National Information Service for Earthquake Engineering
University of California, Berkeley
Note: this paper was presented at the EERC-CUREe Symposium in Honor of Vitelmo V. Bertero, January 31 - February 1, 1997, Berkeley, California.
Key areas of development, required to provide true performance-based capability in future design provisions include the incorporation of a specific serviceability level performance evaluation procedure, verification of the reliability actually inherent in buildings of different structural systems conforming to the provisions and the development and refinement of new analytical evaluation procedures capable of predicting building performance with reduced uncertainty.
Inherently, the performance-based design concept implies the definition of multiple target performance (damage) levels which are expected to be achieved, or at least not exceeded, when the structure is subjected to earthquake ground motion of specified intensity. Though development of the principles of performance based design is in its infancy, guideline documents upon which our future building codes will be based are rapidly focusing and adopting performance-based approaches. Much of the early development effort has taken place in the preparation of the NEHRP Guidelines for Seismic Rehabilitation of Buildings (ATC, 1996), intended as a resource document for use in upgrading the performance of existing buildings. The principles initiated in that document, intended primarily for existing structures, were rapidly extended by SEAOC's Vision 2000 committee and suggested for application to design of new structures, and these same concepts, have now been proposed for adoption into the Commentary to the NEHRP Provisions (BSSC, 1997a.) where they will serve as the stated basis and design intent for earthquake engineering contained in the future International Building Code.
Though the name performance-based engineering is new, the basic concept of developing buildings and structures that will meet expected performance levels under different ground motion scenarios is certainly not. For more than 20 years, SEAOC has indicated that structures designed in accordance with its recommended lateral force requirements (SEAOC, 1996) would be able to meet a number of specific performance objectives, i.e. - resist minor earthquakes without damage; moderate earthquakes with limited structural and nonstructural. damage; major earthquakes with significant damage to structural and non-structural elements, but with limited risk to life safety; and the most severe levels of earthquake ground motion ever likely to effect a site, without collapse. These same basic performance objectives, though more precisely and quantitatively defined, are being adopted by most performance-based engineering guidelines today. It is the quantitative nature of these objectives as adopted in recent efforts and the attempt at precision and reliability that sets contemporary efforts at performance-based engineering apart from earlier practice. In traditional practice, earthquake design has been explicitly performed for only a single design event level, at which a level of performance generally termed "life safety" has been targeted. While attainment of the other performance objectives cited by SEAOC is implied, no specific procedures are provided to allow evaluation of the ability of a structure to actually meet these objectives. Contemporary efforts at performance-based engineering are seeking to provide reliable methods of meeting these multiple performance goals through explicit design procedures.
Although contemporary earthquake engineering procedures, patterned after ATC-3.06 (ATC, 1978), purport to be strength based, in the sense of being an LRFD approach, in reality they are not. In current earthquake engineering procedures, structures are provided with a minimum strength based on a fraction, (I/R), times the theoretical lateral strength demand that would be experienced were the structure to remain elastic. There has never been a serious attempt to define the margin against failure provided by this approach, for the various structural systems for which R values are specified. Instead, these R factors have been set based on judgment and in part, based on observation of structural performance in recent earthquakes, to provide the so- called life safety performance level for design level earthquake ground motions. This life safety level of performance has been defined only qualitatively in terms of poorly stated considerations of limiting damage to structural elements, maintaining egress for occupants, and preventing significant falling hazards.
Both SEAOC's Vision 2000 (SEAOC, 1995) and the NEHRP Guidelines (ATC, 1996) have attempted to provide more quantitative definitions of building performance levels. Both have developed similar systems of designating building performance, though somewhat different terminology has been utilized. Table 1, below, summarizes the performance levels defined by these projects. The NEHRP Guidelines, in particular, have specified quantitative criteria, by which structural performance can be evaluated, relative to these levels. To accomplish this, the various components that comprise the structure are designated as either primary or secondary. Primary components are necessary to the lateral stability and resistance of the structure. Secondary components are not, although they may be necessary to the vertical stability of the structure. In total, secondary elements can not comprise more than 25% of the total lateral force resisting stiffness of the structure, prior to the onset of damage. Consistent with true LRFD approaches, acceptance criteria for the Life Safety and Collapse Prevention performance levels are specified based on desired margins against failure, at the component level. Table 2 summarizes the acceptance criteria for these two performance levels, for both primary and secondary elements.
