Carol A. Roberts

Modelling the flow distribution through automotive catalytic converters

Abstract Conventional computational fluid dynamics (CFD) methods for simulating the flow through automotive exhaust catalysts assume a monolith resistance based on one-dimensional laminar flow. This underpredicts the flow maldistribution in the monolith. Incorporation of an additional pressure loss accounting for entrance effects improves predictions for the maximum flow velocity within the substrate.

Automotive catalyst warm-up to light-off by pulsating engine exhaust:

Abstract This paper presents the results of studies on a range of different catalyst substrates warmed by engine exhaust. Engine speeds were in the range 1200–3000 r/min. One substrate was non-washcoated, four were washcoated but non-reactive and four were washcoated and reactive. The temperature at four locations within the non-reactive substrates was measured. The reactive substrates were warmed to light-off by the pulsating exhaust flow from an engine running fuel rich of stoichiometric. Both substrate temperatures and hydrocarbon conversion were measured. Predicted temperatures and conversion were obtained from a one-dimensional computational fluid dynamics (CFD) model. The model was based on the porous medium approach and incorporated a simple three-way chemical scheme. Comparison was made of measurements with predictions, with particular reference to the time taken to achieve light-off. Pulsing flow CFD predictions were found to be almost identical to steady-flow predictions for the conditions investigated. The CFD predictions were found to be in fair agreement with the engine test results, but using kinetic rate constants higher than previously reported values.

An experimental investigation of oblique entry pressure losses in automotive catalytic converters

Abstract This study investigates oblique entry pressure loss in automotive catalyst monoliths. Experiments have been performed on a specially designed flow rig using different lengths of monolith (17—100 mm) over a range of Reynolds number and angles of incidence (0–75°). Losses were found to be a function of Reynolds number and angle of incidence and a general correlation has been derived. Computational fluid dynamics predictions of the flow distribution across axisymmetric catalyst assemblies have been performed. Incorporating the oblique entry loss provided much better agreement with experimental data with the assumption that such losses were constant above an angle of incidence of 81°.

A Coupled 1D/3D Simulation for the Flow Behaviour inside a Close-Coupled Catalytic Converter

This paper describes the coupling of a 1D engine simulation code (Ricardo WAVE) to a 3D CFD code (STAR-CD) to study the flow behaviour inside a Close-Coupled Catalytic converter (CCC). A SI engine was modelled in WAVE and the CCC modelled in STAR-CD. The predictions of the stand-alone WAVE model were validated against engine bed tests before the coupled 1D/3D simulations were performed at 3000 RPM WOT for both motored and firing conditions. The predicted exhaust velocities downstream of the catalyst monolith in the coupled simulations matched fairly well with Laser Doppler Anemometry (LDA) measurements.

Warm up of an automotive catalyst substrate by pulsating flow: a single channel modelling approach

Abstract Rapid warm up of an automotive catalyst substrate is important for early light off. This work considers the results from a model of warm up in a single channel. The mass flow is pulsating with high amplitude, about 75% of mean flow, but without flow reversal. The flow regime is laminar within the channel. Pulsations occur with frequency in the range 16–100 Hz, and are important in close-coupled systems where the catalyst is located near to the engine and where the rate of rise of gas inlet temperature with time is rapid, about 15 K/s. The use of a single channel model with conjugate heat transfer enables the heat transfer coefficient to be evaluated and compared with results from steady flow simulations. The value of the augmentation factor based on heat flux is found to be less than unity. The value of the augmentation factor based on heat transfer coefficient depends on the method for calculating the mean heat transfer coefficient, but is generally less than unity. The changes caused by pulsations will be small in practical systems. Changes in wall temperature found in the simulations are the result of the cumulative effect of changes in the mass flow rate.

Warm up of automotive catalyst substrates: Comparison of measurements with predictions

Abstract An understanding of the warm up of automotive catalysts is important for accurate prediction of light off. This work describes some experimental studies on warm up in the absence of chemical reactions. In parallel with these experiments, the temperatures of the warmed substrate have been predicted. The problem of warm up is a simple one, capable of analytical description, but since the complete problem of catalyst performance with chemical reactions will ultimately require CFD coding, the simple case is dealt with in this way to form the basis of a more complete model. The studies have found measured heat transfer coefficients which are in the range 15 to 20 W/(m2K) for metallic substrates with sinusoidal channels. This is much lower than standard Nu values suggest. The predictions have also illustrated the significance of the heat transfer coefficint in obtaining accurate agreement with measurements in the simple case of warm up.

Lean NOx trap study on a light-duty diesel engine using fast-response emission analysers

Abstract Storage and regeneration events have been studied using fast-response emission analysers (∼10 ms) for a lean NO x trap (LNT) fitted to a light-duty diesel engine. Tests were conducted at both low and high exhaust temperatures for various storage and purging periods. The use of fast-response analysers has provided detailed information during the short regeneration periods and as combustion switched between rich and lean operating modes. It has also enabled quantification of the storage, reduction, and overall conversion efficiencies, as well as the instantaneous trapping efficiency. With exhaust temperatures of 250 °C, storage efficiency was low (∼30 per cent). During purging, two distinct NO spikes (breakthroughs) were measured downstream of the LNT at the beginning and end of regeneration. For this LNT, the primary reducing mechanism is CO reacting with NO, but CO reacting with ceria and/or water, the water—gas shift reaction, is suspected. With exhaust temperatures of 400 °C, storage efficiencies were high (∼80–90 per cent), except for the long-storage/short-purge case when the trap was near saturation. NO x breakthrough during enrichment depended on storage and purge periods and the availability of catalyst sites. NO2 breakthrough was also observed at the end of regeneration as the combustion switched to lean operation. Generally, for the high-temperature case on this LNT, the primary reducing mechanism is CO reacting with NO2.

