Mashrura Musharraf

A virtual experimental technique for data collection for a Bayesian network approach to human reliability analysis

Bayesian network (BN) is a powerful tool for human reliability analysis (HRA) as it can characterize the dependency among different human performance shaping factors (PSFs) and associated actions. It can also quantify the importance of different PSFs that may cause a human error. Data required to fully quantify BN for HRA in offshore emergency situations are not readily available. For many situations, there is little or no appropriate data. This presents significant challenges to assign the prior and conditional probabilities that are required by the BN approach. To handle the data scarcity problem, this paper presents a data collection methodology using a virtual environment for a simplified BN model of offshore emergency evacuation. A two-level, three-factor experiment is used to collect human performance data under different mustering conditions. Collected data are integrated in the BN model and results are compared with a previous study. The work demonstrates that the BN model can assess the human failure likelihood effectively. Besides, the BN model provides the opportunities to incorporate new evidence and handle complex interactions among PSFs and associated actions.

Assessing offshore emergency evacuation behavior in a virtual environment using a Bayesian Network approach

In the performance influencing factor (PIF) hierarchy, person-based influencing factors reside in the top level along with machine-based, team-based, organization-based and situation/stressor-based factors. Though person-based PIFs like morale, motivation, and attitude (MMA) play an important role in shaping performance, it is nearly impossible to assess such PIFs directly. However, it is possible to measure behavioral indicators (e.g. compliance, use of information) that can provide insight regarding the state of the unobservable person-based PIFs. One common approach to measuring these indicators is to carry out a self-reported questionnaire survey. Significant work has been done to make such questionnaires reliable, but the potential validity problem associated with any questionnaire is that the data are subjective and thus may bear a limited relationship to reality. This paper describes the use of a virtual environment to measure behavioral indicators, which in turn can be used as proxies to assess otherwise unobservable PIFs like MMA. A Bayesian Network (BN) model is first developed to define the relationship between person-based PIFs and measurable behavioral indicators. The paper then shows how these indicators can be measured using evidence collected from a virtual environment of an offshore petroleum installation. A study that focused on emergency evacuation scenarios was done with 36 participants. The participants were first assessed using a multiple choice test. They were then assessed based on their observed performance during simulated offshore emergency evacuation conditions. A comparison of the two assessments demonstrates the potential benefits and challenges of using virtual environments to assess behavioral indicators, and thus the person-based PIFs.

Human Factor Risk Assessment During Emergency Condition in Harsh Environment

This paper presents a quantitative approach to human factors risk analysis during emergency conditions on an offshore petroleum facility located in a harsh environment. Due to the lack of human factors data for emergency conditions, most of the available human factors risk assessment methodologies are based on expert judgment techniques. Expert judgment is a valuable technique, however, it suffers from vagueness, subjectivity and incompleteness due to a lack of supporting empirical evidence. These weaknesses are often not accounted for in conventional human factors risk assessment. The available approaches also suffer from the unrealistic assumption of independence of the human performance shaping (HPS) factors and actions. The focus of this paper is to address the issue of handling uncertainty associated with expert judgments and to account for the dependency among the HPS factors and actions. These outcomes are achieved by integrating Bayesian Networks with Fuzzy and Evidence theories to estimate human error probabilities during different phases of an emergency. To test the applicability of the approach, results are compared with an analytical approach. The study demonstrates that the proposed approach is effective in assessing human error probability, which in turn improves reliability and auditability of human factors risk assessment.

Human performance data collected in a virtual environment

This data article describes the experimental data used in the research article “Incorporating individual differences in human reliability analysis: an extension to the virtual experimental technique” (Musharraf et al., 2017) [1]. The article provides human performance data for 36 individuals collected using a virtual environment. Each participant was assigned to one of two groups for training: 1) G1: high level training and 2) G2: low level training. Participants’ performance was tested in 4 different virtual scenarios with different levels of visibility and complexity. Several performance metrics of the participants were recorded during each scenario. The metrics include: time to muster, time spent running, interaction with fire doors and watertight doors, interaction with hazards, and reporting at different muster locations.