Ultra-Low NOx Emission Prediction for Heavy Duty Diesel Applications Using a Map-Based Approach

As vehicle emissions regulations become increasingly stringent, there is a growing need to accurately model aftertreatment systems to aid in the development of ultra-low NOx vehicles. Common solutions to this problem include the development of complex chemical models or expansive neural networks. This paper aims to present the development process of a simpler Selective Catalytic Reduction (SCR) conversion efficiency Simulink model for the purposes of modeling tail pipe NOx emission levels based on various inputs, temperature shifts and SCR locations, arrangements and/or sizes in the system. The main objective is to utilize this model to predict tail pipe NOx emissions of the EPA Federal Test Procedures for heavy-duty vehicles. The model presented within is focused exclusively on heavy-duty application compression ignition engines and their corresponding aftertreatment setups. The accuracy of the model depends heavily on the ability to gain precise and repeatable test cell data to calculate an expansive SCR conversion efficiency map for the given aftertreatment system. This conversion efficiency map is verifiable based on expected/known chemical and physical properties of SCR aftertreatment systems. For this application, a 2-dimensional map was created, using SCR temperature and space velocity. The model requires several inputs including engine out NOx concentrations, SCR temperature, and exhaust flow/space velocity to input into the table and thus predict the corresponding tail pipe NOx. While different engine calibrations can impact the accuracy of the model, it was found that error in average tail pipe NOx prediction (g/bhp·hr) was approximately +/- 10%. For the purposes of this model, this error was found to be sufficient in providing the proper direction for ultra-low NOx aftertreatment development.