Method and control assembly for operating an exhaust gas system

Abstract

A method for operating an exhaust gas system is provided. The method includes determining an amount of a reducing agent to be supplied to the exhaust gas of an engine and evaluating measurements which indicate a content of nitrogen oxides in the exhaust gas downstream of a catalytic device adapted to diminish the content of nitrogen oxides. A magnitude and a frequency of the measurements are taken into account in determining the amount of the reducing agent to be supplied. A plurality of measurements is captured during a predetermined period of time. A magnitude of a measurement captured within this period of time is related to a quantity derived from the respective magnitudes of the plurality of measurements. The related measurement is utilized to determine the amount of the reducing agent to be supplied. A control assembly for operating an exhaust gas system is also provided.

Claims

1. A method for operating an exhaust gas system, the method comprising the steps of: receiving a signal reading from a nitrogen oxide sensor over a total data sampling time period (), wherein the nitrogen oxide sensor measures a content of nitrogen oxides in exhaust gas flowing downstream of a catalytic device configured to diminish the content of nitrogen oxides in the exhaust gas; obtaining a difference (x.sub.ix.sub.s) between a sampled tail pipe nitrogen oxide concentration (x.sub.i) and a median tail pipe nitrogen oxide concentration (x.sub.s) repeatedly over the total data sampling time period; determining a lumped average relative magnitude frequency (LARMF) by way of a first relationship, $\mathrm{LARMF}=\frac{1}{}{}_0^{}.Math.x_i-x_s.Math.dt;$ establishing a second relationship between the lumped average relative magnitude frequency and a reduction in percent of reducing agent relative to a normal supply of the reducing agent; and adjusting said reduction of reducing agent relative to said normal supply in accordance with said second relationship.

2. The method according to claim 1, wherein a moving median is utilized as the median tail pipe nitrogen oxide concentration (x.sub.s), which takes into account a series of predetermined periods of time.

3. The method according to claim 1, wherein a lower lumped average relative magnitude frequency threshold above zero and an upper lumped average relative magnitude frequency threshold are defined, and an amount of reducing agent to be supplied to the exhaust gas is varied if the total is below the lower threshold or above the upper threshold.

4. The method according to claim 3, wherein the lower lumped average relative magnitude frequency threshold is between 0.1 and 1, and the upper lumped average relative magnitude frequency threshold is between 0.9 and 1.5.

5. The method according to claim 4, wherein the lower lumped average relative magnitude frequency threshold is between 0.3 and 0.8, and the upper lumped average relative magnitude frequency threshold is between 1 and 1.4.

6. The method according to claim 4, wherein the lower lumped average relative magnitude frequency threshold is between 0.4 and 0.6, and the upper lumped average relative magnitude frequency threshold is between 1.2 and 1.3.

7. The method according to claim 3, wherein the amount of the reducing agent to be supplied to the exhaust gas is varied by a PID controller.

8. The method according to claim 1, wherein receiving the signal reading from the nitrogen oxide sensor occurs repeatedly in a range of 5 seconds to 60 seconds.

9. The method according to claim 8, wherein receiving the signal reading from the nitrogen oxide sensor occurs repeatedly in the range of 10 seconds to 30 seconds.

10. The method according to claim 8, wherein receiving the signal reading from the nitrogen oxide sensor occurs repeatedly for 20 seconds.

11. The method according to claim 1, wherein the total data sampling time period is in a range of 2 minutes to 20 minutes.

12. The method according to claim 1, wherein the total data sampling time period is in a range of 5 minutes to 15 minutes.

