Predictive Maintenance: Harnessing the Internet of Things

What is predictive maintenance?

Predictive maintenance involves using machine learning analysis and data points (primarily from sensors) to understand how to keep assets performing. Utilising the Industrial Internet of Things (IIoT) and Big Data, predictive maintenance measures sensor data, such as vibration, temperature, and voltage, from assets so a machine learning algorithm can detect wear on parts to determine what maintenance is needed to minimise downtime. With quality data and the right algorithm, failure can be predicted accurately as far out as four months.

The goals of predictive maintenance are to:

  • Maximise asset uptime by reducing unexpected downtime
  • Optimise time spent on maintenance, thereby reducing operational costs
  • Reduce overall maintenance costs
  • Increase worker safety
  • Enhance overall productivity

Types of maintenance

reactive maintenance vs preventative maintenance vs predicative maintenance graph
While repair costs will plummet, organisations will invest more in maintenance budget to keep up predictive maintenance

There are three main elements of maintenance. In order of increasing complexity, they are:

Corrective: Corrective maintenance is a task performed to rectify a failure and return a working asset to a functional condition.

Preventative: Preventative maintenance is routine upkeep to lessen the likelihood of failure.

Predictive: Predictive maintenance determines, based on operational parameters, how and when maintenance should be performed.

By waiting for failure before engaging in maintenance, an organisation relying on corrective maintenance will encounter greater downtime, lost productivity, and threats to worker safety. While preventative maintenance is a step forward from corrective, it is often costly and unnecessary. For example, fixed maintenance intervals for high-use parts such as bearings or shafts may cause unnecessary downtime if there isn’t significant wear on the part. Runtimes from assets are shortened, uptime is minimised, and costs are increased. Predictive maintenance seeks to avoid this unnecessary cost and downtime by monitoring the ongoing state of an organisation’s assets.

Preventative vs. predictive maintenance

The critical difference in preventative and predictive maintenance is the utilisation of smart asset management. Preventative maintenance relies on schedules and testing (generally from third parties) to estimate when maintenance should be completed. Predictive maintenance uses a combination of historical and current data gathered from an asset to determine if, when, and how maintenance should be conducted. Additionally, proper predictive maintenance can leverage this data to help identify changes to operational processes and procedures to maximise uptime, worker safety, and productivity.

Predictive maintenance & smart asset management

smart asset management

Smart asset management uses computerised data from machines, tools, and workers to maximise performance, efficiency, cost, and safety. The business value of smart asset management cannot be overstated. In an asset-intensive industry (e.g., petrochemicals, mining) the ability to increase the performance of mission-critical assets has a direct effect on revenue. By providing an organisation with the ability to gain critical data about their assets, smart asset management can offer several benefits:

  • Tracking of location, usage, uptime/downtime
  • Improve ROI by increasing asset lifespan
  • Reduced cost-of-ownership by minimising maintenance
  • Monitor workflow and increase organisational efficiency
  • Reduce failures and increase worker safety

Predictive maintenance & Industry 4.0 – The outcome and the paradigm

            Predictive maintenance is a vital element of Industry 4.0, which uses automation and smart asset management to enhance overall productivity. Industry 3.0 implemented computerisation of assets, whilst Industry 4.0 connects and shares the smart data generated by these assets.

Big Data & IIoT

Big Data

is the name given to the use of predictive analytics on large data sets to extract value. Smart assets and a computerised workforce create a huge volume of data, and without the ability to capture, store, identify, and analyse this data its value is diminished. By using analytics (such as artificial intelligence and machine learning) on large sets of data, patterns and trends can be revealed, thereby helping to fine-tune future usage.

The Industrial Internet of Things (IIoT)

is the use of smart devices (primarily sensors) to gather data about manufacturing and industrial processes. This data can include anything from vibrations to temperature to noise, all of which are gathered and sent to a central hub to store the data.

            Big Data and IIoT are two core components of Industry 4.0. IIoT is responsible for gathering the information that is then utilised by Big Data for analysis. These technologies aid in predictive maintenance by enabling assets to be continuously monitored and evaluated, limiting downtime, reducing the risks of failure, and increasing productivity.

Condition-monitoring equipment

Any equipment that measures the state of an asset can be considered condition-monitoring equipment. Predictive maintenance uses condition-monitoring equipment that is connected to IIoT to gather data about an asset’s performance and current state. Condition-monitoring equipment contains a variety of tools, including:

  • Vibration monitors
  • Tachometers
  • Thermometers
  • Hygrometers
  • Phase rotation meters
  • Decibel meters

The equipment used will vary based on industry, but for predictive maintenance purposes, these devices must all be capable of reporting their data back to a CMMS which can then use analytics to determine necessary maintenance schedules.


