How AI, ML and Industry 4.0 affect plant automation

Plant automation, combined with artificial intelligence and machine learning (AI/ML) along with Industry 4.0 concepts, can help companies make many positive strides, but cultural and technological challenges persist.

By Plant Engineering December 15, 2023
Courtesy: Kalypso

Respondents

  • Mithun Nagabhairava, senior manager, data science and AI, Kalypso, Austin, Tex.
  • Daniel Pender, global manager, markets and strategy, Rockwell Automation, Milwaukee
  • Joe Wagner, MS, field application engineer, Red Lion, Sacramento
Left to right: Mithun Nagabhairava, senior manager, data science and AI, Kalypso, Austin, Tex.Daniel Pender, global manager, markets and strategy, Rockwell Automation, Milwaukee Joe Wagner, MS, field application engineer, Red Lion, Sacramento.

Left to right: Mithun Nagabhairava, senior manager, data science and AI, Kalypso, Austin, Tex.
Daniel Pender, global manager, markets and strategy, Rockwell Automation, Milwaukee
Joe Wagner, MS, field application engineer, Red Lion, Sacramento. Courtesy: Kalypso, Rockwell Automation, Red Lion


Question: How can artificial intelligence and machine learning (AI/ML) be leveraged in plant automation to drive process optimization?

Mithun Nagabhairava: Enabled by the latest technological advancements, manufacturers are leveraging AI/ML to solve their complex production challenges and establishing autonomous capabilities. While an automation system is explicitly programmed to perform tasks in the absence of human intervention, an autonomous system can learn how to perform tasks on its own and even adapt to changes in a process or environment. Put simply, the autonomous system replicates the cognitive ability of humans to learn, make decisions, take action against those decisions and improve.

For instance, in a manufacturing process, the model may sense that a critical process parameter is drifting above or below an ideal set point. Based on learnings from what causes this variation and what actions can solve it, the model can evaluate a very large array of possible actions to take and recommend the action that will drive the best outcome.

Question: What role does data analytics and real-time monitoring play in enhancing plant performance and decision-making?

Daniel Pender: Real-time data and analytics mean that operations can understand the abnormalities and effects on plant operations faster. Making decisions faster means better plant performance and reaching desired outcomes faster. Operators can use this data and apply analytics to continuously learn to best optimize performance of a given plant operations. It can mean the difference between a plant-wide shutdown and a small hour repair.

Question: Have you experienced challenges getting information and operational technology (IT/OT) to work together on a project? If so, describe the challenges and solution reached.

Mithun Nagabhairava: We worked with a consumer packaged goods (CPG) manufacturer to utilize AI in identifying operational inefficiencies, bottlenecks, and areas for enhancement in their manufacturing. Consider conveyance systems that merge material flows from various sources and distribute them to multiple workstations or machinery with bypasses between each major section. While each OT control system can make tactical speed adjustments, it lacks awareness of IT data orchestrating material inflow, workforce scheduling, or machine loading.

By integrating IT data with OT data on inventory levels for each major conveyance section, AI can recommend effective bypass operations to alleviate bottlenecks, ensure consistent outflows, and address planned input imbalances. AI can bridge the IT/OT convergence, enabling strategic IT-level decisions about bypass operations while preserving efficient tactical OT-level operations.

Daniel Pender: Yes, converging IT and OT can be difficult when working together on a project. Traditionally, plant operations are set up with separate groups running the business systems (IT) and the plant control systems (OT). The procurement, upkeep, maintenance and setup of server and web-based features for both IT and OT are similar but getting them aligned is a challenge. I once helped to install a power management system where the customer IT group insisted on installing their chosen standard network switch. The customer purchased and installed the IT-approved network switches.

During engineering and testing, we realized that these switches were not communicating robustly enough to meet the electrical control requirements, and they had to be replaced with appropriate switches designed for the application. If the IT/OT teams had collaborated upfront with a network design and strategy, they would have had the proper data to choose which equipment and setup that worked with both of their systems.

Question: What cybersecurity measures should be in place to safeguard automated systems from potential threats or attacks?

Daniel Pender: Like traditional safety measures, a comprehensive cybersecurity strategy is necessary and can positively impact the effectiveness of OT environments. Different processes and operations have a wide variety of connectivity needs and acceptable risk tolerances, so there is no single solution to safeguard all automated systems. A cybersecurity preparedness and risk assessment is a critical first step. While more obvious remediation techniques and approaches include perimeters and access control, 24/7 monitoring and triage capabilities, and response protocols are critical for a comprehensive security strategy.

Question: Describe a success story in which AI was used to improve automation.

Mithun Nagabhairava: A leading tire manufacturer enlisted Kalypso to enhance their tire manufacturing efficiency leveraging AI to minimize downtime and increase throughput. We crafted a business case, harnessing the client’s existing data to review the disparities between anticipated and actual daily production and downtime attributed to splicing failures.

We developed a robust algorithm that adaptively corrected pressure and length settings to minimize the deviation from the desired splice length, and to reduce the skew between the left and right sides. The automatic adjustments by the algorithm have achieved levels of precision and accuracy in a way that a human operator could not consistently do.

The result was a 45% downtime reduction, a daily increase of over 100 tires per machine, and a total of 550,000 extra tires manufactured annually.