Method and apparatus for determining whether a sensor which is responsive to both a reactant and a substance to be reduced by such reactant is responding to either un-reacted portions of the substance or un-reacted portions of the reactant

Abstract

A method wherein a reactant is added to a substance to react with such substance. The product of such reaction along with un-reacted portions of the substance and un-reacted portions of the reactant are directed to a sensor. The sensor produces an output signal in response to detection of both the un-reacted portions of the substance and the un-reacted portions of the reactant. The method includes changing the amount of reactant added to the substance. A measurement is made to determine whether the change in the amount of reactant and the change the output signal are in the same direction or in opposite directions. The changes in reactant and output signal are multiplied an integrated to form a correction on the quantity of reactant to be injected.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application is a divisional application of co-pending U.S. patent application Ser. No. 09/682,445 entitled “Method and Apparatus for Controlling the Amount of Reactant to be Added to a Substance Using a Sensor which is Responsive to both the Reactant and the Substance”, filed on Sep. 4, 2001, the entire subject matter thereof being incorporated herein by reference. This application claims the benefit of the Sep. 4, 2001 filing date of said co-pending U.S. patent application Ser. No. 09/682,445 under the provisions of 35 U.S.C. §120.

TECHNICAL FIELD

[0002] This invention relates generally to methods and apparatus for determining whether a sensor which is responsive to both a reactant and a substance to be reduced by such reactant is responding to either un-reacted portions of the substance or un-reacted portions of the reactant. More particularly the invention relates to methods and apparatus for detecting NOx to be reduced with urea using a sensor which is responsive to un-reacted portions of the NOx and un-reacted portions of the urea.

BACKGROUND

[0003] As is known in the art, in many applications it is desirable to detect the effectiveness of a reaction used to reduce a substance. One such application is in measuring the effectiveness in urea based selective catalytic reduction (SCR) in reducing nitrogen (NOx) in the exhaust gas of a diesel engine. More particularly, an aqueous solution of urea is injected into the exhaust gas of the engine upstream of a catalyst. In order for the method to reduce NOx in the exhaust effectively, it is important that the amount of urea injected into the exhaust be accurately controlled. Injection of too little urea may result in sub-optimal (i.e., incomplete) NOx conversion. Injection of too much urea may produce nitrates in the exhaust which can reduce the life of the exhaust system downstream of the catalyst, may produce an unpleasant odor, and may also produce increases in regulated emissions.

[0004] Thus, it is desirable to have a sensor downstream of the catalyst which can detect the presence of NOx after the reaction. However, the inventors herein have recognized that currently available sensors which are practical, from a size and cost perspective, for automotive use cannot differentiate between NOx and urea. The inventors have further recognized that degraded performance may result when using such a sensor with conventional urea injection methods that adjust the injected urea based on one of measured NOx or measured urea,. In particular, if it is assumed that NOx is measured, yet actually urea is exiting the catalyst, the system may improperly increase injected urea since the system believes excess NOx is being produced and thus more urea is necessary. This can results in increased urea slip. Similarly, if it is assumed that urea is measured, yet actually NOx is exiting the catalyst, the system may improperly reduce injected urea since the system believes excess urea is being produced and thus less urea is necessary. This can result in increased NOx emission.

SUMMARY

[0005] In accordance with the present invention, a method is provided wherein a reactant is added into a substance to react with the substance. The product of such reaction along with un-reacted portions of the substance and un-reacted portions of the reactant are directed to a sensor. The sensor produces an output signal in response to detection of both the un-reacted portions of the substance and the un-reacted portions of the reactant. The method includes changing at least one of the amount of reactant added to the substance and the amount of the substance. A determination is made from said at least one of the change in the amount of reactant and the change in the amount of the substance, and from the change in the output signal, whether the sensor is responding to the substance or to the reactant.

[0006] In one embodiment, a method is provided wherein a reactant is added to a substance to react with such substance. The product of such reaction along with un-reacted portions of the substance and un-reacted portions of the reactant are directed to a sensor. The sensor produces an output signal in response to detection of both the un-reacted portions of the substance and the un-reacted portions of the reactant. The method includes changing the amount of reactant added to the substance and determining from the response of the sensor to such change in the amount of reactant whether the output signal is responding to the substance or to the reactant.

[0007] In accordance with another aspect of the invention, a sensing system is provided wherein a sensor is provided for producing an output signal in response to either a first substance of a reaction or to a second substance of the reaction. The system includes: an injector for changing the level of one of the substances in the reaction; and a processor responsive to the output signal and a signal representative of the change in level of one of the substances, for determining whether the sensor is responding to the first substance or to the second substance.

