Predicting behaviours, saving lives, with AI

Project Summary

Despite decades of training programmes, tightening of procedures and investment in safety cultures, the ability to proactively manage behavioural risk remains one of the greatest challenges facing challenging industrial sectors, not least the oil and gas sector. Behaviours and attitudes to risk can be different between shifts, different sites or offshore facilities or even the time of day. They are hard to quantify, hard to predict and, therefore, hard to address.

OPEX set out to develop a diagnostic tool that could provide key insights into the levels of behavioural risk of personnel and in doing so reveal an organisation’s unique behavioural “DNA”. This behavioural DNA would then be combined with a deep analysis of historic health safety environment (HSE) and asset data from the same organisation in order to provide a clear picture of the critical factors driving behavioural risk across that organisation’s offshore operations. This would then enable operators to target the key risks associated with their organisation’s natural disposition to risk, in turn allowing them to significantly reduce the likelihood of HSE incidents.

The tool utilised the OPEX’s approach to using information to spot patterns. Working with an operator, and combining the skills of data scientists and behavioural psychologists, OPEX identified and quantified levels of behavioural risk across a number of offshore assets and work teams and provided a set of remedial actions to manage the identified risk. The proof-of-concept project drilled down into specific factors which influenced the workforce’s behaviour risk profile – and how that in turn was related to its frequency of incidents.

Industry value:
This could significantly reduce the likelihood of HSE incidents.

Key results / Lessons learned:
The operator was able to get a better understanding of the levels of behavioural risk of the work teams across its assets. It was also better able to understand the critical factors influencing behavioural risk and their relationship with the occurrence of incidents. Remedial actions and interventions to manage the identified risk could then be applied. The operator also gained an increased ability to anticipate, predict and proactively manage the risks posed by human behaviours within its company and operations.

There were challenges, associated with sourcing quality data, availability of which can impact on critical schedules within an asset acquisition timeframe, where this insight is required upfront un order to quickly assess operational risk and optimisation opportunities.