|Ph.D Student||Raviv Gabriel|
|Subject||Modeling of Near Misses Related to Crane Work at|
Construction Sites and their Safety Risk
|Department||Department of Civil and Environmental Engineering||Supervisors||Professor Emeritus Aviad Shapira|
|Professor Barak Fishbain|
|Full Thesis text - in Hebrew|
A near miss can be explained in several ways, all of which lead up to the definition according to which an ongoing chain of events was interrupted and an accident was prevented, leaving the work environment unharmed, albeit with some release of hazard potential. Near misses occur considerably more often than do accidents, and many times they escape becoming a full-scale accident by only a hairbreadth. Thus, the value of studying near misses is equal to the value of studying accidents, but without having to pay the price of bodily and property damages. The construction industry tends to adopt near-miss management systems, but such procedures are relatively new and have not yet been fully explored or understood. The current study implemented qualitative as well as quantitative research methods aiming at studying safety incidents (accidents and near misses) relating to tower cranes at construction sites.
The study began with establishing a comprehensive database of tower-crane-related safety incidents. Leading construction companies in Israel revealed data on near misses and accidents taken from existing files, and also provided the research team with the opportunity to proactively elicit incident stories during field interviews. A large repository of stories was consequently collected, representing various context conditions such as different companies, methods of data collection, and levels of detail. The database ultimately included 241 tower-crane safety incidents, of which 162 were near misses and 79 were full-scale accidents. Qualitative research methods were implemented in classifying the incidents according to basic common themes that define an extensive nomenclature of tower-crane-related incidents organized into categories with variables within each one of the categories.
The quantitative analysis began by assigning the database variables with quantitative values for describing each incident by a set of numbers, termed the “feature vector”. The database was divided into groups (clusters) by implementing the k-means algorithm. An iterative algorithm was developed and consequently the database was divided into five clusters with prominent characters that could be identified and serve as a tool for the incident risk potential evaluation. The risk potential analysis was divided into two terms - the partial risk potential (PRP) and the total risk potential (TRP). The PRP values enabled retrofitting each cluster with certain characteristics and the TRP values graded the clusters according to their hazardous nature.
The study’s most prominent conclusion is the contradiction of the common perception regarding the role of the human factor as a main contributor to construction site accidents. Introducing the risk potential idea, as well as its analysis method, contradicted this common perception by quantitatively assessing the incidence of failures that led to high risk potential incidents.
The study’s contribution can be described from two perspectives, the theoretic contribution and the applied contribution. The theoretic contribution relates to the introduction of a method for the qualitative and quantitative analyses of multi-incident reports. The applied contribution relates to the recognition of hazards revealed by safety incident reports, as well as the introduction of a structured method for assessing the reporting organization’s safety status.