Automotive Three-Way Exhaust Aftertreatment under Transient Conditions – Measurements, Modeling and Simulation

In this work a mathematical model for the dynamic thermal and the macrokinetic conversion behavior of a commercial automotive three-way exhaust catalytic converter is developed and parameterized with independent measurements. In spite of its relative simplicity the developed model is comprehensive enough to cover the major phenomena relevant for catalyst conversion under light-off conditions and – in particular – under warmed-up closed-loop conditions. A one-dimensional, two-phase model, incorporating all major reactions is set up. The model considers in addition to the three legislated pollutants carbon monoxide, hydrocarbons (which are represented by propene and propane) and nitrogen oxide the species hydrogen and oxygen. Between these species combustion, nitrogen oxide reduction, water gas shift and steam reforming reactions are considered. The respective species conversion rate parameters are adjusted to steady state experiments in isothermal flat bed reactors with a superimposed air to fuel ratio modulation of 1 Hz – simulating closed-loop engine control. Representative rich (λ = 0.98) and lean (λ = 1.02) exhaust compositions are retrieved from several emission certification driving cycles (both the European EU98 and the US-American FTP75). Starting with these base compositions individual species’ concentrations are varied within a wide temperature range of 100 ≤ T ≤ 700◦C. Thus kinetic experiments do not just cover the catalytic light-off temperature range (up to 450 ◦C catalyst temperature) but immerse into the high temperature range where homogeneous reaction rates overtake the catalytic reaction rates. Indeed catalytic combustion rates of hydrogen, propene and propane adjusted in the catalytic light-off regime would fail to describe the temperature activation of the combustion reactions at temperatures beyond 600 ◦C. A second combustion mechanism, referred to as heterogeneous-homogeneous coupled, is thus introduced. This novel approach is in good agreement with combustion research. It is shown that all relevant gas-gas reactions proceeding in a catalytic converter at temperatures between 100 and 700 ◦C at rich and lean conditions can be covered with one single set of reaction rate parameters (table 5.1). One main feature in λ -controlled three-way catalytic exhaust conversion is the catalyst’s ability to store oxygen under fuel lean (oxygen surplus) conditions and to release oxygen under fuel rich (oxygen deficient) conditions. Oxygen saturated oxygen storage materials (such as ceria-zirconia solid solutions) decompose once the exhaust oxygen partial pressure falls below the respective material’s equilibrium oxygen partial pressure. Decomposition is accompanied by oxygen release – and thus reduc18 tion of the oxygen storage material. Since oxygen partial pressures in the order of yO2 = 1 · 10−30 . . .1 · 10−10 are necessary for reduction, the consideration of an exact thermodynamic description (which is available for certain relevant materials) would lead to excessive demands on the numerical solution’s precision. Instead a simple sorption type rate expression depending on the richness/leanness – expressed in the mixture’s deviation from stoichiometry (λ −1) and the oxidation extent of the oxygen storage X – is applied for oxygen storage and release. Oxygen storage capacity and kinetics are measured by lean/rich-step experiments in flat-bed reactors under isothermal conditions. It turns out that the respective storage capacity and the rate constant of oxygen uptake are increasing with temperature whereas the rate constant of oxygen release is found to be temperature independent. The resulting model is validated by comparing simulation results with a set of detailed emission certification test results and is able to represent the measured temperature and conversion behavior with good accuracy. This is in particular remarkable since all parameters – except the heat of oxygen storage – are retrieved from catalyst shipping papers, literature and independent, isothermal measurements presented in this work. For a good match of simulated and measured catalyst bed temperatures the heat of oxygen storage was adjusted to ∆hR,os = 200 kJ/(mol O2) which is in good agreement with calorimetric measurements reported in literature. Furthermore the model provides information not directly accessible by measurements like the spatial evolution of the oxidation extent of the oxygen storage material which is a most valuable information for engine control and calibration engineers. Under fuel rich conditions reversible catalyst deactivation is a major issue in exhaust emission control. In this work sulfur poisoning is shown to be of particular importance. Sulfur species – either instantaneously fed to the catalyst or previously stored under lean conditions and subsequent released under rich conditions – deactivate the catalyst. Not just combustion efficiency is affected by sulfur poisoning but also nitrogen oxide reduction, which is expected to be high under fuel rich conditions. The deactivation is slow below 450 ◦C and fast above 550 ◦C and turns out to be reversible by exposure to lean exhaust conditions. However sulfur stored within the catalyst can only be purged at elevated temperatures. The catalyst investigated in this work is purged at 700 ◦C under strong reducing conditions prior to kinetic measurements. A deactivation of water gas shift and steam reforming reactions with rich exhaust is also observed for sulfur free feed at temperatures below 350 ◦C. In accordance with equilibrium conversion of the Boudouard reaction (figure 3.4) this is attributed to the formation of carbonaceous deposits at the active catalyst sites. High temperature deactivation either by extended periods of constant high temperature or of frequent, short, high temperature spikes (e.g. due to deceleration with fuel-cut) cause both a decrease in catalyst activity and oxygen storage capacity. Current on board diagnostics (OBD) of catalyst activity rely on this correlation and measure the aging state dependent oxygen storage capacity. It is shown that the investigated catalyst’s oxygen storage capacity Ω drops from a value of 1.5 g/lcat in its fresh/conditioned state to a value of 0.05 g/lcat after four hours of lean aging at 1300 ◦C (figure B.3). This decrease is due to segregation of active oxygen storage components (such as ceria19 zirconia solid solutions) into e.g. pure phases of cerium and zirconium oxide, both unable to release – and thus to store – oxygen (equilibrium partial pressures above pure cerium oxide see figure 3.7). A relation between the decrease in oxygen storage capacity due to thermal aging and increasing ignition temperatures of carbon monoxide and hydrocarbons is derived (figure B.2). A high temperature burner-fired experimental set-up, which allows to subject the catalyst to high steady state and transient catalyst temperatures (up to 1300 ◦C) at typical exhaust conditions, is developed in this work. Aging with simulated fuel cut decelerations confirm the dominant impact of catalyst temperature (measurable with standard equipment) on catalyst aging: If the exothermic potential is released within the catalyst – and thus the very first section of the catalyst is at a colder temperature compared to the case with the same amount of heat supplied by the hot exhaust – it is less aged than the respective section submitted to higher inlet temperatures (figure B.4). The sensitivity of the linear oxygen sensor (applied for oxygen storage measurements in this work) towards different exhaust constituents is measured and a literature model of the so-called sensor distortion is validated in appendix D. Furthermore the appendix comprises a thorough description of the experimental facilities, physical catalyst data, physical and thermochemical properties of gases, solids and reactions and heat and mass transfer correlations applied in this work.