NOx reduction in grate-fired Waste-to-Energy plants

Due to the great environmental problems related to emission of NOx from combustion processes a large emphasis has been put on minimising the emission over the last decades. Furthermore, in order to accommodate the strict NOx emission levels, which is projected to be tightened even more in the coming years in many countries, new measures must be taken.

It is important to develop models to predict the NOx formation and degradation at varying furnace operating conditions to allow for an optimization of furnace design and operating conditions. Chemical models predicting the NOx formation and degradation in a grate-fired combustor have been developed.

However, these models do not take fluid dynamics into account and are typically too complex to be implementing in computational fluid dynamics (CFD) codes. CFD codes are required to take fluid dynamics into account, thereby predicting NOx formation and degradation more accurately. These modelling short-comings presently limit the industry from obtaining better process control and improved design of combustion facilities.

Initially in this thesis numerous chemical models that describes the NOx formation and reduction are identified. One of these models was selected for further work. The identified detailed chemical kinetic model was reduced using the Simulation Error Minimization Connectivity Method (SEM-CM) and the Path Flux Analysis (PFA)method to yielding a number of skeletal mechanisms with varying size.

It was identified that a skeletal model consisting of 38 species and 251 reactions, developed using the SEM-CM algorithm, offered the best compromise between model size and accuracy. Full scale Waste-to-Energy (W-t-E) plant measurements were conducted identifying the concentration profile of major combustion products and combustibles close to the fuel bed.

Furthermore, the speciation and concentration profile of the NOx precursors were measured close to the bed, and it was shown that accurate predictions of the NOx precursor speciation is essential for accurate NOx predictions. The formation of NOx at the W-t-E plant (Affald+) used for data collection was modelled post process the modelling of the combustion process.

This was done using B&W Vølunds in-house CFD model and the skeletal model developed in this study. The modelling was performed based on process conditions obtained from the plant control system and data collected during measurements at the plant. Results from the modelling showed large differences between the predicted and measured NOx emissions.

It is believed the difference is caused by errors in the predictions of the temperature field in the furnace. The process is very dependent on temperature. Consequently a precise prediction of the temperatures is a precondition for precise results with respect to NOx. A skeletal model for the SNCR process was developed through reduction of the detailed chemical kinetic model using the SEM-CM algorithm.

A skeletal model consisting of 21 species and 50 reaction was identified as the most suitable for CFD modelling as it offered a good compromise between accuracy and size.The efficiency of NOx reduction by injection of NH3 (SNCR) into the flue gas was determined through full scale measurements. Furthermore, the flue gas composition inthe SNCR zone was measured.

Besides normal combustions products (CO2 and H2O),high CO concentrations of up to 6000ppm was measured in the lower part of the SNCRzone.The SNCR process was modelled post process the combustion using the skeletal modeldeveloped in this study. The modelling was performed based on process conditions obtained from the plant control system.

Large differences between predictions by the CFD model and measurements were found. The large differences is believed to be due to inaccuracies in the prediction of the CO concentration field, which the SNCR process ishighly dependent on, and a predicted recirculation of flue gases in the SNCR zone.