, h = 0.45 d = 0.02, f = 0.25 g = 1.67, h = 1.62 g = 9.36, h = 0.61
n n
y y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.29: Matrix plots of cross-correlations: effect of reference response spectrum
(EP hysteresis type, α = 0.10, SAC Los Angeles, design-level, SD soil, NF): (a) AVG
spectrum; (b) DES spectrum.
10 d = 1.85, ρf == 0.50
0.75
10 d = 0.27, ρf == 1.33
0.97
10 d = 3.48, fρ==3.04
0.66 10 d = 3.93, ρf == 0.63
0.63
10 d = 0.27, ρf == 1.32
0.95
10 d = 4.16, fρ==5.41
0.48
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 ρ = 0.75
50 ρ = 0.71
50 ρ = 0.94 50 d = 3.93, ρf == 0.63
0.63
50 g = 65.0, ρh == 0.49
0.48
50 g = 0.87, ρh == 0.43
0.88
d = 1.85, f = 0.50 g = 62.5, h = 0.41 g = 1.64, h = 0.60
µp µp
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.97 ρ = 0.71 ρ = 0.63 ρ = 0.95 ρ = 0.49 ρ = 0.37
4 4 4 4 4 4
d = 0.27, f = 1.33 g = 0.02, h = 2.44 g = 0.12, h = 2.00 d = 0.27, f = 1.32 g = 0.02, h = 2.08 g = 0.11, h = 1.75
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.66 ρ = 0.94 ρ = 0.63 ρ = 0.48 ρ = 0.88 ρ = 0.37
25 25 25 25 25 25
d = 3.48, f = 3.04 g = 0.61, h = 1.67 g = 8.04, h = 0.50 d = 4.16, f = 5.41 g = 1.15, h = 2.32 g = 8.95, h = 0.57
n n
y y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.30: Matrix plots of cross-correlations: effect of reference response spectrum
(SD hysteresis type, α = 0.10, SAC Los Angeles, survival-level, SD soil): (a) AVG spec-
trum; (b) DES spectrum.
134
10 d = 2.23, ρf == 0.54
0.90
10 d = 0.00, ρf == 0.00
0.00
10 d = 0.44, fρ==0.47
0.57 10 d = 6.06, ρf == 0.84
0.91
10 d = 0.00, ρf == 0.00
0.00
10 d = 1.17, fρ==0.66
0.40
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 ρ = 0.90
50 ρ = 0.00
50 ρ = 0.86 50 d = 6.06, ρf == 0.84
0.91
50 g = 0.00, ρh == 0.00
0.00
50 g = 0.33, ρh == 0.36
0.72
d = 2.23, f = 0.54 g = 0.00, h = 0.00 g = 0.89, h = 0.65
µp µp
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.00 ρ = 0.00 ρ = 0.00
ρ = 0.00 ρ = 0.00 ρ = 0.00 4 4 4
4 4 4 d = 0.00, f = 0.00 g = 0.00, h = 0.00 g = 0.00, h = 0.00
d = 0.00, f = 0.00 g = 0.00, h = 0.00 g = 0.00, h = 0.00
µ µ
r r
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.40 ρ = 0.72 ρ = 0.00
25 ρ = 0.57
25 ρ = 0.86
25 ρ = 0.00 25 25 25
d = 1.17, f = 0.66 g = 3.07, h = 2.79 g = 0.00, h = 0.00
d = 0.44, f = 0.47 g = 1.12, h = 1.55 g = 0.00, h = 0.00
n n
y y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.31: Matrix plots of cross-correlations: effect of reference response spectrum
(BE hysteresis type, α = 0.10, SAC Los Angeles, survival-level, SD soil): (a) AVG spec-
trum; (b) DES spectrum.
10 d = 2.48, ρf == 0.65
0.91
10 d = 0.03, ρf == 1.44
0.70
10 d = 0.17, fρ==0.35
0.46 10 d = 3.46, ρf == 0.76
0.95
10 d = 0.03, ρf == 1.67
0.68
10 d = 1.09, fρ==0.62
0.38
µ µ
0 0 0 0 0 0
0 50 0 4 0 25 0 50 0 4 0 25
50 ρ = 0.91
50 ρ = 0.57
50 ρ = 0.77 50 d = 3.46, ρf == 0.76
0.95
50 g = 314, hρ == 1.15
0.54
50 g = 0.26, ρh == 0.32
0.64
d = 2.48, f = 0.65 g = 327, h = 1.13 g = 0.49, h = 0.52
µp µp
0 0 0 0 0 0
0 10 0 4 0 25 0 10 0 4 0 25
ρ = 0.68 ρ = 0.54 ρ = 0.02
4 ρ = 0.70
4 ρ = 0.57
44 ρ = 0.17 4 4 4
d = 0.03, f = 1.67 g = 0.003 g = 0.006
d = 0.03, f = 1.44 g = 0.003 g = 0.004 h = 0.87 h = 1.01
h = 0.88 3 h = 1.10
µ µ
2 r
r
1
0 0 0 0 0 0
0 10 0 50 0 25 0 10 0 50 0 25
ρ = 0.38 ρ = 0.64 ρ = 0.02
25 ρ = 0.46
25 ρ = 0.77
25 ρ = 0.17 25 25 25
d = 1.09, f = 0.62 g = 3.80, h = 3.14 g = 176, h = 0.99
d = 0.17, f = 0.35 g = 2.03, h = 1.91 g = 226, h = 0.91
ny n
y
0 0 0 0 0 0
0 10 0 50 0 4 0 10 0 50 0 4
data data
regression regression
(a) (b)
Figure 9.32: Matrix plots of cross-correlations: effect of reference response spectrum
(BP hysteresis type, α = 0.10, βs = βr = 1/3, SAC Los Angeles, survival-level, SD soil):
(a) AVG spectrum; (b) DES spectrum.
135
50 50
INDspectra
IND spectra: d = 2.76, f = 0.68 INDspectra
IND spectra: d = 2.78, f = 0.69
AVGspectrum
AVG spectrum: d = 1.53, f = 0.59 AVGspectrum
AVG spectrum: d = 1.75, f = 0.63
DES spectrum
DES spectrum: d = 0.38, f = 0.41 DES spectrum
DES spectrum: d = 1.09, f = 0.52
µp
µp
0 0
0 2 4 6 8 0 2 4 6 8
µ µ
4 INDspectra
spectra: d = 0.51, f = 2.91 4 INDspectra
spectra: d = 0.46, f = 2.62
IND IND
AVGspectrum
AVG spectrum: d = 0.42, f = 1.94 AVGspectrum
AVG spectrum: d = 0.32, f = 1.45
DES spectrum: d = 0.33, f = 1.46
DES spectrum DES spectrum: d = 0.30, f = 1.48
DES spectrum
µr
µr
0 0
0 2 4 6 8 0 2 4 6 8
µ µ
25 INDspectra
spectra: d = 5.79, f = 1.30
25 INDspectra
spectra: d = 4.62, f = 1.12
IND IND
AVGspectrum
AVG spectrum: d = 2.34, f = 0.84 AVGspectrum
AVG spectrum: d = 2.30, f = 0.85
DES
DESspectrum
spectrum: d = 0.46, f = 0.47 DES
DESspectrum
spectrum: d = 1.32, f = 0.65
ny
ny
0 0
0 2 4 6 8 0 2 4 6 8
µ µ
50 50
INDspectra
IND spectra: d = 2.87, f = 0.70 INDspectra
IND spectra: d = 3.07, f = 0.68
AVGspectrum
AVG spectrum: d = 1.90, f = 0.65 AVG
AVGspectrum
spectrum: d = 2.42, f = 0.68
DES
DESspectrum
spectrum: d = 0.63, f = 0.49 DES
DESspectrum
spectrum: d = 2.57, f = 0.68
µp
µp
0 0
0 2 4 6 8 0 2 4 6 8
µ µ
4 INDspectra
spectra: d = 0.53, f = 3.61 4
IND INDspectra
IND spectra: d = 0.37, f = 2.63
AVGspectrum
AVG spectrum: d = 0.34, f = 1.49 AVGspectrum
AVG spectrum: d = 0.30, f = 1.82
DES
DESspectrum
spectrum: d = 0.30, f = 1.32 DES
DESspectrum
spectrum: d = 0.44, f = 4.48
µr
µr
0 0
0 2 4 6 8 0 2 4 6 8
µ µ
25 25
INDspectra
IND spectra: d = 5.50, f = 1.42 INDspectra
IND spectra: d = 6.69, f = 1.91
AVGspectrum
AVG spectrum: d = 3.08, f = 1.08 AVGspectrum
AVG spectrum: d = 4.24, f = 1.39
DES spectrum
DES spectrum: d = 1.20, f = 0.71 DES spectrum
DES spectrum: d = 4.32, f = 1.37
ny
ny
0 0
0 2 4 6 8 0 2 4 6 8
µ µ
136
50 50
INDspectra
IND spectra: d = 2.30, f = 0.71 INDspectra
IND spectra: d = 1.91, f = 0.69
AVG
AVGspectrum
spectrum: d = 1.69, f = 0.65 AVG
AVGspectrum
spectrum: d = 1.64, f = 0.67
DES spectrum
DES spectrum: d = 1.49, f = 0.63 DES spectrum
DES spectrum: d = 1.73, f = 0.68
µp
µp
0 0
0 2 4 6 8 0 2 4 6 8
µ µ
4 4
INDspectra
IND spectra: d = 0.46, f = 2.10 INDspectra
IND spectra: d = 0.47, f = 2.84
AVGspectrum
AVG spectrum: d = 0.45, f = 2.30 AVGspectrum
AVG spectrum: d = 0.43, f = 2.10
DES spectrum
DES spectrum: d = 0.40, f = 1.96 DES spectrum
DES spectrum: d = 0.43, f = 2.37
µr
µr
0 0
0 2 4 6 8 0 2 4 6 8
µ µ
25 25
INDspectra
IND spectra: d = 2.44, f = 0.88 INDspectra
IND spectra: d = 1.77, f = 0.81
AVG
AVGspectrum
spectrum: d = 1.87, f = 0.84 AVG
AVGspectrum
spectrum: d = 0.98, f = 0.65
DES
DESspectrum
spectrum: d = 1.56, f = 0.78 DES
DESspectrum
spectrum: d = 1.80, f = 0.88
ny
ny
0 0
0 2 4 6 8 0 2 4 6 8
µ µ
50 50
INDspectra
IND spectra: d = 2.13,
2.87, f = 0.70 INDspectra
IND spectra: d = 2.46, f = 0.77
AVG
AVGspectrum
spectrum: d = 1.69, f = 0.66 AVG
AVGspectrum
spectrum: d = 1.73, f = 0.70
DES
DESspectrum
spectrum: d = 1.45, f = 0.63 DES
DESspectrum
spectrum: d = 2.72, f = 0.82
µp
µp
0 0
0 2 4 6 8 0 2 4 6 8
µ µ
4 4
INDspectra
IND spectra: d = 0.45,
0.53, f = 2.98
3.61 INDspectra
IND spectra: d = 0.40, f = 3.35
AVGspectrum
AVG spectrum: d = 0.38, f = 2.25 AVGspectrum
AVG spectrum: d = 0.32, f = 2.00
DES
DESspectrum
spectrum: d = 0.45, f = 2.66 DES
DESspectrum
spectrum: d = 0.28, f = 1.78
µr
µr
0 0
0 2 4 6 8 0 2 4 6 8
µ µ
25 25
INDspectra
IND spectra: d = 4.71,
5.50, f = 2.04
1.42 INDspectra
IND spectra: d = 4.77, f = 2.83
AVGspectrum
AVG spectrum: d = 3.62, f = 1.64 AVGspectrum
AVG spectrum: d = 3.38, f = 2.13
DES spectrum
DES spectrum: d = 2.74, f = 1.30 DES spectrum
DES spectrum: d = 4.45, f = 2.93
ny
ny
0 0
0 2 4 6 8 0 2 4 6 8
µ µ
137
50
INDspectra
IND spectra: d = 1.70, f = 0.71
AVG
AVGspectrum
spectrum: d = 1.33, f = 0.66
DES spectrum
DES spectrum: d = 2.42, f = 0.81
µp
0
0 2 4 6 8
µ
4
INDspectra
IND spectra: d = 0.56, f = 3.80
AVGspectrum
AVG spectrum: d = 0.32, f = 1.32
DES spectrum
DES spectrum: d = 0.33, f = 1.41
µr
0
0 2 4 6 8
µ
25
INDspectra
IND spectra: d = 1.39, f = 0.85
AVG
AVGspectrum
spectrum: d = 0.57, f = 0.60
DES
DESspectrum
spectrum: d = 0.02, f = 0.25
ny
0
0 2 4 6 8
µ
Figure 9.35: Λ-µ regression curves using IND, AVG, and DES spectra (EP hysteresis type,
α = 0.10, SAC Los Angeles, design-level, SD soil, NF).
138
50 50
INDspectra
IND spectra: d = 1.80, f = 0.48 INDspectra
IND spectra: d = 4.31, f = 0.68
AVG
AVGspectrum
spectrum: d = 1.85, f = 0.50 AVG
AVGspectrum
spectrum: d = 2.23, f = 0.54
DES spectrum
DES spectrum: d = 3.93, f = 0.63 DES spectrum
DES spectrum: d = 6.06, f = 0.84
µp
µp
0 0
0 2 4 6 8 0 2 4 6 8
µ µ
4 4
INDspectra
IND spectra: d = 0.37, f = 1.82 INDspectra
IND spectra: d = 0.00, f = 0.00
AVGspectrum
AVG spectrum: d = 0.27, f = 1.33 AVGspectrum
AVG spectrum: d = 0.00, f = 0.00
DES spectrum
DES spectrum: d = 0.27, f = 1.32 DES spectrum
DES spectrum: d = 0.00, f = 0.00
µr
µr
0 0
0 2 4 6 8 0 2 4 6 8
µ µ
25 25
INDspectra
IND spectra: d = 3.86, f = 3.02 INDspectra
IND spectra: d = 0.19, f = 0.36
AVGspectrum
AVG spectrum: d = 3.48, f = 3.04 AVGspectrum
AVG spectrum: d = 0.44, f = 0.47
DES
DESspectrum
spectrum: d = 4.16, f = 5.41 DES
DESspectrum
spectrum: d = 1.17, f = 0.66
ny
ny
0 0
0 2 4 6 8 0 2 4 6 8
µ µ
(a) (b)
50
INDspectra
IND spectra: d = 4.71, f = 0.86
AVG
AVGspectrum
spectrum: d = 2.48, f = 0.65
DES
DESspectrum
spectrum: d = 3.46, f = 0.76
µp
0
0 2 4 6 8
µ
4
INDspectra
IND spectra: d = 0.03, f = 1.87
AVGspectrum
AVG spectrum: d = 0.03, f = 1.44
DES
DESspectrum
spectrum: d = 0.03, f = 1.67
µr
0
0 2 4 6 8
µ
25
INDspectra
IND spectra: d = 0.20, f = 0.35
AVGspectrum
AVG spectrum: d = 0.17, f = 0.35
DES spectrum
DES spectrum: d = 1.09, f = 0.62
ny
0
0 2 4 6 8
µ
(c)
Figure 9.36: Λ-µ regression curves using IND, AVG, and DES spectra (α = 0.10, SAC Los
Angeles, survival-level, SD soil): (a) SD hysteresis type; (b) BE hysteresis type; (c) BP
hysteresis type, βs = βr = 1/3.
139
CHAPTER 10
In Figures 10.1 through 10.4, inelastic demand spectra are developed using the DES spec-
trum a and b regression coefficients in Table 8.2. The figures are organized in the same order that
the results are presented in Chapter 7: (1) demand spectra for regions with low seismicity, Boston
(Fig. 10.1); (2) demand spectra for regions with high seismicity, Los Angeles (Fig. 10.2); (3)
demand spectra for near-field conditions, NF (Fig. 10.3); and (4) demand spectra for different
hysteresis types, SD, BE, and BP (Fig. 10.4). As described previously in Section 2.3.2, these spec-
tra are constructed using a displacement-based approach by specifying target ductilities, µt, and
using Equations 2.1-2.3, 2.12, and 2.13. Inelastic demand spectra developed based on this proce-
dure can be used within the framework of a capacity spectrum method as demonstrated in the
design example below.
This section presents a design example using the capacity-demand index relationships
developed in this research. The capacity-spectrum design procedure with inelastic demand spectra
is implemented (Sect. 2.3.2). As indicated by Chopra and Goel (1999), a graphical implementa-
tion of this procedure in design is not essential and a purely numerical treatment can be per-
formed. Nevertheless, the procedure in this example is graphically implemented for illustrative
purposes.
140
0.5 0.5
µ=1
SAC Boston, SD soil, design level SAC Boston, SD soil, survival level
a = 1.08 a = 0.65
0.4 b = 0.42 0.4 b = 0.62
ξ = 5% ξ = 5%
µ=1
0.3 0.3
2
S (g)
S (g)
a
a
0.2 0.2 4
2 6 (linea
r-ela
8 stic)
(linea
r-ela
0.1 stic) 0.1
4
6
(a) 0
(b) 0
0 2 4 6 8 10 0 2 4 6 8 10
S (cm) S (cm)
d d
0.8 0.8
SAC Boston, SE soil, design level SAC Boston, SE soil, survival level
µ=1
a = 0.87 a = 0.41
b = 0.43 b = 0.84
0.6 0.6
ξ = 5%
ξ = 5%
µ=1 2
S (g)
S (g)
0.4 0.4
a
a
4
(linea
2 r-ela
6 stic)
8
0.2 (linea 0.2
r-ela
stic)
4
6
8
(c) 0
(d) 0
0 2 4 6 8 10 0 2 4 6 8 10
S (cm) S (cm)
d d
Figure 10.1: Inelastic demand spectra based on DES response spectra (EP hysteresis type, α
= 0.10): (a-b) SAC Boston, SD soil; (c-d) SAC Boston, SE soil.
2.5 2.5
SAC Los Angeles, SD soil, design level SAC Los Angeles, SD soil, survival level
a = 1.71 µ=1
a = 2.04
2 b = 0.59 2 b = 0.79
ξ = 5%
ξ = 5%
1.5 1.5
µ=1
S (g)
S (g)
2
a
(linea
r-ela
1 1 stic)
2 4
(linea 6
r-ela
0.5 stic) 0.5 8
4
6
8
(a) (b)
0 0
0 10 20 30 40 50 0 10 20 30 40 50
S (cm) S (cm)
d d
1.5 1.5
SAC Los Angeles, SE soil, design level µ=1
a = 0.80
b = 0.88 (lin
ea
r-e
las
1 ξ = 5% 1 tic
)
µ=1
2
S (g)
S (g)
a
(linea
r-ela
stic) 4
0.5 2 0.5
6
8
4
(c) 6
8
(d) a = 2.32
b = 1.31
ξ = 5% SAC Los Angeles, SE soil, survival level
0 0
0 10 20 30 40 50 0 10 20 30 40 50
S (cm) S (cm)
d d
Figure 10.2: Inelastic demand spectra based on DES response spectra (EP hysteresis type, α
= 0.10): (a-b) SAC Los Angeles, SD soil; (c-d) SAC Los Angeles, SE soil.
141
1.5
µ=1
SAC Los Angeles, NF, design level
a = 6.23
b = 1.34
1 ξ = 5%
2
S (g)
a
4
6
0.5 8 (linea
r-ela
stic)
0
0 10 20 30 40 50
S (cm)
d
Figure 10.3: Inelastic demand spectra based on DES response spectra (EP hysteresis type, α
= 0.10): SAC Los Angeles, SD soil, NF.
2.5 2.5
SAC Los Angeles, SD soil, survival level SAC Los Angeles, SD soil, survival level
µ=1
a = 2.20 µ=1
a = 3.68
2 b = 0.82 2 b = 1.06
ξ = 5%
ξ = 5%
1.5 1.5
S (g)
S (g)
2
2
a
(linea (linea
r-ela r-ela
1 stic) 1 stic)
4
4
6
6 8
0.5 8 0.5
(a) (b)
0 0
0 10 20 30 40 50 0 10 20 30 40 50
S (cm) S (cm)
d d
2.5
SAC Los Angeles, SD soil, survival level
µ=1
a = 3.00
2 b = 0.97
ξ = 5%
1.5
S (g)
2
a
(linea
r-ela
1 stic)
4
6
8
0.5
(c)
0
0 10 20 30 40 50
S (cm)
d
Figure 10.4: Inelastic demand spectra based on DES response spectra (SAC Los Angeles,
survival-level, SD soil, α = 0.10): (a) SD hysteresis type; (b) BE hysteresis type; (c) BP hys-
teresis type, βs = βr = 1/3.
142
The design process requires the estimation of the seismic demand and the structure capac-
ity as described below. The seismic demand is determined using the IBC 2000 smooth design
response spectra and the regression coefficients presented in Table 8.2. A displacement-based
design procedure prescribing target ductility values, µt, is used in the example.
It is assumed that the global lateral force-displacement relationship of the structures in the
design example can be represented using the EP hysteresis type. The design procedure is illus-
trated for six systems with α = 0.10, defined by two structure periods, T = 0.5 sec. and 1.5 sec.,
and three target ductilities, µt = 2, 4, and 8. The results are presented for capacity-demand index
relationships developed using IND and DES spectra to demonstrate the effect of reference
response spectrum on design. The seismic parameters used in the example are as follows:
The a and b regression coefficients corresponding to the design parameters for the
example (i.e., Los Angeles, SE soil, survival-level) are taken from Table 8.2. The
regression coefficients based on the DES reference spectrum are: a = 2.32, b =
1.31. Similarly, the regression coefficients based on the IND reference spectra are:
a = -0.71, b = 0.94.
The R-µ-T relationships are determined using Equations 2.1-2.3 and the a and b
coefficients from Step 2. These relationships are used in the next step to determine
the inelastic demand spectra.
The inelastic demand spectra are constructed based on inelastic acceleration, Sai,
and displacement, Sdi, demands as shown in Figure 2.4b using Equations 2.12 and
2.13, and the R-µ-T relationships from Step 3. The inelastic demand spectra based
on the DES and IND a and b coefficients used in the example (i.e., Step 2) are
143
1.5 1.5
µ=1 µ=1
T = 0.5 sec. T = 1.5 sec.
a = 2.32 a = 2.32
b = 1.31 b = 1.31
1 1
A (lin (lin
ea ea
r-e 2 r-e
las
S (g)
S (g)
2 tic) las
tic)
a
a
B
4 D
0.5 C
4 0.5
8
8 E
0 0
0 20 40 60 80 100 0 20 40 60 80 100
S (cm) S (cm)
d d
(a)
1.5 1.5
µ=1 µ=1
T = 0.5 sec. T = 1.5 sec.
a = -0.71 a = -0.71
b = 0.94 b = 0.94
1 1
A (lin (lin
ea ea
r-e r-e
las
S (g)
S (g)
tic) 2 las
2 tic)
a
B
0.5 0.5 4 D
C
4
8
8
E
0 0
0 20 40 60 80 100 0 20 40 60 80 100
S (cm) S (cm)
d d
(b)
Figure 10.5: Capacity curve-demand spectra for design example: (a) using DES spectrum;
(b) using IND spectra.
shown in Figures 10.5a and b, respectively. Note that the demand spectra in Figure
10.5a are the same as the spectra in Figure 10.2d.
144
For this example, the required idealized capacity curves that satisfy the target duc-
tilities, µt, for the structures are plotted in Figures 10.5a and b for the DES and
IND spectra, respectively. These capacity curves are constructed as illustrated in
Figure 2.4b. The resulting Sai and Sdi values, which are graphically represented by
the intersection of the capacity curves and the inelastic demand spectra, are sum-
marized in Table 10.1. In comparing Figures 10.5a and b and the values in Table
10.1, for a given target displacement ductility, µt, the IND reference spectra result
in larger R coefficients (i.e., smaller lateral strengths, Fy) to be used in design than
the DES reference spectrum, especially for long-period structures and for larger µt
values (see Table 10.1, structure F). Thus, it is clear that using capacity-demand
index relationships developed based on IND spectra in design could lead to
severely underestimated seismic demands.
Table 10.1: Structure properties and results for the design example
B 0.5 1.35 4 2.2 0.61 3.76 15.0 2.4 0.57 3.56 14.3
E 1.5 0.96 4 3.0 0.32 17.8 71.4 3.9 0.25 13.9 55.5
With the prescribed “target-µ”, µt, it is possible to estimate the residual displace-
ment (µr) and cumulative (µp and ny) damage demands as part of a performance-
based seismic design approach. The µp, µr, and ny demands can be estimated using
Equation 3.43 (with regression coefficients, d and f, from Tables 9.1 and 9.8) and
can be compared with: (1) allowable values to achieve specified performance
goals; and/or (2) corresponding structural capacities.
For this example, the estimated µp, µr, and ny demands using Equation 3.43 and
the d and f coefficients from Tables 9.1 and 9.8 are listed in Table 10.2. Notice that
the structure period, T, and the R coefficient are not required to determine µp, µr,
and ny once µ = µt is known. Additionally, note that the demands for this design
example are relatively independent of the reference response spectrum used.
145
Table 10.2: µp, µr, and ny demands for the design example
T Felas/W µt R µp µr ny R µp µr ny
Structure
(sec) (g)
B 0.5 1.35 4 2.2 13.0 0.9 2.4 2.4 13.0 1.1 2.6
E 1.5 0.96 4 3.0 13.0 0.9 2.4 3.9 13.0 1.1 2.6
146
CHAPTER 11
This chapter compares the seven different ground motion scaling methods described in
Section 3.4 based on the scatter in the dimensional (unit dependent) maximum displacement
demand, ∆nlin, and the non-dimensional maximum displacement ductility demand, µ = ∆nlin/∆y,
from the SDOF nonlinear dynamic time-history analyses. The evaluation of scatter for the MDOF
analyses is presented as normalized floor and roof displacement, ∆, and normalized interstory
drift, θ, response profiles.
The SDOF analysis results are discussed first. Then, the findings from the SDOF analyses
are reinforced using the results from the MDOF analyses. The evaluation of scatter is presented
as: (1) COV-spectra; and (2) γ-spectra, defined as the ratio of the COV-spectra for two different
sets of parameters (i.e., scaling method, hysteresis type, and site conditions). The lateral force
capacities, Fy = Felas/R, of the SDOF systems described in this chapter are determined based on
the linear-elastic acceleration response spectrum for each ground motion (i.e., IND spectra). The
lateral force capacities of the MDOF structures are determined based on smooth design (DES)
response spectra as described in Section 3.1.2. The results are presented for the UND ground
motion ensemble and the SAC Los Angeles design-level stiff (SD) soil near-field (NF) ground
motion ensemble.
Note that, as stated earlier in Section 3.3, the a and b regression coefficients used in the
estimation of the effective structure period, Tµ, for the Sˆa ( T o → T µ ) scaling method take into
account the site conditions and hysteresis types used in the dynamic analyses. These a and b coef-
ficients are listed in Table 3.6.
Figure 11.1 shows the scatter in the maximum displacement demand, ∆nlin, using the
results from the UND SD soil ground motion ensemble for the EP hysteresis type with α = 0.10.
The COV-spectra are presented on the left side of Figure 11.1. The solid lines in the COV-spectra
represent results from the ground motions scaled to the average peak ground acceleration (i.e.,
PGA scaling method) while the dotted lines represent results from the same ensemble for the
other scaling methods. The γ-spectra are presented on the right side of Figure 11.1.
For the PGA, EPA, and A95 scaling methods, the results indicate that the scatter tends to
increase as the structure period increases, similar to previous investigations by Nau and Hall
147
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
) )
PGA EPA
γCOV(∆nlin)
COV(Snlin
COV(Sd)
EPA
d
PGA
COV(∆
γ
(a)
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
PGA
) )
EPV
γCOV(∆nlin)
COV(Snlin
COV(Sd)
EPV
d
PGA
COV(∆
γ
(b)
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
) )
PGA MIV
COV(Snlin
γCOV(∆nlin)
COV(Sd)
MIV
d
PGA
COV(∆
γ
(c)
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
) )
PGA A95
COV(Snlin
γCOV(∆nlin)
COV(Sd)
A95
d
PGA
COV(∆
(d)
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
Sˆ (To)
) )
PGA
COV(Snlin
a
γCOV(∆nlin)
COV(Sd)
Sˆ (To)
d
a PGA
COV(∆
(e)
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
Sˆ (To→T )
) )
PGA
COV(Snlin
a µ
COV(Sd)
γCOV(∆nlin)
Sˆ (To→T )
d
a µ PGA
COV(∆
(f)
0 0
0 T (sec) 2.0 0 T (sec) 2.0
Figure 11.1: Scatter in ∆nlin for the UND SD soil ensemble (EP hysteresis type, α = 0.10)
- PGA method compared to: (a) EPA method; (b) EPV method; (c) MIV method; (d) A95
method; (e) Sˆa ( T o ) method; (f) Sˆa ( T o → T µ ) method.
