industrial//industrial automation//DCS

A DCS (distributed control system) is an industrial control system whose many controllers are spread over the plant and work as one, and it is used to run large industrial processes in which many variables interact. Its typical traits are high availability, redundancy, integrated engineering, alarm management, PID control, a historian, operator stations and a very large number of signals.

A DCS (distributed control system) is an industrial control system whose many controllers are spread over the plant and work as one, and it is used to run large industrial processes in which many variables interact. Its typical traits are high availability, redundancy, integrated engineering, alarm management, PID control, a historian, operator stations and a very large number of signals.

Summing a DCS up as a hundred redundant PLCs connected together, plus alarms, is a good first intuition and a misleading definition. A DCS is not bought as a hundred PLCs wired together. Its controllers, networks, engineering stations, operator stations, alarms, history, redundancy and configuration management are designed from the start to work as one control system with the rest of the plant. The integration is the product. A PLC is a controller you integrate, a DCS an integration you buy: many PLCs plus a SCADA can imitate one, but redundancy, alarm philosophy, change management and diagnostics then become the integrator's problem instead of the product's.

Classic platforms: Emerson DeltaV, Honeywell Experion, Yokogawa CENTUM, ABB System 800xA, Siemens PCS 7 and PCS neo. DeltaV is much more than a big PLC: from one engineering environment the engineer configures controllers, PID loops, alarms, graphics, historization, redundancy, communications and stations.

Its natural terrain is process automation: keep pressure at 5 bar, level at 65 %, temperature at 180 degrees and flow at 200 t/h while the variables interact, which means tens, hundreds or thousands of control loops managed together. Each is a PID block with its own alarms, limits and history (feedback control designs them one at a time), and the platform's job is running all of them without an engineer per loop.

Operation and engineering are separated on purpose: the engineering station changes logic and configuration, the operating center drives the process (HMI).