OCR Text |
Show principle spray direction. The mass, energy, and momentum sources of the inj ectors are included in the CFD model. Additional iterations are performed until the CFD and injector model results converged to a steady-state solution. The results of the injector model are a set of the most probable trajectories and concentration of chemical as a result of droplet evaporation. Droplets lose mass during their lifetimes. The trajectories end with the completion of droplet evaporation. For chemicals such as urea that has a low vapor pressure, vaporization and chemical release occur at this point. From the released chemical concentration, the extent of coverage of active chemicals is estimated. Trajectories also end when droplets impinge on walls. It is critical to avoid droplet impingements onto tube walls to prevent corrosion or erosion. Once the desired droplet size and velocity distributions have been established, injector type and injection parameters such as liquid and air flowrates and pressures are selected and verified through laboratory measurements. Combination of CFD models and CKM Although the CKM has been valuable in identifying key components of the chemical process, the model by itself is limiting in predicting expected performance on units. Spatial variations in temperature and velocity cannot be modelled with the one-dimensional CKM that treats the upper furnace region as a plug flow reactor. It is common to measure greater than 200°C variation in flue gas temperature at a given elevation. In addition, flue gas that flows, for example, over the bullnose of a unit will have much shorter residence time than the flue gas along the front wall. If the NOxOUT kinetic model were applied using average temperatures and residence times, the predicted process performance would be inaccurate. The prediction would worsen for those units with steeper temperature and velocity gradients and features such as secondary air ports. Ideally, inclusion of NOxOUT reaction kinetics into the CFD model would calculate NOx concentrations that account for flow and temperature variations. Such model is yet to be employed due to excessive computational demands. Instead, the upper furnace is divided into many ideal plug flows in parallel; each plug flow has its own temperature and residence time history. This division of the upper furnace region into individual flows is based on the flow streamlines generated by the CFD model. Because each reactor follows the actual gas path, this technique is applicable to complex swirling flows such as tangentially fired or corner fired boilers as well as non-swirling flows. This method, however, does not account for mixing among plug flows. The CFD model generates necessary input to the CKM. Once the CFD model of a furnace is fully converged, a particle, defined so to be fully entrained in the flow, traces the gas flow from various |