148
(1982) and Miranda (1993). This is expected for the PGA and EPA methods since these scaling
methods control the short-period range of the ground motion acceleration spectra. Overall, the
differences between the three scaling methods are negligible, except for short period structures
with small R coefficients.
As compared to the PGA method, the EPV method is, on average, considerably more
effective in reducing the scatter for T > 0.5 sec. (Fig. 11.1b). At T ≅ 1 sec., the EPV method is
most effective in reducing scatter since this method scales the ground motions to the average spec-
tral pseudo-velocity at around a structure period of 1 sec. (see definition of EPV in Section 3.3).
The PGA method is compared to scaling the ground motion ensemble to the average max-
imum incremental velocity (i.e., MIV scaling method) in Figure 11.1c. The results indicate that,
regardless of the R coefficient, the MIV method is modestly more effective in reducing the scatter
for T > 0.5 sec. For T < 0.5 sec., the effectiveness of the MIV method increases with R. The scatter
for the MIV method approaches the scatter for the PGA method as T increases.
The PGA method is compared to scaling the ground motion ensemble to the average spec-
tral intensity at each linear-elastic structure period (i.e., Sˆa ( T o ) method) in Figure 11.1e and to
scaling the ground motion ensemble to the average spectral intensity over a range of structure
periods (i.e., Sˆa ( T o → T µ ) method) in Figure 11.1f. It can be expected that all three methods pro-
duce similar scatter for extremely short-period structures since, at T = 0, the Sˆa ( T o ) and
Sˆa ( T o → T µ ) methods are equivalent to the PGA method. For R = 1 (linear-elastic behavior), the
scatter for the Sˆa ( T o ) and Sˆa ( T o → T µ ) methods is zero for the entire period range since the
ground motion records are scaled to the average linear-elastic spectral acceleration at the linear-
elastic structure period.
Unlike the PGA method which is relatively independent of R, the COV-spectra for the
Sˆa ( T o ) method increase as R increases. With an increase in R, µ increases and, in turn, the effec-
tive structure period, Tµ, increases (see Eq. 3.38). Therefore, the effectiveness of the Sˆa ( T o )
method, based on the linear-elastic structure period, To, decreases for larger R. Since the
Sˆa ( T o → T µ ) method accounts for the period elongation due to the nonlinear behavior expected
in the structure, the scatter is markedly reduced, regardless of the R coefficient for T > 0.75 sec.
For T < 0.75 sec., the effectiveness of the Sˆa ( T o → T µ ) method tends to decrease as R increases.
Comparing the results for the Sˆa ( T o → T µ ) method and the MIV method (solid lines, Fig.
11.2a), it can be seen that the MIV method is more effective for 0.2 < T < 0.3 sec. with R = 6 and
for 0.1 < T < 0.45 sec. with R = 8. The MIV method is even more effective when compared to the
Sˆa ( T o ) method (dashed lines) for 0.1 < T < 0.5 sec. with R = 6 and for 0.1 < T < 0.65 sec. with R
= 8. Thus, it is evident that scaling ground motions based on spectral intensity measures (i.e.,
Sˆa ( T o ) and Sˆa ( T o → T µ ) ) is not necessarily effective for all period ranges and strength levels.
This finding is more prominent for the soft (SE) soil profile (Fig. 11.2c) which will be discussed in
further detail later in Section 11.3. Figures 11.2b and d will also be described later in the chapter.
It was previously reported by Nassar and Krawinkler (1991) that the scatter in the maxi-
mum displacement ductility demand, µ, tends to increase with the R coefficient. Inspecting the
standard deviation, σ, for the µ spectra (Fig. 11.3a), it is evident that the scatter in µ does increase
149
2 2
COV(∆nlin ) R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
COV(∆nlin )
MIV MIV
γ
Sˆ (To ) Sˆ (To )
γ
a a
MIV MIV
Sˆ (To→T ) Sˆ (To→T )
a µ a µ
0 0
0 T (sec) 2.0 0 T (sec) 2.0
(a) (b)
2 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
COV(∆nlin )
COV(∆nlin )
MIV MIV
Sˆ (To ) Sˆ (To )
γ
γ
a a
MIV MIV
Sˆ (To→T ) Sˆ (To→T )
a µ a µ
0 0
0 T (sec) 2.0 0 T (sec) 2.0
(c) (d)
Figure 11.2: Scatter in ∆nlin using the S a ( T o ) and S a ( T o → T µ ) methods compared to
ˆ ˆ
the MIV method (EP hysteresis type, α = 0.10): (a) UND SD soil; (b) UND SC soil; (c)
UND SE soil; (d) SAC SD soil, NF.
as R increases. However, using the standard deviation alone as a measure of scatter can be mis-
leading. The coefficient of variation, COV, which normalizes the sample standard deviation by the
sample mean, is a better means to evaluate scatter. As illustrated in Figure 11.3b, scatter relative
to the sample mean (COV) does not increase with R as much as the sample standard deviation, σ
(Fig. 11.3a).
It should be noted that dimensionless demand estimates, such as the maximum displace-
ment ductility demand, µ = ∆nlin/∆y, are independent of scaling when the individual (IND) linear-
elastic ground motion response spectra are used to determine the lateral force capacity, Fy = Felas/
R (Nassar and Krawinkler, 1991). For example, scaling a ground motion record by a constant fac-
tor will equally affect both the linear-elastic lateral force demand, Felas, and the lateral force
capacity, Fy, (for a given R coefficient) resulting in the same µ. Thus, the scatter in µ shown in
Figure 11.3 is not affected by the method used in the scaling of the ground motion records.
The effect of the structure hysteresis type on the scatter from the different ground motion
scaling methods is examined in this section. The COV-spectra for the scatter in ∆nlin using the
seven scaling methods are presented as is done in the previous section for the EP hysteresis type.
The SD, BE, and BP (βs = βr = 1/3) hysteresis type (α = 0.10) trends, provided in Figure 11.4, are
almost identical to the trends observed for the EP type in Figure 11.1. The γ-spectra for the scatter
in ∆nlin between the different hysteresis types range from 1.0 to 1.25. The γ-spectra for the scatter
150
14 1.5
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
12
10
1
8
COV(µ)
σ(µ)
6
4 0.5
2
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
(a) (b)
Figure 11.3: Scatter in µ (UND SD soil ensemble, EP hysteresis type, α = 0.10): (a) stan-
dard deviation, σ; (b) coefficient of variation, COV.
in µ between the different hysteresis types have similar values as the γ-spectra for the scatter in
∆nlin between the different hysteresis types. Thus, it is concluded that the scatter in ∆nlin and the
scatter in µ are not significantly affected by the hysteresis type or by the amount of inelastic
energy dissipated by the system.
Note that the γ-spectra for the scatter in µ = ∆nlin/∆y between the different hysteresis types
are slightly different from the γ-spectra for the scatter in ∆nlin, because, the use of individual
(IND) linear-elastic ground motion response spectra to determine the lateral force capacities, Fy =
Felas/R, as described before results in a scatter in the yield displacement, ∆y, of the structures. For
cases where the ground motions are scaled using the Sˆa ( T o ) method or where a smooth design
reference response spectrum is used to determine the lateral force capacities, Fy, the yield dis-
placements, ∆y, remain the same regardless of which ground motion is used (i.e., there is no scat-
ter in ∆y). Under these conditions, the γ-spectra for the scatter in µ would be equal to the γ-spectra
for the scatter in ∆nlin.
The effect of site soil characteristics on the scatter from the different ground motion scal-
ing methods is examined in this section. Figures 11.5 and 11.6 show the COV-spectra for the scat-
ter in ∆nlin using the UND SC and SE soil ground motion ensembles, respectively. Comparing
Figures 11.1 and 11.5, the trends observed for the SC soil profile are similar to the trends for the
SD soil profile. The SC soil ensemble exhibits slightly more scatter than the SD soil ensemble for
almost the entire period range, except for the EPV and MIV methods. For the EPV method, the
scatter from the SC soil ensemble is less than the scatter from the SD soil ensemble for T < 0.5 sec.
and is more or less similar to that from the SD soil ensemble for T > 0.5 sec. For the MIV method,
the scatter from the SC soil ensemble is similar to that from the SD soil ensemble for the entire
period range.
151
1.5 R = 1, 2, 4, 6, 8 (thin → thick lines) 2
1.5 (thin →
RR==1,1,2,2,4,4,6,6,88(thin →thick
thicklines)
lines) 2
1.5 RR == 1,
1, 2,
2, 4,
4, 6, (thin →
6, 88 (thin → thick
thick lines)
lines)
nlin))
nlin))
nlin)) EPA
COV(Sd)
PGA PGA
COV(Sd)
d PGA
EPA
d
d
COV(S
COV(S
COV(S
EPA EPA
PGA PGA
EPA
COV(∆
COV(∆
COV(∆
γ
0 0 00
0 T (sec) 2.0 0 T (sec)
(sec) 2.0 0 T (sec) 2.0
nlin))
nlin))
nlin))
COV(Sd)
PGA PGA
EPV
COV(Sd)
PGA EPV
d
d
COV(S
COV(S
COV(S
EPV EPV
PGA EPV
PGA
COV(∆
COV(∆
COV(∆
γ
γ
0 00 0
0 T (sec) 2.0 0 T (sec) 2.0 0 T (sec) 2.0
nlin))
nlin))
COV(Sd)
PGA PGA
COV(Sd)
PGA MIV MIV
d
d
d
COV(S
COV(S
COV(S
MIV MIV
PGA MIV
PGA
COV(∆
COV(∆
COV(∆
γ
γ
0 00 0
0 T (sec) 2.0 0 T (sec) 2.0 0 T (sec) 2.0
nlin))
nlin))
nlin))
COV(Sd)
PGA PGA
COV(Sd)
PGA
A95 A95
d
d
COV(S
COV(S
COV(S
A95 A95
PGA A95
PGA
COV(∆
COV(∆
COV(∆
γ
γ
0 00 0
0 T (sec) 2.0 0 T (sec) 2.0 0 T (sec) 2.0
))
nlin))
(T )
COV(Sd)
PGA
COV(Sd)
PGA
nlin
a o a
d
d
d
Sˆ (To)
COV(S
Sˆ (To) SˆPGA
COV(S
COV(S
PGA (T )
COV(∆
COV(∆
COV(∆
a a a o
γ
γ
0 00 00
0 T (sec) 2.0 0 T (sec) 2.0 0 (sec)
T (sec) 2.0
2.0
nlin))
nlin))
COV(Sd)
COV(Sd)
PGA µ µ
d
d
d
a o a o
Sˆ (To→ T ) Sˆ (To→T ) Sˆ (To→T )
COV(S
COV(S
COV(S
a µ aPGA µ PGA
a µ
COV(∆
COV(∆
COV(∆
0 0 0
0 T (sec) 2.0 0 T (sec) 2.0 00 TT (sec) 2.0
2.0
152
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
) )
γCOV(∆nlin)
COV(Snlin
COV(Sd)
d
COV(∆
γ
PGA EPA
(a) EPA PGA
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
) )
PGA EPV
γCOV(∆nlin)
COV(Snlin
COV(Sd)
EPV
d
PGA
COV(∆
γ
(b)
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
) )
PGA MIV
COV(Snlin
γCOV(∆nlin)
COV(Sd)
MIV
d
PGA
COV(∆
γ
(c)
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
) )
PGA A95
COV(Snlin
γCOV(∆nlin)
COV(Sd)
A95
d
PGA
COV(∆
(d)
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
Sˆ (To)
) )
PGA
COV(Snlin
a
COV(Sd)
γCOV(∆nlin)
Sˆ (To)
d
a PGA
COV(∆
(e)
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
PGA Sˆ (To→T )
) )
µ
COV(Snlin
a
Sˆ (To→T )
COV(Sd)
γCOV(∆nlin)
µ
d
a
PGA
COV(∆
(f)
0 0
0 T (sec) 2.0 0 T (sec) 2.0
Figure 11.5: Scatter in ∆nlin for the UND SC soil ensemble (EP hysteresis type, α = 0.10) -
PGA method compared to: (a) EPA method; (b) EPV method; (c) MIV method; (d) A95
method; (e) Sˆa ( T o ) method; (f) Sˆa ( T o → T µ ) method.
153
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
d )
γCOV(∆nlin)
COV(Snlin
COV(Sd)
COV(∆ )
γ
PGA EPA
(a) EPA PGA
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
d )
γCOV(∆nlin)
COV(Snlin
COV(Sd)
COV(∆ )
γ
PGA EPV
(b) EPV PGA
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
PGA MIV
d )
COV(Snlin
COV(Sd)
)
γCOV(∆nlin)
MIV PGA
COV(∆
γ
(c)
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
d )
nlin
COV(Sd)
)
γCOV(∆nlin)
COV(S
COV(∆
PGA A95
(d) A95 PGA
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
d )
nlin
COV(Sd)
)
γCOV(∆nlin)
COV(S
COV(∆
PGA Sˆ (To)
γ
a
Sˆ (To)
a PGA
(e)
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
PGA Sˆ (To→T )
d )
a µ
nlin
COV(Sd)
Sˆ (To→T )
)
γCOV(∆nlin)
µ PGA
COV(S
a
COV(∆
(f)
0 0
0 T (sec) 2.0 0 T (sec) 2.0
Figure 11.6: Scatter in ∆nlin for the UND SE soil ensemble (EP hysteresis type, α = 0.10) -
PGA method compared to: (a) EPA method; (b) EPV method; (c) MIV method; (d) A95
method; (e) Sˆa ( T o ) method; (f) Sˆa ( T o → T µ ) method.
154
increases in a similar fashion. This period elongation runs the structure into the energy-rich region
of the SE soil ensemble, increasing the nonlinear response and the variability in the response for T
< 1.1-1.5 sec.
Figures 11.6e and f demonstrate that the effectiveness of the Sˆa ( T o ) and the
S a ( T o → T µ ) methods significantly decreases as the period decreases within the ranges of 0.75 <
ˆ
T < 1.5 sec. and 0.75 < T < 1.25 sec., respectively. Note that, as mentioned at the beginning of this
chapter, the a and b regression coefficients used in the estimation of the effective structure period,
Tµ, for the Sˆa ( T o → T µ ) method take into account site soil characteristics. Even so, the scatter in
the demand estimates for T < 0.75 sec. using the Sˆa ( T o → T µ ) method is larger than the scatter
using the PGA method for R = 4, 6, and 8.
In examining Figure A.16, it is evident that there is considerable scatter in the linear-elas-
tic ground motion response spectra for the SE soil ensemble at longer periods. Thus, once a struc-
ture runs into this energy-rich region following yield, the responses themselves tend to have
considerable scatter. It is for this reason that even the Sˆa ( T o → T µ ) method cannot account for
the high variability in the nonlinear response.
Comparing results between the scatter using the Sˆa ( T o → T µ ) method and the MIV
method, it can be seen that the MIV method is more effective for 0.1 < T < 0.5 sec. with R = 6, 8
for the SC soil profile (Fig. 11.2b) and for 0.2 < T < 1.15 sec. with R = 4, 6, 8 for the SE soil profile
(Fig. 11.2c). The MIV method is even more effective when compared to the Sˆa ( T o ) method for
0.1 < T < 0.65 sec. with R = 6, 8 for the SC soil profile and for 0.16 < T < 1.25-1.5 sec. with R = 4,
6, 8 for the SE soil profile. The reduction in the scatter when using the MIV method as compared
to the Sˆa ( T o ) and Sˆa ( T o → T µ ) methods is significant for the SE soil profile at T ≅ 0.5 sec. (γ ≅
0.2 for R = 6, 8). Thus, it is evident that scaling ground motions based on spectral intensity mea-
sures (i.e., Sˆa ( T o ) and Sˆa ( T o → T µ ) methods) is not necessarily effective for all strength levels,
period ranges, and site soil characteristics. This finding will be reinforced using results from the
MDOF analyses later in Section 11.5.
Figure 11.7 shows the COV-spectra for the scatter in ∆nlin using the SAC Los Angeles
design-level SD soil NF ground motion ensemble. Comparing Figures 11.1 and 11.7, the NF
ensemble tends to exhibit less or similar scatter than the far-field ensemble for the PGA, EPA, and
A95 methods, especially for long periods. The dependency of the scatter for these scaling methods
on the period is significantly decreased for the near-field ensemble. Figure 11.7a shows that the
EPA method results in significantly larger scatter as compared to the PGA method, except for
short-period structures with small R coefficients. In contrast, the effectiveness of the EPV method
as compared to the PGA method is highly dependent on both T and R (Fig. 11.7b).
The PGA method is compared to the MIV method in Figure 11.7c and to the A95 method
in Figure 11.7d. The MIV method is modestly more effective in reducing scatter for larger R coef-
ficients in shorter period ranges and for smaller R coefficients in longer period ranges. The differ-
ences between the A95 method and the PGA method are negligible, except for T < 0.5 sec.
155
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
) )
PGA
γCOV(∆nlin)
COV(Snlin
COV(Sd)
EPA
d
COV(∆
γ
EPA
(a) PGA
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
) )
γCOV(∆nlin)
COV(S nlin
PGA
COV(Sd)
EPV
d
COV(∆
γ
(b) EPV
PGA
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
) )
COV(S nlin
PGA
γCOV(∆nlin)
COV(Sd)
MIV
d
COV(∆
γ
(c) MIV
PGA
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
d )
PGA A95
COV(Snlin
γCOV(∆nlin)
COV(Sd)
)
A95 PGA
COV(∆
(d)
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
PGA Sˆ (To)
) )
COV(Snlin
a
COV(Sd)
γCOV(∆nlin)
Sˆ (To)
d
a PGA
COV(∆
(e)
0 0
0 T (sec) 2.0 0 T (sec) 2.0
1.5 2
R = 1, 2, 4, 6, 8 (thin → thick lines) R = 1, 2, 4, 6, 8 (thin → thick lines)
PGA Sˆ (To→T )
) )
µ
COV(Snlin
a
Sˆ (To→T )
COV(Sd)
γCOV(∆nlin)
d
a µ PGA
COV(∆
(f)
0 0
0 T (sec) 2.0 0 T (sec) 2.0
Figure 11.7: Scatter in ∆nlin for the SAC Los Angeles, design-level, SD soil, NF ensemble
(EP hysteresis type, α = 0.10) - PGA method compared to: (a) EPA method; (b) EPV
method; (c) MIV method; (d) A95 method; (e) Sˆa ( T o ) method; (f) Sˆa ( T o → T µ ) method.
156
The PGA method is compared to the Sˆa ( T o ) method in Figure 11.7e and to the
S a ( T o → T µ ) method in Figure 11.7f. As compared to the far-field ground motions in Figures
ˆ
11.1e and f, the effectiveness of both methods with respect to the PGA method is significantly
decreased for the entire range of periods for R > 1, especially for large values of R. Except for
long periods and small R coefficients, the Sˆa ( T o ) method is less effective or only slightly more
effective than the PGA method. As compared to the Sˆa ( T o ) method, the scatter for the
Sˆa ( T o → T µ ) method is significantly reduced, especially for larger R coefficients. The depen-
dence of scatter for both methods on R is highly noticeable.
Figure 11.2d shows that the MIV method does not provide any significant improvement as
compared to the Sˆa ( T o → T µ ) method. However, the MIV method is more effective than the
Sˆa ( T o ) method in reducing scatter for 0.15 < T < 1.15 sec. with R = 6, 8. Thus, the benefit of
using the MIV method over the Sˆa ( T o ) method is still evident for near-field conditions.
As discussed previously in this chapter, the MIV method provides better reduction in scat-
ter than the Sˆa ( T o ) and Sˆa ( T o → T µ ) methods for a significant range of structure periods, espe-
cially on soft (SE) soil profiles and for large R coefficients. This is because the MIV method is
able to capture the impulsive characteristics typical of ground motions on soft soil profiles. To
show that the results obtained from the SDOF analyses are applicable to MDOF systems, the
amount of scatter in the lateral displacements of the four-story and eight-story structures
described in Section 3.1.2 is investigated below.
As described in Section 3.6.2, the UND soft (SE) soil ground motion ensemble is used to
excite the MDOF systems. After scaling each ground motion based on the MIV method or the
Sˆa ( T o → T µ ) method, the entire ground motion ensemble was scaled so that the average linear-
elastic acceleration response spectrum of the ensemble is not less than 1.4 times the 5%-damped
UBC 1997 (ICBO, 1997) Los Angeles design-level SE soil design response spectrum for periods
between 0.2To and 1.5To, where To is the structure fundamental period (see Fig. 3.22).
Figure 11.8 shows the normalized maximum floor and roof lateral displacement, ∆, and
interstory drift, θ, scatter profiles for the four-story and eight-story structures subjected to the
UND SE soil ground motion ensemble scaled as described above. The lateral displacement, ∆, val-
ues are normalized with the maximum mean floor or roof displacement over the height of the
structure, ∆max (see Section 3.7.2). Similarly, the interstory drift, θ, values are normalized with the
maximum mean interstory drift over the height of the structure, θmax. It is evident from the narrow
± standard deviation, σ, bands that, in each case, the MIV method provides a significant reduction
in scatter as compared to the Sˆa ( T o → T µ ) method. This observation is especially apparent for
the four-story structure (Fig. 11.8a). Based on the SDOF results shown previously, the reduction
in scatter as compared to the Sˆa ( T o ) method is expected to be even more apparent.
The reduced scatter in the seismic demands using the MIV method, coupled with its sim-
plicity, makes the MIV method advantageous over the other scaling methods described herein for
a wide range of site and structure characteristics. The advantages of the MIV method over the
157
4 4 4 4
3 3 3 3
floor or roof level
story
story
2 2 2 2
Sa(To→Tµ) MIV Sa(To→Tµ) MIV
1 1 1 1
mean mean mean mean
± std. dev. ± std. dev. ± std. dev. ± std. dev.
data data data data
0 0 0 0
0 2 4 0 2 4 0 2 4 0 2 4
∆i/∆max θi/θmax
(a)
8 8 8 8
mean mean mean mean
± std. dev. ± std. dev. ± std. dev. ± std. dev.
7 data 7 data 7 data 7 data
6 6 6 6
floor or roof level
5 5 5 5
story
story
4 4 4 4
3 3 3 3
2 2 2 2
1 1 1 1
Sa(To→Tµ) MIV Sa(To→Tµ) MIV
0 0 0 0
0 2 4 0 2 4 0 2 4 0 2 4
∆i/∆max θi/θmax
(b)
Figure 11.8: Scatter in MDOF demands for the UND SE soil ensemble using the MIV
method and the Sˆa ( T o → T µ ) method: (a) four-story structure; (b) eight-story structure.
Sˆa ( T o ) and Sˆa ( T o → T µ ) methods come from: (1) the MIV method is based only on ground
motion parameters and not on structure parameters (e.g., linear-elastic structure period, To), which
may not be known in advance; and (2) ground motions scaled using the MIV method can be used
to analyze structures with different properties (e.g., with different To) since the ground motions do
not need to be rescaled for each structure. Thus, it is recommended that nonlinear dynamic analy-
ses are conducted using ground motion records scaled based on the MIV method, particularly for
larger R coefficients and soft soil profiles for the period ranges shown in Figure 11.2.
Figure 11.9 shows the maximum floor and roof displacement, ∆, and interstory drift, θ,
COV profiles for the four-story and eight-story structures, which display similar trends as the cor-
responding normalized displacement and drift values in Figure 11.8. Furthermore, the average
(taken over the height of the structure) COV values of the maximum floor and roof displacement,
158
4 8
5
Sa(T→T
o µ
), COV(∆ )
i
2 4 S (T→T ), COV(θ )
a o µ i
MIV, COV(∆ )
i
3 MIV, COV(θi)
1 Sa(T→T
o µ
), COV(∆ )
i 2
S (T→T ), COV(θ )
a o µ i
MIV, COV(∆ ) 1
i
MIV, COV(θi)
0 0
0 0.5 1 0 0.5 1
COV COV
(a) (b)
Figure 11.9: Covariance in the MDOF demands for the UND SE soil ensemble using the
MIV method and the Sˆa ( T o → T µ ) method: (a) four-story structure; (b) eight-story struc-
ture.
∆, for the four-story and eight-story structures in Figure 11.9 (which are equal to 0.75 and 0.25 for
the four-story structure using the Sˆa ( T o → T µ ) and MIV methods, respectively, and 0.52 and
0.35 for the eight-story structure using the Sˆa ( T o → T µ ) and MIV methods, respectively) com-
pare well with the COV values for the SDOF models at T = 0.49 and 0.87 sec. (i.e., the fundamen-
tal periods of the four-story and eight-story structures), respectively, in Figures 11.6c and f (for R
= 8). Thus, the results from the MDOF analyses seem to satisfactorily verify the results from the
SDOF analyses.
Since the four-story and eight-story structures were designed based on smooth design
response spectra as specified by the UBC 1997 provisions, the observations made above for the
scatter in the maximum lateral floor and roof displacement, ∆, and interstory drift, θ, demands are
also applicable to the scatter in the maximum displacement ductility demand, µ. This is because
the structure lateral force capacity, and thus, the “yield displacement” remain the same regardless
of which ground motion is used. Thus, the scatter in µ is the same as the scatter in ∆.
Note that as the scatter in the maximum displacement ductility demand, µ, increases, the
scatter in the other demand indices (e.g., µp, µr, ny) are expected to increase. Thus, reducing scat-
ter in ∆, θ, and µ is very important for the use of performance-based seismic design procedures
with multiple demand indices, as outlined in Chapter 10.
159
CHAPTER 12
12.1 Summary
The broad objective of this research is to address some of the research needs for the imple-
mentation of performance-based procedures in current U.S. seismic design provisions. The
research has four specific objectives: (1) to develop new nonlinear single-degree-of-freedom
(SDOF) capacity-demand index relationships based on linear-elastic smooth design response
spectra consistent with current seismic code provisions; (2) to develop new relationships that
quantify cumulative damage, hysteretic energy, and residual displacement demands; (3) to investi-
gate the effects of the structure fundamental period, strength level, hysteretic behavior, and site
conditions on the demand estimates; and (4) to investigate the effect of the ground motion scaling
method on the scatter in the demand estimates.
The University of Notre Dame (UND) ground motions are used to investigate the effect of
the ground motion scaling method on the scatter in the demand estimates. The SAC steel project
(SAC) ground motions are primarily used to develop new capacity-demand index relationships,
including relationships based on linear-elastic smooth design response spectra. Finally, the Nassar
and Krawinkler (N&K) ground motions are used to validate the analytical procedure used in this
research by comparing the results with already existing results.
160
In order to investigate the effect of reference response spectra on the capacity-demand
index relationships, three types of linear-elastic acceleration response spectra are used to calculate
the structure lateral force capacity, Fy = Felas/R, namely: (1) response spectra based on the indi-
vidual ground motion records (IND); (2) average ground motion response spectra based on the
ground motion ensembles (AVG); and (3) smooth design response spectra from current U.S. seis-
mic design provisions (DES).
The effect of seven ground motion scaling methods on the scatter in the demand estimates
is examined, including the following parameters: (1) input parameters such as site soil character-
istics and epicentral distance; and (2) structural parameters such as fundamental period, hysteresis
type, and strength level (i.e., response modification coefficient, R).
In addition to the SDOF models, two MDOF models representative of four-story and
eight-story cast-in-place reinforced-concrete special moment-resisting office building frame
structures are considered. These structures are designed according to the UBC 1997 equivalent
lateral force procedure for a region with high seismicity (Los Angeles) and for the soft (SE) soil
profile. Nonlinear dynamic time-history analyses of the MDOF models are conducted to reinforce
the findings from the SDOF analyses on ground motion scaling method. In total, 300,000 SDOF
analyses and 80 MDOF analyses are conducted by the research.
