ISO 13849

ISO 13849-1 has its origins in the mid 1990s when the European Committee for Standardization (CEN) published EN 954-1, Safety of machinery — Safety-related parts of control systems — Part 1: General principles for design[5] in 1996. In 1999, EN 954-1 was transferred to ISO for ongoing development under the Vienna Agreement.

EN 954-1 introduced the original five structural Categories, B, 1-4.

prEN 954-2:1999, Safety of machinery — Safety-related parts of control systems — Part 2: Validation, is the precursor document that eventually became ISO 13849-2 in 2003. This document was never published as a finished standard. The "pr" in "prEN" indicates that the document was a European pre-standard.

In 1999, ISO published the first edition of ISO 13849-1, Safety of machinery — Safety-related parts of control systems — Part 1: General principles for design. The first edition was technically identical to EN 954-1. Within a year after publication, ISO/TC 199 launched a New Work Item Proposal for the revision of the standard. The goal was to add probabalistic requirements to the existing standard.

In 2003, ISO 13849-2, Safety of machinery – Safety-related parts of control systems — Part 2: Validation, was published. This standard included all of the details related to validating the functional safety of a design. In addition, Annexes A-D included key information on basic and well-tried safety principles, well-tried components, and common faults for mechanical, hydraulic, pneumatic, and electrical components.

The second edition of ISO 13849-1 was published in 2006. That edition introduced MTTF_d, DC_avg, and CCF for the first time. The revisions incorporated the recommendations developed through the EU STSARCES project.[6] and [7]

In 2012, ISO 13849-2, Safety of machinery – Safety-related parts of control systems — Part 2: Validation was published. This edition was reaffirmed in 2017 and remains current.

The third edition of ISO 13849-1 was published in 2015. The revision included additional technical explanations, and clarification of the analytical methods. This edition was reaffirmed in 2020, while a new revision was started. The fourth edition was published in May 2023.

Following ISO 13849-1, the design of the safety system is based on a risk assessment performed by the manufacturer of the machine.[8] The risk assessment identifies the safety functions required to mitigate risk and the performance level these functions need to meet to adequately mitigate the identified risks, either completely, or in combination with other safeguards, e.g., fixed or movable guards or other measures.

The Annex A decision tree, Figure A.1, is provided as an example of how the PL_r can be determined. The Annex A method is not a risk assessment tool since the output from the tool is in terms of Performance Level, not risk. Figure A.1 cannot be used for risk assessment. Examples of a risk matrix and a risk decision tree are given in ISO/TR 14121-2.[9] Risk assessment is typically done in at least two cycles, the first to determine the intrinsic risk, and the second to determine the risk reduction achieve by the control measures implemented in the design.

A safety function is a control system function whose failure will result in an immediate increase in risk.[8] ISO 13849-1 includes descriptions of a number of common safety functions, including:

• safety-related stop
• start/restart
• manual reset
• local control
• muting
• response time
• safety-related parameter(s)
• fluctuation, loss and restoration of power sources

Each safety function identified in the risk assessment is assigned a required Performance Level (PL_r) based on the intrinsic risk determined through the risk assessment. The intrinsic risk is the risk posed by the machine if no risk control measures were present, or if the risk control measures fail or are defeated by the user.

Each designated architecture has an associated block diagram. When analyzing SRP/CS designs, a block diagram should be developed to assist the analyst in calculating the MTTF_D of the functional channel(s).

Category B represents the basic category. This category is single-channel, and can include components with MTTF_D = Low or Medium. Components must be suitable for use in the application, and specified appropriately for the conditions of use, i.e., voltage, current, frequency, switching frequency, ambient temperature, pollution class, shock, vibration, etc. Since Category B is single channel, DC_avg = NONE. CCF is not relevant in this category.

The maximum PL = b.

Category 1 achieves increased reliability as compared to Category B through the use of MTTF_D = High components. These components are deemed "well-tried components" and are listed in ISO 13849-2, Annexes A through D. Additionally, components that have been tested by the manufacturer and approved according to the relevant component safety standard, e.g., IEC 60947-5-5, are also considered well-tried. Since Category 1 is single channel, DC_avg = NONE. CCF is not relevant in this category.

The maximum PL = c.

Category 2 is a single-channel architecture that achieves increased reliability by building on Category B, using components with MTTF_D = Low to High, and adding diagnostic capability through the use of test equipment. The DC_avg for Category 2 can be Low to Medium, i.e., 60% ≤ DC < 99%. The diagnostic frequency depends on the demand rate on the safety function, and on the PL_r that must be achieved. A minimum CCF score of 65 is required, see Annex F.

The maximum PL = d.

Category 3 is the first architecture with a redundant structure. Building on Category B, and using components with MTTF_D = Low to High, this architecture introduces cross-monitoring between the two active channels, as well as cyclic monitoring of the output device(s). Category 3 requires DC_avg Low to Medium, i.e., 60% ≤ DC < 99%. A minimum CCF score of 65 is required, see Annex F.

In Category 3, no single component failure is permitted to cause the loss of the safety function.

The maximum PL = e.

Category 4 is also a redundant architecture that builds upon Category B. Using components limited to MTTF_D = High, this architecture includes cross-monitoring between the two active channels, as well as cyclic monitoring of the output device(s). Category 4 requires DC_avg HIGH, i.e., ≥ 99%. A minimum CCF score of 65 is required, see Annex F.

In Category 4, no single component failure is permitted to cause the loss of the safety function.

The PL = e.

The primary differences between Category 3 and 4 are that Category 4 requires:

• MTTF_D components in the functional channels
• DC_avg ≥ 99%
• Accumulation of faults between diagnostic cycles cannot cause the loss of the safety function
• All of the faults that occur between diagnostic cycles must be detected when the diagnostics run