Friday, September 3, 2021
Much of the systems and data for managing industrial plant are about to be moved from ‘on premise’ to the cloud, reckons Emerson’s Peter Zornio. It will be a ‘platform change’ as big as the move from workstations to ‘Wintel’ in the mid 1990s.
Cloud for plant operations is ‘potentially a big platform change that will impact us in the world of operations technology and automation,’ said Peter Zornio, Chief Technology Officer with Emerson Automation.
Mr Zornio envisages that automation systems will go through a platform change similar to the way mainstream computing went from mainframes to workstations, and from workstations to PCs.
‘The problem with a mainframe was when you ran out of horsepower you had to get another mainframe. Everybody dreamed, ‘if I could just have my own computer, so I don’t have to share with other people.”
‘Of course, that happened,’ he says, when we moved from mainframes to UNIX workstations in the engineering world. Everybody could have their own computer, although its processing was done on a central machine.
Then in the mid-90s, there was a move to Windows operating systems on Intel computing infrastructure, what people now call ‘Wintel’.
‘That started a computing environment that has been with us ever since,’ Mr Zornio said. ‘It is still the standard computing environment for the automation world and the IT world.’
‘Wintel’ computers have been used for the past 2 decades, both for personal computers and for servers, handling e-mail and enterprise software.
The first commercial cloud services, launched around 2010, were ‘rent a data centre’, enabling people to avoid the capital expense of having to operate their own. You would run the same software applications in someone else’s data centre, Mr Zornio said.
‘It was like a big virtual mainframe in the sky, but infinitely scalable.’
‘The IT departments of the world, once they saw the advantage of moving in this direction, have been accelerating in this direction, using cloud ever since.’
OT and IT
Much of the drive to move plant systems to the cloud is coming from IT staff, who are increasingly getting involved in the operations technology (OT) world, and are already comfortable running everything on the cloud, he said.
When automation systems did the transition to Windows in the 1990s, it meant the OT and IT worlds were using the same platform – and opened up many years of discussions about whose ‘domain’ it was.
‘IT guys looked at a lot of automation equipment and thought, it looks like the same stuff we use, we should be owning that,’ Mr Zornio said.
But since then, most companies have been keeping the OT world separate to the IT world, having it on separate networks or using firewalls. ‘Systems were connected but weren’t really integrated.’
Emerson tried to provide people in the OT world with pre-configured equipment, so they did not have the complexities of running IT systems.
‘There was a long time of “peaceful co-existence”. OK IT guys, here’s your data, now leave us alone, we want to run our environment.’
In the domain of life science, there has been more integration between IT and OT, he said. Perhaps life sciences has been the most aggressive in moving IT technology to the manufacturing world.
But today, ‘digital transformation’ programs are pushing IT people to get more involved in the OT world in all industry sectors.
‘Nearly everyone we serve has some kind of digital transformation program going on. Many times it’s run by the IT department, sometimes OT and engineering groups.’
Many digital transformation programs have forced a much tighter integration between IT and OT, such as projects delivering data to mobile devices, augmented reality, and bringing in third party connected services.
And as IT people have influence on the OT world, they are bringing in their preferences – including for cloud technology.
‘We see IT departments directly tackling operations problems by pulling in data, running machine learning, trying to solve some of the problems that OT departments have been trying to solve for quite a while.’
Although you normally need a domain expert, someone who understands the process, to solve a problem. ‘A data scientist sometimes comes up with some, quite frankly, silly conclusions or things which are obvious to people in the plant. But the combination of the two does enable solving new problems.’
Some business now have single teams looking after both IT and OT. Probably all companies will go in this direction eventually, as ‘one integrated computing team.’
OT and IT typically work on different time scales, where OT people are used to keeping equipment running until it fails, where IT people are used to implementing new software all the time.
By merging IT and OT, you can get much more IT capabilities in the OT world – such as wireless communications in the field, data analytics in the cloud, and more data storage capability.
If you want cloud hosted services, ‘by definition you have to have IT guys involved – they’re going to want to know about data privacy rules, where data is stored.’
On premise vs cloud
This leads to the question of which software systems should still be kept on premise, if you are moving much of it to the cloud.
The operations technology world has traditionally run everything on premise. It makes sense when you are talking about software which actually runs equipment.
But non-time-critical applications, such as data historians, could be moved to the cloud.
Organisations with data which is not very time sensitive, such as a farm, might be comfortable running all of their systems on the cloud. If anything fails with the digital technology, the corn will still grow.
The automation world is different. The vast majority of Emerson customers ‘want to have the ability for plants to continue to operate without the cloud. If they lose the internet connection they can still reliably run their plant.’
So perhaps what we will see is production plants having a core system on site which runs the plant, including sensors, control systems and visualisations, and everything else going to the cloud, he said.
Other issues to consider are whether you like the low capital requirements of cloud software, how much technical depth you have in your organisation, and how well your organisation is able to adopt change.
Some companies have a specific policy to try to get rid of all their IT hardware and go 100 per cent cloud.
There are increasingly sophisticated data tools available on the cloud, and this strengthens the argument for using cloud software. It is no longer just about renting computers, it can be about renting an operating system, including capability for data storage and data analytics, Mr Zornio said.
Cloud providers are acting more like software companies (or software companies are becoming cloud providers).
For example Microsoft offers tools to ‘virtualise’ data via a cloud service (show data from multiple sources as though it came from the same data store).
There’s still a lot of work you have to do yourself to turn that into a real application. IT people are often willing to do this work, while OT people usually want to only use software which has been built specifically for their equipment, he said.
Cloud hosting companies see an enormous potential market from people who want to ‘ingest’ streaming data. It used to be only manufacturing and industrial sites who were interested in streaming data, but now we have buildings, farms, car parks, and cities steaming sensor data, who want a way to store and analyse it.
Companies like Microsoft and IBM ‘want to build things that go across all the markets, not just manufacturing and industrial,’ Mr Zornio said.
Some people raise concerns over security of cloud data, ‘but many people will argue that cloud vendors have more time, money and people to put into security than any one company does.’
‘This is all they do – store data. They have hundreds of the best people working on security. It’s not perfect, but they probably have more resources to apply than any one company would do.’
One way is to set things up so data from the OT world can only go one way, like a ‘data diode’. Data can come out, but bad things can’t get in.
Emerson has partnered with Microsoft’s Azure as its preferred cloud provider – and about 15 different Emerson solutions are available on Azure as ‘software as a service’.
Emerson has software called ‘Plantweb Optics Data Lake’, which can connect to data streams from sensors, stream the data to the cloud, and ‘historize’ (store) it.
This saves operations technology people from the trouble of working out the best way to connect sensors to cloud systems. This can be very tough in some circumstances, such as where you have an older system, and you don’t want to risk overloading its processing capability with an additional demand.
Emerson has also set up a cloud centre of excellence, where it makes cloud software, and ensures the services are secure and consistent.
‘This is a big platform transition for us,’ Mr Zornio said.