A probabilistic procedure is presented here to address commonly observed but mostly overlooked areas of fabrication and construction deficiency evaluation in civil engineering projects. The advancement in computational power has brought much sophistication to the analysis and design of engineering structures. However, similar sophistication is lacking in the evaluation of human error. Since fabrication and construction deficiencies are essentially human errors, the quality of constructed facilities may not show the same level of improvement. A deterministic evaluation of fabrication and construction deficiencies, in terms of load tests or inspection of the whole project, is not practical. Statistical sampling information can be used to establish the quality, and samples available in exposed areas can be used for this purpose. However, the raw sample data should be used judiciously. The appropriateness of the samples needs to be checked by conducting the χ2 test of homogeneity. The success rate parameter, obtained from the sampling information, is an important input variable in the proposed method. It is a random variable, and can be described by a standard Beta distribution. The parameters of the Beta distribution can be derived from the inspection outcomes. The hyperbinomial distribution, derived in this paper, would be the most appropriate distribution to capture the probabilistic characteristics of fabrication and construction deficiencies. The deficiencies can be quantified with a predetermined confidence level. The size of the samples is very important in the accuracy in the prediction. A large sample size is always beneficial, but the cost-benefit for the collection of large samples should be studied before any decision is made. An example demonstrates that a project with higher quality workmanship may require a smaller sample size to extract the desired probabilistic characteristics. Creative repair work can be used to satisfy all concerned parties. The method was used in several projects, and saved millions of dollars.
|Original language||English (US)|
|Number of pages||8|
|Journal||Journal of Structural Engineering (Madras)|
|Publication status||Published - Apr 2003|
ASJC Scopus subject areas