AI and the Future of PHAs
Six months ago the United States government introduced a ‘DOGE’ policy to eliminate unnecessary bureaucracy. The implementation of this policy left much to be desired, but the basic concern remains: bureaucracy is expensive.
One of the roles of AI in coming years will be to slim down bureaucratic systems ― including the systems we use to manage process safety.
For example, a Process Hazards Analysis (PHA) is a bureaucratic procedure that is expensive. It requires valuable employees to spend days, or even weeks, in lengthy meetings discussing potential failure modes and risk levels for the systems that they operate. The incorporation of information and insights from other management elements such as incident investigations, is difficult, as is understanding the impact of new regulations.
AI could change all that. We could have a system on the following lines:
Currently, we update PHAs every three years or so. In the future the PHA will be updated every time there is a system change ― maybe every few minutes. For example, whenever there a change to the P&IDs or operating procedures, the PHA will be revised. If another facility publishes an incident investigation, or if the Chemical Safety Board publishes a new report, the findings will be incorporated into the PHA.
If a worker has a safety concern or insight, he or she will simply speak it into the AI.
There will still be a need for brainstorming sessions, but these will be much less time consuming than today’s meetings. They will be relatively informal.
Each update will include a full, dynamic fault tree analysis, and will write a new operating procedure, training program and inspection schedule, as needed.
Every morning the facility manager or OIM will look at the risk dashboard, alongside the production and cost reports. If he or she is not happy with the results, he will use his own AI to find out what is going wrong.
Management of Change will lie at the heart of this management system.
