Industrial Internet of Things: data mining for competitive advantage

Forget clever fridges sending you pitiful texts that they’re out of milk.

The Internet of Things is coming; we’ve all heard the buzz, we are queuing to get the latest smart thermostat because we’ve been told so for nearly a decade. While we wait for Alexa to finally be able to turn on our TV, brew our coffee, dim the lights and tune our car radio to our favourite station, the IoT is quietly having a much more significant impact outside our homes: in the enterprise.

Dubbed the Industrial Internet of Things (IIoT), it’s seen factories and plants stuffing their machinery like currant buns with ever smarter, cheaper, and smaller internet-connected sensors.

The idea is that these sensors generate huge amounts of raw data that is analysed in real time and translated into recommendable actions for factory operators.

The aim is two-fold: to optimise performance and boost productivity while cutting costs and waste, and to be able to anticipate imminent machinery failures.

However, this connectivity comes with the risk that the enterprise can be subverted and damaged by malicious hackers.

Not everyone is there yet. According to a survey by PwC, in the US only a third of manufacturers are collecting and using data produced by smart sensors to improve their manufacturing or operating processes. And only a third believe it is “extremely critical” that American manufacturers adopt an IoT strategy in their operations.

The trend is clear, though. A recent worldwide market study by Berg Insight predicts that by 2020 there will be 43.5 million IoT devices in industrial automation.

Already today, automation is reshaping the enterprise landscape, creating big and small intelligent enterprises across industries and markets from retail and transport to the healthcare, manufacturing and mining sectors.

But what good does it do if your tool or engine is internet-connected? How can IoT help the industry?

“In today’s on-demand world, consumers expect shorter delivery times. Simultaneously, the number of shipments is increasing,” says James Morley Smith of Zebra Technologies, a firm that builds tracking technology for making businesses smart. To satisfy the consumers, companies turn to mobile devices and IoT solutions that help the warehouse become more efficient, as well as monitoring equipment to prevent sudden failures, and keeping control of the logistics chain.

“The IoT has enabled everything from improved efficiency for global shipping networks to devices that receive environmental feedback from home appliances and minimise their energy use”

For instance, say your company sends fish from Asia to the US. The fish is first kept in the ocean for a number of days, then loaded onto a third-party lorry, then a train, a ship, another lorry, and finally gets to the firm’s distribution centre somewhere in New York. If you’ve got all the journey’s parts internet-connected, you will know that the fish is kept at a specific temperature and humidity for the entire trip, and you may even have video of the journey in real time.

“The IoT has enabled everything from improved efficiency for global shipping networks to devices that receive environmental feedback from home appliances and minimise their energy use,” says Smith.

Currant-sized sensors are the basis of the industrial automation ecosystem, providing additional data that otherwise can’t be gathered, says Steve Garbrecht, director of product marketing for Brilliant Manufacturing software at GE Digital. IIoT software “allows the integration of operational silos,” he adds, combining vital data that otherwise would be stuck in their own production cycles.

This provides “additional insight and decision-making capabilities in the company,” he adds, while analytics and the ability to extract data from very large datasets offer insights that could not be gleaned before. And then of course there is innovation in hardware, machine learning, wireless connectivity, 3D printing, energy harvesting and cloud solutions, all enabling seamless data exchange between devices, systems and people and thus improving the enterprise’s performance, flexibility and responsiveness.

For instance, when King’s Hawaiian bakery, a client of Rockwell Automation, installed 11 connected machines in a new factory a couple of years ago, it was able to double its output of bread. The machines are all connected via software, which both gives employees remote access to real-time data and shows a detailed picture of the system, enabling them to monitor performance.

The technology has speeded up the time to market, improved asset utilisation and optimisation, cut costs and increased workforce efficiency.

Apart from boosting productivity, all these disruptive technologies can also help companies make sure their enterprise stays healthy – and works for longer.

The more advanced and automated our factories become, the more they can be thought of as a digital network, says Brian Carpizo of the advanced analytics group in McKinsey’s Operations practice. Advanced analytics techniques like machine learning are powerful new tools, he argues, as signals from the noise of machine sensors can spot machinery problems early and predict imminent failures.

Then there’s also the post-?production element of the IIoT concept. Smart products can gather data about their own operation and ‘phone home’ to the manufacturer if something goes wrong. Systemic faults will become visible in real time.

“All of this is resulting in significant savings in time and stress. In fact, initial field trials have already shown that service maintenance interventions with HoloLens can be completed up to four times faster than before”

Some companies, such as Emerson, Honeywell, GE and Yokogawa, try to sell full integrated industrial automation systems, while others, like Electric Imp, Particle and Arrayent, offer platforms that manage the end points of all the billions of ‘things’ that will one day be connected to the internet.

With machine-to-machine communications at the heart of the IIoT, connectivity is the focus of companies such as Jasper Technologies, recently bought by Cisco Systems for $1.4bn, alongside Siemens, Ericsson, Cisco, Belden, Moxa, Schneider Electric and Eaton. Then there is plenty of investment in software platforms and analytics for applications, by companies like GE, ThingWorx and Uptake.

