Predictive Maintenance: A Complete Guide

Predictive Maintenance Guide - Repairing Infrastructure & Equipment

Predictive maintenance monitors and assesses equipment while it’s in use, to detect issues and predict the optimal time to carry out maintenance, preventing failures and reducing maintenance costs.

With so much of modern life and business being reliant on complex and essential pieces of machinery, finding ways to ensure that it stays running is vital, while keeping servicing efficient and cost-effective.

What is predictive maintenance?

Predictive maintenance is a way to minimize both failures and maintenance work, while maximizing machinery and component lifetime.

Predictive maintenance relies on testing and monitoring of equipment while it’s in operation – also called condition-monitoring – to provide data about the current performance of the machine, in order to predict issues and prevent failures.

With real-time monitoring, it can also use machine learning to create a baseline pattern of normal operations and analyse current data to make predictions about maintenance requirements, using complex numerical algorithms.

Predictive maintenance objectives

With predictive maintenance, the goal is to carry out maintenance tasks at the optimal time, which is a combination of two primary competing objectives.

Firstly, predictive maintenance aims to prevent equipment failure, which can potentially cause damage of other machinery or components and increase the total cost of the fix. Further adding to the cost of failure is the downtime experienced by the organisation while the equipment is repaired.

And at the same time, predictive maintenance aims to predict in advance when maintenance will be required, using data from the equipment and predictive algorithms, instead of relying on a fixed maintenance schedule or routine procedures.

Ideally this means that maintenance will be performed less often, and only as required, but before the equipment fails.

The overall purpose of predictive maintenance is to reduce the cost of asset management, whether that’s in the cost of parts, downtime or labor, by performing maintenance at just the right time.

Predictive maintenance vs preventive maintenance

Predictive maintenance differs from preventive maintenance, in that preventive maintenance only addresses the first objective of predictive maintenance – to reduce failures.

It generally does not include a prediction of the timing of future failures, and a matching adjustment to maintenance schedules, based on the current state and performance of equipment.

Instead preventive maintenance relies on time-based scheduled maintenance programs to reduce the risk of failure.

Types of predictive maintenance

Predictive maintenance generally takes one of two forms.

1. Condition measurement

Firstly, equipment can be manually tested non-invasively while it’s in use, on a routine basis, to create a model of when components may fail and provide recommendations around required maintenance tasks.

2. Condition monitoring

Alternatively, machinery can be monitored continuously, using specialized sensors, that provide real-time data to monitoring systems, to allow for detection and prediction of issues.

The two approaches can also be combined, depending on the needs of the organization.

How predictive maintenance works

Preventive maintenance includes a number of important steps for it to be effective.

  1. Monitor equipment – This can be done manually or automatically
  2. Identify “normal” operation
  3. Analyse data
  4. Register changes in measurements
  5. Identify components likely to fail
  6. Recommend maintenance tasks and their timing

As each operating mode or production line etc has unique characteristics, it’s important to have a “baseline” evaluation of the system or equipment in typical or ideal operating conditions. This can be done by skilled operators and also via machine learning models.

The data that is collected via these processes can also be used to forecast component lifespans and give advance warning of equipment failure.

Monitoring methods for predictive maintenance

The measurement and monitoring of systems and equipment is generally done via sensors and specialized testing equipment.

Many of these testing methods are non-invasive and can be done while the equipment is in active use.

Listed below are some of the more common monitoring techniques used in predictive maintenance.

1. Visual inspection

Although visual inspection can’t detect issues with components concealed inside equipment, it is still an important method for detecting obvious issues that may cause failure.

Visual inspections are best conducted by an expert who is very familiar with the machinery and what “normal” should look like.

2. Electrical testing

Various features of electrical systems, including current, voltage, resistance, induction and magnetics fields, can be monitored and measured to detect potential issues with equipment.

An “electrical signature” can be created for the machine or system under normal conditions, and then ongoing measurements are compared to this baseline to detect imperfections and identify the likelihood of failure.

3. Oil analysis

The oil that is passing through machinery in order to lubricate the components as they operate can be analysed for various factors that can indicate the degradation of function.

An increase in the quantity of contaminants and the size of debris particles can indicate a part that is wearing faster than normal, which may precede failure.

4. Vibration analysis

Analysing the vibrations of a piece of equipment while in operation can detect subtle changes in operation that can signal component wear and tear, and an increased risk of failure.

It is most commonly used for machinery that includes rotating components, such as motors and pumps, and can include monitoring of vertical, horizontal or axial movements.

An increase in vibration usually indicates that there is a part – or parts – that are not working together as smoothly as they were.

5. Sound levels

Malfunctioning equipment can create different sounds that can be detected by sound monitoring equipment.

Whether it’s an increase in noise from a component that’s grinding where it shouldn’t be, or a decrease in levels from a component that has stopped working, detecting these changes in sound levels early can reduce the chances of catastrophic failure.

Acoustic analysis can also be used to detect failures in liquid or gas system, that create unusual vibrations and sounds that are transmitted along pipes to sensors.