In summary, this research shows that: (1) previous capacity-demand index relationships
developed using linear-elastic ground motion response spectra can be significantly different than
those developed using smooth design response spectra and can lead to unconservative designs,
particularly for survival-level, soft soil, and near-field conditions; (2) the correlation between the
maximum displacement ductility demand, µ, and the other demand indices, Λ, is relatively strong;
and (3) scaling methods that work well for ground motions recorded on stiff soil and far-field con-
ditions lose their effectiveness for ground motions on soft soil and near-field conditions.
161
12.2 Conclusions
The major conclusions based on the research are summarized below. The conclusions are
presented in the following order: (1) capacity-demand index relationships; (2) effect of reference
response spectra; (3) regression analyses; and (4) ground motion scaling methods.
The following conclusions are based on mean demands from SDOF nonlinear dynamic
time-history analyses where the structural lateral force capacities, Fy = Felas/R, are determined
using the individual (IND) linear-elastic ground motion acceleration response spectra.
(1) The constant-R approach is more conservative than the constant-µ approach even when
an iteration procedure that selects the smallest R coefficient is used. Thus, the constant-R
approach should be used to develop seismic capacity-demand index relationships.
(2) For the EP hysteresis type, increased post-yield stiffness ratio, α, decreases the mean µ
and µr demands. The decrease in the µr demand with an increase in α is especially large for small
values of α and large values of R. The µp demands increase for increased α. The ny demands are
virtually independent of α.
(3) The mean µ demand is slightly larger for the SD hysteresis type than for the EP hyster-
esis type. There are large increases in µp and decreases in ny due to the shooting branch of the SD
type. At higher levels of R and for larger α, the SD type behaves similar to the EP type.
(4) For the BE hysteresis type, all mean demand indices, except µr, are significantly larger
than for the EP hysteresis type due to the lack of inelastic energy dissipation. The BE type is supe-
rior to the EP type in terms of µr since the BE type always returns to the zero displacement posi-
tion after an earthquake event.
(5) As compared to the BE hysteresis type, all mean demand indices, except µr, decrease
somewhat proportionally to the increase in βr for the BP hysteresis type. There is a slight, but
mostly negligible, increase in µr for the BP type with respect to the BE type. Thus, the objective
of reducing the µ, µp, and ny demands while keeping the µr demand within allowable limits can be
reasonably achieved with the BP hysteresis type.
(6) For structures with period T < ~1.0-1.5 sec., the mean µ, µp, and µr demands are larger
for the SE soil profile than for the SD soil profile. For longer period structures (i.e., T > ~1.0-1.5
sec.), the mean µ, µp, and µr demands are smaller for the SE soil profile than for the SD soil pro-
file. This is because individual linear-elastic ground motion acceleration response spectra were
used to determine the structure force capacities, Fy = Felas/R, resulting in increased Fy values for
the structures near the predominant ground motion period.
(7) For seismic demand level, site seismicity, and epicentral distance, the long-period
mean µ, µp, and ny demands for the survival-level, Los Angeles, and near-field ground motion
ensembles are smaller than the corresponding demands for the design-level, Boston, and far-field
162
ensembles, respectively. In these cases, the average linear-elastic acceleration response spectra of
the survival-level, Los Angeles, and near-field ground motion ensembles decrease at a faster rate
(as T increases) than the acceleration response spectra of the design-level, Boston, and far-field
ensembles, respectively, resulting in a significant reduction in the demands as the structure yields.
The following conclusions are based on mean demands from SDOF nonlinear dynamic
time-history analyses where the structure lateral force capacities, Fy = Felas/R, are determined
based on different reference response spectra.
(1) Capacity-demand index relationships developed using individual (IND) ground motion
acceleration response spectra for the basis of Fy can lead to unconservative designs, especially for
survival-level, soft (SE) soil, and near-field conditions. Using smooth design (DES) response spec-
tra to determine Fy provides demand estimates that are more consistent with current design provi-
sions.
(2) In general, the differences between the mean µ demands based on the IND spectra and
the mean µ demands based on the average (AVG) spectra are small.
(3) Modest increases in mean µ demands are usually observed when using smooth (AVG
and DES) response spectra in the development of capacity-demand index relationships for sites
with a stiff (SD) soil profile, with low seismicity (Boston), and under design-level ground motions.
In some cases, the IND spectra can result in slightly larger µ demands as compared with the
smooth reference response spectra.
(4) Using design (DES) response spectra from current U.S. seismic provisions (ICBO,
1997; ICC, 2000) to determine Fy can result in mean µ demands that are extreme and possibly
uncontrollable for sites with a soft (SE) soil profile, with high seismicity (e.g., Los Angeles),
within close proximity to an active fault (NF), or under survival-level ground motions. Thus,
either the DES spectra in current seismic provisions need to be modified (e.g., using AVG spectra
instead) or the R coefficients recommended in the provisions should be reduced under these con-
ditions.
(5) The differences between the mean µ demands based on the IND spectra and the
smooth (AVG and DES) response spectra are not significantly affected by the hysteresis type.
(6) Cross-correlations between µ and the other demand indices, Λ = µp, µr, and ny, are
dependent on the reference response spectra used to determine Fy = Felas/R. The Λ-µ relationships
are significantly affected by reference response spectra for: (1) µ versus µp and µ versus ny rela-
tionships for the Boston design-level SD and SE soil ground motions using the EP hysteresis type;
(2) µ versus ny relationship for the Los Angeles design-level SD soil NF ground motions using the
EP hysteresis type; and (3) µ versus ny relationship for the Los Angeles survival-level SD soil
ground motions using the BE and BP hysteresis types.
163
Regression Analyses
The following conclusions are based on regression analyses of mean demand estimates
from SDOF nonlinear dynamic time-history analysis results.
(1) R-µ-T relationships were developed based on the dynamic analysis results using the
two-step nonlinear regression analysis scheme and the form of the regression equations by Nassar
and Krawinkler (Eqs. 2.1-2.3). The resulting R-µ-T relationships provide a reasonable representa-
tion of the mean demands from the dynamic analyses. The R-µ spectra from these regression anal-
yses reinforce the previous findings for the mean R-µ spectra from the dynamic analyses.
(2) Relationships between the demand indices, µ, µp, µr, and ny, were developed in the
form of Equations 3.43 and 3.44 using a simple one-step nonlinear regression analysis. The corre-
lation between µ and the other demand indices is relatively strong, especially for µp. In some
cases, the cross-correlations between the demand indices show weak to no correlation, indicating
that these demand indices can carry independent measures of seismic demand. Designers should
be careful in using the regression relationships developed for these cases.
(3) The regression relationships developed by the research, which take into account a wide
variety of seismic and structural parameters, can be integrated into current and future seismic
design approaches. As an example, a “capacity spectrum procedure” can be easily implemented
either by using the inelastic demand spectra provided in Figures 10.1 through 10.4 or by direct
calculation using Equations 2.1-2.3, 2.12, and 2.13 (with regression coefficients a and b from
Table 8.2), and smooth design response spectra from current seismic design provisions. Further-
more, using the relationships between µ and the other demand indices (Eqs. 3.43 and 3.44 and
regression coefficients d, f, g, and h from Tables 9.1-9.4 and 9.7-9.10), a more extensive perfor-
mance-based seismic design procedure can be realized. For example, for a given “target-µ” as part
of a displacement-based design procedure, it is possible to design (or redesign) structures with
enhanced objectives in mind (e.g., limiting residual displacement and/or cumulative damage). A
design example that demonstrates this procedure is provided in the research.
The following conclusions are based on demand estimates from SDOF and MDOF nonlin-
ear dynamic time-history analyses under ground motions scaled using different methods.
(1) For the very dense (SC) and stiff (SD) soil profiles at T ≅ 1 sec., the EPV method is sig-
nificantly more effective in reducing scatter in the maximum displacement demand than the PGA
method. This can be expected since the EPV method scales the ground motions to the average
spectral pseudo-velocity at around a period of 1 sec.
(2) For the SC, SD, and soft (SE) soil profiles at longer periods, the MIV method is more
effective in reducing the scatter in the maximum displacement demand than the PGA method,
regardless of the R coefficient. The effectiveness of the MIV method increases with R for shorter
periods. The scatter for the MIV method approaches the scatter for the PGA method as T
increases.
164
(3) Except for short and very short-period structures, the Sˆa ( T o ) method is significantly
more effective in reducing the scatter than the PGA method for the SC and SD soil profiles, espe-
cially for small R coefficients.
(4) Since the Sˆa ( T o → T µ ) method accounts for the amount of nonlinear behavior
expected in the structure, the scatter using this scaling method is smaller than the scatter using the
Sˆa ( T o ) method, particularly for larger R values. However, the differences between the two meth-
ods are usually not very large.
(5) The SD, BE, and BP hysteresis type trends are almost identical to the trends observed
for the EP type. The dependency of scatter (in terms of covariance of the maximum displacement
demand and covariance of the maximum displacement ductility demand) on the hysteretic behav-
ior of the structure is very small. Thus, it is concluded that the amount of scatter in the maximum
displacement demand and the amount of scatter in the maximum displacement ductility demand
are not significantly affected by the hysteresis type or by the amount of inelastic energy dissipated
by the structure.
(6) For the SC, SD, and SE soil profiles, the MIV method is more effective in reducing the
scatter in the maximum displacement demand than the Sˆa ( T o ) and Sˆa ( T o → T µ ) methods for a
wide range of periods, especially for larger R coefficients. The reduction in the scatter when using
the MIV method is particularly significant for the SE soil profile. Thus, it is evident that scaling
ground motions based on spectral intensity measures (i.e., Sˆa ( T o ) and Sˆa ( T o → T µ ) ) is not nec-
essarily effective for all site soil characteristics, structure lateral strengths, and periods.
(7) For the SE and near-field (NF) ground motion ensembles, the effectiveness of the
Sˆa ( T o ) and Sˆa ( T o → T µ ) methods with respect to the PGA method is significantly decreased.
Except for long periods, the Sˆa ( T o ) and Sˆa ( T o → T µ ) methods are less effective than the PGA
method for the SE soil profile. For the NF ground motion ensemble, the effectiveness of the
Sˆa ( T o ) and Sˆa ( T o → T µ ) methods in the long period range decreases as R increases.
(8) In general, the dependency of the scatter in the maximum displacement demand on R is
larger for the NF and SE soil ground motion ensembles than for the SC and SD soil ensembles.
(9) For the PGA, EPA, and A95 scaling methods, the dependency of the scatter in the max-
imum displacement demand on the period, T, is significantly decreased for the near-field (NF)
ground motion ensemble.
(10) The COV values (i.e., the sample standard deviation normalized by the sample mean)
for the scatter in the maximum lateral displacements of the MDOF four-story and eight-story
structures using the MIV and Sˆa ( T o → T µ ) methods compare well with the corresponding COV
values from the SDOF models. Thus, the results from the MDOF analyses seem to satisfactorily
verify the results from the SDOF analyses.
(11) The ratios between the COV values based on the MIV method and the Sˆa ( T o → T µ )
method using the MDOF structures compare well with the ratios using the SDOF models. Thus,
the MDOF analyses support the finding from the SDOF analyses that the MIV method provides a
165
significant reduction in scatter as compared to the Sˆa ( T o → T µ ) method for the SE soil profile
within a wide range of periods, especially for larger R coefficients.
(12) The increased effectiveness of the MIV method over the Sˆa ( T o ) and Sˆa ( T o → T µ )
methods, coupled with its simplicity, make the MIV method advantageous over the other scaling
methods described herein for a wide range of site and structure characteristics. The advantages of
the MIV method over the Sˆa ( T o ) and Sˆa ( T o → T µ ) methods come from: (1) the MIV method is
based only on ground motion parameters and not on structure parameters (e.g., linear-elastic
structure period, To), which may not be known in advance; and (2) ground motions scaled using
the MIV method can be used to analyze structures with different properties (e.g., with different
To) since the ground motions do not need to be rescaled for each structure. Thus, it is recom-
mended that nonlinear dynamic analyses are conducted using ground motion records scaled based
on the MIV method, particularly for larger R coefficients and soft soil profiles for the period
ranges given in the research.
(13) The scatter in the maximum displacement ductility demand, µ, in terms of the sample
standard deviation, σ, increases as R increases. The scatter for µ in terms of the coefficient of vari-
ation, COV, which normalizes the sample standard deviation by the sample mean, does not
increase with R as much as the increase in terms of the sample standard deviation, σ.
(14) Dimensionless demand estimates, such as µ = ∆nlin/∆y, are independent of the scaling
method when the individual (IND) linear-elastic ground motion response spectra are used to
determine the lateral force capacity, Fy = Felas/R.
166
REFERENCES
1. Ang, A. H-S. and Tang, W. H., 1975, Probability Concepts in Engineering Planning and
Design, John Wiley and Sons, New York.
2. American Society of Civil Engineers (ASCE), 2000, Prestandard and Commentary for the
Seismic Rehabilitation of Buildings, FEMA 356, prepared for the Federal Emergency Manage-
ment Agency, Washington, DC.
3. Applied Technology Council (ATC), 1978, Tentative Provisions for the Development of Seis-
mic Regulations for Buildings, ATC 3-06.
4. ATC, 1996, Seismic Evaluation and Retrofit of Concrete Buildings, ATC 40.
5. Arias, A., 1969, “A Measure of Earthquake Intensity,” Seismic Design of Nuclear Power
Plants, The Massachusetts Institute of Technology, Cambridge Press.
6. Astarlioglu, S., Memari, A. M., and Scanlon, A., 1998, “Use of Inelastic Design Spectra in
Seismic Design of Precast Concrete Wall Panel Structures,” Proceedings of the 6th U.S.
National Conference on Earthquake Engineering, Seattle, Earthquake Engineering Research
Institute, Oakland. (CD-ROM)
7. Attalla, M. R., Paret, T. F., and Freeman, S. A., 1998, “Near-Source Behavior of Buildings
Under Pulse-Like Earthquakes,” Proceedings of the 6th U.S. National Conference on Earth-
quake Engineering, Seattle, Earthquake Engineering Research Institute, Oakland. (CD-ROM)
8. Bardet, J. P., Ichii, K., and Lin, C. H., 2000, “EERA: A Computer Program for Equivalent-
Linear Earthquake Site Response Analyses of Layered Soil Deposits,” Department of Civil
Engineering, University of Southern California, [WWW Document], URL http://
geoinfo.usc.edu/gees/Software/EERA2000/Default.htm
9. Bates, D. M. and Watts, D. G., 1988, Nonlinear Regression Analysis and Its Applications,
John Wiley and Sons, New York.
10. Bendat, J. S. and Piersol, A. G., 1986, Random Data: Analysis and Measurement Procedures,
John Wiley and Sons, New York.
11. Borzi, B. and Elnashai, A.S., 2000, “Refined Force Reduction Factors for Seismic Design,”
Engineering Structures, 22, 1244-1260.
167
12. Bozorgnia, Y. and Mahin, S. A., 1998, “Ductility and Strength Demands of Near-Fault
Ground Motions of the Northridge Earthquake,” Proceedings of the 6th U.S. National Confer-
ence on Earthquake Engineering, Seattle, Earthquake Engineering Research Institute, Oak-
land. (CD-ROM)
13. Building Seismic Safety Council (BSSC), 1994, NEHRP Recommended Provisions for Seis-
mic Regulations for New Buildings, Parts 1 and 2, FEMA 222A and FEMA 223A, Federal
Emergency Management Agency, Washington, DC.
14. BSSC, 1998, NEHRP Recommended Provisions for Seismic Regulations for New Buildings
and Other Structures, Part 1 and 2, FEMA 302 and FEMA 303, Federal Emergency Manage-
ment Agency, Washington, DC.
15. Carballo, J. E. and Cornell, C. A., 1998, “Input to Nonlinear Structural Analysis: Modification
of Available Accelerograms for Different Source and Site Characteristics,” Proceedings of the
6th U.S. National Conference on Earthquake Engineering, Seattle, Earthquake Engineering
Research Institute, Oakland. (CD-ROM)
16. Cernica, J. N., 1995, Geotechnical Engineering: Foundation Design, John Wiley and Sons,
New York.
17. Chopra, A. K., 1995, Dynamics of Structures: Theory and Applications to Earthquake Engi-
neering, Prentice Hall, Inc., New Jersey.
18. Chopra, A. K. and Goel, R. K., 1999, “Capacity-Demand-Diagram Methods Based on Inelas-
tic Demand Spectrum,” Earthquake Spectra, 15, 637-656.
19. Clough, R. W. and Penzien, J., 1993, Dynamics of Structures, Second Edition, McGraw-Hill,
Inc., New York.
20. Elghadamsi, F. E. and Mohraz, B., 1987, “Inelastic Earthquake Spectra,” Earthquake Engi-
neering and Structural Dynamics, 15, 91-104.
21. Fajfar, P., 1999, “Capacity Spectrum Method Based on Inelastic Spectra,” Earthquake Engi-
neering and Structural Dynamics, 28, 979-993.
23. Fleischman, R. B. and Farrow, K. T., 2001, “Dynamic Behavior of Perimeter Lateral-System
Structures with Flexible Diaphragms,” Earthquake Engineering and Structural Dynamics, 30,
745-763.
24. Foutch, D. and Shi, S., 1998, “Effects of Hysteresis Type on the Seismic Response of Build-
ings,” Proceedings of the 6th U.S. National Conference on Earthquake Engineering, Seattle,
Earthquake Engineering Research Institute, Oakland. (CD-ROM)
168
25. Freeman, S. A., Nicoletti, J. P., and Tyrell, J. V., 1975, “Evaluation of Existing Buildings for
Seismic Risk - A Case Study of Puget Sound Naval Shipyard, Bremerton, Washington,” Pro-
ceedings of 1st U.S. National Conference on Earthquake Engineering, Earthquake Engineer-
ing Research Institute, Berkeley, 113-122.
26. Freeman, S. A., 1998, “Development and Use of Capacity Spectrum Method,” Proceedings of
the 6th U.S. National Conference on Earthquake Engineering, Seattle, Earthquake Engineer-
ing Research Institute, Oakland. (CD-ROM)
27. Fu, Y. and Cherry, S., 1999, “Simplified Seismic Code Design Procedure for Friction-Damped
Steel Frames,” Canadian Journal of Civil Engineering, 26 (1), 55-71.
28. Ghobarah, A., Abou-Elfath, H., and Biddah, A., 1999, “Response-Based Assessment of Struc-
tures,” Earthquake Engineering and Structural Dynamics, 28, 79-104.
29. Hachem, M., 2000, “BISPEC Help Manual,” [WWW Document], URL http://www.ce.berke-
ley.edu/~hachem/bispec/index.html
30. Hadidi-Tamjed, H., 1987, “Statistical Response of Inelastic SDOF Systems Subjected to
Earthquakes,” Ph.D. Dissertation, Department of Civil Engineering, Stanford University, Palo
Alto, CA.
31. Hall, J. F., Heaton, T. H., Halling, M. W., and Wald, D. J., 1995, “Near-Source Ground Motion
and its Effects on Flexible Buildings,” Earthquake Spectra, 11, 569-605.
32. Hart, G. C. and Wong, K., 2000, Structural Dynamics for Structural Engineers, John Wiley
and Sons, Inc., New York.
33. International Code Council (ICC), 2000, International Building Code 2000 (IBC 2000), Falls
Church, VA.
34. International Conference of Building Officials (ICBO), 1994, Uniform Building Code, Whit-
tier, CA.
36. Kasai, K., Fu., Y., and Watanabe, A., 1998, “Passive Control Systems for Seismic Damage
Mitigation,” ASCE Journal of Structural Engineering, 124 (5), 501-512.
37. Kawashima, K., MacRae, G.A., Hoshikuma, J., and Nagaya, K., 1998, “Residual Displace-
ment Response Spectrum,” Journal of Structural Engineering, ASCE, 124 (5), 523-530
38. Kennedy, R. P., Short, S. A., Merz, K. L., Tokarz, F. Z., Idriss, I. M., Powers, M. S., and
Sadigh, K., 1984, “Engineering Characterization of Ground Motion- Task I: Effects of Char-
acteristics of Free-Field Motion on Structural Response,” NUREG/CR-3805, U.S. Regulatory
Commission, Washington D.C.
169
39. Krawinkler, H. and Gupta, A., 1998, “Effect of Stiffness Degradation on Deformation
Demands for SDOF and MDOF Structures,” Proceedings of the 6th U.S. National Conference
on Earthquake Engineering, Seattle, Earthquake Engineering Research Institute, Oakland.
(CD-ROM)
40. Krawinkler, H. and Rahnama, M., 1992, “Effects of Soft Soil on Design Spectra,” Proceed-
ings, 10th World Conference on Earthquake Engineering, 5841-5846.
41. Krawinkler, H., 1994, “New Trends in Seismic Design Methodology,” Proceedings of 10th
European Conference on Earthquake Engineering, 2, Balkema, Rotterdam, 821-830.
42. Kunnath, S. K. and Gupta, B., 1998, “Effect of Hysteretic Model Parameters on Inelastic Seis-
mic Demands,” Proceedings of the 6th U.S. National Conference on Earthquake Engineering,
Seattle, Earthquake Engineering Research Institute, Oakland. (CD-ROM)
43. Kurama, Y., Sause, R., Pessiki, S., and Lu, L. W., 1997, “Seismic Design and Response Eval-
uation of Unbonded Post-Tensioned Precast Concrete Walls,” Research Report No. EQ-97-01,
Department of Civil and Environmental Engineering, Lehigh University, Bethlehem, PA.
44. Kurama, Y. C., 2001, “Simplified Seismic Design Approach for Friction-Damped Unbonded
Post-Tensioned Precast Walls,” ACI Structural Journal, 98 (5), 705-716.
45. Lam, N., Wilson, J., and Hutchinson, G., 1998, “The Ductility Reduction Factor in the Seis-
mic Design of Buildings,” Earthquake Engineering and Structural Dynamics, 27, 749-769.
46. Loh, C. and Ho, R., 1990, “Seismic Damage Assessment Based on Different Hysteretic
Rules,” Earthquake Engineering and Structural Dynamics, 19, 753-771.
47. MacRae, G. A. and Kawashima, K., 1997, “Post-Earthquake Residual Displacements of Bilin-
ear Oscillators,” Earthquake Engineering and Structural Dynamics, 26, 701-716.
48. Mahin, S., and Lin, J., 1983, “Construction of Inelastic Response Spectra for Single Degree of
Freedom Systems,” Report No. UCB/EERC-83/17, Earthquake Engineering Research Center,
University of California, Berkeley, CA.
49. Makris, N., 1997, “Rigid-Plasticity-Viscosity: Can Electrorheological Dampers Protect Base-
Isolated Structures from Near-Source Ground Motions?,” Earthquake Engineering and Struc-
tural Dynamics, 26, 571-591.
50. Malhotra, P. K., 1999, “Response of Buildings to Near-Field Pulse-Like Ground Motions,”
Earthquake Engineering and Structural Dynamics, 28, 1309-1326.
51. Martinez-Rueda, J. E., 1998, “Scaling Procedure for Natural Accelerograms Based on a Sys-
tem of Spectrum Intensity Scales,” Earthquake Spectra, 14 (1), 135-152.
170
53. McCann, M. W. and Shah, H. C., 1979, “Determining Strong-Motion Duration of Earth-
quakes,” Bulletin of the Seismological Society of America, 69 (4), 1253-1265.
54. Miranda, E., 1993, “Site-Dependent Strength-Reduction Factors,” Journal of Structural Engi-
neering, American Society of Civil Engineers, 119, 3503-3519.
55. Miranda, E. and Bertero, V. V., 1994, “Evaluation of Strength Reduction Factors for Earth-
quake-Resistant Design,” Earthquake Spectra, Earthquake Engineering Research Institute, 10
(2), 357-379.
56. Montgomery, D. C. and Peck, E. A., 1991, Introduction to Linear Regression Analysis, John
Wiley and Sons, New York.
57. Naeim, F., 1995, “On Seismic Design Implications of the 1994 Northridge Earthquake
Records,” Earthquake Spectra, 11, 91-109.
58. Naeim, F. and Anderson, J., 1993, “Classification and Evaluation of Earthquake Records for
Design,” The 1993 NEHRP Professional Fellowship Report, Earthquake Engineering
Research Institute.
59. Nassar, A. and Krawinkler, H., 1991, “Seismic Demands for SDOF and MDOF Systems,”
John A. Blume Earthquake Engineering Center, Report 95, Dept. of Civil Engineering, Stan-
ford University.
60. Nassar, A., Osteraas, L., and Krawinkler, H., 1992, “Seismic Design Based on Strength and
Ductility Demands,” Proceedings, 10th World Conference on Earthquake Engineering, 5861-
5866.
61. Nau, J. M. and Hall, W. J., 1984, “Scaling Methods for Earthquake Response Spectra,” Jour-
nal of Structural Engineering, ASCE, 110 (7), 91-109.
62. Newmark, N. M. and Hall, W. J., 1973, “Procedures and Criteria for Earthquake Resistant
Design,” Building Sci. Series No. 46, National Bureau of Standards, U.S. Dept. of Commerce,
Washington, D.C.
63. Osteraas, J. D. and Krawinkler, H., 1991, “Strength and Ductility Considerations in Seismic
Design,” John A. Blume Earthquake Engineering Center, Report 90, Dept. of Civil Engineer-
ing, Stanford University.
64. Paulay, T. and Priestley, M. J. N., 1992, Seismic Design of Reinforced Concrete and Masonry
Buildings, John Wiley and Sons, Inc., New York.
65. Peng, M. H., Elghadamsi, F. E., and Mohraz, B., 1988, “A Stochastic Procedure for Nonlinear
Response Spectra,” Proceedings, 9th World Conference on Earthquake Engineering, Tokyo-
Kyoto, Japan, 5, 1069-1074.
171
66. Prakash, V., Powell, G., and Campbell, S., 1993, “DRAIN-2DX Base Program Description
and User Guide: Version 1.10,” Report UBC/SEMM-93/17, Department of Civil Engineering,
University of California, Berkeley, CA.
67. Rahnama, M. and Krawinkler, H., 1993, “Effects of Soft Soil and Hysteresis Model on Seis-
mic Demands,” John A. Blume Earthquake Engineering Center, Report 108, Dept. of Civil
Engineering, Stanford University.
68. Reinhorn, A. M., 1997, “Inelastic Analysis Techniques in Seismic Evaluations,” P. Fajfar and
H. Krawinkler (eds.), Seismic Design Methodologies for the Next Generation of Codes,
Balkema, Rotterdam, 277-287.
69. Riddell, R. and Newmark, N. M., 1979, “Statistical Analysis of the Response of Nonlinear
Systems Subjected to Earthquakes,” Structural Research Series No. 468, Dept. of Civil Engi-
neering, University of Illinois, Urbana, IL.
70. Riddell, R., Hidalgo, P., and Cruz, E., 1989, “Response Modification Factors for Earthquake
Resistant Design of Short Period Structures,” Earthquake Spectra, Earthquake Engineering
Research Institute, 5 (3), 571-590.
71. Sarma, S. K. and Yang, K. S., 1987, “An Evaluation of Strong Motion Records and a New
Parameter A95,” Earthquake Engineering and Structural Dynamics, 15, 119-132.
72. Shome, N. and Cornell, C. A., 1998, “Normalization and Scaling Accelerograms for Nonlin-
ear Structural Analysis,” Proceedings of the 6th U.S. National Conference on Earthquake
Engineering, Seattle, Earthquake Engineering Research Institute, Oakland. (CD-ROM)
73. Shome, N., Cornell, C. A., Bazzurro, P., and Carballo, J. E., 1998, “Earthquakes, Records, and
Responses,” Earthquake Spectra, 14, 469-500.
74. Singh, J., 1985, “Earthquake Ground Motions: Implications for Designing Structures and
Reconciling Structural Damage,” Earthquake Spectra, Earthquake Engineering Research
Institute, 1 (2), 239-270.
75. Somerville, P., Smith, N., Punyamurthula, S., and Sun, J., 1997, “Development of Ground
Motion Time Histories for Phase 2 of the FEMA/SAC Steel Project,” Report SAC/BD-97/04,
SAC Joint Venture, Sacramento.
76. Structural Engineers Association of California (SEAOC), 1996, Recommended Lateral Force
Requirements and Commentary, Sixth Edition, Seismology Committee, Sacramento, CA.
172
78. Uang, C.-M., 1995, “A Balanced Seismic Steel Design Procedure for Strength and Ductility
Requirements,” Restructuring: America and Beyond, Proceedings of the 1995 Structures Con-
gress, American Society of Civil Engineering, 421-426.
