How Process Industries Can Catch Up in AI

Most mining, chemical manufacturing, and oil and gas companies have failed to adopt AI, despite its proven track record.

Applying AI to heavy industry’s most vexing problems has resulted in improved efficiency, throughput, and other metrics, along with waste reduction – all to the tune of 15% or more. Continuous process companies can no longer ignore this industry advantage, unless they want their competitors to unlock value while they fall further and further behind.

“The good news is that the relative lack of adoption means AI still offers significant untapped value – if companies take the steps necessary to implement it.”

To capitalize on the AI advantage, your company must gather pertinent data, put that data within reach of the right people and convince the human workforce to embrace change.

You’ve got to get your data in front of the right people if you want to benefit from AI.

Your operators may be the best and most experienced in the industry, but they can still benefit from the bird’s-eye and granular views that only data can provide. Too many process companies collect only sparse, low-quality data that’s spread throughout the company, and changing one sensor or machine changes the whole system. To benefit from your data, it must be integrated and available to the people who will benefit the most from seeing it. Data management is critical.

“The scientific relationships between process inputs and outcomes guarantee that a specific intervention will impact results – for better or worse – in ways that can be quantified and tracked over time.”

Continuous processes can be trickier to track than batch processes, but they also allow for continuous data and quick, conclusive experiments that lead to real improvement. Combine that with cloud computing (which keeps real-time data available and all in the same place) and industry 4.0 technologies, and decision-making will improve exponentially. Just be sure you’re collecting data on the right factors: raw material characteristics, control levers, and other external variables, with overall process KPIs among them.

Sometimes the biggest obstacle to AI-enhanced operations is the resistance of the human workforce. Ease their minds if you want to attain AI success.

Your process operators and engineers are experts, and their methods are usually deterministic, with a meticulous understanding of the physical systems involved in their work. It makes sense that they’d be suspicious of machine learning, which leans away from the precise and toward probabilistic.

“Probabilistic models will be sometimes right and sometimes wrong – but overall, more right than wrong.”

Human decision-makers need reassurance that AI-informed changes are iterative and may start imperfectly, but will likely lead to overall improvements over time. Most importantly, they need to know they won’t be punished for implementing imperfect models – imperfect changes are part of the process. Workers may also feel concern that your company wants to replace the human workforce with machines. It’s vital that they embrace new ways of working, but they’ll only be able to do so if they’re reassured that AI will enhance human understanding of the options, not replace human decision-making entirely.