What is predictive maintenance? All industrial machines are subject to wear and tear. Whether its abrasion, corrosion or fatigue, the health of your machinery will be affected and you will inevitably lose money due to downtime.
What can you do to mitigate an issue that naturally occurs?
You could already be pursuing a strategy of preventative maintenance – periodic inspection of machinery could identify problems before they happen, reducing the time spent on stoppages. However, this still won’t tell you exactly when issues occur.
However, as we’ve entered the fourth industrial revolution, predictive maintenance is a way to prevent you from losing money that downtime causes.
Through the industrial Internet of Things (IoT), you can connect sensors to machinery that will collect data, and use advanced analytics to take this data and create meaningful insights.
How manufacturers can benefit from predictive maintenance
Analysts believe predictive maintenance will be big business – according to Big Market Research (BMR), the global predictive maintenance market size will be worth $24,015.7m by 2026.
As a manufacturer, why should you be interested? The answer is value-added services and servitisation.
Certainly, predictive maintenance has the capability of powering new as-a-service models, in which the manufacturers of capital equipment can bundle in maintenance and repair as part of the overall cost of a system.
There is excellent value in the data you can collect from predictive maintenance, but you need to look closely at the back-office expertise you need to make the best use of it. You need to have the right data available, frame your problems appropriately and evaluate the predictions properly.
Best practice dictates that every sensor on your machinery must have an algorithm aligned with the monitored piece of equipment. If not, you won’t be able to report individual red/amber/green situations effectively.
The algorithm is usually defined by the equipment’s manufacturer.
The next step is to collate multiple instances of fail scenarios to understand why they occur and anticipate them. Many manufacturers successfully collect data to understand why numerous instances of their products fail.
However, it would be more effective if users worked together to collect similar data across all of their estates, so they can understand why chains of equipment from different manufacturers might fail.
3. Machine learning
The next step in maintenance optimisation is to apply machine learning and predictive analytics to these cases, but machine learning is exponentially dependent on data quality.
Optimising a complex operational chain requires increasing numbers of variables and parameters to be analysed in real time for incremental degrees of benefit.
Using predictive maintenance for the circular economy
The circular economy is a system that looks to eliminate waste and the continual use of resources.
As opposed to a traditional linear economy in which we make, consume and throw away, moving circular is all about creating a circle where we reduce waste and pollution by keeping products and materials in use for as long as possible and finding ways to create new resources from what we discard.
Being part of the circular economy is a good thing from a moral standpoint. But, more to the point, you may be getting pressure from customers and regulatory bodies to do more sustainable business that is better for the environment.
Customers are increasingly appreciating and looking for links with companies that show they care, and the long-lasting relationships you can build like this are like gold dust.
According to the Ellen MacArthur Foundation, the manufacturing industry could save between 10% to 15% on direct materials required for production.
Managing the economics of the circular economy demands that your business understands and remodels your supply chain. You must become experts in predicting failure rates, both individual units and at scale.
And you must create new reverse supply chains, from the consumer back to base or to redistribution centres. Both of these are new disciplines that benefit from predictive maintenance.
How to start using predictive maintenance
Predictive maintenance could help you unlock the ability to reuse product profitably in an environmentally friendly way while maintaining a positive customer experience.
For predictive maintenance, you will need the systems and equipment in place to collect data from connected devices to report back to base. You then have the option to proactively service a unit and extend its lifetime, or to recommend replacement and then bring a unit into its secondary market economy.
Through predictive logistics, you can put this foresight to work, optimising reverse supply chains, and turning them from reactive to proactive. And according to analysts Bain & Co, shippers with optimised distribution networks could expect to increase margins by up to 10%.
In the words of Talking Logistics: “Shippers and service providers must operate smarter, not bigger. This is where profit will be made, and where the leaders will separate from laggards.”
Core to predictive maintenance is collecting, storing and analysing data sets. On a high level, predictive maintenance relies on three layers:
- Collection – the accumulation of data, which is likely to come from sensors used on devices through the Internet of Things (IoT)
- Interpretation – data transferred through the collection is stored, analysed and interpreted
- Prediction – you can predict events through the calculation of event entry probabilities.
Here are some simple steps to getting started:
- Start small. If you’re starting from scratch, you’re bound to get some resistance with a new way of working. Start with looking at one or two critical assets initially with your predictive maintenance implementation – it’ll mean you can take care and analyse mistakes in the process before pursuing a large scale roll-out.
- Identify what assets to use for predictive maintenance. You need to ask yourself what assets would work for a predictive maintenance approach. When you first start, it’s probably worth using low-cost or expendable assets to test the reliability of your predictive maintenance approach before going to more mission-critical assets.
- Identify what resources you need. Of course, you need to put in place the tools and technology for predictive maintenance – such as sensors, cameras and software. But it’s not just technology around collecting data from assets you need. You also need to look at what you need in terms of people, training, materials, and physical space.
As a distributor, you should recognise that you’re more under the spotlight when it comes to sustainability and the environment. But there is evidence that the circular economy can work from a business point of view.
Gatwick Airport in the UK, for example, implemented a 10-year green strategy in 2010. By 2017, as well as a 10.5% reduction in carbon emissions and a 5% reduction in energy consumption per passenger, it achieved the goal of sending zero untreated waste to landfill.
With predictive maintenance, you can make efficiency gains that will help you save money and increase profitability. You have a significant role to play in engineering secondary markets, and the use of IoT data and AI for predictive applications could be central to your success.
The potential for predictive maintenance is inspiring and it’s certainly worth looking carefully at what it can do for you.