[0008] In one embodiment, the sensing system is a NOx sensing system. Such system includes: a NOx sensor for producing an output signal in response to NOx and urea and a processor responsive to the output signal for determining whether the sensor is responding to NOx or to urea.

[0009] In accordance with another aspect of the invention, a method is provided wherein a reactant is added to a substance to react with such substance. The product of such reaction along with un-reacted portions of the substance and un-reacted portions of the reactant are directed to a sensor. The sensor produces an output signal in response to detection of both the un-reacted portions of the substance and the un-reacted portions of the reactant. The method includes changing the amount of reactant added to the substance, determining whether the change in the amount of reactant and the change in the output signal are in the same direction or in opposite directions, and from such determination determining whether the output signal is responding to the substance or to the reactant.

[0010] In one embodiment, if the directions are the same, reactant is added to the substance. If the directions are opposite, reactant is reduced from the substance.

[0011] In one embodiment of the invention, a change is made in the amount of a reactant added to a substance to be reacted and thereby reduced by, the reactant, the time history of such change in the reactant being represented by:

[0012] Û(t)

[0013] where t is time.

[0014] The products of the reaction, along with un-reacted portions of the reactant and the un-reacted portions of the substance are passed by a sensor. The sensor is unable to effectively differentiate between the un-reacted reactant portions and the un-reacted portions of the substance. That is, the sensor produces an output signal V(t) in response to the un-reacted portions of the reactant and the un-reacted portions of the substance. The change in the output signal from the change in the amount of reactant added to the substance may be represented as:

[0015] {circumflex over (V)}(t)

[0016] The change in the amount of the reactant is recorded over a period of time T. The delay between the change in the amount of the reactant occurring at a time t0, to the time the reaction is effected by such change in the amount of reactant is a time delay, D. A signal, I, is produced to determine whether the output signal increases or decreases with the changed amount of reactant. In one embodiment, 1 I = ∫ t0 t0 + D + T ⁢   ⁢ e ⁢   ⁢ x ⁢   ⁢ c ⁢ V ^ ⁡ ( t + D ) ⁢ u ^ ⁡ ( t ) ⁢ ⅆ t

[0017] where Texc is the time duration of the excitation û(t).

[0018] If the change in the level of the reactant is in the same sense (i.e., direction) as the resulting change in the output signal, (i.e., if the amount of reactant increases and the output signal increases, or if the amount of reactant decreases and the output signal decreases), I>0. In such case, the sensor is determined to be responding to the un-reacted reactant.

[0019] On the other hand, if the change in the level of reactant is in the opposite to resulting change in the output signal, (i.e., if the amount of reactant increases and the output signal decreases, or if the amount of reactant decreases and the output signal increases), I<0. In such case the sensor is determined to be responding to un-reacted substance.

[0020] If I=0, there was no change in the reactant and therefore it may be concluded that the injector is faulty.

[0021] Further, if V is much greater than 0 (i.e., V is greater than some calibratable threshold), it may be concluded that both the un-reacted substance, and un-reacted reactant are both being measured and therefore a catalyst used to facilitate the reaction is determined to be faulty.

[0022] Thus, from the signal I, a determination may be made as to whether the sensor is responding to the un-reacted reactant, to the un-reacted substance, and whether the injector or the catalyst is faulty.

[0023] The details of one or more embodiments of the invention are set forth in the accompanying drawings and the description below. Other features, objects, and advantages of the invention will be apparent from the description and drawings, and from the claims.

DESCRIPTION OF DRAWINGS

[0024] FIG. 1 is a diagrammatical sketch of an engine exhaust system according to the invention having a processor programmed for determining the effectiveness of injected urea in reducing NOx produced by the engine;

[0025] FIG. 2 is a diagram showing the relationship between an output signal produced by a sensor used in the exhaust system of FIG. 1 as a function of either nitrogen monoxide (NO), nitrogen dioxide (NO2) or urea;

[0026] FIG. 3 is a diagram showing the output signal produced by the sensor used in the exhaust system of FIG. 1 in response to increases and decreases in the amount of urea injected into the engine exhaust in the system of FIG. 1; and

[0027] FIG. 4 is a flow diagram of the process used by the processor in FIG. 1 to determine the effectiveness of injected urea in reducing NOx produced by the engine exhaust in the system of FIG. 1;

[0028] FIG. 5 is a functional block diagram of a NOx reduction system according to another embodiment of the invention;

[0029] FIG. 6 are diagrams showing as a result of computer simulations various parameters produced in the system of FIG. 5 for an assumed Urea to NOx ratio correction factor of +0.32; and

[0030] FIG. 7 are diagrams showing as a result of computer simulations various parameters produced in the system of FIG. 5 for an assumed Urea to NOx ratio correction factor of −0.3.

