Вы находитесь на странице: 1из 6

8 th ASCE Specialty Conference on Probabilistic Mechanics and Structural Reliability

PMC2000-311

RELIABILITY AND PERFORMANCE BASED DESIGN FOR SEISMIC LOADS


S-Y. Yun University of Illinois at Urbana-Champaign, Urbana, IL 61801 s-yun1@uiuc.edu D. A. Foutch and K. Lee University of Illinois at Urbana-Champaign, Urbana, IL 61801 d-foutch@uiuc.edu, kihaklee@uiuc.edu
Abstract Unexpected brittle fracture behavior of the welded steel moment frame connections during the 1994 Northridge Earthquake has raised questions about the performance of the system. To predict and to evaluate the performance of the moment frame systems, a procedure for seismic performance evaluation based on nonlinear dynamics and reliability theory was developed. It features full integration over the three key stochastic models: ground motion hazard curve, nonlinear dynamic displacement demand, and displacement capacity. Further, both epistemic and aleatory uncertainties are evaluated and carried through the analysis. A suite of uncertainty analyses are input to the procedure such as period, live load, material properties, damping, analysis procedure, and orientation of the structure. Several limit states are defined instead of the traditional single state. Not simply a mean estimate of the annual probability, but rather the percentage of confidence that the building will meet the performance objective is obtained through the procedure.

Introduction Structural failures observed in the Northridge and Kobe earthquakes have exposed the weakness of current design and construction procedures and shown the need for new concepts and methodologies for evaluation of building performance and design. A central issue is proper treatment and incorporation of the large uncertainty in both seismic loading and building resistance in the evaluation and design process. The state of the art of statistical and reliability methods that can be used for this purpose have been reviewed and several critical issues directly related to the mission of SAC project have been discussed in the report Critical Issues in Developing Statistical Framework for Evaluation and Design (Wen and Foutch, 1997). Based on the review, a statistical and reliability framework for the purpose of comparing and evaluating predictive models for structural performance evaluation and design was developed. This was further advanced by Hamburger (1998) and Jalayer and Cornell (1999). From this basis, the load and resistance factor approach described below has been adopted by the SAC Phase 2 Project. Acceptance Criteria For the Collapse Prevention performance level, the desired performance is to prevent global or local collapse. Global collapse is assumed to have occurred when the numerical calculation of dynamic r esponse becomes unstable or a drift of 10% in any story has been reached. Local collapse is assumed to have occurred when the rotation at each end of a

Yun, Foutch and Lee

girder is so large that the gravity load carrying capacity is lost. It is based on experiments and is different for each connection type. The acceptance criteria may be written in equation form as

con =
where:
D = estimate of median drift demand

C a D

(1)

con = confidence factor k = slope of the hazard curve RC = standard deviation of the natural log of the drifts due to randomness in capacity

C = estimate of median drift capacity


= resistance factor = R U = e
2 k RC 2 2 k UC 2

= demand factor = e

2 k RD 2 2 k a 2

a = analysis demand factor = e

UC = standard deviation of the natural log of the drifts due to uncertainty in capacity

Estimate of Median Capacity, C


The global stability limit is determined using the Incremental Dynamic Analysis (IDA) technique developed by Cornell and his associates (Luco and Cornell, 1999). This requires the following steps: 1. Choose a suite of ten to twenty accelerograms representative of the site and hazard level. The SAC project developed typical accelerograms for Los Angeles, Seattle and Boston sites (Somerville, 1997). These might be appropriate for similar sites. 2. Perform an elastic time history analysis of the building for one of the accelerograms. Plot the point on a graph whose vertical axis is the spectral ordinate for the accelerogram at the first period of the building and the horizontal axis is the maximum calculated drift at any story. Draw a straight line from the origin of the axis to this point. The slope of this line is referred to as the elastic slope, Se 3. Perform a nonlinear time history analysis of the building subjected to this accelerogram. Plot this point on the graph. Call this point 1 . 4. Increase the amplitude of the accelerogram and repeat step 3. This may be done by multiplying the accelerogram by a constant that increases the spectral ordinates of the accelerogram by 0.1g. Plot this point as 2 . Draw a straight line between points 1 and 2 . If the slope of this line is less than 0.2 Se then 1 is the global drift limit.

