Qatargas Operating Company Limited (Qatargas) and Total Research Center Qatar (TRC-Q) established a collaboration to study Predictive Emissions Monitoring System (PEMS) algorithms for pilot application on a Qatargas turbine to predict NOx emissions. The approach for this Pilot Study was a blind benchmarking comparison of three main PEMS algorithms (first principle, statistical and neural networks) to define the most appropriate one(s) for application at the pilot gas turbine. This study was intended to demonstrate to local authorities that PEMS can be a reliable monitoring technique in both an alternative or complimentary capacity to CEMS. The assessment of PEMS models developed as part of this study and their corresponding performance will be discussed in a separate paper when the study is completed. This initial paper describes the challenges and lessons learned during the preparatory phase of PEMS model development. It describes the importance of a well-planned preparatory stage as this significantly affects the quality and validity of collected turbine operational and emissions data for PEMS model construction. It is important to undertake internal quality assurance checks on collected operational and monitored emissions data prior to model development. This paper describes the important role played by maintenance and calibration of measuring instruments such as stack emissions analyzers in ensuring reliability and accuracy of measured data. To build robust PEMS models, individual correlations between NOx emissions and various turbine operational parameters need to be assessed in the preparatory stage. The PEMS models developed for the pilot turbine were initiated on an incremental basis using variables with significant correlation and then optimized using other secondary parameters to improve correlation between the predicted and measured NOx emissions. This paper notes that the influence of turbine operational parameter on NOx emissions varies depending on its role in the formation of NO_x as part of the combustion process. These include fuel gas composition and flow rate, as well as ambient air temperature and humidity.
展开▼