[0031] Like reference symbols in the various drawings indicate like elements.

DETAILED DESCRIPTION

[0032] Referring now to FIG. 1, is a diagrammatical sketch of an engine exhaust system 10 is shown having a processor 12 programmed for determining the effectiveness of injected urea in reducing NOx produced by an engine 14. The system 10 includes a reactant injector 16, here a urea injector, adapted to inject urea into the exhaust 18 of the engine in response to an excitation signal u(t), where t is time. The injection signal u(t) is made up of two components: a nominal level u0 and a changeable component û(t). The nominal level u0 may be produced by providing a sensor 13 in the engine exhaust upstream of the injection of the urea. Thus, here u0=k_base*NOx1, where NOx1 is the signal produced by this upstream sensor and k_base is a conversion factor stored in a look-up table 15 relating the a priori assumed stoichiometric amount of urea needed to convert the NOx in the upstream engine exhaust.

[0033] The exhaust 18, together with the injected urea, is directed to a catalyst 20 though the engine's exhaust pipe. The catalyst 20 is used to facilitate the reaction between the NOx and the urea. The injected urea reacts with NOx which may be present in the exhaust 18 in the catalyst 20 to produce a product 22 which includes the reaction products as well as any un-reacted portions of the injected urea or un-reacted portions of the NOx.

[0034] As will be described, the processor 12 is also programmed to determine whether the injector 16 is faulty and whether the catalyst 20 is faulty.

[0035] More particularly, the output of the catalyst 22 is sensed by a second NOx sensor 26. While it would be desirable that the sensor 26 sense only the presence of any NOx at the output of the catalyst, current practical automotive NOx sensors, as noted above, produce an output signal in response to NOx and urea. Thus, these sensors cannot differentiate between NOx and urea. That is, the sensor 26 is a non-selective NOx sensor. This is illustrated from FIG. 2. Curve 30 shows the relationship, i.e., transfer function, between sensor 26 output signal (V, volts) and measured urea in the presence of only urea. It is noted that the slope of curve 60 is here k_urea. Curve 32 shows the relationship, i.e., transfer function, between sensor 26 output signal (V, volts) and measured NO in the presence of only NO. It is noted that the slope of curve 60 is here k_no Curve 34 shows the relationship, i.e., transfer function, between sensor 26 output signal (V, volts) and measured NO2 in the presence of only NO2. It is noted that the slope of curve 60 is here k13 no2. (It is noted that the response of the sensor to NOx is interpolated from the responses to NO and NO2. For example a 50-50 mixture of NO and NO2 would result in the slope (k_no+k_no2)/2. The family of interpolated slopes is collectively referred to as k_nox.

[0036] We have discovered a method which enables differentiation between NOx and urea detection using theses practical NOx non-selective sensors. More particularly, we have discovered that a urea excitation technique to be described allows one to determine whether the non-selective sensor 26 is responding to NOx or to urea. According to the method, the amount of urea injected into the engine exhaust 18 is modulated, or changed in a particular, a priori known direction, or sense. The output of the non-selective NOx sensor 26 is processed by processor 12 along with knowledge of the direction in the change in the amount of injected urea. The result of the processing yields an indication of whether the sensor is sensing urea or NOx by determining whether the change in urea injection and the change in sensor 26 output signal are in the same direction or in opposite directions.

[0037] Here, in the system shown in FIG. 1, a small, periodic negative amplitude excitation signal û(t) of duration Texc is superimposed on the excitation signal u0 in a summer 28. The summer 28 produces u(t) which provides the excitation signal for the injector urea injector 16. Since during these negative excursions in the excitation signal u(t) the amount of urea is reduced, less NOx is being reduced, if any. If the sensor 26 output signal, here a voltage V(t), goes lower in response to the negative excursion in the excitation signal u(t), too much urea is being injected into the exhaust 18 and the sensor 26 output signal change {circumflex over (V)}(t) was due to un-reacted urea. If, on the other hand, the sensor 26 produces a higher voltage V(t), not enough urea was being injected into the exhaust to reduce all of the NOx and the sensor output signal change {circumflex over (V)}(t) was due to un-reacted NOx. Thus, if we reduce the amount of urea and the voltage V produced by the sensor reduces, (i.e., the direction of the change in urea is in the same direction in the change in the output signal, V(t), of the sensor) the sensor 26 is now known to be responding to urea. On the other hand, if we reduce the amount of urea and the voltage V(t) produced by the sensor increases, (i.e., the direction of the change in urea is in the opposite direction to change in the output signal, V(t), of the sensor) the sensor 26 is now known to be responding to NOx. FIG. 3 illustrates this effect.

