Safety systems have come a long way in recent times.
They have evolved from simple mechanistic shut-down functions to technologies such as safety capable logic, which can react to machine conditions and improve productivity.
To use modern safety systems effectively, designers need new tools.
Current international safety standards provide quantitative methods to calculate risk and reliability.
This is a big shift from the simple qualitative approach of EN 954, which did not require designers to assess the reliability of safety components.
There are many compelling reasons to adopt international safety standards. The most obvious include meeting the requirements of a global market and to lay the groundwork for future expansion.
For example, machines exported to Europe must comply with International Organisation for Standardization (ISO) 13849-1 or International Electrotechnical Commission (IEC) 62061 (also known as AS 62061-1:2005).
Manufacturers can also take advantage of the framework provided by international safety standards to homogenise the operation of their plants around the world.
This leads to cost savings in training and maintenance, as well as increased safety for workers and equipment.
The quantitative approaches of ISO 13849-1 and IEC/AS 62061:2005 are also useful for engineers seeking to explain the need for a particular safety system in an application, or to justify the cost of a safety upgrade in terms of actual risk reduction.
International standards allow companies to demonstrate compliance to customers, and give them confidence that their machines will operate safely, with reduced down-time resulting from component failures.
This can be augmented by employing engineers, such as myself, who have been certified as a Functional Safety Engineers by industry bodies. I obtained my training and certification through TÜV Rhineland.
Sometimes there are competing international standards governing an aspect of the design process. This is illustrated by the two competing safety standards in Europe.
Both of these standards contain a framework and tools to analyse the functional safety of a system — the parts of the control system that ensure the safety of plant and personnel. For designers, the choice of which standard to apply can be confusing.
The ISO and IEC recognise the problem, and are participating in a joint working group to merge ISO 13849-1 and IEC 62061:2005. The process began in 2011 and it is likely that it will take several years to complete.
In the mean time, the main consideration for engineers is to choose a standard that they feel comfortable working with and select safety systems that meet the requirements of the operating environment and machine function.
The IEC standard already operates in Australia as AS 62061-1:2005. It applies to programmable devices, such as safety PLCs, and should be used for these applications.
The standard describes risk, and the ability of the system to reduce it, in terms of Safety Integrity Levels (SIL). SIL 1 is the lowest risk and SIL 3 the highest.
The IEC standard is useful for applications in the petrochemical or power generation industry, as these industries are familiar with the concept of SIL.
In the process industry, risks can exceed SIL 3, so IEC 61508, and the process specific standard, IEC 61511, include SIL 4.
ISO 13849-1 is also applicable in Australia, and is referred to in AS4024:1504. It applies to electrical, mechanical, pneumatic and hydraulic systems.
Under ISO 13849-1, mean time to dangerous failure (MTTFd) for the system is calculated in years.
Instead of SIL, risk and system performance are described using Performance Levels (PL). There are five levels, ranging from PL a to PL e, where PL e is the highest. Table 1 shows an approximate relationship between PL and SIL.
A key difference between these two international standards is the work involved in the calculations of system performance.
Unlike ISO 13849-1, AS 62061-1:2005 does not consider mean time to failure in years, and uses considerably more complex methods to determine the probability of dangerous failure per hour (PFHD).
A free software tool, called SISTEMA (Safety Integrity Software Tool for the Evaluation of Machine Applications), has been produced by the Institute for Occupational Safety and Health of the German Social Accident Insurance (IFA) to perform automatic calculations of Performance Levels under ISO 13849-1.
It can be downloaded from the IFA website.
Safety Lifecycle
Integrating safety and machine functionality during the concept and design phase can deliver a system that minimises risk, meets functional specifications and reduces training requirements.
Specific gains can be made though sharing components between the standard and safety parts of the application and using intelligent safety systems to enhance operations.
The safety lifecycle of a machine starts with a system risk assessment which then flows into the development of the functional requirements for the system.
Once the risk assessment and functional specification are complete, it is time to put your chosen international standard to work in the design and verification process.
For this article, I will use the design flow of ISO 13849-1 to describe what happens next.
