Condition-Based Maintenance: a comprehensive guide

By Richard Crowter

03 April, 2023

Introduction

In this guide to Condition-Based Maintenance (CBM), we discuss how it is defined and what it is not. How Industrial IoT has brought about the digital transformation of CBM. The advantages of Condition-Based compared to Reactive and Preventive strategies, and how it improves upon the P-F interval. The merits and downsides of Predictive Maintenance as an alternative to CBM are also considered. Implementation tips are provided, based on innius’ experience and customer success story.

What is Condition-Based Maintenance?

Condition-Based Maintenance (CBM) is a strategy where maintenance is determined based on the actual condition of a machine or part. This relies on accurate monitoring to detect signs of wear and tear, so that maintenance can be conducted proactively before potential problems escalate and cause failures. By only doing what is needed, CBM aims to use maintenance resources more efficiently, to reduce machine downtime, cut repair costs and extend the life of machines.

This contrasts with a Reactive maintenance strategy where maintenance only occurs when something has broken, or a Preventive maintenance strategy where maintenance activities are scheduled based on set time or work intervals.

Condition-Based Maintenance is not Predictive Maintenance

Be aware that in some places, particularly older material, CBM is referred to as ‘a predictive maintenance strategy’. We would caution against this. Nowadays, Predictive Maintenance is seen as a separate strategy, which employs technologies such as Artificial Intelligence and Machine Learning. We’ll consider and compare these strategies in more detail further in this article.

The impact of Industrial IoT on CBM

Industrial IoT (IIoT) has radically transformed Condition-Based Maintenance. But it is worth noting that CBM is not entirely dependent upon IIoT. When for example, a maintenance technician puts a screwdriver to a machine and listens to it to check if it is running smoothly, CBM is being conducted. From experience the technician knows how the machine should sound, and if something isn’t right maintenance can be planned.

Using IIoT, such as Innius, takes CBM to the next level, which is the subject of this article. Digital sensors can detect things a technician never could by looking and listening. So, lots of new opportunities have been created as many maintenance use cases can be moved from a Reactive or Preventive to a more efficient Condition-Based Maintenance strategy.

Why use Condition-Based Maintenance?

Like all strategies, Condition-Based Maintenance is not a one-size-fits-all answer to every maintenance scenario there is. Instead CBM should be seen as part of a holistic approach to maintenance. This means that maintenance use cases and failure scenarios need to be categorized one by one and the most appropriate maintenance strategy for each needs to be applied. This could be reactive, preventive, condition-based or predictive.

The decision to use CBM instead of reactive or preventive maintenance strategies is largely to made on the expected return on investment (ROI). CBM requires an initial investment, which can be more if special sensors need to be installed, and then there are recurring costs to process and store the data. With innius these are not substantial costs, but they are costs not incurred with Preventive or Reactive Maintenance. But because CBM ultimately reduces equipment downtime, the costs of CBM can be justified if downtime costs are high enough. CBM is therefore more suited to parts and machinery which are highly critical to business operations. Another consideration for the ROI is that CBM reduces the workload on maintenance staff and the severity of repairs, which also save costs.

Increasing the P-F interval with CBM

A key comparison is that with Condition-Based Maintenance defects can be discovered, and potential failures diagnosed before preventive maintenance would. This is demonstrated by the P-F curve, a graph which shows a how machine condition (or of a part) deteriorates over time and how different maintenance strategies can detect the potential failure (P) before the actual failure (F) occurs. The difference between these two points is known as the P-F interval. These graphs show a typical P-F curve, although each potential failure actually has its own P-F curve.

Here you see that the P-F interval for Condition-Based Maintenance is longer than for Preventive Maintenance. This extra time and accurate diagnosis of the problem, allows the correct parts to be sourced in time, which reduces the Mean Time To Repair (MTTR) when the maintenance takes place. It also allows the maintenance to be planned when the staff are available and ideally when the machinery is not scheduled to be used, thus maintaining Overall Equipment Effectiveness (OEE).

Other advantages of Condition-Based Maintenance

Lower severity maintenance

Dealing with impending failures earlier, also means that on average these have a lower level of severity, because they have not escalated and caused further damage. Generally, maintenance incidents with lower severity are cheaper in terms of parts and time to perform. Again, quicker maintenance tasks reduce the Mean Time To Repair (MTTR). Take for example a bearing which seizes, in turn breaking a pump shaft, causing the flow of coolant to stop, spoiling a batch of product. This would be a costly chain of events which would have been prevented had the bearing been replaced before it seized.

By reducing a machine’s downtime and consequently raising OEE, a machine’s useful life is also extended. This is what happened to innius customer Itho Daalderop.

With maintenance tasks only conducted when required, staff save time and use their precious expertise more effectively. If the maintenance team work remotely, remote monitoring allows them to cut down unnecessary journeys. Ideal if they are offering maintenance as a service.

Another advantage of lowering the severity of maintenance incidents is that these are less dangerous and so worker safety is improved.

Comparing Condition-Based Maintenance and Predictive Maintenance

The P-F curve demonstrates that the earlier potential failures are detected, the better. But the scale of the graph is also important, as the P-F interval needs to be long enough to be able to respond. As this diagram shows, the P-F interval for Predictive Maintenance is even longer than for CBM. This begs the question why Predictive Maintenance is not always the preferred strategy compared to Condition-Based Maintenance. Especially as both make use of sensors, data streams and software.

