Because the Industrial Internet of Things (IIoT) spans so many technical areas, it helps potential users to understand how others see it, which can point out the most useful ways to implement it in individual applications. Here are the expert assessments of several suppliers:
"The challenge now is we have many more devices connected to the Internet that weren't connected before. We also have many brownfield sites with lots of legacy equipment that's adapting to new uses with open architectures and communications, as well as mobile and edge computing. The question is how to deploy and manage all these devices," says Alex Kuo, product sales manager for North America at Advantech. "We can help with computing hardware that's compatible with open architectures, and can communicate via serial, OPC UA, EtherNet/IP, Profinet, EtherCAT, MQTT and other protocols to reach cloud-computing services like Microsoft Azure and Amazon Web Services (AWS)."
"IIoT is designed into new plant areas because it's faster than the virtual machines (VM) with everyday duplex communications that have been used for 10 years. IIoT devices are also less costly, lightweight, Ethernet-based, and enable future upgrades," says Keith Flynn, senior project management director at Aspen Technology, which recently acquired Mnubo Inc. that provides AI-based IIoT and analytics software, and Sabisu Ltd. that provides enterprise visualization and workflow software. "We've been moving to provide both onsite secure-edge and cloud-native solutions that can connect users to their data and let them build models faster. Edge technologies and artificial intelligence (AI) can enhance basic, first-principles models, while the digital realm's increasingly unlimited computing power can produce simulations and immediately send alerts to operators, even for equipment that's 30-60 years old."
So users don't have to develop AI independently, Flynn adds that AspenTech provides subscriber-based, cloud-native applications, including engineering models, manufacturing execution systems (MES) and asset performance management (APM) products. "We want to give users the self-optimizing, semi-autonomous plant, which uses AI for edge, cloud, advanced process control (APC), and multi-plant analytics," he says.
"We see the same IIoT trends as everyone, such as users asking for technologies that don't have to be onsite. On one end side, we have what we consider to be the edge, which is local sensors and on-premise networks. On the other side, we have the cloud, where data is moved offsite and outside local networks," says Scott Kornblue, lead solution designer at Belden. "In some cases, users need high-speed communications that are close to the action. In others, they're dealing with remote applications, such as oil fields, wells and pipelines out in the field. Exceptional requirements need an IIoT strategy that can address them, such as Persistent Data Network (PDN) from our ProSoft division, which has decades of experience in IoT/IIoT and secure remote access applications. PDN can bridge devices anywhere, so they all believe they're on the same network. There may be some slight latency delays, but a well-designed network will be OK with them, and will likely be used for less-critical applications. This demonstrates that today's IIoT consists of multiple virtual devices in one location, and runs with hybrid, edge computing networked by several high-speed protocols."
Kornblue reports that ProSoft can equip process applications or facilities with PDN hardware and software, run multiple protocols on one secure network, and keep its connections active by tunneling between different protocols. "Bridging networks lets protocols talk in a simper way," says Kornblue. "The edge usually has many protocols such as EtherNet/IP or Profinet that need to be converted behind the scenes to OPC UA, MQTT or MQTT SN connectionless. Physical architecture and segmentation may incorporate wireless and use data diodes and/or firewalls for cybersecurity, such as those available from Belden’s Tripwire, Hirschmann and Tofino divisions. This lets all of a user's devices and processes talk to each other securely. With standardized communications, it also matters less which supplier's protocol a user standardizes on."
Of course, the main snag that goes with setting up all these new network links is increased risk of cyber-probes, intrusions and attacks, which mean risk assessments (RA) are needed for cloud-computing and multiple software packages running in one box—just as RAs were needed for earlier networks and technologies. "Security has to go with network bridging," explains Kornblue. "We tell users converting data from devices and protocols that they must address cybersecurity, such as using Tripwire for passive anomaly detection. These aren't active queries, so they don't create nuisance traffic. It just builds a behavior baseline, so when unusual behavior occurs, it can automatically disable ports or adjust firewall rules on the fly. We also have deep packet inspection on our Tofino devices, along with whitelisting, so only certain industrial protocol values can be read and written."
Kornblue adds that future industrial automation design strategies will likely combine industrial computing, PLC logic, network switching, firewall, MQTT broker, OPC UA server, I/O ports, data collection and protocol conversion. Future solutions will also be scalable, and let users add software apps they need, such as Docker containers for running functions on the edge. "This is the future. Users will browse the industrial equivalent of the App Store, and download what they need onto edge IIoT gateway hardware," says Kornblue. "Once they convert to a unified architecture and all their devices are on the same network, they can aggregate their data to a cloud-computing service, perform analytics, find efficiencies and inefficiencies, and make better decisions to optimize their operations."