79. Vidic, T., Fajfar, P., and Fischinger, M., 1994, “Consistent Inelastic Design Spectra: Strength
and Displacement,” Earthquake Engineering and Structural Dynamics, 23, 507-521.
80. Wu, S., 1995, “Seismic Analysis and Retrofit of Non-Ductile Reinforced Concrete Frame
Buildings,” Ph.D. Dissertation, Dept. of Civil and Env. Eng., Lehigh Univ., Bethlehem, PA.
81. Vuetic, M. and Dobry, R., 1991, “Effect of Soil Plasticity on Cyclic Response,” Journal of the
Geotechnical Engineering Division, ASCE, 111 (1), 89-107.
173
APPENDIX A
Table A.1: University of Notre Dame (UND) very dense (SC) soil ensemble
Peak
Max. Effective
Epicentral Ground
Ground motion Site soil Incremental Peak Accel.,
Station Abbreviation distance Accel.,
record description Vel., MIV EPA
(km) PGA
(cm/sec) (g)
(g)
Loma Prieta, 1989 San Francisco-Presidio LPPR rock 102 0.20 45.4 0.16
Quebec, 1988 St. Andre Du Lac-St. Jean SASJ bedrock 64 0.09 1.83 0.04
Michoacan, 1985 Zihuatanero-Aeropuerto MIZI rock 166 0.17 26.4 0.21
San Francisco, 1957 San Francisco--Golden Gate Park SFSF bedrock, chert 16 0.08 5.50 0.09
Parkfield, 1966 Parkfield--Cholame Shandon #2 PACH alluvium over 21 0.49 100.59 0.38
sstone
San Fernando, 1971 Castaic--Old Ridge Road SFCA bedrock, sand- 27 0.32 25.60 0.34
stone
Morgan Hill, 1984 Gilroy #6 MHGI bedrock 37 0.42 24.65 0.33
Central Chile, 1985 Valparaiso CCVA bedrock, volcanic 26 0.18 22.41 0.26
Michoacan, 1985 La Union MILU bedrock, meta- 80 0.15 16.90 0.25
andesite brec
Michoacan, 1985 La Villita MILV bedrock, tonalite 40 0.12 23.02 0.11
San Salvador, 1986 National Geographic Institute SSNG bedrock, fluviatile 14 0.53 115.54 0.43
pumice
San Salvador, 1986 Institute of Urban Construction SSUC bedrock, fluviatile 15 0.68 81.52 0.53
pumice
San Salvador, 1986 Geotechnical Research Center SSGR bedrock, fluviatile 15 0.69 91.42 0.66
pumice
Whittier, 1987 Mt. Wilson WHMW quartz diorite 18 0.17 3.67 0.06
Loma Prieta, 1989 Corralitos LPCO alluvium, land- 1 0.48 83.78 0.44
slide deposits
Loma Prieta, 1989 Santa Cruz LPSC bedrock, lime- 23 0.41 29.72 0.52
stone
Loma Prieta, 1989 San Francisco--Cliff House LPCH bedrock, fran- 104 0.11 33.65 0.07
ciscan melange
Loma Prieta, 1989 San Francisco--Pacific Heights LPPH bedrock, fran- 100 0.06 23.39 0.04
ciscan melange
Loma Prieta, 1989 San Francisco--Rincon Hill LPRH bedrock, fran- 98 0.09 13.30 0.07
ciscan melange
Loma Prieta, 1989 Yerba Buena Island LPYB bedrock, fran- 99 0.07 14.43 0.06
ciscan melange
174
Table A.2: University of Notre Dame (UND) stiff (SD) soil ensemble
Peak
Max. Effective
Epicentral Ground
Site soil Incremental Peak Accel.,
Ground motion record Station Abbreviation distance Accel.,
description Vel., MIV EPA
(km) PGA
(cm/sec) (g)
(g)
Northridge, 1994 Newhall-LA Co. Fire Station NONW alluvium 20 0.59 153.07 0.70
San Fernando, 1971 Orion Blvd. SFOR deep cohesionless 21 0.25 45.81 0.29
Kern County, 1952 Taft--Lincoln School Tunnel KCLS alluvium 56 0.15 20.32 0.19
San Fernando, 1971 Ave. of the Stars SFAS silt and sand layers 38 0.14 11.09 0.17
Whittier, 1987 Tarzana--Cedar Hill Nursery WHTA alluvium over silt- 110 0.54 21.77 0.48
stone
Whittier, 1987 7215 Bright Ave. WHBT alluvium 21 0.61 54.88 0.61
Imperial Valley, 1979 Bonds Corner IVBC alluvium 3 0.58 75.96 0.68
Imperial Valley, 1979 James Road IVJR alluvium 22 0.52 52.04 0.52
Imperial Valley, 1979 Imperial Valley College IVIV alluvium 21 0.33 64.30 0.25
Central Chile, 1985 El Almendral CCEA compacted fill 84 0.29 47.18 0.33
Whittier, 1987 Long Beach--Ranchos Los WHLC alluvium 47 0.24 17.56 0.23
Cerritos
Loma Prieta, 1989 Oakland--two story office bldg LPOA alluvium 49 0.26 54.59 0.21
175
Table A.3: University of Notre Dame (UND) soft (SE) soil ensemble
Peak
Max. Effective
Epicentral Ground
Site soil Incremental Peak
Ground motion record Station Abbreviation distance Accel.,
description Vel., MIV Accel., EPA
(km) PGA
(cm/sec) (g)
(g)
Loma Prieta, 1989 Foster City LPFO bay mud 65 0.28 75.10 0.24
Bucharest, Romania, 1977 Bucharest RUBU soft 174 0.21 120.46 0.13
Michoacan, Mexico City, 1985 Secretaria de Comunica- MISE soft clay 400 0.17 117.96 0.09
ciones Transportes
Loma Prieta, 1989 Treasure Island LPTR fill 98 0.16 57.15 0.14
Eureka, 1954 Basement, Eureka Federal EUF1 partially consoli- 1102 0.17 37.70 0.20
Bldg. dated sediment
Eureka, 1954 Basement, Eureka Federal EUF2 partly consoli- 1102 0.26 36.33 0.32
Bldg. dated sediment
Loma Prieta, 1989 Oakland, Outer Harbor OHW1 alluvium, bay 98 0.27 72.11 0.21
Wharf mud/fill
Loma Prieta, 1989 Oakland, Outer Harbor OHW2 alluvium, bay 98 0.29 57.58 0.24
Wharf mud/fill
Bucharest, Romania, 1977 Building Research Institute RUB1 soft 107 0.11 15.05 0.10
Bucharest, Romania, 1977 Building Research Institute RUB2 soft 107 0.18 53.46 0.13
Michoacan, Mexico City, 1985 Tlahuac Bombas TLB1 alluvial, soft soil, 381 0.11 59.41 0.05
Xochimilco clay
Michoacan, Mexico City, 1985 Tlahuac Bombas TLB2 alluvial, soft soil, 381 0.14 65.49 0.06
Xochimilco clay
Michoacan, Mexico City, 1985 Tlahuac Deportivo MIDX alluvial, soft, clay 97 0.11 44.67 0.06
Xochimilco
Loma Prieta, 1989 Oakland--Outer Harbor LPOH bay mud 95 0.27 72.11 0.21
Wharf
Michoacan, Mexico City, 1985 Central de Abastos--Frig- MIFI soft clay 389 0.10 72.61 0.05
orifico
Michoacan, Mexico City, 1985 Central de Abastos--Oficina MIOF soft clay 389 0.08 74.31 0.04
Loma Prieta, 1989 San Francisco Int. Airport LPIA bay mud 79 0.33 48.90 0.29
Loma Prieta, 1989 San Francisco comm. build- LPCB fill over mud 95 0.16 28.65 0.15
ing
176
Table A.4: Nassar and Krawinkler (N&K) 15s very dense (SC) soil ensemble
Peak
Max. Effective
Epicentral Ground
Site soil Incremental Peak Accel.,
Ground motion record Station Abbreviation distance Accel.,
description Vel., MIV EPA
(km) PGA
(cm/sec) (g)
(g)
Kern County, 1952 Taft Lincoln School Tunnel eq03 rock 43.0 0.16 20.32 0.14
Kern County, 1952 Taft Lincoln School Tunnel eq04 rock 43.0 0.18 27.05 0.15
Lower California, 1934 Imperial Valley eq05 rock 64.0 0.16 22.29 0.15
Lower California, 1934 Imperial Valley eq06 rock 64.0 0.18 21.56 0.17
Western Washington, 1949 Olympia Washington High- eq07 rock 16.0 0.16 23.53 0.18
way
Western Washington, 1949 Olympia Washington High- eq08 rock 16.0 0.28 32.97 0.22
way
Puget Sound, 1965 Olympia Washington High- eq09 rock 61.0 0.20 16.40 0.16
way
San Fernando, 1971 Castaic Old Ridge eq10 rock 29.0 0.27 37.30 0.22
Long Beach, 1933 Public Utilities Bldg. eq11 rock 27.0 0.20 34.01 0.18
Long Beach, 1933 Public Utilities Bldg. eq12 rock 27.0 0.16 23.61 0.16
Imperial Valley, 1979 Holtville P.O. eq14 rock 19.0 0.25 58.42 0.23
Imperial Valley, 1979 Calexico Fire Station eq15 rock 15.0 0.27 28.93 0.25
Coyote Lake, 1979 San Yasidro School eq16 rock 12.0 0.25 48.26 0.21
Coyote Lake, 1979 San Yasidro School eq17 rock 12.0 0.23 37.60 0.21
Coalinga, 1983 Parkfield Zone 16 eq20 rock 39.1 0.18 21.96 0.16
177
Table A.5: SAC Boston design-level stiff (SD) soil ensemble
Peak
Max. Effective
Epicentral Ground
Site soil Incremental Peak
Ground motion record Station Abbreviation distance Accel.,
description Vel., MIV Accel., EPA
(km) PGA
(cm/sec) (g)
(g)
New Hampshire, 1982 Franklin Falls bo05 firm 8.4 0.58 22.84 10.75
New Hampshire, 1982 Franklin Falls bo06 firm 8.4 0.32 20.73 10.75
178
Table A.6: SAC Boston survival-level stiff (SD) soil ensemble
Peak
Max. Effective
Epicentral Ground
Site soil Incremental Peak Accel.,
Ground motion record Station Abbreviation distance Accel.,
description Vel., MIV EPA
(km) PGA
(cm/sec) (g)
(g)
179
Table A.7: SAC Boston design-level soft (SE) soil ensemble
Peak Effective
Max.
Epicentral Ground Peak
Ground motion Site soil Incremental
Station Abbreviation distance Accel., Accel.,
record description Vel., MIV
(km) PGA EPA
(cm/sec)
(g) (g)
180
Table A.8: SAC Boston survival-level soft (SE) soil ensemblea
Peak
Max. Effective
Epicentral Ground
Site soil Incremental Peak Accel.,
Ground motion record Station Abbreviation distance Accel.,
description Vel., MIV EPA
(km) PGA
(cm/sec) (g)
(g)
a Ground motions generated using EERA site response analysis program (Bardet et al., 2000), as
described in Section 3.3
181
Table A.9: SAC Los Angeles design-level stiff (SD) soil ensemble
Peak
Max. Effective
Epicentral Ground
Site soil Incremental Peak
Ground motion record Station Abbreviation distance Accel.,
description Vel., MIV Accel., EPA
(km) PGA
(cm/sec) (g)
(g)
Imperial Valley, 1979 Array #5 la03 firm 4.1 0.39 103.0 1.01
Imperial Valley, 1979 Array #5 la04 firm 4.1 0.49 74.75 1.01
Imperial Valley, 1979 Array #6 la05 firm 1.2 030 106.4 0.84
Imperial Valley, 1979 Array #6 la06 firm 1.2 0.23 81.73 0.84
North Palm Springs, 1986 dhsp la19 firm 6.7 1.02 99.50 2.97
North Palm Springs, 1986 dhsp la20 firm 6.7 0.99 150.3 2.97
182
Table A.10: SAC Los Angeles survival-level stiff (SD) soil ensemble
Peak
Max. Effective
Epicentral Ground
Site soil Incremental Peak
Ground motion record Station Abbreviation distance Accel.,
description Vel., MIV Accel., EPA
(km) PGA
(cm/sec) (g)
(g)
Kobe, 1995 Kobe JMA la21 firm 3.4 1.28 275.1 1.15
Kobe, 1995 Kobe JMA la22 firm 3.4 0.92 241.8 1.15
Loma Prieta, 1989 Los Gatos la23 firm 3.5 0.42 87.07 0.82
Loma Prieta, 1989 Los Gatos la24 firm 3.5 0.47 210.8 0.82
183
Table A.11: SAC Los Angeles design-level soft (SE) soil ensemble
Peak Effective
Max.
Epicentral Ground Peak
Ground motion Site soil Incremental
Station Abbreviation distance Accel., Accel.,
record description Vel., MIV
(km) PGA EPA
(cm/sec)
(g) (g)
184
Table A.12: SAC Los Angeles survival-level soft (SE) soil ensemblea
Peak
Max. Effective
Epicentral Ground
Site soil Incremental Peak
Ground motion record Station Abbreviation distance Accel.,
description Vel., MIV Accel., EPA
(km) PGA
(cm/sec) (g)
(g)
a Ground motions generated using EERA site response analysis program (Bardet et al., 2000), as
described in Section 3.3
185
Table A.13: SAC Los Angeles near-field design-level stiff (SD) soil ensemble
Peak Effective
Max.
Epicentral Ground Peak
Ground motion Site soil Incremental
Station Abbreviation distance Accel., Accel.,
record description Vel., MIV
(km) PGA EPA
(cm/sec)
(g) (g)
Loma Prieta, 1989 Los Gatos nf03 firm 3.5 0.72 268.87 0.59
Loma Prieta, 1989 Lexington Dam nf05 firm 6.3 0.69 243.65 0.35
Northridge, 1994 Olive View nf15 firm 6.4 0.73 142.39 0.65
Elysian Park 1 Elysian Park 1 nf21 firm 17.5 0.86 186.04 0.97
Elysian Park 2 Elysian Park 2 nf23 firm 10.7 1.80 582.84 1.46
Elysian Park 3 Elysian Park 3 nf25 firm 11.2 1.01 251.98 1.08
Elysian Park 4 Elysian Park 4 nf27 firm 13.2 0.92 417.15 0.96
Elysian Park 5 Elysian Park 5 nf29 firm 13.7 1.16 497.39 1.00
Palos Verdes 1 Palos Verdes 1 nf31 firm 1.5 0.97 430.46 0.64
Palos Verdes 2 Palos Verdes 2 nf33 firm 1.5 0.97 438.77 0.57
Palos Verdes 3 Palos Verdes 3 nf35 firm 1.5 0.87 404.87 0.59
Palos Verdes 4 Palos Verdes 4 nf37 firm 1.5 0.79 295.83 0.54
Palos Verdes 5 Palos Verdes 5 nf39 firm 1.5 0.92 394.90 0.64
186
LPPR SASJ
acceleration (g) 0.2 0.1
0.1 0.05
0 0
−0.1 −0.05
−0.2 −0.1
0 10 20 30 40 0 10 20 30
MIZI SFSF
0.2 0.1
acceleration (g)
0.1 0.05
0 0
−0.1 −0.05
−0.2 −0.1
0 20 40 60 80 0 10 20 30 40
PACH SFCA
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 20 40 60 0 20 40 60 80
MHGI CCVA
0.5 0.2
acceleration (g)
0.1
0 0
−0.1
−0.5 −0.2
0 20 40 60 0 20 40 60 80
MILU MILV
0.2 0.2
acceleration (g)
0.1 0.1
0 0
−0.1 −0.1
−0.2 −0.2
0 20 40 60 80 0 20 40 60 80
time (sec) time (sec)
Figure A.1: Ground motion records: University of Notre Dame (UND) very
dense (SC) soil ensemble.
187
SSNG SSUC
1 1
acceleration (g) 0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 10 20 30 0 5 10 15
SSGR WHMW
1 0.2
acceleration (g)
0.5 0.1
0 0
−0.5 −0.1
−1 −0.2
0 2 4 6 8 10 0 5 10 15 20
LPCO LPSC
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 10 20 30 40 0 10 20 30 40
LPCH LPPH
0.2 0.1
acceleration (g)
0.1 0.05
0 0
−0.1 −0.05
−0.2 −0.1
0 10 20 30 40 0 10 20 30 40
LPRH LPYB
0.1 0.1
acceleration (g)
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
0 10 20 30 40 0 10 20 30 40
time (sec) time (sec)
188
NONW LPHO
1 0.2
acceleration (g)
0.5 0.1
0 0
−0.5 −0.1
−1 −0.2
0 20 40 60 0 20 40 60
LAYE NOSY
0.5 1
acceleration (g)
0.5
0 0
−0.5
−0.5 −1
0 20 40 60 80 0 20 40 60
SFOR la02
0.5 1
acceleration (g)
0.5
0 0
−0.5
−0.5 −1
0 20 40 60 0 20 40 60
KCLS SFFI
0.2 0.2
acceleration (g)
0.1 0.1
0 0
−0.1 −0.1
−0.2 −0.2
0 20 40 60 0 20 40 60
SFHO SFAS
0.5 0.2
acceleration (g)
0.1
0 0
−0.1
−0.5 −0.2
0 20 40 60 80 0 20 40 60
time (sec) time (sec)
Figure A.2: Ground motion records: University of Notre Dame (UND) stiff
(SD) soil ensemble.
189
WHTA WHBT
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 5 10 15 20 0 10 20 30 40
IVBC IVJR
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 10 20 30 40 0 10 20 30 40
IVIV CCEA
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 10 20 30 40 0 20 40 60 80 100
WHLC LPOA
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 5 10 15 20 0 10 20 30 40
WHAH WHAD
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 5 10 15 20 0 5 10 15 20
time (sec) time (sec)
190
LPFO RUBU
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 20 40 60 0 5 10 15 20
−4 MISE LPTR
x 10
2 0.2
acceleration (g)
1 0.1
0 0
−1 −0.1
−2 −0.2
0 50 100 150 200 0 20 40 60
EUF1 EUF2
0.2 0.5
acceleration (g)
0.1
0 0
−0.1
−0.2 −0.5
0 20 40 60 80 0 20 40 60 80
OHW1 OHW2
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 10 20 30 40 0 10 20 30 40
RUB1 RUB2
0.2 0.2
acceleration (g)
0.1 0.1
0 0
−0.1 −0.1
−0.2 −0.2
0 5 10 15 20 0 5 10 15 20
time (sec) time (sec)
Figure A.3: Ground motion records: University of Notre Dame (UND) soft
(SE) soil ensemble.
191
TLB1 TLB2
0.2 0.2
0 0
−0.1 −0.1
−0.2 −0.2
0 20 40 60 80 100 0 20 40 60 80 100
WHA1 WHA2
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 10 20 30 40 0 10 20 30 40
MIDX LPOH
0.2 0.5
acceleration (g)
0.1
0 0
−0.1
−0.2 −0.5
0 20 40 60 80 100 0 10 20 30 40
MIFI MIOF
0.1 0.1
acceleration (g)
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
0 20 40 60 0 50 100 150 200
LPIA LPCB
0.5 0.2
acceleration (g)
0.1
0 0
−0.1
−0.5 −0.2
0 10 20 30 40 0 10 20 30 40
time (sec) time (sec)
192
eq03 eq04
0.2 0.2
acceleration (g)
0.1 0.1
0 0
−0.1 −0.1
−0.2 −0.2
0 20 40 60 0 20 40 60
eq05 eq06
0.2 0.2
acceleration (g)
0.1 0.1
0 0
−0.1 −0.1
−0.2 −0.2
0 20 40 60 80 100 0 20 40 60 80 100
eq07 eq08
0.2 0.5
acceleration (g)
0.1
0 0
−0.1
−0.2 −0.5
0 20 40 60 80 100 0 20 40 60 80 100
eq09 eq10
0.2 0.5
acceleration (g)
0.1
0 0
−0.1
−0.2 −0.5
0 20 40 60 80 100 0 20 40 60 80
eq11 eq12
0.2 0.2
acceleration (g)
0.1 0.1
0 0
−0.1 −0.1
−0.2 −0.2
0 20 40 60 80 100 0 20 40 60 80 100
time (sec) time (sec)
Figure A.4: Ground motion records: Nassar and Krawinkler 15s very dense
(SC) soil ensemble.
193
eq14 eq15
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 10 20 30 40 0 10 20 30 40
eq16 eq17
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 10 20 30 0 10 20 30
eq20
0.2
acceleration (g)
0.1
−0.1
−0.2
0 20 40 60 80
Figure A.4 (continued): Ground motion records: Nassar and Krawinkler 15s
very dense (SC) soil ensemble.
194
bo01 bo02
0.2 0.1
acceleration (g) 0.1 0.05
0 0
−0.1 −0.05
−0.2 −0.1
0 10 20 30 0 10 20 30
bo03 bo04
0.2 0.2
acceleration (g)
0.1 0.1
0 0
−0.1 −0.1
−0.2 −0.2
0 10 20 30 0 10 20 30
bo05 bo06
1 0.5
acceleration (g)
0.5
0 0
−0.5
−1 −0.5
0 5 10 15 20 0 5 10 15 20
bo07 bo08
0.1 0.1
acceleration (g)
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
0 10 20 30 0 10 20 30
bo09 bo10
0.1 0.1
acceleration (g)
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
0 5 10 15 20 0 5 10 15 20
time (sec) time (sec)
Figure A.5: Ground motion records: SAC Boston design-level stiff (SD) soil
ensemble.
195
bo11 bo12
0.2 0.2
acceleration (g)
0.1 0.1
0 0
−0.1 −0.1
−0.2 −0.2
0 5 10 15 20 0 5 10 15 20
bo13 bo14
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 5 10 15 20 0 5 10 15 20
bo15 bo16
1 0.5
acceleration (g)
0.5
0 0
−0.5
−1 −0.5
0 10 20 30 0 10 20 30
bo17 bo18
0.2 0.5
acceleration (g)
0.1
0 0
−0.1
−0.2 −0.5
0 10 20 30 40 0 10 20 30 40
bo19 bo20
0.2 0.5
acceleration (g)
0.1
0 0
−0.1
−0.2 −0.5
0 10 20 30 40 0 10 20 30 40
time (sec) time (sec)
Figure A.5 (continued): Ground motion records: SAC Boston design-level stiff
(SD) soil ensemble.
196
bo21 bo22
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 10 20 30 0 10 20 30
bo23 bo24
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 10 20 30 0 10 20 30
bo25 bo26
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 10 20 30 0 10 20 30
bo27 bo28
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 10 20 30 0 10 20 30
bo29 bo30
0.2 0.5
acceleration (g)
0.1
0 0
−0.1
−0.2 −0.5
0 5 10 15 20 0 5 10 15 20
time (sec) time (sec)
Figure A.6: Ground motion records: SAC Boston survival-level stiff (SD) soil
ensemble.
197
bo31 bo32
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 5 10 15 20 0 5 10 15 20
bo33 bo34
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 5 10 15 20 0 5 10 15 20
bo35 bo36
2 1
acceleration (g)
1 0.5
0 0
−1 −0.5
−2 −1
0 10 20 30 0 10 20 30
bo37 bo38
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 10 20 30 40 0 10 20 30 40
bo39 bo40
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 10 20 30 40 0 10 20 30 40
time (sec) time (sec)
198
bs01 bs02
0.2 0.1
acceleration (g) 0.1 0.05
0 0
−0.1 −0.05
−0.2 −0.1
0 10 20 30 40 50 0 10 20 30 40 50
bs03 bs04
0.2 0.1
acceleration (g)
0.1 0.05
0 0
−0.1 −0.05
−0.2 −0.1
0 10 20 30 40 50 0 10 20 30 40 50
bs05 bs06
0.1 0.1
acceleration (g)
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
0 10 20 30 40 50 0 10 20 30 40 50
bs07 bs08
0.1 0.1
acceleration (g)
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
0 5 10 15 20 25 0 5 10 15 20 25
bs09 bs10
0.1 0.1
acceleration (g)
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
0 5 10 15 20 25 0 5 10 15 20 25
time (sec) time (sec)
Figure A.7: Ground motion records: SAC Boston design-level soft (SE) soil
ensemble.
199
bs11 bs12
0.6 0.4
acceleration (g)
0.4 0.2
0.2 0
0 −0.2
−0.2 −0.4
0 5 10 15 20 25 0 5 10 15 20 25
bs13 bs14
0.2 0.6
acceleration (g)
0.1 0.4
0 0.2
−0.1 0
−0.2 −0.2
0 5 10 15 20 25 0 5 10 15 20 25
bs15 bs16
0.4 0.4
acceleration (g)
0.2 0.2
0 0
−0.2 −0.2
−0.4 −0.4
0 10 20 30 40 50 0 10 20 30 40 50
bs17 bs18
0.2 0.4
acceleration (g)
0.1 0.2
0 0
−0.1 −0.2
−0.2 −0.4
0 10 20 30 40 50 0 10 20 30 40 50
bs19 bs20
0.4 0.4
acceleration (g)
0.2 0.2
0 0
−0.2 −0.2
−0.4 −0.4
0 10 20 30 40 50 0 10 20 30 40 50
time (sec) time (sec)
Figure A.7 (continued): Ground motion records: SAC Boston design-level soft
(SE) soil ensemble.
200
bs21 bs22
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 10 20 30 0 10 20 30
bs23 bs24
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 10 20 30 0 10 20 30
bs25 bs26
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 10 20 30 0 10 20 30
bs27 bs28
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 10 20 30 0 10 20 30
bs29 bs30
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 5 10 15 20 0 5 10 15 20
time (sec) time (sec)
Figure A.8: Ground motion records: SAC Boston survival-level soft (SE) soil
ensemble (generated using EERA site response analysis program, Bardet et al.,
2000).
201
bs31 bs32
0.5 0.2
acceleration (g)
0.1
0 0
−0.1
−0.5 −0.2
0 5 10 15 20 0 5 10 15 20
bs33 bs34
0.5 1
acceleration (g)
0.5
0 0
−0.5
−0.5 −1
0 5 10 15 20 0 5 10 15 20
bs35 bs36
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 10 20 30 0 10 20 30
bs37 bs38
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 10 20 30 40 0 10 20 30 40
bs39 bs40
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 10 20 30 40 0 10 20 30 40
time (sec) time (sec)
202
la01 la02
0.5 1
acceleration (g)
0.5
0 0
−0.5
−0.5 −1
0 20 40 60 0 20 40 60
la03 la04
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 10 20 30 40 0 10 20 30 40
la05 la06
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 10 20 30 40 0 10 20 30 40
la07 la08
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 20 40 60 80 100 0 20 40 60 80 100
la09 la10
1 0.5
acceleration (g)
0.5
0 0
−0.5
−1 −0.5
0 20 40 60 80 100 0 20 40 60 80 100
time (sec) time (sec)
Figure A.9: Ground motion records: SAC Los Angeles design-level stiff (SD)
soil ensemble.
203
la11 la12
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 20 40 60 0 20 40 60
la13 la14
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 20 40 60 0 20 40 60
la15 la16
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 5 10 15 0 5 10 15
la17 la18
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 20 40 60 0 20 40 60
la19 la20
2 1
acceleration (g)
1 0.5
0 0
−1 −0.5
−2 −1
0 20 40 60 0 20 40 60
time (sec) time (sec)
Figure A.9 (continued): Ground motion records: SAC Los Angeles design-
level stiff (SD) soil ensemble.
204
la21 la22
2 1
0 0
−1 −0.5
−2 −1
0 20 40 60 0 20 40 60
la23 la24
0.5 0.5
acceleration (g)
0 0
−0.5 −0.5
0 10 20 30 0 10 20 30
la25 la26
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 5 10 15 0 5 10 15
la27 la28
1 2
acceleration (g)
0.5 1
0 0
−0.5 −1
−1 −2
0 20 40 60 0 20 40 60
la29 la30
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 20 40 60 0 20 40 60
time (sec) time (sec)
Figure A.10: Ground motion records: SAC Los Angeles survival-level stiff
(SD) soil ensemble.
205
la31 la32
2 2
acceleration (g) 1 1
0 0
−1 −1
−2 −2
0 10 20 30 0 10 20 30
la33 la34
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 10 20 30 0 10 20 30
la35 la36
1 2
acceleration (g)
0.5 1
0 0
−0.5 −1
−1 −2
0 10 20 30 0 10 20 30
la37 la38
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 20 40 60 0 20 40 60
la39 la40
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 20 40 60 0 20 40 60
time (sec) time (sec)
Figure A.10 (continued): Ground motion records: SAC Los Angeles survival-
level stiff (SD) soil ensemble.
