Using vibration analysis in predictive maintenance

Production and maintenance engineers are constantly seeking to maximize the performance of their machines and minimize downtime. Predictive maintenance is increasingly being used to achieve these goals. As vibrations in equipment can indicate problems, collecting vibration data is key to this approach.

Vibration analysis allows engineers to assess the condition of devices, such as pumps and motors. By continuously gathering data on vibrations, engineers can predict when equipment needs maintenance. Vibration data is easy to gather and is very effective at identifying problems in equipment that use motors.

What is predictive maintenance?

Predictive maintenance is designed to assess the condition of a piece of equipment and determine when maintenance tasks need to be performed. The equipment’s status is assessed using periodic or continuous condition monitoring, based on non-destructive techniques such as acoustic, infrared and vibration analysis.

The predictive aspect of the method reflects the goal of predicting how the device’s condition will change over time, based on statistical techniques. The aim is to conduct maintenance on the asset at the most cost-effective time to ensure it continues to perform as required, while minimizing interruptions to production.

As such, predictive maintenance is a condition-based maintenance process. This compares to preventive maintenance, a time-based method dictated by pre-set maintenance intervals that pay little attention to the actual health of the machine. As these maintenance tasks do not necessarily align with production schedules, they can be disruptive and thus lead to losses.

Completing preventive maintenance to a fixed schedule also runs the risk of parts being replaced before they need to be, leading to extra costs. Working on a machine unnecessarily can also cause issues due to the danger of parts not being correctly replaced or that components become misaligned. Although preventive maintenance can be easier to plan, it can also lead to using more time, parts, and money than necessary.

Benefits of predictive maintenance

Many industries such as food and beverage and water and wastewater treatment rely on continuous operation of critical assets such as motors and pumps. These devices need to work to ensure customers get the food products or water supply services they require. Failure to supply can lead to these companies suffering financial penalties or legal sanctions.

A predictive maintenance programme seeks to avoid these consequences by eliminating unexpected failures and therefore unplanned downtime. When downtime is not planned, people and assets are used in an ‘emergency mode’, leading to inefficiencies. Maintenance technicians are devoted to fixing the failed asset and are not available to conduct planned maintenance on other machines.

If the asset was not expected to fail, there may not be spares for the machine components in question. Although there are suppliers who can supply spare parts quickly, such as Farnell, this introduces inevitable delays in getting the machine back into working order. In contrast, a company may have too many spare parts in stock for some machines, tying up working capital unnecessarily.

Predictive maintenance gives insights into which machine and which component on the machine is likely to fail and when. This allows maintenance staff to investigate the machine’s condition more effectively, schedule maintenance tasks to comply with production schedules and conduct any repairs before the machine fails.

A properly conducted predictive maintenance programme will dramatically reduce or even eliminate unplanned downtime caused by machine failures. It will also help improve the use of the workforce, allowing their tasks to be scheduled more rationally and effectively. A major benefit is the increased production capacity that can be gained. For example, if a machine is only taken out of service when absolutely necessary rather than according to the diktats of a prearranged schedule, available time to do useful work increases.

Similarly, moving away from preventive maintenance and toward a culture of predictive maintenance can help eliminate routine tasks that are adding no value, reducing overall maintenance costs. With a better insight into the actual condition of the machine, it can be serviced in the way it requires, increasing its lifespan.

These benefits can be quite significant. In Plant Engineer’s Handbook, R. Keith Mobley (2001) cites numerous benefits of predictive maintenance including:

• Maintenance costs – down by 50%
• Unexpected failures – reduced by 55%
• Repair and overhaul time – down by 60%
• Spare parts inventory – reduced by 30%
• Mean Time Between Failures (MTBF) – increased by 30%
• Uptime – increased by 30%

For a typical manufacturing plant, a 10% reduction in maintenance costs can produce the same financial benefit as a 40% increase in sales.

The importance of vibration analysis in predictive maintenance

Vibration analysis for predictive maintenance monitors the vibrations of machines such as motors. By analysing these vibrations, engineers can gain insight into the condition of the machine and predict of when it might fail. This allows them to carry out maintenance to prevent the possible failure and extend the machine’s working life.

Some industries are particularly reliant on rotating machinery and the equipment they drive, such as pumps and compressors. The food and beverage industry for example, uses many compressors, chillers, and pumps during processing.

The water industry needs to avoid downtime in its pumps to maintain water quality and supply adequate water pressure to customers. Oil and gas facilities make use of many motors, which need to be reliable to enable the company to meet its supply commitments.

