Overview
ISO 3086:2006 - "Iron ores - Experimental methods for checking the bias of sampling" defines statistical and experimental procedures to determine whether a sampling method for iron ores is biased compared with a reference stopped-belt sampling method (per ISO 3082). The standard explains how to design paired-sample experiments, compute differences, test for outliers and build confidence intervals to decide if a candidate routine sampling procedure has an acceptably small bias.
Key topics and requirements
- Reference method: A stopped-belt sampling method in accordance with ISO 3082:2000 is used as the unbiased reference (method A).
- Test method: The method under test (method B) - e.g., moving-belt samplers, ship/wagon transfer sampling - is compared against method A on the same material.
- Minimum data: At least 10 paired sets of measurements are required. Additional tests may be needed depending on outliers and statistical results.
- Pre-test inspection: A mechanical inspection of the sampling system is recommended before conducting bias tests.
- Data analysis steps:
- Compute paired differences (d = xA − xB), mean (d̄) and standard deviation (Sd).
- Apply Grubbs’ test to detect outliers; follow rules for reinstating/excluding outliers based on assignable causes and the proportion of remaining data (minimum retained percentage considered).
- Determine a confidence interval for the mean difference using Student’s t (the standard recommends assessing a 90% confidence interval for the true average bias).
- Decision rule: Compare the confidence interval for the average bias to a pre‑defined relevant bias (δ). If the CI lies entirely within ±δ, method B is acceptable; if CI excludes zero, method B is biased and must be adjusted; if CI includes zero but extends beyond ±δ, further sampling is required.
Practical applications
- Establishing QA/QC for mine sampling programs and bulk material trading where accurate iron content, moisture, size distribution or other quality characteristics are critical.
- Validating new mechanical samplers, transfer-point sampling procedures, or modified sampling systems to ensure no systematic bias versus stopped-belt reference.
- Investigating differences between sampling locations (e.g., loading vs unloading) or testing sample-preparation and sieving bias when used with the corresponding ISO methods.
Who uses this standard
- Mining sampling engineers and metallurgists
- Laboratory managers and analysts responsible for sample preparation and assay QA/QC
- Quality assurance auditors, commodity traders and material buyers/sellers assessing contractual sampling risk
Related standards
- ISO 3082:2000 - Sampling and sample preparation procedures (reference sampling methods)
- ISO 3085:2002 - Experimental methods for checking precision of sampling and measurement
- ISO 4701 - Hand sieving methods (relevant where size distribution bias is tested)
Keywords: ISO 3086:2006, iron ores sampling bias, stopped-belt sampling, Grubbs' test, confidence interval, statistical test for bias, ISO 3082, ISO 3085.