206
ls01 ls02
0.4 0.4
acceleration (g)
0.2 0.2
0 0
−0.2 −0.2
−0.4 −0.4
0 20 40 60 80 100 0 20 40 60 80 100
ls03 ls04
0.4 0.4
acceleration (g)
0.2 0.2
0 0
−0.2 −0.2
−0.4 −0.4
0 10 20 30 40 50 0 10 20 30 40 50
ls05 ls06
0.4 0.4
acceleration (g)
0.2 0.2
0 0
−0.2 −0.2
−0.4 −0.4
0 10 20 30 40 50 0 10 20 30 40 50
ls07 ls08
0.5 0.4
acceleration (g)
0.2
0 0
−0.2
−0.5 −0.4
0 20 40 60 80 100 0 20 40 60 80 100
ls09 ls10
0.5 0.4
acceleration (g)
0.2
0 0
−0.2
−0.5 −0.4
0 20 40 60 80 100 0 20 40 60 80 100
time (sec) time (sec)
Figure A.11: Ground motion records: SAC Los Angeles design-level soft (SE)
soil ensemble.
207
ls11 ls12
1 0.5
acceleration (g)
0.5
0
0
−0.5 −0.5
0 20 40 60 80 100 0 20 40 60 80 100
ls13 ls14
1 0.4
acceleration (g)
0.5 0.2
0 0
−0.5 −0.2
−1 −0.4
0 20 40 60 80 100 0 20 40 60 80 100
ls15 ls16
0.5 0.4
acceleration (g)
0.2
0
0
−0.5
−0.2
−1 −0.4
0 5 10 15 20 25 0 5 10 15 20 25
ls17 ls18
0.5 0.5
acceleration (g)
0
0
−0.5
−1 −0.5
0 20 40 60 80 100 0 20 40 60 80 100
ls19 ls20
0.5 0.5
acceleration (g)
0
0
−0.5
−1 −0.5
0 20 40 60 80 100 0 20 40 60 80 100
time (sec) time (sec)
Figure A.11 (continued): Ground motion records: SAC Los Angeles design-
level soft (SE) soil ensemble.
208
ls21 ls22
2 2
acceleration (g) 1 1
0 0
−1 −1
−2 −2
0 20 40 60 0 20 40 60
ls23 ls24
0.5 1
acceleration (g)
0.5
0 0
−0.5
−0.5 −1
0 10 20 30 0 10 20 30
ls25 ls26
1 2
acceleration (g)
0.5 1
0 0
−0.5 −1
−1 −2
0 5 10 15 0 5 10 15
ls27 ls28
2 2
acceleration (g)
1 1
0 0
−1 −1
−2 −2
0 20 40 60 0 20 40 60
ls29 ls30
0.5 1
acceleration (g)
0.5
0 0
−0.5
−0.5 −1
0 20 40 60 0 20 40 60
time (sec) time (sec)
Figure A.12: Ground motion records: SAC Los Angeles survival-level soft
(SE) soil ensemble (generated using EERA site response analysis program,
Bardet et al., 2000).
209
ls31 ls32
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 10 20 30 0 10 20 30
ls33 ls34
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 10 20 30 0 10 20 30
ls35 ls36
2 2
acceleration (g)
1 1
0 0
−1 −1
−2 −2
0 10 20 30 0 10 20 30
ls37 ls38
1 2
acceleration (g)
0.5 1
0 0
−0.5 −1
−1 −2
0 20 40 60 0 20 40 60
ls39 ls40
1 2
acceleration (g)
0.5 1
0 0
−0.5 −1
−1 −2
0 20 40 60 0 20 40 60
time (sec) time (sec)
Figure A.12 (continued): Ground motion records: SAC Los Angeles survival-
level soft (SE) soil ensemble (generated using EERA site response analysis
program, Bardet et al., 2000).
210
nf01 nf03
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 20 40 60 0 10 20 30
nf05 nf07
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 20 40 60 0 20 40 60
nf09 nf11
0.5 1
acceleration (g)
0.5
0 0
−0.5
−0.5 −1
0 10 20 30 0 20 40 60
nf13 nf15
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 5 10 15 0 20 40 60
nf17 nf19
2 1
acceleration (g)
1 0.5
0 0
−1 −0.5
−2 −1
0 20 40 60 0 20 40 60
time (sec) time (sec)
Figure A.13: Ground motion records: SAC Los Angeles design-level near-field
(NF) ensemble.
211
nf21 nf23
1 2
acceleration (g)
0.5 1
0 0
−0.5 −1
−1 −2
0 20 40 60 0 20 40 60
nf25 nf27
2 1
acceleration (g)
1 0.5
0 0
−1 −0.5
−2 −1
0 20 40 60 0 20 40 60
nf29 nf31
2 1
acceleration (g)
1 0.5
0 0
−1 −0.5
−2 −1
0 20 40 60 0 20 40 60 80 100
nf33 nf35
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 20 40 60 80 100 0 20 40 60 80 100
nf37 nf39
1 1
acceleration (g)
0.5 0.5
0 0
−0.5 −0.5
−1 −1
0 20 40 60 80 100 0 20 40 60 80 100
time (sec) time (sec)
Figure A.13 (continued): Ground motion records: SAC Los Angeles design-
level near-field (NF) ensemble.
212
2 100
UND SC soil UND SC soil
mean ξ = 0.05 mean
ξ = 0.05
Sv (cm/s)
Sa (g)
1 50
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
Acceleration Velocity
Figure A.14: Response spectra: University of Notre Dame (UND) very dense
(SC) soil ensemble.
2 100
UND S soil ξ = 0.05 UND SD soil
D
mean mean
ξ = 0.05
Sv (cm/s)
Sa (g)
1 50
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
Acceleration Velocity
Figure A.15: Response spectra: University of Notre Dame (UND) stiff (SD)
soil ensemble.
213
2 100
UND SE soil UND SE soil
mean ξ = 0.05 mean
ξ = 0.05
Sv (cm/s)
Sa (g)
1 50
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
Acceleration Velocity
Figure A.16: Response spectra: University of Notre Dame (UND) soft (SE) soil
ensemble.
2 100
N&K N&K
mean ξ = 0.05 mean
ξ = 0.05
Sv (cm/s)
Sa (g)
1 50
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
Acceleration Velocity
Figure A.17: Response spectra: Nassar and Krawinkler 15s very dense (SC)
soil ensemble.
214
4 300
SAC Boston, SAC Boston,
S soil, design level
D
ξ = 0.05 SD soil, design level
mean mean
ξ = 0.05
Sv (cm/s)
Sa (g)
2 150
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
Acceleration Velocity
Figure A.18: Response spectra: SAC Boston design-level stiff (SD) soil ensem-
ble.
4 300
SAC Boston, SAC Boston,
S soil, survival level
D
ξ = 0.05 S soil, survival level
D
mean mean
ξ = 0.05
Sv (cm/s)
Sa (g)
2 150
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
Acceleration Velocity
Figure A.19: Response spectra: SAC Boston survival-level stiff (SD) soil
ensemble.
215
4 300
SAC Boston, SAC Boston,
S soil, design level
E
ξ = 0.05 SE soil, design level
mean mean
ξ = 0.05
Sv (cm/s)
Sa (g)
2 150
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
Acceleration Velocity
Figure A.20: Response spectra: SAC Boston design-level soft (SE) soil ensem-
ble.
4 300
SAC Boston, SAC Boston,
S soil, survival level
E
ξ = 0.05 S soil, survival level
E
mean mean
ξ = 0.05
Sv (cm/s)
Sa (g)
2 150
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
Acceleration Velocity
Figure A.21: Response spectra: SAC Boston survival-level soft (SE) soil
ensemble (generated using EERA site response analysis program, Bardet et al.,
2000).
216
4 300
SAC Los Angeles, SAC Los Angeles,
S soil, design level
D
ξ = 0.05 SD soil, design level
mean mean
ξ = 0.05
Sv (cm/s)
Sa (g)
2 150
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
Acceleration Velocity
Figure A.22: Response spectra: SAC Los Angeles design-level stiff (SD) soil
ensemble.
4 300
SAC Los Angeles, SAC Los Angeles,
S soil, survival level
D
ξ = 0.05 S soil, survival level
D
mean mean
ξ = 0.05
Sv (cm/s)
Sa (g)
2 150
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
Acceleration Velocity
Figure A.23: Response spectra: SAC Los Angeles survival-level stiff (SD) soil
ensemble.
217
4 300
SAC Los Angeles, SAC Los Angeles,
S soil, design level
E
ξ = 0.05 SE soil, design level
mean mean
ξ = 0.05
Sv (cm/s)
Sa (g)
2 150
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
Acceleration Velocity
Figure A.24: Response spectra: SAC Los Angeles design-level soft (SE) soil
ensemble.
4 300
SAC Los Angeles, SAC Los Angeles,
S soil, survival level
E
ξ = 0.05 S soil, survival level
E
mean mean
ξ = 0.05
Sv (cm/s)
Sa (g)
2 150
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
Acceleration Velocity
Figure A.25: Response spectra: SAC Los Angeles survival-level soft (SE) soil
ensemble (generated using EERA site response analysis program, Bardet et al.,
2000).
218
4 300
SAC Los Angeles, SAC Los Angeles,
S soil, design level, NF
D
ξ = 0.05 S soil, design level, NF
D
mean mean
ξ = 0.05
Sv (cm/s)
Sa (g)
2 150
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
T (sec) T (sec)
Acceleration Velocity
Figure A.26: Response spectra: SAC Los Angeles design-level stiff (SD) soil
near-field (NF) ensemble.
219
APPENDIX B
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% [cdspec.m]
% SDOF Nonlinear Dynamic Time-History Analysis Program
% This program will calculate the response of a range of
% specified SDOF oscillators. The program uses:
% 1. Preprocessor [cdspec.m]
% 2. Ground motion scaling [eqscale.m]
% 3. Smooth design spectrum [nehrpdes.m]
% 4. Nonlinear dynamic time-history analysis [lnlthist.m]
% 5. Postprocessor [cdspecpost.m]
% first created: KF 9/27/99
% last revised: KF 11/17/00
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
clear
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Define variables
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
model=input(’Which hysteresis model? (1-ep, 2-be, 3-bp, 4-sd):’);
tfree=5 % length of free vibration (s)
tend=25 % length of time history analyses (s)
nstruc=30; % number of structural periods
T=[0.1 3.0]; % range of periods (s)
aT=exp(log(T(2)/T(1))/(nstruc-1)); % multiplier for structural points
for n=1:nstruc
Tj(n)=T(1)*aT^(n-1); % structural periods (s)
end
g=386.4; % gravitational acceleration (k/in^2)
ko=1000 % const. initial stiffness (k/in)
mj=Tj.^2*ko/4/pi^2; % structural masses (k-s^2/in)
omegaj=2*pi./Tj; % structural frequencies (rad/s)
xsi=0.05 % damping ratio
cj=2*xsi*mj.*omegaj; % structural damping (k-s/in)
posty=input(’What post-yield stiffness ratio? (1-0.00, 2-0.05, 3-0.10):’);
if posty==1
alpha=0.00
elseif posty==2
alpha=0.05
elseif posty==3
alpha=0.1 % initial/post-yield stiffness ratio
end
R=[1 2:2:8] % strength reduction factors
Tj
if model==3
br=input(’Which beta_r? (1-1/6, 2-1/3, 3-1/2):’);
bs=input(’Which beta_s? (1-3, 2-beta_r):’);
if br==1
betar=1/6
elseif br==2
betar=1/3
elseif br==3
betar=1/2
end
if bs==1
betas=3
elseif bs==2
betas=betar
end
else
betar=0; % hysteretic energy factor, beep model
betas=0; % stiffness factor, beep model
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Define Earthquakes
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
220
group=input(’What earthquake group? (1-stiff, 2-medium, 3-soft, 4-near fault, 5-test group, 6-SAC/LA/SD/10in50, 7-SAC/LA/SD/2in50, 8-Nassar & Krawinkler 15s, 9-SAC/
BOS/SE(B)/10in50), 10-SAC/LA/SE(B)/10in50), 11-cyclic, 12-SAC/BO/SD/10in50, 13-SAC/BO/SD/2in50, 14-SAC/BO/SE/2in50), 15-SAC/LA/SE/2in50:’);
if group==1
gr=’stiff’
eqs=[’LPPR’ ’SASJ’ ’MIZI’ ’SFSF’ ’PACH’ ’SFCA’ ’MHGI’ ’CCVA’ ’MILU’ ’MILV’ ’SSNG’ ’SSUC’ ’SSGR’ ’WHMW’ ’LPCO’ ’LPSC’ ’LPCH’ ’LPPH’
’LPRH’ ’LPYB’];
elseif group==2
gr=’medium’
eqs=[’NONW’ ’LPHO’ ’LAYE’ ’NOSY’ ’SFOR’ ’la02’ ’KCLS’ ’SFFI’ ’SFHO’ ’SFAS’ ’WHTA’ ’WHBT’ ’IVBC’ ’IVJR’ ’IVIV’ ’CCEs’ ’WHLC’ ’LPOA’
’WHAH’ ’WHAD’];
elseif group==3
gr=’soft’
eqs=[’LPFO’ ’RUBU’ ’MISs’ ’LPTR’ ’EUF1’ ’EUF2’ ’OHW1’ ’OHW2’ ’RUB1’ ’RUB2’ ’TL1s’ ’TL2s’ ’WHA1’ ’WHA2’ ’MIDs’ ’LPOH’ ’MIFs’ ’MIOs’
’LPIA’ ’LPCB’];
elseif group==4
gr=’nearfault’
eqs=[’nf01’ ’nf03’ ’nf05’ ’nf07’ ’nf09’ ’nf11’ ’nf13’ ’nf15’ ’nf17’ ’nf19’ ’nf21’ ’nf23’ ’nf25’ ’nf27’ ’nf29’ ’nf31’ ’nf33’ ’nf35’ ’nf37’ ’nf39’];
elseif group==5
gr=’stiff’
eqs=[’LPPR’ ’PACH’];
elseif group==6
gr=’SACLASD10in50’
eqs=[’la01’ ’la02’ ’la03’ ’la04’ ’la05’ ’la06’ ’la07’ ’la08’ ’la09’ ’la10’ ’la11’ ’la12’ ’la13’ ’la14’ ’la15’ ’la16’ ’la17’ ’la18’ ’la19’ ’la20’];
elseif group==7
gr=’SACLASD2in50’
eqs=[’la21’ ’la22’ ’la23’ ’la24’ ’la25’ ’la26’ ’la27’ ’la28’ ’la29’ ’la30’ ’la31’ ’la32’ ’la33’ ’la34’ ’la35’ ’la36’ ’la37’ ’la38’ ’la39’ ’la40’];
elseif group==8
gr=’n&k15s’
eqs=[’eq03’ ’eq04’ ’eq05’ ’eq06’ ’eq07’ ’eq08’ ’eq09’ ’eq10’ ’eq11’ ’eq12’ ’eq14’ ’eq15’ ’eq16’ ’eq17’ ’eq20’];
elseif group==9
gr=’boss’
eqs=[’BS01’ ’BS02’ ’BS03’ ’BS04’ ’BS05’ ’BS06’ ’BS07’ ’BS08’ ’BS09’ ’BS10’ ’BS11’ ’BS12’ ’BS13’ ’BS14’ ’BS15’ ’BS16’ ’BS17’ ’BS18’ ’BS19’ ’BS20’];
elseif group==10
gr=’lass’
eqs=[’LS01’ ’LS02’ ’LS03’ ’LS04’ ’LS05’ ’LS06’ ’LS07’ ’LS08’ ’LS09’ ’LS10’ ’LS11’ ’LS12’ ’LS13’ ’LS14’ ’LS15’ ’LS16’ ’LS17’ ’LS18’ ’LS19’ ’LS20’];
elseif group==11
gr=’soft’
eqs=[’cy05’ ’cy20’];
elseif group==12
gr=’SACBOSD10in50’
eqs=[’bo01’ ’bo02’ ’bo03’ ’bo04’ ’bo05’ ’bo06’ ’bo07’ ’bo08’ ’bo09’ ’bo10’ ’bo11’ ’bo12’ ’bo13’ ’bo14’ ’bo15’ ’bo16’ ’bo17’ ’bo18’ ’bo19’ ’bo20’];
elseif group==13
gr=’SACBOSD2in50’
eqs=[’bo21’ ’bo22’ ’bo23’ ’bo24’ ’bo25’ ’bo26’ ’bo27’ ’bo28’ ’bo29’ ’bo30’ ’bo31’ ’bo32’ ’bo33’ ’bo34’ ’bo35’ ’bo36’ ’bo37’ ’bo38’ ’bo39’ ’bo40’];
elseif group==14
gr=’SACBOSE2in50’
eqs=[’bs21’ ’bs22’ ’bs23’ ’bs24’ ’bs25’ ’bs26’ ’bs27’ ’bs28’ ’bs29’ ’bs30’ ’bs31’ ’bs32’ ’bs33’ ’bs34’ ’bs35’ ’bs36’ ’bs37’ ’bs38’ ’bs39’ ’bs40’];
elseif group==15
gr=’SACLASE2in50’
eqs=[’ls21’ ’ls22’ ’ls23’ ’ls24’ ’ls25’ ’ls26’ ’ls27’ ’ls28’ ’ls29’ ’ls30’ ’ls31’ ’ls32’ ’ls33’ ’ls34’ ’ls35’ ’ls36’ ’ls37’ ’ls38’ ’ls39’ ’ls40’];
end
221
interval=5
end
else
filename=’Your’;
end
ns=input(’Additional constant scalar (for MDOF comparison)? ’,’s’);
if ns==’y’
newscale=input(’Enter new scale factor: ’)
EQi=EQi*newscale;
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Use which reference spectra?
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
fel=ones(neqs,1)*100000000; % initialize to infinite strength
anal=input(’(a)verage spectra, (e)lastic response or (n)ehrp spectra?’,’s’);
if anal==’n’
stype=input(’soil type? (1-SB, 2-SC, 3-SD, 4-SE, 5-NF):’);
perf=input(’performance level? (1-design, 2-survival):’);
zone=input(’seismic region? (1-LA, 2-SEA, 3-BOS):’);
[felnehrp]=nehrpdes(mj,Tj,neqs,perf,stype,zone);
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% nehrp spectra AND scaling?
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
if anal==’n’ % nehrp reference spectra?
if method==1 % PGA scaling?
PGAnhp=max(mean(felnehrp)./mj)/2.5/g;% nehrp pga (g)
PGAeqi=max(max(abs(EQi)));% gm pga (g)
scalar=PGAnhp/PGAeqi % new scale factor
EQi=EQi*scalar; % scale gm’s again
elseif method==5
Sanhp=mean(felnehrp)./mj/g;
for j=1:nstruc
scalar(j)=Sanhp(j)./Saavg(j,j);
EQi(:,:,j)=EQi(:,:,j)*scalar(j);
end
end
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Determine Elastic Force Demand/Inelastic Displacement Demand
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
diary([date ’sdofecho.txt’])
time=clock;
time=time(:,4:end);
[filename ’ run has started! Current time is ’ num2str(time)]
diary off
222
dpmaxij(:,j,:)=vpmi;
nyinelij(:,j,:)=nyi;
nrinelij(:,j,:)=nrevi;
nzinelij(:,j,:)=nzi;
end
if (saveflag==’y’ & rem(j,interval)==0)
!klog -password sodesne
save(filename)
end
waitbar2(j/nstruc)
end
felavg=ones(neqs,1)*mean(felasij);
close(hh)
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Post-process data
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
diary([date ’sdofecho.txt’])
time=clock;
time=time(:,4:end);
[filename ’ run is completed! Current time is ’ num2str(time)]
diary off
[mu,mubar,disp,muc,mur,mup,muec]=cdspecpost(delasij,dinelij,felasij,finelij,R,Tj,omegaj,mj,alpha,neqs,nstruc,gr,vijelas,vijinel,fijelas,fijinel,dpinelij,nyinelij,nrinelij,nzelasij,nzin
elij,tfree,ti,dpmaxij,tend);
if saveflag==’y’
save(filename)
end
figure(1)
subplot(2,1,1)
title([date ’ Filename: ’ filename ])
223
B.2 EQSCALE.M: Ground Motion Scaling Function
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% [eqscale.m]
% Ground motion scaling program
% This program scales earthquakes according to specified
% method.
% first created: KF 9/27/99
% last revised: KF 9/27/00
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [EQi,ti,Saavg] = eqscale(gr,eqs,neqs,method,Tj,R,alpha,tfree,tend)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Define time new time vector
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
dtnew=0.01; % specify same time step (s)
if (method==2 | method==7)
dtref=dtnew; % specify time step to revert to for MIV (s)
dtnew=0.0025; % specify smaller step for MIV (s)
trefn=0:dtref:tend; % reference time vector (s)
end
tnewn=0:dtnew:tend; % new time vector (s)
ti=tnewn;
Ni=length(tnewn); % number of points in ti
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Load earthquake data
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
h = waitbar(0,’Loading and modifying earthquake data...’);
for n = 1:neqs
eqn=[eqs(4*n-3:n+3*n) ’prep’];
if method==6
eqspecn=[eqs(4*n-3:n+3*n) ’specl’];
else
eqspecn=[eqs(4*n-3:n+3*n) ’spec’];
end
path1=[’../groundmotions/final/’ gr ’/’ eqn];
path2=[’../groundmotions/final/’ gr ’/’ eqspecn];
load(path1,’-ascii’)
load(path2,’-ascii’)
eval([’tn=’ eqn ’(:,1);’]) % time vector (s)
eval([’an=’ eqn ’(:,2);’]) % acceleration vector (g)
eval([’Tn=’ eqspecn ’(:,1);’]) % period vector (s)
eval([’San=’ eqspecn ’(:,5);’]) % pseudo acceleration spectra (g)
eval([’Svn=’ eqspecn ’(:,2);’]) % pseudo velocity spectra (in/s)
dt=diff(tn); % time step (s)
dt=dt(1);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Modify earthquakes
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
t0=tn(1); % beginning time (s)
tmaxn=max(tn); % check if record is long enough
if tmaxn<tend % if not, make it long enough
tn1=t0:dt:tend;
nn=length(tn);
n1=length(tn1);
an1=zeros(n1,1);
nmodif=nn+1:n1;
ndiff=length(nmodif);
an1(1:nn)=an;
an1(nmodif)=zeros(ndiff,1);
clear an tn
an=an1; % new acceleration vector (g)
tn=tn1; % new time vector (s)
end
tshiftn=tn-t0; % shift time to relative scale (s)
anewn=spline(tshiftn,an,tnewn); % new acceleration vector (s)
Ist=find(tnewn==tend-tfree); % start of free vibration
anewn(Ist+1:end)=zeros(length(anewn)-Ist,1); % zero accel
Snewn=spline(Tn,San,[Tj]); % new acceleration spectrum vector (g)
Tnsec=[Tj , 3:.1:7];
Snsec=spline(Tn,San,Tnsec); % new acceleration spectrum vector (g)
Svnewn=spline(Tn,Svn,[Tj]); % new velocity spectrum vector (in/s)
EQi(n,:)=anewn;
Si(n,:)=Snewn;
Svi(n,:)=Svnewn;
Siorig(n,:)=Snsec;
waitbar(n/neqs)
end
close(h)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Scale to PGA
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
if method==1 | method==8
amax=max(abs(EQi),[],2) % find max accel’s
aavg=mean(amax) % find mean of max accel’s
if method==1
label=’PGA’;
sctype=input(’Scale to (a)verage or (s)pecified PGA? ’,’s’);
if sctype==’a’
224
sfactor=aavg./amax% determine scale factor
else
aavg=input(’specify PGA (in g): ’);
sfactor=aavg./amax
end
else
label=’none’;
sfactor=ones(neqs,1)
end
for n=1:neqs % scale records
EQi(n,:)=EQi(n,:)*sfactor(n);
Si(n,:)=Si(n,:)*sfactor(n);
end
Saavg=mean(Si); % find response spectra avg.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Comparison Plot
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
figure(1)
subplot(2,1,1)
plot(ti,EQi)
hold on
plot([0 tend],[aavg aavg],’r--’)
plot([0 tend],-[aavg aavg],’r--’)
xlabel(’t (sec)’)
ylabel(’acceleration (g)’)
title([date ’ Groundmotions: ’ gr ’ Scaling: ’ label])
hold off
subplot(2,1,2)
plot([Tj],Si)
hold on
plot([Tj],Saavg,’--’,’LineWidth’,5)
plot([0 Tj(end)],[aavg aavg],’--’)
xlabel(’T (sec)’)
ylabel(’S_a (g)’)
title(’Response Spectra, \xi=0.05’)
hold off
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Scale to MIV
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
elseif method==2
scalar=981; % change units from g->cm/s^2
acel=EQi*scalar;
N=length(acel);
h = waitbar(0,’Scaling using MIV method...’);
IV=zeros(neqs,Ni); % initialize
tIV=zeros(neqs,Ni);
for i = 1:neqs
counter=0;
for n = 2:N
if (acel(i,n)<=0 & acel(i,n-1)>=0) | (acel(i,n)>=0 & acel(i,n-1)<=0)
incross=n;
counter=counter+1;
for m = incross+1:N
if acel(i,incross)==acel(i,incross-1)
break
end
if (acel(i,m)<=0 & acel(i,m-1)>=0) | (acel(i,m)>=0 & acel(i,m-1)<=0)
outcross=m-1;
break
end
end
IV(i,counter)=sum(acel(i,incross:outcross))*dtnew;
tIV(i,counter)=ti(incross);
end
end
[maxIV(i),I(i)]=max(abs(IV(i,:)));
maxt(i)=tIV(i,I(i));
waitbar(i/neqs)
end
close(h)
maxIV’
maxt
IVavg=mean(maxIV)
sfactor=IVavg./maxIV’ % determine scale factor
for n=1:neqs % scale records
EQin(n,:)=spline(ti,EQi(n,:),trefn)*sfactor(n);
Si(n,:)=Si(n,:)*sfactor(n);
end
ti=trefn;
clear EQi
EQi=EQin;
Saavg=mean(Si); % find response spectra avg.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Comparison Plot
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
figure(1)
subplot(2,1,1)
plot(ti,EQi)
xlabel(’t (sec)’)
ylabel(’acceleration (g)’)
title([date ’ Groundmotions: ’ gr ’ Scaling: MIV’])
225
subplot(2,1,2)
plot([Tj],Si)
hold on
plot([Tj],Saavg,’--’,’LineWidth’,5)
xlabel(’T (sec)’)
ylabel(’S_a (g)’)
title(’Response Spectra, \xi=0.05’)
hold off
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Scale to EPA
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
elseif method==3
Iepa=find((Tj>=.1)&(Tj<=.5)); % define spectral range
EPA=mean(Si(:,Iepa),2)/2.5
EPAavg=mean(EPA)
sfactor=EPAavg./EPA % determine scale factor
for n=1:neqs % scale records
EQi(n,:)=EQi(n,:)*sfactor(n);
Si(n,:)=Si(n,:)*sfactor(n);
end
Saavg=mean(Si); % find response spectra avg.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Comparison Plot
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
figure(1)
subplot(2,1,1)
plot(ti,EQi)
hold on
plot([0 tend],[EPAavg EPAavg],’r--’)
plot([0 tend],-[EPAavg EPAavg],’r--’)
xlabel(’t (sec)’)
ylabel(’acceleration (g)’)
title([date ’ Groundmotions: ’ gr ’ Scaling: EPA’])
hold off
subplot(2,1,2)
plot([Tj],Si)
hold on
plot([Tj],Saavg,’--’,’LineWidth’,5)
plot([0 Tj(end)],[EPAavg EPAavg],’--’)
xlabel(’T (sec)’)
ylabel(’S_a (g)’)
title(’Response Spectra, \xi=0.05’)
hold off
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Scale to EPV
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
elseif method==4
Iepv=find((Tj>=.8)&(Tj<=1.2)); % define spectral range
EPV=mean(Svi(:,Iepv),2)/2.5
EPVavg=mean(EPV)
sfactor=EPVavg./EPV % determine scale factor
for n=1:neqs % scale records
EQi(n,:)=EQi(n,:)*sfactor(n);
Si(n,:)=Si(n,:)*sfactor(n);
end
Saavg=mean(Si); % find response spectra avg.