[0038] The processor 12 (FIG. 1) is provided to detect the direction of the change in injected urea u(t) relative to the change in the output signal V(t) of the sensor 26. That is, the processor determines whether the direction of the change in output signal V(t) of the sensor 26 is the same as, or opposite to, the direction of the change in injected urea u(t). As described above, if they are the same direction, the sensor 26 is detecting urea whereas if the directions are opposite one another the sensor 26 is detecting NOx. Here such determination is made by multiplying a signal representative of {circumflex over (V)}(t) (i.e., the change in the output signal V(t) produced by the sensor 28) with the change in the excitation signal û(t). Thus, the change in the output of the sensor, {circumflex over (V)}(t), is proportional to the change in the output signal of the sensor V(t). Here, the determination in {circumflex over (V)}(t) is made when the system is in a steady-state condition which can be checked by monitoring rate of change of rpm, load, space velocity, etc.

[0039] Thus, for time [t0-Tss, t0], we determine the steady-state output voltage, Vss= 2 ∫ t0 - T ⁢   ⁢ s ⁢   ⁢ s t ⁢   ⁢ o ⁢ V ⁡ ( t ) ⁢ ⅆ t

[0040] Thus, during time from t0, to t0+D+Texc:

{circumflex over (V)}(t)=V(t)−Vss,

[0041] Where Vss is the steady state output voltage of the sensor 28 and where the excitation is repeated every Ttot seconds, where Ttot>D+Tss+Texc where Tss is used to take into account the time used to determine Vss.

[0042] If the mathematical product of {circumflex over (V)}(t) and û(t) is positive, the sensor 26 is determined by the processor 12 to be detecting urea. If, on the other hand, the mathematical product is negative, the sensor 26 is determined by the processor 12 to be detecting NOx. In FIG. 1, the multiplication is shown by a multiplier 41 which is fed {circumflex over (V)}(t) produced by the high pass filter 47 (i.e., a signal proportional to the change in sensor output signal) and by a signal representative of the change in the urea excitation, i.e. û(t). It should be understood that the processor 12 is preferably a digital processor which performs the process shown in FIG. 4.

[0043] It should be noted that due to adsorption and desorption of urea in the catalyst 20 and reaction kinetics the time history profile of the voltage V(t) produced by the sensor 26 will not be a rapidly changing, pulse, but more a low pass filtered version of a pulse. Further, due to transport delay (i.e., the delay between the time the urea is injected into the exhaust 18 and the time the of reaction in the catalyst 20), the voltage V(t) produced by the sensor will be delayed an amount D from the commencement of the change in the injection of the urea. The delay D may be mapped a priori as a function of engine operating conditions, e.g., engine speed using a look-up table.

[0044] Here, the output of the multiplier 41 is integrated in the processor 12 as represented by integrator 43. Thus, if the pulse-width of the change in the urea excitation signal û(t) has a pulse width Texc and starts at a time t0, the output of the integrator 43 may be represented as: 3 I = ∫ t0 t0 + D + T ⁢   ⁢ e ⁢   ⁢ x ⁢   ⁢ c ⁢ V ^ ⁡ ( t + D ) ⁢ u ^ ⁡ ( t ) ⁢ ⅆ t

[0045] Thus, if I>0, (i.e., if the integrated mathematical product of {circumflex over (V)}(t) and û(t) is positive), the sensor 26 is detecting urea. Here, such condition is indicated by a logic 1 produced by comparator 44. Thus, here the amount of urea is V(t)*k_urea

[0046] If, on the other hand, I<0 (i.e., the integrated mathematical product of {circumflex over (V)}(t) and û(t) is negative), the sensor 26 is detecting NOx. Here, such condition is indicated by a logic 1 produced by comparator 46. Thus, here the amount of NOx is V(t)*k_nox

[0047] Having discriminated between sensing urea and sensing NOx, if I<0, the sensor 26 output signal V(t) can be fed to the transfer function shown by curves 32 or 34 in FIG. 2 and a measurement of the NOx may be obtained. If, one the other hand, I≦0, the sensor 26 output signal V(t) can fed to the transfer function shown by curve 30 in FIG. 2 and a measurement of the urea may be obtained.

