Rockwell Automation Techtalk ///

Connecting marketers to the World Oil audience What is this?

World Oil TechTalk allows marketers to connect directly with the World Oil audience by enabling them to create content and participate in the conversation on the World Oil website. Each TechTalk article is produced by the marketer. For more information on TechTalk packages, click here.

Computers at the Wheel? See How AI Can Improve Your Equipment Maintenance Strategies

When you hear the words “artificial intelligence,” do you conjure up images of science fiction movies or robots taking over the world?

You’re not alone. Wondering what AI really means? Or what “agents” are? Companies like yours are already taking advantage of this technology to improve operations. Remaining competitive in the future depends on it.

How Can AI Help Your Pipeline Operations Stay Competitive?

AI is comprised of intelligent agents, also referred to as simply “agents,” which are autonomous pieces of software that take in information from sensors, detect patterns and anomalies, and then use that information to take action by requesting human intervention or modifying operating parameters.

There are tons of business applications for agents. Pipeline companies have always struggled with equipment maintenance as they (by nature) operate across large distances, and often in remote and uninhabitable areas.

Equipment maintenance is a natural fit for beginning to implement agents and AI. Agents can monitor performance of rotating equipment such as electric motors, gas generators, pumps and compressors; balance-of-plant equipment; and in some cases, even performance of the pipeline itself: measuring flow, temperature, pressure, and density of the pipeline contents.

Consider this:

  • 85% of equipment fails despite calendar-based maintenance (Boeing study)
  • One-third of all maintenance dollars are wasted through ineffective maintenance management methods. (Maintenance Fundamentals 2nd Edition – R. Keith Mobley)

What an opportunity for improvement!

Without the time and expense of physically sending a professional to remote corners of the globe to monitor performance and maintenance needs of a piece of equipment, an agent can be programmed to detect issues with equipment located anywhere in the world before failure occurs. Once a potential issue is detected, the agent can alert your staff to act.

What is an Agent?

There are two types of agents frequently used for predictive maintenance: failure agents and anomaly agents.

Failure agents are programmed to look for patterns in data that are already associated with a known failure mode. When this pattern is detected, the agent alerts your team of impending failure and prescribes the appropriate maintenance activity.

Anomaly agents look for data that is deviating from that associated with normal operations and alerts your team when a new pattern is found so it can be investigated.

Creating failure agents and anomaly agents is a fairly simple process:

  • Determine the data needed
  • Install sensors to capture this data
  • Configure failure agents to recognize known failure patterns
  • Configure anomaly agents to recognize abnormal data patterns

Predictive Monitoring to Ensure Peak Performance

While humans can identify simple and dramatic patterns using trending and historical-analysis tools, pipeline systems are often too complex for human pattern recognition to detect issues before failure occurs.

Considering this, pipeline companies around the world are turning to mathematical models for predictive analytics, and some are even implementing machine learning tools to identify when the pipeline is operating abnormally and help ensure continuous peak performance.

Want to learn more about how AI can shift your maintenance strategy from calendar-based to predictive and help you enhance pipeline safety and operational efficiency?

I’ll be hosting a webinar on May 31, 2018. Sign up here! And check out our whitepaper (PDF) on the ways to best leverage smart equipment in your pipeline operations.

More From This Sponsor ///More

Comments ///

comments powered by Disqus