Industrial IoT Intelligence Aims to Save Lives by Preventing Disasters
On March 23, 2005, a BP Plc oil refinery in Texas blew up, killing 15 people and injuring more than 170. The explosion's cause was BP employees overfilling and overheating an important piece of oil processing equipment, The Guardian reported at the time. BP ultimately shelled out more than $3 billion to pay fines, settle lawsuits, and improve the refinery, which Marathon Petroleum Corporation bought in early 2013 for $2.5 billion.
This is just one example in a long list of disasters because, in the energy sector, plants and similar facilities can explode catastrophically. That's long been a fact of life. However, vendors pioneering Industrial Internet of Things (IIoT) infrastructure are looking to help prevent accidents such as these through automation. IIoT platforms provide real-time monitoring and preventive maintenance, which help enable plant owners and operators to react more quickly to emergency situations. This reaction time is sped up further via virtualization, which helps reduce latency in IoT applications that operate in facilities such as oil refineries and power plants. Decisions on equipment in power plants and chemical plants must be made in real time to protect the safety and security of the surrounding population.
In order for oil refineries and power plants to avoid catastrophes, a safety instrumented system (SIS) from companies such as Schneider Electric can "monitor these critical variables that would be indicative of an exothermic reaction," said Christopher Lyden, Senior Vice President of Strategy and Portfolio at Schneider Electric. "If they sense that those variables are changing too quickly, then they take actions to actually shut the process down."
Lyden is in an excellent position to comment on these situations because Schneider Electric is an automation supplier that provides the automation equipment specifically for power plants, oil production plants, and oil refineries. Vendors such as Emerson Electrics and Honeywell International also offer SIS platforms.
A SIS acts as a "brake" on a plant's operation, according to Lyden. "Basically, a SIS is there to ensure the plant shuts down before a catastrophe or critical incident can occur," Lyden said. "It monitors the performance of the operation and the operating assets. If the process begins to speed up or if something starts to lose control, then the SIS takes over and brings the plant down."
Schneider Electric's SIS, the EcoStruxure Triconex Safety System, is a combination of edge controller hardware and software that helps maintain the uptime of plants. The system can provide warning of fires or other combustible events as well as other events, such as toxic gas leaks, and help rectify the situations. Although SIS platforms do not connect to data networks due to security concerns, they still play a role in IIoT by providing data to operators in plants and refineries to help them make critical decisions.
"For example, operators can receive alerts via dashboard on their smartphone, telling them the plant or a particular plant asset is at risk. They can then take the necessary action to avert an incident," Lyden said. "We're helping them understand their safety threshold, [meaning] how far they can drive the process and their assets before the plant reaches an unsafe state."
Redundancy, Autonomy, and Fast Failover
According to Lyden, in order to protect against the type of disaster that occurred at the BP oil refinery mentioned earlier, operators should maintain fast failover and automated control, as well as consider implementing redundancy on virtual machines (VMs). To accomplish this, Schneider Electric deploys Wind River's Titanium Control on-premises cloud infrastructure platform. "Having redundancy on the VMs is highly important, and having this fast failover means that they never lose view of the plant for long enough to give them anxiety," Lyden said.
A distribution control system from a company such as Schneider Electric brings autonomous functions to chemical plants and power plants. Schneider Electric's programmable logic controllers running Wind River's VxWorks real-time operating system (OS) let power plants control their operations autonomously. The autonomous functions of a distribution control system help power plants and oil facilities control the pressure, temperature, and flow of energy. Lyden refers to this as "heartbeat control." In fact, Schneider Electric and Wind River are working on a next-generation process controller. This type of control technology handles failover in plants, when standby equipment takes over due to failure in primary infrastructure.
Safety and Security of Critical Infrastructure
Wind River helps customers such as Schneider Electric integrate the diverse hardware and control apps of industrial plants on a single platform. Plants can also leverage virtualization and containers to maintain optimal availability. Wind River specializes in real-time OSes as well as the virtualization technologies needed to drive intelligence to the edge, keeping systems available and connected to the cloud.
