An Agent-based Decision Support Tool to Identify Near-Miss Incidents and Accessibility Issues in Accelerated Bridge Construction

Project Information 

Link to Latest Report: Coming Soon!

Background:

The construction industry workforce carries a large share of the overall industry workforce and experiences very high levels of injury rates, unlike other industries. Based on recent statistics, construction industry experiences nearly one-fifth share of the overall workplace fatalities. Ensuring safety at active construction sites is a major concern all over the world. Construction safety is a complex concept that involves the integration of uncertainties and unknowns associated with construction activities. The construction industry requires expertise from diverse professionals who differ widely in hazard identification, risk perception, and situational awareness . The empirical literature offers extensive analyses of infrastructural, behavioral, and organizational modules on safety practices at construction sites. Many researchers have interpreted construction hazards in terms of site layout screening, construction stages, and activities. The recent paradigm for construction safety relies more on the cognitive models and the behavioral impacts of the construction stakeholders. However, comprehensive safety analyses for active construction sites call for the integration of stakeholder perception along with the site risk screening approach.

Technological advancement has paved the way for predictive analysis of construction hazards. The array of research on modeling of near-miss events has progressed from theoretical analysis to visualization with the help of the Building Information Modeling (BIM) tool. Construction site characteristics vary widely depending on the size and settings. Nevertheless, the components of the site (i.e. workers, materials, equipment, etc.) have similarities in features. Therefore, the concept of site hazard modeling follows the grouping or zoning of the site layout constituents having relevant attributes. A recent study proposed a Discrete Event Simulation (DES) model for improving construction site safety and productivity in real-time for static and dynamic site components. Another study used various scheduling intervals to develop cell-based site layout optimization. A cell-based simulation model has also been used to derive the congestion and productivity of an existing site layout using Agent-based Modeling (ABM). On the other hand, Shen and Marks have utilized equipment footprints to determine hazardous zone boundaries in a construction site.

In recent years, accessibility in construction and/or transportation has also been gaining more and more attention from researchers, Transportation Research Board, National Academies, state DOTs, U.S. DOT, ASCE, and the private sector to eliminate adverse impacts experienced by underrepresented and marginalized groups including construction workers. The existing literature appears fragmented and does not provide a system-wide perspective to assess and measure accessibility in construction, both general and accelerated. Although ABC is expected to make construction more accessible and socially equitable, no such assessment technique is currently available. Thus, the proposed decision support tool will be first in the country in the ABC area, providing a system-wide understanding of construction accessibility.

The proposed agent-based modeling (ABM) techniques support the inclusion of complex agent-agent interactions and behavior into the analysis of near miss incidents and accessibility issues associated with ABC projects. Since modeling the construction environment involves construction workers’ attributes, ABM has been preferred by a wide range of researchers to identify latent contributing variables of construction hazards. For example, it has been explored to determine the influence of safety behavior of workers in workplace productivity. ABM models have also been used to investigate a comparative analysis of the efficiency of different types of safety investments. On the other hand, some researchers have identified the impact of the worker-management relationship on the overall safety behavior of workers. While these studies provide insights on integrating human behavior into the modeling framework and reveal the potential of ABM, however, in many instances, the incorporation of construction site components (such as movement of workers, equipment, materials, among others), environmental (such as weather conditions), and other characteristics are missing, especially in the context of ABC.

Objective:

The objective of the proposed research is to utilize an agent-based analytical modeling framework to conduct behavioral experiments and identify near-miss incidents and accessibility issues at an active ABC site. The I-235 Broadway Extension Corridor Widening project near Oklahoma City that recently used ABC method will be considered as a case study in this project.

Scope:

Task 1 –   Develop a Comprehensive Agent-based Modeling (ABM) Framework

The research team will develop a comprehensive agent-based modeling framework to integrate different components of an active ABC site as shown in Figure 1. The I-235 Broadway Extension Corridor Widening project near Oklahoma City that recently used ABC method will be considered as a case study in this project. Moreover, the ABC-UTC project database (https://utcdb.fiu.edu/search/) will be explored to collect relevant data on ABC site details from recently completed ABC projects. To incorporate behavioral characteristics of diverse background of workers and others on site into the ABM framework, five broad classes of information will be considered: (1) work related information (i.e., organization type, role, safety culture); (2) risk perception (i.e., risk rating); (3) type of project (project location, scope, duration, etc.); (4) important project execution variables (i.e. construction schedule); and (5) socio-demographics (i.e. age, gender, race, education, language). In order to develop a meaningful and realistic ABM framework, the research team will also consider national databases such as Bureau of Labor Statistics, Occupational Safety and Health Administration (OSHA) among others.

Task 2 – Identify the Scale and Scope of the Safety Performance to be Inspected

Once the proposed ABM framework is developed, appropriate scales and scopes of analyses will be selected before conducting the agent-based behavioral experiments in Task 3. For example, we will conduct a systematic literature review of recent ABC studies and/or projects where different safety issues or near miss incidents have been reported. In this step, we will also identify critical safety performance metrics to be analyzed through agent-based behavioral experiments in the next step.

Task 3 – Perform Agent-based Behavioral Experiments and Analyses

The research team will use traditional ABM software (NetLogo, AnyLogic, etc.) that are publicly available to run different agent-based behavioral experiments for the selected ABC sites. The input data obtained from Task 1 will be used to perform these experiments based on the scales and scopes selected in Task 2. Several near-miss scenarios will be considered to identify different resilience, safety, vulnerability, and accessibility metrics.

Task 4 – Compilation of Results and Reporting

The results and findings of the network experiments will be reported in a manner consistent with the existing protocols followed by ABC-UTC. The proposed tool will be developed in such a manner that for practitioners can readily use in the field.

Research Team:
Principal Investigator: Dr. Arif Mohaimin Sadri
Co-Principal Investigator: Dr. Musharraf Zaman,
Student Assistant: TBD