6. Ultrasound

Ultrasound technology can be used to “see” defects in structures or equipment before they become visible to the human eye.

It can detect physical issues like corrosion and flawed welds that may indicate impending failure, and the subtle ultrasonic sound changes that can result from leaking fluids and gases and malfunctioning bearings.

7. UV emissions (Corona Detection)

Certain kinds of equipment, such as those involved in high-voltage electrical systems, such as power transmission lines and electric trains, may emit corona discharges when not operating correctly.

Some of these emissions may be in the UV spectrum, meaning that they are not visible to the human eye. Measuring the corona using specialized UV cameras can ensure that any resultant damage is detected and rectified promptly.

8. Temperature / Thermography

Irregularities in the operating temperature of a piece of equipment can be an early warning sign of malfunctioning components.

Excessive overheating may lead to melting or burning of components, that will cause severe damage to equipment if not detected promptly.

Temperature changes also affect the amount of radiation emitted by equipment, which can be easily detected via infrared cameras.

Predictive Maintenance Guide - Monitoring Temperature

Additional monitoring methods may be used, depending on the equipment and business needs, including radiography, laser interferometry, flow rates and output volumes.

Predictive maintenance technology & tools

An essential feature of predictive maintenance is the technology used to monitor and analyse the data about the current state of the equipment.

With the growth in the Internet of Things (IoT), sensors are able to collect and share measurements online with software systems that store, analyse and respond to the data.

Analysis of the data can be conducted manually, however this is generally labor intensive, inefficient, unable to scale and the accuracy depends on the team members.

Machine learning is taking a much larger role these days in predictive maintenance, using the data provided to:

  • Detect component issues in real-time and alert maintenance teams
  • Build models of “normal” expected operations
  • Forecast component life and prioritize component replacement, using predictive algorithms
  • Optimize the timing of planned maintenance tasks and downtime

The data collected can potentially also be used to optimize the operating efficiency of the equipment or systems.

The tools and technologies required for a successful predictive maintenance program may include:

  • Software for collecting, processing and storing data
  • Algorithms to monitor measurements in real time
  • Algorithms to make predictions based on current and past data
  • Integration of your monitoring systems with maintenance processes

Predictive maintenance examples & industries

Although predictive maintenance can be used across a wide range of industries, some concrete examples help to demonstrate why it’s becoming more important to organisations.

Refrigerated shipping containers

Shipping companies can use predictive maintenance approaches to assess the fuel and power consumption of refrigerated containers and optimize operations.

By analysing the data collected and exploring the efficiency of various modes of operation, one shipping company discovered that they could save $6.5M each year by running multiple generators at a lower capacity instead of overworking a smaller number of generators.

Utility maintenance

Utility companies are using drones carrying specialized sensors to monitor the infrastructure network, and identify risks such as trees that could fall on power lines, a transformer that has an elevated corona discharge, or a turbine that is suffering from thermal fluctuations that will lead to failure in 6-12 months.

By collecting data about the assets across the entire network over time, predictive analysis can identify these issues before they cause power failures, irreversible equipment damage and unplanned maintenance.

Oil and gas mining

Much of the world is still dependent on oil and gas to power their everyday activities. The extraction, refinement and transportation of oil and gas resources involves specialized, expensive equipment.

Failure of this equipment can lead to massive environmental damage and loss of life, as demonstrated by various catastrophes over the years including Piper Alpha in 1988, Exxon Valdez in 1989 and Deepwater Horizon in 2010.

By monitoring equipment closely, and providing detailed information on the current condition of systems to all concerned, the chances of similar tragedies is greatly reduced.

Railway operations

By monitoring data from signal boxes, locomotives and other transportation assets, railway operators can identify deviations that suggest imminent faults, and prioritize maintenance activities to minimize disruptions to cargo movements and passenger travel.

Detecting issues early also increases safety for railway personnel and decreases the cost of reactive, unplanned maintenance.

Predictive Maintenance Guide - Railway Operations

Other industries that are seeing benefits from predictive maintenance include manufacturing, food & beverage, IT services, automotive, airlines and ports.

Benefits of predictive maintenance services

The examples above illustrate a number of benefits that organisations can realize from implementing predictive maintenance techniques.

1. Increased lifespan of assets

One of the main advantages of predictive maintenance is the increase in lifespan of equipment, as a result of targeted maintenance that addresses key deteriorations before they lead to catastrophic failure.

2. Reduced equipment damage

Unexpected failure of equipment or components can directly cause further damage to the equipment and connected systems, leading to expensive repairs, multiple part replacement, or even retirement of the entire machine. By catching issues early, the likelihood of collateral damage from equipment failure is greatly reduced.

3. Fewer replacement parts needed

Because equipment is being repaired proactively, the reduction in collateral damage means that fewer replacement parts are required, reducing the overall cost of repairs and maintenance.

4. Reduced downtime

The reduction in unexpected failures combined with the reduction in the frequency of maintenance that comes from a preventive maintenance approach means a commensurate reduction in downtime. Equipment is being repaired before it fails, but only as required, meaning that is offline for as little time as possible.

