Real Time AI/ML Solutions – Roadmap, Challenges and Process Industry Applications

Real time AI/ML solutions are being incorporated into the Process Manufacturing industry framework. Operations are expected to have AI/ML solutions providing operational insights on profitability, sustainability, safety, and reliability. Key components to consider while planning for its adoption are:

• Clarity of purpose/objective
• Selection of technology
• Implementation approach based on technology and purpose
• Ongoing maintenance and use
• Value capture, baseline and tracking plan
• Identify and assign executive champion who will take ownership of the solution application and drive the necessary workflow transformation within the organization

Fitting the technology stack to the purpose allows faster and effective value realization. Too often the approach of letting technology strengths dictate the solution/use results in information not pertinent/actionable.

Hybrid models incorporating the best of what is known from deterministic first principles and combining it with AI/ML real-time models, along with Data Engineering and Decision Engineering, should form the basis of any real time AI/ML implementation program, especially the ones targeting process manufacturing with expected key result areas addressing Production, Efficiency, Reliability, Safety and Sustainability.

An effective implementation provides Augmented Intelligence to operations managers, engineers, and operators in the Oil & Gas Industry (Upstream, Midstream & Downstream), that would otherwise have taken years of operational experience.

ML / AI need ongoing maintenance and analysis. Post deployment MLOps support is rapidly being seen as a necessity. Combining this as an integral part of the implementation can lead to greater acceptance and understanding on an ongoing basis.

A quick review of implementation of such systems for a US Petrochemical manufacturer (Ethylene Complex), Gas Separation Units, Delayed Coker Unit, Olefins Complexes and VCM-PVC Complexes that did increase product yield, reduce asset downtime, increase production, and reduce emissions, without any major capital expenditure, within the first year will be done. Key takeaways from these sample projects will be provided.