01 July 2003
Knowing plant's assets, predictive maintenance allow users to reap big benefits.
By Tony Chambers
Arjo Wiggins is a global manufacturer of paper products. Its Stoneywood Mill in Aberdeen, U.K., came to a conclusion four years ago. They needed to reevaluate the plant's maintenance and asset management strategy.
In the time since they first analyzed the situation, the Stoneywood plant, which produces fine writing papers on five paper machines, has seen the long-term benefits of condition monitoring as a tool to help reduce downtime and maintenance costs.
For the past four years, Arjo Wiggins' Condition Monitoring Engineer Colin Angus has used the Rockwell Automation Entek Emonitor Odyssey condition-monitoring system. This system comprises software for logging and analyzing vibration and other plant data from portable data collectors.
Every three weeks engineers gather information from the data collectors, which routinely sample vibration data at 2,500 points in the papermaking plant. The Windows-based software downloads, stores, and analyzes this data to produce detailed information on the status of each item.
Through an array of diagnostic tools, the program identifies the mechanical fault corresponding to a vibration peak.
The plant's software provides a visual display of equipment status using mimic diagrams highlighting each data collection point.
In 1999, Arjo Wiggins began to look at how this system could monitor critical machinery even better, and how it could improve health and safety by avoiding the need to manually collect data from key points behind or underneath the large rotating plant. After analyzing a significant amount of downtime experienced at the Stoneywood plant in 1998, the company approved a plan to use an online surveillance system with the existing manual data collection regime.
As a result, in December 1999, the plant's Paper Machine 9 received the new system. In October 2000, the company expanded the system to cover Paper Machine 8.
The bottom-line objective of extending the existing predictive maintenance system using portable data collectors was to reduce equipment downtime, thereby increasing machine availability and output.
The system used permanently installed vibration transducers and data acquisition units to automatically collect condition-monitoring data according to a predetermined schedule.
Online surveillance is an intermediate condition-monitoring technique that sits between walk around, manual systems and more sophisticated continuous monitoring and protection installations that constantly monitor high speed critical machinery and provide an instant alarm if vibration exceeds safe limits.
Online surveillance provides a number of benefits to manufacturers. Users can regularly monitor machinery that is remote from the site; users can collect vibration data from locations that are too dangerous for engineers to enter, due to the proximity of rotating machinery or other hazards; and they can collect more data without making excessive demands on overstretched engineering and maintenance departments.
Online monitoring also improves the accuracy and repeatability of vibration readings compared to manual collection. You can adjust the schedule of data acquisition so the readings occur when the machine is running at the same speed. This ensures data is comparable, and the frequency of data collections from specific points can be quickly increased.
The system also enables one channel of data to constantly update any workstation, enabling remote investigation and analysis of any problems highlighted by the routine surveillance.
FINDING WEAK POINTS
Arjo Wiggins uses seven 16-channel data acquisition units to collect vibration data from 107 points on Paper Machine 9 (PM 9) and ten 16-channel units for 159 points on Paper Machine 8 (PM 8).
PM 9 is a twin-wire machine, and the system monitors top- and bottom-wire rolls; the first, second, and third press rolls; and various pumps, screens, and agitators. On PM 8, the system monitors all drive motors and gearboxes and the press roll section, including wire rolls and pumps, screens, and agitators.
Three types of vibration data are measured at each point: millimeters per second, g's Spike Energy (gSE) (gSE is a signal-processing technique developed by Entek to measure the spike energy produced by surface flaws in rolling-element bearings, metal-to-metal contacts, and insufficient bearing lubrication), and voltage of the accelerometer. The device also measures the revolutions per minute of the machine to ensure data is only gathered at normal running speed.
Detecting deterioration before the machine generates significant levels of vibration, gSE monitoring is a complement to vibration analysis, which detects other types of defects such as shaft misalignment or machinery imbalance.
The system gathers data on a six-hour cycle and logs it in the database every twelve hours.
All points have individual alarm levels set to protect machinery if they exceed preset levels or the trend in readings indicates they are likely to exceed them in the near future.
The 17 data acquisition units collect the data and download it via a local area network to the software, where it integrates with the information collected manually by engineers using portable data collectors. The combined data produces a complete picture of the health of the plant. This information appears on several PC workstations providing an intuitive, visual display of plant status at a glance.
Over a twelve-month period, the company fine-tuned the systems to reflect the nature of the various types of equipment monitored. They set alarm levels in the database and installed speed inputs on the driven rolls and main drive unit.
The plant achieved its objective of reducing plant downtime by installing the new system, Angus said.
"We have had some notable successes with the online surveillance system," Angus said. "On PM 9, for example, the Enwatch detected damage to the under wire roll bearings due to water ingress. The rolls were removed at a planned shut before the possibility of wire failure, which would have involved a twelve-hour shut if it happened during a production run."
The system also detected a problem with the top-wire box vacuum pump on the same machine.
"Increased vibration levels indicated that drive end bearing had started to deteriorate quickly," Angus said. "It was greased, but this made no difference to the readings, so it was changed at the next planned shut. This avoided around six hours of downtime that would have resulted if it had failed during a production run."
The system detected damage to under wire roll bearings on PM 9, preventing a failure that could have cost twelve hours lost production.
The system also detected a failure in the main drive bearings on PM 9 due to the rapid rise in vibration readings.
"These bearings failed on two separate occasions, the first being when the non-drive bearing suddenly began to fail after being in the position for a number of years and the second when the drive end bearing lost its fit on the shaft," Angus said. "In both cases the fault was between planned shuts, but due to the increasing magnitudes on each six-hour collection, it was decided to shut the machine at an opportune time. Because of this there was no cost saving, but disruption to production was kept to a minimum as we were able to carry out the repair through the day and also avoided secondary damage to the equipment."
On PM 8, the system detected something as simple as a loose locknut on a shaft, again preventing what could have been a major interruption in a production run.
"A sudden rise in vibration on a fan pump jackshaft was investigated at the planned shut," Angus said. "The bearing was OK, but the locknut had come slack and allowed the bearing to move on the shaft, which had sent the gSE reading into alarm. The locknut was tightened, and the readings came back down to normal. If this had caused a failure during production it could have led to four hours of downtime—or even more if the shaft had been damaged as a result."
The system also alerted maintenance engineers to a failed bearing on the couch pit pump on PM 8, again enabling officials to change the bearings at the next planned shut. Because the pump is under the machine in a confined space, a permit to work and multiple lock-offs of the machine are required for access. As a result, the job would have caused six hours of downtime if the bearing had failed during production.
As Arjo Wiggins found, the reduced downtime and cost savings are considerable, and the payback period on the cost of the installation can be as short as eighteen months. Equally important, the company minimized interruptions to production. IT
Behind the byline
Tony Chambers is an integrated conditioning monitoring specialist at Rockwell Automation. He is an electrical engineer with over twenty-five years of experience. His e-mail is firstname.lastname@example.org.
Return to Previous Page