WO1999036689A1

WO1999036689A1 - LEAN REGENERATION OF NOx STORAGE UNITS

Info

Publication number: WO1999036689A1
Application number: PCT/EP1998/008290
Authority: WO
Inventors: Ekkehard Pott
Original assignee: Volkswagen Aktiengesellschaft
Priority date: 1998-01-19
Filing date: 1998-12-17
Publication date: 1999-07-22
Also published as: DE19801815A1; EP1049861A1; DE59806001D1; EP1049861B1

Abstract

The invention relates to a method for cleaning the exhaust gas of a μ-controlled, lean-mix internal combustion engine which has a NOx storage catalytic converter (2) and a μ probe. According to said method, in the presence of stoichiometric or lean exhaust gas (5) with a relatively high concentration of oxygen, each volume element (3) of the NOx storage catalytic converter is impinged alternately with rich (4) or lean (5) exhaust gas in a time and position-dependent manner. The μ value oscillates in the direction of the time axis about a mean value μ-m, which mean value μ-m is greater than or equal to 1.

Description

Aging regeneration of NOx storage

The invention relates to a method for regeneration of a NOx storage catalytic converter in lean-burn internal combustion engines. An internal combustion engine can be operated lean if it is used at least for a subset of all conceivable speed-load combinations (in particular 1.5 x idle speed to 0.25 x nominal speed, 0.05 to 0.15 x Pme, max) Lambda> 1, 1 and particularly advantageously Lambda> 1, 3 can be operated over periods of> 10 seconds and particularly advantageously> 30 seconds.

The air-fuel ratio in the exhaust gas from lambda-controlled gasoline engines is usually monitored by one or more lambda sensors arranged in the exhaust line before and / or after the catalytic converter (s). Conventional jump probes show a pronounced voltage jump at lambda = 1, which is used in the engine control to shift the injection quantity towards "lean" at high voltage and towards "rich" at low voltage. With the control frequency of the probe, slightly rich and slightly lean exhaust gas is thus generated in the engine and pushed out into the exhaust system. Moreover, in real engine operation, all cylinders do not run without a deviation from one another or without a deviation from the desired lambda signal. Since the gas columns are mixed on the way through the exhaust gas aftertreatment, with stoichiometric control the catalyst does not have sharply separated exhaust gas qualities, but clouds with rich and lean exhaust gas.

All of the current technical and patent literature, in particular that of Toyota, specifies a rich or stoichiometric air-fuel ratio as a prerequisite for the regeneration of NOx storage catalytic converters. An oxygen-free environment and the presence of a reducing agent (HC or CO) are required for regeneration. With a homogeneous inflow, such as on synthesis gas test stands, this boundary condition can be represented over the entire catalyst cross section for any length of time; this is not possible on the motor. The invention is therefore based on the object of providing a regeneration method for the NOx accumulator of an internal combustion engine which also works in an oxygen-containing exhaust gas atmosphere.

The object is solved by the subject matter of claim 1. Preferred embodiments of the invention are the subject of the dependent claims.

In the method according to the invention for cleaning the exhaust gas of a lean-operated, λ-controlled internal combustion engine with a NOx storage catalytic converter and a λ probe, each volume element of the NOx storage catalytic converter alternates with time and place depending on the stoichiometric or lean exhaust gas with a relatively high oxygen concentration in the exhaust gas rich and lean exhaust gas.

Experiments with NOx storage catalytic converters on a lean-out engine with operating state-dependent λ control have shown that safe NOx regeneration is indeed possible with stoichiometric exhaust gas and relatively high residual oxygen concentrations. Each element of the NOx storage catalytic converter is alternately charged with rich and lean exhaust gas depending on the time and location. The NOx regeneration takes place in the time portions with a rich load. This effect is used in accordance with the method according to the invention in order to be able to clean NOx storage catalytic converters even in the case of lean exhaust gas. The duration of regeneration depends on the proportion of time the fat is applied; compared to globally rich exhaust gas, an increase in the regeneration duration can be expected in the method according to the invention.

The λ value advantageously oscillates in the direction of the time axis around an average value λ _m , the average value λ _{m being} greater than or equal to one, in particular> 1.05. The oscillation of the λ value around the mean value λ _{m can} be a sinusoidal oscillation or, for example, a triangular oscillation, such as a sawtooth.

In order to obtain a specific regeneration behavior of the NOx store, the amplitude of the oscillation is advantageously changed. Furthermore, the frequency of the vibration can also be made variable. In other words, it can Amplitude or frequency modulation of the λ function take place. In the corresponding application, amplitude and frequency modulation can be combined.

