One task that’s leveraging artificial intelligence (AI) is predictive maintenance, especially when it comes to the Internet of Things (IoT). The challenge is especially important to asset-intensive industries. One company that’s addressing the space is IBM with its Watson IoT technology.
To find out more about IBM’s work in this area, I talked with Greg Knowles, Program Director for the Watson IoT Portfolio Strategy.
Greg Knowles, Program Director, Watson IoT Portfolio Strategy, IBM
Why do companies need to think about maintenance differently? What are the challenges that companies in asset-intensive industries have with their assets?
A digital re-invention is occurring in asset-intensive industries that will change operating models in disruptive ways. Plant managers and reliability engineers are being pushed to improve operational efficiency by balancing costs with risk and performance. There’s a tension between spending more on maintenance programs that can lead to over-maintaining assets and cost cutting, which they know may leave them with costly exposures such as unplanned downtime—that can cost an average of $260,000 per hour! This tension is a real dilemma in the industry.
We’re finding that even organizations with the most mature Enterprise Asset Management solutions and programs are still experiencing mounting pressures to improve availability and predictability to reduce unscheduled downtime, manage costs of maintenance programs, and mitigate the risk of catastrophic failures that can impact operations.
These sound like challenges that the industry has been struggling with for a long time. What’s changed?
Three major factors have changed in recent years:
First, assets and equipment are getting smarter, more sophisticated, and sensor-enabled. The cost of sensors has drastically decreased, while OEMs themselves are delivering IoT-enabled equipment. This means there’s more and more data available about the health of assets—often in real time.
Second, easy and inexpensive connectivity with 5G on the horizon, cloud computing including multicloud technologies, and feature-rich data-management tools. These are all more readily available, cheaper to integrate, and getting easier to implement each day—this middle layer is extremely important for continuous data aggregation and fast time to value. In other words, it’s easier to collect data from these smarter assets and then apply tools and analytics.
Third, proven analytics, AI, and machine-learning technologies now enable us to learn from the data collected. Most industries have examples of leaders who are getting really smart about how they use all of the data now available to make better decisions. Technologies can easily be leveraged to understand current health of assets and predict failures in the future, in addition to many more use cases that we deliver to customers.
Many have heard about Enterprise Asset Management (EAM). This sounds like different technologies and solutions? Why is it a critical next step for people with an established EAM solution? Why adapt?
We call this set of analytics and AI-driven technologies “Asset Performance Management” (APM). APM is all about building on your EAM foundation to drive better asset decisions and optimize your maintenance strategy. Asset-intensive organizations now have the tools available to effectively minimize unplanned repair work and costly downtime, increase asset resiliency, lower maintenance costs by decreasing over-maintenance of assets/blindly performing preventative maintenance, and reduce the risk of equipment and system failure.
For many organizations, the shift to leveraging analytics and AI to improve asset-related decisions will be a long-term journey. They struggle to know where to start. These organizations can have hundreds of different asset classes and thousands of assets that they track. In our experiences, it’s very important to start on this journey today. The value driven by data collection and AI is exponential, and delays in starting may mean uncompetitive operating costs, greater unplanned downtime, and poor return on assets in the future. Organizations often will start with an important asset class, look for quick wins, and use these wins to prove out the ROI.
Many organizations have become very good at maintenance execution. But to reach new levels of business performance, they will need to get even more out of their assets. Industry leaders have already started, and those that delay will find themselves at a competitive disadvantage.
What's the role of AI in an Asset Performance? Why is AI so important?
AI will transform how organizations financially optimize asset performance decisions. So many variables—both internal and external—can impact asset performance. The sheer number and dynamics of these variables mean that standard analytics just can’t keep up as conditions change.
Take a manufacturing scenario in an environment where demand is variable. When should a critical asset on the manufacturing line be taken out of service for routine maintenance? What’s the risk of failure and how does this change depending on operational cycles and the maintenance schedule? What’s the impact to the rest of the supply chain if maintenance activities are performed?
AI is needed in this situation. In highly dynamic environments, AI enables customers to better identify critical assets (plan), understand current asset condition (health), anticipate failures before they happen (predict), and optimize maintenance and repair (assist). Simply put, AI enables the delivery of actionable insights that helps organizations get more out of their critical equipment.
The bottom line: AI-embedded solutions provide a faster payback and significantly greater ROI.
What’s required to get an APM solution up and running--how does an organization know it's ready for APM?
When it comes to APM, there are many entry points, depending on customer requirements and maturity. Organizations should start with condition monitoring and anomaly detection to better see the health of their assets; predictive maintenance to minimize unplanned downtime and optimize maintenance scheduling based on predicted asset failure; or AI-based solutions to better inform their technicians as they diagnose and repair increasingly complex equipment.
We see customers implement one module at a time, and it’s important to get quick wins. To determine the best starting point, organizations need to work with their vendors to understand the unique challenges and develop a list of highly critical assets that will deliver the highest and quickest ROI.
Can you talk about examples and benefits you have seen from clients that have embraced APM? What are the long-term benefits?
APM can benefit organizations across industries such as energy and utilities, oil and gas, travel and transportation, and industrial manufacturing. We have worked with several local water utilities that are struggling to keep up with manual asset inspections. These utilities are now leveraging condition-monitoring solutions to improve the accuracy and efficiency of inspections, so that they can optimize maintenance versus replacement decisions.
We’re also working with a global elevator company that’s leveraging predictive maintenance to increase first-time fix rate by 25% and decrease customer complaints by 60%. One mining equipment OEM has even incorporated predictive maintenance into its equipment, which can help operators anticipate needed repairs. The benefits are so substantial with up to a 10% increase in productivity that this OEM can offer such a capability as a new revenue-generating value-added aftermarket service.
Based on our observations and experiences, the benefits of APM are very real. Circling back to our earlier point, it’s important that organizations start on this journey today. Identify those critical assets today and start with projects that will lead to quick wins and help you prove the ROI.
Greg Knowles is the Program Director for Watson IoT Platform Analytics and is responsible for analytics offerings and strategy for the Watson IoT Platform. He works closely with customers to understand industry use cases and deliver analytics and cognitive offerings to help them leverage asset and production data and to gain insights and drive bottom-line business value. He has delivered solutions for a variety of industries including oil & gas, energy & utilities, healthcare, automotive, heavy equipment, transportation, and retail. Greg has been with IBM for 20 years, serving in various roles from offering management, solution delivery, and technical sales to architecture and development.