ISO 13849

On this page 7

ISO 13849 is the international standard for the design and validation of safety-related parts of control systems (SRP/CS) in machinery, and it is one of the core references in robot safety. It expresses how reliably a safety function performs as a Performance Level (PL) from a to e, derived from architecture, component reliability, diagnostics and resistance to common cause failures.

In robotics, the standard answers a practical question: is the circuit that stops a robot when a light curtain is interrupted reliable enough for the risk it controls? Robot standards such as ISO 10218 refer to it directly, so integrators use ISO 13849 to prove that emergency stops, guard interlocks, speed monitoring and other functions meet their required Performance Level (PLr).

Structure and scope of ISO 13849

ISO 13849 consists of two main parts with distinct roles. Part 1 sets out design principles and the method for estimating the achieved PL, while Part 2 describes validation: analysis and testing that confirm the design actually delivers the specified safety function. The current edition of Part 1 was published in 2023 and refines the earlier 2015 text rather than replacing its method.

The standard is technology neutral. It covers electrical, electronic, programmable electronic, hydraulic, pneumatic and mechanical control parts, which suits robot cells that combine safety PLCs, valves, brakes and sensors. Under the EU Machinery Directive and the new Machinery Regulation, ISO 13849-1 is a harmonised standard, so compliance gives a presumption of conformity for control system safety requirements.

The following table summarizes what each part contributes to a robot safety project.

PartMain contentTypical output
ISO 13849-1Design principles, risk graph, PL estimationRequired and achieved PL per function
ISO 13849-2Validation by analysis and testingValidation plan and test records
Technical reportsWorked examples and calculation guidanceReference architectures
Parts of the standard and their purpose

Determining the required Performance Level

The required Performance Level follows from the risk assessment of each hazard. ISO 13849-1 offers a risk graph with three parameters: severity of injury (S1 slight and reversible, S2 serious or irreversible), frequency or duration of exposure (F1 seldom or short, F2 frequent or long) and the possibility of avoiding the hazard (P1 possible, P2 scarcely possible). The 2023 edition also allows the probability of the hazardous event to lower the result in justified cases.

Robot cells usually land high on this graph because moving arms carry substantial kinetic energy and operators enter the cell regularly for loading or teaching. As a result, ISO 10218-2 generally requires PL d with Category 3 architecture for robot safety functions unless the risk assessment justifies another value.

The table shows typical outcomes of the risk graph for robot-related hazards.

Hazard situationS / F / PTypical PLr
Crushing by a large industrial arm during loadingS2 / F2 / P2PL e
Entry through an interlocked cell doorS2 / F2 / P1PL d
Pinching at a slow collaborative gripperS1 / F2 / P1PL b
Occasional maintenance near a stopped conveyorS2 / F1 / P1PL c
Example risk graph outcomes in robot cells

Categories, MTTFd, DCavg and common cause failure

The achieved Performance Level is estimated from four quantities. The Category (B, 1, 2, 3 or 4) describes the architecture and its behavior under faults; MTTFd is the mean time to dangerous failure of each channel; DCavg is the average diagnostic coverage; and CCF measures protection against failures that defeat redundant channels at once.

Category 3 means a single fault does not cause loss of the safety function, and Category 4 additionally requires detection of faults and tolerance of accumulated faults. Category 2 relies on periodic testing of a single channel, while Categories B and 1 have no fault tolerance at all and depend on component quality. For Categories 2, 3 and 4, a CCF score of at least 65 points is mandatory.

The parameter bands used in the simplified estimation method appear below.

ParameterBandRange
MTTFd per channelLow / medium / high3-10 / 10-30 / 30-100 years
DCavgNone / lowBelow 60% / 60% to below 90%
DCavgMedium / high90% to below 99% / 99% or more
CCF scoreMinimum for Categories 2-465 of 100 points
Parameter bands for PL estimation

Combining these values gives an average probability of dangerous failure per hour (PFHd), which maps to a PL. Engineers often perform this calculation with tools such as SISTEMA, but the inputs must come from manufacturer data such as B10d values for valves and contactors, not from assumptions.

Performance LevelPFHd range per hourTypical robot use
PL a10^-5 to below 10^-4Rarely used for robot functions
PL b3 x 10^-6 to below 10^-5Low-energy auxiliary functions
PL c10^-6 to below 3 x 10^-6Some maintenance functions
PL d10^-7 to below 10^-6Default for robot safety functions
PL e10^-8 to below 10^-7High-energy, frequent-exposure hazards
PL grades and their PFHd ranges

How ISO 13849 relates to other robot safety standards

ISO 13849 is a generic machinery standard, while robot standards define what the safety functions must do. ISO 10218 sets requirements for industrial robots and their integration, ISO/TS 15066 adds biomechanical limits for power and force limiting, and both point to ISO 13849 or IEC 62061 for the reliability of the control implementation.

IEC 62061 is the main alternative and expresses reliability as a Safety Integrity Level (SIL) derived from IEC 61508. The two approaches are broadly comparable: PL d corresponds roughly to SIL 2, and PL e to SIL 3. Many integrators choose ISO 13849 because it handles mixed technologies such as pneumatics more naturally.

StandardReliability metricMain strength
ISO 13849-1PL a to eAll technologies, simplified architectures
IEC 62061SIL 1 to 3Complex electronic and programmable systems
IEC 61508SIL 1 to 4Basic standard for any industry
ISO 10218Refers to PL or SILRobot-specific functional requirements
Comparison with related functional safety standards

Common mistakes when applying ISO 13849

Most failures in practice come from treating PL as a component property rather than a property of the complete safety function. A PL e safety relay does not make a function PL e if the sensor, wiring or output contactors are weaker links.

  • Ignoring the whole chain: the input device, logic unit and output actuator must all be included in the calculation.
  • Skipping validation: a correct calculation without ISO 13849-2 testing leaves wiring and software faults undetected.
  • Overlooking systematic failures: safety software needs specification, code review and testing, not only hardware numbers.
  • Wrong usage assumptions: B10d-based MTTFd depends on the real number of operations per year, which robot cells often exceed.
  • Bypassed devices: operators defeating interlocks invalidate every calculation, so design must make bypassing hard.

The simple explanation

ISO 13849 is a rulebook for building safety circuits in machines, including robots. It asks two questions. First, how dangerous is the situation? Second, how dependable is the circuit that protects people from it?

The answer is a grade from a to e, called the Performance Level. A higher grade means the protection is less likely to fail when it is needed. Think of it like a bridge rating: a bridge for heavy trucks needs stronger design than a footbridge. In the same way, a large robot that people approach every hour needs a more dependable stopping system than a small machine that is rarely touched.

How it is used in practice

An integrator builds a palletizing cell with a 180 kg payload robot. Operators replace pallets several times per hour through an opening protected by a light curtain. The risk assessment rates the hazard S2, F2 and P2, so the required level is PL e, while the door used for maintenance receives PL d.

The team designs a dual-channel Category 4 function: a Type 4 light curtain, a safety PLC and two monitored contactors that cut drive power. Calculation gives a PFHd of about 2 x 10^-8 per hour. Validation then includes fault injection tests, such as disconnecting one contactor feedback wire, to confirm that the cell refuses to restart. The outcome is a documented, validated PL e function that supports the CE declaration of conformity.