How much control goes to the cloud?

Cloud computing is gaining ground as industrial plants become more efficient, but it’s important to recognize where computing is needed and where it should be taking place.
By Bob McIlvride, Skkynet Cloud Systems Inc. September 6, 2018

How much control should go to the cloud? All graphics courtesy: Skkynet Cloud Systems Inc.Many process engineers would be happy to apply the mantra "What happens in Vegas stays in Vegas" principle to their systems, meaning, "What happens in operations stays in operations." To process engineers, all automation, control, and instrumentation applications should stay in the plant. It’s more reliable and secure that way. And that’s the way it was―until a few years ago. A shift is taking place.

In a drive for efficiency and having a competitive advantage, companies are turning to cloud computing as a way to gather production data, crunch the numbers, and feed selected results to management, to analysts, to suppliers, to vendors, and in some cases, back to the plant. Call it the Industrial Internet of Things (IIoT), Industrie 4.0, or enhanced supervisory control and data acquisition (SCADA), but the digital transformation of industrial production is well underway.

Because this radically departs from the way things have been for decades, many questions arise like: What about security? Are the connections reliable? Isn’t this just what we’ve always been doing, with a new name?

As companies move beyond the pilot stage and begin to implement full-scale IIoT and Industrie 4.0 systems, another question often coming up is: How much control goes to the cloud? Or, how much data processing should be done in the cloud? (Figure 1)

Cloud computing for industrial systems

Some cloud-computing proponents assert that the more computing that can be done in the cloud, the better. However, that approach does consider the realities of industrial control systems (ICSs). It would be foolhardy to attempt low-level or time-sensitive control from the cloud, as well as most types of supervisory control. The security, latency, and reliability of an internet connection can’t match an in-plant network. Also, the volume and rate of data pouring in from a typical industrial system would consume enormous amounts of cloud resources, resulting in a much higher cost.

One of the latest trends in cloud computing isn’t in the cloud at all, but at the edge. Edge computing can mean different things to different people. From the viewpoint of an IIoT cloud, the edge is often considered to be the border of the industrial system, such as the gateway that connects to the cloud. From within an industrial control system itself, the edge could be a device, like a sensor, actuator, or perhaps a remote terminal unit (RTU) out in the field that collects data from a number of devices. However, the edge is defined, the idea is if processing power is inserted there, a lot of time and money can be saved by filtering, conditioning, and aggregating data before it is passed on to the next level of analysis.

Everything doesn’t have to be done in the cloud. In fact, most automation engineers would agree it’s better to put computing power where it is needed. Local computing keeps responses closer to real-time, cuts bandwidth, and reduces the uncertainties of network connections. Consider these four areas where processing can take place:

Balance the data load at each step of the process.Device: Adding computing power at the device level can help reduce the amount of data that needs to be sent to the plant’s upstream applications and the cloud by filtering or conditioning the data at the source. In addition, processing at the device can abstract the data from the different field protocols into a common protocol. This means upstream applications do not need to know the specific protocols of the field devices providing them with information, which makes the data available to a wider range of clients.

Plant: Traditionally, this is where most industrial computing has taken place, with SCADA and human-machine interface (HMI) systems providing supervisory control and visualization. Now, to satisfy new requirements, these systems are increasingly being used to create metadata, such as device status, connection status, and system health scores, as well as target production tracking.

Gateway: Computing at the gateway is an effective way to apply the cost savings of data reduction and conditioning to established infrastructure that may not be able to support added computing resources themselves. If an organization doesn’t want to disrupt the legacy system, adding data processing at the point where the data leaves the plant makes sense. (Figure 2)

Cloud: When proper steps to reduce, manage, and enhance the quality of the data from plant systems and remote devices is done at the source, cloud computing resources can be used more effectively to aggregate data from multiple locations, store and analyze the data, and present it in a form best suited to the client needs.

The latest generation of IIoT cloud services also provides secure, bidirectional connections, which allows the cloud to send data and analytics back to authorized end users at any location. Not all cloud services offer this, but the benefits can be significant. Cloud services can store data on a scale that can’t be matched by in-house systems. Combining that with a broad range of cloud analytics shows how integrating plant data and cloud services can enhance process knowledge and guidance.

Industrial control, cloud services

Making the most of the new era of cloud services for industrial control will depend on how the cloud data needs to be managed and the data that needs to be received from the cloud. Applying resources at the appropriate level to condition and optimize the data sent to the cloud will reduce costs and generate a quicker round-trip time for analytical data supplied back to the plant. Abstracting the data from multiple source protocols will make it available to more client applications in the plant and in the cloud.

The days of "what happens in the plant stays in the plant" are numbered. Driving process data into the cloud and getting meaningful answers back is already the goal of many integration projects. Balancing the data load at each step in the process seems to be the key to a successful implementation and adding edge computing where it is needed will pull things together.

Bob McIlvride is director of communications at Skkynet Cloud Systems Inc., a CSIA member. The CSIA is a CFE Media content partner. Edited by Emily Guenther, associate content manager, Control Engineering, CFE Media, eguenther@cfemedia.com.

MORE ANSWERS

KEYWORDS: cloud computing, industrial analytics 

Cloud computing for industrial systems

Areas where computing power takes place for industrial systems.

Consider this

How does your facility balance and analyze data at each step to optimize operations?

About the author: Bob McIlvride is director of communications at Skkynet Cloud Systems Inc., a global leader in real-time data information systems. He has been working as a professional technical writer in the industrial process control sector for more than 15 years.