12.07.2015 Views

Report - PEER - University of California, Berkeley

Report - PEER - University of California, Berkeley

Report - PEER - University of California, Berkeley

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

to allow a quantitative assessment <strong>of</strong> their relative importance). It can be noted that infact their influence is almost insignificant in this case, which is another way <strong>of</strong>confirming that displacement-related response quantities such as εccare only weaklydependent on strength-related mechanical parameters.For what concerns the shear demand, after yielding <strong>of</strong> the column extremities, asexpected the curve tends to flatten. Again, the adjacent plots show the sensitivities <strong>of</strong>the shear demand with respect to f cand f . While the first one is on the averageyzero, the second one shows that for a positive variation <strong>of</strong> the yield stress the increasein shear demand is independent <strong>of</strong> the PGA value. This is expected, since thevariation in shear force due an increase in the yield stress remains constant along thehardening branch.Finally, it is worth commenting that the sensitivities <strong>of</strong> the response with respectto the capacity variables εcc, uand ε Vhave not been computed. This fact comesunavoidably from a limitation <strong>of</strong> the available analytical tools, which do not allow toaccount in the course <strong>of</strong> the analysis for the modification <strong>of</strong> the response due to theattainment <strong>of</strong> the capacity in some <strong>of</strong> the members. Hence, any perturbation in thecapacity parameters would go undetected during the analysis, yielding identicallyzero derivatives.3.4.1 Fragility CurvesOnce the demand variables are statistically determined, reliability analysis canproceed as indicated in Section 2. The most straightforward and accurate way toevaluate the system probability <strong>of</strong> failure in Eq. (10) is to resort to Monte Carlosimulation. It is recalled that at this stage no more structural analyses are needed andthat a trial <strong>of</strong> the MC simulation simply consists <strong>of</strong> sampling from the jointdistribution <strong>of</strong> x (in this case f , f ,ε , ε ) and checking the state <strong>of</strong> the system. Itc y cc, u Vis worth observing that the evaluation <strong>of</strong> the entire fragility by MC simulation at thislast stage <strong>of</strong> the procedure usually involves less effort than a single non-lineardynamic analysis.Figure 6 (Left) contains the fragility curves for the structure, evaluated by Eq.(10), as well as by simpler alternative procedures. In particular, the simplest one isthat based on the assumption <strong>of</strong> independence among the failure modes (Eq. (3)),while in the second alternative the sensitivities are ignored, as suggested by theirmodest influence on the response (see Figure 5).One can note that the independence assumption leads to quite different resultsfrom the other curves, that are considerably more severe in the upper part <strong>of</strong> thefragility.The closeness between curves (a) and (c), i.e., between considering or ignoringthe dependence <strong>of</strong> the demands on x through the sensitivities, is a further indicationthat in many cases, such as the present one, the dominant effect is the ground-motioninducedvariability, which tends to overshadow that related to structural randomness.230

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!