Print this article

Comment

01 October 2003

Grasping distributed control

By Samuel Herb

In early fluid processing plants, controlling the process frequently required many operators. They continuously circulated around each process unit, observing locally mounted instruments and manipulating the valves. Overall plant operations would often require operators to "tour the plant" with a clipboard, manually recording key parameters. At the end of the first pass, following appropriate calculations, the operator would make a second trip to adjust controls.

Technology improved, and the ability to transmit pneumatic signals became a reality, so the control room came into being at larger plants. Operators began moving the large indicating instruments to one location, along with some controls that transmitted signals back to valves in the plant. Operators could now record their readings in logbooks and make adjustments in the operating process without touring the plant as frequently. They still had to tour the plant to adjust the more distant valves, dampers, and other end elements.

Following World War II, electronic controls became more rugged, and new types of sensors measured parameters not previously measurable. Controllers got smaller, so more of them could fit on a panel. Further, as computers became less expensive, they became more common in large facilities. The centralized control room became increasingly common and steadily more complex.

Video technologies, with their ability to display data or allow the operator to initiate control actions, made possible the onset of distributed control. Thus, the central control room could provide centralized information without having all the processing in one location; this distributed the overall risk to the control system while reducing the cost and complexity of wiring.

Now operators no longer have to tour the plant. They can literally "let their fingers do the walking" as they call up each controller or group of controllers on their screens to check the progress of their process. If necessary, they can easily make set point and output changes from their keyboard, as well as respond to any alarms if a process is "off normal."

Distributed control systems distribute risk by dispersing control throughout the plant: A problem at location X needn't ruin operations at location Y. Further, plants receive data and act on it for virtually all practical purposes in real time. High-quality operator interfaces enable a knowledgeable operator to know what's going on in his plant with little more effort than a glance.

DCSs have some noteworthy weaknesses, though. First, they are expensive to purchase and install. What's more, the operating system and communications protocols are proprietary. The operator must select wisely lest, as Ben Franklin liked to say, he repent at leisure.

In the early to mid-1990s, when a process required a high level of discrete action, along with sophisticated process action, architectures that could combine distributed controllers and PLCs in the same network began to appear. The advantage of these hybrid systems is they allow the more complex system user to tailor the equipment selection by unit operation requirements. It also segregates the addressing of safety issues, splitting the configuration and troubleshooting tasks between the DCS portion and the PLC as a separate part of the system so they don't interfere with each other.

A big disadvantage of these hybrid architectures is that they're more complex. There are high integration costs for configuring, programming, documenting, and training. It may be necessary to match different systems with no common protocols. And you will have to rely on data links between these PLCs and DCSs. IT

Samuel Herb is a process control specialist in the Foxboro Automation Platform Marketing group of Invensys.

There are over 100 distributed control systems in use. Different vendors assemble varied architectures for process control.