Urea Decomposition for Urea-SCR Applications

Doctoral thesis, 2010

Volatile emissions produced by the combustion of fossil fuels may be converted on a catalyst to less harmful products. During lean combustion an excess of oxygen in the exhaust gases together with a catalyst effectively oxidizes carbon monoxide and hydrocarbons into carbon dioxide and water. This leaves a deficit of reductants to reduce nitrogen oxides. In order to manage the conversion of nitrogen oxides into nitrogen a commonly used technology is selective catalytic reduction (SCR). In the SCR system an extra reductant such as ammonia or hydrocarbons are added to complete the conversion of nitrogen oxides. During the last few years SCR-systems for mobile applications using urea as an onboard source of ammonia have been developed and commercialized. Commonly, a solution of urea and water (commercially known as AdBlue) is injected trough an atomizer upstream of the catalyst. However, several problems still exist with the urea-SCR system. During conversion of the urea into ammonia and isocyanic acid stable high molecular compounds that can accumulate within the exhaust gas system may form. The abundance of these by-products depend on the conditions under which the urea is injected. In order to optimize spraying conditions new models and more fundamental knowledge of the urea decomposition processes are needed. In this thesis flow reactor experiments using Differential Scanning Calorimetery (DSC), Fourier Transformed Infrared spectroscopy (FT-IR) and Mass Spectrometry (MS) have been undertaken to study the pyrolysis of urea. Results show that the conditions under which urea is pyrolyzed governs which by-products are formed. Computational Fluid Dynamics (CFD) has been used to simulate the injection and decomposition of AdBlue for a urea-SCR application. New models for the decomposition and evaporation of urea have been developed and exhibit remarkable agreement with data found in the literature. The results from the CFD simulations were found to be sensitive to which sub models that were used.