Download files
Abstract
While numerous studies have strived to identify generic defect causes, analysis of the complex mechanisms acting among these causes, in which the most influential patterns are identified, is lacking. To address this deficiency, this research aims to develop the analytical understanding of the complex mechanisms of defect generation so as to identify and quantify defects most influential root causes, and accordingly develop effective defect prevention strategies.
In order to trace these mechanisms this thesis initially presents a classification protocol that categorizes the defective acts (i.e., errors and violations) that are directly linked to the occurrence of a defect. Knowing these direct causes provides a platform for thoroughly tracking construction defects to their root causes. Following this basis, the thesis develops a methodology that utilizes Fault Trees and risk Importance Measures (IMs) to identify the high frequency (i.e., most common) and the high magnitude (i.e., most likely to cause defects) root causes. The thesis further develops a complementary methodology that utilizes Social Network Analysis (SNA) to formulate the interrelationships between these root causes, and identify the high pathogenicity (i.e., most likely to case defects through intermediaries) causes based on their position within the network of sequential events leading to defects.
Both methodologies were tested on four residential projects in Dubai, which confirmed their applicability as effective tools for their intended purposes. Thence, perceptions of 106 industry professionals on a questionnaire survey were collected and analyzed using the aforementioned methods to demonstrate the methods applicability upon larger datasets, while identifying the most influential root causes in terms of frequency, magnitude and pathogenicity. The mechanics and sequence of events associated with these causes are discussed and graphically visualized using SNA. From this, practical defect management recommendations are provided. The identified causal patterns were further validated using a different methodological approach; namely, Bayesian Belief Networks (BBN). This validation procedure highlighted the benefit of the joint utility of SNA and BBN. In essence, this thesis contributes to knowledge in the area of construction engineering and management by proposing methodological approaches that is targeted to providing both mathematical and visualization analyses needed to better understand the complex mechanisms of defect generation.