For its part, GE’s allegiance to IIoT is clear: the company estimates that a mere 1 per cent boost in productivity across its global manufacturing base will bring $500m in annual savings. Even such an infinitesimal productivity improvement could add $10-15tn to worldwide GDP over the next 15 years.

GE uses IIoT solutions both externally, for customers, and internally. Its Brilliant Factory data feedback system comprises a wide range of sensors connected to industrial machines. The information the sensors generate is sifted using big data analytics and predictive algorithms, with one main aim: to improve productivity. For instance, at GE’s Durathon battery factory in Schenectady, New York, thousands of sensors on the assembly line and in every single battery help managers find out the status of production in real time, locating bottlenecks and taking corrective actions to improve the production flow.

GE has even created an ‘app store’ called Predix.io, which is based on GE’s Predix cloud operating system and makes it possible for industrial customers to develop their own apps. Predix is aimed at larger-scale activities such as power distribution and generation, aviation fleet management, railway logistics, oil and gas operations, hospital resource optimisation and manufacturing.

Procter & Gamble, for example, uses GE’s software to create scorecards that can be customised for specific purposes such as monitoring production compared to the original goals.

Another pioneer in the area is ThyssenKrupp, which uses the IIoT to improve servicing its lifts – especially when it comes to providing precise diagnostics and transparent and immediate data, but also to enable predictive maintenance to keep downtime to a minimum. “We are already incorporating this in the field through MAX – our predictive maintenance solution based on Machine Learning and Microsoft Azure Cloud technology – which aims to cut unavailability periods by half and further improves efficient mobility in our cities,” says ThyssenKrupp chief executive Andreas Schierenbeck.

Data collected in real time from millions of connected sensors in lifts is sent to Microsoft’s intelligent cloud, where an algorithm calculates the remaining lifetime of key systems and components in each lift. Since MAX can instantly identify the need for replacements in components and systems before the end of their lifecycle, it can effectively halve downtime by making the lift ‘tell’ service technicians its needs, in real time, says Schierenbeck.

"the idea of connecting the physical and digital worlds to drive innovation, efficiencies and global economic growth is the Intelligent Enterprise"

The company is also beginning to use Microsoft’s HoloLens internet-connected virtual reality goggles, to enable service technicians to visualise and identify problems with lifts ahead of a job, and have remote, hands-free access to technical and expert information when on site. “All of this is resulting in significant savings in time and stress. In fact, initial field trials have already shown that service maintenance interventions with HoloLens can be completed up to four times faster than before,” says Schierenbeck.

Honeywell has jumped on the IIoT train too, working with Intel to develop solutions for the retail industry to improve logistics, inventory visibility and supply chain efficiency, by connecting the data from sensors, laptops, cloud-based software, processors, bar code scanners, and RFID tags and readers.

One of the aims is to help retailers align their online and in-store sales operations, with better visibility of where items are available so that they can be bought online but picked up in-store.

Honeywell has also teamed up with Aereon, working on systems to help oil and gas companies improve the safety, efficiency and reliability of their operations.

As the IIoT spreads from one industry to another, how will it progress? “One of the big next steps is to be able to take this information and model it in order to provide a simulation or series of ‘what ifs’ for the operation,” says Garbrecht. “What if I choose to take the plant down for maintenance next week versus waiting to the following month? What if I leverage my manufacturing facility in Puerto Rico versus Germany for producing this product? Will it be more profitable?”

Big data analytics will go beyond simply providing information to help individuals make better decisions. Soon, IIoT analytics will involve a lot more automation, with machines making decisions, while humans monitor the process and intervene when necessary.

First, we’ll have to deal with the IIoT’s shortcomings, such as the need to push computing power to the edge and the need to be able to deal with network frailty and outages of the connection between plants and the cloud.

“Occasionally a digger or backhoe will dig up a network cable, causing the connection from the plant to the cloud to be cut,” says Garbrecht. “At some point everything will be wireless, and there will be no such thing as a network outage. Until then, customers are asking for a two-tier system, where mission-critical applications can run on site but also connect to the cloud for more advanced analytics.”

Businesses eager to get their factories internet-connected also need to assess cyber-security risks. “As connectivity grows in the industrial sector, so does the issue of cyber security,” says Paul Rogers, head of Wurldtech and general manager of GE Industrial Cyber Security. Technology always has to be backed up by secure systems to safeguard data, and ensure data integrity and application access. If an enterprise wants to use third-party outsourcing, sending data out for analysis and insights, it’s crucial to make sure the data is properly secured and conforms to regional legislation.

Also, before making their enterprise connected, companies need to be sure they are prepared. “Ultimately, their business models will change, their revenue curve will change and their people will need to change,” says Charles Joel, IBM lead consultant for Industry 4.0 and Internet of Things. “The typical failure rate on connected devices is 0.5 to 3 per cent; if you make one million connected devices and 3 per cent fail, that could be 30,000 service calls. Organisations need to ask themselves if they are ready to manage this.”

In the end, “the idea of connecting the physical and digital worlds to drive innovation, efficiencies and global economic growth is the Intelligent Enterprise,” says Smith. Businesses need to adapt quickly when it comes to the adoption of new technologies, a process he describes as ‘storming, norming and performing’. “We are certainly in the middle of the norming phase right now!”

This article was first published in E&T magazine