Automated Hazard Identification for Construction Safety Management

Abstract

The National Institute for Occupational Safety and Health (NIOSH) suggested several contributing factors to construction injuries and fatalities such as lack of hazard identification, lack of coordinates of activities, inexperienced workers, etc. Among these contributing factors, hazard identification is the first step of safety management. Construction safety management process mainly consists of hazard identification, risk assessment, and risk countermeasure. Although all three steps are important, identifying hazards becomes a fundamental step because unidentified hazards negate the risk assessment and the risk countermeasure process. As a result, an unidentified hazard can cause harm to construction workers who will remain unprotected until it is eliminated.

When workers start their works, they move and orient their activity areas. In this situation, a construction worker encounters a hazard during both moving and working. This study briefly classifies hazards, which a worker can encounter during moving and working, into two following categories: 1) activity-oriented hazard

and 2) non-activity-oriented hazard. These two types of hazards have different characteristics. Therefore, different approaches are taken to identify each type of hazard. In this regard, the objective of this disseratation develops a hazard identification system that can deal with two types of hazards.

Activity-oriented hazards can be identified by using past accident cases. To retrieve similar past accident cases and apply to corresponding site, indices for generating queries are firstly selected. The indices are extracted from building information model (BIM) and project management information system (PMIS). Based on extracted indices, queries for retrieving past similar accident cases can be generated. After generating queries, search engine retrieves cases and calculates similarity. The results include time and geometric information, as well. Also, they are automatically provided to the workers by using the push system.

The non-activity-oriented hazard system identifies potentially hazardous areas (PHA) on workers path by using the RTLS of workers and a building information model (BIM). The suggested system identifies PHA by using the deviation between the optimal route (the shortest path), which is determined by extracting nodes from BIM objects, and the real movement path of a worker, which is tracked by radio frequency identification. After this process, PHA can be divided into uncontrolled PHA and already controlled PHA. The information about uncontrolled PHA is provided to safety managers so they can establish proper safety countermeasures and manage the area.

Consequently, this study will contribute to enhance the efficiency of hazard identification of safety management by automated data processing procedure that can eliminate iterative and repetitive process. The automated procedure can identify a hazard quickly, so that the existing time a hazard without any countermeasure diminishes. It means that the time a worker exposed to risk decreases.