The big data boom is transforming the industrial sector although managing that data can be a challenge.
The Internet of Things' industrial applications promise even more value than the consumer applications like smart homes, connected cars, or wearable devices that garner more public attention.
The industrial internet of things (IIoT) is poised to transform everything from the factory floor to the logistics of shipping, receiving, and maintaining products at the customer site. Isn’t it about time you got serious about taking the IIoT to work with you?
Manufacturers are already using the IIoT for real-world applications, such as preventative maintenance on manufacturing equipment. Manufacturers traditionally would use informed guesswork to time maintenance intervals. With IIoT and predictive analytics, they can time this maintenance based on a granular analysis of data, such as contextual information around prior downtime incidents. For example, you might be able to anticipate that Production Line 1 is more likely to break down this week because it produced X, Y, and Z at a variety of temperatures. But Line 2 only managed to churn out Product X at a consistent temperature, so the model indicates it can wait four more months for maintenance even though it is using the same types of equipment.
The massive amounts of data generated by embedded connected sensors are enabling new insights, and presenting new challenges, in this approach to predictive maintenance. While many are tempted to use all of the data, that is often not necessary. Homing in on the right data, however, can present many surprises.
One chemical manufacturer, for instance, was having difficulty identifying why batches were failing intermittently. Combing through plant data yielded no clues. But adding context brought surprising insights. When the firm added weather data to the model, it became clear that humidity levels outside the factory were the biggest predictor of batch failure. Now the company alters its production schedule when the weather forecast calls for high humidity.
A Proliferation of Data Drives New Opportunities
Ongoing improvements in sensor technologies—including miniaturization, performance, standardization in communication protocols, cost and energy consumption—have made intelligent products more accessible. Manufacturers are increasingly installing sensors in production facilities and in the industrial goods they sell. And these sensors are producing data at an unprecedented rate.
Consider that a twin-engine Boeing 737 aircraft generates 333 GB of data per minute per engine, which means a flight from Los Angeles to New York generates roughly 200 TB of data. An IIoT-ready oil and gas drilling rig produces 7 to 8 TB of operational data a day. Connected automobiles can generate more than one petabyte (PB) of operational data daily.
By applying advanced analytics to the generated data and to relevant external data, companies can gain a better understanding of the interactions among input variables and what it takes to achieve desired business outcomes. Using that knowledge helps develop innovative products, services, and new business models such as ongoing monitoring contracts or remote maintenance services.
What’s involved in enabling advanced analytics on IIoT data?
But realizing the promise of the IIoT won’t be easy. The technological complexity involved in applying analytics to the data can seem staggering to those who are familiar with only one side of the equation – either the sensor side or the big data and analytics side. Success means bringing the two sides together.
Making effective use of IIoT involves ingesting (loading and processing) data at rates that far exceed what is usual today. It also upends dominant analytical patterns. Traditionally, analytics have been performed on data at rest—data that’s loaded into and consolidated in a single place at one time. Analysis in the context of IIoT shifts the emphasis to data in motion—so-called “streaming analytics.” This shift to streaming analytics requires a shift to new technologies, new skills, new capabilities with existing resources, and considerable learning and creativity.
For industrial companies, extracting meaningful insights from streaming data is made simpler and more accessible with purpose-built applications that address perceptual and production quality, field quality, and asset performance analytics. These solutions bring together numerous enabling technologies—from streaming analytics to rule-driven decision automation capabilities—all in one package.
Without such a platform, organizations often struggle to cobble together an IIoT-enabled infrastructure from scratch. This requires combining large numbers of disparate technologies to enable data management, model development, model deployment and model management.
Companies must also develop new practices for dealing with streaming data on a massive scale—processes that typically don’t already exist inside the organization. Managing this information with effective governance is a significant challenge, as is deciding which information is relevant and which can be discarded. Analytic models that can be “tracked and traced,” as well as continuously improved, are the models that will help an organization to grow their analytics capabilities sustainably.