Yun, Foutch and Lee

This can be thought of as the point at which the inelastic drifts are increasing at 5 times the rate of elastic drifts. 5. Repeat step 4 until the straight line slope between consecutive points i and i+1 , is less than 0.2 Se. When this condition is reached, i is the global drift capacity for this accelerogram. If i+1 > 0.10 then the drift capacity is taken as 0.10. 6. Choose another accelerogram and repeat steps 2 through 5. Do this for each accelerogram. The median capacity for global collapse is the median value of the calculated set of drift limits. An illustration is shown in Figure 1. The open triangles represent the IDA for an accelerogram where the 0.2 Se slope determined the capacity. The open circles represents a case where the default capacity = 0.10 applies. The factors that effect the curve of the IDA are P- effects, increment used for the analysis, ground motions used, strain hardening ratio, shifting of fundamental period due to nonlinearity, higher mode effects, and shifting of maximum story drift location
1.0

0.8

slope=0.2 Se 4 3 2 1
0.050

5 slope=0.2 Se

0.6 Sa, (g) 0.4

0.2

slope=Se slope=0.2 Se

0.0 0.000

0.100

0.150 drift ratio

0.200

0.250

0.300

Figure 1. Illustration of Two IDA Analyses

Uncertainties The standard deviation of the natural log of the variable, , for each source of uncertainty and randomness as determined for the SAC project will be given below for the 2/50 hazard level. The analyses are based on studies for three buildings designed for the LA site according to the 1994 UBC Code. Uncertainty in estimating the period and the damping of the structure was performed using collection of measured data. The major part of the data is from the report by Goel and Chopra (1997). The collected data were plotted by height verses the variable, in this case damping. Some of the data that were largely scattered from the median values were omitted from the study. The best-fit line from the regression analysis of the data was

Yun, Foutch and Lee

calculated for damping and is shown in Figure 2. A constant standard deviation was assumed. The changes in drift values were then calculated due to the variation in the variables. The procedure for calculating the variance of the natural log of the maximum drift associated with different sources of uncertainty is as follows: 1. Run 10 or 20 time histories using the mean of the variable and call this value 1 . 2. Find the mean of the natural log of the corresponding maximum drift, m1. 3. Run 10 or 20 time histories using the mean+stdev of the variable and call this value 2 . 4. Find the mean of the natural log of the corresponding maximum drifts, m2. 5. Calculate the sensitivity which is

(m 2 m1 ) ( 2 1 )

(2)

6. The variance of the natural log of the drift with respect to the variable is the square of the sensitivity times the variance of the variable.
( m2 m1 ) 2 2 VAR = var = (m 2 m1 ) ( var + var ) var
2

(3)

Therefore, ln Drift

(m 2 m1 ) 2 = var = m 2 m1 ( var + var ) var


2

(4)

For damping, the lnDrift values are 0.024 for 3-story, 0.030 for 9-story and 0.034 for 20story. A similar procedure was taken for calculating the variation in drift due to period estimation. The period values and the regression line are shown in Figure 3. The lnDrift values are 0.47 for 3-story, 0.29 for 9-story and 0.15 for 20-story building. Variation of maximum drifts due to the randomness in the orientation of the structures to the ground motion was investigated. The structures were excited by ten fault-parallel components and ten fault-normal components of ground motions representing the 2/50 hazard level. The procedure for the orientation is similar to the procedure previously described for damping and period except that the orientation is a uniformly distributed function instead of a log-normally distributed function. The orientation of fault parallel (00 ) and fault normal (900 ) were selected for the study. The equation is

ln Drift
where, 0 o

( o 0o ) 2 = 90 o ori 90 0 o
2

(5)

= median drift for the 00 rotated (fault parallel) ground motions

Yun, Foutch and Lee

90 o ori

= median drift for the 900 rotated (fault normal) ground motions = standard deviation of uniform distribution from 0o to 90o = 260

The lnDrift for the orientation of the building was 0.21 for 3 -story, 0.19 for 9-story and 0.26 for 20-story. All of these s are incorporated into the demand factor a.
8
3-story

data points linear best fit


9-story 20-story

7 6 5
20-story

7 6 Damping, (%) 5 4 3 2 1 0 0

y = 0 .028 x 0.849

Period, (sec)

