AVEVA Plant SCADA (formerly Citect SCADA) is a client-server SCADA package. Everything an engineer builds lives in projects, and a topology decides which computers run the I/O, alarm, trend and report server processes, which computers are clients, and how the servers are grouped into clusters and paired for redundancy. This article explains that structure from AVEVA's Plant SCADA 2026 documentation. Where Plant SCADA sits among AVEVA's products: What is AVEVA.
1. What Plant SCADA is and what it used to be called
AVEVA describes Plant SCADA as SCADA software that runs either as a standalone system or as a distributed client-server system with built-in redundancy and load sharing. The 2020 R2 release notes record the move from the Citect SCADA brand to AVEVA Plant SCADA. The release notes section of the 2026 documentation lists Citect SCADA 7.0 to 2018 R2 among the earlier versions of Plant SCADA. The first monthly patch for Plant SCADA 2026 is dated 11 November 2026.
Two changes in the 2026 release matter on a running plant. Communication between runtime processes, computers and CtAPI is now encrypted by default, which needs a System Management Server and the Runtime Manager running as a service on every computer that hosts a server process. AVEVA Industrial Graphics, the HTML5 graphics added in 2020 R2, has been removed. A separately installed Integration Assistant can turn a project's equipment tags into System Platform objects (InTouch HMI or System Platform). Upgrade paths from Citect SCADA and earlier Plant SCADA versions are covered in Upgrading Citect SCADA to AVEVA Plant SCADA.
2. Projects, included projects and Plant SCADA Studio
Everything configured for a system is stored in projects: graphics pages, variable tags, alarms, communication settings, equipment and code. A large plant can be split into projects, for example one per production line, which are then included in a main project for the whole factory, so each line can be developed and tested on its own. New projects created from a starter project automatically include AVEVA's system projects with page templates and controls.
Projects are built in Plant SCADA Studio, which groups the work into activities, among them:
| Activity | Used for |
|---|---|
| Projects | create, back up, restore, compile and run projects |
| Topology | computers, clusters, I/O devices and the Profile Wizard |
| System Model | equipment, variable tags, alarms, trend tags, SPC tags, accumulators |
| Security | user roles and user or group permissions |
| Setup | alarm categories, devices, groups, languages, events, reports |
| Deployment | versions of the runtime project sent to computers |
| License | AVEVA Enterprise License Manager or a Sentinel key update |
In the System Model, equipment names use dot notation, as in AVEVA's example Plant.Mixer.Line1.Pump, and each piece of equipment becomes the parent of its tags.
3. Topology: computers, addresses, clusters and server processes
A topology lists the computers and I/O devices and the role each plays at runtime. AVEVA's order of work: computers and their network addresses, clusters, server processes (each assigned to a cluster and a network address), redundancy if required, then I/O devices. Rules worth knowing before changing a running system:
- Each network address needs a unique name and an IP address or computer name; localhost and fully qualified domain names are not accepted. A computer with two network cards gets one address per card.
- At startup, runtime compares the topology with the computer's own addresses, once, to decide which server roles it runs; if nothing matches, the computer runs as a client. AVEVA therefore recommends static IP addresses for all Plant SCADA computers.
- A cluster can hold only one redundant pair each of alarm, trend and report servers, each pair on two different computers, but any number of I/O servers.
- If every tag name in the project is unique, tags can be used without naming their cluster; a tag that is not unique and has no cluster context shows #COM.
- Each server type has its own default port: 2080 for alarms, 2082 for I/O, 2084 for reports and 2085 for trends.
4. Server roles and computer roles
The four server process types are I/O, alarm, trend and report. The I/O server is the one that talks to the field: it reads its I/O devices regularly, keeps the values in a cache and answers clients from that cache, so clients never read a PLC directly.
Each runtime computer gets one of four roles in the Setup Wizard, saved as [Client]ComputerRole in Citect.ini:
| Role | What the computer does |
|---|---|
| Server and Control Client | runs server processes and a control client, in one process or as separate processes |
| Control Client | client only; not available on a computer with a server process assigned |
| View-only Client | read-only client |
| HMI (standalone) | one computer with networking disabled |
A computer in single process mode cannot run as a Windows service, and AVEVA recommends the single-process setup only for short-term use, demonstrations and tests.
5. The typical system scenarios
AVEVA describes seven simplified layouts; a real project usually combines several:
| Scenario | In practice |
|---|---|
| Standalone | every server and the client on one computer |
| Distributed I/O | an I/O server at each remote site, or near devices with slow serial links, still in one cluster |
| Client-server | servers and clients spread over a LAN, each found by its address |
| Redundant server | primary and standby I/O servers, plus alarm, trend and report processes on a common primary and standby computer, in one cluster |
| Clustered control | one cluster per site with its own redundant servers; the control room has only clients and sees every cluster |
| Redundant and distributed control | each site's standby servers at another site or in a central location, so one site can be monitored from another |
| Load sharing | two computers acting as standby for each other across two clusters |
The sizing rule from the distributed I/O page: a second cluster is needed only when the project needs more than one redundant pair of alarm, trend or report servers.
