The project was primarily concerned with the development of acceptable prediction procedures for the performance of coal fired utility boilers. These procedures were to be applicable to existing boilers, to examine the implications of changes to operating procedures, or of retrofitted modifications. As the firing of coal blends, where the coals are sourced internationally, is the norm in many locations in Europe, it would be necessary to be able to predict the effects of modifications to the coal mix. Particularly the effect of a more extensive use of lower grade and/or cheaper coals would become more and more important. Finally the retrofit application of environmental protection technologies such as coal over coal reburn is foreseen. Thus the modelling should be able to take account of this technology. Finally the prediction procedures were also to be of use in the optimisation of design concepts for the boilers of the new millennium, with the aim to produce viable, flexible designs, capable of accepting multiple fuels sources, including biomass and waste fuels in an environmentally friendly manner. It was clearly recognised that the use of more and more powerful digital computers to run more extensive numerical models, in no way precluded the need for extensive experimental research both to provide a basis for improved phenomenological understanding, and extensive data bases for model validation
The COCA code is capable of predicting pulverised coal combustion in a 3-dimensional utility boiler geometry. Additionally, there are also other forms of the code that can be applied on 2-dimensional axisymmetric furnaces. The algorithm is based on a Eulerian description for the gas phase, for which the continuity, momentum, the species concentration (oxygen, carbon dioxide, water, volatiles) and radiation transport equations are solved by means of the SIMPLE algorithm. The coal particles and their motion are described using the Lagrangian approach. During the motion of the particles there is interaction with the ambient gas, which produces momentum and energy source terms, properly accommodated in the equations of the continuous phase. The gas phase equations are discretised by applying Cartesian computational grids to the furnace geometry, in conjunction with the porosity methodology near inclined surfaces. The local grid refinement method, which has been lately implemented, saves an important amount of memory and computational time, by dealing with the most important areas of the flow inside the furnace. The gas phase equations are discretised by integration over the cell control volumes. The code uses several higher order accurate difference schemes for the calculation of the transported variables on the cell faces. Schemes such as the BSOU (bonded second order upwind) and VONOS (variable order non oscillatory scheme) are used for the computations. The advantage of higher order discretisation schemes is that they provide accurate solutions in standard computational cells and CPU time. Additionally, the nitrogen oxide emissions are predicted by means of a post-processor. The nitrogen oxide post-processor takes into account the turbulent fluctuations of the temperature and oxygen field, but it also has the possibility of a deterministic nitrogen oxide emissions calculation. The effect of temperature and oxygen turbulent fluctuations is taken into account by a statistical approach, assuming two statistically independent variables with an associated density-weighted probability density function (pdf).