The Internet of Things (IoT) has gone viral in manufacturing, from mahogany row to the blue-collar muscle on the front lines. Yet many still struggle with justifying these projects. Meanwhile, visionary manufacturers and fast followers have found their big payback in what I call the Analytics of Things.
Here are some ways that companies are justifying IoT projects:
Condition-based maintenance (CBM) is often the first step in the Industrial Revolution 4.0. Whether in the product being built or the manufacturing line itself, sensor data is used to optimize product quality and production. CBM uses the sensor data to predict component failures via advanced algorithms. This allows repairs to be scheduled before the device breaks down. For any large machine – engines, cars, trains, etc. – CBM shifts many repairs to planned maintenance instead of unplanned. Repair labor is applied with fewer panics, and more proactively. This yields less downtime, lower costs, and happier customers. If your company hasn’t deployed CBM solutions, start immediately. Companies like PTC Axeda and SeeControl have been helping manufacturers do this for years.
Root Cause Analysis
It’s hard for engineering teams to design for low cost and high quality in just the right balance. With sensor data, they are finally getting the information they need to move beyond gut feeling. Did they pick a low-quality part? Did the supplier subtly change its goods? Were there three good parts that don’t work well together? That’s one root cause that stumps engineers most: multiple high-quality components that, together, add up to a failure. But data and a good statistician can spot correlations, anomalies, and false positives. One manufacturer using sensor data told me, “Now, every product generation is truly better than the last. This also gives us a stronger position during supplier negotiations.” His relationship with suppliers becomes more collaborative. They share data when a component supplied is less than ideal. They help the supplier to design its product better.
Integrate Data – and Check on the Customer
Manufacturers get useful results from analyzing sensor data in isolation. But that’s the low-hanging fruit. According to Michael Porter and James Heppelmann, “This new-product data is valuable by itself, yet its value increases exponentially when it is integrated with other data, such as service histories, inventory locations, commodity prices, and traffic patterns.”
Everyone we have talked to says the toughest quality and efficiency problems exist across the entire supply chain. One manufacturer told me, “That’s where we need the bigger magnifying glass.” It’s where huge paybacks and competitive advantage lurk. To find these, sensor data is combined with supplier, production, shipping, and customer-usage data. This means a lot of executive backing and data are needed. Industry-specific consultants with analytic experience are needed to help executives see the possibilities. It starts out as executive vision. Properly executed, it ends up as solid ROI.
Consider, for example, a hard disk drive (HDD) manufacturer and its OEM buyers. The HDD field quality engineers work constantly with Dell, HP, and other OEMs. Now imagine that several batches of PCs exhibit HDD failures in corporate offices three months after installation. A tiger team working with the OEM vendor shares customer support data using serial numbers of the HDDs and PCs. The data eventually correlates customer complaints, pointing to a single batch of 1,000 disk drives. Or maybe it shows that the PC manufacturer changed a cooling fan in new models. The tiger team identifies the cause of the defect, traces product distribution, and replaces specific units via the OEM’s customer support channel. The tiger team now prevents further equipment failures. They can also proactively notify consumers of any safety concerns. And data feeds back into the root cause analysis. The result is increased loyalty of the OEM companies and consumers.
With the lower cost of sensors and analytics these days, there are plenty of ways to avoid risks. When sensors were $30 apiece, it wasn’t feasible to have them throughout the product or value chain. Today, sensors are approaching trivial cost levels – we can’t afford not to deploy them. I’m not saying the cost of sensors and analytics is small. Au contraire. I’m saying the return on investment is huge.
For example, poor designs or bad parts show up on the balance sheet as warranty costs of returned goods. The CFO maintains cash reserves to cover possible liabilities. It’s idle cash, insurance. And every few weeks, a major product recall slams a manufacturer – publicly. Corporations find themselves replacing thousands of products for safety or failure reasons. The costs of replacement goods, shipping, labor, and refurbishing are staggering. These problems go directly to the bottom line of the financial statement.
But with sensor data properly leveraged, product returns drop while quality soars. I met one manufacturer who said warranty costs are half of the nearest competitor, to the tune of $10-$20 million annually. He said that paid for the sensors and analytics in the first year.
While much of the Industrial Revolution 4.0 began with heavy assets, it’s not limited to them. A few quick examples:
- Abbott Diabetes Care is a sensor that people wear on the upper arm to track blood glucose without the daily finger prick problems. Doctors download measurements from patient’s mobile phones to provide more accurate treatments.
- Utility companies using smart meters have been able to analyze meter readings every 5-15 minutes to identify stolen meters.
New Business Models
Using sensor data only for cost efficiency and product quality misses half the value of the Analytics of Things. With a wave as big as the Industrial Revolution 4.0, we must all think of new business models and new revenue streams.
Most people haven’t noticed the unusual startups in Silicon Valley. They go by names like WalMart, BMW, Nokia, Bosch, and GE. A large number of manufacturers are establishing software outposts in order to innovate beyond their comfort zone. They are becoming Silicon Valley hipsters. Manufacturing companies are adding software products to their portfolios. Many are investing in the Analytics of Things.
Bosch claims that as many as 600 programmers are building IoT software. In one case, its apps track GPS sensors on industrial torque wrenches. It also remotely monitors fluids in vehicles. Consider AT&T’s moves on the connected car. AT&T supplies a software kit for developing IoT applications. Analytic feedback goes to the consumer or the manufacturer. AT&T helps “monitor cargo, homes, vehicles, and containers around the globe.” AT&T has inserted itself in the Analytics of Things dialogue. Watch out Amazon! And Siemens gets “reduced labor costs and improved fix-time/fix-rates” from digitalization of its Internet of Trains.
The Analytics of Things helps to keep engineers on a steady diet of data. Mathematics is in their heart anyway. The good news is the Analytics of Things is still emerging. You haven’t fallen behind yet. There’s plenty of time to be visionary or a mainstream adopter. Find a profitable use case that fits and start the proof of concept and pilot projects. Keep it small. Fail forward. Learn. Win.
With over 30 years in IT, Dan Graham has been a DBA, Strategy Director in IBM’s Global BI Solutions division, and General Manager of Teradata’s high end servers. He is currently Technical Marketing Director for the Internet of Things at Teradata.
Subscribe to Data Informed for the latest information and news on big data and analytics for the enterprise.