If infrastructure could talk, what would it tell us?
The idea of communicating with assets is no longer far-fetched. Advances in industrial internet of things (IIoT) technology mean intelligent infrastructure is fast becoming the norm according to Bhupinder Singh, newly appointed Chief Product Officer for international infrastructure software company Bentley Systems.
“Infrastructure is essentially dumb; when it gets wired with sensors, they're talking to you so the infrastructure becomes intelligent,” Singh says.
And the ‘conversations’ made possible through the IIoT offer the potential for massive operational, safety and productivity benefits.
“When you retrofit an existing bridge with sensors, you can measure things like vibration, sound, displacement and strain gauges,” Singh explains.
“You can get a good idea about exactly the stress in that joint or that particular point in the bridge and then you can compare it to the theoretical stress from its design.
“You'd be able to tell if it's operating at a level of efficiency that's 30% of its design or 50% of its design or 80% of its design, and then you would be able to be proactive and efficient in the safety rehabilitation work.”
There are obvious safety and savings benefits to having access to an asset’s digital DNA.
“If you look at any government that has public infrastructure, where you have a safety component, think about how more efficient you can be and how safer you can be doing something around sensors versus doing things with just human inspection,” Singh says.
“The other dimension is around very expensive infrastructure; offshore oil platforms, for example, where you really would love to extend the life of these platforms another decade.”
Water networks, where vast amounts of infrastructure are buried away from the human eye and are difficult to access, are also a prime target for the IIoT.
The growth of the IIoT
Some estimates put the number of internet of things (IoT) devices in circulation by 2020 at 24 billion; according to Singh, this is likely to be a gross understatement.
“I was at Microsoft in March of this year and they told us that, by March, they had put in more storage and more computer infrastructure in their cloud, in their data centres, in those three months than they had done all of last year,” he recalls.
“That's insane, because these providers that are reacting to demand.”
But what is causing this IoT boom? Singh says it is a confluence of consumerization – where the popularity and hence the profitability of personal devices like the smartphone are driving the development of the technology – and opportunities created through cloud computing.
“Companies are spending billions of dollars to build these vast data centres that can capture, store and distribute petabytes worth of data efficiently,” Singh explains.
“Sensors have gone from dollars to cents; people are sprinkling them into the concrete on new decks of bridges in universities during research. These RFID tags (radio frequency identification tags) are getting so cheap, they're getting put on every bar of steel that's shipped sending out of a mill.
“So if you think about low-cost sensors you can put anywhere, these things are going to capture a large amount of data that they're going to transmit.
“With cloud computing, you can store it efficiently. I think those are two magic things that are causing an acceleration of the stuff coming to market.”
What next?
Singh predicts machine learning – the science of getting computers to act without being explicitly programmed – will be one of the fastest growing facets of the IIoT.
“I think what's evolving very, very quickly are machine learning algorithms because the amount of data that's being generated through all of this is immense,” Singh predicts.
“You can't use traditional techniques when you have petabytes' worth of data that are being generated to make sense of it and provide actionable information.
“Being able to do some kind of pattern recognition and intelligent suggestions will involve new types of algorithms and AI techniques – which they're calling machine learning or deep learning.
“I think that's an area that will evolve very, very rapidly over four or five years. There’s examples of this today with some of these supercomputers that are playing chess and beating grandmasters and they're using certain types of techniques that are going to get applied to our world very quickly.”
Image: Bhupinder Singh