5. Reduced cost of downtime

Being able to predict when equipment needs attention means that maintenance work can be scheduled at times when it is more convenient and cost-effective for the business to pause operations, reducing the per unit cost of downtime. Avoiding unexpected failure also generally means that less time is required to fix the equipment.

6. Parts and resources available

Planning ahead for equipment maintenance also allows for the necessary parts and equipment to be available in advance, ensuring that repair work can be carried out as efficiently as possible, with the right parts and skills for the job.

7. Increased safety

Minimizing unexpected failures also leads to an increase in safety for a number of reasons. Firstly, staff operating or working close to the machinery are less likely to be injured if equipment does not fail without warning.

In addition, staff working on the repair of equipment will also be safer if this is a planned task with minimal damage, rather than an emergency job that may to be completed in less than ideal conditions.

And finally, if the well-being of the general public would be affected by the failure or downtime of this equipment, their safety is also improved by predictive maintenance.

Predictive maintenance may also lead to a lower environmental impact, waste reduction, improved quality of output, higher morale and an average in performance over time.

Disadvantages of predictive maintenance

Predictive maintenance aims to reduce the overall cost of maintenance in a number of ways, but there are still some disadvantages to consider.

Depending on the systems and equipment involved these risks and costs may outweigh the cost benefits realized by a preventive maintenance approach.

1. Expensive monitoring equipment

The monitors and testing equipment required for some of the predictive maintenance methods can be quite expensive to purchase and install, making the upfront costs of a predictive maintenance program quite high.

2. Expertise of technicians & analysts

Making use of the data collected by the condition monitoring used in predictive maintenance requires specialized expertise to ensure that data is correctly interpreted.

3. Measurement limitations

The danger of more complex monitoring and analysis technologies is an increasing reliance on them and an expectation of infallibility.

With any technology, it’s important to recognize that no measurement or prediction can be 100% accurate, and failures may still occur and must be planned for.

Monitoring models may also be unable to take into account the general context of operations, such as the age of the equipment or the current weather conditions.

Predictive Maintenance Guide - Condition Monitoring

Considerations for predictive maintenance programs

Along with the benefits and drawbacks outlined above, an important consideration when deciding whether to implement a preventive maintenance program, is whether the approach is a good fit for the asset.

If failure of the equipment will not cause excessive harm or costs to operations, then it may not be necessary to work to avoid failures at all.

In cases such as these, the machinery can be allowed to run until it breaks down and then repaired, or the frequency and costs of failure can be tracked over time and balanced with the frequency and costs of maintenance.

How to implement a predictive maintenance program

Once you have decided to proceed with a predictive maintenance program for your business, there’s some key steps that you’ll want to complete to ensure success.

1. Create a team & define goals

To create a successful predictive maintenance program, you’ll need to put together a team who can champion the program, represent areas of the business and define the goals of the exercise.

2. Conduct a pilot program

If possible, test the idea on a section of your operations to test the impact and value to your organization of a predictive maintenance program. This will also help you to iron out any issues before you scale it up the entire business.

3. Prioritize assets

Some equipment will be easier to add to a preventive maintenance program than others, but it’s just as important to consider how critical a piece of equipment is to operations, or if a predictive approach is even appropriate.

Choose your equipment on priority, complexity of maintenance, how often they’re failing, replacement and repair costs, and their history and condition.

4. Identify resources

To set up a preventive maintenance program, you’ll need the right tools and people. Evaluate monitoring technology and data collection and analysis systems based on your goals and chosen equipment.

And then choose the right people to implement and run the program and provide all the necessary training to ensure that your program has the greatest chances of success.

5. Set up data collection

The next step is to set up your monitoring and measuring equipment and processes and begin collecting data. You can also add in historical data to some systems to provide value more quickly.

6. Create algorithms

As you begin to collect data, you’ll want to use this to create baseline models of normal operations, to identify what failure what looks like. You’ll also want to design a model for predicting future failures and requisite maintenance tasks.

7. Assess performance & improve

As your predictive maintenance program progresses, you will be able to learn from your successes and failures to make your program even more effective. Make sure you establish a process for continual review and improvement, with a view to creating a consistent approach over the long-term.

Predictive maintenance: The future of maintenance?

Predictive maintenance is an evolution of our approach to maintenance, that aims to reduce the frequency of both equipment failure and equipment maintenance, to maximize efficiency and minimize costs.

There are many different methods available to measure and monitor equipment, in order to detect issues and predict the timing of failures, to allow for maintenance work to be planned in advance and completed with minimal downtime.

There are many concrete benefits of preventive maintenance that organisations are realizing through the implementation of preventive maintenance programs, and these need to be weighed up against the costs and limitations of the approach.

Implementing a preventive maintenance program can lead to cost-savings and create a competitive advantage for your business.

Need to streamline your field service operations?

Talk to us today about how NextService can help your business improve the efficiency of your field service operations and maintenance services.

And download the NextService product guide below.

Download Brochure Here

 

References