PLC, DCS and SCADA compared: what each does, where control lives, and how to choose for a plant or a network of remote sites
The difference between PLC, DCS and SCADA systems in scope, control, database and communications, where each sits in the ISA-95 levels, the questions that point toward one, SCADA and remote sites over radio and cellular links, and what to plan for over the life of a control system, from obsolescence to backups and cyber security. With a drawing of the three architectures.
Key takeaways
- A PLC controls a machine or process cell, a DCS controls a whole process plant as one system, and SCADA supervises controllers spread over a wide area. The difference is scope and architecture more than the hardware.
- In a DCS the control is in the system's own controllers, sharing one database. In SCADA the control is mostly in the PLCs and RTUs at each site, and the master supervises them over long, unreliable links.
- The lines have blurred. PLC based systems now offer much of what a DCS does, and a DCS can supervise remote sites, so the choice is about the plant and the people more than the label.
- On an operating site, the existing platform usually wins. Spares, skills, the object library and the engineering standard are worth more than a feature list.
- Whatever is chosen, the cost over its life is in obsolescence, backups, cyber security and the skills to fault find it at night. Plan for those on day one.
A processing plant adds a new circuit and the project buys PLCs because they are cheaper per controller than extending the existing DCS. The control works. Then the operators find the new circuit's alarms in a different list, its trends in a different historian and its screens in a different style, and the maintenance team needs a second set of software, licences, spares and skills to fault find it. The saving on hardware is spent many times over in the years that follow.
This guide compares PLC, DCS and SCADA systems: what each is, how they differ in scope, control, database and communications, where each sits in the plant's layers, the questions that point toward one, and what to plan for over the life of the system. The lines between the three have blurred, so the comparison describes the typical case. The standards named are reference points, and the right choice depends on the plant, the site's existing systems and the people who will maintain it.
Three architectures
The three are easiest to tell apart by their architecture. A PLC sits beside the machine it controls, wired to its inputs and outputs, with a local HMI panel. A DCS spreads controllers across a plant on a redundant control network, with operator and engineering stations that all share one database. SCADA puts a master in a control room and talks to PLCs or RTUs at sites that may be hours apart by road.
The hardware inside can be similar. Modern DCS controllers and high end PLCs run on similar processors, and many SCADA remote sites run on PLCs. What differs is how the system is engineered and where its boundaries are, and that decides how it behaves when something fails and how much work it is to maintain.
The label matters less than the boundary. Ask where control happens, where the database lives, and what happens when the link drops.
The three side by side
The table sets out the typical case. Real systems mix them: a DCS supervising remote pump stations, a PLC based system with a single database and redundant servers that behaves like a DCS, or SCADA software running a single plant.
| Aspect | PLC | DCS | SCADA |
|---|---|---|---|
| What it is | A rugged industrial computer that runs control logic for a machine or process cell | A plant wide control system of controllers, operator stations and engineering tools sold as one system | Software that supervises and gathers data from controllers and remote units spread over a wide area |
| Typical scope | One machine, a conveyor line, a packaging cell or a pump station | A whole continuous or batch process plant | Pipelines, water and power networks, rail, and mine sites with remote assets |
| Where control happens | In the PLC, scanning its program every few milliseconds | In the DCS controllers, running continuous regulatory control and sequences | Mostly in the PLCs and remote units at each site. SCADA sends setpoints and commands |
| Database | Tags in each PLC, with the HMI and historian configured separately | One system wide database shared by controllers, displays, alarms and history | A tag database at the SCADA master, mapped to the registers in each remote device |
| Engineering | IEC 61131-3 languages such as ladder, function block and structured text, one controller at a time | The vendor's integrated tools, with libraries of control modules and displays | Configuration of the master, the communications and the displays, plus the remote PLC programs |
| Communications | Industrial Ethernet and fieldbus on the plant floor | The vendor's redundant control network | Radio, cellular, satellite and fibre over long distances, often with DNP3, IEC 60870-5-104, Modbus or MQTT |
| Redundancy | Available, usually added per controller | Built in across controllers, networks and servers | Built in at the master, with store and forward at the remote end |
| Cost profile | Low entry cost that grows with each controller and interface | High entry cost, lower cost per loop on a large plant | Driven by the number of sites and the communications |
Where each sits in the plant
The ISA-95 model, also published as IEC 62264, and the Purdue reference model place control systems in levels. PLCs, DCS controllers and RTUs are basic control at level 1. SCADA and DCS operator stations are supervisory control at level 2. Above them sit the historian and MES, then the business systems such as the ERP and the CMMS. The levels matter for integration and for OT security, because they set which systems may talk to which.
| Level | What it does | Typical systems |
|---|---|---|
| Level 4 | Business planning and logistics | ERP, CMMS and enterprise reporting |
| Level 3 | Operations management | MES, historian, laboratory and production reporting |
| Level 2 | Supervisory control | SCADA, DCS operator stations, HMIs and alarm management |
| Level 1 | Basic control | PLCs, DCS controllers, RTUs and safety controllers |
| Level 0 | The process | Field instruments, drives, valves and motors |
Choosing between them
On a new plant the choice follows from the process and the geography. On an existing site the installed platform usually wins, and for good reason. A second platform doubles the spares, the licences, the training and the number of things that can go wrong at night, and splits the alarms, history and screens that operators rely on. A new platform is worth that cost only when the existing one cannot do the job or is near the end of its supported life.
