The Internet of Things (IoT) is at the height of its hype cycle, along with its associated buzz word, Big Data. It is instructive to look at the hype and see how IoT will be used outside of manufacturing and what issues and problems will be seen in the general use area and then relate those applications, issues, and problems to an Industrial Internet of Things, which will be quite different and have different applications than the general Internet of Things may have.
We will discuss how the Internet of Things and its variant, the Industrial Internet of Things, will affect automation and control systems and the process industries and manufacturing in general.
Beginning fifteen or twenty years ago, people discussed the benefits of interconnecting all the sensors everywhere, giving us Machine-to-Machine control (M2M). Your refrigerator could call and order your supplies directly from the grocer without you having to do anything other than make sure the money to pay for it was in your account. Less improbably, you HVAC could interact automatically with energy suppliers, making demand-response power pricing practical.
The Internet of Things was originally called M2M, now M2M is considered a core part of the IoT. However, the number of sensors and other nodes that will have to be connected together to form an Internet of Things, or an Internet of Everything, is so large that it will require wholesale adoption of IPv6 (Internet Protocol version 6). To date, IPv4 stubbornly continues to be the protocol version in use, even though there are no new IP addresses. A variety of workarounds have been established to make it possible to continue to use IPv4.
In order to produce the IoT devices, the following have to apply: very low power radio networks, very inexpensive "lick and stick" sensors, and intelligent final control elements. The sheer numbers of these devices required for the IoT will feed back into the design and availability of these sensors for the Industrial Internet of Things, exactly the same way as advances in design and economies of scale for automotive sensors have reduced prices for many devices used outside of the automotive environment.
The Industrial Internet of Things, working with Smart Manufacturing systems, will be able to produce a revolution in the way manufacturing is done, especially in discrete manufacturing and batch processing, but also effect a considerable change in process manufacturing as well. Based on the way the Internet of Things is designed, an "app-based" design approach may well produce the agile, limber process environment and process control systems that have been called for in the past ten to fifteen years.
The aggregation of sensors and data in the Industrial Internet of Things will first to able to revolutionize the process control lifecycle. Completely automating maintenance work orders, diagnostics, and calibration will be among the first major effects of the IIot. Connecting to vendor purchase networks automatically, for replacement and repair, will be another major effect. This will permit maintenance and operations personnel to concentrate on causing the control system to work in an optimized fashion and not spend time collecting and aggregating data and inputting data into dissimilar systems.
Using RFID technologies, inventory can be made entirely automatic. Delivery of raw or intermediate materials using robot-guided vehicles can also be made practical and will improve time to market and agility. RDID technologies can also be used to improve worker and asset safety by providing location services both personnel and critical assets such as fire trucks and safety gear.
It IIot will also affect how simulation and modeling can interact with the real time process. Models can be much more detailed with large amounts of data available from the IIot and simulation can be morphed into ways to meta-control the process in real time.
But...There is a problem with security. Increasing drastically the number of sensor and controller nodes on a control system network and extending the network beyond the physical boundaries of the plant to include suppliers and supply chain networks, increases the potential for threat to the system in a topologically complex way. Increasing the number of sensors and controllers, as well as other network nodes, increases the threat surface available to invaders of the system. It also opens the network and the control system to physical and cyber-physical attack, not just cyber attack.
The control systems of today cannot be made safe with the number of sensors and controllers and the limited complexity of industrial networks currently in existence or in design. In order to operate safely within the Industrial Internet of Things, control systems and industrial networks must be re-designed from the beginning to enhance safety and security and prevent both accident and cyber-intrusion. This will require an entirely new class of control system.
What will the IoT and IIot mean for vendors? The implications of the Internet of Things, Big Data, and the Industrial Internet of Things are enormous. They will create a completely different vision of control systems and how to control process plants based on the amount of data and the availability of data and the ability to mine and refine that data into usable information.
The theory of Big Data brings to process control and manufacturing not only the concept of complex systems, but the complex systems themselves in practice.
The IIoT will make the entire sensor network, including final control elements, and the safety instrumented system, and the control system a single complex system. Adding to the complexity will be the integral interconnections to the supply chain and to the enterprise. This will especially be true if, as is predicted, it sill become commonplace for the business systems to be seamlessly connected to the control systems.
