Putting data to work with predictive maintenance

One of the areas that big data and analytics has the greatest potential to transform manufacturing environments is in the world of predictive maintenance. The potential of data science to improve how machines are monitored and maintained has long been recognised, with the roots of today’s data-driven approaches to condition monitoring being formed in the aerospace and defence sectors over 30 years ago. 

However, the labour intensive nature of analysing and forecasting machine health has limited its use to heavily regulated sectors and the most critical production assets only. It has, traditionally, relied on gathered and spot the crucial signs that a known fault might occur again in the future. 

Although expensive and time-consuming, this approach did enable organisations to understand when a component or machine was getting close to failure and even predict when it would no longer be able to perform its intended function. The discipline, called prognostics, improved maintenance by enabling engineers to apply the right intervention before failures could occur. 

Recent advances in Artificial Intelligence (AI), automation and analytics,has automated prognostic analysis and it can now be achieved by a computer and deployed affordably at scale in more down to earth settings.

It is now possible to collect and analyse a range of data sets from most machines which can contain the clues required to spot early signs of wear and tear.

Software can be introduced to analyse data collected from machines automatically, using smart, self-learning algorithms to diagnose and predict failures.

This automated approach to predictive maintenance is transformative for how manufacturers manage and invest in machinery. Factories can now undertake condition monitoring automatically across thousands of machines, minimising downtime by ensuring machines are scheduled for maintenance ahead of failure without dedicating hundreds of human hours to gathering and analysing specific information.

Simple alerts can be set up to provide information about matching failure models and the remaining useful life of each piece of machinery. Engineers can look at a simple dashboard each morning to see where their efforts would be best applied and when. An automated approach to condition monitoring and predictive maintenance should offer an accessible way to reduce machine downtime and increase the overall efficiency and performance of their production environment.

Before introducing predictive maintenance or condition monitoring, manufacturers first have to understand exactly what they want to achieve. Qualifying the goal is key to working out the validity of condition monitoring. 

Cost is one of the best ways to quantify potential savings and justify the use of condition monitoring but this isn’t always a straightforward process as downtime can have many hidden costs beyond simply the loss of production, including the price of spare parts, labour and scrap product.

The gains from condition monitoring must clearly outweigh the cost, but there is more to consider than the cost of implementation and downtime avoided. Businesses should determine the total potential savings, including prices of assets with a longer lifespan as part of their return on investment figures. The initial financial gains will increase over time due to continuous improvement.

While an automated approach to prognostics and condition monitoring represents a new way of working for many, the benefits more than justify the change. Organisations that have already embraced this approach have reduced maintenance costs by 40%, halved their levels of machine downtime and delivered dramatic improvements in throughput, quality and margin.

While predictive maintenance has been practised for 30 years, it is only in recent years that advances in prognostics and automation allow such sophisticated approaches to condition monitoring to be applied factory-wide. Given the financial and operational benefits of introducing predictive maintenance at scale are now much greater than the cost of implementing and managing such programmes, it is hard to conceive any large scale manufacturer not using them in the decades to come.