"Fit-for-purpose sensors and wireless networking are enabling many new entrants to the IIoT space, but we're also seeing many low-cost products with poor signal strength and battery life, as well as broken sensors, false alarms and devices not picking up what they should," says Shane Hale, global business development director for Pervasive Sensing wireless products at Emerson. "Most IIoT components from non-traditional suppliers aren't designed to run in harsh settings, such as a vibration sensor from a major online consumer retailer that you can tell isn't designed for use in heavy industry. The problem is users see a good price, but the device ends up being more trouble than it's worth, and they need a replacement solution that’s actually viable, often within a few months."
Hale adds a good case can be made for fit-for-purpose, IIoT-enabled sensors, but users must examine their tradeoffs, such as accuracy, update rates, communication frequencies, battery life and others. "You don't always need 0.05% accuracy or 1-second update rates," explains Hale. "If you usually check on a process every eight hours, then a regular wireless pressure gauge (WPG) that updates every 30-50 seconds is sufficient. There's also the question of fitting IIoT devices into existing systems, and users have to determine if they're compatible. We use WirelessHART because most plants already have 4-20 mA HART devices, and it's often easier to adopt WirelessHART that's essentially an extension of the HART protocol. Many new-entrant suppliers require users to buy their proprietary protocol gateways and cloud services, which don't communicate with devices from other suppliers. WirelessHART is better because it lets users mix and match devices from different vendors on a common infrastructure. Also, WirelessHART gateways can talk to existing Modbus, HART-IP or OPC UA hosts, and send production and performance data to the cloud via MQTT. This is also how IIoT connects fit-for-purpose sensors to plant networks without using regular Internet or third-party subscription cloud services.”
Hale adds that Emerson recently worked with Florida Power and Light on manual rounds at its facilities, each with about 200 final points, such as monitoring calibration gas cylinder pressures and pump vibration levels. FPL and Emerson converted from manual monitoring to using Rosemount WPGs, vibration and acoustic sensors, and other multi-input devices to monitor operations. "The solution has been in place for over a year as we completed this project just as COVID-19 was impacting operations," says Hale. "All the points connect to WirelessHART gateways, and are relayed via Modbus to historians and to a central control room with Emerson Ovation DCS, allowing operations and maintenance personnel to make better operating and maintenance decisions based on improved alerts and alarms."
Hale explains there are five basic steps for implementing IIoT devices and networks:
Determine what new measurements could produce value, and decide what to monitor. This can be done by looking at manual-round tasks and areas where staff is presently spending their time, and ask what else could they be doing instead? This will show where IIoT-based measurement could be of value.
Work out measurement priorities. There are many possible measurements, so which should be done first? Organize them according to three categories: production, safety and environment/emissions.
Map where to install wireless gateways, and settle other physical networking issues.
Decide where to send measurements and other data, whether it's to a historian or another system. Work with IT colleagues on how much data to relay to higher levels because it's likely too costly to send up all the data from every I/O, control system and production application.
Specify, install and configure gateways to send data via Modbus, OPC UA or other protocols, and set up separate entries to the IT, cloud and business sides.
"This is the basis for the very high return on investment (ROI) that IIoT and digital transformation can achieve," says Hale.
"We don't usually define IIoT as just a microprocessor, Ethernet and an Internet protocol (IP) connection. We consider to be a cyber-secure ecosystem that lets us remotely access, operate and monitor assets and overcome productivity and personnel challenges," says Harshal Haridas, principal software engineer and chief cybersecurity architect on the global architecture team at Honeywell Process Solutions. "We also consider it to be a end-to-end architecture of formerly disconnected and isolated devices and processes with fixed parameters in our IIoT solutions. In the past year, COVID-19 made remote connectivity, remote accessibility and mobility essential. This is why we say IIoT is a complement and a transition to an overall, connected environment, where systems can run locally, remotely or in the cloud."
Haridas reports that Honeywell's three-year-old Measurement IQ virtualization software runs in the cloud, and decouples software and hardware. Once a process facility's assets are connected, data is relayed to Honeywell's Secure Data Center, which can help run a DCS, provide insights and recommendations, and deliver customized alerts via mobile devices. Equipment located to the field can report measurements from I/O points to Honeywell's Control Edge remote terminal units (RTU).
"Each user has different priorities and requirements, so they have to evaluate how different IIoT options can benefit them, reduce costs, and maintain their cybersecurity and safety," explains Haridas. "Once an IIoT solution makes a good business case, the next step is following IEC 62443 best practices and meeting its requirements, such as seeking risks to the architecture, determining if there are any new attack surfaces, segmenting processes and plants into different protection layers separated by firewalls, and implementing zero-trust networking."