This clearly has implications for the design and operation of plant control systems. Control systems have always been somewhat isolated from the business systems of the plant, as the Purdue model and its many variants have shown. The Industrial Internet of Things will will force the control system to be a part of a "network of networks" and be capable of interacting easily and in an agile manner with all the other networks that surround it in the business enterprise, however large.
Automation system vendor, as some already have, must embrace the IIoT by whatever name the vendor wants to call it. The vendor must also embrace the theory of Smart Manufacturing, again, by whichever of the many names currently is use the vendor prefers to use.
The IIot will finally do for sensors and networking what the PC did for control systems. The introduction of the PC produced a COTS (Commercial Off The Shelf Systems) platform onto which the control system software could run. IIot will provide the COTS sensors and networks that will be usable with no or minor modification in the industrial environment. The reason for this is that the sensors and networks will have to be more robust, not less, than the current technologies for sensors and sensor networks because they will be used in electric grid, building automation, and home automation systems where the level of training and support will be significantly lower than the standard in process automation.
Big Data and the Future of Control Systems Architecture
Ever since the development of large rational databases, theoreticians have discussed what could be done if the database were to be large enough or if multiple databases were interconnected in useful ways.
Big Data is at the height of its hype. It can do anything for anybody. It is like magic and not entirely well understood. But Gartner says, "Through 2015, 85% of Fortune 500 organizations will be unable to exploit big data for competitive advantage." This is true of both business systems and process control systems.
Arguably, the automation systems vendor that first, or best, applies the theory of Big Data and its ability to improve operation of both the control system and the enterprise will have a significant strategic advantage over automation systems vendors who maintain the status quo regarding system design and use. There is, thankfully, meaning at the bottom of the hype about Big Data. We'll concentrate on the sues of Big Data in process automation. As the Internet of Things will provide the same injection of COTS enablement that the PC provided in DCS and SCADA systems, Big Data's analytical tools for process automation will be modifications and extensions of the COTS tools that have been developed over the past decade to sue the ocean of data we swim in.
Big Data applications in industrial control systems
There are hundreds of potential applications in the industries process automation serves, including design of processes, design of products, and the actual design of process plants and discrete automaton systems for factory automation.
Big Data will have a major impact on modeling and simulation both off-line and in real time. Big Data will make possible relatively inexpensive and high-resolution 3D training environments using the predictive analytics that have already been developed for consumer analysis and using COTS data for things like piping, values, etc. so that such objects do not have to be developed independently. Big Data can reduce costs to the point that using online modeling and simulation tools to predict outcomes of batch reactions becomes practical. Big Data will be used to improve operations as well as gaining insights that are not entirely obvious or easily apparent.
Another major implication for Big Data (and the Internet of Things) is how the architecture of supply chains will be changing. Currently, most people (and most product developers and product users) visualize the plant, enterprise, and supply chains as two-dimensional representations. Big Data, and the Internet of Things, will force us all to see these constructs as three dimensional, moving through time-in other words, for dimensional constructs.
Big Data is, simply, relational databases, and databases of databases, containing extremely large amounts of data. The amount of data beyond a certain point isn't important--what are important are the tools that are available to analyze the data and use the data to reach important conclusions that are not obvious available from simple data scans.
An example of this is the movement from real time data analysis for maintenance and diagnostics to including in the data historical data, data from other sources outside the plant, manufacturers' data, and other information to allow the maintenance managers to make even more informed decisions.
This example leads inextricably to the conclusion that two-dimensionality of the Purdue model, and its modified successors, is not correct. As data flows from sensors, its path no longer can be linear. Currently, the flow of data is conceived to be linear and unidirectional. How it really will flow is best described as a web that is not linear and not single connection in-and-out. Each piece of data will have multiple connections, multiple uses, and reside in multiple databases. Visually, this data flow will be indistinguishable from the data flow in the larger Internet. This concept alone will require a major rethinking of how process and discrete control systems interact with data.
Typically, data feeds to control systems have been confined to field sensors and feedback from final control elements. Sometime,s OPC feeds from special purpose computers and APC control information are also fed to the plant control system. The control system of the future will require inputs from all those and also from the plant and from other plants in the enterprise.
The architecture of the control system of the future will be completely determined by the Internet of Things, and the uses Big Data can bring to the industrial plant or factory setting.
Article above is from the March edition of Industrial Automation INSIDER.