HomeChannelsFundamentalsSlowly, slowly, slowly – Software AG’s three principles of IIoT success

Slowly, slowly, slowly – Software AG’s three principles of IIoT success

German software company Software AG reckons the industrial internet-of-things (IIoT) has reached a tipping point. Enterprises have bought into its promise, and are starting to carry through on tech blueprints to connect to improve efficiencies and productivity.

“We see a turning point in the industry, after many years of piloting and proofs-of-concept,” explains Bernd Gross, senior vice president of IoT and cloud at Software AG. “We have noticed this for 12-18 months now. The focus is shifting to deployment.”

However Gross, speaking at the 2018 IoT Tech Expo in London last week, observes also the extant challenges facing ambitious enterprises looking to harness new IoT technologies. Sixty-five per cent of all IoT projects fail their original business plans, he says, quoting Boston Consulting Group. “That’s a large number,” notes Gross.

Software AG has completed hundreds of IIoT projects, he says, and has learned a few things. Here, Gross offers sage advice on how to give new industrial tech initiatives the best chance of success.

Step 1: Test

More than anything, Software AG advocates caution. Enterprises must learn to walk, first, before the technology runs away from them, and well-established processes unravel. “The first thing is to establish a discrete separate application environment,” explains Gross.

Enterprises might engage in isolated application management, alarm management, or condition monitoring to start with, he suggests. “Whether it’s a wind turbine, or a compressor, or an electric mountain bike, it is important to run applications in a discrete fashion – the environment must be stable, before it is more widely integrated with other functions.”

The risk is enterprises introduce unreliable new functions into well-oiled old systems, which introduce points of failure that can destabilise business operations. “If you implement too fast, you will fall into that category of failed IoT projects,” he says. “Our recommendation is very straightforward – start the project in a discrete application environment, a step removed from the rest of the business.”

Step 2: Integrate

In the end, under-cooked IoT solutions can undermine, expose and derail critical business functions. Enterprises must run new connectivity solutions in isolation for an extended period, and edit and patch as they go, until they are comfortable with the technology, and its impact. Gross put the test-time at six to nine months.

“Once you have it under control for that kinds of period, once you know the solution works and is reliable, and you know you can manage it – all the new modems, and 2G, 3G, and LoRa devices – then you move to the second step,” he says. The second step is integration, or “data-driven process integration”, as Gross puts it.

Suddenly, enterprises can plug in with confidence, and electrify their processes with insights. “Only then can you can start to automate your business in a completely new way, and bring transparency to it – which includes operational technologies (OT) and machines, as well as IT and back-office systems,” says Gross.

“You can then start to connect these worlds, and drive new levels of automation and efficiency.”

Step 3: Expand

Intelligence takes time; even machines must be taught. Too often, enterprises are awe-struck by their ultimate destination – in the shape of total automation, achieved by sophisticated artificial intelligence and machine learning – and get lost on the back-roads along their journey to it.

“Unfortunately, at the moment, we see too many projects that aim immediately at machine learning initiatives before they have even done the real deployment and connectivity and data virtualisation,” explains Gross. These first steps most be completed, he cautions again, before industrial capabilities can be expanded to include more sophisticated analyticals techniques.

“We see these projects take a year, or a year and a half before there is a commercially real available machine learning implementation. So take your time, develop your project plans. It’s a step by step approach. Start isolated, separate your learning in a few worlds, add then integrate, and then add in capabilities like machine learning.”

This is where the promise of IoT gets real, and where new horizons come into view. “Then you can innovate,” says Gross.

Previous post
Telia uses sensors to improve efficiency at construction sites in Norway
Next post
Cumulocity IoT platform retro-fitted to London buses, joins Telefónica IoT portfolio