Search Results

Diesel Oxidation Catalysts (DOC) are used on heavy duty diesel engine applications and experience large internal temperature variations from 150 to 600°C. The DOC oxidizes the CO and HC in the exhaust to CO2 and H2O and oxidizes NO to NO2. The oxidation reactions are functions of its internal temperatures. Hence, accurate estimation of internal temperatures is important both for onboard diagnostic and aftertreatment closed loop control strategies. This paper focuses on the development of a reduced order model and an Extended Kalman Filter (EKF) state estimator for a DOC. The reduced order model simulation results are compared to experimental data. This is important since the reduced order model is used in the EKF estimator to predict the CO, NO, NO2 and HC concentrations in the DOC and at the outlet. The estimator was exercised using transient drive cycle engine data. The closed loop EKF improves the temperature estimate inside the DOC compared to the open loop estimator.

Numerical models of aftertreatment devices are increasingly becoming indispensable tools in the development of aftertreatment systems that enable modern diesel engines to comply with exhaust emissions regulations while minimizing the cost and development time involved. Such a numerical model was developed at Michigan Technological University (MTU) [1] and demonstrated to be able to simulate the experimental data [2] in predicting the characteristic pressure drop and PM mass retained during passive oxidation [3] and active regeneration [4] of a catalyzed diesel particulate filter (CPF) on a Cummins ISL engine. One of the critical aspects of a calibrated numerical model is its usability - in other words, how useful is the model in predicting the pressure drop and the PM mass retained in another particulate filter on a different engine without the need for extensive recalibration.

Optimizing the performance of the aftertreatment system used on heavy duty diesel engines requires a thorough understanding of the operational characteristics of the individual components. Within this, understanding the performance of the catalyzed particulate filter (CPF), and the development of an accurate CPF model, requires knowledge of the particulate matter (PM) distribution throughout the substrate. Experimental measurements of the PM distribution provide the detailed interactions of PM loading, passive oxidation, and active regeneration. Recently, a terahertz wave scanner has been developed that can non-destructively measure the three dimensional (3D) PM distribution. To enable quantitative comparisons of the PM distributions collected under different operational conditions, it is beneficial if the results can be discussed in terms of the axial, radial, and angular directions.

Selective catalytic reduction (SCR) systems are in use on heavy duty diesel engines for NOx control. An SCR NOx reduction efficiency of higher than 95% is required to meet the proposed increasingly stringent NOx emission standards and the 2014-2018 fuel consumption regulations. The complex engine exhaust conditions including the nonuniformity of temperature, flow, and maldistribution of NH3 present at the catalyst inlet need to be considered for improved performance of the SCR system. These factors cause the SCR to underperform negatively impacting the NOx reduction efficiency as well as the NH3 slip. In this study, the effects of the nonuniformity of temperature, flow velocity and maldistribution of NH3 on the SCR performance were investigated using 1-dimensional (1D) model simulations for a Cu-zeolite SCR. The model was previously calibrated and validated to reactor and steady-state and transient engine experimental data.

This paper focuses on the development of an Extended Kalman Filter for estimating internal species concentration and storage states of an SCR using NOX and NH₃ sensors. The motivation for this work was twofold. First, knowledge of internal states may be useful for onboard diagnostic strategy development. In particular, significant errors between the outlet NOX or NH₃ sensors, reconstructed from estimated states, and the measured NOX or NH₃ concentrations may aid OBD strategies that attempt to identify particular system failure modes. Second, the EKF described estimates not only stored ammonia but also NO, NO₂ and NH₃ gas concentrations within and exiting the SCR. Exploiting knowledge of the individual species concentrations, instead of lumping them together as NOX, can yield improved closed loop urea controller performance in terms of reduced urea consumption and better NOX conversion.

Numerical modeling of aftertreatment systems has been proven to reduce development time as well as to facilitate understanding of the internal physical and chemical processes occurring during different operating conditions. Such a numerical model for a catalyzed diesel particulate filter (CPF) was developed in this research work which has been improved from an existing numerical model briefly described in reference. The focus of this CPF model was to predict the effect of the catalyst on the gaseous species concentrations and to develop particulate matter (PM) filtration and oxidation models for the PM cake layer and substrate wall so as to develop an overall model that accurately predicts the pressure drop and PM oxidized during passive oxidation and active regeneration. Descriptions of the governing equations and corresponding numerical methods used with relevant boundary conditions are presented.

A high-fidelity multi-step global kinetic Selective Catalytic Reduction (SCR) model which can predict SCR performance in engine exhaust systems is desirable for optimizing the SCR system, designing on-vehicle control systems and on-board diagnostic (OBD) functions. In this study, a Cu-zeolite SCR catalyst in the exhaust of a 2010 Cummins 6.7L ISB diesel engine was experimentally studied under both steady-state and transient conditions. Steady-state engine tests spanned SCR inlet temperatures from 250 to 400°C with a constant space velocity of 60 khr-1. A 1-D Cu-zeolite model originally developed from reactor data was improved and calibrated to the steady-state engine experimental data. The calibrated model is capable of predicting NO/NO₂ reduction, NH₃ slip, and NH₃ storage associated phenomena.

