A new artificial intelligence (AI)-powered asset monitor from IBM gives manufacturers the chance to move into the next phase of the Internet of Things (IoT) evolution. The hard part will be everything else: Making sense of the data, revamping existing processes, and retraining technicians and engineers.
Kareem Yusuf, general manager of the IBM IoT business unit, said IBM’s goal for the Maximo Asset Monitor is for companies to draw insights and take action based on data from the assets that they’re managing in Maximo.
“We want to help clients get new data or make existing data understandable,” he said. “It’s not just about collecting data but bringing it into a cohesive stream and linking it to historical data in a consistent way.”
“Companies are ready to take this on with multiple asset classes and with multiple processes and really move things forward,” he continued. “These assets become core to how these companies operate, and you have to have a holistic way of managing the physical assets and knowing their state.”
Reid Paquin, research director, IDC Manufacturing Insights, said asset-intensive companies such as utilities, oil and gas, metals, mining, pulp and paper, and chemical producers will find this mind of monitoring most helpful in eliminating downtime.
While Brian Hopkins, a VP principal analyst at Forrester, said that the Maximo news is an example of the trend of tech vendors embedding AI into use cases and industry-specific solutions. Hopkins added that this new capability is quickly becoming table stakes in the industry and that success with AI solutions depends on how well the underlying data is managed.
SEE: Special report: Sensor’d enterprise: IoT, ML, and big data (free PDF)
“Data management in industrials is difficult because so much data comes from devices and never makes it into enterprise data lakes,” Hopkins said. “We think that buying a solution like Maximo, without also undertaking data management at the edge may be more frustrating than beneficial.”
IBM plans to build on Maximo’s capabilities by integrating OpenShift hybrid cloud capabilities from RedHat, Watson Studio, Watson ML, and other core technologies into the platform.
The early phases of digital transformation
For manufacturers, there are four basic steps in the evolution to this new data-centric business model:
- Doing an asset inventory
- Implementing data collection
- Organizing and synthesizing the data
- Using the data to develop predictive analytics and operational processes
Yusuf said that the bulk of IBM’s clients are in the first phase of their digital transformations and are using an enterprise asset management system for basic tasks. A smaller group of companies are just beginning to capture data from existing assets.
“We are at the cusp now where people need to track the instrumentation problem,” Yusuf said.
Paquin agreed that the majority of manufacturers still have work to do. “Many manufacturers still need to make infrastructure upgrades to put a Strategic Asset Management program in place,” he said. “Also when the assets are connected, being able to manage all of the data that is now accessible is still a top challenge we see when talking with manufacturers.”
The next phase of the transformation is having enough coherent data to monitor the health of various assets and build a system of predictive maintenance alerts. Yusuf said only a few companies are using the data to predict equipment failure.
“Some of our clients have gotten to that level, but I would say that this is in terms of pilot projects rather than dramatically at scale,” he said.
Barriers to digital transformation
Hopkins of Forrester said manufacturers need to move beyond the idea that technology modernization is a sufficient data strategy.
“In our CIO 2020 predictions, we predict that advanced firms are going to recognize the true enterprise cost of data management and double or triple their data strategy budgets,” Hopkins said.
Yusuf said he sees two common hurdles that companies face at the start of the process of connecting processes and machines to each other and to the web. The first one is deciding where to start with the transformation, given the scope of the work.
“About 80% of the equipment out there has to be retrofitted to bolt connectivity onto it and then it has to be connected,” Yusuf said.
Companies have to decide which asset classes to start with and map out a plan for collecting data and making it useful at the same time.
Hopkins said that CIOS will need to significantly increase data storage and processing capabilities.
“The cost of data transaction processing in industrial IoT scenarios will outstrip storage and centralized cloud processing,” he said.
Another crucial element in this transformation is ensuring that it benefits the technicians who fix and maintain the physical systems.
“You have lots of data being thrown at technicians and engineers, and they want to know how this transition will make their jobs easier,” Yusuf said.
Yusuf added that improving safety on the manufacturing floor and establishing collaboration platforms for knowledge transfer are priorities for the next phase of IBM’s work.