The average λ value can advantageously be generated by cylinder-selective control of the internal combustion engine. That is, some of the cylinders are operated with a rich λ value, while the other part of the cylinders are operated with a lean λ value. The individual λ values of both the rich cylinder and those of the lean cylinder can differ from one another and from one another and are adapted to the respective requirement. Furthermore, the λ value of the individual cylinder can be changed from cycle to cycle.

Furthermore, the control of the lean exhaust gas can advantageously be generated by a change in the leaning rate or by a change in the dead times of the change in the injection quantity.

The control frequencies of the λ vibration are currently in the order of 0.1 to 20 Hz and are ultimately a function of the response times of the λ probes used. With the development of probes with faster reaction times, it will be possible to increase the control frequency, whereby very high control frequencies can have a negative effect on the regeneration times with an average lean exhaust gas due to the decreasing "cloud formation".

Preferred embodiments of the invention are explained below with reference to the drawings.

FIGS. 1-5 each schematically show a storage catalytic converter in the upper part and the corresponding λ function in the lower part;

6 shows an amplitude modulation of the λ signal,

7 shows a frequency modulation of the λ signal, 8 shows waveforms of the λ signal in the form of a left-hand and right-hand saw tooth,

Fig. 9 shows the course of the λ signal when the lean regulation

Exhaust gas by changing the dead times of the injection quantity change,

10 shows the course of the λ signal when the lean is controlled

Exhaust gas by changing the leaning rate,

11 shows the adjustment of λ> 1 by cylinder-selective

Influencing the injection quantity in broadband lambda sensors, and

Fig. 12 shows the adjustment of λ> 1 by cylinder-selective

Influencing the injection quantity in step response lambda sensors.

Figures 1-5 graphically show the underlying mechanisms of lean regeneration of NOx stores. An exhaust system 1, which has a NOx storage catalytic converter 2, is shown in the respective upper part of FIGS. 1-5. An idealized catalytic converter element 3 is considered in the storage catalytic converter, the flow through the catalytic converter element 3 being shown with the different exhaust gas qualities. The corresponding λ values are plotted against time t in the lower part of FIGS. 1-5. The origin of the time axis is located at λ value one. The λ values greater than one (lean exhaust gas) are shown above and the λ values less than one (rich exhaust gas) are shown below. Furthermore, the mean value λ _{m is shown} as a dashed line.

1 shows an exclusively rich exhaust gas 4, represented by the hatching of the entire exhaust system 1. With this rich regeneration, the shortest regeneration times are possible thanks to time and location-resolved almost 100% rich flow. Because of the complete flow through the storage catalytic converter 2 with rich exhaust gas 4, any catalytic converter element 3 is always flowed through with rich exhaust gas 4 in both temporal and spatial resolution. In the lower part of Fig. 1 is the time course of λ am Catalyst element 4 shown. Due to the control frequency of the λ probe, the λ value oscillates around an average value λ _m , the amplitude of the λ oscillation always being in the rich range, ie λ <1. In other words, as is shown schematically in the upper part of FIG. 1, the catalytic element 3 is always flowed through with rich exhaust gas, the exhaust gas being periodically more or less rich.

FIG. 2 shows the situation in an exhaust system 1 with a NOx storage catalytic converter, through which rich exhaust gas 4 and lean exhaust gas 5 flow. This is shown schematically by rich exhaust gas clouds 4 (hatched areas) which are surrounded by lean exhaust gas clouds 5 (shown as white areas). It is therefore clear that any catalytic converter element 3 is flowed through in a temporally and spatially resolving manner, alternately with rich exhaust gas 4 and lean exhaust gas 5. That is, there are already small, lean fractions in the exhaust gas, so that time and location-resolved flow through all catalyst zones is not always rich. Shown in the lower part of FIG. 2, this means that the mean value λ _{m is} closer to λ = 1 and the amplitudes of the λ value exceed the stoichiometric value of λ = 1. Regeneration therefore takes place in the hatched areas I, in which the λ value falls below the value 1, while no regeneration takes place in the areas II, in which the λ value is exceeded by 1. On average, the NOx storage is regenerated based on the average value λ _m close to 1.

Fig. 3 shows a similar situation as Fig. 2 with the difference that the mean value λ _m = 1. The time components of rich and lean exhaust gas 4, 5 are thus approximately the same, which leads to approximately the same areas I of regeneration and areas II of no regeneration in the lower part of FIG. 3. The NOx storage is also regenerated here, however, the regeneration duration continues to increase.

Fig. 4 shows the situation with globally leaner exhaust gas, shown schematically by the fact that the number of clouds of rich exhaust gas 4 is less than the lean exhaust gas 5. The time and location components of the rich exhaust gas 4 continue to decrease and the regeneration duration increases increasingly . In the lower representation of FIG. 4, this means that the amplitudes of λ are largely above the value 1 and only a small part of the values of λ are below the value 1. The middle one Value λ _m is above 1. The regeneration areas I are smaller than the areas II in which no regeneration takes place. However, the NOx storage is still regenerated here.

