Moving toward edge computing adoption
For the control engineer, it's worth asking what the benefits are from moving intelligence to the edge and how they can achieve success in brownfield plants.
According to Silvia Gonzalez, development leader, machine automation solutions, Emerson, industrial edge controllers can provide significant advantages in many applications. Their ability to gather and analyze vast amounts of process data, and deliver actionable insights allows edge controllers to facilitate better decision-making and improved operational performance, when compared with more traditional operations technology industrial controllers, such as programmable logic controllers (PLCs) and process automation controllers (PACs).
“The main principle of edge computing is moving the computing intelligence as close as possible to the device, machine or application,” Gonzalez said. “The computing intelligence then performs data pre-processing, analytics, visualization and other computing tasks, right at the edge of the data communications network. This helps to eradicate process latency, because data does not need to be sent from the edge to a central processing system and then back to the edge.”
The result is more efficient data processing, because large amounts of data can be processed near the source, reducing Internet bandwidth requirements. This can help reduce costs and allows more applications to be implemented, especially in remote locations.
“There is also less reliance on the stability of outside networks,” Gonzalez said. “First and foremost, edge intelligence allows sub-second decisions to be made by the controllers, which can help optimize operations. Manufacturing has been using edge-based control for a long time, with edge controllers running logic to maintain safe and efficient operations. However, in addition to just controlling the machine or process, edge devices are increasingly analyzing and diagnosing what is happening in these processes. Greater access to and analysis of operational and diagnostic data enables operators to optimize production and allows maintenance personnel to identify potential issues. The latest generation of edge controllers offer Industrial Internet of Things (IIoT) functionality with non-deterministic edge applications, analytics and cloud connectivity to significantly expand the range of services provided by the controller.”
Gonzalez went on to explain that, typically, these controllers will have two separate operating systems – one that processes the deterministic control functions and the other running non-deterministic applications. “Data can be collected from the device or machine and aggregated with other data from sensors outside the control platform, such as vibration, pressure and temperature sensors, that can be used to determine process problems, or diagnose equipment health issues,” she said .
Gonzalez went on to point out that, within many brownfield applications, obtaining operational and diagnostic data from equipment at the edge of the network has been challenging, with systems being expensive to design, operate and maintain. “This task has been simplified by the latest edge controllers, which can be retrofitted into brownfield systems to provide analytical and IoT capabilities.
“Alternatively, other edge components, such as industrial computing devices and gateways, can provide cost-effective solutions, working in conjunction with an existing control system and providing analytics and data communication functions,” Gonzalez said. “An edge computing device will be implemented when the analytics demands high fidelity data, where the high bandwidth requirements and the cost of cloud systems and storage is prohibitive. This localized solution also overcomes any potential network connectivity issues. Edge gateways, on the other hand, simply collect the data, package it and forward to industrial clouds. This provides visibility to geographically distributed assets, enabling access to consolidated data that can be used to check equipment health without the need for site visits.”
Lower capital expenditure
According to Andreas Hennecke, product marketing manager at Pepperl+Fuchs, intelligence at the edge of process plants – in smart sensors and actuators – has been available for some time. “Processor power enables multi-variable instruments,” he said. “For the engineer, this results in lower capital expenditure for the instrument, including attributable cost and effort for engineering, piping, fittings and even spares holding. Valve position sensors can report break away and run time to enable maintenance teams to adopt a predictive approach to plant upkeep leading to lower operational expenditure.”
Valves with a Foundation Fieldbus H1 interface can run a PID algorithm locally based on the inputs from the flow or level instrumentation on the same segment in the field. “However, utilization of this intelligence to its full capacity is limited today – among other things due to the limited bandwidth and lack of a flat network technology.
“Ethernet-based industrial protocols and infrastructure enable flat networks and open architectures. Only recently have these technologies included attributes required for process automation, which now enables deployment of intelligence at the edge within process plants,” Hennecke said.
Today, an Ethernet-based, flat network can be deployed to the field within process plants. The Ethernet advanced physical layer, or short Ethernet-APL is ruggedized power and communication on two wires. It is an open and complete package of standards available to any vendor and user providing high bandwidth, long cable lengths and explosion protection with integrated intrinsic safety. Being just a physical layer, Ethernet-APL carries any and all Ethernet-based protocols.
Today’s industrial protocols offer ‘profiles’ for simple engineering such as a standardized description of instrument types independent of make, type and vendor – controllers can be reconfigured ‘in run’ for continuous operations. “Profile-based communications such as Profibus PA or Profinet enable operation and instrument exchange as simple as users know it from 4-20 mA – actually even eliminating the need for PV scaling,” Hennecke said.
He goes on to explain that, with this combination of technologies deployed, users are able to unlock data that was previously unavailable. While controller and instrument maintain their cyclic data exchange, an asset management can log historical data or store configuration settings via HART IP or OPC-UA protocols. At the same time, a maintenance person can access an instrument’s web page via smart phone to get health information. Open communications via Ethernet-based physical layer and protocols will enable application and handling benefits for operators and maintenance teams that were much harder to achieve with the existing portfolio of technologies.
Hennecke says that, for brownfield upgrades, a significant benefit lies in protection of the investment in existing assets while at the same time upgrading communications to the controller via Ethernet. “This enables or improves access to data currently ‘locked in’ the instrumentation. The costs are the same as for any typical brownfield system upgrade,” he said. “Ruggedized remote IO such as the Pepperl+Fuchs LB/FB Remote IO systems installed in the field connect to the same backbone and provide access to modern-day instrumentation with 4-20 mA interface. They require external power and a standard or fiber-optical Ethernet back to the control room. In addition to the primary measurement value, the remote IO station can translate health and diagnostic data, secondary measurement values or historical data traces from HART to the Ethernet-based protocol of choice. This access is as fast as it can get, as the Remote IO utilizes individual HART masters per instrument and thus eliminates bus cycle times.
“For Profibus PA instrumentation, the rail field switch from FieldConnex offers dual-purpose spur/instrument ports. This option will automatically detect Profibus PA or Profinet over APL communication and adapt automatically. A single infrastructure can connect and operate both generations of communications technologies simultaneously saving significant investment costs in instrumentation and infrastructure. Even minor upgrades are possible, where the value of fast, Ethernet-based communication is required, e.g. for specific mission-critical sensors or actuators. Ethernet-APL field switches can be installed in junction boxes close to the instrumentation and can utilize the same communications and power link back as Remote IOs.
“Ethernet-based communication, via standardized protocols such as Profinet, EtherNet I/P, OPC-UA and HART-IP provides access to the intelligence already located at the edge, in the sensors and actuators already installed in plants. Users can harness this data from any process plant and without constraints to improve efficiency of standard operating procedures, turning from reactive to predictive and proactive plant upkeep, monitoring and optimization.”