13. The method according to claim 1, wherein the total data sampling time period is 10 minutes.

14. The method according to claim 1, wherein the exhaust gas system is of a vehicle.

15. A control assembly for operating an exhaust gas system comprising: an injector to supply an amount of a reducing agent to exhaust gas of the exhaust gas system of an engine, and a controller for executing stored instructions to: receive a signal reading from a nitrogen oxide sensor over a total data sampling time period (), wherein the nitrogen oxide sensor measures a content of nitrogen oxides in exhaust gas flowing downstream of a catalytic device configured to diminish the content of nitrogen oxides in the exhaust gas, determine a lumped average a relative magnitude frequency (LARMF) by way of a first relationship, $\mathrm{LARMF}=\frac{1}{}{}_0^{}.Math.x_i-x_s.Math.dt;$ establish a second relationship between the lumped average relative magnitude frequency and a reduction in percent of reducing agent relative to a normal supply of the reducing agent; and adjust said reduction of reducing agent relative to said normal supply in accordance with said second relationship.

16. The control assembly according to claim 15, wherein the exhaust gas system is of a vehicle.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 is a control assembly for operating an exhaust gas system, in which an evaluation unit evaluates signals of a NOx sensor to adjust the amount of a urea-water-solution supplied into the exhaust gas upstream of an SCR catalyst;

(2) FIG. 2 illustrates three curves indicating the NOx concentration downstream of the SCR catalyst for three different scenarios;

(3) FIG. 3 illustrates curves showing an example of the signal of the NOx sensor and a moving median for a scenario with insufficient supply of ammonia to the SCR catalyst;

(4) FIG. 4 illustrates a curve which shows total values of the differences between the signal and the moving median curve shown in FIG. 3;

(5) FIG. 5 illustrates curves showing a signal of the NOx sensor and the moving median for an aged SCR catalyst with normal ammonia supply;

(6) FIG. 6 illustrates a curve showing the absolute values of differences between the signal and the moving median curve of FIG. 5;

(7) FIG. 7 is a qualitative relationship between a quantity derived from curves like those shown in FIG. 4 and FIG. 6 and an insufficient supply of ammonia to the exhaust gas;

(8) FIG. 8 illustrates zones for a closed-loop ammonia dosing control strategy;

(9) FIG. 9 is a schematic view of a logic structure of the closed-loop dosing control strategy; and

(10) FIG. 10 illustrates examples of dosing correction factors applied in the closed-loop dosing control.

DETAILED DESCRIPTION OF THE DRAWINGS

(11) FIG. 1 shows a control assembly 10 for operating an exhaust gas system of a vehicle. Exhaust gas from an engine (not shown) is treated in a catalytic device in the form of an SCR catalyst 12. In the SCR catalyst 12 ammonia (NH.sub.3) reacts with nitrogen oxides

(12) NOx in the exhaust gas in a selective catalytic reduction (SCR) reaction to form nitrogen and water. A reducing agent in the form of a urea-water solution is injected to the exhaust gas upstream of the SCR catalyst 12 by a dosing unit 14. The urea-water solution which is also called diesel exhaust fluid (DEF) releases the ammonia in a hydrolysis reaction after being mixed with the hot exhaust gas.

(13) Input signals 16 are provided to a controller 18 of the dosing unit 14 depending on engine operating parameters. The control assembly 10 is designed to detect insufficient ammonia supply to the SCR catalyst 12 and to closed-loop control the dosing of the diesel exhaust fluid or such a reducing agent into the exhaust gas.

(14) The closed-loop control is based on a magnitude-frequency analysis of signals or measurements provided by a NOx sensor 20 which measures the NOx concentration in ppm (part per million) in a tail pipe 22. The tail pipe 22 is a section of an exhaust gas pipe of the exhaust gas system, which is located downstream of the SCR catalyst 12. The analysis looks at the magnitude and frequency of the NOx sensor's 20 signal reading and, thus, of the measured NOx concentration downstream of the SCR catalyst 12.

(15) If the SCR catalyst's 12 ammonia supply is sufficient and within a normal range, the tail pipe 22 NOx sensor's 20 signal reading is generally low and smooth because of a good NOx conversion efficiency of the SCR catalyst 12. With the right amount of ammonia supplied to the SCR catalyst 12, not only the SCR catalyst 12 will have good NOx conversion, but more importantly, there will be extra ammonia stored in the SCR catalyst 12. This stored ammonia provides a damping effect by maintaining good conversion when there is a sudden increase of inflowing NOx or when there are sudden NOx fluctuations. The stored ammonia thus maintains a smooth tail pipe 22 NOx output. The more ammonia that is stored in the SCR catalyst 12 the better is the damping effect and the smoother the tail pipe NOx output.

