Operating Regimes with Rapidly Pulsed Reductants in Diesel NOx Reduction with Lean NOx Traps: Experiment and Modeling
Amin Reihani, University of Michigan
Operating Regimes with Rapidly Pulsed Reductants in Diesel NOx Reduction with Lean NOx Traps: Experiment and Modeling
Authors: Amin Reihani1, Brent Patterson2, Galen B. Fisher2*, John W. Hoard1
1Mechanical Engineering Dept., 2Chemical Engineering Dept., Univ. of Michigan, Ann Arbor, MI 48109
Lean NOx Traps (LNTs) are often used to reduce NOx on smaller diesel passenger cars where urea-based Selective Catalytic Reduction (SCR) systems may be difficult to package. However, the performance of LNTs at temperatures above 400°C needs to be improved. Rapidly Pulsed Reductants (RPR, also known as Di-Air [1]) is a process in which hydrocarbons (HC) are injected in rapid pulses ahead of the LNT in order to improve its performance at higher temperatures and space velocities. A further goal of RPR is to reduce the fuel penalty associated with the fuel injections to regenerate the LNT.
Figure 1. Improvement of NOx conversion with ethylene reductant across a large temperature range by increasing the pulsing frequency (i.e, by reducing the shown total cycle period, P). The inset box defines the rich/lean total cycle period (P) and the duty cycle (DC % = %rich/P). DC is 15% for these experiments.
The use of rapidly pulsed reductants on a research Pt/Rh LNT formulation 90 gpcf of precious metal (Pt/Pd/Rh = 8/0/1) , and no ceria was studied experimentally using a synthetic flow reactor and a high frequency injection system [2]. The effects of the pulsing frequency, amplitude, duty cycle, temperature, space velocity, and NOx concentration on the NOx conversion and the reaction products were investigated. Fig. 2 (a) and (b) demonstrates NOx/ethylene conversion and product distribution as a function of pulsing frequency at constant fuel penalty and 600°C inlet temperature. We find that at any given temperature, fuel penalty, and space velocity there exists an optimal pulsing frequency which maximizes the NOx conversion.
For the temperature range of 450-600°C, the overall frequency response of the RPR process can be divided into two regimes: low frequencies (i.e., frequencies lower than the optimal frequency) and high frequencies (i.e., frequencies higher than the optimal frequency. The existence of an optimal frequency can be explained as a trade-off between NOx storage efficiency and NOx reduction effectiveness in the low and high frequency regimes as follows; (1) By decreasing the frequency relative to the optimal frequency, the NOx reduction and purging of the LNT becomes very efficient; however, as the lean periods become longer this results in lower NOx storage efficiency and lower overall cycle-averaged NOx conversion. (2) By increasing the frequency relative to the optimal frequency, the storage becomes more efficient due to shorter lean periods, but the effectiveness of the rich pulses for converting the stored NOx and purging the LNT decreases. This is mainly due to axial mixing of the rich pulse which results in less capability for the pulse to produce a sufficiently rich mixture to effectively react with the stored NOx.
Figure 2. RPR frequency sweep results at 600°C inlet temperature, 15% duty cycle, λpulse = 0.83, and SV=40,000 h-1. (a) NOx/ethylene conversion and catalyst temperature (b) product distributions
Previous work has suggested a mechanism based on the formation of stable isocyanate (-NCO) intermediates at high temperatures which promotes the NOx conversion efficiency [3]. However, here, we are determining whether a global kinetic model based on the conventional (low frequency) storage and reduction mechanism would be able to predict the NOx conversion improvements at high frequencies. The model was tuned to fit the low frequency response of a research Pt–Rh/Ba/Al2O3 LNT monolith. For this purpose, a series of long duration lean/rich/lean switches were used. The data was taken from a differential reactor (at space velocity of 150,000 h-1) by varying feed flow temperature and concentrations of NO, NO2, CO2, O2, and propylene (during reductant injection) at three levels.
Figure 3. Cycle-averaged NOx conversion from bench reactor and the model. The kinetic model captures the low frequency regime while the pulse mixing should be included to capture the high frequency regime.
As indicated in Fig. 3, which illustrates cycle-averaged NOx conversion in the pulsing frequency domain, the global kinetic model is able to predict improvement in cycle-averaged NOx conversion as the pulsing frequency is increased, due to shorter NOx storage periods and the improvement in its storage efficiency. However, kinetics alone does not indicate any reduction in NOx conversion as the frequency is further increased. The mixing and attenuation of reductants pulses, which is considered to be the main reason for the decrease in NOx conversion at higher frequencies, must be included in the model to predict the reduction in NOx conversion at high frequencies and predict the optimal pulsing frequency.
References:
[1] Bisaiji, Yuki, et al. “Development of Di-Air – A New Diesel deNOx System by Adsorbed Intermediate Reductants.” SAE International Journal of Fuels and Lubricants 5.2011-01-2089 (2011): 380-388.
[2] Reihani, Amin, et al. “Rapidly Pulsed Reductants in Diesel NOx Reduction by Lean NOx Traps: Effects of Mixing Uniformity and Reductant Type.” SAE International Journal of Engines 9.2016-01-0956 (2016): 1630-1641.
[3] Inoue, Mikio, et al. “deNOx Performance and Reaction Mechanism of the Di-Air System.” Topics in Catalysis 56.1-8 (2013): 3-6.