| OCR Text |
Show A number of codes have been developed to predict the rate of volatile release and the composition of key species. T w o examples are FG-DVC12,13 and F L A S H C H A I N 1 4 . Both these models use functional group look-up tables to predict volatile yields and release rates. W e have explored the use of F G - D V C for predicting char, tar and gas yields and for a wide range of coal ranks. Figure 4 shows the predicted partitioning of fuel-N between char, tar and gases, and the char nitrogen results agree well with drop-tube data. The predicted gas yields and compositions are in good agreement with pyroprobe-GC data. 2.2. Volatile Combustion and NOx formation Turbulent non-premixed coal flames generated from volatile combustion often exhibit significant departures from chemical equilibrium through phenomena such as super-equilibrium radical concentrations, pollutant formation, extinction and sooting. One possible route to include these effects in turbulent flow calculations is the laminar flamelet approach17. Here, the local instantaneous structure of the reaction zone in the turbulent flow is assumed to be the same as that in a quasi steady one-dimensional laminar flame, despite convection and distortion by the turbulent velocity field. In the flamelet model, the principle influence of turbulence on the chemical structure of the laminar flame is by the amount of stretching that the flame is subjected to as a result of turbulent flow-field straining. In the flamelet model developed for the gas-phase volatile combustion of coal, the local temperature and major species concentrations are given by the mixture fraction. The profiles of the different thermo-chemical variables are highly dependent on the rate of strain and the location of the peak flame temperature almost occurs at the same position in mixture fraction space. The approach developed provides an alternative way of incorporating finite rate chemistry effects in turbulent coal volatile combustion modelling. The turbulent flame is assumed to consist of an ensemble of laminar flamelets which are stretched and distorted by the hydrodynamic flow-field. The detailed structure of the turbulent and laminar flamelets is assumed to be the same. Then state-relationships between mixture fraction and reactive scalars, obtainable from laminar flame calculations or alternatively from experimentation measurements, can be used for turbulent coal combustion modelling. Before performing the CFD modelling for the reactive flow-field, the data-base relating the thermo-chemical variables to the instantaneous mixture fraction (in look-up tables) must be prepared in the required format for the code. The code is modified to accept the variables listed to any size of the instantaneous mixture fraction array. This flexibility in mixture fraction array size is necessary in order for the code to be able to communicate with the data extracted from different pre-processors. The pre-processor used in the present work is the laminar opposed diffusion flame code, OPPDIF, from S A N D I A National Laboratories18. Nitrogen oxide formation during combustion of the volatiles was modelled using the OPPDIF code. The fuel input comprised the volatiles released when South Brandon coal was heated at a rate of 4 x 104 K s"1 to a final temperature of 1923 K. The oxidiser used was either normal air 6 |
