Emerging Leaders: Graduate Student Research Expo

Construction projects possess a lot of uncertainties given the interdisciplinary nature of the industry. That being said, variance performance happens throughout the project lifecycle as the execution progresses. Therefore, in this research we are developing a warning dashboard that could be used to provide an early signal to project managers whenever there is one of the identified factors impacting productivity. Also, one of the functions of this dashboard is an imbedded AI algorithm that could forecast the productivity factor (P.F). Finally, it provides the user with a list of recommended strategies to mitigate each of the flagged factors in order to improve P.F. To achieve this, the research team collected 315 survey responses filled by a major construction company in North America. However, PF was reported by discipline; therefore, “Piping” was selected to filter the dataset. This narrowed down the dataset to 157 survey responses out of which 125 were used to train an AI model and the rest for testing. The survey was designed to collect the ranking of the factors impacting productivity and then matching it with the respective PF reported for this week. Thereafter, correlation analysis was performed to list the highly impactful factors which was then linked to the dashboard. Finally, a list of strategies was reported as ranked by the industry partners to mitigate the factors. This dashboard could be automatically updated each time a new dataset is collected. Also, it fills in the gap between the current conditions on site and their impact on productivity; therefore, it acts as a proactive control method on site to improve the productivity.

Due to the numerous conditions of construction operations at any site, the interactions between these conditions are highly complex and uncertain. To overcome this challenge, some research was conducted to integrate operations research with advanced mathematical models, with the use of control theory, to deal with the varying conditions of production patterns and enhance safety analytics. To overcome these limitations, a predictive analytic framework that will revolve around control theory models, machine learning and discrete event simulation will constitute the decision-making backbone to support safety and production planning and control decisions on-site with a clear and holistic approach. Detailed research will need to be undertaken for the predictive analytics framework to explore the effective integration of control theory models with either machine learning algorithms or discrete-event simulation models. This will enable evidence-based, knowledge-grounded construction of production predictive models smoothly integrated to management processes and that makes sense to decision makers. The predictive analytic framework will explore and adapt machine learning classifiers and other type of algorithms to identify and sort safety risks on-site. The safety and production dimensions of the predictive analytics framework fed with data gathered in real-time, or near real-time, will provide new decision-making capabilities to engineers on-site.

 Compared to completed buildings, fire safety on construction sites is more complex and has a higher risk, due to the dynamic nature of construction sites and the lack of a permanent firefighting system. Although this field is gaining increasing attention, the availability of well-defined frameworks to guide research and practical applications remains limited. To address this problem, a scoping review is conducted to systematically map the existing research in this area and identify gaps. An iterative search will be conducted in selected databases, including Scopus, Web of Science, and Google Scholar. Furthermore, grey literature will be collected by exploring the websites of national agencies and industry associations involved in fire safety on construction sites. Based on the extraction and analysis of information, the following questions will be answered: What are the features of fire on construction sites compared to the completed building? What are the challenges of fire safety management? Which challenges in this field have been addressed, and what are the existing gaps that still need to be explored?

Construction constraints are the factors that restrict, limit or regulate the progress of construction activities from achieving construction products within the agreed time, cost, and quality. While the initial discussion of constraint management (CM) can be traced back to the 1990s, constraint management has yet to reach sufficient implementation levels in most projects. This research aims to systematically gain an in-depth understanding of research trends and reveal challenges and potential solutions in this area. To achieve this objective, the authors explore the state-of-the-art overview of CM-related research by conducting a systematic literature review incorporating quantitative and qualitative methods. A total of 99 papers are collected using a structured data collection approach from Scopus and the lean construction-related conference, i.e., the IGLC (International Group for Lean Construction) conference, and then fed into a bibliometric analysis to construct science maps. Subsequently, an in-depth qualitative analysis combined with the quantitative analysis results is presented to provide deeper insights into the CM-related research topics. Towards the end, future research directions are proposed based on the in-depth analysis of review results. 

