A systematic approach to assessing measurement uncertainty for CO2 emissions from coal-fired power plants: Missing contributions from the Theory of Sampling (TOS)

An augmented measurement uncertainty approach for CO2 emissions from coal-fired power plants with a focus on the often forgotten contributions from sampling errors occurring over the entire fuel-to-emission pathway is presented. Current methods for CO2 emission determination are evaluated in detail, from which a general matrix scheme is developed that includes all factors and stages needed for total CO2 determination, which is applied to the monitoring plan of a representative medium-sized coal-fired power plant. In particular sampling involved significant potential errors, as identified and assessed by the Theory of Sampling (TOS), which also shows how these can be eliminated and/or minimised. Since coal-related CO2 emission calculations not only require analytical results of the carbon content of coal itself but also of the by-products fly ash and bottom ash, sampling procedures of these three materials were also given full attention. A systematic error (bias) is present in the current sampling approach, which increases the present uncertainty estimate unnecessarily. For both primary sampling and analytical sample extraction steps, random variations, which hitherto only have been considered to a minor extent, have now also been fully quantified and included in the overall uncertainty. Elimination of all identified sampling errors lead to modified CO2 determination procedures, which indicate that the actual CO2 emission is approximately 20,000 t higher than the present estimate. Based on extensive empirical sampling experiments, a fully comprehensive uncertainty estimate procedure has been devised. Even though uncertainties increased (indeed one particular factor is substantially higher, the so-called “emission factor”), the revised CO2 emission budget for the case plant complies with the official pre-determined uncertainty levels maxima in the EU guidelines.