Various models aiming to predict the amount of unburned particles (solids) during heavy fuel-oil combustion have been developed. The parameters taken into consideration are generally asphaltenes precipitated by normal heptane or pentane and Conradson carbon as well as the metals content having a known catalytic effect on cenosphere combustion in the combustion chamber. The Exxon and Shell models can be mentioned, which were developed respectively in 1979 and 1981 (Chapter II). Other models also give consideration to the fuel-oil composition, the way it is atomized and diffused in the chamber and the combustion kinetics (research done by the MIT Energy Laboratory published in 1986). However, the above parameters are not the only ones involved. For some fuel oils, experience has shown that the state of dispersion of asphaltenes may also play an important role particularly for combustion installations with mechanical injection for which the dispersion of fuel-oil droplets is not very great and does not affect the structures built up by asphaltene aggregates. This influence of asphaltene dispersion on combustion was revealed in the past by the use of dispersant additives, and more recently in the combustion of heavy fuel oils made up by the dilution of pentane-precipited asphalts with a light cycle oil (LCO). These fuel oils are considered in this article because a divergence has been found between prediction and the measurement of solid unburned hydrocarbons as the result of a more or less dispersed state of asphaltenes, depending on the conditions of diluted-asphalt preparation with a fixed fuel oil/LCO ratio. The goal has thus been to add on a term representative of this state of dispersion to the terms normally considered (asphaltenes, Conradson carbon, metals). To assess the state of dispersion of asphaltenes in fuel oils, pictures implying a special preparation of sample (taken by Total) were examined. These photos give a fairly representative picture of aggregate distribution in the fuel oil. To assess this dispersion, a fractal approach, which had already been applied successfully to describe structures with comparable aspects, was tried, but we came up against difficulties stemming from the exploration method and from the unmatching of asphaltenic and fractal structures. We finally chose a visual determination based on the photos in which the asphaltene agglomerates are clearly represented as they occur in the fuel oil (set of photos in the article). This laborious exploration method (liable to be replaced by image-scanning software) nevertheless enabled a more complete model to be designed for this type of production. This model was based on a serie of 25 fuel oils, ten of which were burned in a 2 MW boiler and 15 in a 100 kW furnace. The characteristics of these fuel oils are given in Table I. The designations tau(s) and tau(v) represent the rates of surface and volume dispersion of the fuel oils expressed respectively in agg/mu-m2 and agg/mu-m3 (agg = agglomerates). Table II has to do with the nature of the fuel oils used (origin of the crude and method of preparation). The prospection methodology in creating the overall model is diagramed in a flowchart. The two series of fuel oils were burned respectively in a 2 MW boiler and a 100 kW furnace. The aim was to reduce them to a single serie of 25 measurement so that the model would have greater significance. The particles emission between the two installations were appreciably different, and so they had to be made comparable by a "passage equation" between the actual values obtained for the 2 MW boiler and the fictive values predicted by the intermediate model used (model of the 100 kW furnace). With the help of the passage equation, we were thus able to build a model on all the data. The model concerning the 100 kW furnace was taken as the intermediate model, and thus the values predicted by the overall model are directly applicable to the 100 kW furnace. However, the reverse "passage equation" had to be applied to the fictive predicted values for the 2 MW boiler to obtain the particle emission corresponding to this latter installation.
