Numerical Simulation of Deposit Formation in Coal-Fired Utility Boilers with Biomass Co-Combustion (English shop)

Preview

Table of Contents, Datei (47 KB)
Extract, Datei (130 KB)

The use of fossil fuels for energy generation leads to the emission of greenhouse gases. The resulting environmental damage can be reduced by replacing coal as an energy carrier in power plants with CO2-neutral biomass. However, the efficiency of energy conversion in the thermal utilisation of pure biomass is still insufficient compared with the use of coal. A further alternative is the co-firing of biomass in existing coal-fired power plants. Yet the co-firing of biomass tends to lead to increased formation of ash deposits on the surfaces of the heat exchangers. Especially in the region of the superheater heating surfaces, the ash deposits impair the heat transfer from the flue gas to the water/steam cycle and thereby limit the efficiency of the steam generator. In order to be able to reduce these negative consequences, a fundamental understanding of the formation of the ash deposits is required. Above all, the high proportion of volatile potassium released during the combustion of the biomass contributes to the formation of ash deposits. The released potassium can react with ash particles before or during the condensation process. This leads to the melting of ash particles and thus ultimately to deposition on the surfaces of the heat exchangers. The released potassium can also condense on the cooled tube surfaces and form a sticky layer there, which further accelerates the formation of additional deposits.

In recent years, the modelling of combustion processes in combination with numerical fluid mechanics (Computational Fluid Dynamics) has proven to be a very efficient and reliable tool. This tool is therefore widely applied in order to predict the flow field, temperature distribution, flue gas composition and particle motion in combustion chambers at a high spatial resolution. In addition, the ash deposition process can be investigated comprehensively within it through the integration of deposition models. This approach is particularly interesting because carrying out corresponding simulations enables both a saving of time and a reduction of costs compared with purely experimental investigations.

The main objective of the present work consists in the modelling of deposit formation in coal-fired steam generators during the co-firing of biomass. The focus was placed in particular on the release of alkaline constituents and their influence on the build-up of the deposit. The corresponding models were integrated into the 3D combustion simulation code AIOLOS. The models for deposit formation were implemented as a post-processing step. In the first step, a base simulation is carried out which describes both the release of the potassium compounds and the flow field as well as the temperature and concentration distribution. Subsequently, the trajectories of a large number of particles in the combustion chamber are tracked until they either touch the wall surfaces and adhere or leave the combustion chamber with the flue gas. The data generated in the base simulation – such as the concentration of the solids or the properties of the gas phase – are now used as input data for the actual model of the deposition mechanism.

Both heterogeneous and homogeneous reactions are taken into account for the description of the alkali release. The heterogeneous reactions are divided into a first step of alkali release and a subsequent reaction between the released alkali species and the aluminosilicates of the fly ash. A global model by Tomeczek is used to describe the sulphation of the released alkali species. The validation of the models employed was carried out on the basis of simulations of a small-scale entrained flow reactor. The model for sodium release was evaluated by means of simulations using a German hard coal as fuel. The simulation results agree well with the measurements – in particular at a residence time of less than 10 s. For the validation of the proposed model for potassium release, simulations were carried out with lignite with a low aluminosilicate content and with hard coal as fuel at short residence time. The calculated potassium release shows good agreement with the experimental data. However, it also becomes apparent that a reaction rate which fits well for a particular coal type cannot be transferred to other coal types. Different ash compositions can affect the behaviour of the ash constituents with regard to the control of alkali release.

As far as the formation of the deposits is concerned, both the influence of the alkali species on the sticking probability of the ash particles and the contribution of the released alkali species to the deposit layer are modelled. Two essential deposition mechanisms are taken into account. These are the inertial impaction of the pulverised coal particles and the condensation of the released alkali compounds. On this basis, simulations of biomass co-firing in coal-fired steam generators were carried out. The simulation results are discussed with a focus on the properties of the biomass employed and the share of the biomass in the total thermal output, and are compared with experimental data.

With regard to the base simulation, it turns out that the calculated temperature profile agrees well with the measured values. Certain deviations can be attributed to the uncertainties in determining the swirl number of the secondary air. Two condensation models were employed and assessed with respect to their applicability for predicting the deposition process in coal-fired steam generators during biomass co-firing. In general, the condensation rate calculated with a model based on film theory is five orders of magnitude lower than the condensation rate calculated with the model according to Goldbrunner. With Goldbrunner’s model, the deposit consists predominantly of condensable species. However, this does not agree with the real test cases considered. For this reason, the model based on film theory was used to describe the condensation of K2SO4. Furthermore, it could be shown that the film theory model is able to predict the condensation rate at the superheater heating surfaces in steam generators. Nevertheless, deviations between experiment and simulation were observed. These could be attributable to the influence of aerosol formation, which, however, was not taken into account in this work.