In cases where statistically sufficient data are available, the problem of safety evaluation can be formulated with the application of the theory of probability and statistics. For example, Berens et al. presented results of a comprehensive statistical study of crack growth.?The unique feature of most civil engineering structures is that they are not only much larger and more complex than aircraft structures but also design dependent. In the great majority of cases, the same design is used and built only once. Therefore, it is not economically feasible to collect statistically sufficient data. In fact, the structural behavior near failure remains unknown for most civil engineering structures because it is not practical to conduct even a single full-scale destructive test of such large and massive structures.

In civil engineering practice, a given structure can be studied either experimentally or analytically or both whenever one or more interested so wish as (a) a result of observing signs of distress or failure or ( b) a part of periodic inspection procedure. The experimental studies can be field surveys and/or laboratory tests.?Field surveys include the determination of exact locations of failed elements and other signs of distress, the conduct of various nondestructive testing techniques to the existing structure, the discovery of poor workmanship and defective details, and proof-load and other load testing of a portion of a large structure. On the other hand, specimens can be collected from the field and tested in the laboratory for the evaluation of material strengths and other mechanical?properties. Analytical studies may consist of the examination of the original design calculations and drawings, the review of project specifications, the performance of additional analyses using newly acquired field observations and test data, and the possible explanation and description of the event under consideration.

In 1981, Guedelhofer presented a comprehensive summary of methods which are applicable for the strength evaluation of distressed structures. He classified these methods into the following three types:?(a) analytical evaluation, (b) in-situ load testing, and ( c) model load testing. Moreover, he suggested that it is necessary to follow an integrated systematical approach involving the following three frequently iterating steps: (i) condition survey, (ii) structural evaluation, and ( iii) correlation. The condition survey should produce a comprehensive documentation of the in-situ and as-built condition of the structure being evaluated. The structural evaluation is the determination of the structural behavior including the effects of all relevant field conditions. The correlation step is the description of the relationship between predicted structural behavior with the observed distress conditions.

In the mid-seventies, a safety evaluation program was developed.?The structural and fire evaluation model ( SAFEM) was developed to provide a broad overview of potential safety problems for more than 10, 000 buildings for a governmental agency in the States. A building can be classified into (a) 'green' requiring only routine scrutiny,?(b) "yellow requiring some attention, and ( c) 'red' requiring immediate attention and improvement. Authors emphatically stated that 'SAFEM is not a substitute for an engineering analysis, but it directs attention to buildings which require engineering analysis on a priority basis'. The procedure consists of (i) collection of such data?as building size, cost, number of occupants, address, and predetermined exposure to natural hazards; (ii) ranking buildings on the basis of priorities; (iii) choosing buildings which should undergo field surveys; (iv) performing field surveys and recording survey results in the computer file; (v) re-ranking buildings on the basis of priorities and requesting engineering studies for buildings with the largest potential problems, (vi) performing engineering studies and producing the final priority rankings, and (vii) allocating funds for upgrading these structures following these priorities. A detailed computer program is developed on the basis of professional experience to combine numbers ranging from 0 to 9 for hazards (geophysical, intrinsic, and local), exposure, and vulnerability. The SAFEM profiles include one each on fire, structural, and miscellaneous (glass safety, electrical system, elevator system, etc.). Although this program covers a broad scope and many detailed considerations, it is difficult to understand how the computer program evolved with the many subjective inputs.

Several recent damage assessment studies are reviewed in a comprehensive manner by Scholl et al., who recommended a general procedure. By computing the maximum floor responses, they estimate the component damage by using a component motion-damage library.?The damage to a structure is defined as the sum of the component damage which in turn are sums of sub-component damage.

Recently, Hart et al. proposed the use of reliability indices for the evaluation of structural damage. Their evaluation process consists of the following parts: (a) identification of possible failure modes,?(b) selection of analytical models representing these failure modes,?(c) quantification of the mean values and uncertainties for respective loads and resistances, (d) calculation of the reliability index for each?failure mode, and (e) application of decision theory techniques to evaluate the potential structural damage.