The Vision 2000 document recommends that buildings be constructed, based on their intended occupancies and uses, to meet the performance objectives indicated in Figure 1. In the figure, each combination of an earthquake return period and performance level, indicated by a red diamond, represents a specific design performance objective. The intent is that ordinary buildings provide a low risk that life be endangered as a result of the performance of the building in any earthquake likely to effect it; and that for frequent earthquakes, the building user not be burdened with extensive repairs or loss of use; that buildings required for emergency response and essential public function have a low risk of being damaged beyond a level that would permit their use, and; that facilities housing systems and materials that would pose a hazard to many persons if released, have a low risk of damage resulting in such release. The NEHRP Guidelines suggest similar performance objectives as the basis for rehabilitation design for existing structures and specifically recommend that a performance evaluation be performed for each specifically intended performance objective. The performance evaluation consists of a structural analysis with computed demands on structural elements compared against specific acceptance criteria provided for each of the various performance levels. This is in contrast to the approach taken by current building code provisions, wherein a single performance evaluation is required, for the Life Safety performance level at a specified level of ground motion, termed the Design Basis Earthquake.
The single performance evaluation inherent in the current building codes is appealing to those responsible for developing, adopting and enforcing them as it aligns well with the basic role of public safety protection intended for these documents. However, because the Life Safety performance level is relatively poorly defined in terms of the margin against failure provided, this performance evaluation has little technical meaning. As our future codes move towards a more closely performance-based concept it would be preferable to abandon the so-called Life Safety basis for design and adopt an approach that is truer to LRFD methods. Specifically, as with all LRFD design methods, two performance states should be considered - a serviceability state, similar to the Operational level of the NEHRP Guidelines, and a failure, or collapse state. Structures should be proportioned such that they provide an appropriate margin against the collapse state under maximum expected, or considered, levels of load and such that they not exceed the serviceability state under frequent levels of load.
The concept of gradation of performance objectives based on building occupancy and use, as suggested by Vision 2000, is an appropriate one. However it is not necessary to adopt four independent levels of performance, as suggested by Vision 2000, in order to attain the enhanced performance desired for such structures. The two basic LRFD levels, serviceability and collapse are sufficient for this purpose. For relatively more important structures, the design margin against the collapse state for maximum expected loads should be increased relative to ordinary structures. Similarly, for such important structures, the load level at which serviceability performance is required should also be increased, such that the probability that the serviceability level is exceeded, is reduced.
In the development of the 1997 Provisions, it was decided to abandon the concept of a design basis earthquake ground motion with uniform probability of exceedance throughout the nation. Instead, in the development of new hazard mapping for use by the Provisions, the United States Geologic Survey (USGS) was directed to depict ground motion response parameters for a maximum considered earthquake (MCE) ground motion. This MCE motion is typically defined as having a 2% probability of exceedance in 50 years, as it was deemed that consideration of less probable levels of ground motion would be inconsistent with the level of risk adopted by society with regard to other hazards. The exception to this definition of MCE motion is in areas close to known active faults capable of producing large magnitude events. In such locations, where the source of the hazard is well defined, it was felt that rather than resorting to a probabilistic definition of MCE ground motion it would be more appropriate to base NICE ground motion on a maximum, or characteristic, magnitude earthquake on the fault. Specifically, it is taken as 150% of the ground shaking obtained from median attenuation relationships for the characteristic earthquake. In essence, using the terminology discussed relative to LRFD approaches, the NICE ground motion is the maximum expected loading.
