OCR Text |
Show 3 introduction of fuel and combustion air such that there is a relatively large fuel rich zone in the furnace. In this zone the fuel nitrogen is released as nitrogen containing compounds; however, because there is insufficient oxygen available, the vast majority of the nitrogen is further converted to molecular nitrogen (N2 ). The mechanism by which this happens is not well understood; however, it is well accepted that hydrogen cyanide (HCN) and ammonia (NH3) playa major role [4]. These compounds result from the reaction of fuel bound nitrogen and the hydrocarbons released during devolatilization and, depending on the prevailing conditions, they can increase or decrease the amount of NOx being formed. Low NOx burner technology attempts to take advantage of the current understanding of the formation mechanisms to create the conditions which will minimize NOx production. Computational Fluid Dynamic (CFD) modelling of the combusting flow fields has been used extensively to assist in the understanding of the important parameters in NOx formation and in the application of certain technologies to full-scale combustion geometries. In order to do this, many models have been developed which predict the formation of NOx in flames and are used in conventional CFD computer codes. Because of the complexity of the processes by which NOx is formed, predictions . are routinely compared to measurements for validation purposes. In order to reduce the measurement costs, the measurement/prediction comparisons are usually made using a pilot-scale furnace. The confidence gained in the pilot-scale comparisons can then be used when making predictions at full-scale. The important issue of pilot-scale measurements is addressed in this paper. Specifically, the work undertaken in this study is directed towards the development and validation of NOx prediction models. To this end complete measurements of the temperature, composition and heat transfer fields were made in the combustor under study. Whereas it is understood that correct predictions of |