[0048] Here, as noted above, the excitation is repeated every Ttot seconds, where Ttot>D+Texc+Tss.

[0049] It should be noted that if I=0, the commanded change in urea was not applied and a diagnosis is that of a faulty injector. Here, such condition is indicated by a logic 1 produced by comparator 48.

[0050] If V>>0 (i.e., greater than some calibrate threshold), one may conclude that both the engine exhaust NOx and the urea are still measured after the catalyst 20 and one can conclude a faulty catalyst 20. Here, such condition is indicated by a logic 1 produced by comparator 50.

[0051] It should be noted that since here we decrease the amount of urea for a short time, an increase in tail pipe emissions may result. This is the case if I<0. This can be compensated by applying a positive correction+û after T0+T1+T2, when I<0. If I>0, too much urea is being injected and a positive correction is not required. This procedure keeps the overall tailpipe emissions neutral.

[0052] It should also be noted that the procedure works best in the steady state, when, as noted above, it is easy to determine {circumflex over (V)}(t), by subtracting the mean over the previous period. If knowledge of engine emissions and catalyst behavior is very accurate, the method may be applied to transients. In this case, it may be able to gain further information about the system catalyst conditions, urea injection system, and NOx sensor by measuring the transient sensor response.

[0053] Referring now to FIG. 4, the processor is programmed in accordance with the flow diagram shown therein. Thus, a determination is made as the open loop quantity of urea u0 to be added. Next, a negative excitation signal is applied to the urea injector for T2 seconds. A measurement is made as to the change in the sensor 26 output signal. The signal I is computed as described above. If I<0, the NOx transfer function is used and a positive, increase, is made in the amount of urea. If, on the other hand, I<0, the urea transfer function is used.

[0054] Having described a method and system for determining whether the sensor 26 is responding to NOx or Urea, a method and system using these techniques will be described which controls the amount of urea to be injected into the engine exhaust to produce correct stoichiometric urea and diagnostics.

[0055] Referring now to FIG. 5, a functional block diagram is shown of a NOx reduction system 10′. Here a NOx sensor 60 is included upstream of the point 19 where urea is added into the engine 14 exhaust 18 via injector 16. The NOx sensor 26 is disposed downstream of the catalyst 20, as shown. As noted above in connection with FIG. 1, the NOx sensors 26 and 60 are sensitive to both urea and NOx. The catalyst 20 is positioned between the point 19 where the urea is injected and the position of the NOx sensor 26. Output signals nox1, nox2 (where nox2 is referred to as V in connection with FIGS. 1-4) produced by sensors 60, 26 respectively are processed by a programmed processor 12′, in a manner to be described, to produce the urea injection signal to urea injector 16. It is noted that part of the output nox2 is due to urea slip.

[0056] Here, the processor 12′ is shown by a functional block diagram for purposes of understanding the signals processed by such processor 12′. It should be understood that preferably the processor 12′ is a digital processor programmed to execute an algorithm to be described below. Suffice it to say here that the processor 40′ includes a square wave generator 62 which produces an excitation voltage u_exc (where u_exc was referred to as û(t) in connection with FIGS. 1-4) which may be represented as:

u—exc:=A—du*k—nox/k—urea if t<0.5*T—du

u—exc:−=A—du if t>0.5*T—du;

[0057] where:

[0058] A_du is the amplitude of the negative portion of the steady state excitation signal fed to the urea injector 16 (i.e., a negative excitation signal reduces the amount of urea injected into the engine exhaust);

[0059] T_du is the period of the excitation signal u_exc; (T_du=Texc×2).

[0060] k_urea is the sensitivity of the NOx sensor 26 with respect to urea;

[0061] k_NOx is the sensitivity of the NOx sensor 26 with respect to NOx.

[0062] Thus, a correction factor k_nox/k_urea is inserted to account for different sensitivities of the sensor 26 to urea and NOx.

[0063] The output of the square wave generator 62 is fed, via a switch 64, to a low pass filter 66. The function of the low pass filter 66 is to take into account a reaction delay, D, between the time urea is injected into the exhaust and the reaction in the catalyst 20. The processor 40′ provides this low pass filter 66 digitally in accordance with:

u—rk:=kf—rk*u—rk+(1−kf—rk)*u—exc,

[0064] where:

[0065] u_rk is the output of the low pass filter 66;

[0066] kf_rk is time constant of NOx-urea reaction kinetics.