Svavg=mean(Svi);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Comparison Plot
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
figure(1)
subplot(2,1,1)
plot(ti,EQi)
xlabel(’t (sec)’)
ylabel(’acceleration (g)’)
title([date ’ Groundmotions: ’ gr ’ Scaling: EPV’])
subplot(2,1,2)
plot([Tj],Svi)
hold on
plot([Tj],Svavg,’--’,’LineWidth’,5)
plot([0 Tj(end)],[EPVavg EPVavg],’--’)
xlabel(’T (sec)’)
ylabel(’S_v (g)’)
title(’Response Spectra, \xi=0.05’)
hold off
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Scale to Sa1
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
elseif method==5
nstruc=length(Tj); % number of structural periods
h = waitbar(0,’Scaling using Sa1 method...’);
for j=1:nstruc
intensity=Si(:,j);
intavg=mean(intensity);
sfactor=intavg./intensity; % determine scale factor
for n=1:neqs % scale records
EQij(n,:,j)=EQi(n,:)*sfactor(n);
Sij(n,:,j)=Si(n,:)*sfactor(n);
end
Saavg(j,:)=mean(Sij(:,:,j)); % find response spectra avg.
waitbar(j/nstruc)
end
226
clear Si EQi
EQi=EQij;
Si=Sij;
close(h)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Comparison Plot
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
nsub=5; % number of subplots
nfigs=nstruc/nsub; % number of figures
if nfigs<1 % if there are less than 5 periods
nfigs=1;
nsub=nstruc;
end
for j=1:nfigs
figure(j)
for k=1:nsub
subplot(nsub,1,k)
strucno=j*k + (j-1)*(nsub-k);
plot(Tj,Saavg(strucno,:),’--’,’LineWidth’,5)
hold on
plot(Tj,Si(:,:,strucno))
hold off
end
end
figure
plot([Tj],Saavg’)
xlabel(’T (sec)’)
ylabel(’S_a (g)’)
title(’Response Spectra, \xi=0.05’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Scale to SaTsec and plot comparison
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
elseif method==6
methodman=input(’(k)ennedy or (y)our method? ’,’s’);
key=2; % preparing to call nkregression
interval=3; % specify plot interval
nstruc=length(Tj); % number of structural periods
h = waitbar(0,’Scaling using SaTsec method...’);
for k=1:length(R)
Rk=R(k); % select R factor
for j=1:nstruc
nsamp=j;% select period
if gr([1:2])==’so’
[muk]=nkregression_s(nsamp,alpha,Tj,key,Rk);
elseif gr([1:2])==’ne’
gr
[muk]=nkregression_nf(nsamp,alpha,Tj,key,Rk);
elseif gr([1:2])==’me’
gr
[muk]=nkregression_me(nsamp,alpha,Tj,key,Rk);
elseif gr([1:2])==’st’
gr
[muk]=nkregression_st(nsamp,alpha,Tj,key,Rk);
else
[muk]=nkregression(nsamp,alpha,Tj,key,Rk);
% predicted ductility demand
end
if methodman==’y’
Tsec=Tj(j)*sqrt(muk/(muk*alpha+(1-alpha)));
else
Tsec=Tj(j)*sqrt(muk);
end
% predicted secant period
jrange=find(Tnsec>=Tj(j) & Tnsec<=Tsec);% det. range of T
intensity=Si(:,j);% intensity at T
intrange=Siorig(:,jrange);% intensity over range
intrngavg2=mean(intrange);% avg. over range
intrngavg=mean(intrange,2);% avg. over range
intavg=mean(intrngavg);% avg. intensity
Saavg(k,j)=intavg;
sfactor=intavg./intrngavg;% determine scale factor
for n=1:neqs% scale records
EQijk(n,j,k,:)=EQi(n,:)*sfactor(n);
Sijk(n,j,k,:)=Si(n,:)*sfactor(n);
Sijkorig(n,j,k,:)=Siorig(n,:)*sfactor(n);
end
if rem(j,interval)==0
figure(j/interval)
subplot(length(R),1,k)
sa(:,:)=Sijkorig(:,j,k,:);
plot(Tnsec,sa)
hold on
plot(Tnsec(jrange),intrngavg2,’--o’,’Linewidth’,3)
hold off
end
end
waitbar(k/length(R))
end
clear Si EQi
EQi=EQijk;
Si=Sijk;
close(h)
227
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Scale to A95 (Sarma and Yang 1987)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
elseif method==7
scalar=981; % change units from g->cm/s^2
acel=EQi*scalar;
acel2=acel.^2; % square acceleration record (g^2)
Es=sum(acel2,2)*dtnew; % arias intensity (g^2-sec)
A95=0.764*Es.^.438 % A95 per Sarma and Yang Regression (g)
A95avg=mean(A95); % average A95 (g)
sfactor=A95avg./A95 % determine scale factor
for n=1:neqs % scale records
EQin(n,:)=spline(ti,EQi(n,:),trefn)*sfactor(n);
Si(n,:)=Si(n,:)*sfactor(n);
end
ti=trefn;
clear EQi
EQi=EQin;
Saavg=mean(Si); % find response spectra avg.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Comparison Plot
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
figure(1)
subplot(2,1,1)
plot(ti,EQi)
xlabel(’t (sec)’)
ylabel(’acceleration (g)’)
title([date ’ Groundmotions: ’ gr ’ Scaling: A95’])
subplot(2,1,2)
plot([Tj],Si)
hold on
plot([Tj],Saavg,’--’,’LineWidth’,5)
xlabel(’T (sec)’)
ylabel(’S_a (g)’)
title(’Response Spectra, \xi=0.05’)
hold off
end
228
B.3 NEHRPDES.M: Smooth Design Reference Spectrum Function
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% [nehrpdes.m]
% Smooth Design Reference Spectrum
% This program will calculate the design force for each
% SDOF oscillator using smooth design spectra.
% first created: KF 11/10/99
% last revised: KF 12/21/99
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [fel]=nehrpdes(mj,Tj,neqs,perf,stype,zone)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Load Spectra
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
if perf==1
load(’../groundmotions/final/NEHRPspectra/nehrpdes’)
elseif perf==2
load(’../groundmotions/final/NEHRPspectra/nehrpsur’)
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Choose spectra and calculate design forces
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
g=386.4; % gravity (in/sec^2)
sa(1,:)=Sa(stype,zone,:); % spectrum
sa=spline(Torig,sa,Tj); % resample
fel=ones(neqs,1)*(sa.*mj*g) ; % design force
figure(1)
subplot(2,1,2)
hold on
plot(Tj,sa,’r’,’LineWidth’,3)
bar(Tj,sa,’r’)
hold off
229
B.4 LNLTHIST.M, LNLTHISTSD.M: Nonlinear Dynamic Time-History Analysis Functions
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% [lnlthist.m]
% Linear Acceleration Program
% This program calculates the response of a SDOF oscillator using the
% linear acceleration method.
% first created: KF 9/27/99
% last revised: KF 9/26/00
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [vmaxi,fmaxi,amaxi,telapsei,vrec,frec,crit,fl,vpli,nyi,nrevi,nzi,vpmi]=lnlthist(EQi,neqs,ti,ko,m,co,om,R,alpha,fel,flag,method,model,betar,betas)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Initialize
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
scalar=386.4; % change g->in/sec^2
niter=6; % number of maximum iterations
alphao=alpha; % original post-yield stiffness
if flag==1 % determine if elastic or inelastic
rstart=1;
rend=1;
tol=0.01 % error tolerance
elseif flag==2
rstart=2;
rend=length(R);
tol=0.01 % error tolerance
end
if method==6
EQorig=EQi;
clear EQi
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% reduction factor loop
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
for kk=rstart:rend
Ro=R(kk) % select reduction factor
if method==6
EQi(:,:)=EQorig(:,kk,:);
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% ground motion loop
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
for e=1:neqs
fy=fel(e)/Ro; % determine yield strength (k)
vy=fy/ko; % determine yield displacement (in)
vy1=0;
vy2=0;
P=EQi(e,:); % select earthquake (g)
timeo=ti; % time (sec)
ho=diff(timeo); % time increment,dt (s)
ho=ho(1);
No=length(timeo); % no. of points
po=-m*P*scalar; % force vector (k)
vmax=0; % initialize
fmax=0;
amax=0;
telapse=0;
tstart=zeros(size(clock));
tend=zeros(size(clock));
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% iteration loop
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
for l=2:niter % iteration loop
M=2^(l-2)% interpolation
h=ho/M;% modified dt (s)
p=interp(po,M);% modified load vector
time=interp(timeo,M);% modified time vector
N=No*M;% modified no of points
v=0;% init disp
a=0;% init acel
vdi=0;% init vel
f=0;% init spring force
k=ko;% initialize stiffness
c=co;% iintialize damping
fu=fy;% set yield surface
flag=0;% init reversal code
fl=0;% init reversal code
crit=0;% init yield criteria
vd=0;% init vel
vpl=0;% init plastic defo
vpm=[0 0];% init max plastic defo
nrev=0;% init total yield reversals
ny=0;% init yield events
nzero=0;% init zero crossings
r=0;% init yield reversal code
rev=0;% init yield reversal code
if model==3
alpha2=0.00001;% post-yield stiffness, ep
if (betar==betas | Ro==1)
k1o=ko/(1+betas);% elastic stiffness, be
alpha=alphao*(1+betas);% modified post-yield stiffness
230
else
k1o=ko/(1+betar);% elastic stiffness, be
alpha=alphao*(1+betar);% modified post-yield stiffness
end
k2o=betas*k1o;% elastic stiffness, ep
k1=k1o;% initialize stiffness, be
k2=k2o;% initialize stiffness, ep
k=k1+k2;% total stiffness, beep
flag1=0;% init reversal code, be
flag2=0;% init reversal code, ep
f1=0;
f2=0;
r1=0;
r2=0;
rev1=0;
rev2=0;
vy1=fy/(1+betar)/k1;% yield displacement, be
vy2=fy/(1+betar)/k2*betar;% yield displacement, ep
end
ai=(p(1)-k*v(1)-c*vdi)/m;% acceleration
khat=6*m/h^2 + 3*c/h +k;% pseudo stiffness
tstart=clock;% start time
hh = waitbar1(0,[’Calculating response: EQ ’ num2str(e) ’, Iteration ’ num2str(l-1)]);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% linear acceleration loop
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
for i=2:N
dp=p(i)-p(i-1);% change in loading force
khat=6*m/h^2 + 3*c/h +k;% pseudo stiffness
dphat=dp+(6*m/h + 3*c)*vdi+(3*m + c*h/2)*ai;% change in pseudo force
dv=dphat/khat;% change in displacement
dvd=ai*h + 3*(dv-vdi*h-ai*h^2/2)/h;% change in velocity
da=(dv-vdi*h-ai*h^2/2)*6/h^2;% change in acceleration
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Update variables
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
a(i)=ai;
ai=ai+da;% acceleration
v(i)=v(i-1)+dv;% displacement
vdi=vdi+dvd;% velocity
vd(i)=vdi;
df=k*dv;% change in spring force
if i==2
fu=-sign(v(i))*vy;
fu2=-sign(v(i))*vy2;
flag=0;
vd(1)=sign(v(i))*.000000001;
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Check yielding and reversals
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
if model==1% bilinear elasto-plastic
[df,fu,k,flag,dpl,y,r,rev]=epdisp(flag,fu,vd(i),vd(i-1),ko,alpha,vy,v(i),dv,r,rev);
elseif model==2% bilinear elastic
[df,k,flag,dpl,y,r,rev]=bedisp2(flag,vd(i),vd(i-1),ko,alpha,vy,v(i),dv,r,rev);
elseif model==3% bilinear elasto-plastic
[df1,k1,flag1,dpl,y,r1,rev1]=bedisp2(flag1,vd(i),vd(i-1),k1o,alpha,vy1,v(i),dv,r1,rev1);
[df2,fu2,k2,flag2,dpl,y,r2,rev2]=epdisp(flag2,fu2,vd(i),vd(i-1),k2o,alpha2,vy2,v(i),dv,r2,rev2);
f1(i)=f1(i-1)+df1;
f2(i)=f2(i-1)+df2;
df=df1+df2;
k=k1+k2;
fu=fu2;
flag=flag2;
rev=rev2;
rev2=0;
end
vpl=vpl+dpl;% accumulative plastic defo
ny=ny+y;% number of yield events
vpm(ny+1)=vpm(ny+1)+dpl;% plastic defo per excursion
vpm(ny+2)=[0];
nrev=nrev+rev;% number of total yield reversals
rev=0;% reset counter
f(i)=f(i-1)+df;% spring force
if (sign(f(i))~=sign(f(i-1)))
nzero=nzero+1;% zero crossing
end
fl(i)=flag;
crit(i)=fu;
if (fl(i)>fl(i-1) & model~=3)
ai=(p(i)-f(i)-c*vdi)/m;% correct acceleration
end
if rem(i,100)==0
waitbar1(i/N)
end
end
close(hh)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Check error
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
tend=clock;% end time
telapse=etime(tend,tstart);% elapsed time
231
vmax(l)=max(abs(v));% maximum displacement (absolute)
fmax=max(abs(f));% maximum force
vpmax=max(abs(vpm));% maximum plastic excursion
at=a-p/m;% total acceleration
amax=max(abs(at));% maximum acceleration
if Ro~=0
err=(vmax(l)-vmax(l-1))/vmax(l)% error
else
vcheck(:,l)=spline(time’,v’,ti’);
err=max(abs((vcheck(:,l)-vcheck(:,l-1)))./vmax(l))% error
end
if abs(err)<tol% check if error is within tolerance
break
end
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Store final values
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
if l>=niter
diary([date ’sdofecho.txt’])
’CHECK CONVERGENCE’
[’error= ’ num2str(err) ’ iter= ’ num2str(l) ’ delta= ’ num2str(vmax(l)) ’ eq= ’ num2str(e)]
diary off
end
telapsei(e,1,kk)=telapse;
vmaxi(e,1,kk)=vmax(l);
fmaxi(e,1,kk)=fmax;
amaxi(e,1,kk)=amax;
vpli(e,1,kk)=vpl;
vpmi(e,1,kk)=vpmax;
nyi(e,1,kk)=ny;
nrevi(e,1,kk)=nrev;
nzi(e,1,kk)=nzero;
vrec(e,1,kk,:)=spline(time,v,ti);
frec(e,1,kk,:)=spline(time,f,ti);
end
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% [lnlthistsd.m]
% Linear Acceleration Prog
% This program calculates the response of a SDOF oscillator using the
% linear acceleration method.
% first created: KF 9/27/99
% last revised: KF 9/26/00
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [vmaxi,fmaxi,amaxi,telapsei,vrec,frec,crit,fl,vpli,nyi,nrevi,nzi,vpmi,vri]=lnlthistsd(EQi,neqs,ti,ko,m,co,om,R,alpha,fel,flagg,method,model,betar,betas)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Initialize
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
scalar=386.4; % change g->in/sec^2
tol=0.01 % error tolerance
niter=6; % number of maximum iterations
alphao=alpha; % original post-yield stiffness
if flagg==1 % determine if elastic or inelastic
rstart=1;
rend=1;
elseif flagg==2
rstart=2;
rend=length(R);
end
if method==6
EQorig=EQi;
clear EQi
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% reduction factor loop
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
for kk=rstart:rend
Ro=R(kk) % select reduction factor
if method==6
EQi(:,:)=EQorig(:,kk,:);
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% ground motion loop
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
for e=1:neqs
fy=fel(e)/Ro; % determine yield strength (k)
vy=fy/ko; % determine yield displacement (in)
vy1=0;
vy2=0;
P=EQi(e,:); % select earthquake (g)
timeo=ti; % time (sec)
ho=diff(timeo); % time increment,dt (s)
ho=ho(1);
No=length(timeo); % no. of points
po=-m*P*scalar; % force vector (k)
vmax=0; % initialize
fmax=0;
amax=0;
telapse=0;
232
tstart=zeros(size(clock));
tend=zeros(size(clock));
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% iteration loop
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
for l=2:niter % iteration loop
M=2^(l-2)% interpolation
h=ho/M;% modified dt (s)
p=interp(po,M);% modified load vector
time=interp(timeo,M);% modified time vector
N=No*M;% modified no of points
v=0;% init disp
a=0;% init acel
vdi=0;% init vel
f=0;% init spring force
k=ko;% initialize stiffness
c=co;% iintialize damping
fu=fy;% set yield surface
fsh=[-fy fy];% init shoot-through force
vsh=[-vy vy];% init shoot-through displ
flag=0;% init reversal code
fl=0;% init reversal code
crit=0;% init yield criteria
vd=0;% init vel
vpl=0;% init plastic defo
vr=0;% init residual disp
vpm=[0 0];% init max plastic defo
nrev=0;% init total yield reversals
ny=0;% init yield events
nzero=0;% init zero crossings
r=0;% init yield reversal code
rev=0;% init yield reversal code
if model==3
alpha2=0.00001;% post-yield stiffness, ep
if (betar==betas | Ro==1)
k1o=ko/(1+betas);% elastic stiffness, be
alpha=alphao*(1+betas);% modified post-yield stiffness
else
k1o=ko/(1+betar);% elastic stiffness, be
alpha=alphao*(1+betar);% modified post-yield stiffness
end
k2o=betas*k1o;% elastic stiffness, ep
k1=k1o;% initialize stiffness, be
k2=k2o;% initialize stiffness, ep
k=k1+k2;% total stiffness, beep
flag1=0;% init reversal code, be
flag2=0;% init reversal code, ep
f1=0;
f2=0;
r1=0;
r2=0;
rev1=0;
rev2=0;
vy1=fy/(1+betar)/k1;% yield displacement, be
vy2=fy/(1+betar)/k2*betar;% yield displacement, ep
end
ai=(p(1)-k*v(1)-c*vdi)/m;% acceleration
khat=6*m/h^2 + 3*c/h +k;% pseudo stiffness
tstart=clock;% start time
hh = waitbar1(0,[’Calculating response: EQ ’ num2str(e) ’, Iteration ’ num2str(l-1)]);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% linear acceleration loop
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
for i=2:N
dp=p(i)-p(i-1);% change in loading force
khat=6*m/h^2 + 3*c/h +k;% pseudo stiffness
dphat=dp+(6*m/h + 3*c)*vdi+(3*m + c*h/2)*ai;% change in pseudo force
dv=dphat/khat;% change in displacement
dvd=ai*h + 3*(dv-vdi*h-ai*h^2/2)/h;% change in velocity
da=(dv-vdi*h-ai*h^2/2)*6/h^2;% change in acceleration
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Update variables
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
a(i)=ai;
ai=ai+da;% acceleration
v(i)=v(i-1)+dv;% displacement
vdi=vdi+dvd;% velocity
vd(i)=vdi;
df=k*dv;% change in spring force
f(i)=f(i-1)+df;% spring force
if i==2
fu=-sign(v(i))*vy;
fu2=-sign(v(i))*vy2;
fsh=sign(v(i))*fsh;
vsh=sign(v(i))*vsh;
flag=0;
vd(1)=sign(v(i))*.000000001;
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Check yielding and reversals
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
if model==1% stiffness-degrading
233
[df,fu,k,flag,dpl,y,r,rev,fsh,vsh]=sddisppl(flag,fu,vd(i),vd(i-1),ko,alpha,vy,v(i),dv,r,rev,f(i),f(i-
1),fsh,vsh,k);
elseif model==2% bilinear elastic
[df,k,flag,dpl,y,r,rev]=bedisp2(flag,vd(i),vd(i-1),ko,alpha,vy,v(i),dv,r,rev);
elseif model==3% bilinear elasto-plastic
[df1,k1,flag1,dpl,y,r1,rev1]=bedisp2(flag1,vd(i),vd(i-1),k1o,alpha,vy1,v(i),dv,r1,rev1);
[df2,fu2,k2,flag2,dpl,y,r2,rev2]=epdisp(flag2,fu2,vd(i),vd(i-1),k2o,alpha2,vy2,v(i),dv,r2,rev2);
f1(i)=f1(i-1)+df1;
f2(i)=f2(i-1)+df2;
df=df1+df2;
k=k1+k2;
fu=fu2;
flag=flag2;
rev=rev2;
rev2=0;
end
vpl=vpl+abs(dpl);% accumulative plastic defo
vr=vr+dpl;% residual disp
ny=ny+y;% number of yield events
vpm(ny+1)=vpm(ny+1)+abs(dpl);% plastic defo per excursion
vpm(ny+2)=[0];
nrev=nrev+rev;% number of total yield reversals
rev=0;% reset counter
f(i)=f(i-1)+df;% spring force
if (sign(f(i))~=sign(f(i-1)))
nzero=nzero+1;% zero crossing
end
fl(i)=flag;
flag;
crit(i)=fu;
if (fl(i)>fl(i-1) & model~=3)
ai=(p(i)-f(i)-c*vdi)/m;% correct acceleration
end
if rem(i,100)==0
waitbar1(i/N)
end
end
close(hh)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Check error
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
tend=clock;% end time
telapse=etime(tend,tstart);% elapsed time
vmax(l)=max(abs(v));% maximum displacement (absolute)
fmax=max(abs(f));% maximum force
vpmax=max(abs(vpm));% maximum plastic excursion
at=a-p/m;% total acceleration
amax=max(abs(at));% maximum acceleration
if Ro~=0
err=(vmax(l)-vmax(l-1))/vmax(l)% error
else
vcheck(:,l)=spline(time’,v’,ti’);
err=max(abs((vcheck(:,l)-vcheck(:,l-1)))./vmax(l))% error
end
if abs(err)<tol% check if error is within tolerance
break
end
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Store final values
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
if l>=niter
diary([date ’sdofecho.txt’])
’CHECK CONVERGENCE’
[’error= ’ num2str(err) ’ iter= ’ num2str(l) ’ delta= ’ num2str(vmax(l)) ’ eq= ’ num2str(e)]
diary off
end
telapsei(e,1,kk)=telapse;
vmaxi(e,1,kk)=vmax(l);
fmaxi(e,1,kk)=fmax;
amaxi(e,1,kk)=amax;
vpli(e,1,kk)=vpl;
vri(e,1,kk)=vr;
vpmi(e,1,kk)=vpmax;
nyi(e,1,kk)=ny;
nrevi(e,1,kk)=nrev;
nzi(e,1,kk)=nzero;
vrec(e,1,kk,:)=spline(time,v,ti);
frec(e,1,kk,:)=spline(time,f,ti);
% end
end
234
B.5 CDSPECPOST.M: Post-Processing Function
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% [cdspecpost.m]
% SDOF Post-processing Program
% This program plots results of the SDOF analyses.
% method.