4 3
3-story 9-story

2 1 measured period nonlinear regression 0 300 400 0 100 200 300 400 Story Height, (ft) 500 600

y = 0 .0089 x + 4. 66
100 200 Story Height, (ft)

Figure 2. Damping vs. Story Height

Figure 3. Period vs. Story Height

Evaluation Process for New Buildings The procedure for evaluation of a 9-story building designed according to the 1997 NEHRP Provisions and built using reduced beam section would be as follows: 1. Determine Ss and S1 for the site from maps or the USGS web site. Determine the design response spectrum following the 1997 NEHRP provisions, but replace the 2/3 factor by 1.0 for determining SDS and SD1. 2. Calculate the maximum drift demand, m, using NEHRP linear static procedure. The demand drift, D , is then calculated as D = CB m = 0.034 (6) where m CB = the maximum story drift angle, for all stories = 0.028 = bias factor = 1.20 for 2/50 hazard level.

3. Get C =0.09 (global), = 0.85 (global), and C =0.07 (local), = 0.90 (local).
4. For CP performance level get a = 1.07 and = 1.21 5. Using Equation (1), calculate which gives con = 1.74 for global and 1.43 for local. 6. Get UT which is 0.40 for CP against global collapse and 0.33 against local collapse for 2/50 hazard level. The confidence level in achieving the performance objective resulted in 97% for global stability and 93% for local stability 93%. Decide if the

Yun, Foutch and Lee

confidence is acceptable. If not, redesign the frame to make it stiffer and, therefore, reduce D . The new buildings performed very well against CP performance level. An existing building designed according to the 1994 UBC resulted in a confidence level of 70% for the global stability and 42% for the local stability which is much lower. Concluding Remarks The reliability of the building systems in general can be improved by increasing the mean capacity against the limit state and/or reducing the uncertainty in the capacity. The increase in mean capacity can be achieved by more stable configuration, more redundancy of the systems and stronger and stiffer components including connections against overload and low-cycle fatigue failures. This increase would result in lower response at each level and hence lower limit state probability.
Acknowledgements This work was funded by the SAC Joint Venture through a grant from the Federal Emergency Management Agency. This support is gratefully acknowledged. Any result, findings and conclusions are strictly those of the authors and do not necessarily represent those of the sponsors. References Foutch, D.A. (1999), Performance Prediction and Evaluation of Steel Moment Frames Under Seismic Loads, SAC Joint Venture Report No. XX, FEMA XXX. Goel, R. K. and Chopra, A. K. (1997), Vibration Properties of Buildings Determined From Recorded Earthquake Motions, Report No. UCB/EERC-97/14, University of California at Berkeley, CA Hamb urger, R.A. (1998), Performance Based Analysis and Design Procedure for Moment Resisting Steel Frames, Internal SAC Document. Jalayer, F. and Cornell, C. A. (1998), Development of a Probability-Based Demand and Capacity Factor Design Seismic Format, Draft. 8/11/98. Publication as a SAC Background Document pending Lee, K. and Foutch, D. A. (2000), Performance Prediction and Evaluation of Special Moment Frames, SAC Joint Venture Report No. XX, FEMA XXX. Luco, N. and Cornell, C. A. (1998), Seismic Drift Demands for Two SMRF Structures with Brittle Connections, 6 th U.S. National Conference on Earthquake Engineering, Seattle, WA Somerville, P., Smith, N., Punyamurthula, S., and Sun, J. (1997). Development of Ground Motion Time Histories for Phase 2 of the FEMA/SC Steel Project. SAC Joint Venture, Sacramento, CA Wen, Y. K. and Foutch, D. A. (1997), Proposed Statistical and Reliability Framework for Comparing and Evaluating Predictive Models for Evaluation and Design, and Critical Issues in Developing such Framework, Report No. SAC/BD-9703 SAC Joint Venture. Richmond, CA Yun, S-Y. and Foutch, D. A. (2000), Performance Prediction and Evaluation of Low Ductility Moment Frames, SAC Joint Venture Report No. XX, FEMA XXX.

Yun, Foutch and Lee

Вам также может понравиться