Warming automotive catalysts with pulsating flows

Abstract Temperatures of an automotive catalyst substrate warmed by convection pre-light-off have been measured. Direct comparison has been made of warm-up by steady and pulsating flow for a one-dimensional flow case. The 32 Hz pulsating mass flow did not feature flow reversal. Pulsations were achieved by interruption of the airflow by a rotating disc. Very small differences between steady and pulsating cases were observed because the effect of mass flow pulsations on heat transfer is minimal. Two different computational fluid dynamics methods were used to predict temperature. A one-dimensional porous medium model, which required input of an assumed heat transfer coefficient, was compared with a single-channel model. Predictions agreed closely and there was also qualitative agreement with measurements. Similar mass flow pulsations in the range 32-100 Hz have been studied for a case with a larger diameter automotive catalyst supplied via a conical diffuser. The radial flow distribution is controlled by pulsation frequency and the effect of frequency on temperature at different depths in the substrate was observed experimentally. Pulsations will affect catalyst warm-up in practical systems because of their effect on flow distribution, rather than on heat transfer.

An Assessment of CFD Applied to Steady Flow in a Planar Diffuser Upstream of an Automotive Catalyst Monolith

Flow maldistribution across automotive exhaust catalysts significantly affects their conversion efficiency. Flow behaviour can be predicted using computational fluid dynamics (CFD). This study investigates the application of CFD to modelling flow in a 2D system consisting of a catalyst monolith downstream of a wide-angled planar diffuser presented with steady flow. Two distinct approaches, porous medium and individual channels, are used to model monoliths of length 27 mm and 100 mm. Flow predictions are compared to particle image velocimetry (PIV) measurements made in the diffuser and hot wire anemometry (HWA) data taken downstream of the monolith. Both simulations compare favourably with PIV measurements, although the models underestimate the degree of mixing in the shear layer at the periphery of the emerging jet. Tangential velocities are predicted well in the central jet region but are overestimated elsewhere, especially at the closest measured distance, 2.5 mm from the monolith. The individual channels model is found to provide a more consistently accurate velocity profile downstream of the monolith. Maximum velocities, on the centre line and at the secondary peak near to the wall, are reasonably well matched for the cases where the flow is more maldistributed. Under these conditions, a porous medium model remains attractive because of low computational demand.

Spatial Conversion Profiles within an SCR in a Test Exhaust System with Injection of Ammonia Gas Modelled in CFD using the Porous Medium Approach

Modelling of SCR in diesel exhaust systems with injection of urea spray is complex and challenging but many models use only the conversion observed at the brick exit as a test of the model. In this study, the case modelled is simplified by injecting ammonia gas in nitrogen in place of urea, but the spatial conversion profiles along the SCR brick length at steady state are investigated. This is a more rigorous way of assessing the ability of the model to simulate observations made on a test exhaust system. The data have been collected by repeated engine tests on eight different brick lengths, all which were shorter than a standard sized SCR. The tests have been carried out for supplied NH 3 /NOx ratios of α 1.5, excess ammonia, α 1.0, balanced ammonia, and α 0.5, deficient ammonia. Levels of NO, NO 2 and NH 3 have been measured both upstream and downstream of the SCR using a gas analyser fitted with ammonia scrubbers to give reliable NOx measurements. A CFD model based on the porous medium approach has incorporated a kinetic scheme available in the open literature. Comparison of CFD simulations with observed data is presented and the results are discussed. NOx conversion is significant in the first 30 mm of the brick for α 0.5 and in the first 90 mm for α 1.0 and 1.5. The ammonia level influences NOx conversion, which is generally under-predicted by the current model for α 0.5 and over-predicted for α 1.5. Measurements show that NO 2 conversion exceeds NO conversion in the first section of the monolith, which the model fails to predict. INTRODUCTION Selective catalytic reduction, SCR, with aqueous urea has become one of the primary methods of NOx reduction for both heavy and light-duty diesel applications over the past few years. The urea is thermally hydrolysed in the vehicles exhaust to produce ammonia, which in turn reacts with the NOx over a catalyst. SCR technology is advantageous for light-duty applications as it has a minimal impact on fuel consumption and costs, yet has high NOx reduction efficiencies [1], and [2]. NOx reduction is affected by the exhaust gas temperature. For low gas temperatures the fast SCR reaction is the dominant reaction where approximately equi-molar quantities of NO and NO2 are present in the exhaust. The amount of NH3 injected requires careful control, and can be critical under some conditions. Over-injection of NH3 into the exhaust could potentially provide a high NOx conversion, but is likely to result in NH3 slip [3]. The addition of an NH3 oxidation catalyst downstream of the SCR is an option, but this may cause NO formation, which in turn will result in the vehicle not complying with emissions standards [4]. Underinjection of NH3 will avoid NH3 slip, but will provide an inferior NOx conversion. The control of the NO2: NOx ratio in the exhaust stream and the amount of NH3 injected into the exhaust at any one time are therefore very important. This can be investigated experimentally, but this is a time consuming process and is not cost effective [5]. Simulation modelling can be considered a useful tool in reducing development time and costs. Spatial Conversion Profiles within an SCR in a Test Exhaust System with Injection of Ammonia Gas Modelled in CFD using the Porous Medium Approach 2010-01-2089 Published 10/25/2010 M. P. Sturgess, S. F. Benjamin and C. A. Roberts Coventry University Copyright