By Ranjit Nair, CTO and co-founder of Altizon Systems
IIoT adoption is growing, with expectations that it will reach $200 billion by 2021. Yet, those in industries using IIoT aren’t totally confident about their own implementations. According to a Cisco study, only quarter of respondents believe that their IIoT initiatives have been successful so far.
Why is the enthusiasm for IIoT offset by a lack of successful implementation?
It may be that industry is struggling to integrate the data collected from IIoT systems with backend systems. Until that happens, the various industries that stand to gain benefits from the IIoT won’t be able to build the stickiness needed to derive the full value of the data.
“Extracting insights from multiple data sources is a new goal for manufacturers,” explained Diego Tamburini, principal manufacturing industry lead, Azure Industry Experiences Team, wrote in a blog post. “But tracking data is often relegated to more than one solution—and each solution is created to solve a different problem.”
Tamburini also pointed out that industry might be making things more difficult—and more costly—in the way they use Key Performance Indicators (KPIs). Industries use multiple KPIs in production, meaning more expenses and less efficiency. “And,” he added, “given the disconnected states of these apps, there is no way to see the entire picture of operations.”
Building the right KPIs
“You have all these amazing protocols—every robot, every machine, every sensor is talking a proprietary language. The data has to be normalized across all of these different protocols,” explained Bryan Tantzen, a senior director at Cisco Systems in an Enterprise IoT Insights article.
Eliminating these silos and creating the IIoT stickiness necessary for improved implementation begins with building the right KPIs. But how do you know what is “right” or will enable optimum efficiency of your IIoT platform?
It takes a holistic approach—recognizing how the KPIs affect the entire system and then establishing a clear purpose for the performance measures. It’s also knowing what processes should be measured. Is it performance? Product availability? Manufacturing cycle? Should the KPI be built on a Time-to-Value model?
The KPI itself is only as good as the data it is tracking. And there’s the problem—when the IIoT and backend systems have data in silos, it is difficult to get a good understanding of the data. Seamless collaboration and constant communication, rather than silos, enable improved data collection and big data that is actually useful.
Data analytics is a top reason why many organizations turn to the IIoT, but they are getting minimum value from the data generated from sensors. As Medium reported, “How do we explain why in a gas rig with 30,000 sensors, only 1% of data is being used for decision making? Or why Industrial IoT data management remains a leading challenge for companies year after year?”
Good questions.
Intelligent data
The data is there, but it is under-utilized. To maximize benefits, data generated from IIoT sensors and stored in the cloud needs to be used in tandem with data coming from the backend systems. The problem is that these two types of systems have different business-operations goals and aren’t designed to work in tandem. In order to achieve data intelligence, organizations need to rethink their approach to IIoT implementation and how to best connect it with legacy systems.
Case in point: JK Tyre, a leading manufacturer, wanted to digitalize and IoT-enable its plant assets to bring visibility and predictability into its manufacturing value chain, improving asset utilization and reducing conversion cost-per-unit of production. In short, JK Tyre needed to integrate the IoT with ERP systems. By implementing a manufacturing-intelligence platform, JK Tyre is able to securely connect and process IoT data at scale. Most importantly, the company now has a 360-degree view of the line from the productivity, quality, energy, maintenance and traceability perspectives.
How to make data stickiness happen
IIoT data stickiness begins with the right IIoT platform. Everything needs to work like a well-oiled engine in order for the implementation of IIoT to be successful. It requires the right framework, one that’s easy to operate and scalable to multiple lines, function units, or event multiple plants. Key factors for choosing an IIoT platform include:
- Understanding your needs. The IIoT should be added only if it will bring value to production and if the data can be useful.
- Scalability. Can your system keep up with the amount of data you’ll be generating?
- Deployment flexibility. An organization wants an on-premise deployment to retain control or it may choose a cloud-based option for higher efficiency and cost savings. Above all, the platform must be flexible in providing different deployment options.
- Interoperability. It is this feature that brings IIoT in tandem with backend systems to create data-driven decisions using real-time data.
Backend systems often rely on older databases, with much of the data manually entered and manipulated. For IIoT implementations to run harmoniously with these systems, we must reconsider how we approach database design. Traditional databases aren’t able to keep up with the amount of data generated in a smart industrial setting and aren’t designed to keep up with real-time data.
Could AI be key to KPIs and data intelligence?
One of the main benefits of artificial intelligence in an industrial setting is the data analysis with minimal human interaction. AI can make sense of the vast amount of data generated by the IIoT sensors and sort through the data points that are most valuable to operations.
As IIoT makes a stronger presence in industrial settings, its success will depend on how well this technology is integrated with legacy systems. IIoT is all about data, and an organization’s ability to manage and coordinate that data will be the difference between success or going back to the drawing board to find another solution.
