Sensitivity analysis for coal combustion
Overview
Most energy conversion systems operate by using liquid sprays or pulverized fuel in the combustion processes, which requires an improved understanding of the interaction process between droplet/particles essential in predicting the ignition, the combustion behaviour, and even the formation of pollutants. Simulating such large complex systems is expensive as it involves many models, e.g., chemical kinetic models and transportation models. These models have many parameters that are measured by experiments or obtained from simulations with certain uncertainties. Thus, it becomes necessary to quantify the uncertainties and the sensitivities of the models’ parameters. Computing models’ sensitivities helps to reduce the models’ complexity and determines the reliability region of the models. Based on that, the main goal of this research is to identify the sensitivities and the uncertainties of typical parameters used in combustion models, e.g., the burning rate of a particle with respect to its surface temperature.
 |  |
|---|
Left: Schematic picture of the single particle combustion model. Right: Sensitivity map of burning rate with respect to Damköhler number Dag and surface temperature TS.
Methodology
The methods and tools are developed on small scale models, the mass, energy and momentum balance equations are simulated for single char particle (primal problem) in steady and unsteady state. Three different methods to extract sensitivities are implemented, e.g., finite difference method, which is also known as a brute force method. The second method is the forward method, which is based on solving the linearized version of the primal problem with respect to certain parameter (e.g. thermal conductivity, particle radius, …etc.). Since combustion models have many parameters the third method, the adjoint method, is the most efficient method as it computes the sensitivities for all of the model parameter with a single simulation.
In our simple model the burning rate of the coal particle is the highlighted quantity of interest. To identify the most sensitive parameter to the burning rate a sensitivity analysis for a single operating point is carried out. This analysis shows that the surface temperature is the most pronounced parameter. Experiments and numerical simulations have shown that various starting conditions can influence the temperature at the surface of the particle in the steady state limit and hence alter the local conditions of the system (e.g. surface temperature, species composition in the far field, …etc.), for which these sensitivities are extracted. Therefore, in order to provide a more comprehensive picture, these sensitivities should be extracted for all possible operating conditions rather than their local counterparts leading to the sensitivity map of the system.
Person of Contact
Ahmed Hassan