The intent of an LRFD approach is to design for a high confidence of a low probability of failure at maximum expected load. Determination of the probability of failure for a multi-degree of freedom, nonlinear structural system such as a building, in response to complex dynamic loading such as earthquake ground motion is an exceedingly difficult task and has never been performed in a comprehensive manner for the wide range of structural systems covered under the scope of the NEHRP Provisions. However, it was the judgment of members of the Seismic Design Procedures Group (SDPG), a joint task force of the BSSC and USGS engaged in development of the new hazard maps and provisions, that buildings designed and constructed in conformance with the procedures of the 1994 NEHRP Provisions would be able to resist a loading at least P/2 times larger than the design ground motion, without collapse. Therefore, it was decided in the 1997 NEHRP Provisions to specify a design ground motion (loading) that is 11(l.5) or 2/3 of the MCE ground motion, such that for maximum expected loading (the MCE ground motion), a high confidence of a low probability of failure would exist.
Another important feature of the proposed 1997 Provisions is the introduction of an occupancy importance factor, I, to regulate the amount of margin provided in a design depending on the use and importance of the building. The I factor is introduced into the base shear as a modifier on the response modification coefficient, R, such that an effective response modification coefficient (R/I) is used to determine design force levels. The value of I varies from 1 for ordinary structures to 1.5 for essential structures. This has the effect of increasing the minimum presumed margin of 1.5 for ordinary structures to a value of (1.5) squared, or 2.25, for essential structures. In addition to increasing the inherent safety margin in essential structures, the I factor also has the effect of increasing the load level at which elastic behavior can occur and therefore results in a raising of the threshold level at which damage is expected to initiate, consistent with the design goals of providing reduced damage and improved safety in important structures.
The incorporation of a meaningful serviceability evaluation procedure should receive a high priority in future design procedure development. Recent California earthquakes have demonstrated a largely satisfactory reliability for our modem structures with regard to collapse avoidance, however, the failure of many modem structures to remain serviceable following moderate to severe ground shaking has been a source of concern to the business, financial and emergency management communities, and in fact, has largely lead to the current motivation for development of performance-based design procedures. Conceptually, it should be relatively simple to develop a serviceability evaluation procedure for future design provisions. Since by definition, serviceability implies limited damage, structures must behave in an elastic, or near--elastic manner at the service level and our existing linear analysis methods are probably adequate for the performance of a meaningful serviceability evaluation. However, considerable work must be performed to define an appropriate load level, or earthquake exceedance probability, at which serviceability should be obtained for structures of different occupancy and use. Further, much more research into the behavior of nonstructural building components including architectural mechanical and electrical components, is required to permit development of appropriate design parameters and acceptance criteria for these important building elements. Although the determination of acceptable levels of structural response and damage at a service level of performance is an engineering task, the most difficult task, determination of the service load level or exceedance probability is really beyond the sole province of the structural engineer and requires the participation of the financial, social planning and regulatory communities.
Development of reliable evaluation procedures for the failure, or collapse state, capable of accounting for the important velocity pulse and ground motion duration effects must also clearly occur. In addition, there is a need to develop a series of prototype, or model buildings, representative of our various structural systems, so that the reliability of various analytical and design procedures can be evaluated on a consistent basis. Finally, more comprehensive determination of the reliability inherent in our design procedures must be performed so that appropriate margins of safety can be maintained as our engineering procedures evolve.
ATC (1996). "NEHRP Guidelines for the Seismic Rehabilitation of Buildings FEMA-273 Ballot Version." September, 1996.
BSSC (1995) "NEHRP Recommended Provisions for Seismic Regulations for New Buildings," FEMA-222." May, 1995.
BSSC (1997a). "Commentary to the NEHRP Recommended Provisions for Seismic Regulation of New Buildings and Other Structures, FEMA-223." publication pending
BSSC (1997b). "NEHRP Recommended Provisions for Seismic Regulation of New Buildings and Other Structures, FEMA-222." publication pending
SEAOC (1995). "Vision 2000 - A Framework for Performance Based Earthquake Engineering." Vol. 1, January, 1995.
SEAOC (1988). "Recommended Lateral Force Requirements and Commentary." 1988