(16) If there is a shortage in the amount of ammonia supplied to the SCR catalyst 12, the level of the stored ammonia is reduced. If short supply gets severe enough, there is no stored ammonia left in the SCR catalyst 12 at all. Under such conditions, due to the reduced damping capacity, with an inflowing NOx variation, the tail pipe NOx sensor's 20 signal reading becomes noisy. Magnitude and frequency of the noise increases with the increased degree of shortage in ammonia supply.

(17) FIG. 2 shows a graph 24 in which an ordinate 26 indicates the NOx concentration in ppm in the tail pipe 22. On an abscissa 28 the time in seconds is indicated. A first curve 30 shows the measurements provided by the NOx sensor 20 when the SCR catalyst 12 is comparatively new or fresh and the ammonia supply is within a normal range. According to the curve 30 the tail pipe 22 NOx level is low and the signal is smooth. A second curve 32 shows the tail pipe 22 NOx concentration for an aged SCR catalyst 12 with normal ammonia supply. The tail pipe 22 NOx level is overall high but the signal is still relatively smooth. A third curve 34 shows the tail pipe 22 NOx signal if the ammonia supplied to the fresh SCR catalyst 12 is reduced. Accordingly, the tail pipe 22 NOx signal becomes both higher and noisier.

(18) This general behavior of the NOx sensor's 20 signal reading is utilized to detect the SCR catalyst's 12 ammonia supply condition. In other words the tail pipe 22 NOx sensor 20 signal's magnitude and frequency characteristics are analyzed to detect underdosing or overdosing of ammonia into the exhaust gas. The detected magnitude-frequency analysis result is utilized to adjust the dosing control in a closed-loop approach accordingly and brings it back to a normal level if ammonia supply deviates from a normal level to either the rich or the lean side.

(19) Thus, the reducing agent supply can be changed to a limit without notifying a driver of the vehicle. The reason is that insufficient ammonia supply could be partially caused by a hardware malfunction which could lead to increased warranty costs if investigated, even though the malfunction is tolerable. If the supply increases above the limit, the driver can be notified, for example by a check engine light. The limit is based on the degree of tolerable malfunction of the exhaust gas aftertreatment system without damaging the system hardware.

(20) A measurement tool for evaluating the level or amount of ammonia supplied to the SCR catalyst 12 is an integrated quantity which is called lumped average relative magnitude-frequency or in short LARMF. The LARMF is expressed below:

(21) $\mathrm{LARMF}=\frac{1}{}{}_0^{}.Math.\left(x_i-{\tilde{x}}_s\right).Math.\mathrm{dt}$

(22) where x.sub.i=the tail pipe NOx sensor's 20 signal at data sampling time point i x.sub.s=the signal's moving median at time i for s sampling time period =the total time period of data sampling.

(23) The LARMF quantity uses relative magnitude-frequency analysis instead of absolute magnitude-frequency analysis. Specifically the measurement x.sub.i is related to a quantity x.sub.s, which is the moving median at time point i for a predetermined period of time or sampling time period s. The utilization of this relative magnitude-frequency analysis results in a high detection resolution with either a fresh SCR catalyst 12 or an aged SCR catalyst 12.

(24) The moving median x.sub.i is a virtual signal. The moving median x.sub.i depends on the length of the predetermined period of time and thus on the number of measurements taken within this period of time. The moving median x.sub.i can for example be the mathematic average of s data points or measurements which are symmetrically arranged around the measurement x.sub.i. The moving median calculation is arbitrary in the sense that the median value depends on how many measurements are used, i.e., how the predetermined period of time is selected. A larger predetermined period of time makes LARMF bigger and vice versa. If the sampling time period is too small, LARMF will be too small to reliably detect insufficient dosing of ammonia to the exhaust gas.