Computerized Maintenance Management System

A computerised maintenance management system (CMMS) is the software that organises and controls information about operations and maintenance. The CMMS holds all the information shared by the many sensors equipped to smart assets and ensures that it is readily accessible by maintenance technicians and other software systems. CMMS software provides many elements necessary for predictive maintenance:

  • Data collection and organisation
  • Alert and work order generation
  • Data interpretation and reporting

Barriers to predictive maintenance

predictive maintenance roadmap
The building blocks of becoming predictive

While predictive maintenance can provide an organisation with cost savings, enhanced productivity, and reduced risk to operational capability and worker safety, it is easier said than done. It is vital that before attempting the transition to predictive maintenance, executive management understands the challenges posed by the conversion.

Lack of standards

A significant hurdle in the implementation of predictive maintenance is a lack of standardisation across smart assets. As predictive maintenance is a new and developing technology, there are several challenges to multiple aspects of standardisation. These include both software standards and standardisation of how data is handled by an organisation.

Smart assets come in a variety of combinations, often incorporating vendor-specific tools and software. These tools and their software must then interface with an organisation’s data management system and its predictive maintenance software. As many of these technologies have been created as bespoke solutions, this presents a genuine barrier.

Similarly, an organisation must develop its internal data governance, data management, safety, and productivity standards to account for predictive maintenance. Implementing a predictive maintenance solution is not as simple as installing an upgraded part or computer – it requires thorough examination and modification of existing standards.

While standards can be a challenge, it cannot be understated how vital they can be in reducing the risk to workers and productivity. If older standards put in place for preventative maintenance are used for predictive maintenance, this can impair productivity. At best productivity will be unchanged, at worst it can be harmed by unnecessary asset downtime and expensive and unwarranted repairs. Likewise, if maintenance is not adequately completed, workers can be put in very real danger. Machines may suffer catastrophic failure or old practices may put workers in harm’s way. Business leadership is required to ensure that as collaboration between man and machine becomes more prevalent, both can function in a productive and safe environment.

People and culture

Anytime a new tool, procedure, or process is implemented, it is vital to make sure the people in an organisation are properly trained and given the skills to function successfully. Upskilling workers is a common barrier to implementing new maintenance software, tools, and procedures, particularly one as complicated as predictive maintenance. Taking small steps, like maximising preventative maintenance procedures and metrics first, can smooth the transition. Additionally, a quality CMMS solution will allow an organisation to track training and skills of maintenance workers to make sure no one is lost in the process.

Experienced business leaders know that the main challenge in implementing many new solutions aren’t the technical skills but the social changes that may result. A worker may view their role in the organisation as changing or fear being replaced by new technologies. Resistance to change is a common struggle, but it is best countered by making sure workers are involved in the change and allowed to be active participants, creating buy-in and giving them a meaningful stake in any change. By repeatedly engaging in this behaviour, employees will grow to understand that changes at an organisation aren’t made to replace or challenge them, but to improve the workflow and increase their safety and productivity. This helps to establish a culture that is open to innovations and flexible when meeting challenges, which is beneficial for any business, regardless of maintenance tools.


Implementing a predictive maintenance system can be expensive. Companies in the engineering field are sometimes relying on decades-old equipment, and replacement with smart assets (or upgrading existing assets with smart technology) can present a daunting cost barrier. However, the Wall Street Journal said that unplanned downtime cost the manufacturing industry $50 billion per year, while General Electric found that unplanned shutdown costs oil and gas companies $42 million per year – or $88 million in the worst-case scenarios.

Using the $42 million as a benchmark – anything your organisation can invest up to that number will have benefits for your company in the long-term. And, depending on the company’s maintenance maturity, you can expect to spend close to that number in the first year alone

reliability-centered maintenance diagram
Each stage of this flow-diagram requires cash investment, not just in technology but in overhauling people and process.

Using data published by the U.S. Department of Energy, it can be estimated that predictive maintenance can provide the following financial benefits:

  • Reduction in maintenance costs between 25% to 30%
  • Eliminate 70% to 75% of breakdowns
  • Reduce asset downtime by 35% to 45%
  • Increase productivity 20% to 25%

Even though initial costs might be considerable, the long-term benefits of predictive maintenance make financial sense.

Would you like to provide feedback on this content?

Predictive maintenance: Where to start?

Starting a predictive maintenance program from the ground up is a daunting task. Luckily, almost all organisations have some of the basics already in preventative maintenance procedures, processes, software, and equipment. The first step towards predictive maintenance is enhancing existing preventative maintenance.

Focus on metrics

Utilising a CMMS will allow maintenance statistics to be tracked and analysed. If an organisation doesn’t know how long maintenance tasks are taking, there is no way to know if there is an improvement. There are three maintenance metrics which should be the focus.

Mean time to repair:

This metric measures the average time it takes workers/technicians to troubleshoot and repair an asset which has failed. Mean time to repair measures the downtime of equipment while workers perform the following tasks:

  • Notice equipment failure
  • Diagnose/troubleshoot the issue
  • Implement a fix
  • Reassemble the equipment
  • Align/calibrate equipment (if needed)
  • Test the equipment

Mean time to repair should be a metric that is concentrated on to encourage the uptime of assets, particularly mission-critical ones. Assets that are non-functional are costing productivity and time, and the failure and repair of an asset can endanger worker safety. Note that this metric requires corrective maintenance. In preventative or predictive maintenance this measure is important but should come up less frequently.