"The proliferation of data and the many ways of producing and consuming it are tearing down the seven-layer Purdue Open System Interconnect (OSI) model's paradigm and silos," says Dave Eifert, IIoT business development manager at Phoenix Contact. "As more data connects with users more quickly, it's also contextualized. When it shows up, the people and/or equipment consuming it already know how it can help them."
Eifert reports that unified namespace (UNS) software has emerged as one of the primary ways to handle large data volumes from many different sources because it acts like a centralized repository. It's often combined with MQTT Sparkplug publish-subscribe specification that allows any device or application to subscribe, and create a foundation for scalable, digitalization applications.
"Sensors and instruments that send data to SCADA systems, PLCs and DCSs typically use a lot of custom formats and different protocols, which is a mess because they use different units and timestamps," explains Eifert. "Now they're using more standard communications like OPC UA and common IoT protocols like MQTT Sparkplug. Plus, adding some software lets them use UNS, which serves as a traffic cop that makes data consistent and coordinates by normalizing different sources and protocols, so the data that comes out has common units and timestamps."
Eifert adds that Phoenix Contact is also developing a free MQTT Sparkplug Library that will let users load software onto the Linux side of a controller without going through its usual PLC side. This will allow people who are unfamiliar with traditional PLC programming languages to use simple programming methods such as Node-RED to facilitate communications with to servers like Ignition Perspective from Inductive Automation.
PLCnext Edge Gateway makes it simple to connect whatever process they're monitoring or controlling, and send select I/O and tags to a cloud-computing service, such as AWS, Microsoft Azure, Google IoT Cloud, etc. These tasks can be done securely by using a dedicated security appliance with a firewall, passwords and certificates like Phoenix Contact's FL mGuard or the one built into PLCnext, which also includes network routing functions. This allows it to set up a virtual private network (VPN) or firewall on its PLC side or host a broker on its Linux side.
Phoenix Contact also recently launched its BL2 BPC 1500 industrial PC (IPC) that can run Ignition Edge software. Likewise, it's working with ClearBlade's Intelligent Assets edge-computing platform and cloud service to provide edge applications that can also run on PLCNext./p>
"The key is that one system integrator or one in-house person can get data from its producer, broker or consumer, and relay it to whichever operator, manager, manufacturing execution system (MES), enterprise resource planning (ERP) system or other subscriber needs it," says Eifert.
"Whether it's called IoT, IIoT or the newer 'artificial intelligence of things' (AIoT), it's all the same thing—a series of hardware and software for aiding tasks," says Doug Sinclair, business development director for facilities, transportation and infrastructure at Radix Engineering. "IoT adds connectivity and mobility to hardware that they didn't have before with new elements like virtual and cloud computing. It allows users to create environments for real-time data analysis and process optimization. Some IIoT devices perform monitoring, while others are beginning to manage control, but both processes are being augmented by AI and machine learning (ML)."
Sinclair reports that Radix has been working with the University of Massachusetts Amherst to develop an energy command center that consolidates several data sources, such as field flow instruments, power meters, plant equipment and fuel price data. These data sources feed efficiency and optimization models, and enable key performance indicators and forecasts, which are displayed in real-time on dashboards that can be shared and viewed remotely. The university's central heating plant (CHP) provides all steam heat and 70% of campus electricity with a combined-cycle process that runs three boilers fueled by natural gas and ultra-low, No. 2 fuel oil. They each fire at 125,000 pounds per hour, and work with one heat recovery steam generator that adds 100,000 pounds per hour. Total steam generated approaches 1.2 billion pounds per year.
"To reduce costs by better understanding the CHP and its building management and automation, Radix implemented an advisory system that lined up with its operations. It runs real-time models and dashboards, simulates different operating configurations, and applies the best solutions and controls to the assets," says Sinclair. "This solution is based on OSIsoft's PI System software, Thermoflow's Thermoflex heat-balance software and Radix's X Power software. It's designed to save UMass about $1 million per year, and has already reduced operating costs and environmental waste."
Sinclair reports that any overall strategy employing IIoT devices and software should begin by assessing the user's present business plan, and developing a macro strategy. "We can help companies determine a scope for an effective change management program, which also addresses culture and governance for continuous improvement," adds Sinclair. "This scope evaluating existing infrastructure, executing a gap analysis to determine what technology is needed to accomplish the initial plans, and implementing the program and organizational standards required to meet the macro strategy and overall business goals."
"IIoT is like having many different kinds of apples. For example, to achieve cloud-to-cloud interoperability, users also have to understand the data in each type, and decide what is redundant and what needs to be rationalized. Because IIoT technologies vary, it's essential for users, their clients and suppliers to normalize how they're using IIoT-related terminology, and agree on common terms, classifications and metadata as part of their governance process."