% first created: KF 10/25/99
% last revised: KF 9/29/00
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [mu,mubar,mud,muc,mur,mup,muec] =
cdspecpost(delasij,dinelij,felasij,finelij,R,Tj,omegaj,mj,alpha,neqs,nstruc,gr,vijelas,vijinel,fijelas,fijinel,dpinelij,nyinelij,nrinelij,nzelasij,nzinelij,tfree,ti,dpmaxij,tend)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Plotting options
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
supress=1; % supress plots (1=y, 0=n)
range1=[0 3.5 0 14]; % specify plotting ranges
range2=[0 3.5 0 1.6];
range3=[0 3.5 0 1.6];
range4=[0 3.5 0 4];
range5=[0 3.5 0 100];
range6=[0 3.5 0 1.2];
range7=[0 3.5 0 50];
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Determine R-L-T relationship
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
mur2=zeros(neqs,nstruc,length(R)); % initialize matrix
intog=386.4; % conversion factor
dinelij(:,:,1)=delasij; % set elastic displacement
finelij(:,:,1)=felasij; % set elastic force
nzinelij(:,:,1)=nzelasij; % set elastic zero crossings
umax=max(vijinel,[],4); % maximum displacements (in)
umin=min(vijinel,[],4); % minimum displacements (in)
fmax=max(fijinel,[],4); % maximum force (k)
fmin=min(fijinel,[],4); % minimum force (k)
umax(:,:,1)=max(vijelas,[],3); % include max elastic displacements (in)
umin(:,:,1)=min(vijelas,[],3); % include min elastic displacements (in)
fmax(:,:,1)=max(fijelas,[],3); % include max elastic force (in)
fmin(:,:,1)=min(fijelas,[],3); % include min elastic force (in)
Ist=find(ti==tend-tfree); % find point where free vibration starts
ures=abs(mean(vijinel(:,:,:,Ist+1:end),4)); % residual displacement (in)
for k=1:length(R)
dy(:,:,k)=delasij./R(k); % yield displacements
fy(:,:,k)=felasij./R(k); % yield force
Rbar(:,k)=ones(nstruc,1)*R(k);
for j=1:nstruc
mu(:,j,k)=dinelij(:,j,k)./dy(:,j,k);% displacement ductility ratio
ainelij(:,j,k)=finelij(:,j,k)./mj(j)/intog;% accel spectra
muc(:,j,k)=(-umin(:,j,k)+umax(:,j,k))./dy(:,j,k) - 1;% cyclic displacement ductility
mur(:,j,k)=ures(:,j,k)./dy(:,j,k);% residual displacement ductility
mup(:,j,k)=dpinelij(:,j,k)./dy(:,j,k) + 1;% cumulative plastic deformation ductility
muec(:,j,k)=dpmaxij(:,j,k).*fy(:,j,k)*2./(.5.*umax(:,j,k).*fmax(:,j,k)+.5.*umin(:,j,k).*fmin(:,j,k));% max cyclic energy
I1=find(alpha.*(mu(:,j,k)-1)<1);% residual disp condition 1
I2=find(alpha.*(mu(:,j,k)-1)>=1);% residual disp condition 2
mur2(I1,j,k)=abs(ures(I1,j,k)./((mu(I1,j,k)-1).*(1-alpha).*dy(I1,j,k)));
mur2(I2,j,k)=abs(ures(I2,j,k)./((1-alpha).*dy(I2,j,k)./alpha));
end
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Calculate mean and dispersion measures
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
mubar(:,:)=mean(mu); % mean displacement ductility
mucbar(:,:)=mean(muc); % mean cyclic ductility
murbar(:,:)=mean(mur); % mean residual displacement ductility
mur2bar(:,:)=mean(mur2); % mean residual displacement ductility
mupbar(:,:)=mean(mup); % mean cumulative plastic deformation ductility
muecbar(:,:)=mean(muec); % mean maximum cyclic energy ductility
deltabar(:,:)=mean(dinelij); % mean displacement
acelbar(:,:)=mean(ainelij); % mean acceleration
nybar(:,:)=mean(nyinelij); % mean no. of yield events
nrbar(:,:)=mean(nrinelij); % mean no. of full yield reversals
nzbar(:,:)=mean(nzinelij); % mean no. of zero crossings
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
musig(:,:)=std(mu); % std. dev. of displacement ductility
mucsig(:,:)=std(muc); % std. dev. of cyclic ductility
mursig(:,:)=std(mur); % std. dev. of residual displacement ductility
mur2sig(:,:)=std(mur2); % std. dev. of residual displacement ductility
mupsig(:,:)=std(mup); % std. dev. of cumulative plastic deformation ductility
muecsig(:,:)=std(muec); % std. dev. of max cyclic energy ductility
deltasig(:,:)=std(dinelij); % std. dev. of displacement
acelsig(:,:)=std(ainelij); % std. dev. of acceleration
nysig(:,:)=std(nyinelij); % std. dev. of no. of yield events
nrsig(:,:)=std(nrinelij); % std. dev. of no. of full yield reversals
nzsig(:,:)=std(nzinelij); % std. dev. of no. of zero crossings
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
for k=1:length(R)
235
mucov(:,k)=musig(:,k)./mubar(:,k); % COV of displacement ductility
muccov(:,k)=mucsig(:,k)./mucbar(:,k); % COV of cyclic ductility
murcov(:,k)=mursig(:,k)./murbar(:,k); % COV of residual displacement ductility
mur2cov(:,k)=mur2sig(:,k)./mur2bar(:,k); % COV of residual displacement ductility
mupcov(:,k)=mupsig(:,k)./mupbar(:,k); % COV of cumulative plastic deformation ductility
mueccov(:,k)=muecsig(:,k)./muecbar(:,k); % COV of max cyclic energy ductility
deltacov(:,k)=deltasig(:,k)./deltabar(:,k); % COV of displacement
acelcov(:,k)=acelsig(:,k)./acelbar(:,k); % COV of acceleration
nycov(:,k)=nysig(:,k)./nybar(:,k); % COV of no. of yield events
nrcov(:,k)=nrsig(:,k)./nrbar(:,k); % COV of no. of full yield reversals
nzcov(:,k)=nzsig(:,k)./nzbar(:,k); % COV of no. of zero crossings
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
mud(:,:)=std(log(mu)); % dispersion of displacement ductility
mucd(:,:)=std(log(muc)); % dispersion of cyclic ductility
murd(:,:)=std(log(mur)); % dispersion of residual displacement ductility
mur2d(:,:)=std(log(mur2)); % dispersion of residual displacement ductility
mupd(:,:)=std(log(mup)); % dispersion of cumulative plastic deformation ductility
muecd(:,:)=std(log(muec)); % dispersion of max cyclic energy ductility
deltad(:,:)=std(log(dinelij)); % dispersion of displacement
aceld(:,:)=std(log(ainelij)); % dispersion of acceleration
nyd(:,:)=std(log(nyinelij)); % dispersion of no. of yield events
nrd(:,:)=std(log(nrinelij)); % dispersion of no. of full yield reversals
nzd(:,:)=std(log(nzinelij)); % dispersion of no. of zero crossings
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Nassar & Krawinkler regression comparison
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
nsamp1=1; % where to sample 1st period
nsamp2=4; % where to sample 2nd period
munk=0:.1:10;
key=1; % prepare to retrieve reduction factors
T1=Tj(nsamp1);
T2=Tj(nsamp2);
[Rnk1]=nkregression(nsamp1,alpha,Tj,key,R); % 1st R factor
[Rnk2]=nkregression(nsamp2,alpha,Tj,key,R); % 2nd R factor
Rbar1=Rbar(nsamp1,:); % 1st R factor, anal
Rbar2=Rbar(nsamp2,:); % 2nd R factor, anal
mubar1=mubar(nsamp1,:); % 1st mu factor, anal
mubar2=mubar(nsamp2,:); % 2nd mu factor, anal
musig1=musig(nsamp1,:); % 1st mu factor, anal
musig2=musig(nsamp2,:); % 2nd mu factor, anal
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Plot R-L-T relationships
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
if supress==0
lastfig=gcf;
for i=1:neqs
figure(lastfig+i)
for k=1:length(R)
plot(Tj,dinelij(i,:,k),’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’S_d (in) ’)
title([date ’ Displacement Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
hold off
end
236
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
lastfig=gcf;
figure(lastfig+1)
subplot(3,1,1)
for k=1:length(R)
plot(Tj,mubar(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’\mu’)
title([date ’ Displacement Ductility Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range1)
hold(’off’)
subplot(3,1,2)
for k=1:length(R)
plot(Tj,mud(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’†elta’)
title([date ’ Dispersion Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
subplot(3,1,3)
for k=1:length(R)
plot(Tj,mucov(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’Coefficient of Variation’)
title([date ’ COV Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
lastfig=gcf;
figure(lastfig+1)
subplot(3,1,1)
for k=1:length(R)
plot(Tj,mucbar(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’\mu_c’)
title([date ’ Cyclic Displacement Ductility Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range1)
hold(’off’)
subplot(3,1,2)
for k=1:length(R)
plot(Tj,mucd(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’†elta_c’)
title([date ’ Dispersion Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
subplot(3,1,3)
for k=1:length(R)
plot(Tj,muccov(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’COV_c’)
title([date ’ Coefficient of Variation Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
lastfig=gcf;
figure(lastfig+1)
subplot(3,1,1)
for k=1:length(R)
plot(Tj,murbar(:,k),’b’,’LineWidth’,k)
hold on
end
237
grid on
xlabel(’T (sec)’)
ylabel(’\mu_r’)
title([date ’ Residual Displacement Ductility Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range4)
hold(’off’)
subplot(3,1,2)
for k=1:length(R)
plot(Tj,murd(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’†elta_r’)
title([date ’ Dispersion Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
subplot(3,1,3)
for k=1:length(R)
plot(Tj,murcov(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’COV_r’)
title([date ’ Coefficient of Variation Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
lastfig=gcf;
figure(lastfig+1)
subplot(3,1,1)
for k=1:length(R)
plot(Tj,mur2bar(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’\mu_r_2’)
title([date ’ Residual Displacement Ductility Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range6)
hold(’off’)
subplot(3,1,2)
for k=1:length(R)
plot(Tj,mur2d(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’†elta_r_2’)
title([date ’ Dispersion Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
subplot(3,1,3)
for k=1:length(R)
plot(Tj,mur2cov(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’COV_r_2’)
title([date ’ Coefficient of Variation Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
lastfig=gcf;
figure(lastfig+1)
subplot(3,1,1)
for k=1:length(R)
plot(Tj,mupbar(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’\mu_p’)
title([date ’ Cumulative Plastic Deformation Ductility Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range5)
hold(’off’)
subplot(3,1,2)
for k=1:length(R)
238
plot(Tj,mupd(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’†elta_p’)
title([date ’ Dispersion Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
subplot(3,1,3)
for k=1:length(R)
plot(Tj,mupcov(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’COV_p’)
title([date ’ Coefficient of Variation Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
lastfig=gcf;
figure(lastfig+1)
subplot(3,1,1)
for k=1:length(R)
plot(Tj,muecbar(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’\mu_e_c’)
title([date ’ Maximum Cyclic Energy Ductility Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range4)
hold(’off’)
subplot(3,1,2)
for k=1:length(R)
plot(Tj,muecd(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’†elta_e_c’)
title([date ’ Dispersion Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
subplot(3,1,3)
for k=1:length(R)
plot(Tj,mueccov(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’COV_e_c’)
title([date ’ Coefficient of Variation Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
lastfig=gcf;
figure(lastfig+1)
subplot(3,1,1)
for k=1:length(R)
plot(Tj,deltabar(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’‡elta (in)’)
title([date ’ Displacement Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range1)
hold(’off’)
subplot(3,1,2)
for k=1:length(R)
plot(Tj,deltad(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’†elta_‡elta (in)’)
title([date ’ Dispersion Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
239
subplot(3,1,3)
for k=1:length(R)
plot(Tj,deltacov(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’COV_‡elta’)
title([date ’ Coefficient of Variation Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
lastfig=gcf;
figure(lastfig+1)
subplot(3,1,1)
for k=1:length(R)
plot(Tj,acelbar(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’a (g)’)
title([date ’ Acceleration Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range6)
hold(’off’)
subplot(3,1,2)
for k=1:length(R)
plot(Tj,aceld(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’†elta_a (g)’)
title([date ’ Dispersion Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
subplot(3,1,3)
for k=1:length(R)
plot(Tj,acelcov(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’COV_a’)
title([date ’ Coefficient of Variation Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
lastfig=gcf;
figure(lastfig+1)
subplot(3,1,1)
for k=1:length(R)
plot(Tj,nybar(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’n_y’)
title([date ’ Number of Inelastic Excursions Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range7)
hold(’off’)
subplot(3,1,2)
for k=1:length(R)
plot(Tj,nyd(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’†elta_n’)
title([date ’ Dispersion Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
subplot(3,1,3)
for k=1:length(R)
plot(Tj,nycov(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’COV_n’)
title([date ’ Coefficient of Variation Spectra for ’ gr])
240
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
lastfig=gcf;
figure(lastfig+1)
subplot(3,1,1)
for k=1:length(R)
plot(Tj,nrbar(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’n_r’)
title([date ’ Number of Yield Reversals Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range7)
hold(’off’)
subplot(3,1,2)
for k=1:length(R)
plot(Tj,nrd(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’†elta_n’)
title([date ’ Dispersion Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
subplot(3,1,3)
for k=1:length(R)
plot(Tj,nrcov(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’COV_n’)
title([date ’ Coefficient of Variation Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
lastfig=gcf;
figure(lastfig+1)
subplot(3,1,1)
for k=1:length(R)
plot(Tj,nzbar(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’n_z’)
title([date ’ Number of Zero Crossings Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range5)
hold(’off’)
subplot(3,1,2)
for k=1:length(R)
plot(Tj,nzd(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’†elta_n’)
title([date ’ Dispersion Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
subplot(3,1,3)
for k=1:length(R)
plot(Tj,nzcov(:,k),’b’,’LineWidth’,k)
hold on
end
grid on
xlabel(’T (sec)’)
ylabel(’COV_n’)
title([date ’ Coefficient of Variation Spectra for ’ gr])
legend(’R=1’,’R=2’,’R=4’,’R=6’,’R=8’)
axis(range2)
hold(’off’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Plot Against Nassar & Krawinkler Regression
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
lastfig=gcf;
figure(lastfig+1)
subplot(2,1,1)
plot(mubar1,Rbar1)
241
hold on
plot(mubar1+musig1,Rbar1,’--’)
plot(mubar1-musig1,Rbar1,’--’)
plot(munk,Rnk1,’r-.’,’Linewidth’,2.5)
grid on
xlabel(’\mu’)
ylabel(’R’)
title([date ’ Comparison with N&K for T=’ num2str(T1) ’ sec; \alpha=’ num2str(alpha)])
legend(’\mu_a_v_g’,’\mu+\sigma’,’\mu-\sigma’,’N&K’)
axis([0 10 0 8])
hold off
subplot(2,1,2)
plot(mubar2,Rbar2)
hold on
plot(mubar2+musig2,Rbar2,’--’)
plot(mubar2-musig2,Rbar2,’--’)
plot(munk,Rnk2,’r-.’,’Linewidth’,2.5)
grid on
xlabel(’\mu’)
ylabel(’R’)
title([date ’ Comparison with N&K for T=’ num2str(T2) ’ sec; \alpha=’ num2str(alpha)])
legend(’\mu_a_v_g’,’\mu+\sigma’,’\mu-\sigma’,’N&K’)
axis([0 10 0 8])
hold off
242
B.6 SDREG.M: R-µ-T Nonlinear Regression Program
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% [sdofreg.m]
% SDOF Results Plotting Program
% This program plots results of the SDOF analyses.
%
% first created: KF 01/03/01
% last revised: KF 02/09/01
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Initialize
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
path(path,’/afs/nd.edu/user30/kfarrow1/m-files’)
keepold=input(’Would you like to keep the old normalizing data?’,’s’);
if keepold==’n’
clear
cfile=input(’Which comparison data file?’,’s’);
nfile=input(’Which normalizing data file?’,’s’);
’LOADING NORMALIZING DATA’
load(nfile)
filenameo=filename;
gro=gr;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Determine R-L-T relationship
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
clear dy fy mu mur mup
’CALCULATING DEMAND INDICES’
ures=zeros(neqs,nstruc,length(R));% initialize matrix
intog=386.4; % conversion factor
dinelij(:,:,1)=delasij; % set elastic displacement
finelij(:,:,1)=felasij; % set elastic force
Ist=find(ti==tend-tfree); % find point where free vibration starts
if model~=2
ures=abs(mean(vijinel(:,:,:,Ist+1:end),4));% residual displacement (in)
end
if anal==’a’ % effective R using smooth spectra
Reff=felasij./felavg;
elseif anal==’n’
Reff=felasij./felnehrp;
else
Reff=ones(size(felasij));
end
Inot=find(Reff<1);
elas=ones(size(Inot));
Reff(Inot)=elas;
hh = waitbar2(0,[’Calculating demand indicies...’]);
for k=1:length(R)
dy(:,:,k)=delasij./Reff./R(k);% yield displacements
fy(:,:,k)=felasij./Reff./R(k);% yield force
for j=1:nstruc
mu(:,j,k)=dinelij(:,j,k)./dy(:,j,k);% displacement ductility ratio
ainelij(:,j,k)=finelij(:,j,k)./mj(j)/intog;% accel spectra
mur(:,j,k)=ures(:,j,k)./dy(:,j,k);% residual displacement ductility
mup(:,j,k)=dpinelij(:,j,k)./dy(:,j,k) + 1;% cumulative plastic deformation ductility
end
waitbar2(k/length(R))
end
close(hh)
mu(:,:,1)=ones(neqs,nstruc);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Calculate mean and dispersion measures
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
mubaro(:,:)=mean(mu); % mean displacement ductility
murbaro(:,:)=mean(mur); % mean residual displacement ductility
mupbaro(:,:)=mean(mup); % mean cumulative plastic deformation ductility
deltabaro(:,:)=mean(dinelij); % mean displacement
acelbaro(:,:)=mean(ainelij); % mean acceleration
nybaro(:,:)=mean(nyinelij); % mean no. of yield events
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
mudo(:,:)=std(log(mu)); % dispersion of displacement ductility
murdo(:,:)=std(log(mur)); % dispersion of residual displacement ductility
mupdo(:,:)=std(log(mup)); % dispersion of cumulative plastic deformation ductility
deltado(:,:)=std(log(dinelij)); % dispersion of displacement
aceldo(:,:)=std(log(ainelij)); % dispersion of acceleration
nydo(:,:)=std(log(nyinelij)); % dispersion of no. of yield events
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
musigo(:,:)=std(mu); % std. dev. of displacement ductility
mursigo(:,:)=std(mur); % std. dev. of residual displacement ductility
mupsigo(:,:)=std(mup); % std. dev. of cumulative plastic deformation ductility
deltasigo(:,:)=std(dinelij); % std. dev. of displacement
acelsigo(:,:)=std(ainelij); % std. dev. of acceleration
nysigo(:,:)=std(nyinelij); % std. dev. of no. of yield events
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
for k=1:length(R)
mucovo(:,k)=musigo(:,k)./mubaro(:,k); % COV of displacement ductility
murcovo(:,k)=mursigo(:,k)./murbaro(:,k); % COV of residual displacement ductility
mupcovo(:,k)=mupsigo(:,k)./mupbaro(:,k); % COV of cumulative plastic deformation ductility
deltacovo(:,k)=deltasigo(:,k)./deltabaro(:,k); % COV of displacement
243
acelcovo(:,k)=acelsigo(:,k)./acelbaro(:,k); % COV of acceleration
nycovo(:,k)=nysigo(:,k)./nybaro(:,k); % COV of no. of yield events
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
muo=mu;
muro=mur;
mupo=mup;
nyo=nyinelij;
else
cfile=input(’Which comparison data file?’,’s’);
save plottemp *o cfile
clear
load plottemp
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Load comparison data
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
’LOADING COMPARISON DATA’
load(cfile)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Determine R-L-T relationship
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
clear dy fy mu mur mup
’CALCULATING DEMAND INDICES’
ures=zeros(neqs,nstruc,length(R));% initialize matrix
intog=386.4; % conversion factor
dinelij(:,:,1)=delasij; % set elastic displacement
finelij(:,:,1)=felasij; % set elastic force
Ist=find(ti==tend-tfree); % find point where free vibration starts
if model~=2
ures=abs(mean(vijinel(:,:,:,Ist+1:end),4));% residual displacement (in)
end
if anal==’a’ % effective R using smooth spectra
Reff=felasij./felavg;
elseif anal==’n’
Reff=felasij./felnehrp;
else
Reff=ones(size(felasij));
end
Inot=find(Reff<1);
elas=ones(size(Inot));
Reff(Inot)=elas;
hh = waitbar2(0,[’Calculating demand indicies...’]);
for k=1:length(R)
dy(:,:,k)=delasij./Reff./R(k);% yield displacements
fy(:,:,k)=felasij./Reff./R(k);% yield force
for j=1:nstruc
mu(:,j,k)=dinelij(:,j,k)./dy(:,j,k);% displacement ductility ratio
ainelij(:,j,k)=finelij(:,j,k)./mj(j)/intog;% accel spectra
mur(:,j,k)=ures(:,j,k)./dy(:,j,k);% residual displacement ductility
mup(:,j,k)=dpinelij(:,j,k)./dy(:,j,k) + 1;% cumulative plastic deformation ductility
gammamui(:,j,k)=mu(:,j,k)./muo(:,j,k);
gammamuri(:,j,k)=mur(:,j,k)./muro(:,j,k);
gammamupi(:,j,k)=mup(:,j,k)./mupo(:,j,k);
gammanyi(:,j,k)=nyinelij(:,j,k)./nyo(:,j,k);
end
waitbar2(k/length(R))
end
close(hh)
mu(:,:,1)=ones(neqs,nstruc);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Calculate mean and dispersion measures
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
mubar(:,:)=mean(mu); % mean displacement ductility
murbar(:,:)=mean(mur); % mean residual displacement ductility
mupbar(:,:)=mean(mup); % mean cumulative plastic deformation ductility
deltabar(:,:)=mean(dinelij); % mean displacement
acelbar(:,:)=mean(ainelij); % mean acceleration
nybar(:,:)=mean(nyinelij); % mean no. of yield events
gammamu(:,:)=mean(gammamui);
gammamur(:,:)=mean(gammamuri);
gammamup(:,:)=mean(gammamupi);
gammany(:,:)=mean(gammanyi);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
mud(:,:)=std(log(mu)); % dispersion of displacement ductility
murd(:,:)=std(log(mur)); % dispersion of residual displacement ductility
mupd(:,:)=std(log(mup)); % dispersion of cumulative plastic deformation ductility
deltad(:,:)=std(log(dinelij)); % dispersion of displacement
aceld(:,:)=std(log(ainelij)); % dispersion of acceleration
nyd(:,:)=std(log(nyinelij)); % dispersion of no. of yield events
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
musig(:,:)=std(mu); % std. dev. of displacement ductility
mursig(:,:)=std(mur); % std. dev. of residual displacement ductility
mupsig(:,:)=std(mup); % std. dev. of cumulative plastic deformation ductility
deltasig(:,:)=std(dinelij); % std. dev. of displacement
acelsig(:,:)=std(ainelij); % std. dev. of acceleration
nysig(:,:)=std(nyinelij); % std. dev. of no. of yield events
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
for k=1:length(R)
mucov(:,k)=musig(:,k)./mubar(:,k); % COV of displacement ductility
murcov(:,k)=mursig(:,k)./murbar(:,k); % COV of residual displacement ductility
mupcov(:,k)=mupsig(:,k)./mupbar(:,k); % COV of cumulative plastic deformation ductility
244
deltacov(:,k)=deltasig(:,k)./deltabar(:,k); % COV of displacement
acelcov(:,k)=acelsig(:,k)./acelbar(:,k); % COV of acceleration
nycov(:,k)=nysig(:,k)./nybar(:,k); % COV of no. of yield events
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Calculate normalized spectra
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
if (size(gro)==size(gr) & sum(gro==gr)==length(gr))
gratio=’individual’
else
gammamu=mubar./mubaro;
gammamur=murbar./murbaro;
gammamup=mupbar./mupbaro;
gammany=nybar./nybaro;
gratio=’mean’
end
gammadel=deltacov./deltacovo;
gammaacc=acelcov./acelcovo;
gammamud=mucov./mucovo;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Plot R-L-T relationships
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
rangemu=[0 3.5 0 14]; % specify ranges
rangemup=[0 3.5 0 100];
rangemur=[0 3.5 0 4];
rangeny=[0 3.5 0 50];
rangedd=[0 2 0 2];
rangega=[0 3.5 0 4];
rangecov=[0 2 0 1.5];
close all
filename
filenameo
rename=input(’change these names? ’,’s’)
if rename==’y’
filename=input(’enter comparison filename: ’,’s’)
filenameo=input(’enter normalization filename: ’,’s’)
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
figure(1)
subplot(2,1,1)
plot(0.1,0.1,’w:’)
hold on
for k=1:length(R)
plot(Tj,mubaro(:,k),’-’,’LineWidth’,k)
hold on
plot(Tj,mubar(:,k),’r--’,’LineWidth’,k)
end
grid on
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\mu’,’FontName’,’Times’)
title([filenameo ’ (solid) vs. ’ filename ’ (dashed)’],’FontName’,’Times’)
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
ightarrow thick lines)’)
axis(rangemu)
set(gca,’FontName’,’times’)
hold off
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
subplot(2,1,2)
plot(0.1,0.1,’w:’)
hold on
for k=1:length(R)
plot(Tj,mupbaro(:,k),’-’,’LineWidth’,k)
hold on
plot(Tj,mupbar(:,k),’r--’,’LineWidth’,k)
end
grid on
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\it\mu_p
m’,’FontName’,’Times’)
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
ightarrow thick lines)’)
axis(rangemup)
set(gca,’FontName’,’times’)
hold off
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
figure(2)
subplot(2,1,1)
plot(0.1,0.1,’w:’)
hold on
for k=1:length(R)
plot(Tj,murbaro(:,k),’-’,’LineWidth’,k)
hold on
plot(Tj,murbar(:,k),’r--’,’LineWidth’,k)
end
grid on
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\it\mu_r
245
m’,’FontName’,’Times’)
title([filenameo ’ (solid) vs. ’ filename ’ (dashed)’],’FontName’,’Times’)
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
ightarrow thick lines)’)
axis(rangemur)
set(gca,’FontName’,’times’)
hold off
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
subplot(2,1,2)
plot(0.1,0.1,’w:’)
hold on
for k=1:length(R)
plot(Tj,nybaro(:,k),’-’,’LineWidth’,k)
hold on
plot(Tj,nybar(:,k),’r--’,’LineWidth’,k)
end
grid on
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\itn_y
m’,’FontName’,’Times’)
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
ightarrow thick lines)’)
axis(rangeny)
set(gca,’FontName’,’times’)
hold off
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Plot R-g-T relationships
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
baseline=ones(size(Tj));
lastfig=gcf;
figure(3)
subplot(2,1,1)
plot(0.1,0.1,’w:’)
hold on
plot(Tj,baseline)
hold on
for k=2:length(R)
plot(Tj,gammamu(:,k),’r-’,’LineWidth’,k)
hold on
end
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\it„amma_\mu
m’,’FontName’,’Times’)
title([’„amma = ’ filename ’/’ filenameo],’FontName’,’Times’)
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
ightarrow thick lines)’)
axis(rangega)
set(gca,’FontName’,’times’)
hold(’off’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
subplot(2,1,2)
plot(0.1,0.1,’w:’)
hold on
plot(Tj,baseline)
hold on
for k=2:length(R)
plot(Tj,gammamup(:,k),’r-’,’LineWidth’,k)
hold on
end
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\it„amma_\mu_p
m’,’FontName’,’Times’)
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
ightarrow thick lines)’)
axis(rangega)
set(gca,’FontName’,’times’)
hold(’off’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
figure(4)
subplot(2,1,1)
plot(0.1,0.1,’w:’)
hold on
plot(Tj,baseline)
hold on
for k=2:length(R)
plot(Tj,gammamur(:,k),’r-’,’LineWidth’,k)
hold on
end
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\it„amma_\mu_r
m’,’FontName’,’Times’)
title([’„amma = ’ filename ’/’ filenameo],’FontName’,’Times’)
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
246
ightarrow thick lines)’)
axis(rangega)
set(gca,’FontName’,’times’)
hold(’off’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
subplot(2,1,2)
plot(0.1,0.1,’w:’)
hold on
plot(Tj,baseline)
hold on
for k=2:length(R)
plot(Tj,gammany(:,k),’r-’,’LineWidth’,k)
hold on
end
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\it„amma_n_y
m’,’FontName’,’Times’)
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
ightarrow thick lines)’)
axis(rangega)
set(gca,’FontName’,’times’)
hold(’off’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Calculate Regression Constants Using N&K Model
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
figure(5)
subplot(2,1,1)
plot(0.1,0.1,’w:’)
hold on
betao=[1]; % initial guess for c constant
Rj=ones(nstruc,1)*R; % R matrix
tstart=15; % index of control periods
tend=30;
for j=tstart:tend
X=Rj(j,:);
Y=mubar(j,:);
cfit(j)=nlinfit(X,Y,’muhat’,betao);% calculate c const.
end
betao=[0.8 0.29]; % initial guess for a & b constants
X=Tj(tstart:tend);
Y=cfit(tstart:tend);
betahat=nlinfit(X,Y,’chat’,betao);% calculate reg consts.
for k=1:length(R)
X=[Tj’ Rj(:,k)]; % regression indep vars
mufit(:,k)=farrow(betahat,X);
plot(Tj,mubar(:,k),’-’,’Linewidth’,k)
hold on
plot(Tj,mufit(:,k),’r--’,’Linewidth’,k)
end
axis(rangemu)
grid on
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\it\mu
m’,’FontName’,’Times’)
title([ ’Regression for ’ filename ’: a = ’ num2str(betahat(1)) ’, b = ’ num2str(betahat(2))],’FontName’,’Times’)
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
ightarrow thick lines)’)
set(gca,’FontName’,’times’)
hold off
betahati=betahat
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
subplot(2,1,2)
plot(0.1,0.1,’w:’)
hold on
betao=[1]; % initial guess for c constant
for j=tstart:tend
X=Rj(j,:);
Y=mubaro(j,:);
cfit(j)=nlinfit(X,Y,’muhat’,betao);% calculate c const.
end
betao=[0.8 0.29]; % initial guess for a & b constants
X=Tj(tstart:tend);
Y=cfit(tstart:tend);
betahat=nlinfit(X,Y,’chat’,betao);% calculate reg consts.
for k=1:length(R)
X=[Tj’ Rj(:,k)]; % regression indep vars
mufit(:,k)=farrow(betahat,X);
plot(Tj,mubaro(:,k),’-’,’Linewidth’,k)
hold on
plot(Tj,mufit(:,k),’r--’,’Linewidth’,k)
end
axis(rangemu)
grid on
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\it\mu
m’,’FontName’,’Times’)
title([ ’Regression for ’ filenameo ’: a = ’ num2str(betahat(1)) ’, b = ’ num2str(betahat(2))],’FontName’,’Times’)
247
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
ightarrow thick lines)’)
set(gca,’FontName’,’times’)
hold off
betahato=betahat;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Compare Scatter
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
figure(6)
subplot(2,1,1)
plot(0.1,0.1,’w:’)
hold on
for k=1:length(R)
plot(Tj,acelsigo(:,k),’-’,’LineWidth’,k)
hold on
plot(Tj,acelsig(:,k),’r--’,’LineWidth’,k)
end
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\it\sigma(S_d)
m’,’FontName’,’Times’)
title([filenameo ’ (solid) vs. ’ filename ’ (dashed)’],’FontName’,’Times’)
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
ightarrow thick lines)’)
%axis(rangega)
set(gca,’FontName’,’times’)
hold off
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
subplot(2,1,2)
plot(0.1,0.1,’w:’)
hold on
for k=1:length(R)
plot(Tj,musigo(:,k),’-’,’LineWidth’,k)
hold on
plot(Tj,musig(:,k),’r--’,’LineWidth’,k)
end
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\it\sigma(\mu)
m’,’FontName’,’Times’)
title([’„amma = ’ filename ’/’ filenameo],’FontName’,’Times’)
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
ightarrow thick lines)’)
axis(rangemu)
set(gca,’FontName’,’times’)
hold off
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
figure(7)
subplot(2,1,1)
plot(0.1,0.1,’w:’)
hold on
for k=1:length(R)
plot(Tj,acelcovo(:,k),’-’,’LineWidth’,k)
hold on
plot(Tj,acelcov(:,k),’r--’,’LineWidth’,k)
end
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\itCOV(S_d)
m’,’FontName’,’Times’)
title([filenameo ’ (solid) vs. ’ filename ’ (dashed)’],’FontName’,’Times’)
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
ightarrow thick lines)’)
axis(rangecov)
set(gca,’FontName’,’times’)
hold off
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
subplot(2,1,2)
plot(0.1,0.1,’w:’)
hold on
plot(Tj,baseline)
hold on
for k=1:length(R)
plot(Tj,gammaacc(:,k),’r-’,’LineWidth’,k)
hold on
end
grid on
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\it„amma_C_O_V_(_S_d_)
m’,’FontName’,’Times’)
title([’„amma = ’ filename ’/’ filenameo],’FontName’,’Times’)
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
ightarrow thick lines)’)
axis(rangedd)
set(gca,’FontName’,’times’)
hold(’off’)
248
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
figure(8)
subplot(2,1,1)
plot(0.1,0.1,’w:’)
hold on
for k=1:length(R)
plot(Tj,mucovo(:,k),’-’,’LineWidth’,k)
hold on
plot(Tj,mucov(:,k),’r--’,’LineWidth’,k)
end
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\itCOV(\mu)
m’,’FontName’,’Times’)
title([filenameo ’ (solid) vs. ’ filename ’ (dashed)’],’FontName’,’Times’)
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
ightarrow thick lines)’)
axis(rangecov)
set(gca,’FontName’,’times’)
hold off
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
subplot(2,1,2)
plot(0.1,0.1,’w:’)
hold on
plot(Tj,baseline)
hold on
for k=2:length(R)
plot(Tj,gammamud(:,k),’r-’,’LineWidth’,k)
hold on
end
grid on
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\it„amma_C_O_V_(_\mu_)
m’,’FontName’,’Times’)
title([’„amma = ’ filename ’/’ filenameo],’FontName’,’Times’)
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
ightarrow thick lines)’)
axis(rangedd)
set(gca,’FontName’,’times’)
hold(’off’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Demand Index Cross-correlations
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
N=0; % set counter
deg=2; % degree of polynomial fit
nx=30; % number of discrete points
mumax=10; % maximum controllable mu
labels=[’\mu \mu_r\mu_pn_y ’];% matrix titles
Tmax=24; % maximum period for residual disp
[Imax]=find(mubar<mumax);
submubar=mubar(1:Tmax,:);
submurbar=murbar(1:Tmax,:);
submupbar=mupbar(1:Tmax,:);
subnybar=nybar(1:Tmax,:);
[Jmax]=find(submubar<mumax);
lambda(:,:,1)=mubar; % set up demand index matrix
lambda(:,:,2)=murbar;
lambda(:,:,3)=mupbar;
lambda(:,:,4)=nybar;
sublambda(:,:,1)=submubar; % set up demand index matrix
sublambda(:,:,2)=submurbar;
sublambda(:,:,3)=submupbar;
sublambda(:,:,4)=subnybar;
figure(9)
for i=1:4
for j=1:4
N=N+1;
subplot(4,4,N)
lambda1=lambda(:,:,i);
lambda2=lambda(:,:,j);
sublambda1=sublambda(:,:,i);
sublambda2=sublambda(:,:,j);
if i==j
axis([0 2 0 2])
axis off
text(1,1,[’\fontname{times}\it’ labels(5*i-4:5*i)])
else
if (i==2 | j==2)
lambda1=sublambda1(Jmax);
lambda2=sublambda2(Jmax);
else
lambda1=lambda1(Imax);
lambda2=lambda2(Imax);
end
plot(lambda2,lambda1,’.’)