Mean time between failures:

This metric is a measure of how long an asset is up between breakdowns. Mean time between failures does not include planned outages, such as preventative maintenance, inspections, or recalibrations. The mean time between failures for an asset should be as high as possible. Many vendors provide their estimates of the mean time between failure for their products, but there are often the result of laboratory testing and may not be applicable for real-world usage. There is also the human factor – workers can cause equipment failure through misuse, improper maintenance, and poor repairs.

Overall equipment effectiveness:

Overall equipment effectiveness measures three factors: availability, performance, and quality.

  • Availability: A measure of when equipment is up and functional. It is driven lower by breakdowns, idle time, and stoppages. It can be improved through better preventative maintenance, proper worker training, and efficiency analysis.
  • Performance: The measure of a system’s throughput compared to its maximum possible throughput. This is harmed by inefficient work processes, poorly conducted maintenance, and older systems. Improvements are gained through better maintenance procedures, more efficient work processes, or replacement with newer equipment.
  • Quality: The number of useable units produced compared to the number of units started. Poorly maintained equipment, improper calibration, and failures will drive this metric down. Maintaining and calibrating equipment will increase quality.

Measuring overall equipment effectiveness will help to understand overall system productivity. A higher effectiveness score indicates that machines are available and effective. A low score indicates that maintenance procedures may need to be examined, workers might require additional training, or equipment is out-of-date and requires replacement. Executive leadership should be aware of this metric in particular as it can be an indicator of the health of an organisation’s maintenance infrastructure.

Predictive maintenance machine learning: Break down data silos

Data silos are isolated and fixed data repositories, generally under the control of a single unit or department. If the purchasing department and the engineering department each had their own data storage locations and the two were not in communication, they each have siloed data. Data silos are a disruptive business and information practice, and they make it virtually impossible to implement predictive maintenance. Executive leadership can disrupt this behaviour through strong governance principles and policies.

Data governance:

The best tool to disrupt data silos is a data governance solution. Data governance is the systems and philosophies related to the monitoring and decision-making of an organisation’s data. A robust data governance plan involves creating a framework that manages a strategy designed to enable interconnected data management across an organisation’s units. Implementing a data governance plan also requires the assistance of advanced software to aid in consolidating data, enforcing and tracking policies and procedures, and meeting quality and regulatory metrics.

Information management:

Like data governance, information governance involves the high-level management of an organisation’s information. Less specific than data governance, however, information governance is the umbrella term for the systems and philosophies related to the management of all information created by an organisation. This includes record keeping, business structures, compliance, management strategies, and risk policies. Without a formalised and specific set of information governance systems, proper data governance cannot be implemented, and data silos will continue.

Predictive maintenance tools & software

Enterprise mobility management

As predictive maintenance relies on IIoT and Big Data, it is crucial that everyone, from workers to executives, has the necessary technological tools and training to operate in a data-driven enterprise.

Training and culture:

It is imperative that all employees receive the proper training in the tools and software necessary to work towards achieving predictive maintenance. Creating training programs that help all employees receive the requisite knowledge and skills to flourish in their role will not only help ease the transition to predictive maintenance but will allow for a more flexible and adaptive enterprise. By integrating technological solutions (e.g., data governance software, CMMS) into an organisation, the culture will become more open and receptive to the change towards an information-intensive environment.

Operations management software:

Enhancing the day-to-day operation of an organisation is best achieved through operations management software, which is considered any software solution that helps to streamline business operations. Operations management software is useful for helping to unlock more efficient operations at all levels of a business. This is accomplished through several mechanisms, such as decision analysis, supply chain management, scheduling, or quality control. While this software generally doesn’t implement predictive maintenance, it is a vital tool in helping an organisation to work towards achieving the high-level benefits of predictive maintenance, and no organisation should be without it.


Today’s industrial organisation is now more information-oriented than ever. Smart tools, statistical analysis, and metrics have pushed businesses into the next generation: Industry 4.0. Any organisation that isn’t actively pursuing these technologies will be missing out on their potential.

Predictive maintenance will help to enhance productivity and safety, streamline operations and maintenance, and provide a competitive advantage. The ability to maximise asset uptime, minimise time to repair, and increase overall efficiency while saving money is vital to modern industrial organisations. This can only be achieved by enhancing operations using software and smart assets that enable the free flow of data. While predictive maintenance is a lofty goal for many organisations, working on the building blocks – information management, data governance, and culture – are all easily achievable with the right partners.

  • On-Demand Webinar available now

    Breaking down the silos between finance and projects is crucial to improving outcomes - but how do you do it? Find out in this webinar

  • Never Miss an Update

  • Recent Posts