This content is not included in your SAE MOBILUS subscription, or you are not logged in.
CFD Modeling of Tailpipe NOx Sensor Accuracy
ISSN: 1946-3936, e-ISSN: 1946-3944
Published August 08, 2018 by SAE International in United States
Citation: Kalyankar, A., Munnannur, A., and Liu, Z., "CFD Modeling of Tailpipe NOx Sensor Accuracy," SAE Int. J. Engines 11(4):435-446, 2018, https://doi.org/10.4271/03-11-04-0029.
In a modern diesel aftertreatment system, a sensor for nitrogen oxides (NOx) placed downstream of the selective catalytic reduction (SCR) catalyst is necessary to determine if the tailpipe NOx concentration remains below the applicable On-board diagnostic (OBD) threshold. Typically the same NOx sensor also provides feedback to the dosing control module to adjust diesel exhaust fluid (DEF) dosing rate thereby controlling tailpipe NOx and ammonia emissions. However, feedback signal sent by the tailpipe NOx sensor may not always be accurate due to reasons including non-uniformity in NOx and ammonia distributions at SCR outlet. Flow based metrics from computational fluid dynamics (CFD) analyses, that are typically used to qualitatively assess NOx sensor accuracy in different designs are often inadequate. In this work, an improved CFD analysis procedure has been developed for assessing NOx sensor accuracy. This approach enables a direct comparison of NOx sensor accuracy between different sampling probe and sensor designs. This improved modeling approach was first validated against test data without spray effects by injecting gaseous NOx in a 5″ pipe. The impact of various numerical and geometrical sensitivities on model predictions was considered. Afterwards, studies were conducted to study the impact of non-uniformity of NOx and velocity distribution on sensor accuracy. Species distribution index (SDI) and hole and hole mass flow gamma (γ) were found to be useful metrics directly affecting NOx sensor accuracy. Lastly, the modeling approach was validated against test data on a representative aftertreatment system. Model predictions of sensor accuracy error were in agreement with test data (<5% error) at multiple operating points for two different sensor designs.