Using Big Data to improve mill maintenance is a growing opportunity—and a significant challenge. Today’s mills can dramatically boost their maintenance performance and improve operational efficiency through data analysis, but managing that data can be daunting and sometimes impossible. For example, management at one mill I worked with felt they had to stop collecting data on their equipment because too much equipment had already failed, and they were not equipped to find and solve more problems. Other mills don’t have the resources or training to properly analyze data that can improve maintenance practices.
Fortunately, there is a solution to the “data paralysis” mills are experiencing. Modern data analysis software can largely automate the mill data collection and analysis process; properly analyzed data can, in turn, fuel reliability-centered maintenance (RCM) programs that not only predict critical equipment failures but also provide mills with the tools they need to prevent failures from occurring in the first place.
The best place for mills to start is by realizing that all equipment begins to fail as soon as it is installed. In addition, some equipment will wear out over time, or it can fail randomly. Given this reality, mills must find a way to predict and prevent potential failures, prioritize and triage key failure indicators, and take action to avoid negative impacts.
For example, one North American paper mill producing about 2,400 tons per day of fine paper had an excellent traditional vibration analysis program with excellent in-house vibration technicians; however, given the amount of rotating equipment in the mill, vibration techs could only take readings from each piece of equipment on six-week intervals. Nevertheless, a six-week check was more than adequate for most traditional failure modes the vibration crew was tasked with finding because most of these failure modes—such as normal bearing fatigue and pump impeller wear—provide up to six months’ worth of warning.
However, after our company, Strategic Maintenance Reliability (SMR), began working with the mill, using our RCM2™ engineered maintenance strategy, we discovered a significant problem. RCM2™ determines when an asset will no longer meet its function and provides a solution to identify failing assets before their performance falls below desired levels. Unfortunately, during one of our RCM2™ analyses at the fine paper mill, we identified a few failure modes that provided minimal warning—fewer than seven to 10 days with traditional vibration analysis.
One particular challenge was a failure mode associated with a coupling between a motor and a vacuum pump that provided a one-week warning before failure. The failure mode would have required the vibration crew to take readings every three to four days. The vibration crew supervisor, who was also the site’s PI / Process Book champion, knew that the vacuum pump train had on-line vibration monitoring that was not being actively utilized because analyzing the volume of data being generated and stored was not practical.
At the time, the supervisor was part of an RCM2™ team that had analyzed the vacuum pump train and identified the failure mode. As a result, daily polling for the on-line vibration readings was set up. The supervisor programmed calculated indicators in the mill’s analysis software that looked for a 5% increase in vibration readings daily and another indicator that looked for a 2% increase over a six-week rolling period. Both indicators triggered automatic e-mail notifications if thresholds were exceeded.
Over the next month, the day-to-day indicator went into alarm, and e-mails were generated. The maintenance staff responded with prepared standard jobs from the CMMS library to repair a problem with the coupling during a scheduled clothing change. If RCM2™ had not been deployed, the potential failure indicator would have gone unnoticed and produced 16 hours of unscheduled downtime on the paper machine.
This is one example of how Big Data can be used effectively. Still, for that to happen, maintenance teams must understand how and where to apply the data, run software to perform the analysis, and take appropriate action. That’s where SMR’s partner, Pulmac, comes in.
Pulmac’s Data Analytics and Its Easy Integration with RCM2™
Pulmac has partnered with Seeq®, an advanced analytics application, and together they provide an automated, detailed analysis of mill maintenance data that would take engineers weeks or months to prepare manually. The data analyzed by Pulmac is then integrated with RCM2™, creating a maintenance development program. Once we know the strategies and indicators to tie to each asset, the Pulmac/RCM2™ solution can provide detailed, effective recommendations.
Pulmac’s high-level data analysis using Seeq®, combined with the RCM2™ solution, can provide paper mills with the experience level that once required a crew of seasoned, 35-year employees. In addition, data integration using Pulmac’s analytics team means that most solutions can be handled programmatically (automatically via advanced software) and not biologically (using manual data analysis by teams of people that many mills cannot afford). This helps mills move away from investigating failures via root cause analysis and into a proactive prediction of pending failures, helping avoid costly unplanned downtime. As a result, reliability engineers, superintendents, supervisors, managers, planners, and schedulers can meet weekly to effectively plan, schedule and execute corrective work on a controlled basis.