Phase 4: CPU selection and Control System Architecture

In this article we discuss how to select the CPU and describe the main options for the Control System Architecture (index of design phases).


4-CPU selection and design of Control System Architecture

The selection of CPU and PLC / PAC model depends directly on several parameters. Some of the main ones are:

-The number of inputs and outputs, the number of racks and whether they are local and / or remote I/Os.

-The size of the memory, the power and the speed of program execution (cycle time or scan).

-If we need to program complex or regulation algorithms (PIDs for example).

-If the CPU module must incorporate any communications port and / or any special feature.

-If we have a redundant CPU configuration and what type of redundancy is specified.

-If it is a fail-safe PLC.

Another aspect to consider is whether we are talking about a traditional PLC or PC-based control (“embedded PC” or “softPLC”).

Control System Architecture

The term Control System Architecture is very broad and covers many aspects, from the configuration of the PLC (local and remote racks, redundancy, etc.) to the type of network in its different levels.

Choosing an architecture will depend firstly on what is required in the technical specification, and in addition, on the different aspects of design made so far. Here are some cases:

-PLC / PAC with local I/O racks and without redundancy.

This is the simplest case, from a single chassis with its power supply, CPU and I/O modules to a configuration with several local expansion chassis.

-PLC / PAC with local I/O racks and with redundancy.

The redundancy can be CPU, power supply and I/Os. Normally the term redundancy is associated with the increase in availability (logic 2oo2, 2oo3, 2oo4) of the system, but we can also speak of redundancy to increase safety (logic 1oo2).

When designing a redundant system we must analyze correctly that there are no common mode failures, i.e., no single failure can affect the two redundant channels at the same time. If this happens our system is no longer totally redundant. This is one of the points that require more attention and analysis when designing a redundant control system. In addition, the design is not exclusive to the PLC or control cabinet, but to the Plant as a whole.

Reducing common mode failures to zero is an almost impossible task and in each case the designer should focus on those most likely to occur.

-Control System Architecture with decentralized controllers – Remote I / O

There may be different reasons for decentralizing I/Os: i) saving wiring; ii) redundancy criteria; iii) distribution and segmentation of I/Os; etc.

It is also possible to use fieldbus to communicate the CPU directly with some devices (drives, etc.)

In large installations, the decentralization of controllers by production areas is a good practice. Nowadays fiber optic networks give us a very high quality of communication in almost any type of network.

-Basic industrial network

In the simplest configuration we will have a PLC with local I/O and a PC / SCADA or terminal for the operator connected to the communications port of the CPU.

-Industrial network (control level)

Manufacturers offer different typologies (star, ring, bus or a combination).

Typically, at this level of control (“control level”) we connect the controllers, the SCADA or the operating stations, and the engineering station. Today Industrial Ethernet is being imposed in its different versions of protocols (Ethernet/IP, Profinet, EtherCAT, PowerLink, Modbus-TCP, etc.)

One of the aspects to be considered (not always defined by the technical specification) is the determinism of the network to ensure that the information is transmitted from one node to another at a specific time. Classic industrial networks such as ControlNet or Profibus are deterministic, but not the Ethernet of the IT world. Many of the industrial protocols that use Ethernet have been adapted to be deterministic and are suitable for real time but there is still no standard for this. There are many expectations put into Ethernet TSN that can be in short time a standard for the industry.

– Industrial network (field bus)

Used to connect I/O blocks and intelligent devices (sensors, actuators, drives, MCCs, etc.)

The most used are the following: Profibus DP, Modbus-RTU, CC-Link, CAN, DeviceNet, etc.

-DCS (Distributed Control System)

In large installations, it is used the Distributed Control Systems (DCS), consisting of at least two controllers with their I/Os, one or two servers, the operating stations and an engineering station. We will have at least one fiber optic network and sometimes also one or several fieldbus. In this type of Plant the location of the controllers by zones usually provides interesting advantages.

-Wireless Network

There are Plant Wireless Networks normally destined to interconnect devices like camcorders, tablets, mobileworkers, tracking of assets, etc. and Field Wireless Networks designed to interconnect field sensors and final control elements for process measurement and control. Field Wireless Networks can be integrated into plant networks via Ethernet or radio cable. The two major Wireless standards for the process industry are ISA 100.11a and WirelessHart.

It is a solution to take into account mainly in modifications and extensions of existing plants and for the monitoring of non-critical signals.

Link to next phase.