This study utilizes natural language processing (NLP), text mining, and deep learning methods to analyze 821 incidents and 12,024 inspection reports of a project from 2015 to 2018 from a construction project in Alberta, Canada. Each incident is linked to its respective past inspection, and sentence bidirectional encoder representations from transformers (SBERT) are used to generate embeddings and compute the similarity between incident and inspection descriptions. Root cause analysis (RCA) is conducted and the incidents are classified into process safety management (PSM), SMS, and human factor analysis and classification system (HFACS) frameworks. In addition, n-gram models are used to extract meaningful information and identify patterns from incident and inspection reports. The results of the study provide new perspectives on identifying the missing safety leading indicators in inspection reports with a promising application in risk analysis and building a safety early warning system.  By embracing a transdisciplinary approach, the research techniques applied in this study can be effectively applied in construction and various other industries, such as oil and gas. This transdisciplinary perspective allows for integrating diverse knowledge, methodologies, and best practices from different fields, enabling organizations to address safety challenges holistically and foster collaboration across disciplines.

This poster presents a safety management decision support system that utilizes early warning systems. The construction industry is known for its high rate of incidents, necessitating a proactive approach to prevent safety hazards. The methodology leverages data fusion integrating various data sources including project data, HR data, incident data, and safety indicators to drive a comprehensive analysis. The system incorporates an accident causation model to understand underlying factors contributing to incidents and employs early warning systems to detect and communicate potential risks. The decision support system aims to enhance decision-making processes by providing actionable information and proactive measures to mitigate safety hazards in construction sites. Implementation of the proposed system can lead to improved safety performance, reduced costs, and a safer work environment in construction sites. 

Construction-related businesses within the province of Alberta, Canada are economically hurdled by the changing workplace policies and economic effects brought by the efforts in handling the COVID-19 pandemic. These changes in workplace policies aim to reduce the spread and exposure of the airborne virus leading to a reduction in labor exposure hours since the emergence of the pandemic in 2020. However, it is uncertain whether the adjust-ed count of claims related to workplace injuries, illnesses, and fatalities in the construction sector was significantly reduced despite the decreased labor exposure hours. The compounded risks of uncontrolled construction hazards and COVID-19 exposure have adverse on the following levels: (1) Alberta’s economy due to the total amount of claims expenditure; (2) financial burden on construction-related businesses due to lost productivity and costs associated with hiring and replacement of an injured worker; and (3) the societal level where workers and their families suffer lost wages due to disabling injuries. The present study investigates the impact of COVID-19 on health and safety claims within the Construction and Construction Trades Sector of Alberta. Moreover, the main sources for incidents are re-examined while analyzing which demographic group is exposed to greater risks. This is analyzed by aggregating the prevalence claims rate (non-COVID claims) with incidence rate claims (claims due to COVID-19) from the Workers’ Compensation Board of Alberta dataset. The study aims to provide evidence-based in-sights that will guide investment decisions, prevention efforts, and proactive programs in reducing workplace injuries, illnesses, and fatalities.

Compared to completed buildings, fire safety on construction sites is more complex and has a higher risk, due to the dynamic nature of construction sites and the lack of a permanent firefighting system. Although this field is gaining increasing attention, the availability of well-defined frameworks to guide research and practical applications remains limited. To address this problem, a scoping review is conducted to systematically map the existing research in this area and identify gaps. An iterative search will be conducted in selected databases, including Scopus, Web of Science, and Google Scholar. Furthermore, grey literature will be collected by exploring the websites of national agencies and industry associations involved in fire safety on construction sites. Based on the extraction and analysis of information, the following questions will be answered: What are the features of fire on construction sites compared to the completed building? What are the challenges of fire safety management? Which challenges in this field have been addressed, and what are the existing gaps that still need to be explored?

This study aims to see if an Artificial Neural Network (ANN) trained on a company's incident database can accurately predict incident severity, based on details currently captured in near-miss reporting. By better understanding how near miss features correlate to potential incident severity, it allows insight into what details need to be captured in these near miss reports, which work divisions or activities are more prone to high-impact events, how these events can be better mitigated, and how these reports can be improved moving forward.