As rotating components are critical to many types of industrial equipment and are often part of bottleneck processes that cannot be avoided, motors are frequently the cause of unplanned downtime due to failures if not maintained correctly.

A major cause of these failures is wear and misalignment in components such as bearing races, gear wheels and shafts. As these components move, a lack of lubrication can cause them to clash against each other rather than slide smoothly, causing damage that leads to large or unusual vibrations. For example, a rutted roller bearing race can create movement each time a bearing travels over the damaged area. Left unchecked, these vibrations can lead to further damage and ultimately failure.

Analysing and understanding these vibrations and gaining insights into what they mean for the performance and future health of the motor, is the basis of predictive maintenance.

As well as the motors themselves, vibration analysis can be used to monitor other parts of a process, machine or drive system, including bearings on conveyors, gear boxes, drive shafts and free wheels. Vibration monitoring is also used to monitor the structural health and safety of infrastructure assets such as bridges, pipes, and turbine blades.

In addition to vibration analysis, there are several other options for monitoring motor health. One of the more prominent methods is Motor Current Signature Analysis (MCSA). The technique senses an electrical signal that has current components. The current signals are processed to obtain a frequency spectrum for the motor. If there is a fault, this frequency spectrum will differ from that exhibited by the healthy motor.

MCSA can be used to diagnose several types of motor fault, including bearing faults and broken rotor bars as well as inconsistent air gaps between the rotor and stator.

Despite the usefulness of MCSA, continuous vibration analysis systems using fixed sensors are easier to deploy and use. They can be used readily by untrained personnel, offer maintenance more time for planning, and can be integrated with factory automation systems.

Vibration analysis data is vital for predictive maintenance. The vibration data can be analysed using various techniques to give different insights into the motor’s condition. For example, time domain analysis looks at metrics such as RMS acceleration, crest factor, velocity RMS and displacement RMS. RMS acceleration is a popular and useful metric for vibration health monitoring, while an increasing crest factor can indicate bearing failure.

Velocity RMS is the main metric tracked over time to monitor vibration health and inform predictive maintenance programmes. Displacement RMS is useful in assessing rotating equipment, as unbalanced shafts and other components can cause significant displacements.

To analyse the many vibration frequencies that a motor or other machine can exhibit, Fast Fourier Transform (FFT) algorithm is used. The technique decomposes the signal into all its constituent frequencies, converting it from the time domain into the frequency domain. This gives investigators a deeper understanding of the vibration profile and thus more clues about the condition of the system under test – in fact, most vibration analysis will be done in the frequency domain.

Using vibration analysis in predictive maintenance

Capturing vibration data and using mathematical techniques to analyse it to gain insights into the health of the motor or other system components are just the first stages. A fully operational and useful predictive maintenance programme needs these activities to be integrated into a management scheme that can use this information effectively.

This is the role of a Computerized Maintenance Management System (CMMS). This takes the form of a software package that maintains a database tracking maintenance operations. As such, it helps a company organise its maintenance activities effectively to achieve the goals of maximum machine productivity, extended life, and minimum costs. It does this by helping maintenance workers keep track of which machines require maintenance tasks, and where tools, measurement equipment and spares are located.

It also helps maintenance managers calculate the costs of repairing a machine following a breakdown as opposed to conducting preventive maintenance, allowing the most effective use of resources. A CMMS also records data to keep track of assigned and completed tasks.

An example of a CMMS is the eMaint software from Fluke Reliability. eMaint is a powerful asset reliability platform that helps organisations increase uptime through the easy integration of maintenance tools and software.

With eMaint, users can establish asset hierarchies, track the maintenance history of their assets, and prioritise work. Based on this, the solution can be used to assign work orders and generate work requests.

Integrating with condition monitoring tools such as vibration sensors, eMaint can be used to define the boundaries for equipment operation, import readings from instruments and sensors, and display graphs of the results. Dashboards and reports allow maintenance engineers to analyse trends and make decisions based on data. Predictive maintenance is also enabled through the solution’s ability to visualise asset data easily in different ways, such as schematics or a site map.

Managing and tracking spare parts, assigning predictive maintenance tasks, tracking inspections, and allowing access on mobile devices are all features offered by eMaint.

Vibration sensors, such as the extensive range available from Farnell plus a CMMS package by eMaint, form a complete predictive maintenance solution, helping companies protect their assets while ensuring maximum productivity and efficiency.