[0067] The output signal produced by the sensor 26 is here, as noted above, represented as: nox2. The signal nox2 is fed to a high pass filter 42 to produce an output signal nox2_hp representing the change in the output signal produced by the sensor 26. Here, the processor 40 provides this high pass filter output in accordance with:

[0068] nox2_hp:=nox2−nox2_lp; (where nox2_hp was referred to as {circumflex over (V)}(t) in FIGS. 1-4).

[0069] where nox2_lp:=kf_lp_nox2*nox2_lp+(1−kf_lp_nox2)nox2;

[0070] where: kf_lp_nox2 is the filter gain for the low pass filter nox2.

[0071] The output of the high pass filter 42 (which represents the change in the output signal produced by the sensor 26) and the signal produced by the low pass filter 66 (which represents the change in the excitation signal u_exc to be fed to the urea injector 16, in a manner to be described, are fed to a multiplier 41. It should be noted that if the direction of change in the excitation signal u_rk is the same as the direction in the change in the sensor 26 signal, the mathematical product produced by the multiplier 41 will be positive. On the other hand, if the direction of change in the excitation signal u_rk is opposite to the direction in the change in the sensor 26 signal, the mathematical product produced by the multiplier 41 will be negative.

[0072] The mathematical product produced by the multiplier 41 is normalized by the signal nox1 produced by the sensor 60 in a divider 45 to produce:

dydu:nox−hp*u—rk/nox1

[0073] The signal dydu is scaled by ki and integrated by integrator 43 to produce a correction signal k_corr in accordance with:

k_corr:=∫(ki*dydu)dt;

[0074] where: ki>0 is the integral gain used to correct the nominal urea:NOx ratio (which is k_base)

[0075] Thus, if dydu>0 (i.e., the change in the sensor 26 and the change in the urea excitation are the in the same direction), the amount of urea should be reduced whereas, if dydu<0, (i.e., the change in the sensor 26 and the change in the urea excitation are the in opposite directions),the amount of urea should be increased.

[0076] The signal k_corr is used to add (if k_corr>0) or subtract from (if k_corr<0) the nominal a priori determined amount of urea for correct stoichiometry which is k_base*nox1. The actual amount of urea added by the injector is k_base*nox1*k_injector, where k_injector is the injector transfer function which is unknown because of aging, etc. Thus, with the processor 40, the signal k_corr together with u_exc will both modulate the signal k_base*nox1 to produce the correct stoichiometric urea independent of k_injector. and thus automatically adjust the amount of urea which should be added to the exhaust.

[0077] More particularly, the final urea quantity to be applied via the injector 16 is:

u—tot—ppm:=(k—base+k—corr)*nox1+u—exc;

[0078] where k_base is the nominal urea:NOx ratio.

[0079] To put it another way, a priori determined injection signal u0=nox1*k_base is modulated by both the correction signal k_corr*nox1 and the square wave signal u_exc. At correct stoichiometry, (k_corr+k_base)*k_injector*nox1 results in the injector 16 delivering stoichiometric urea to the engine exhaust upstream of the catalyst 20.

[0080] The signal u_tot_ppm is in parts per million of urea and is converted to mg/sec of urea by using mass air flow (Maf) of the exhaust, the upstream temperature (TMP) of the catalyst and the fuel flow (Wf) in a converter 61.

[0081] At the end of each period T_du, the integral int_dydu_last is evaluated in a comparator 47 as follows: 4 ∫ t - T ⁢   ⁢ _ ⁢   ⁢ d ⁢   ⁢ u t ⁢ ⅆ y ⁢ ⅆ u ⁡ ( t ) ⁢ ⅆ t

[0082] between the limits t−T_du and t, where t is the current time. The integral represents the incremental correction to the urea:NOx ratio accumulated over the last excitation period T_du. If int_dydu_last<k_dydu_thres, (where k_dydu_thres is the threshold for correction contribution to determine whether further adjustment is needed), this incremental correction is considered small enough to terminate adjusting the urea and the switch 64 is open from its initial closed condition. Otherwise, the switch 64 remains closed.

[0083] If k_corr>k_corr_lmx, where k_corr_lmx is the maximum limit for the correction factor k_corr to be declared a failure, (i.e., a blocked injector 16 or a catalyst 20 failure) as determined by comparator 47, a system failure has occurred (i.e., the catalyst 20 is inactive or the injector 16 is blocked).

[0084] If k_corr<k_corr_lmn<0, where k_corr_lmn is the minimum limit for the correction factor k_corr to declare a system failure for a leaking urea injector, as determined by comparator 49, a system failure has occurred, i.e., the injector is leaking, and excess urea will shorten the life of the exhaust system. For example, for detection of a 50 percent increase, k_corr_lmx is set=0.5.

