Since Manufacturing AUTOMATION has a readership in the automation business, I tend to write about manufacturing and automation in my columns. I do, however, have other interests. I wander off into venture capital, chaos, autonomous agents, supercomputers, rapid transit, movies and biology. Most of these subjects have a lot in common. They are “self-similar,” meaning that the processes of control resemble each other, but are described with different jargon. The sensing processes for automation, buildings, hunting dogs and physics all seem self-similar.
Sensing means to be aware of something or some event. We need to know about sensing for water systems, beer processing and discrete manufacturing. These interests are narrow considering the complete universe of measurement. Let’s see if, by examining another set of applications, it helps us understand ours.
In my youth, I was involved in the design of controllers for HVAC (heating, ventilation and air conditioning). The actual benefit to controlling smart buildings was to ensure comfort, safety and status. Earlier sensing for buildings was odor (our nose) and vision, while today we use smoke detectors. Co-ordinated means of response and control varied from loud noises from neighbours to radio and computer systems. We have gone from independent proprietary networks in the 1990s to integrated interoperability using the Internet Protocol Suite (known as TCP/IP – Transmission Control Protocol and the Internet Protocol) with open graphical user interfaces. Most have agreed that the universal acceptance of today’s networks is here to stay.
Although SCADA is not the term used in HVAC, the system similarity with beer and building make them kissing cousins. Today, HVAC, security, video monitoring, status and comfort can be integrated into a single system. Control of these integrated systems can report alarms, security breaches and after-hours access, as well as control lighting. We need these self-similar requirements in most large systems.
Let’s look forward. All future systems will be wireless, with global secure access anywhere and any-when. The technology needs to incorporate sufficient technology, but the choice of which technology used is irrelevant. Does the buyer really care if thin “one-minute” servers are used or that Intel is inside? Of course not. We will just assume that the available bandwidth is “infinite.” The hardware costs disappear into the mud. More transistors were made in 2008 than grains of rice. Imagine a container of rice grains with a tera-rice capacity. It is a system-oriented sensing and measuring world out there — not one of components. Even SCADA is undergoing change. SCADA is sometimes defined as the interface between silicon life and carbon life (humans). We are now treating it as an interface, not an embedded technology.
An upcoming SCADA conference will examine the human side of the interface. Questions such as display size, controls, alarm prioritization, room temperature and work shifts will be examined. Sensing is a necessary element, but it is useless unless we can act on the information.
Physical bandwidth alone is not a good metric. We need information. Bandwidth compression uses techniques that update the old data in your system. This lets movies be transmitted over low physical bandwidth systems. Using this form of virtual bandwidth substantially improves system performance. Deal with change, not magnitude. Motion detection systems work. As in sailing, dead reckoning with occasional static updates works for boats and submarines.
How do we improve the response of sensing and measurement systems? By being faster than real time. Simulation and modelling work for weather forecasters and predicting election returns. Testing the model using postdiction techniques works well. Using models is not a slam-dunk, but it is better than no guess at all.
Using inference in measurement should be one of the best tools in our toolbox. All single measurements are only good to two digits, and anything to the right of the decimal point is an error. The biological world uses multiple sensors of low accuracy to define the environment. Multiple sensing has an exponential relationship to the accuracy of the observation, while a single measurement is scalar. This is particularly important for large systems. For instance, we can know the size of an iceberg by looking at the ice-salt content and the size above the surface of the ocean.
Applications using advanced techniques are economics, biomedical, pharmaceutical, military and social measurements.
Although I am enamoured with inference measurement, the standard single point data capture will be with us for a long time. Inference allows you to look at your spouse and say, “You don’t look well.” Another way is to take your spouse’s temperature. Both are valid.
Flow, for example, is needed in all processes. The traditional flow meter technologies include differential pressure, turbines, variable area and open channel flow meters. New technologies include coriolis, magnetic, ultrasonic, vortex and thermal offerings, with optical and sonar on the horizon.
Other measurements needed for single point measurements are area, length, time, temperature and pressure. These and more should be treated as components in an advanced sensor application.
I consider engineering necessary, but irrelevant. The science and system concerns are not irrelevant. Using set theory, modelling, computing power and our single point technology will let us deliver robust and better process systems. Do not be afraid.
Dick Morley is the inventor of the PLC, an author, speaker, automation industry maverick and a self-proclaimed ubergeek. E-mail him at email@example.com.