A Simulator and Training Technique for Diagnosing Plant Failures from Control Panels

Area of Science:

• Human Factors Engineering
• Industrial Psychology
• Cognitive Science

Background:

• Traditional methods for identifying plant failures from control panel indications face simulation fidelity challenges.
• Existing signal identification techniques may bias problem-solving descriptions by requiring serial information processing.
• Operator stress during diagnosis is a critical factor influenced by environmental hazards and time constraints.

Purpose of the Study:

• To develop a simulator and training technique for plant failure identification from control panel indications.
• To address simulation fidelity issues related to input features, operator stress, and diagnostic time constraints.
• To evaluate the effectiveness of the developed training technique in improving diagnostic accuracy and reducing stress.

Main Methods:

• Development of a simulator using life-size projection of control panel mock-ups.
• Implementation of an adaptive cumulative training regime with a high volume of failure array exposure.
• Performance assessment through testing 24 hours post-training, analyzing diagnostic strategies and error rates.

Main Results:

• Trainees demonstrated high diagnostic accuracy, with most making no or minimal errors in post-training tests.
• The training regime allowed trainees to process significantly more failure arrays in a shorter time compared to operational settings.
• Subject reports revealed diverse diagnostic strategies, including pattern recognition and heuristic-based instrument checking.

Conclusions:

• The developed simulator and training technique are effective in enhancing plant failure identification skills.
• The training approach appears to reduce diagnostic stress by enabling performance with fewer discriminatory dimensions.
• Improved diagnostic robustness in dangerous environments is achievable by minimizing the dimensions required for accurate fault identification.