A component of an IIoT infrastructure, real-time OSes typically focus on safety and mission-critical apps. They react to their environment on a micro-second scale, and are ideal for devices and apps that can't fail. "Real-time OSes can ensure that the compute, memory, and the cache always gets distributed on a priority basis," said Jim Douglas, President and CEO of Wind River.Running real-time OSes in parallel with Linux lets companies apply machine learning (ML) at the edge where plants have "high safety criticality," Douglas said. Although real-time OSes and Linux can be operated separately, when used together Linux can run the non-safety-critical parts of a device or app while the real-time OS handles the mission-critical functions. Linux is useful with embedded systems because of its lower system requirements and higher performance capabilities, according to Douglas. Because of these higher-performance capabilities, you can now find various flavors of Linux being used on factory floors, in programmable controls, in airplanes, and in flight-control systems.
More compute is occurring at the edge to avoid this delay. "You can't have that latency," Douglas said. "If something happens, you're going to have a catastrophe."
Automated and Unmanned
These technologies are evolving rapidly enough that Lyden predicts that some gas plants may soon become unmanned to protect against disasters. "The technology today is not such that people are confident to do that. However, we're starting to see it offshore," Lyden said. "So you'll see all of the operations on a group of offshore oil platforms operated from a central mother platform with unmanned daughter platforms."
Lyden also noted that a number of small gas plants are now remotely managed. "We're driving, I think, toward this notion of autonomy, meaning that the control system has everything in it that would allow control, but still allow safety without humans there," Lyden said.
"This notion of intelligent, autonomous, edge devices that are self-diagnosing," he continued, "that not only do [they] control but they can also then begin to do things like manage the condition of the physical plant assets. Those kind of capabilities are necessary to get to this vision of unmanned plants."
In addition, edge computing and IIoT will create opportunities for humans and autonomous machines to coexist. In fact, artificial intelligence (AI) at the edge will be at the heart of IIoT, according to Douglas. The first wave of IIoT involved connecting edge machines to enterprise networks. Then came analytics and data visualization. "We could start to do analytics using software [and] write things like visualization packages to make it easier to detect anomalies," Douglas said.
"The next wave is, you're going to have machines that are either fully autonomous or partially autonomous where they can actually start doing more sophisticated tasks on their own, and you can have people be more focused on higher-level tasks and let the robots do the lower-level tasks," Douglas continued. "That's the big transformation. That's where AI comes in, where when you have enough compute power on the edge that you can start to make this machinery a lot smarter. And have it take over more of these type of tasks that [today] take a lot of human interaction."
Analytics and Equipment Health
Currently, manufacturing plants and refineries have a variety of equipment that helps control the manufacturing process, including compressors, gauges, pumps, and valves. Sensors enable these components to become intelligent and share information about their operating performance. According to Lyden, in the future, chemical plants will have pumps generating analytics that tell personnel when the pumps are shorting out the starters or using too much current.
"You can expect pumps to be instrumented in a way that tells you if it's consuming the power or if the pump is becoming less efficient," Lyden said. "And all of those things would be run from a common edge device that both controls the operation of the pump and diagnoses the pump."
As organizations remove people from power plants, and the prices of sensors and measurement devices go down, there will need to be more management of the plants through IIoT to avoid failure of the unmanned plants.
"It's things of that nature that I think the IoT will enable, because we'll see much more equipment health management than we see today," Lyden said. "The next step then, as the IIoT matures, is to impart actual control capability into these assets. What we're talking about is each of these assets becoming a cyber-physical system that is able to control itself autonomously." In addition, going forward, chemical plants will connect their blowers, heat exchanges, motors, and pumps, according to Lyden.
Next to physical IIoT developments, analytics will develop and play a bigger role in managing the performance of pumps. The combination of increased connectivity, computing power, and analytics will help plants and refineries manage the health of process equipment, improve decision making, and boost the reliability of critical infrastructure.
The aforementioned incident at BP's oil refinery, as well as an Exxon Mobile refinery incident on Feb. 18, 2015, in California, in which a release of hydrocarbon caused an explosion, show a need for IIoT technology. The intelligence it brings can help prevent these types of disasters.
This article originally appeared on PCMag.com.