Predictive Maintenance

What is predictive maintenance?

Predictive maintenance is a type of condition-based maintenance in which assets are monitored with sensor devices that supply data (i.e. vibrational frequency) about the asset’s condition. This data is used to predict when the asset will require maintenance to prevent equipment failure.

Predictive maintenance workflow

Note: Predictive maintenance workflow exmaple

Overview

Predictive maintenance (PdM) is the most advanced type of maintenance currently available. With time-based maintenance, organizations run the risk of performing too much maintenance or not enough. And with reactive maintenance, maintenance is performed when needed, but at the cost of unscheduled downtime. Predictive maintenance solves these issues. Maintenance is only scheduled when specific conditions are met and before the asset breaks down.

Organizations started using predictive maintenance around the start of the twenty first century. To monitor conditions of assets, organizations used a periodic or offline approach. A study that documents three PdM case studies from 2001 states: “The vibration measurements are taken periodically—one time per month in general—and vibration is monitored by comparing previous measurements to new ones.”

Today, a continuous or online approach is used to monitor conditions of assets. Remote monitoring is also possible by connecting an IoT sensor device to maintenance software. When specific conditions are met, a work order for an inspection is triggered.

How to implement predictive maintenance

Before predictive maintenance is implemented on the facility floor, ROI cases are presented to management. Maintenance staff and machine operators are also trained with how to use the PdM technology. After this happens, true implementation begins.

  1. Establish baselines. The maintenance team establishes acceptable condition limits for assets that will have sensors.
  2. Install IoT devices. The relevant sensor is affixed to the asset. For instance, a vibration meter is affixed to a mechanical asset with gears and a temperature sensor is attached to a boiler.
  3. Connect devices to software. The IoT device is connected to a CMMS or remote dashboard where data is collected and analyzed.
  4. Schedule maintenance. Inspections are automatically triggered by a CMMS when the condition limit is exceeded or the person monitoring the dashboard schedules the inspection manually.

Note: Different types of IoT Sensors by Monnit

Types of predictive maintenance

Vibrational analysis

Machine Speed: High | Machine Type: Mechanical | Cost: Medium

This is the go-to type of analysis for predictive maintenance inside manufacturing plants with high-rotating machinery. Because it’s been around longer than other types of condition monitoring, it’s relatively cost-effective. In addition to detecting looseness like in the example above, vibrational analysis can also discover imbalance, misalignment, and bearing wear.

Note: Using vibration analysis in predictive maintenance on a motor

Acoustical analysis (sonic)

Machine Speed: Low, High | Machine Type: Mechanical | Cost: Low

This type of analysis requires less money to implement and is used for low- and high-rotating machinery. It’s particularly popular among lubrication technicians.

According to an article by Machinery Lubrication, “Acoustic analysis is similar to vibration analysis; however, its focus is not to detect causes for rotating equipment failure by measuring and monitoring vibrations at discrete frequencies and recording data for trending purposes.

Instead, acoustic bearing analysis is intended for the lubrication technician and focuses on proactive lubrication measures.”

Acoustical analysis (ultrasonic)

Machine Speed: Low, High | Machine Type: Mechanical, Electrical | Cost: High

While sonic acoustical analysis borders on the line of proactive and predictive maintenance, ultrasonic acoustical analysis is solely used for predictive maintenance efforts. And because it can identify sounds related to machine friction and stress in the ultrasonic range, it’s used for electrical equipment that emit subtler sounds as well as mechanical equipment. It’s argued that this type of analysis predicts imminent breakdowns better than vibration or oil analysis.

Infrared analysis

Machine Speed: Low, High | Machine Type: Mechanical, Electrical | Cost: Low

This type of analysis is not dependent on an asset’s rotational speed or loudness. Therefore it’s suitable for many different types of assets. When temperature is a good indicator of potential issues, infrared analysis is the most cost-effective tool for predictive maintenance. It’s often used to identify problems related to cooling, air flow, and even motor stress.

Example of predictive maintenance

A centrifugal pump motor in a coal preparation plant is a vital asset for day-to-day operations. To prevent unscheduled downtime, the maintenance team decides to use predictive maintenance technology. Because it’s a large piece of mechanical equipment that performs heavy rotations, the obvious choice is to monitor vibrations with vibration meters.

The team attaches a vibration meter close to the pump’s inner bearing and establishes a normal baseline measurement, visualized through a waveform graph (below, left). A few months later, the vibration meter identifies a spike in acceleration (below, right). The maintenance team reviews this new data remotely and schedules an inspection. The technician who performs the inspection finds a loose ball-bearing and repairs it.

Moving forward, the team connects the vibration meter to its CMMS. Now, when the same spike is identified, a fault with the ball-bearing is predicted and a work order is automatically triggered to perform the repair.

Note: This example is inspired from a real use case documented in this study.

The ROI of predictive maintenance

According to a paper by the US Department of Energy, “a well-orchestrated predictive maintenance program will all but eliminate catastrophic equipment failures.” Compared to a preventive maintenance program, cost savings are 8 to 12 percent higher; and compared to a reactive maintenance program, cost savings range from 30 to 40 percent.

Other numbers stated by the Department of Energy include:

  • Return on investment: 10 times
  • Reduction in maintenance costs: 25% to 30%
  • Elimination of breakdowns: 70% to 75%
  • Reduction in downtime: 35% to 45%
  • Increase in production: 20% to 25%

Note: Increase production by 25% by reducing downtime, breakdowns and maintenance cost

Conclusion

Predictive maintenance is not for every organization, especially those that have yet to implement planned maintenance. But for larger organizations that have outgrown traditional PMs and have additional budget, predictive maintenance can provide an ROI that turns the maintenance department into a source of cost-savings and higher profits.