hold on
P=polyfit(lambda2,lambda1,deg);% find coeff’s
mi=min(lambda2);
ma=max(lambda2);
249
dx=(ma-mi)/nx;
xfit=[mi:dx:ma];
yfit=polyval(P,xfit);
coeffs(i,j,:)=P;
plot(xfit,yfit,’r’)
set(gca,’FontName’,’times’)
hold off
end
end
end
suptitle([’\fontname{times}Cross-correlations, ’ filename ’: data points and regression lines’])
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
[Imax]=find(mubaro<mumax);
submubar=mubaro(1:Tmax,:);
submurbar=murbaro(1:Tmax,:);
submupbar=mupbaro(1:Tmax,:);
subnybar=nybaro(1:Tmax,:);
[Jmax]=find(submubar<mumax);
lambda(:,:,1)=mubaro; % set up demand index matrix
lambda(:,:,2)=murbaro;
lambda(:,:,3)=mupbaro;
lambda(:,:,4)=nybaro;
sublambda(:,:,1)=submubar; % set up demand index matrix
sublambda(:,:,2)=submurbar;
sublambda(:,:,3)=submupbar;
sublambda(:,:,4)=subnybar;
N=0; % set counter
figure(10)
for i=1:4
for j=1:4
N=N+1;
subplot(4,4,N)
lambda1=lambda(:,:,i);
lambda2=lambda(:,:,j);
sublambda1=sublambda(:,:,i);
sublambda2=sublambda(:,:,j);
if i==j
axis([0 2 0 2])
axis off
text(1,1,[’\fontname{times}\it’ labels(5*i-4:5*i)])
else
if (i==2 | j==2)
lambda1=sublambda1(Jmax);
lambda2=sublambda2(Jmax);
else
lambda1=lambda1(Imax);
lambda2=lambda2(Imax);
end
plot(lambda2,lambda1,’.’)
hold on
P=polyfit(lambda2,lambda1,deg);% find coeff’s
mi=min(lambda2);
ma=max(lambda2);
dx=(ma-mi)/nx;
xfit=[mi:dx:ma];
yfit=polyval(P,xfit);
coeffso(i,j,:)=P;
plot(xfit,yfit,’r’)
set(gca,’FontName’,’times’)
hold off
end
end
end
suptitle([’\fontname{times}Cross-correlations, ’ filenameo ’: data points and regression lines’])
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Compare Scatter, part II
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
figure(11)
subplot(2,1,1)
plot(0.1,0.1,’w:’)
hold on
for k=1:length(R)
plot(Tj,deltacovo(:,k),’-’,’LineWidth’,k)
hold on
plot(Tj,deltacov(:,k),’r--’,’LineWidth’,k)
end
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\itCOV(S_d)
m’,’FontName’,’Times’)
title([filenameo ’ (solid) vs. ’ filename ’ (dashed)’],’FontName’,’Times’)
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
ightarrow thick lines)’)
axis(rangecov)
set(gca,’FontName’,’times’)
hold off
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
subplot(2,1,2)
plot(0.1,0.1,’w:’)
hold on
plot(Tj,baseline)
250
hold on
for k=1:length(R)
plot(Tj,gammadel(:,k),’r-’,’LineWidth’,k)
hold on
end
grid on
xlabel(’\itT_o
m (sec)’,’FontName’,’Times’)
ylabel(’\it„amma_C_O_V_(_S_d_)
m’,’FontName’,’Times’)
title([’„amma = ’ filename ’/’ filenameo],’FontName’,’Times’)
legend(’\fontname{times}\itR
m = 1, 2, 4, 6, 8 (thin
ightarrow thick lines)’)
axis(rangedd)
set(gca,’FontName’,’times’)
hold(’off’)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Save Regression Constants?
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
compile=input(’save these regression constants? ’,’s’)
if compile==’y’
save([’./constants/’ filename],’betahati’,’coeffs’)
save([’./constants/’ filenameo],’betahato’,’coeffso’)
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Save Figures?
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
hardcopy=input(’save these plots? ’,’s’)
if hardcopy==’y’
check=ls([’./figures/’ filename ’*’])
if (check(1:2)~=’No’)
addon=input([filename ’ exists. Please enter modifier text or press CTRL-C:’],’s’)
filename=[filename addon]
end
for f=1:11
figure(f)
orient tall
print(’-depsc’,[’./figures/’ filename num2str(f)])
end
end
251
B.7 DEMANDREG.M: Cross-Correlation Nonlinear Regression Program
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% [demandreg.m]
% SDOF Regression Program
% This program performs nonlinear regression analyses for the demand indices.
%
% first created: KF 06/19/01
% last revised: KF 06/19/01
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Initialize
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
path(path,’/afs/nd.edu/user30/kfarrow1/m-files’)
init=input(’Load new data?’,’s’)
if init==’y’
clear
nfile=input(’Which data file?’,’s’);
’LOADING DATA’
load(nfile)
filenameo=filename;
gro=gr;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Determine R-L-T relationship
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
clear dy fy mu mur mup
’CALCULATING DEMAND INDICES’
ures=zeros(neqs,nstruc,length(R)); % initialize matrix
intog=386.4; % conversion factor
dinelij(:,:,1)=delasij; % set elastic displacement
finelij(:,:,1)=felasij; % set elastic force
Ist=find(ti==tend-tfree); % find point where free vibration starts
if model~=2
ures=abs(mean(vijinel(:,:,:,Ist+1:end),4)); % residual displacement (in)
end
if anal==’a’ % effective R using smooth spectra
Reff=felasij./felavg;
elseif anal==’n’
Reff=felasij./felnehrp;
else
Reff=ones(size(felasij));
end
Inot=find(Reff<1);
elas=ones(size(Inot));
Reff(Inot)=elas;
hh = waitbar2(0,[’Calculating demand indicies...’]);
for k=1:length(R)
dy(:,:,k)=delasij./Reff./R(k); % yield displacements
fy(:,:,k)=felasij./Reff./R(k); % yield force
for j=1:nstruc
mu(:,j,k)=dinelij(:,j,k)./dy(:,j,k);% displacement ductility ratio
ainelij(:,j,k)=finelij(:,j,k)./mj(j)/intog;% accel spectra
mur(:,j,k)=ures(:,j,k)./dy(:,j,k);% residual displacement ductility
mup(:,j,k)=dpinelij(:,j,k)./dy(:,j,k) + 1;% cumulative plastic deformation ductility
end
waitbar2(k/length(R))
end
close(hh)
mu(:,:,1)=ones(neqs,nstruc);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Calculate mean and dispersion measures
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
mubaro(:,:)=mean(mu); % mean displacement ductility
murbaro(:,:)=mean(mur); % mean residual displacement ductility
mupbaro(:,:)=mean(mup); % mean cumulative plastic deformation ductility
nybaro(:,:)=mean(nyinelij); % mean no. of yield events
mupbaro=mupbaro-1; % mean cumulative plastic deformation ductility
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
mudo(:,:)=std(log(mu)); % dispersion of displacement ductility
murdo(:,:)=std(log(mur)); % dispersion of residual displacement ductility
mupdo(:,:)=std(log(mup)); % dispersion of cumulative plastic deformation ductility
nydo(:,:)=std(log(nyinelij)); % dispersion of no. of yield events
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
musigo(:,:)=std(mu); % std. dev. of displacement ductility
mursigo(:,:)=std(mur); % std. dev. of residual displacement ductility
mupsigo(:,:)=std(mup); % std. dev. of cumulative plastic deformation ductility
nysigo(:,:)=std(nyinelij); % std. dev. of no. of yield events
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
for k=1:length(R)
mucovo(:,k)=musigo(:,k)./mubaro(:,k); % COV of displacement ductility
murcovo(:,k)=mursigo(:,k)./murbaro(:,k); % COV of residual displacement ductility
mupcovo(:,k)=mupsigo(:,k)./mupbaro(:,k); % COV of cumulative plastic deformation ductility
nycovo(:,k)=nysigo(:,k)./nybaro(:,k); % COV of no. of yield events
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
else
save plottemp *o
clear
load plottemp
252
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Plot R-L-T relationships
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
ranges=[10 4 50 25]; % specify ranges
close all
filenameo
rename=input(’change this name? ’,’s’)
if rename==’y’
filenameo=input(’enter filename: ’,’s’)
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Demand Index Cross-correlations
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
N=0; % set counter
deg=2; % degree of polynomial fit
nx=30; % number of discrete points
mumax=10; % maximum controllable mu
labels=[’\mu \mu_r\mu_pn_y ’]; % matrix titles
Tmin=15; % minimum period
Tmax=24; % maximum period (residual disp)
[Imax]=find(mubaro<mumax);
submubar=mubaro(Tmin:Tmax,:);
submurbar=murbaro(Tmin:Tmax,:);
submupbar=mupbaro(Tmin:Tmax,:);
subnybar=nybaro(Tmin:Tmax,:);
[Jmax]=find(submubar<mumax);
lambda(:,:,1)=mubaro; % set up demand index matrix
lambda(:,:,2)=murbaro;
lambda(:,:,3)=mupbaro;
lambda(:,:,4)=nybaro;
sublambda(:,:,1)=submubar; % set up demand index matrix
sublambda(:,:,2)=submurbar;
sublambda(:,:,3)=submupbar;
sublambda(:,:,4)=subnybar;
betao=[0.8 0.29]; % initial guess for a & b constants
f=25
figure(f)
for j=1:4
for i=j:4
N=4*i+j-4;
M=4*j+i-4;
lambda1=lambda(:,:,i);
lambda2=lambda(:,:,j);
sublambda1=sublambda(:,:,i);
sublambda2=sublambda(:,:,j);
if i==j
subplot(4,4,N)
axis([0 2 0 2])
axis off
text(1,1,[’\fontname{times}\it’ labels(5*i-4:5*i)])
else
lambda1=lambda1(Imax);
lambda2=lambda2(Imax);
subplot(4,4,N)
plot(lambda2,lambda1,’.’)
hold on
subplot(4,4,M)
plot(lambda1,lambda2,’.’)
hold on
mi2=min(lambda2);
ma2=max(lambda2);
dx2=(ma2-mi2)/nx;
xfit2=[mi2:dx2:ma2]’;
mi1=min(lambda1);
ma1=max(lambda1);
dx1=(ma1-mi1)/nx;
xfit1=[mi1:dx1:ma1]’;
if (j==1 | i==1)
keys=[1 3];
[betahat2,r2]=nlinfit(lambda2,lambda1,’lambdahat1’,betao);% calculate reg consts.
[betahat1,r1]=nlinfit(lambda1,lambda2,’lambdahat3’,betao);% calculate reg consts.
else
keys=[2 4];
[betahat2,r2]=nlinfit(lambda2,lambda1,’lambdahat2’,betao);% calculate reg consts.
[betahat1,r1]=nlinfit(lambda1,lambda2,’lambdahat4’,betao);% calculate reg consts.
end
betahati=[betahat2 betahat1];
resi=sum(abs([r2 r1]));
[Yk,Ik]=min(resi);
betahat=betahati(:,Ik);
coeffs2(i,j,:)=betahat;
coeffs2(j,i,:)=betahat;
res(i,j)=resi(Ik);
res(j,i)=resi(Ik);
subplot(4,4,N)
xfit=xfit2;
yfit=lambdahat(betahat,xfit,keys(1));
plot(xfit,yfit,’r’)
axis([0 ranges(j) 0 ranges(i)])
set(gca,’FontName’,’times’)
hold off
253
subplot(4,4,M)
xfit=xfit1;
yfit=lambdahat(betahat,xfit,keys(2));
plot(xfit,yfit,’r’)
axis([0 ranges(i) 0 ranges(j)])
set(gca,’FontName’,’times’)
hold off
end
end
end
suptitle([’\fontname{times}Cross-correlations, ’ filenameo ’: data points and regression lines’])
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Save Regression Constants?
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
coeffs2
res
compile=input(’save these regression constants? ’,’s’)
if compile==’y’
save([’./constants/’ filenameo],’coeffs2’,’-append’)
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Save Figures?
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
hardcopy=input(’save these plots? ’,’s’)
if hardcopy==’y’
check=ls([’./figures/’ filenameo ’*’])
if (check(1:2)~=’No’)
addon=input([filenameo ’ exists. Please enter modifier text or press CTRL-C:’],’s’)
filename=[filenameo addon]
end
figure(f)
orient tall
print(’-depsc’,[’./figures/’ filenameo num2str(f)])
end
254
B.8 NKREGRESSION.M, NKREGRESSION_ST.M, NKREGRESSION_ME.M,
NKREGRESSION_SO.M, NKREGRESSION_NF.M: Regression Functions
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% [nkregression.m]
% Nassar & Krawinkler R-mu-T regression
% This program calculates the R or mu for a given T depending
% on the user specification.
% first created: KF 3/10/00
% last revised: KF 3/10/00
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [nkout] = nkregression(nsamp,alpha,Tj,key,R)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Nassar & Krawinkler regression
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
aldata=[0 .02 .1]; % strain hardening pts
adata=[1 1.01 .8]; % a coeff. data
bdata=[.42 .37 .29]; % b coeff. data
alnk=0:.01:.1; % strain hardening fit
ank=spline(aldata,adata,alnk); % a coeff. fit
bnk=spline(aldata,bdata,alnk); % b coeff. fit
I=find(alnk==alpha); % pick point
cnk=Tj.^ank(I)./(Tj.^ank(I)+1)+bnk(I)./Tj;
T=Tj(nsamp); % sampled period
cnk=cnk(nsamp); % c parameter
if key==1
munk=0:.1:10;
Rnk=(cnk*(munk-1)+1).^(1/cnk); % R factor
nkout=Rnk; % output R
elseif key==2
Rnk=R;
munk=(Rnk.^cnk-1)./cnk+1; % mu factor
nkout=munk; % output mu
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% [nkregression_st.m]
% Nassar & Krawinkler R-mu-T regression
% This program calculates the R or mu for a given T depending
% on the user specification.
% This was modified for the UND very dense soil ensemble results
%
% first created: KF 2/22/01
% last revised: KF 2/22/01
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [nkout] = nkregression_st(nsamp,alpha,Tj,key,R)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Nassar & Krawinkler regression
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
ank=1.4628;
bnk=0.5778;
cnk=Tj.^ank./(Tj.^ank+1)+bnk./Tj;
T=Tj(nsamp); % sampled period
cnk=cnk(nsamp); % c parameter
if key==1
munk=0:.1:10;
Rnk=(cnk*(munk-1)+1).^(1/cnk); % R factor
nkout=Rnk; % output R
elseif key==2
Rnk=R;
munk=(Rnk.^cnk-1)./cnk+1; % mu factor
nkout=munk; % output mu
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% [nkregression_me.m]
% Nassar & Krawinkler R-mu-T regression
% This program calculates the R or mu for a given T depending
% on the user specification.
% This was modified for the UND stiff ensemble results
%
% first created: KF 2/22/01
% last revised: KF 2/22/01
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [nkout] = nkregression_me(nsamp,alpha,Tj,key,R)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Nassar & Krawinkler regression
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
ank=1.4925;
bnk=0.4619;
cnk=Tj.^ank./(Tj.^ank+1)+bnk./Tj;
T=Tj(nsamp); % sampled period
cnk=cnk(nsamp); % c parameter
if key==1
munk=0:.1:10;
Rnk=(cnk*(munk-1)+1).^(1/cnk); % R factor
nkout=Rnk; % output R
255
elseif key==2
Rnk=R;
munk=(Rnk.^cnk-1)./cnk+1; % mu factor
nkout=munk; % output mu
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% [nkregression_so.m]
% Nassar & Krawinkler R-mu-T regression
% This program calculates the R or mu for a given T depending
% on the user specification.
% This was modified for the UND soft soil ensemble results
%
% first created: KF 2/22/01
% last revised: KF 2/22/01
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [nkout] = nkregression_so(nsamp,alpha,Tj,key,R)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Nassar & Krawinkler regression
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
ank=-0.40898;
bnk=0.94534;
cnk=Tj.^ank./(Tj.^ank+1)+bnk./Tj;
T=Tj(nsamp); % sampled period
cnk=cnk(nsamp); % c parameter
if key==1
munk=0:.1:10;
Rnk=(cnk*(munk-1)+1).^(1/cnk); % R factor
nkout=Rnk; % output R
elseif key==2
Rnk=R;
munk=(Rnk.^cnk-1)./cnk+1; % mu factor
nkout=munk; % output mu
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% [nkregression_nf.m]
% Nassar & Krawinkler R-mu-T regression
% This program calculates the R or mu for a given T depending
% on the user specification.
% This was modified for the SAC near field ensemble results
%
% first created: KF 2/22/01
% last revised: KF 2/22/01
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [nkout] = nkregression_nf(nsamp,alpha,Tj,key,R)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Nassar & Krawinkler regression
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
ank=0.3490;
bnk=0.8949;
cnk=Tj.^ank./(Tj.^ank+1)+bnk./Tj;
T=Tj(nsamp); % sampled period
cnk=cnk(nsamp); % c parameter
if key==1
munk=0:.1:10;
Rnk=(cnk*(munk-1)+1).^(1/cnk); % R factor
nkout=Rnk; % output R
elseif key==2
Rnk=R;
munk=(Rnk.^cnk-1)./cnk+1; % mu factor
nkout=munk; % output mu
end
256
B.9 EPDISP.M, BEDISP.M, SDDISP.M: Hysteretic Rule Functions
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% [epdisp.m]
% Bilinear Elasto-Plastic Hysteretic Rule
% This program determines the yielding rules for a SDOF oscillator.
%
% first created: KF 9/18/00
% last revised: KF 9/26/00
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [df,fu,k,flag,dpl,y,r,rev]=epdisp(flag,fu,vdi,vdo,ko,alpha,vy,vi,dv,r,rev)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Yielding condition
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
if flag==1 % yielding
y=0; % yield counter
if (sign(vdi)~=sign(vdo)) % reversal
flag=0; % unload
df=alpha*ko*dv; % unloading force (k)
fu=vi; % redefine yield surface (in)
k=ko; % unloading stiffness (k/in)
dpl=0; % no plastic defo
r=1; % full reversal counter
else % inelastic
df=alpha*ko*dv; % incremental force (k)
fu=vi; % redefine yield surface (in)
k=alpha*ko; % secondary stiffness (k/in)
dpl=abs(dv); % plastic defo (in)
r=0; % full reversal counter
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Non-yielding condition
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
elseif flag==0 % elastic
if abs(vi-(fu))>=2*vy % check for yielding
y=1; % yield counter
dvo=abs(vi-(fu))-2*vy; % abs plastic disp (in)
pct=dvo/abs(dv); % plastic percentage
dv1=dv*pct; % plastic disp (in)
dv2=dv-dv1; % elastic disp (in)
df=ko*dv2+alpha*ko*dv1; % incremental force (k)
flag=1; % yielding
fu=vi; % redefine yield surface (in)
k=alpha*ko; % secondary stiffness (k/in)
dpl=abs(dv1); % plastic defo (in)
if r==1 % check for full reversal
rev=1;
end
r=0; % full reversal counter
else % elastic
y=0; % yield counter
k=ko; % elastic stiffness (k/in)
df=ko*dv; % incremental force (k)
dpl=0; % no plastic defo
if (sign(vdi)~=sign(vdo)) % reversal
if sign(fu)==sign(fu-sign(vdi)*2*vy)
mult2=sign(vdi);
else
mult2=sign(fu);
end
fu=fu-mult2*2*vy;% redefine yield surface (in)
r=0;% full reversal code
end
end
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% [bedisp.m]
% Bilinear Elastic Hysteretic Rule
% This program determines the yielding rules for a SDOF oscillator.
%
% first created: KF 9/19/00
% last revised: KF 9/26/00
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [df,k,flag,dpl,y,r,rev]=bedisp(flag,vdi,vdo,ko,alpha,vy,vi,dv,r,rev)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Yielding condition
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
if flag==1 % yielding
y=0; % yield counter
vo=vi-dv; % disp before yielding
if (sign(vo)~=sign(vi) & abs(dv)>=2*vy)
dv2=dv-sign(dv)*2*vy; % plastic disp (in)
dv1=dv-dv2; % elastic disp (in)
flag=0; % unload
df=ko*dv1+alpha*ko*dv2; % incremental force (k)
k=ko; % elastic stiffness (k/in)
257
dpl=abs(dv2); % plastic defo (in)
r=1; % full reversal counter
elseif abs(vi)<=vy % elastic
flag=0; % unload
dv1=vi-sign(vo)*vy; % elastic disp (in)
dv2=dv-dv1; % plastic disp (in)
df=ko*dv1+alpha*ko*dv2; % incremental force (k)
k=ko; % elastic stiffness (k/in)
dpl=abs(dv2); % plastic defo (in)
r=1; % full reversal counter
else % inelastic
k=alpha*ko; % secondary stiffness (k/in)
df=alpha*ko*dv; % incremental force (k)
dpl=abs(dv); % plastic defo (in)
r=0; % full reversal counter
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Non-yielding condition
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
elseif flag==0 % elastic
if abs(vi)>=vy % yielding
y=1; % yield counter
flag=1; % yielding
dvo=abs(vi)-vy; % abs plastic disp (in)
pct=dvo/abs(dv); % plastic percentage
dv1=dv*pct; % plastic disp (in)
dv2=dv-dv1; % elastic disp (in)
df=ko*dv2+alpha*ko*dv1; % incremental force (k)
k=alpha*ko; % secondary stiffness (k/in)
dpl=abs(dv1); % plastic defo (in)
if r==1 % check for full reversal
rev=1;
end
r=0; % full reversal counter
else % elastic
y=0; % yield counter
k=ko; % elastic stiffness (k/in)
df=ko*dv; % incremental force (k)
dpl=0; % no plastic defo
if (sign(vdi)~=sign(vdo)) % reversal
r=0;% full reversal code
end
end
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% [sddisp.m]
% Stiffness-Degrading Hysteretic Rule
% This program determines the yielding rules for a SDOF oscillator.
%
% first created: KF 3/14/01
% last revised: KF 3/14/01
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [df,fu,k,flag,dpl,y,r,rev,fsh,vsh]=sddisp(flag,fu,vdi,vdo,ko,alpha,vy,vi,dv,r,rev,fi,fo,fsh,vsh,k)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Yielding condition
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
if flag==1 % yielding
y=0; % yield counter
if (sign(vdi)~=sign(vdo)) % reversal
flag=0; % unload
df=alpha*ko*dv; % unloading force (k)
fu=vi; % redefine yield surface (in)
k=ko; % unloading stiffness (k/in)
dpl=0; % no plastic defo
r=1; % full reversal counter
fsh=fliplr(fsh); % reverse shoot-through force vector
vsh=fliplr(vsh); % reverse shoot-through displ vector
else % inelastic
df=k*dv; % incremental force (k)
fu=vi; % redefine yield surface (in)
dpl=dv; % plastic defo (in)
r=0; % full reversal counter
vsel=[vi,vsh(2)]; % candidates for shoot-through point
fsel=[fi,fsh(2)];
[Ysh,Ish]=max(sign(dv)*vsel);% select farthest point
vsh(2)=vsel(Ish); % shoot-through displacement
fsh(2)=fsel(Ish); % shoot-through force
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Non-yielding condition
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
elseif (flag==0 | flag==3) % elastic
if abs(vi-(fu))>=2*vy % check for yielding
y=1; % yield counter
dvo=abs(vi-(fu))-2*vy; % abs plastic disp (in)
pct=dvo/abs(dv); % plastic percentage
dv1=dv*pct; % plastic disp (in)
dv2=dv-dv1; % elastic disp (in)
if flag==3
k=abs((fsh(2)-(fo+k*dv2))/(vsh(2)-(vi-dv1)));% shoot-through stiffness (k)
258
flag=2;
else
k=alpha*ko;
flag=1;
end
df=ko*dv2+k*dv1; % incremental force (k)
fu=vi; % redefine yield surface (in)
dpl=dv1; % plastic defo (in)
if r==1 % check for full reversal
rev=1;
end
r=0; % full reversal counter
elseif (abs(vsh(1))~=abs(vsh(2)) & sign(fi)~=sign(fo))% check for shoot-through
flag=2; % shoot-through
y=0; % yield counter
dfo=fi-fo; % initial incremental force (k)
pct=abs(fo/dfo); % elastic percentage
dv1=dv*pct; % elastic disp (in)
dv2=dv-dv1; % plastic disp (in)
k=abs(fsh(2)/(vsh(2)-(vi-dv2)));% shoot-through stiffness (k)
df=ko*dv1+k*dv2; % incremental force
r=0; % full reversal counter
dpl=dv2; % plastic defo (in)
else % elastic
y=0; % yield counter
k=ko; % elastic stiffness (k/in)
df=ko*dv; % incremental force (k)
dpl=0; % no plastic defo
if (sign(vdi)~=sign(vdo)) % reversal
if sign(fu)==sign(fu-sign(vdi)*2*vy)
mult2=sign(vdi);
else
mult2=sign(fu);
end
fu=fu-mult2*2*vy;% redefine yield surface (in)
r=0;% full reversal code
fsh=fliplr(fsh);% reverse shoot-through force vector
vsh=fliplr(vsh);% reverse shoot-through displ vector
end
end
elseif flag==2
if abs(fi)>=abs(fsh(2))
y=1; % yield counter
dvo=abs(vi-vsh(2)); % abs plastic disp (in)
pct=dvo/abs(dv); % plastic percentage
dv1=dv*pct; % plastic disp (in)
dv2=dv-dv1; % elastic disp (in)
df=k*dv2+alpha*ko*dv1; % incremental force (k)
flag=1; % yielding
fu=vi; % redefine yield surface (in)
k=alpha*ko; % secondary stiffness (k/in)
dpl=dv1; % plastic defo (in)
vsh(2)=vi; % shoot-through displacement
fsh(2)=df+fo; % shoot-through force
else
y=0; % yield counter
df=k*dv; % incremental force (k)
dpl=dv; % plastic defo
if (sign(vdi)~=sign(vdo)) % reversal
fu=vi;% redefine yield surface (in)
r=0;% full reversal code
fsh=fliplr(fsh);% reverse shoot-through force vector
vsh=fliplr(vsh);% reverse shoot-through displ vector
k=ko;% back to elastic stiffness
flag=3;% elastic unloading on shoot-through
end
end
end
259
STRUCTURAL ENGINEERING RESEARCH REPORT SERIES
LIST OF TECHNICAL REPORTS
260