Performance Level Structure (Categories)
Performance levels are the basis for quantifying the ability of the safety related parts of a system to respond to risk.
They are based on the system architecture (category); the reliability of the system, represented by the mean time to dangerous failure (MTTFd); and the effectiveness of the system in checking for faults using Diagnostic Coverage (DC) and Common Cause Failure (CCF).
Many engineers are familiar with the use of categories to describe control system architecture. This terminology was used in the now obsolete EN954-1 but remains an integral part of ISO 13849-1.
A graph relating Performance Levels to Categories and average mean time to failure is shown in Diagram 1 (below).
In Diagram 1 above, DCavg is the average diagnostic coverage which is a measure of the test quality applied to components of the system.
The risk graphs from ISO 13849-1 and EN 954-1 are shown in Diagrams 2 (top) and 3 (bottom) alongside.
In these diagrams, S1 refers to the risk of an incident resulting in a minor injury, such as a cut finger, and S2 to incidents with more serious outcomes.
The main difference is that the S2 branch now subdivides, requiring more careful consideration of the safety measures for those systems have inherently lower risks.
Five categories
For a safety related control system, there are five categories: B, 1, 2, 3 and 4. The system behaviour for each of these categories is described in Table 2 (below).
Category B has no specific fault tolerance, but is the basis for the higher categories.
In Category 1 systems, fault prevention is achieved through the use of simple design, and stable and predictable components and materials.
The three highest categories (2, 3 and 4) require that if faults cannot be prevented, they must be detected and the system must react appropriately.
Redundancy, diversity and monitoring are the key concepts employed in reaching this outcome.
System reliability
As mentioned earlier, ISO 13849-1 uses MTTFd (years) as a measure of system reliability.
The MTTFd of a single channel system or subsystem is the average of the MTTFd of each of its elements.
This value can be calculated using the simplified formula provided in the standard.
For a dual channel system or subsystem, the MTTFd of each channel needs to be calculated separately.
Often, the PFHD of systems and subsystems are available from the manufacturer and can be entered directly into SISTEMA.
The MTTFd is limited by the standard to 100 years, although in some cases it may be higher.
The average MTTFd of each system or subsystem is categorised as low, medium or high depending on its value, as shown in Table 3 (below).
This reliability range can then be used to determine PL as shown in Diagram 1.
Diagnostic Coverage
As discussed earlier, the different categories of safety system have varying levels of diagnostic testing.
Diagnostic Coverage (DC) is the term used to describe the system's effectiveness in detecting faults. The failure rate within a system is expressed as Lambda λ.
DC is defined as the ratio of dangerous failures which are detected (λdd) to total dangerous failures (λd) expressed as a percentage (DC = λdd / λd). The failures that pose the greatest threat are the dangerous undetected hazards (λdu).
DC is divided into four basic ranges, as shown in Table 4 (below) and Diagram 1 (above).
One of the important principles of ISO 13849-1 is the need for designers to determine whether the possibilities of faults in both channels of a dual channel system are separate and unrelated.
If failure of a component in one system causes faults in other systems or components, this is considered a single failure.
Events which cause more than one component of the system to fail are called common cause failures (CCF).
CCFs are many and varied, and it is necessary for engineers to employ a diverse arsenal of methods to combat them.
Technological diversity
The approach outlined in the standard is qualitative, and summarised in Table 5 (below).
Simply put, designers need to analyse the possible CCF of their application and mitigate the risk of them occurring.
Annexe F of ISO 13849 lists various measures — including the technological diversity of the design, physical separation of signal paths and electromagnetic compatibility — which can be taken to minimise CCF and assigns a score to each type.
To demonstrate compliance with the standard, designers need to achieve a score of 65 or greater.
International standards not only support global markets and complex safety technologies.
They give designers tools to quantify risk and provide a structured framework to implement integrated safety lifecycle design.
There is no right or wrong when choosing between ISO 13849-1 and IEC/AS 62061:2005: engineers must do their research and decide which standard best suits their design parameters and provides the most workable tools for their application.
[The five tables that appear in this article have been taken from Safebook 4, Rockwell Automation, 2011.]
[Wayne Pearse is Safety Consultant, Rockwell Automation.]