In order to further characterize and optimize the performance of Selective Catalytic Reduction (SCR) aftertreatment systems used on heavy-duty diesel engines, an accurately calibrated high-fidelity multi-step global kinetic SCR model and a reduced order estimator for on-board diagnostic (OBD) and control are desirable. In this study, a Cu-zeolite SCR catalyst from a 2010 Cummins ISB engine was experimentally studied in a flow reactor using carefully designed protocols. A 2-site SCR model describing mass transfer and the SCR chemical reaction mechanisms is described in the paper. The model was calibrated to the reactor test data sets collected under temperatures from 200 to 425 °C and SCR space velocities of 60000, 90000, and 120000 hr-1. The model parameters were calibrated using an optimization code to minimize the error between measured and simulated NO, NO₂, N₂O, and NH₃ gas concentration time histories.

Active regeneration experiments were carried out on a production 2007 Cummins 8.9L ISL engine and associated diesel oxidation catalyst (DOC) and catalyzed particulate filter (CPF) aftertreatment system. The effects of SME biodiesel blends were investigated to determine the particulate matter (PM) oxidation reaction rates for active regeneration. The experimental data from this study will also be used to calibrate the MTU-1D CPF model [1]. The experiments covered a range of CPF inlet temperatures using ULSD, B10, and B20 blends of biodiesel. The majority of the tests were performed at a CPF PM loading of 2.2 g/L with in-cylinder dosing, although 4.1 g/L and a post-turbo dosing injector were also investigated. The PM reaction rate was shown to increase with increasing percent biodiesel in the test fuel as well as increasing CPF temperature.

A 2007 Cummins ISL 8.9L direct-injection common rail diesel engine rated at 272 kW (365 hp) was used to load the filter to 2.2 g/L and passively oxidize particulate matter (PM) within a 2007 OEM aftertreatment system consisting of a diesel oxidation catalyst (DOC) and catalyzed particulate filter (CPF). Having a better understanding of the passive NO₂ oxidation kinetics of PM within the CPF allows for reducing the frequency of active regenerations (hydrocarbon injection) and the associated fuel penalties. Being able to model the passive oxidation of accumulated PM in the CPF is critical to creating accurate state estimation strategies. The MTU 1-D CPF model will be used to simulate data collected from this study to examine differences in the PM oxidation kinetics when soy methyl ester (SME) biodiesel is used as the source of fuel for the engine.

The catalyzed particulate filter (CPF), used in conjunction with a diesel oxidation catalyst (DOC) is an important aftertreatment device used to meet Environmental Protection Agency (EPA) heavy-duty diesel emission standards for particulate matter (PM). Numerical modeling of these exhaust after-treatment devices decreases the time and cost of development involved. Modeling CPF active regeneration gives insight into the PM oxidation kinetics, which helps in reducing the regeneration fuel penalty. As seen from experimental data, active regeneration of the CPF results in a significant temperature increase into the CPF (up to 8°C/sec) which affects the oxidation rate of particulate matter (PM). PM oxidation during active regeneration was determined to be a function of filter PM loading, inlet temperature and inlet hydrocarbon concentration.

Active regeneration experiments were performed on a production diesel aftertreatment system containing a diesel oxidation catalyst and catalyzed particulate filter (CPF) using blends of soy-based biodiesel. The effects of biodiesel on particulate matter oxidation rates in the filter were explored. These experiments are a continuation of the work performed by Chilumukuru et al., in SAE Technical Paper No. 2009-01-1474, which studied the active regeneration characteristics of the same aftertreatment system using ultra-low sulfur diesel fuel. Experiments were conducted using a 10.8 L 2002 Cummins ISM heavy-duty diesel engine. Particulate matter loading of the filter was performed at the rated engine speed of 2100 rpm and 20% of the full engine load of 1120 Nm. At this engine speed and load the passive oxidation rate is low. The 17 L CPF was loaded to a particulate matter level of 2.2 g/L.

A quasi-steady 1-dimensional computer model of a catalyzed particulate filter (CPF) capable of simulating active regeneration of the CPF via diesel fuel injection upstream of a diesel oxidation catalyst (DOC) or other means to increase the exhaust gas temperature has been developed. This model is capable of predicting gaseous species concentrations (HC's, CO, NO and NO2) and exhaust gas temperatures within and after the CPF, for given input values of gaseous species and PM concentrations before the CPF and other inlet variables such as time-varying temperature of the exhaust gas at the inlet of the CPF and volumetric flow rate of exhaust gas.

