Basic concepts of redundant architecture of PLC

In this article we will go over the most usual options to configure a redundant architecture. We will focus on some  of the most commons and we will have occasion to study them more deeply in a future post.

Depending on the application we will have to determine whether what we want is a “high availability system”, ”a high safety system” or both of them. Also we must always consider the cost of architecture.

There are different technologies that we will analyze in the future: TMR (Triple Modular Redundant), QMR (Quadruple Modular Redundant), FMR (Flexible Modular Redundancy), XMR, etc. and different logics 1oo2, 2oo2, etc.

1-Redundant CPU and non redundant I/Os
It is a very simple architecture and it is designed with MTBF criterion (Meantime Between Failures) because it is substantially lower in a CPU than in an input/output module. It is a value given by the manufacturer based on a study of probabilities.

2-Redundant CPU and non redundant I/O but well distributed
In this case we take a step forward and we distribute the inputs and outputs always thinking about field devices. For example, if we have a unit with two pumps at 50% capacity, we will not mix I/O of each pump in the same modules so that if the module fails we will only lose one of the pumps. Also, if possible, we shall place each pump module in a different rack.

3-Redundant CPU and well distributed mixed I/O
We will use a “2 out of 3 logic” (2oo3) for inputs and, optionally, a dual logic or H-shaped for digital outputs.

2oo3 logic philosophy is a simple system of voting. For example, we have to design a pump shut-down when having high level and we have three level sensors placed in the same point. The pump only will be stopped when at least two sensors detect high level, but not if only one is detecting it. This avoids unnecessary shut-downs in case of failure of field instrument or PLC input.

It is important to place each of the three inputs in a different module and, if possible, in a different rack. We must decide in which inputs we will use 2oo3 logic and the type “digital or analog“. There is not a general optimum criterion because it depends on each application and on the cost we are willing to pay.

We should at least, triple the field signals that cause general shut-downs of the process or of the unit.

4-Redundant CPU and Redundant I/O and/or 2oo3 logic
The difference with the previous case is the redundant I/O that we will use for a certain group of I/O looking for its “high availability”. Not all manufacturers allow this type of architecture because it is expensive although there are critical applications that require it.

We will treat deeply in next articles this topic of redundant architecture of PLC and explain how to use IO Builder.