Choice of urea-spray models in CFD simulations of urea-SCR systems

https://doi.org/10.1016/j.cej.2008.12.003Get rights and content

Abstract

The sensitivity of modeling choices to obtained results for Eulerian–Lagrangian CFD simulations of urea-SCR systems has been investigated for a system consisting of an AdBlue-spray located at the exhaust pipe wall, directed into the exhaust gas flow. The decomposition of urea is modeled as being heat transfer limited and taking place at a constant temperature (425 K). It is shown that modeling choices may affect the predicted extent of wall hit, which types of droplets that are predicted to hit the wall, and also where they will do so.

The influence of the different forces due to drag, buoyancy, lift effects, thermophoresis and history effects was investigated, proving that only the forces due to drag and buoyancy are necessary to correctly describe droplet motion within this system. It is necessary to use a droplet drag coefficient that takes the current level of droplet distortion into account.

A stochastic particle tracking model will describe the effects of turbulent dispersion, but also make the simulation results sensitive to the quality of the turbulence model's prediction of the turbulent fluctuating velocities. Using such a model will also resolve some of the enhancement of heat and mass transfer caused by the continuous acceleration/deceleration of droplets by turbulent eddies.

Introduction

The increasing demand on the automotive industry to reduce emissions from diesel engines entails more knowledge about the modeling of exhaust gas after treatment systems. One of the major pollutants in diesel exhaust is nitric oxides (NOx). NOx has adverse effects on both human health and the environment.

A promising technique for reducing NOx emissions is selective catalytic reduction (SCR) with urea. With this approach, a water–urea solution is sprayed into the exhaust pipe in front of the catalyst. Urea will decompose into ammonia, which is the reducing agent needed to transform NOx to N2 on the catalyst.

Computational fluid dynamics (CFD) simulations will most probably play a crucial role in the design and optimization of future urea-SCR systems. The work presented here has investigated droplet–flow interactions in a urea-SCR-spray originating from nozzle located at the exhaust pipe wall, using commercial CFD software (Fluent 6.3.26). The main body of this work is devoted to determining the most suitable and least complex models for Eulerian–Lagrangian CFD simulations of urea-SCR systems, and to critically assess the actual impact of choosing a specific model on all studied results.

In order to identify the main characteristics of the urea-SCR system from a modeling perspective and – equally important – identify what that is too poorly known to be accurately modeled, a short description of the chemistry and physics of the urea-SCR system is necessary, and will follow below.

In the urea-SCR system, a solution of 32.5 wt% urea in water (AdBlue) is sprayed into the pipe ahead of the SCR catalyst. Water is evaporated and the resulting solid urea melts and starts to decompose thermally [1]:NH2–CO–NH2 (s)  NH3 (g) + HNCO (g), ΔH = 185.5 kJ/mol

This results in release of gaseous ammonia that can take part in the SCR reactions. The resulting isocyanic acid will also produce ammonia, through hydrolyzation on the SCR catalyst (or in the gas phase at high temperatures) [1]:HNCO (g) + H2O (g)  NH3 (g) + CO2 (g)

This reaction, being limited mainly by external and internal mass transfer [2], is much faster than the SCR reactions, which means that it is believed that every mole of urea will result in two moles of ammonia for the SCR reactions.

It has been proposed that, since urea ideally decomposes into ammonia, ammonia will be the active reducing agent [1], and thus the SCR reactions for the urea-SCR system will be the same as for an NH3-SCR system.

When the water has more or less fully evaporated from the AdBlue droplet, it is not yet completely known what the droplet looks like. The various suggestions in the literature include a particle of solid urea [1], [2], [3] or molten urea [2], [3], [4] or just a very concentrated water solution of urea [5]. Urea has a melting temperature of approximately 406 K [6]. There are many suggestions in the literature for the temperature when urea starts to decompose thermally, for example: 406 K [7], 416 K [8], 425 K [9] and 433 K [10], [11]. In fact, urea decomposition has been observed at low rates at temperatures as low as 353 K [12]. In previous studies, urea decomposition has been modeled (in order of decreasing complexity) either as evaporation with a saturation pressure curve determined from experimental data [13], with an Arrhenius expression [7], [14], as controlled by the turbulent mixing process [3], or simply with a conversion efficiency factor determined from experimental data [2]. The Arrhenius expression approach has resulted in urea decomposition modeled as taking place at temperatures as high as above 600 K [14]. For the modeling in this work, given the above-presented background, it was decided to model urea decomposition as a heat transfer limited process at a constant temperature of 425 K. This approach evades the need for empirical parameters that need to be trimmed to experimental data from a specific system.

The conversion efficiency of urea into ammonia changes with temperature. This is believed to be caused by the fact that urea decomposes in at least two stages [7]. The first stage, which is the one starting somewhere above 406 K, results in the formation of isocyanic acid and ammonia. Further heating will cause polymerization into biuret (from 433 K) and cyanuric acid and ammelide (from 448 K) [9]. The second stage starts at 523 K, where decomposition into ammonia and isocyanic acid stops. Urea may thereafter form cyanurates, ammeline, ammelide and melamine [9]. Ammonia can even be consumed during these transitions [15]. For these reasons, urea should ideally decompose at temperatures close to the start of the first stage of decomposition, in order to gain maximum urea conversion efficiency into ammonia. It is also of interest to notice that contact with the catalyst is reported to speed up the first stage, so that there will be almost no decomposition of urea at temperatures of 523 K and above [15].