| Question | Points toward |
|---|---|
| Is it one machine or a small cell? | PLC with a local HMI |
| Is it a large continuous or batch process, with many loops and a central control room? | DCS, or a PLC based system engineered to behave like one |
| Are the assets spread over kilometres, with patchy communications? | SCADA over PLCs or RTUs at each site |
| Does the process need tight loop control, batch management and one database for alarms and history? | DCS |
| Is the plant mostly discrete, sequencing motors, conveyors and valves? | PLCs, with SCADA or a supervisory HMI over them |
| Is the site already standardised on one platform? | Stay on it unless there is a strong reason not to. Spares, skills and the library matter more than features |
SCADA and remote sites
SCADA is built for distance. Water networks, pipelines, rail, power distribution and the remote bores, pump stations and substations of a large mine site are all spread over areas no plant network can reach. The links are radio, cellular, satellite or leased fibre, and they drop out. SCADA protocols such as DNP3 and IEC 60870-5-104 handle that with time stamped events and store and forward, so nothing is lost when the link returns. Newer designs use MQTT for the same reason, as the plant protocols guide covers.
Because the link can fail, the control has to live at the remote site. The PLC or RTU at each pump station runs the pump on its own, and the SCADA master sends setpoints and gathers data. A design that depends on the master to close a control loop over a radio link will fail on the first bad day.
The life of a control system
A control system runs for twenty years or more, far longer than the computers and networks inside it. Most of its cost over that life is in keeping it supported, backed up, secure and understood by the people on site. These belong in the design from the start, not in a list of issues found after handover. The obsolescence guide and the legacy modernisation guide cover the last two rows in more depth.
| Concern | What goes wrong | What to plan |
|---|---|---|
| Obsolescence | Controllers, IO cards, operator stations and operating systems go out of support on different timelines | A lifecycle register with end of support dates and last time buy notices, reviewed yearly |
| Backups | Programs and configurations change in the field and the only current copy is in the controller | Version controlled backups after every change, restored and tested on a schedule |
| Cyber security | Remote access, flat networks and unpatched stations | Zones and conduits under IEC 62443, managed remote access, and patching through change control |
| Skills | Each platform needs people who can fault find it at night | Fewer platforms, a site standard, and training kept current |
| Migration | Replacing a control system while the plant keeps running | A staged migration plan, with the IO and logic documented before it starts |
Where control system choices go wrong
- A second platform added to save on hardware, doubling the spares, licences and skills the site has to hold.
- Control loops closed over a SCADA radio link instead of at the remote site.
- PLC programs with no version control, so the only current copy is in the controller.
- Alarms, history and screens for a new area kept separate from the rest of the plant.
- Remote access added for a vendor and never removed.
- The end of support date for the controllers found when a card fails and no spare exists.
- A choice made on features in a brochure rather than on who will fault find it at three in the morning.
Common questions
What is the difference between a PLC and a DCS?
A PLC is a controller that runs logic for a machine or process cell, programmed and configured largely on its own, with the HMI and historian set up separately. A DCS is a plant wide control system sold as one product: its controllers, operator stations, alarms and history share a single database and engineering environment, with redundancy built in. PLCs suit discrete and smaller processes. A DCS suits large continuous and batch process plants with many control loops and a central control room.
What is the difference between SCADA and a DCS?
Both have operator stations, alarms and history. The difference is where control happens and how far apart the equipment is. A DCS runs control in its own controllers on a fast, reliable plant network. SCADA supervises PLCs and RTUs at sites that may be kilometres apart, over radio, cellular or satellite links, and those remote devices do most of the control. SCADA is built to cope with slow and unreliable communications, using protocols such as DNP3 and IEC 60870-5-104.
Is SCADA a PLC?
No. A PLC is a controller that runs the logic for the equipment around it. SCADA is supervisory software, running on servers and workstations, that gathers data from many PLCs and RTUs, displays it to operators, raises alarms, stores history and sends commands and setpoints. A SCADA system is usually built on top of PLCs, and the PLCs keep running their logic if the link to SCADA is lost.
What is an RTU?
A remote terminal unit is a controller designed for remote sites in a SCADA system. It reads and drives local inputs and outputs, runs local logic, and communicates with the SCADA master over long distance links, usually with protocols such as DNP3 or IEC 60870-5-104 that support time stamped events and store and forward when the link drops. Many modern PLCs can do the same job, and the line between an RTU and a PLC is now mostly about the communications and the environment.
Which is better, PLC or DCS?
Neither is better in general. A DCS is usually the better fit for a large continuous or batch process with many loops, tight integration of alarms and history, and a long life under one vendor. PLCs are usually the better fit for machines, discrete processes and smaller plants, and give more choice of vendor and lower entry cost. On an existing site the installed platform, the spares holding and the skills of the people who maintain it often decide the question.
Key terms
Plain-language definitions from our glossary for the concepts this article leans on.
Standards and further reading
- IEC 61131-3:2025 Programmable controllers, programming languages (IEC)
- IEC 62264-1:2013 Enterprise-control system integration, models and terminology (IEC)
- ISA/IEC 62443 series, security for industrial automation and control systems (ISA)
- NIST SP 800-82 Rev. 3, Guide to operational technology security (NIST)
Related case studies and tools
- Telemetry and monitoring for remote assets (case study)
Related reading
Working through something like this?
See how we approach these initiatives, or tell us what you are dealing with.