Importance of nitrates in Cu-SCR modelling: A validation study using different driving cycles

Highlights

• Calibrated Cu-CHA model with and without nitrates to the same experimental data.

• Applied models to predict various driving cycles (WHTC, FTP, NRTC, ETC, WNTE).

• Inclusion of nitrates allows for improved prediction of colder cycles (i.e., cWHTC).

• Prediction of hotter cycles (i.e., NRTC) the same for both models since no nitrates stored.

• Model accurately predicts increase in NOx conversion with dosing strategy.

Abstract

Both a mechanistic model, which accounts for ammonia, nitrate, and ammonium nitrate storage, and a global model, which only accounts for ammonia storage and no nitrates, were calibrated to the same data collected across a Cu-CHA catalyst. Once calibrated, the models were directly applied to simulate various driving cycles (cold WHTC, hot WHTC, FTP, NRTC, ETC, and WNTE), with different catalyst layouts (i.e., washcoat loading and cpsi) as part of the model validation process.

It was demonstrated that the mechanistic model provides a notable improvement in the prediction of the NO_x conversions for colder cycles (i.e., cold WHTC, hot WHTC, and FTP) due to its inclusion of nitrates and ammonium nitrate. The quality of prediction of the hotter cycles (i.e., NRTC, ETC, and WNTE) was close to the same for both models, since no ammonium nitrate accumulates at these higher temperatures. Despite the additional reactions and surface species included in the mechanistic model, no significant increase in simulation time is observed compared to the global model.

Finally, the mechanistic model is applied to calibrate a simple NH₃ dosing strategy. The dosing strategy is compared to a constant NH₃/NO_x approach, where it was shown through simulations, and confirmed experimentally, that the dosing approach allows for an increase in NO_x conversion. Overall, this demonstrated that the model accurately predicts the increase in NO_x conversion and that the model is a valuable tool to establish the catalyst’s true potential during driving cycles.

Introduction

The advancement of more efficient engines, which results in increasingly lower exhaust gas temperatures, as well as stricter government regulations, has challenged the automotive industry to further improve their aftertreatment systems. To meet the NO_x emissions standards from diesel vehicles, research is continuously being conducted to modify the SCR catalyst formulations to improve the activity of the main SCR reactions (Standard SCR (1), Fast SCR (2), and NO₂ SCR (3)) by understanding the location and role of the active sites and the SCR reactions’ rate limiting steps [1,2].

4NH₃ + 4NO + O₂ → 4N₂ + 6H₂O

2NH₃ + NO + NO₂ → 2N₂ + 3H₂O

4NH₃ + 3NO₂ → 3.5 N₂ + 6H₂O

In addition to improving catalyst formulations, the design of the aftertreatment system plays an important role in the minimization of NO_x emissions. Numerical simulation has proven to be extremely beneficial in reducing the time and cost of achieving the latter. Examples of such practical simulation studies include the influence of geometrical properties on deNOx performance [3,4], the optimal combination of an SCR catalysts (i.e., Fe- and Cu-zeolite) for zoned or layered architectures [5,6], and catalyst screenings with optimized ammonia dosing operating strategies during driving cycles [7].

The completion of meaningful simulations requires models that describe the catalyst behavior under a wide variety of conditions. A significant amount of global models have been published to account for the SCR activity across vanadium, Fe-, and Cu-zeolite catalysts [[8], [9], [10], [11],5,1]. Most of these models capture the ammonia storage dynamics, are then calibrated to describe the SCR steady state activity, and sometimes the model prediction of a driving cycle is shown.

In this work, we focus on the role of nitrate species in Cu-CHA catalyst modelling. Nitrates have been shown to be a central intermediate in the mechanisms of Fast and NO₂ SCR [12,13]. The reaction mechanisms involving copper nitrates have been established using transient test protocols [14] and kinetic models have been developed that reproduce the transient response data, hence providing further quantitative confirmation of the proposed reaction mechanisms [14,15]. In the presence of ammonia, ammonium nitrate precursors containing copper nitrate and NH₃ (the so-called ammonia blocking species) have been proposed to form [12,14]. The exact chemical nature of these species is not yet fully understood. One challenge with the Cu-CHA catalysts is that they show a slow transient deactivation owing to ammonium nitrate accumulation, which is particularly prominent when the catalyst is operated at low temperatures with high NO₂/NO_x inlet ratios [[16], [17], [18]]. The observed NO₂ inhibition on Cu-CHA catalysts contrasts with Fe-zeolite and vanadium catalysts, where the catalyst activity can be increased by dosing additional ammonium nitrate into the catalyst [[19], [20], [21]].

We have recently shown that under operation with fluctuating NO₂/NO_x ratio, the formation of ammonium nitrate on the Cu-CHA catalyst can have a beneficial effect on the NO_x conversion since ammonium nitrate can act as a buffer that cushions fluctuations in the NO₂/NO_x ratio and in this way allows a constant NO_x consumption close to the rate of Fast SCR [18].

Only recently, a few studies have been published that include nitrate dynamics in the kinetic models used for exhaust system design and the prediction of catalyst performance during realistic drive cycles [[22], [23], [24], [25]]. All three kinetic models were validated against a single type of drive cycle and all three models achieved close to quantitative prediction for this drive cycle. In [23], we compiled state of the art submechanisms for Standard, Fast and NO₂ SCR and a few more relevant reactions as well as NH₃ storage and the storage of the different nitrate species into a semi-mechanistic kinetic model. The model was calibrated to steady state and transient response lab bench data and provided a good prediction of catalyst performance during test cycles run on an engine without any additional recalibration. Daya et al. [24,25] developed a detailed global model that included ammonium nitrate formation as an additional global reaction. By taking into account two different Cu species (ZCu^IIOH and Z₂Cu^II) and assuming that the only effect of ageing is changing the concentrations of these two copper species [24], they can describe the performance of two differently aged catalysts with a single parameter set. In the long run this approach will allow them to provide a physical description of hydrothermal catalyst ageing within the kinetic model [26].

To our knowledge, to date, there has been no paper published demonstrating the improvement in predicting driving cycles when applying a Cu-SCR model that accounts for nitrates compared to a model that only accounts for ammonia storage.

The goal of this work is to demonstrate the necessity of including nitrates when simulating cold cycles over a Cu-SCR catalyst. To this end we calibrate a global model not considering nitrates using the same steady state- and ammonia storage data used previously to calibrate our mechanistic model [23]. The performance of the two models is then compared for a variety of different drive cycles and it is shown that the low temperature cycles are better predicted by the detailed model containing nitrate chemistry while the higher temperature cycles are well-predicted by both models. At the same time, our study shows that the old approach of validating a new model versus a single drive cycle is insufficient and that several different cycles should be applied in model validation to identify where the model needs to be improved.

Finally, to further demonstrate the accuracy of the detailed model and to show the value of the model for practical simulation studies, we apply the model to compute an optimized NH₃ dosing strategy [7], and validate the simulated results against an experiment run with the optimized dosing strategy.