[0085] Referring now to FIG. 6, such FIG. 6 shows, as a result of computer simulations various parameters produced in the system of FIG. 5 for a resulting Urea to NOx ratio correction factor of +0.32 (i.e., k_base=0.68) FIG. 7 shows, as a result of computer simulations various parameters produced in the system of FIG. 5 for a resulting Urea to NOx ratio correction factor of −0.3 (i.e., k_base=1.3).

[0086] The process described above in connection with FIG. 5 may be summarized as follows:

[0087] The following measured inputs are used:

[0088] Nox1:Nox sensor signal measured before the SCR brick (in ppm)

[0089] Nox2:Nox sensor signal measured after the SCR brick (part of this output is due to urea slip) (in ppm)

[0090] MAF:mass air flow

[0091] T1:temperature upstream of the SCR brick

[0092] Wf:fuel flow

[0093] (Maf, T1 and Wf are used to convert the urea ppm quantity to a quantity in mg/sec.)

[0094] The following gains are used and to be calibrated based on experimental data. They may dependent on engine operating conditions (speed and load) and exhaust temperature.

[0095] A_du:the amplitude of the negative part of the excitation

[0096] T_du:the period of the excitation is T_du

[0097] k_urea:sensitivity of the Nox sensor wrt urea

[0098] k_nox:sensitivity of the Nox sensor wrt urea

[0099] (A_du, k_urea and k_nox determine the amplitude of the positive part of the excitation).

[0100] kf_lp_nox2:filter gain for low pass filtered nox2.

[0101] kf_rk:time constant of nox-urea reaction kinetics.

[0102] k_base:nominal urea:nox ratio

[0103] ki:integral gain to correct the nominal urea:nox ratio

[0104] k_dydu_thres:threshold for correction contribution to determine whether further adjustment is needed.

[0105] k_corr_lmx:maximum limit for correction factor to declare an OBD failure (blocked injector or catalyst malfunction).

[0106] k_corr_lmn:minimum limit for correction factor to declare a system failure (leaking injector).

[0107] As noted above, the processor 12′ is a digital processor. The following flow diagram describes the program executed by such digital processor:

THE PROCESS FLOW

[0108] Step 1 The excitation signal is the repetitive signal with period T_du defined by:

[0109] u_exc:=A_du*k_nox/k_urea if t<0.5*T_du

[0110] u_exc:=−A_du if t>0.5*T_du

[0111] The correction factor k_nox/k_urea is inserted to account for different sensitivities to Nox and urea. These gains follow from sensor characteristics.

[0112] Step 2 Compute the switch on_logic (specified later).

[0113] Step 3 If on_logic=TRUE, apply u_exc, otherwise don't and return to 1.

[0114] Step 4 Compute the excitation signal delayed by reaction kinetics:

[0115] u_rk :=kf_rk*u_rk+(1−kf_rk)*u_exc.

[0116] Step 5 Obtain the measurement voltage from the second Nox sensor, nox2.

[0117] Step 6 Compute the low pass filtered version nox2_lp, with filter constant kf_lp_nox2:

[0118] nox2_lp:=kf_lp_nox2*nox2_lp+(1−kf_lp_nox2)*nox2.

[0119] Step 7 Compute the high pass filtered

[0120] nox2_hp:=nox2−nox2_lp.

[0121] Step 8 Obtain the measurement voltage from the first Nox sensor:nox1.

[0122] Step 9 Compute the convolution:

[0123] dydu:=nox2_hp*u_rk/nox1

[0124] Step 10 Integrate the convolution into a correction factor:

[0125] k_corr:=k_corr+ki*dydu

[0126] Step 11 At the end of every period T_du evaluate the integral

[0127] int_dydu_last:=ki*∫tt−T—dudydu(t) dt between the limits t−T_du and t, where t is the current time. This integral represents the incremental correction to the urea:nox ratio accumulated over the last excitation period. If int_dydu_last<k_dydu_thres, this incremental correction is small enough to stop adjusting, and set on_logic=false.

[0128] Otherwise on_logic is true.

[0129] Step 11 If

[0130] k_corr>k_corr_lmx

[0131] a system failure has occurred:the catalyst is inactive or the injection system is blocked.

[0132] If

[0133] k_corr<k_corr_lmn

[0134] a system failure has occurred: the injection system is leaking. The limit k_corr_lmx corresponds directly to OBD limits. For detection of a 50% increase in a regulated exhaust component, set k_corr_lmx=0.5.