(25) However if the predetermined period of time is too long and thus s is too large, this may lead to a LARMF value that is too big, if the SCR catalyst 12 is aged. A proper selection of the sampling period of time and thus the median averaging the measurements within this period of time is important in order to have both good ammonia supply level detection resolution and good separation from the impact of aging on the SCR catalyst 12. In other words, a good selection of the predetermined period of time and thus the number of measurements taken within this period of time should result in a LARMF value which is large enough to detect an ammonia supply insufficiency, but which is still small enough, if an aged SCR catalyst 12 is monitored. The relative properties of the LARMF value include the utilization of an integration function which is constructed by the difference between the signal or measurement x.sub.i and its moving median x.sub.s instead of the signal's absolute magnitude.

(26) In another graph 36 in FIG. 3 again the tail pipe 22 NOx concentration in ppm is shown on the ordinate 26 and the time in seconds on the abscissa 28. A curve 38 shows the signal x.sub.i and a curve 40 is a plot of the corresponding moving median x.sub.s.

(27) FIG. 4 shows another graph 42 in which a curve 44 visualizes the integration function, i.e., the relative magnitude-frequency term utilized in the LARMF equation. This integration function represents the absolute values of the differences between the measurement x.sub.i and the moving median x.sub.s. The LARMF value is than a sum or total of all these absolute values collected over the sampling period .

(28) For an aged SCR catalyst 12 the tail pipe 22 NOx signal's absolute magnitude is high but relatively smooth, if the SCR catalyst 12 is supplied with sufficient ammonia. Thus, the difference between the signal x.sub.i and its moving median x.sub.s will be very small, although the signal's absolute magnitude is high. As a consequence the integration function in the LARMF equation is very small too, and the resulting LARMF value is also very small.

(29) Thus, an aged SCR catalyst 12 with a normal amount of ammonia supplied to the exhaust gas will not be mistakenly detected as an ammonia short supply case since the LARMF value is very small.

(30) A corresponding scenario is illustrated in FIG. 5 and FIG. 6. A graph 46 in FIG. 5 shows a plot in the form of a curve 48 representing the NOx tail pipe 22 concentration measurement of an aged SCR catalyst 12. Another curve 50 represents the corresponding moving median. Even though the absolute magnitude of the signals represented by the curve 48 is rather high, the signal is very smooth and does almost not deviate from the moving median. This is visualized in another graph 52 in FIG. 6, in which a curve 54 indicates the integration function.

(31) For a fresh SCR catalyst 12 with normal ammonia supply the tail pipe 22 NOx signal is low and smooth. However, even if the signals absolute magnitude changes and is high or low, this has little impact on the LARMF value. On the other hand the smoothness and thus the frequency of the signal does have an impact on the LARMF value. As a consequence as long as the ammonia supply is normal both a fresh and an aged SCR catalyst 12 will exhibit a smooth tail pipe 22 NOx signal. Hence, both SCR catalysts 12 will result in a small LARMF value.

(32) There is a correlation between the LARMF value and the ammonia supply level which can be established through experimental tests. In general a higher degree of ammonia insufficiency corresponds to a larger LARMF value. This general relationship is illustrated by FIG. 7. The LARMF value is indicated on an ordinate 56, whereas a reduction of the ammonia supply compared to a normal supply in percent is indicated on an abscissa 58.

(33) A correlation curve 60 illustrates the aforementioned relationship.

(34) With reference to FIG. 1 an evaluation unit 62 of the control assembly 10 includes a calculation unit 64 which determines the LARMF value based on the NOx signals or measurements provided by the NOx sensor 20. The evaluation unit 62 also includes a dosing compensation logic 66. With this dosing compensation logic 66 the evaluation unit 62 provides dosing adjustment information to the controller 18 of the dosing unit 14. Thus, a closed-loop control utilizing the LARMF value is established. With this closed-loop control even under tightened emission regulations and OBD (on-board diagnosis) regulations an optimal ammonia supply may be maintained which is critically important to assure compliance with both emission and OBD regulations.

