Most companies today are already collecting large amounts of high quality operational data in real-time. But instead of looking to drive value with this existing data, many are focusing on collecting more. They should realise the opportunities of using advanced analytics with the data they already got, to drive improvements in maintenance and operational processes.
As the President of Digital Solutions at Maintpartner, can you briefly describe what do you want to achieve with Maintpartner? Who can benefit from Maintpartner, in what ways? What do you offer to your customers?
Maintpartner specializes in industrial maintenance outsourcing and project services. In the Digital Solutions business unit, we help the customers to minimize downtime and solve operational issues by using their data with advanced analytics.
Most of our customers are taking their first steps in digitalization and are typically moving their operational data to cloud for the first time in our projects. This obviously raises concerns e.g. on data security and involves a lot of learning from the organisation. It’s also very important to show tangible results from these first projects to ensure top management support and funding.
Using existing data with advanced analytics is a low-hanging fruit for most industrial companies. It doesn’t require large up-front investments and can result in very fast payback leading to more funding and more success – a virtuous cycle speeding up the digitalization process.
What are the challenges your customer face during their Digitalization journey? And how do you help to tackle them?
One of the main challenges our customers face is making their data available in real time for external partners. As mentioned, most companies are just getting started in storing operational data in the cloud and this often requires different parties to come together to make it happen. We have been working with many customers in the same situation and can advise them on options and their pros and cons.
The other technical challenge related to this, is data security. While use of cloud-based solutions is becoming the norm today across different enterprise applications, using cloud for operational data is seen as requiring much tighter security measures. Operational data often includes very specific insights and intellectual property that are strategic to the business and should be protected accordingly. This is a discussion we always take up in the first meetings to ensure security aspects are not neglected.
While these technical challenges require a lot of effort, the most important change is that you need to be able to react to new insights in the operational level. This requires organizational and management support. Operational and maintenance personnel are dealing with lot of information and systems – and if they are not engaged to the activity, they could see the new data as ‘just another screen’ and additional work. As part of the deployment we need to understand how the solutions fit into the day-to-day activities and ensure that the insights are acted upon. This is how the business benefits are realized.
To ensure that this happens and customer sees that value, we deploy the solutions for a minimum of 6-12 months where we require our customers to commit resources so that they see the real benefits. Understanding that using analytics for operational data is not only a technical project, but also a transformation which touches on maintenance and/or operational processes, is important.
In your keynote, you will present real-life examples of applying machine learning to existing process data to minimise downtime and improve process efficiency. Can you briefly explain how can companies apply machine learning to existing processes? And how can machine learning decrease unplanned downtime and improve process efficiency?
I will talk about how we are using machine learning and anomaly detection in the industrial domain today. Detecting very small process disturbances enables our customers to react early to issues that might otherwise lead to unplanned downtime, higher production costs or quality issues for example.
By using large amounts of reliable process data and our fully automated end-to-end analytics pipeline, we can quickly build a picture of what ‘normal’ looks like and start observing for any deviations from that ‘normal’. This approach enables us to identify very small deviations when all individual parameter values are still completely normal, but the combination of multiple correlating values is somehow unique.
One example is from power plants we work with. In one situation we identified turbine vibration levels that while well within acceptable limits, they were abnormal compared to previously observed vibration levels in a similar situation in the past. As a result, the turbine was inspected, a serious problem identified and unplanned downtime prevented. In my presentation, I will also share findings from metal and pharmaceutical industries.
You will also host a roundtable discussion on the typical challenges companies are facing when trying to apply Machine Learning to their operational process data and how to overcome these. Are there any usual suspects when it comes to challenges?
Yes, we will be touching upon for example how to transfer operational data to the cloud in real-time and what challenges we have seen in the field. We will discuss security aspects related to making data available. And finally we will also explore what organisational issues are related to the deployment.
What are the biggest trends in Enterprise Asset Management now? Why is that? And how will they evolve in the future?
From our point of view, we see most traction in
- Using advanced analytics for predictive maintenance.
- Using advanced analytics for OEE optimization (performance, quality)
- IoT: deploying new sensors and visualising meaningful data in the factory floor for immediate decision making
Some of the trends that people I spoke with mentioned IoT-enabled assets, predictive maintenance, smart logistics, sensors, smart glasses and flexible asset management. What are your thoughts on this? Do you have any customers that really excel in these areas? Can you give specific examples?
Augmented/Virtual Reality (AR/VR) is getting quite a lot of press but has had limited impact for our customers as of yet. I believe it will take a few years to have a real impact as technology matures but once it does, it opens up interesting opportunities e.g. for remote assistance in maintenance. In addition to AR/VR, using 360 video and laser scanning to model sites for planning and virtual trainings are interesting areas to keep a watch on.
What made you decide to speak at EAM NXT?
We are currently dealing with customers in the energy and industrial sectors mainly in the Nordics but are expanding to Europe. What we are doing in the analytics space is suited for asset-heavy industries anywhere and we would like to learn more about their requirements and get their views on what we are doing. It is a great opportunity to engage with leading companies and understand what they are doing in the analytics area.
What are your expectations from EAM NXT?
To have meaningful discussions and learn. Especially about how industrial companies are approaching digitalization: what steps are they taking, what solutions are they looking at, how they have organized, etc.
Any message you want to share the participants of EAM NXT?
There is already a lot of data already collected by everyone who will be joining the conference. We are looking forward to having meaningful discussions and seeing how they can use this data.