Smart Weapons for Maintenance

Maintenance practices and philosophies have evolved greatly throughout the years, driven by the need of industry to reduce maintenance costs, to increase productivity through reduced downtime, and to improve product quality.

Originally, up until the 1950s, ‘run to failure’ was the dominant maintenance philosophy, and the resulting high costs of secondary damage, production downtime, spares inventory and overtime labour were considered to be just the cost of doing business.

Next came preventive or calendar-based maintenance, where overhauls were performed and parts were replaced at regular time intervals, whether required or not, in order to avoid machine failures and control downtime. This flawed concept often results in the needless overhaul of machines which are in good condition. Additionally, there is a high probability of equipment failure immediately after an overhaul, due to human error in the rebuild process and poor lubrication on start-up.

Rotating machinery will usually provide some warning signs prior to failure, such as changes in temperature, in vibration level and pattern, in motor amperage and changes in the lubricant. With regular periodic monitoring, some of these signs can be detected well in advance, and then trended to predict when failure is most likely to occur. This makes it possible to order parts and schedule a repair at the optimum time, to correct the problem before the failure occurs.

Many organizations in a wide range of different industries have begun to implement this predictive, or condition-based, maintenance philosophy in order to extract the greatest value from their equipment assets. This approach allows them to strike a balance between the production benefits obtained from running machines as long as possible, the risk of failures, and the resulting costs in secondary damage and unscheduled downtime.

For a condition-based maintenance approach to be successful, data regarding the condition of the equipment must be collected at regular intervals. Most condition monitoring (CM) programs are based primarily on vibration data obtained by personnel who follow specified routes through the plant. Each route covers a specific group of machines and is repeated on a regular time interval.

Using this route-based method, there is typically a time lapse of 30 days or more between consecutive measurements taken on any given machine. While this is usually sufficient, in some cases it is not enough. Production-critical machines with no backup, machines where a serious fault has been detected, and machines with intermittent vibration problems are just a few examples of situations that need to be followed more closely.

In these cases, it is beneficial to install an online monitoring system that can continuously record and evaluate machine data. When the subject of online systems comes up, many people still think of traditional protection systems designed to shut down machines if vibration exceeds pre-determined limits. While these systems will always play an important role — guarding against catastrophic failures of critical equipment — a new generation of CM systems has evolved to fill many different roles, with far more extensive data analysis capabilities. These ‘smart weapons’ in the war on unscheduled downtime and production losses can be adapted for a particular application, and to target specific problems.

Case history

At the Domtar fine paper mill in Windsor in Quebec’s Eastern Townships, the experienced and innovative predictive maintenance team have implemented a unique strategy and application for an online system. Thanks to its long-established route-based predictive maintenance program, the team was often able to detect machine faults that were still in their early stages.

The next question the team members often faced, especially for production-critical machines, was: “Can we successfully operate this machine through to the next scheduled shutdown?” Thus, they began to see a need for some way to closely monitor the status of these ailing critical machines through to the next planned period of downtime.

Their innovative solution to this problem? A portable online system. They combined an advanced online condition monitoring system with a panel PC for on-the-spot viewing and trending of data, and mounted the combination on a rolling cart that can be wheeled up next to practically any machine in their extensive mill complex, and set up to start monitoring within minutes.

The system that they selected for this task has inputs for up to eight vibration sensors, as well as three inputs for 4 mA to 20 mA signals, and two each for displacement sensors and digital inputs. The unit also has inputs for RPM and Key Phasor, and a digital output. Although the system has both Ethernet and serial interfaces, in order to avoid the inconvenience of temporary cable runs, or wireless data transmission limitations, a memory stick is used for periodic data transfer to the network.

The system can easily be programmed to extract peak vibration levels from frequency bands of fully-adjustable width, centred on any chosen frequency. Also, alarm levels in these bands can be adjusted independently of the rest of the base band.

Since Domtar uses portable data collectors to perform the initial problem diagnosis, the source and cause of the vibration (e.g. a drive-end bearing inner race fault) will typically already be known, prior to calling in the online system. This provides the vibration analysts with the opportunity to configure the online system’s narrow-band measurements and alarm levels to the specific frequency or frequencies where the problem is found.

Order spectrums (vibration spectrums with the frequency expressed in multiples of the machine’s running speed) allow the defined frequency bands to always follow the problem as machine speeds change. In cases where the load and speed of a machine fall into two separate and distinct operating states, these states can be defined in the system. Once this has been done, the system will continually segregate measured data into the appropriate operating state, based on the speed and load at the time. It also then becomes possible to define different alarm levels for each state.

Typically, when there are frequent large variations in load and speed, the resulting variations in the frequency and amplitude of the vibration make vibration trends difficult — if not impossible — to interpret. However, due to the separation of data into different operating states, data for each state can then be trended separately, thus providing meaningful trends and facilitating analysis.

In addition to the role described above, this system has also been used by Domtar to troubleshoot intermittent problems with pump cavitation and paper machine drive system vibration.

Other similar systems are used around the world for applications as diverse as monitoring wind turbines, the drive systems of ocean-going ships and the drilling rigs on offshore drilling platforms. These systems have a multitude of communications options currently available, such as serial (PPP), Ethernet, ModBus TCP and RTU; as well as an ever-growing range of wireless solutions for short-, medium- and long-distance communication.

The potential for applications of such systems is vast and wide-ranging, and many exciting possibilities remain to be explored.

Tom Burke, eastern Canadian sales manager, and Ted Pater, application engineer, are with Hyatt Industries, Vancouver, B.C., a supplier of Pruftechnik condition monitoring products. For more information, visit www.hyatt-ind.com or use the reply number below.