VGB Statistical guidelines for emissions compliance evaluation

Final report of the vgbe Research project 409

For a range of pollutant emissions to air, the Large Combustion Plant Best Available Techniques (BAT) Reference document, the LCP BREF, defines Associated Emission Levels (BAT-AELs), for daily and annual averaging periods, as concentration ranges. The Competent Authority in each Member State sets Emission Limit Values (ELVs) from these BAT-AEL ranges. The Industrial Emissions Directive (IED) specifies that the 95% Confidence Interval of a single measurement (hourly average) shall not exceed a defined percentage of the Daily ELV. Towards the lower end of the BAT-AEL concentration range, the measurement uncertainty could be higher than required when monitoring pollutants continuously using Automated Measuring Systems (AMS).

The IED requires validation of daily averages by subtraction of the specified Confidence Interval from the hourly average concentration, to allow for measurement uncertainty, although the detailed methodology varies between Member States. Compliance is then typically assessed based on these validated averages for Hourly, Daily and Monthly ELVs, noting that there may only be a few hours of operation within a day or a few days of operation within a month. However, for annual BAT-AELs it needs to be confirmed that it is acceptable to validate an annual average emission concentration using the IED Confidence Interval, as suggested by the JRC Reference Report on Monitoring.

If the Confidence Interval is subtracted for some averaging periods, and not others, it is more difficult for Operators to manage emissions, on a day-to-day basis, against a background of increasingly intermittent operation and reduced annual generation. This was not envisaged by Operators when the LCP BREF was developed and could affect both operations and investment decisions.

There are two main reasons for applying Confidence Intervals when reporting annual average emission concentrations. The first is the question of legal certainty and this is crucially important for compliance assessment as it needs to be clear to both the Competent Authority and the Operator when permit conditions are breached. The second is to account for the systematic uncertainty that is applicable to annual average concentrations due to AMS calibration.

These issues are explored by considering the uncertainty contributions that arise from the AMS Quality Assurance Levels which are defined in EN 14181: QAL1 (certification), QAL2 (calibration) and QAL3 (control) – along with an Annual Surveillance Test (AST calibration check).

Provided that a suitably QAL1 certified AMS is available, the measurement uncertainty is dominated by systematic uncertainty associated with the QAL2 calibration in which a Standard Reference Method (SRM) is employed by an accredited test laboratory to calibrate the AMS during a three day test campaign. The underlying uncertainty of the SRM is transferred to the AMS for a five year period within which the AMS calibration is allowed to shift within a defined tolerance. Residual uncertainty associated with single-point SRM sampling within the stack also needs to be taken into account, where applicable. The uncertainty associated with correction to the emissions reporting conditions also needs to be considered and this is especially important for gas turbines. It is assumed that the AMS is not subject to QAL3 drift under normal operation.

The example uncertainty analysis, employing relatively conservative assumptions relating to systematic uncertainty contributions, demonstrates that it is entirely appropriate to apply the IED Confidence Interval to all reported emissions, including annual emissions, and this provides both the necessary legal certainty for compliance assessment and simplicity with regards to plant operation. When it is not possible to achieve the required uncertainty at low concentrations then the implementation of a fixed absolute uncertainty is recommended, for each pollutant, as already mandated in The Netherlands.

More specifically, it is recommended that the Annual ELV is used instead of the Daily ELV for AMS assessment since this is more representative of the typical emission concentration. It is also recommended that SRM single point sampling should always be from the same point within the stack, during QAL2 and AST test campaigns, to minimise the influence of positional uncertainty.

However, it is also clear that a number of challenges remain in relation to minimising and assessing measurement uncertainty: i) re-validation of SRM standards at lower concentrations and the development of new test methods with lower limits of detection; ii) improved SRM uncertainty assessment methodologies; iii) certification of new AMS with improved sensitivity and discrimination between pollutants; iv) improved AMS calibration procedures at low concentration, especially for reactive trace gases and Dust (requiring improvements within EN 14181).