Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
ISO 29473:2010 — Fire Tests — Uncertainty of Measurements in Fire Tests
Applying GUM methodology to fire testing: measurement uncertainty principles for fire safety engineering
Introduction to Measurement Uncertainty in Fire Testing
ISO 29473:2010 provides a framework for evaluating and expressing measurement uncertainty in fire test methods developed by ISO/TC 92, based on the approach presented in the ISO/IEC Guide 98-3 (GUM — Guide to the Expression of Uncertainty in Measurement). The standard addresses the unique challenges of fire testing — high temperatures, aggressive combustion products, transient phenomena — which create uncertainty sources not typically encountered in conventional metrology.
Fire test results are used for regulatory compliance, product development, and fire safety engineering. Without quantified uncertainty, it is impossible to determine whether differences between two test results are significant or merely experimental noise.
The standard applies to quantitative tests producing results in engineering units — such as heat release rate measured by oxygen consumption calorimetry (ISO 5660-1). It explicitly excludes tests producing pass/fail or index-based results, where uncertainty plays a different role.
Uncertainty Evaluation Framework
| Component | Definition | Fire Test Example |
|---|---|---|
| Type A evaluation | Statistical analysis of repeated measurements | Replicate cone calorimeter runs on identical specimens |
| Type B evaluation | Evaluation by other means (specifications, calibration data, literature) | Oxygen analyser manufacturer’s stated accuracy, thermocouple tolerance |
| Combined standard uncertainty (uc) | Root-sum-square of all component uncertainties | Combined effect of all measurement inputs on heat release rate |
| Expanded uncertainty (U) | uc × coverage factor k (typically k = 2 for 95 % confidence) | Reported uncertainty range for the test result |
The standard uses the cone calorimeter heat release rate measurement as its primary worked example, deriving an uncertainty budget that includes contributions from: oxygen mole fraction measurement, exhaust flow rate, orifice coefficient, ambient conditions, and data acquisition resolution.
Fire tests present unique challenges for uncertainty estimation — transient combustion behaviour, specimen heterogeneity, and the difficulty of replicating exactly identical fire scenarios mean that Type A uncertainties can be significantly larger than those in conventional physical testing.
Engineering Applications and Practical Considerations
For fire safety engineers, understanding measurement uncertainty is essential when using test data as input for fire modelling or performance-based design. The standard deviation of heat release rate measurements from a cone calorimeter — typically 5–15 % of the mean value depending on material homogeneity — directly affects the confidence interval of predicted fire growth and spread.
For testing laboratories accredited to ISO/IEC 17025, estimation of measurement uncertainty is a mandatory requirement. ISO 29473 provides the sector-specific guidance needed to fulfil this requirement for fire test methods, including templates for uncertainty budgets and worked examples that can be adapted to specific test configurations.
The standard includes an informative annex on basic concepts of measurement uncertainty, making it accessible to fire test engineers who may not have a formal metrology background. The mathematics involved (sensitivity coefficients, partial derivatives, effective degrees of freedom via Welch-Satterthwaite) is presented with clear fire-test-specific context.
A notable feature of ISO 29473 is its honest acknowledgment of the limitations: “It is not always possible to quantify the uncertainty of fire test results as some sources of uncertainty cannot be accounted for.” This transparency is crucial for maintaining scientific rigour in a field where complete uncertainty quantification remains an ongoing challenge.
Frequently Asked Questions
Q: Is ISO 29473 applicable to all fire test methods?
A: It applies to quantitative tests producing results in engineering units (kW, MJ/m², g/s). It does not apply to tests producing pass/fail or index-based results, where different uncertainty concepts apply.
A: It applies to quantitative tests producing results in engineering units (kW, MJ/m², g/s). It does not apply to tests producing pass/fail or index-based results, where different uncertainty concepts apply.
Q: How is the cone calorimeter used as an example in the standard?
A: The standard derives a complete uncertainty budget for heat release rate measurement using oxygen consumption calorimetry, showing how uncertainties in oxygen concentration, flow rate, and ambient conditions propagate through the calculation formula.
A: The standard derives a complete uncertainty budget for heat release rate measurement using oxygen consumption calorimetry, showing how uncertainties in oxygen concentration, flow rate, and ambient conditions propagate through the calculation formula.
Q: What is a coverage factor and how is it chosen?
A: The coverage factor (k) is the multiplier applied to combined standard uncertainty to obtain expanded uncertainty. For approximately 95 % confidence, k = 2 is typically used. The exact value depends on the effective degrees of freedom of the uncertainty estimate.
A: The coverage factor (k) is the multiplier applied to combined standard uncertainty to obtain expanded uncertainty. For approximately 95 % confidence, k = 2 is typically used. The exact value depends on the effective degrees of freedom of the uncertainty estimate.
Q: How can I implement ISO 29473 in my laboratory?
A> Start by identifying all input quantities that affect your fire test results, classify them as Type A or Type B, quantify each uncertainty component, combine them using root-sum-square, and apply an appropriate coverage factor. The standard provides practical examples to guide this process.
A> Start by identifying all input quantities that affect your fire test results, classify them as Type A or Type B, quantify each uncertainty component, combine them using root-sum-square, and apply an appropriate coverage factor. The standard provides practical examples to guide this process.
