AI-powered analytics and automated condition monitoring are revolutionizing reliability. By shifting to continuous asset health monitoring, reliability teams gain real-time insights, predictive capabilities and the vast data sets needed to fuel AI engines.
Condition monitoring insights
- Artificial intelligence is transforming industrial reliability, but its success hinges on access to frequent, high-quality data — something manual, route-based methods can no longer provide.
- To meet this demand, plants are adopting automated condition monitoring systems that continuously collect and analyze data, enabling lean teams to leverage AI for predictive maintenance and improved operational performance.
- By shifting to continuous asset health monitoring, reliability teams gain real-time insights, predictive capabilities and the vast data sets needed to fuel AI engines.
The rise of artificial intelligence (AI) is generating a paradigm shift in industrial operations. The capabilities and potential of AI dominate headlines across the globe, with scientists, journalists and politicians speculating and exploring the many ways in which this new technology will change the world.
The process manufacturing industry is no exception to the influx of excitement around AI. Technology providers have gained a renewed enthusiasm for innovation, developing new use cases for neural networks and large language models almost daily. End users are leveraging those technologies for advanced analytics both at the edge and in the cloud, particularly in reliability, where AI tools can drive more effective predictive maintenance for better outcomes.
It seems as if there is nothing AI cannot accomplish, but that assumption is an illusion. There is one key thing AI cannot do: collect data from the field. For AI, data is everything. The AI agents that drive the best modern reliability technologies depend upon massive amounts of data — a need for hour-by-hour or even minute-by-minute data flow is not unreasonable for the most in-depth results.
This insatiable appetite for data comes just as process manufacturing plants are facing a key challenge. As retirements claim the most experienced plant personnel, fewer new workers are taking their place. Not only are small, lean teams unable to collect the near-real-time data necessary to feed the technologies driving competitive advantage, many can barely find the time to collect the monthly data necessary to simply maintain awareness of the overall health of their assets.
To address this issue, most plants are moving away from manual, route-based data collection in favor of automated online condition monitoring. A combination of both wireless and wired condition monitoring sensors — with a varying array of capabilities — empowers reliability teams to set whatever cadence of data collection best fits their unique circumstances. This type of solution also prepares these teams to implement the modern technologies that will drive operational excellence and competitive advantage.
Route-based data collection falls short
Historically, reliability teams have created scheduled routes to collect vibration data from assets on a regular cadence. A technician would visit each asset across the facility on scheduled days, with a handheld vibration monitoring tool and collect the data. They would then bring it back to the main office where data analysts would pore over the spectrums and waveforms, tracking and trending asset health to identify issues that required attention.
Such a system has been in place for decades because it worked then and it still works now. The problem is, though it works, it is inefficient and leaves gaps in visibility that can decrease performance while increasing costs, time to repair and the likelihood of unplanned downtime. If a plant has thousands of assets and just a few technicians — not uncommon in these days of staff shortages — they may spend the first three weeks of a month collecting data and then analysts may need to spend the last week analyzing that data.
By the time the reliability team has results, a new month has started and they need to begin the process all over again. Additionally, if only 1% to 2% of those thousands of assets have a problem — a common situation — teams have spent a lot of time traveling around the plant just to confirm that most machinery is performing as expected.
Moreover, most route-based data collection is set at once a month, which creates another potential problem. If a technician visits a gearbox on the first of the month, collects data with the analyzer, then moves on, it is possible that a fault will start developing minutes or hours later. In such a case, if that fault continues to worsen over the next 29 days, it might create a blind spot that could lead to increased wear and tear on assets or even an unplanned outage.
For teams looking to shorten the time to analysis and delivery of actionable information, AI tools might seem to be the perfect fix. However, AI tools rely on a foundation of continuous, near-real-time data, so individual data points coming in every 30 days simply will not be enough. AI can help, but only if the data foundation is already solid.
Online condition monitoring
Though it may not have received the attention that AI has experienced in recent years, online condition monitoring technology also continues to evolve. Many modern solutions exist, offering reliability teams a wide variety of options for collecting the more regular data necessary to free up personnel for higher value tasks and to feed their analytics technologies.
One key technology used by reliability teams for continuous data capture is wireless vibration monitoring. A wireless vibration monitor can be installed quickly and easily right at the asset, typically by a plant’s own technicians. Modern wireless vibration monitors collect spectrum and waveform data from balance of plant assets and send it directly to technicians and analysts from the field to eliminate the need for them to visit each asset and collect data. The most advanced wireless vibration monitors not only deliver raw data but also offer an intuitive health score for each asset, providing decision support to technicians of any experience level.
For assets needing additional insights, often those which rank higher in criticality, many teams are installing edge analytics devices to collect continuous data. Like wireless vibration monitors, edge analytics devices collect vibration data and send it — either via wireless signal or Ethernet cable — from the field directly to technicians and analysts. However, edge analytics devices also contain on-board machine learning for deeper analysis at the edge.
Using built-in analytics, these devices can automatically identify common issues, such as imbalance and under-lubrication, in the most common assets: fans, motors, gearboxes, pumps and other rotating machinery. Additional features, like a hardwired power supply to ensure continuous data collection and the ability to automatically shut down assets when a problem is detected, make edge analytics devices a critical element of any continuous condition monitoring strategy (see Figure 1).
Condition monitoring improved with data
Achieving optimal plant performance is dependent upon reliable access to good data. Some teams simply want to improve the efficiency of their reliability and predictive maintenance efforts. Others may want to implement the newest AI technologies to capture competitive advantage and strive for top quartile reliability. In either case, reliability teams need a way to regularly collect data that does not require their personnel to spend hours in the field manually gathering it.
With lean teams, tight budgets and increased competition, technicians’ and analysts’ time can be better spent actively improving and maintaining plant equipment, rather than inspecting it. Continuous condition monitoring empowers teams to do just that, unlocking modern AI technologies and providing reliability personnel with the insights necessary to perform at their very best.
