Toward Performance-based Engineering
A relationship, which is shown below graphically, is developed between the performance objective, type of facility, probability of earthquake occurence, which is then tied to response parameters related to each performance objective. These parameters are identified and some initial estimates are quantified.
As shown, the performance objective increases (i.e. there should be less damage) for a high probability earthquake (one that may occur several times during the life of the structure) or for an important structure or dangerous occupancy (i.e. a hospital or dynamite plant). Conversely, more damage is acceptable for a rare, severe earthquake or for less critical or temporary facilities. Thus, a building would be expected to suffer more damage if it were subjected to a more severe, less likely earthquake. Also, a more critical building would be exepected to have less damage for the same earthquake probability.
A basic structure would be expected to have essentially no damage if subjected to an even with a 10% probability of occurence in 30 years whereas it would be near collapse if subjected to an event with a 10% probability within 100 years. One can substitute more appropriate numbers for a particular project, or upgrade the characterization of the structure (to an essential facility, for instance, where the structure would be designed to remain life safe during the very rare event).
This method removes some of the ambiguity from the current SEAOC recommendations. The method still needs to indicate what performance parameters to consider (drift, stress, plastic hinge rotation, acceleration, etc. and what limits are to be imposed to achieve a particular performance objective. Some information on performance parameters was provided in Vision 2000 for basically the first time, but it was for the most part based on consensus rather than on test data or quantitative field observation.
Vision 2000 has four performance (i.e. limit) states:
Three occupancy types are considered in Vision 2000:
The earthquake intensity is described quantitatively in probabilistic terms as follows:
* need not exceed mean + 1 standard deviation for the maximum deterministic event
Drift limits have recently been added to Vision 2000, and are shown below.
These limits raise several questions. How do we calculate the maximum drift (or maximum permanent drift) and prove we satisfy these criteria? Why are these criteria selected? Will a building at 2.6% drift collapse?
Since there is a large jump in damage from the Operational limit state to the Life Safe limit state, another limit state related to repairability may need to be added. In this new state, damage would be limited to make repair quick and/or economically feasible. However, since damage is difficult to quantify and economics issues are owner sensitive, intermediate states are difficult to incorporate in a code.
Vision 2000 approach has several limitations. First, it does not suggest
analytical approaches nor methods to assure the reliability of the structure.
Also, intermediate limit states are difficult to quantify. In addition,
note that Vision 2000 is an uncoupled approach - that is, we end up with
a deterministic procedure based on a probabilistically determined spectrum.
Load and resistance factors still remain to be determined to provide desired
reliability. Lastly, identification of limit states by a subjective name
(i.e. fully operational) may lead to legal problems. The Japanese have
avoided this potential problem by using a letter system (category A, B,
C, etc. performance objective). Probabilistic specification fo response
parameters may be better, however.