Experimental analysis and multidimensional modeling of water condensation due to low-pressure exhaust gas recirculation activation during engine cold starts

Resum

The increasing use of the exhaust gas recirculation during the last years due to the impact reduction on the NOx emissions and the high demand on solutions to fulfill the homologation restrictions has revealed the need of studying deeply the system to operate continuously and independently of the boundary conditions. As a consequence of this and knowing that the main drawback of employing the low pressure exhaust gas recirculation is the condensation generation that causes erosion on the compressor impeller, research is required to understand and to predict the condensation during the exhaust gas recirculation activation. This work has been focused on generating and improving the modelling condensation prediction; and also, the development of experimental techniques that validates the proposed models for calculating the condensation water at cold warm-up homologation cycles. Concerning the experimental work presented on the thesis, the eorts has been focused on the condensation measurements in two dierent locations. On one hand, on the low pressure exhaust gas recirculation cooler, changing inlet conditions while simulating warm-up conditions on an engine. On the other hand, the condensation has been measured at the three-way junction outlet where the exhaust gas recirculation is mixed with fresh air coming from the ambient. Also, novel methodologies has been developed for measuring with optical techniques and with indirect measurements the mixing condensation. In terms of the condensation modeling, this thesis has been based in two main topics. The first topic regarding modeling was a 0D condensation model for calculating the condensation on heat exchangers and in particular, on low pressure exhaust gas recirculation coolers. Following this topic, it has been a quantitative validation of a 3D-CFD condensation submodel by means of experimental measurements. The second topic has been centered on developing new statistical methodologies for reducing the condensation calculation complexity on multi-dimensional simulations and reducing multi-fidelity parameters with the reduction of the computational cost. These methodologies have allowed to calculate the condensation on homologation cycles and to perform sensitivity analysis with different conditions.