[0135] Step 12 The final urea quantity to be applied is:

[0136] u_tot_ppm:=(k_base+k_corr)*nox1+u_exc.

[0137] Step 13 Use MAF, Wf, T1 to convert u_tot_ppm to a quantity in mg/sec: u_tot_mgsec.

[0138] Step 14 Go to Step 1.

[0139] A number of embodiments of the invention have been described. Nevertheless, it will be understood that various modifications may be made without departing from the spirit and scope of the invention. Accordingly, other embodiments are within the scope of the following claims.

Claims

1. A method, comprising:

(a) adding a reactant into a substance to react with the substance, the product of such reaction along with un-reacted portions of the substance and un-reacted portions of the reactant being directed to a sensor, such sensor producing an output signal in response to detection of both the un-reacted portions of the substance and the un-reacted portions of the reactant, such method including changing at least one of the amount of reactant added to the substance and the amount of the substance; and

(b) determining from said at least one of the change in the amount of reactant and the change in the amount of the substance, and from the change in the output signal, whether the sensor is responding to the substance or to the reactant.

2. A method, comprising:

(a) adding a reactant into a substance to react with the substance, the product of such reaction along with un-reacted portions of the substance and un-reacted portions of the reactant being directed to a sensor, such sensor producing an output signal in response to detection of both the un-reacted portions of the substance and the un-reacted portions of the reactant, such method including changing the amount of reactant added to the substance; and

(b) determining from the change in the amount of reactant and the change the output signal whether the sensor is responding to the substance or to the reactant.

3. A sensing system, comprising:

(a) a sensor for producing an output signal in response to either a first substance of a reaction or to a second substance of the reaction;

(b) an injector for changing the level of one of the substances in the reaction;

(c) a processor responsive to the output signal for determining whether the sensor is responding to the first substance to the second substance.

4. A NOx sensing system, comprising:

(a) a NOx sensor for producing an output signal in response to NOx and urea;

(b) a processor responsive to the output signal for determining whether the sensor is responding to NOx or to urea.

5. A method, comprising:

(a) adding a reactant to the substance to react with a substance, the product of such reaction along with un-reacted portions of the substance and un-reacted portions of the reactant being directed to a sensor, such sensor producing an output signal in response to detection of both the un-reacted portions of the substance and the un-reacted portions of the reactant, such method including changing the amount of reactant added to the substance; and

(b) determine whether the change in the amount of reactant and the change the output signal are in the same direction or in opposite directions.

6. A method for use in reducing a substance with a reactant added to the substance to react with such substance, the product of such reaction along with un-reacted portions of the substance and un-reacted portions of the reactant being directed to a sensor, such sensor producing an output signal in response to detection of both the un-reacted portions of the substance and the un-reacted portions of the reactant, such method comprising:

(a) changing the amount of reactant added to the substance;

(b) determining whether the output signal increases or decreases with the changed amount of reactant; and

(c) determining therefrom whether the sensor is responding to the un-reacted reactant or to the un-reacted substance.

7. A method, comprising:

changing an amount of a reactant added to a substance to be reacted, and thereby reduced by, the reactant, the time history of such change in the reactant being represented by:

Û(t)

where t is time;

such reaction taking place in the presence of a catalyst;

wherein un-reacted portions of the reactant and the un-reacted portions of the substance are passed by a sensor, such sensor producing an output signal in response to the un-reacted portions of the reactant or the un-reacted portions of the substance, the change in the output signal from the change in the amount of reactant added to the substance may be represented as:

{circumflex over (V)}(t)

wherein such change takes place over a period of time T, and with a delay between the change in the amount of the reactant, T0, to the time the reaction is effected by such change in the amount of reactant being represented by D;

producing a signal, I, to determine whether the output signal increases or decreases with the changed amount of reactant, such signal being produced in accordance with

5 I = ∫ t0 t0 + D + T ⁢   ⁢ e ⁢   ⁢ x ⁢   ⁢ c ⁢ V ^ ⁡ ( t + D ) ⁢ u ^ ⁡ ( t ) ⁢ ⅆ t

determining whether, I>0 or I<0;

concluding that the sensor is responding to the un-reacted reactant if I>0 and concluding that the sensor is responding to the un-reacted substance if I<0.

8. The method recited in claim 7 including determining whether I=0, and is so concluding that the injector is faulty.

9. The method recited in claim 7 including determining whether V is greater than a calibratable threshold and if so concluded that the catalyst is faulty.