Fig. 5 shows the situation with very lean exhaust gas 5. In the illustration, there are no more rich exhaust gas clouds, or the time and place components of the rich exhaust gas are too small for a practical regeneration period. A net storage discharge is only possible if the NOx mass flow converted by the time and place-resolved regeneration is greater than the lean NOx storage. In the lower λ representation, the course of λ and the mean value λ _{m are} now completely above 1, ie there is only area II without regeneration.

6 shows the increase in the amount of reducing agent at a predetermined mean value λm by increasing the λ amplitude. As the proportion of pollutants rises, the regeneration duration tends to decrease. It can be clearly seen that the amplitude of λ increases in the direction of the arrow and therefore the regeneration areas I also increase in area, i. H. the proportion of pollutants required for regeneration increases in the direction of the arrow. The regeneration speed also increases in the direction of the arrow. The increase in amplitude of λ can also be referred to as amplitude modulation.

FIG. 7 shows an optimization of the regeneration speed in NOx stores with oxygen storage capability by changing the control frequency, represented by a λ oscillation in which the control frequency is varied. In the present example, the control frequency is reduced, with λ fluctuating around an average value λ _m . The frequency of the change between rich and lean flow (wobble frequency) influences the regeneration time. As the oxygen storage capacity of the NOx storage increases, a decreasing wobble frequency also causes a decrease in the regeneration times. Regeneration areas I, non-regeneration areas II, and areas III can be seen in the areas defined by the oscillation of λ, regeneration also not taking place in areas III below the stoichiometric λ value, since stored O2 is consumed in the case of an oxygen-storing NOx store . In the direction of the arrow, there is therefore a decrease in the control frequency, an increase in the usable amount of reducing agent and a Increase in the rate of regeneration. As already mentioned, the control frequencies are currently in the order of 0.1 to 20 Hz.

Fig. 8 shows the variation of the pollutant reduction properties by a shape of λ. It has been shown that shaping the fat-lean jumps influences the regeneration behavior. The left graph in FIG. 8 shows a rapid enrichment and subsequent slow leaning. In the illustration, this results in the sawtooth curve falling rapidly after fat, also referred to as right-hand sawtooth, which shortens the NOx regeneration by starting the NOx conversion quickly, but there is a risk of HC and CO breakdowns. The sawtooth shown on the right represents a rapid leaning and then slow enrichment (left-sided sawtooth). Such a course that drops rapidly after fat causes a slower start of the NOx conversion with better control of the HC and CO breakthroughs and brings about a longer regeneration time in comparison to the left sawtooth course of the rapid enrichment. In contrast to the idealized image representation, the emaciation and enrichment need not necessarily be linear. Furthermore, areas I of regeneration and those without regeneration II are shown again.

The corresponding exhaust gas quality is generated controlled by a step response or broadband lambda probe.

9 shows the control of a lean exhaust gas by changing the dead times of the injection quantity change in the case of a step response probe. With such a step response probe, the setting of the globally lean exhaust gas takes place through different dead times between the detection of lean exhaust gas and the command to enrich the mixture and the detection of rich exhaust gas and the command to lean the mixture. Plotted in FIG. 9 is the probe voltage Vs of a step response probe versus time t in the upper graph and the corresponding injection quantity signal E versus time in the lower part of FIG. 9, namely once for an average value λ _m = 1 and in the right part for λ _m > 1. T1 means the dead time until the detection of "lambda lean", T2 the dead time until the readjustment of the injection after rich, T3 the dead time until the detection of "lambda rich" and T4 the dead time until the readjustment of the injection after skinny. In this way, the proposals of FIGS. 4 and 6 can be implemented. The arrow on the right shows it Part of FIG. 9 for λm> 1 for a longer dead time T4 with lean exhaust gas in comparison to T4 of the left part of FIG. 9 for λ _m = 1.

FIG. 10 shows a control of a lean exhaust gas with a step response probe by changing the leaning rate, that is to say by means of different rates of change during enriching and leaning, whereby the ancestor of FIG. 8 is realized. The meaning of the reference symbols T1, T2, T3 and T4 corresponds to that of FIG. 9. The two arrows in the right part of FIG. 10 for λ _m > 1 indicate a faster and stronger emaciation with lean exhaust gas.

In the case of broadband probes, the target signal of the average lambda λ _{m is set} to the desired value (lambda> 1) and the time components of rich exhaust gas are monitored via frequency and amplitude.