The difference between Condition-Based and Predictive Maintenance

To answer that question, it is important to firstly understand the difference between CBM and Predictive Maintenance. CBM applies human experience or trial and error to accurately define measurement thresholds. The accuracy of the ‘prediction’ will only ever improvement if operators adjust the settings. But with Predictive Maintenance once a threshold or breakdown scenario has been defined by the human operator, AI or machine learning will try to identify patterns in the data stream to give advance warning. Over time, as data builds and perhaps with corrective input from operators, the predictions of failures will get more accurate, and the P-F interval should further increase.

The downside to Predictive Maintenance

The downside to Predictive Maintenance, when compared to Condition-Based, is that it requires a lot of data to be built up before patterns can be detected. Often this also means data streams at a high frequency. Processing data quickly and storing large amounts is expensive. So, which strategy is best is mainly based on the Return on Investment. For example, it has been used in industry to monitor frequency changes in electric motors, but the data costs were so expensive it was cheaper to simply keep a spare motor on standby.

When to use Predictive Maintenance

Predictive Maintenance can be the right solution for incredibly critical assets, where the cost of downtime and repair are high, and a long P-F interval is needed to plan and conduct maintenance. Predictive can also be a solution where changes in condition are difficult to detect with other methods.

How to use CBM with Industrial IoT and overcome the challenges

To successfully implement a digitalised Condition-Based Maintenance strategy, using Industrial IoT, our experience suggests the following steps:

1. Identify a failure scenario for a part or machine.
2. Identify a measurable condition (or more than one) which indicate the potential failure in advance.
3. Is the condition already being monitored by a sensor in the machine? If so, new sensor is needed.
4. If you need a new sensor, consider that different sensor types will be able to detect potential failures at different points, and so have different P-F intervals. Crucial to success is ensuring the P-F interval is long enough to respond. Another consideration is that the sensor(s) need to be able to withstand the conditions of the operating environment. Choose from the following measurement/sensor types:
  1. - Pressure
    - Temperature
    - Vibration
    - Oil quality
    - Electrical
    - Ultrasonic
    - Infrared
    - Acoustic
    - Operational performance
5. Choose Condition-Based Maintenance software which meets your needs, consider the following:
  1. - Is it an off-the-shelf solution or does it require custom work? This has a big impact on the setup time and costs.
    - The initial and recurring costs.
    - Are you only going to use the software for CBM or other business applications? If it is only for CBM avoid software which is too heavily geared for other uses.
    - How easy is for operators to adjust the thresholds and notifications for example, without requiring help from a technical consultant?
    - Does the CBM software offer dashboards, reports and mobile apps with push notifications?
    - Is It possible to combine multiple data streams and monitor this with a single threshold and notification? Innius does.
6. Connect the sensor to the CBM software. In the case of innius this is done by placing an Ewon modem in the PLC.
7. Configure the data streams in the CBM software and start collecting baseline data. At this point having experienced maintenance staff who are familiar with the machines and can interpret the data will be a huge help and speed up the implementation process.
8. Install mobile and tablet apps and hang up dashboards as needed to create awareness amongst operators and maintenance staff.
9. Thresholds and notifications can then be accurately defined. Initially you want notifications to be set so that they are slightly more sensitive than required, and then tune them down with experience. But still you don’t want staff to be bombarded with unnecessary notifications or alternatively breakdowns occurring because the thresholds are not sensitive enough.
10. Long term, you need the commitment of maintenance staff to continue fine-tuning the thresholds so that notifications are limited to those which are actionable.
11. Fine-tuning also helps to extend the P-F interval, which gives the maintenance staff more time to plan work and reduce the impact of irregular schedules due to notification-based work.

Our customer’s example of using Condition-Based Maintenance

Innius customer, Itho Daalderop was able to save a significant amount of money by implementing Condition-Based Maintenance on their production line. Itho Daalderop makes boilers and a machine used to test boilers for leaks, the leak tester, was old and broke down regularly. Using innius to monitor existing sensors in the machine, the maintenance team are warned by mobile notifications of potential breakdowns in advance. Often repairs can take place during production line breaktimes, thus minimising downtime. The result is that the operating life of the leak tester has been extended and the machine will not need to be replaced after all. So, in this case the ROI was calculated based on the reduction in costly production downtime and the alternative solution, which was to buy a new machine at hundreds of thousands of Euros.

References:

Teixeira, Humberto Nuno, Isabel Lopes, and Ana Cristina Braga. “Condition-based maintenance implementation: a literature review.” Procedia Manufacturing 51 (2020): 228-235.
Sakib, Nazmus, and Thorsten Wuest. “Challenges and opportunities of condition-based predictive maintenance: a review.” Procedia Cirp 78 (2018): 267-272.
Prajapati, Ashok, James Bechtel, and Subramaniam Ganesan. “Condition based maintenance: a survey.” Journal of Quality in Maintenance Engineering 18.4 (2012): 384-400.

Richard Crowter, UX Designer & Marketing Coordinator at innius

Curious what innius means for your business?

This is Richard, he’s happy to advise you! Please fill out the form and he will be in touch!