Active regeneration of a catalyzed particulate filter (CPF) is affected by a number of parameters specifically particulate matter loading and inlet temperature. The MTU 1-D 2-Layer CPF model [1] was used to analyze these effects on the pressure drop, oxidation and filtration characteristics of a CPF during active regeneration. In addition, modeling results for post loading experiments were analyzed to understand the difference between loading a clean filter as compared to a partially regenerated filter. Experimental data obtained with a production Cummins regenerative particulate filter for loading, active regenerations and post loading experiments were used to calibrate the MTU 1-D 2-Layer CPF model. The model predicted results are compared with the experimental data and were analyzed to understand the CPF characteristics during active regeneration at 1.1, 2.2 and 4.1 g/L particulate matter (PM) loading and CPF inlet temperatures of 525, 550 and 600°C.

Active regeneration experiments were performed on a Cummins 2007 aftertreatment system by hydrocarbon dosing with injection of diesel fuel downstream of the turbocharger. The main objective was to characterize the thermal oxidation rate as a function of temperature and particulate matter (PM) loading of the catalyzed particulate filter (CPF). Partial regeneration tests were carried out to ensure measureable masses are retained in the CPF in order to model the oxidation kinetics. The CPF was subsequently re-loaded to determine the effects of partial regeneration during post-loading. A methodology for gathering particulate data for analysis and determination of thermal oxidation in a CPF system operating in the engine exhaust was developed. Durations of the active regeneration experiments were estimated using previous active regeneration work by Singh et al. 2006 [1] and were adjusted as the experiments progressed using a lumped oxidation model [2, 3].

Steady state loading characterization experiments were conducted at three different engine load conditions and rated speed on the cracked catalyzed particulate filter (CPF). The experiments were performed using a 10.8 L 2002 Cummins ISM-330 heavy duty diesel engine. The CPF underwent a ring off failure, commonly seen in particulate filters, due to high radial and axial temperature gradients. The filters were cracked during baking in an oven which was done to regenerate PM collected after every loading characterization experiment. Two different configurations i.e. with and without a diesel oxidation catalyst (DOC) upstream of the CPF were studied. The data were compared with that on an un-cracked CPF at similar engine conditions and configurations. Pressure drop, transient filtration efficiency by particle size and PM mass and gaseous emissions measurements were made during each experiment.

In this paper, a model-based linear estimator and a non-linear control law for an Fe-zeolite urea-selective catalytic reduction (SCR) catalyst for heavy duty diesel engine applications is presented. The novel aspect of this work is that the relevant species, NO, NO2 and NH3 are estimated and controlled independently. The ability to target NH3 slip is important not only to minimize urea consumption, but also to reduce this unregulated emission. Being able to discriminate between NO and NO2 is important for two reasons. First, recent Fe-zeolite catalyst studies suggest that NOx reduction is highly favored by the NO 2 based reactions. Second, NO2 is more toxic than NO to both the environment and human health. The estimator and control law are based on a 4-state model of the urea-SCR plant. A linearized version of the model is used for state estimation while the full nonlinear model is used for control design.

Steady-state particulate loading experiments were conducted on an advanced production catalyzed particulate filter (CPF), both with and without a diesel oxidation catalyst (DOC). A heavy-duty diesel engine was used for this study with the experiments conducted at 20, 40, 60 and 75 % of full load (1120 Nm) at rated speed (2100 rpm). The data obtained from these experiments were used and are necessary for calibrating the MTU 1-D 2-Layer CPF model. These experimental and modeling results were compared to previous research conducted at MTU that used the same engine but an earlier development version of the combination of DOC and CPF. The motivation for the comparison of the two systems was to determine whether the reformulated production catalysts performed as good or better than the early development catalysts. The results were compared to understand the filtration and oxidation differences between the two DOC+CPF and the CPF-only aftertreatment systems.

A Cummins ISB 5.9 liter medium-duty engine with cooled EGR has been used to study an early extrusion of an advanced ceramic uncatalyzed diesel particulate filter (DPF). Data for the advanced ceramic material (ACM) and an uncatalyzed cordierite filter of similar dimensions are presented. Pressure drop data as a function of mass loadings (0, 4, and 6 grams of particulate matter (PM) per liter of filter volume) for various flow rate/temperature combinations (0.115 - 0.187 kg/sec and 240 - 375 °C) based upon loads of 15, 25, 40 and 60% of full engine load (684 N-m) at 2300 rpm are presented. The data obtained from these experiments were used to calibrate the MTU 1-D 2-Layer computer model developed previously at MTU. Clean wall permeability determined from the model calibration for the ACM was 5.0e-13 m2 as compared to 3.0e-13 m2 for cordierite.

Model-based control strategies are important for meeting the dual objective of maximizing NOx reduction and minimizing NH3 slip in urea-SCR catalysts. To be implementable on the vehicle, the models should capture the essential behavior of the system, while not being computationally intensive. This paper discusses the adequacy of two different reduced order SCR catalyst models and compares their performance with a higher order model. The higher order model assumes that the catalyst has both diffusion and reaction kinetics, whereas the reduced order models contain only reaction kinetics. After describing each model, its parameter identification and model validation based on experiments on a Navistar I6 7.6L engine are presented. The adequacy of reduced order models is demonstrated by comparing the NO, NO2 and NH3 concentrations predicted by the models to their concentrations from the test data.