(35) The theoretical minimum LARMF value is zero. However the control assembly 10 should be able to avoid an overdosing scenario. With a LARMF value near zero it is difficult to tell, whether the ammonia is oversupplied or not. Therefore, the LARMF value is preferably kept within a certain range above zero. This allows assuring that the ammonia dosing stays optimal.

(36) For example, the LARMF based closed-loop dosing control may include three zones as shown in FIG. 8. A first dosing holding zone 68 is bordered by an upper threshold 70 or high LARMF limit and a lower threshold 72 or low LARMF limit. If the LARMF value is above the upper threshold 70 and therefore greater than the holding zone 68 high limit, a positive ammonia dosing adjustment is fed to the controller 18. The dosing adjustment can in particular be incremental. A positive dosing change will push the LARMF value gradually into the holding zone 68 and the controller 18 will then hold to whatever the dosing level is. If the LARMF value is below the lower threshold 72 and thus smaller than the holding zone 68 low LARMF limit, ammonia dosing will be reduced to push the LARMF to the holding zone 68. The LARMF based closed-loop ammonia dosing control strategy thus includes an increasing dosing zone 74 and a decreasing dosing zone 76 (see FIG. 8). The LARMF value target zone or holding zone 68 can in particular be between 0.5 and 1.25.

(37) The LARMF based closed-loop reducing agent dosing control logic structure is shown schematically in FIG. 9. A LARMF target 78 corresponds to the holding zone 68 LARMF values, i.e., is arranged between the upper threshold 70 and the lower threshold 72. A PID controller 80 makes adjustment to the reducing agent dosing based on differences between the target value and the actual LARMF values until the actual LARMF values fall in the holding zone 68. The input signals 16 correspond to an open-loop reducing agent dosing. From the signals measured by the NOx sensor 20, variations 82 or changes in the NOx concentrations downstream of the SCR catalyst 12 are detected. Based on these measurements the calculation unit 64 determines the LARMF value and provides adjustment information.

(38) FIG. 10 illustrates the expected behavior for ammonia supply correction factors of LARMF-based reducing agent dosing with closed-loop under different ammonia supply scenarios. The correction factor is indicated on an ordinate of a graph 86 in FIG. 10, whereas the time in seconds is indicated on an abscissa 88. In a first scenario 90 an increasing correction factor is applied due to underdosing of ammonia. In a second scenario 92 the correction factor is reduced again as normal ammonia dosing is attained. In a third scenario 94 of ammonia overdosing the correction factor is further decreased to values below 1. Thus, normal ammonia dosing has a correction factor of 1, underdosing will result in a correction factor greater than 1, and overdosing in a correction factor below 1.

(39) A total time for the sampling period to determine the LARMF value can in particular be about 10 minutes. Within this total sampling period the predetermined periods of time during which the moving median is calculated can be about 20 seconds. Thus, after 10 minutes of data sampling a LARMF value is generated and the dosing is adjusted accordingly until a new LARMF value is calculated and so on.

(40) Data sampling preferably takes places if there are approximately steady state operating conditions. This avoids a situation where highly transient operating conditions exist which would have an unwanted influence on the detection accuracy.

LIST OF REFERENCE SIGNS

(41) 10 control assembly 12 SCR catalyst 14 dosing unit 16 input signals 18 controller 20 NOx sensor 22 tail pipe 24 graph 26 ordinate 28 abscissa 30 curve 32 curve 34 curve 36 graph 38 curve 40 curve 42 graph 44 curve 46 graph 48 curve 50 curve 52 graph 54 curve 56 ordinate 58 abscissa 60 correlation curve 62 evaluation unit 64 calculation unit 66 dosing compensation logic 68 holding zone 70 upper threshold 72 lower threshold 74 increasing dosing zone 76 decreasing dosing zone 78 target 80 PID controller 82 variations 84 ordinate 86 graph 88 abscissa 90 scenario 92 scenario 94 scenario