Type B Uncertainty

Type B uncertainty evaluation uses any relevant information other than repeated measurement data to quantify an uncertainty component. Sources include manufacturer accuracy specifications, calibration certificate data for reference standards, handbook values for material properties (thermal expansion coefficients), resolution of digital displays, environmental measurement data, and engineering judgment based on experience.

Type B evaluations require assuming a probability distribution for each uncertainty source. Common distributions include rectangular (uniform) for situations where only upper and lower bounds are known (e.g., instrument specifications, resolution), normal (Gaussian) for calibration certificate uncertainties, triangular for quantities known to have a most likely value near the center of the bounds, and U-shaped for quantities equally likely to be at either extreme. The standard uncertainty is then derived from the distribution parameters.

For calibration uncertainty budgets, Type B components often dominate the combined uncertainty. They include the reference standard uncertainty (from its calibration certificate), resolution of the unit under test, thermal expansion effects, and environmental corrections. Properly evaluating Type B uncertainties requires understanding the source information and selecting appropriate probability distributions. While Type B evaluation involves judgment, it should be based on documented, defensible reasoning, not arbitrary guesses. The GUM and ISO 17025 treat Type A and Type B contributions equally in the combination process.