Could the Industrial Internet of Things (IIoT) change the way automation and operational procedures work? Yes, it could. But only if manufacturers are willing and able to change the structural framework upon which decades of industrial processes have been built. To be clear, that doesn’t mean a rip and replace of machines, control systems and networks. It means opening it all up and adding to it.
Data is quickly becoming the most significant strategic asset on the plant floor. That’s because, when applied correctly, it can create new business models and new streams of revenue. According to a report from McKinsey Global Institute, the potential economic impact of IoT applications will be as much as $11.1 trillion per year by 2025. Factories, which will focus on operations and equipment optimization, represent the largest potential of up to $3.7 trillion.
Sound far-fetched? Think about this: Ninety-nine percent of all data in an oil rig is lost before reaching decision makers, according to the McKinsey report. In other words, only one percent of data from an oil rig with 30,000 sensors is examined. This data is used for anomaly detection and control, not optimization and prediction, which provide the greatest value.
The obstacle in front of the industry today is connectivity, and many groups and vendors are working diligently to solve interoperability issues. But interoperability only gets you so far. In order to obtain value from the 99 percent of data lost on that rig, you need analytics.
Enter Sight Machine, a four year old company that has developed an analytics engine dedicated to manufacturing. The Sight Machine Manufacturing Analytics platform, which was released in the last year, includes proprietary models to perform analysis on the most common types of discrete, batch and continuous process operations.
Sight Machine captures and collects data from a variety of plant floor sources, including PLCs, sensors, barcodes, even unstructured formats like audio files and photos. Data acquisition happens through plug-in representational state transfer (REST) APIs and adapters for connecting to legacy systems, from homegrown MES to custom factory IT systems. The key here is the technology’s open source model, which co-founders Jon Sobel and Nathan Oostendorp are very familiar with.
Sobel has been on the management teams of Tesla Motors, Yahoo, and the open source community SourceForge. Oostendorp worked as an architect for SourceForge and co-founded Slashdot.org. Oostendorp also worked at an automotive factory in college doing quality control programming—which was on his mind several years later when he began to dabble in big data.
“The key to what we do is combining structured and unstructured data types in the same way as Google or Facebook, but we do it for manufacturing,” said co-founder Sobel, who is also the CEO of Sight Machine. The challenge in the manufacturing environment is the volume, velocity and variety of data types, which makes it difficult to work with and analyze the data at scale, he said.
The complexity of managing all that data might be the reason why many companies have yet to tackle this aspect of IIoT.
“Most IoT [companies] talk about communication and moving things around, but we believe that problem is well on its way to being solved,” Sobel said. “The core challenge is how to effectively analyze data and use the same models over and over, from one plant to another, to [provide] useful insights.”
Sight Machine is currently engaging with companies in the pharmaceutical, automotive and apparel industries. In these early implementations, customers are asking the company to help with capacity planning and making sure processes work efficiently and reliably. In fact, many companies want help with the exact same problem: They have similar assets in different locations with performance discrepancies. The same machines—the same processes-- but what is going well in one place is not going so well in the other.
The Sight Machine engine is a set of data models replicating how a production process works. It takes in hundreds of parameters, organizes it all, and transforms the data on the other side. The “other side” being a server on the plant floor providing real-time big data analysis of quality, traceability, and operational information. The cloud also factors into the equation for data storage.
According to Gartner, which recently named Sight Machine in its 2015 Cool Vendors in Manufacturing Operations report, the analytics layer includes machine-learning algorithms that enable the rapid pinpointing of issues in mass production. This is what drives real-time alerts for operators and “is a far cry from the traditional OEE dashboard reporting that [just] provides a snapshot of production performance.”
Most importantly, applying the Sight Machine engine does not require re-architecting the plant floor systems. The point is to provide a way to add data acquisition, management and analysis without disrupting the existing IT infrastructure. “We call what we do a ‘data blanket’ because we sit on top of everything,” Sobel said. “There is no rip and replace.”
Indeed, Sight Machine could be seen as a seamless evolution in the industry, and perhaps an introduction to the future of modern manufacturing that includes IIoT.
