Predictive maintenance system delivers 8-month ROI

Mark Willliamson

Maintenance and operations staffs share responsibility for maintenance analysis, decisions and actions.

Many mills today have undergone a maintenance evolution, from corrective maintenance only to a modern, proactive philosophy where the servicing is done for well-founded reasons to prevent unexpected failures. However, both operations and maintenance people need the tools to measure, analyze and predict when maintenance is required and, equally important, when it is not required. When a liquid packaging board machine was rebuilt, the mill decided to upgrade the machinery and runnability analysis capability and to make it more user friendly for production operators as well as maintenance staff. Thus, these online tools would help them to implement their proactive maintenance philosophy. The extra monitoring capability was needed anyway since the rebuilt machine was much more complex with the addition of three coating stations.

Selection criteria

A new machinery condition and runnability analysis system was purchased so it would be available for the machinery startup. The online information would help to fine tune the machine and adjust speeds and settings to avoid critical resonance frequencies.

When considering alternative suppliers three main selection criteria were evaluated:

  • User friendliness. It had to be intuitive and easy to use. The Windows-based system was familiar and easy to navigate through.
  • Alarming methods had to be simple and clear enough so non-expert users like paper machine operators could see a problem and alert maintenance staff.
  • The system should be able to isolate the exact cause of a vibration problem in a complex mechanical system like a multi-nip press or calender.

The analysis system has over 300 monitoring points, from the stock delivery through to the reel. There are three PC-based user terminals – one in the machine's wet end control room, one in the maintenance office and one in the shift supervisor's office. The machine room terminal is linked to the system database server via fiber optic cables. The latter two are networked via the corporate information network. The system can be accessed from the supplier's offices for software updates or remote servicing.

Non-expert users

As specified, the new system allows a non-expert user to spot a developing problem. An alarm is first identified in a machinery group, like the press section; then it is isolated graphically to a certain roll, bearing or other mechanical component. On the same screen the user can see time trends and other analyses which show the severity of the problem. For instance, circle plots can clearly show roll corrugations or other roundness problems.

After this first-glance diagnosis of a problem, a maintenance specialist can then investigate more deeply into the problem using spectral analysis and other signal analysis tools. The user interface is shown in Figure 1.

Figure 1: User interface isolates vibration problems to a specific component. This display shows STA signals, spectra and circle plots of a third press bottom roll. Nip vibrations are also monitored by clicking on blue buttons.
STA analysis pinpoints root causes

In a complex mechanical system where everything is vibrating looking for a root cause is very difficult. But, with the STA analysis, one can determine the source. When a calender, press section or other multi-roll and multi-nip machine section vibrates, all of the rolls may have large vibration amplitudes. If a roll is corrugated or out-of-round it creates vibrations in adjacent rolls and the vibration is then transmitted. STA is used to determine which roll is the source of the vibration problem. Using a magnetic rotation sensor as a trigger, numerous signal periods are collected and averaged. All the vibration frequencies that are not specific to that roll's rotation period are eliminated. The remaining signal and its frequency characteristics are specific to that roll. In this way a roll in need of grinding can be positively identified and replaced.

This STA analysis feature has been used effectively at this mill, particularly when diagnosing press roll cover corrugations. The STA circle plots are a good tool for problem visualization. Once the problem and its severity are identified the roll cover is placed on a priority list for replacement and grinding at the appropriate time before paper quality suffers or a failure is possible. Roll vibration trends and spectra are saved in the system's database to track roll histories over several regrinding cycles. Figure 2 shows how STA analysis identifies roll corrugations.

Figure 2: A third press bottom roll was showing severe corrugations on the STA analysis circle plot.
Analysis of stock preparation pulsations

In addition to paper machine rolls, nips and bearings, the paper machine stock preparation system is monitored for vibrations and pulsations. For instance a headbox pulsation can be related to vibrations or pulsations in the fan pumps or screens. In one case more serious paper quality problems were avoided. In this case fouling on the rotor of the fan pump could be related directly to headbox pulsations. This pulsation showed up in the paper quality but it was under normal quality tolerances. The rotor was cleaned during a planned shutdown. If the problem were left untouched, it would have worsened, causing more severe quality problems and potentially an unplanned machine shutdown. Figure 3 show the user interface for the stock preparation analysis.

Figure 3: Pulsations caused by screens and fan pumps can be monitored and alarmed. These pulsations can have an effect on paper quality
Maintenance, operating staff work together

Avoiding damaging failures and lengthy production downtime is what the mill's predictive maintenance program is all about. The operating staff alerts the maintenance shift coordinator to abnormal conditions and alarms on weekends or overnight. Then the maintenance specialists can analyze the information in more detail and determine if the problem needs immediate attention or if it can wait until the next planned maintenance shutdown. If the problem is considered severe enough a work order is generated in the mill's maintenance planning system.

The mills predictive maintenance culture is shared by operating and maintenance staff. The operating staff trust in the judgment of the maintenance specialists. And that judgment is backed up by the information from the online system.

The machinery condition information is discussed in the morning meetings. Sometimes machinery vibration can be reduced by changing felt tension or nip loading. In some rare cases machine speed can be lowered to avoid a problem before the next shutdown.

Because of the high potential for damage the mill's staff has a high awareness of abnormal machinery condition analysis. That awareness extends to the lubrication system where wrong flow set points or clogged drains can cause bearings to run without lubrication. The mill is documenting many cases where maintenance has been properly scheduled and failure has been avoided by early diagnosis of the machinery condition. Figure 4 shows how a bearing outer race damage was detected. There are also records about paper machine operational conditions that could have led to potential damage. These could be detected only by the online system.

Figure 4: Dryer cylinder bearing damage, shown in the above picture was detected before failure. The damaged outer race is identified by a spectral analysis technique called envelope analysis.
Pinpoint needs

The mill keeps records of the scheduled maintenance actions which are based on information from the online system. These actions have been based on well documented information– not because of failure. If these problems were not detected ahead of time the lost time for repair would be four to five times as long. The return on investment for the system is calculated to be conservatively 8 months. The shutdown periods have been shorter because they can pinpoint the maintenance needs more precisely.

Video screens and pictures are courtesy of Metso.