Predictive maintenance

Perspective: QVARTZ Analytics

Predictive maintenance has been around for 20 years, so it is by no means a new concept. However, factors such as lowered maintenance costs with reduced downtime and the rapid expansion of the Internet of Things (IoT) are expected to greatly increase the demand for predictive maintenance solutions and services. Adding to this the immense development that has taken place within technology during the last couple of years, predictive maintenance is more relevant today than ever before, and consequently, more and more companies are investing in it.

In its simplest terms, predictive maintenance provides predictions of asset or component failure based on machine learning algorithms. These algorithms analyse the historical data patterns from various sensors embedded in and around the asset, along with relevant external parameters, e.g. weather and temperature. Predictive maintenance has demonstrated promising results including eliminating breakdowns by 35% to 45% – and reducing downtime by up to 75% across industries. Furthermore, it provides an opportunity to optimise planned maintenance by providing relevant insights, which can enable the maintenance staff to skip the regular maintenance cycle by utilising the predictions made by the machine learning algorithms.

Predictive maintenance requires preparation

Even though technical advances have made predictive maintenance more available than ever before, most companies are not sufficiently prepared to take full advantage of the technology and apply it to their own operations. Typically, two key aspects are missing:

First, a clear understanding of scenarios where predictive maintenance will be useful and the potential cost savings must be in place. Predictive maintenance can enable you to predict the failure, but you will not get significant savings by avoiding the cost of spare parts required to repair the failure. The failure will in all likelihood still take place, however; the significant cost savings are realised by other means, e.g. reductions in unplanned downtime, which can have a far bigger impact on your operations.
Second, it requires a certain level of maturity from the company: of having the necessary data stored and available in real-time. Without such maturity, there is no solid basis for building a reliable model for predicting when maintenance is needed. Just knowing that you have a sensor in your assets will not suffice; you need to ensure that you have the right quality and reliability of data – and that you can read your data, even if it is in a propriety data format from the original equipment manufacturer.

Beside the technology and data, you also need to address the adoption of
technology and the new ways of working for your maintenance team.

Getting prepared and succeeding with predictive maintenance

With this perspective, QVARTZ aims to highlight that predictive maintenance is about more than mere technology. It is not just about the data and analytics, but more about the choices you make at various stages to maximise the value it can create for your business. In order to succeed with predictive maintenance, you need to go through the four steps described below. Each step is in itself essential – and all steps are equally important.

FIGUR 1: Investeringer i kraftnettet frem til 2025 (mrd. NOK)

Wondering if your organisation is ready to go on a predictive maintenance journey?

Found the above words intriguing? You’ll love these.

Digital transformation To navigate in the era of digital transformation, we need to understand both the motions and the key drivers that enable us to respond to the digitisation of our society and our businesses. By looking at the digitisation pioneers among Scandinavian companies and studying their best practices, we can learn from their experience – and we can create a framework to better understand the initiatives and the progress we are making towards the new world.