6. What each server does on failover
| Server | Normal operation | When the primary fails |
|---|---|---|
| I/O | only the primary talks to the PLCs, so the PLC network load is not doubled | clients switch to the standby; when the primary returns, device communication follows each I/O device's primary or standby setting |
| Alarm | only the main server updates the alarm database; synchronisation keeps the standby up to date | the standby becomes main, and the returning server becomes standby |
| Trend | both servers collect trends in parallel | a restarted trend server fills gaps in its files from its partner |
| Report | scheduled reports run only on the primary, unless the standby is set to run them in parallel | scheduled reports run on the standby; report data is not copied between the two |
Below the I/O server, an I/O device can have standby definitions with a promotion priority, and a second data path can be configured to the same PLC. Each computer can also have two network cards on separate subnets. Two alarm server cases need care: if the pair loses contact, both may run as main, and AVEVA warns against configuration changes until they reconnect; and before the first start, with an empty alarm database, both servers should be connected, or they may create different IDs for the same objects. How other platforms handle this layer: SCADA server redundancy.
7. I/O drivers to the PLCs
AVEVA's advice is to look for a native driver first, then a generic driver such as Modbus or DNP, then an OPC server. Since 2020 R2 the installation media carry a reduced set of core drivers by default; the others are downloaded from the AVEVA Knowledge & Support Center. For Allen-Bradley plants, the ABCLX driver connects to ControlLogix over Ethernet. Its 2026 page lists support for ControlLogix firmware 36, 37 and 38, and notes that the driver downloads and subscribes all PLC tags before it comes online, so it does not suit slow or unreliable links. A route path is needed when the CPU is not in the chassis of the Ethernet module the I/O server talks to.
8. Licensing as the documentation describes it
The computer roles above map to licence types. A server licence also covers one control client, and a server computer needs one licence however many server processes it runs. Licences come either from AVEVA Enterprise Licensing (a browser-based License Manager and a License Server, with activated licences) or from legacy Sentinel USB keys updated with CiUSAFE, never both on one computer. The Floating License Manager is not supported from Plant SCADA 2023. Usage is measured by a dynamic point count: a variable tag counts the first time it is requested at runtime, counting is per tag and not per PLC address, and memory, local and disk I/O variables count only when an external client writes to them. With no licence available, Plant SCADA starts in a time-limited demo mode. AVEVA licensing in general: AVEVA licensing.
9. Compile, run and deploy
A project is compiled after every change, and errors appear in the Compile Messages area. At runtime, the Runtime Manager controls the running state of each component. A deployment server stores each added project as a version and sends the runtime files to deployment clients over an encrypted connection, so it needs a System Management Server.
What TDL checks first
These are our own working habits on an existing Plant SCADA or Citect system, not AVEVA requirements:
- We check the exact release and patch against the release notes, because the 2026 release turns on encryption by default and removes Industrial Graphics.
- We take a full project backup (.ctz) with the included projects and the configuration files, so the Citect.ini files travel with the project.
- We list the clusters, server processes and network addresses, and check that server IP addresses are static.
- We check the I/O driver and its version for each PLC, and for ControlLogix the controller firmware against the driver page.
- We check the licence source, Sentinel key or Enterprise Licensing, and the point count in use, which the Kernel's General page or the CitectInfo function shows.
- We check the Windows version of each computer against AVEVA's installation requirements.
How TDL approaches it
TDL Industrial Automation is a member of the AVEVA Partner Ecosystem, and our AVEVA service page lists AVEVA Plant SCADA (formerly Citect) among the platforms we build, extend and maintain. At the Pha Lai II thermal power plant, we migrated the
Frequently asked questions
Is Citect SCADA the same product as AVEVA Plant SCADA?
Yes. AVEVA's 2020 R2 release notes describe that release as the move from the Citect SCADA brand to AVEVA Plant SCADA, and the 2026 documentation lists Citect SCADA 7.0 to 2018 R2 as earlier versions.
How many alarm servers can one cluster have?
One redundant pair, on two different computers. The same limit applies to trend and report servers, while a cluster can have any number of I/O servers.
Do both I/O servers of a redundant pair talk to the PLC?
No. Only the primary communicates with the PLCs, so the PLC network load is not doubled, and the standby takes over the device communication if the primary fails.
Does a standby report server copy report data from the primary?
No. While both run, scheduled reports run only on the primary unless the standby is set to run them in parallel, and AVEVA does not copy report data between the two.
How does Plant SCADA count points for licensing?
With a dynamic point count: each variable tag counts the first time it is requested at runtime, and counting is per tag rather than per PLC address.
Sources
- AVEVA, About Plant SCADA, Plant SCADA Release Notes, Plant SCADA 2026 Release Notes, Plant SCADA 2026 Monthly Patches, Release Notes for Previous Versions and Plant SCADA 2020 R2 Release Notes, Plant SCADA 2026 documentation
- AVEVA, Configure Plant SCADA Projects, Plant SCADA Activities and Build a System Model
- AVEVA, Define a Topology, Topology, Add Network Addresses, Clusters, Server Processes and Network Redundancy
- AVEVA, I/O Devices, Configure the Computer Role Page and Standalone System
- AVEVA, Typical System Scenarios, Distributed I/O System, Client-Server System, Redundant Server System, Clustered Control System, Redundant and Distributed Control System and Load Sharing System
- AVEVA, I/O Server Redundancy, Report and Trend Server Redundancy, Alarms Server Redundancy, Alarm Server Side Synchronization, Trends Server Redundancy and Reports Server Redundancy
- AVEVA, I/O Device Promotion and Data Path Redundancy
- AVEVA, Determining Which Driver to Use, ABCLX Driver and ControlLogix PLCs via Ethernet
- AVEVA, Licensing, AVEVA Enterprise Licensing, Update a Sentinel Key with CiUSAFE, Dynamic Point Count and Demo Mode
- AVEVA, Compile and Run a Project, Deploy a Project, What Should I Install?, Back Up a Project and Runtime Server Installation Requirements
Read 13 November 2026. This article paraphrases the sources and does not quote them; read the original documents before designing or working on a system.