FIG. 11 shows the regulation of λ _m > 1 by influencing the cylinder-selective injection quantity in broadband lambda sensors, the mean value λ _{m being represented} by the dashed line. In addition to the application of the same injection signal to all cylinders, it is also conceivable that with an n-cylinder engine 1 to (n-1) cylinders (here cylinders 1, 2, 3 and 4) receive an injection signal which differs from the other cylinders, whereby more than two different injection signals can also be specified (e.g. Ix very rich, Ix slightly rich, 2x lean). In the case of broadband lambda probes, the lambda signal continues to be used for global regulation to λ _m > 1. When controlling the rich and / or stoichiometrically running cylinders, lambda is regulated by regulating at least one lean-running cylinder. 11 is divided into areas A, B and C, which have the following meanings:

A: Control of rich cylinders 1 and 4, adjusted regulation of cylinders 2 and 3 (ideal case).

B: Readjustment of cylinder 2 if cylinder 4 deviates from the specified setpoint.

C: A deliberately different control of the rich cylinders 1 and 4 is compensated for by a modified regulation of the cylinders 2 and 3. Finally, FIG. 12 shows a cylinder-selective injection quantity influencing for the generation of a lean exhaust gas with a step response probe. The thick line means the λ of the individual cylinders, the dashed line an average λ over a cycle and the double solid line means the average λm over several cycles. Here, too, the origin of the time axis is λ = 1. In the case of step response probes, the cylinders running bold and / or stoichiometrically are also controlled and the cylinders running lean are regulated. Since no information about the degree of leanness can be obtained with the step response probe, a leaner than the mean desired lambda value λm (1 + 2 * Δλ, ie. Lambda 1, 06 with desired lambda 1.03) is first approached in a controlled manner by the Cylinders 2 and 3 are operated very lean. S denotes the step response of the probe.

The leaning of cylinders 2 and 3 is gradually reduced over the following work cycles until the step response probe detects globally rich exhaust gas. Then the highest lean value of the lean-running cylinders is set again. 12 illustrates this procedure in principle, although different cylinder-specific leanings / enrichments are also conceivable here.

In the case of lean gasoline and diesel engines, this procedure not only reduces consumption but also HC and CO breakthroughs during regeneration; in diesel engines, moreover, particle emissions.

If all or part of the reduction in consumption is converted into a more frequent regeneration, additional advantages in sulfur poisoning are to be expected, since a part of the deposited sulfates is also removed again with each NOx regeneration. The released sulfur is increasingly emitted in the form of SO ₂ ; the odor-causing H ₂ S formation is largely suppressed. Likewise, only slight NH ₃ formation is to be expected. LIST OF REFERENCE NUMBERS

1 exhaust system

2 NOx storage catalytic converter

3 catalyst element

4 Fat exhaust gas

5 Lean exhaust gas

I regeneration area

II Non-regeneration area

III Non-regeneration area

T1 dead time

T2 dead time

T3 dead time

T4 dead time λm mean λ value s probe voltage

E injection quantity signal

S step response signal

Claims

PATENT CLAIMS

1. A method for cleaning the exhaust gas of a lean, λ-controlled internal combustion engine with a NOx storage catalyst (2) and a λ probe, characterized in that each volume element of the NOx storage catalyst with stoichiometric or lean exhaust gas with a relatively high oxygen concentration in the exhaust gas depending on the time and location, alternating between rich and lean exhaust gas.

2. The method according to claim 1, characterized in that the λ value oscillates in the direction of the time axis around an average value λ _m , the average value λ _{m being} greater than or equal to one.

3. The method according to claim 1 or 2, characterized in that the oscillation of λ is a sinusoidal oscillation.

4. The method according to claim 1 or 2, characterized in that the oscillation of λ is a triangular oscillation.

5. The method according to any one of claims 3 or 4, characterized in that the amplitude of the λ vibration is changed.

6. The method according to any one of claims 2-5, characterized in that the frequency of the λ vibration is variable.

7. The method according to any one of the preceding claims, characterized in that the average value λ _{m is} generated by a cylinder-selective control of the internal combustion engine.

8. The method according to claim 7, characterized in that some of the cylinders are operated with a rich λ value, while the other part of the cylinders are operated with a lean λ value.

9. The method according to claim 8, characterized in that both the λ values of the rich cylinders and those of the lean cylinders are different from each other.

10. The method according to any one of claims 1-6, characterized in that the control of the lean exhaust gas is generated by a change in the leaning rate.

11. The method according to any one of claims 1-6, characterized in that the control of the lean exhaust gas is generated by a change in the dead times of the change in injection quantity.

12. The method according to any one of the preceding claims, characterized in that the frequency of the vibration is greater than or equal to 0.1 Hz.