Even if the thermal decomposition of urea starts already in the exhaust pipe just after injection, it will most probably not be complete by the time it reaches the catalyst entrance. Depending on the temperature of the exhaust gases, as little as 20% conversion of reaction (1) can be expected (at 330 °C), while isocyanic acid is stable enough in the gas phase to remain totally unreacted until reaching the catalyst surface [16]. In one study, no more than 65% of the injected urea had decomposed before the catalyst entrance at a flue gas temperature of 440 °C and a residence time in the exhaust pipe of 90 ms [1]. As urea decomposition only liberates half of the total molar amount of ammonia, the maximum total conversion efficiency prior to the catalyst is theoretically 50% (unless there is substantial hydrolyzation in the gas phase); this has also been supported by experimental observations [2].

Deposits in the exhaust pipe in front of the catalyst have been observed in many urea-SCR systems [1], [17], and the formation of melamine could be one major explanation for this [15]. It is generally believed that deposit formation is the result of a poorly adjusted AdBlue-spray [1], [18]. However, it is still unclear what happens if and when the spray hits the walls, and this has been addressed only recently [13]. Depending on temperatures and droplet deformation, the effects may range from film formation (which could lead either to finalized decomposition and hydrolyzation or to deposit formation) to splashing. Today, CFD softwares often include wall-film models that can be used to predict the interaction between a spray and a wall, but these could produce ambiguous results if not all aspects of the complex pathways of urea decomposition are taken into account. Hence, the work presented here is only directed towards estimating the extent of wall hit and to clarify the underlying reasons.

Section snippets

The system

Since the aim of the current work has been to study the influence of different modeling choices for CFD simulations of urea-SCR systems, it was decided to use a very simple and basic system setup, so as to not overshadow the implications of the choices made by the effects of an unnecessarily complex system design.

The system simulated in this work is depicted in Fig. 1. It is a straight exhaust gas pipe with a diameter of 120 mm. A single-phase (i.e. not air-assisted) AdBlue-injector is situated

Material data

When droplets are injected and transported by the exhaust gas stream they are heated and water evaporates until only pure urea is left. During the evaporation phase droplets will contain a mixture of urea and water. Since material properties depend on both temperature and mass fraction of urea, good material data is needed to make good model predictions.

The material data used is summarized in Table 4 for the gaseous phase and Table 5 for the droplet phase. This is the – to the authors’

Typical Eulerian–Lagrangian simulation results

In order to discuss changes in the simulation results due to different modeling choices, the typical results from Eulerian–Lagrangian simulations of the AdBlue-spray should first be established.

The smaller droplets will be dragged away by the main flow more quickly and will thus appear on the side of the pipe closest to the injector. Since the smallest droplets also are the ones that will lose their water and start to decompose first, most of the ammonia released ahead of the catalyst entrance

Discussion

The first and most important observation is of course that the choice of models influences the simulation results. Thus, the anticipated droplet fates in this system are also a function of the different sub-models used for the discrete phase in the performed Eulerian–Lagrangian simulations. It is therefore of high priority to understand what is causing these differences, both in order to be able to perform accurate simulations, as well as to be able to correctly assess the value of the

Conclusions

The aim of the presented work has been to assess the fidelity and influence of the most commonly used models for Eulerian–Lagrangian CFD-simulations of urea-SCR systems. It has been shown that:

  • Modeling choices may affect the predicted extent of wall hit, which types of droplets that are predicted to hit the wall, and also most possibly the simulated effects of wall-wetting if a spray/wall-interaction model is to be included in the simulations.

  • In Eulerian–Lagrangian modeling of the urea-SCR

Acknowledgements

This work has been performed within the Competence Centre for Catalysis, which is financially supported by Chalmers University of Technology, the Swedish Energy Agency and the member companies: AB Volvo, Volvo Car Corporation, Scania CV AB, GM Powertrain Sweden AB, Haldor Topsoe A/S and The Swedish Space Agency.

Thanks also to Linda Hellström for helping out with the graphics.

References (47)

  • M. Chen, S. Williams, Modelling and optimization of SCR-exhaust aftertreatment systems. SAE Technical Paper (2005)...
  • N.A. Chigier, The Physics of Atomization, in: Proceedings of the 5th International Conference on Liquid Atomization and...
  • M. Koebel, M. Elsener, G. Madia, Recent advances in the development of urea-SCR for automotive applications, SAE...
  • D.R. Linde (ed.), CRC Handbook of Chemistry and Physics, Internet Version 2007, 87th ed., Taylor & Francis, Boca Rato,...
  • T.Y. Oh et al.

    Design optimization of the mixing chamber in SCR system for marine diesel engine

  • W.H. Sun et al.

    Small scale test results from new selective catalytic NOx reduction process using urea

  • J.L. Calabrese, J.A. Patchett, K. Grimston, G.W. Rice, G.W. Davis, The influence of injector operating conditions on...
  • S. Sluder et al.

    Urea decomposition and SCR performance at low temperature

  • F. Birkhold, U. Meingast, P. Wassermann, O. Deutschmann, Analysis of the injection of urea–water-solution for...
  • C.S. Sluder, J.M.E. Storey, S.A. Lewis, L.A. Lewis, Low-temperature urea decomposition and SCR performance, SAE...
  • J.C. Ball, A toxicological evaluation of potential thermal degradation products of urea, SAE Technical Paper (2001)...
  • G. Stiesch

    Modelling Engine Spray and Combustion Processes

    (2003)
  • Fluent, Inc., Fluent 6.3 Documentation,...
  • Cited by (117)

    View all citing articles on Scopus
    View full text