GB2592624A

GB2592624A - Method for monitoring an exhaust after treatment device

Info

Publication number: GB2592624A
Application number: GB2003114.2A
Authority: GB
Inventors: Bovi Philippe
Original assignee: Delphi Automotive Systems Luxembourg SA
Current assignee: BorgWarner Luxembourg Automotive Systems SA
Priority date: 2020-03-04
Filing date: 2020-03-04
Publication date: 2021-09-08
Also published as: CN113356983A; CN113356983B; GB202003114D0

Abstract

A control unit receives temperature signals from a pair of temperature sensors (20, 22, fig 4) arranged upstream and downstream of an exhaust treatment device (16, fig 4). The control unit runs a temperature model to estimate a temperature of the exhaust treatment device, taking into account the thermal inertia of the exhaust treatment device and receiving as input an inlet temperature determined by the upstream temperature sensor. The thermal model uses a nominal inertia corrected by a correction coefficient, the correction coefficient being determined by a learning procedure on the basis of a deviation between modelled and measured outlet temperatures during a transient engine operating phase, such as at cold start. The exhaust treatment device may be a catalyst or particulate filter. An internal combustion engine (12, fig 4) comprises an exhaust system with the control unit, temperature sensors and exhaust treatment device.

Description

METHOD FOR MONITORING AN EXHAUST AFTER TREATMENT DEVICE FIELD OF THE INVENTION

The present invention generally relates to the field of exhaust gas treatment in internal combustion engines.

BACKGROUND OF THE INVENTION

Modern internal combustion engines are featured with various exhaust after treatment devices to reduce the toxicity of emissions from the engine. Components typically used for treating the exhaust gas include: * a catalytic converter to break down gaseous pollutants in the exhaust gas into less harmful components; * a particulate filter (or soot filter) to remove the fine, solid particles in the exhaust gas (especially in diesel engines).

For optimal operation of these components, it is necessary, among other things, to regulate the exhaust gas temperature and/or the exhaust after treatment devices in predetermined ranges. For example, a catalytic activity of known catalysts for converting pollutants in the exhaust gas is strongly temperature-dependent. This means that a minimum temperature at the catalytic converter must be exceeded (so-called light-off temperature) in order to make it possible to reduce pollutant emissions.

Furthermore, with stringent emission limits imposed to vehicles powered by internal combustion engine, exhaust temperature is becoming more and more critical for proper after-treatment operation.

Several temperature sensors are usually installed in the exhaust line, at the most relevant locations but temperature models are required to estimate the temperature for after-treatment control and diagnostic.

For example, temperature models can be used in the following applications: -estimation of the temperature at locations not fitted with sensors (catalyst bed temperature...) - control of active heat-up strategies: catalyst light-off, DPF regeneration...

- diagnostic of catalysts and plausibility diagnostic of temperature sensors Temperature models conventionally consist of estimating the heat losses, but also of modeling the delay and dynamic change due to the inertia of the components. Common way to estimate temperature is to use map based coefficients associated to physical formulas such that a catalyst for instance is characterized by nominal heat transfer coefficients and a nominal inertia.

Whereas inertia has a low importance in case of steady operating conditions resulting in slow temperature changes, it becomes critical during transient operation. If the thermal inertia of the exhaust treatments device (high inertia components are the catalysts and the particulate filters) varies from the nominal value (production tolerance, part removal, aftermarket component...), the dynamic of the estimated temperature will be different and can lead to large temperature error compared to the actual value. A large temperature error can lead to bad temperature control or to wrong diagnostic reporting.

Hence it would be desirable to dispose of ways to more properly estimate the temperature of exhaust after treatment devices.

SUMMARY OF THE INVENTION

The present invention relates to a method of operating an exhaust piping system of an internal combustion engine, wherein the exhaust piping system comprises at least one exhaust treatment device and a control unit receiving temperature signals from two temperature sensors arranged respectively upstream and downstream of the exhaust treatment device. A control unit runs a temperature model to estimate a temperature of the exhaust treatment device, the temperature model taking into account the thermal inertia of the exhaust treatment device and receiving as input an inlet temperature determined by the upstream temperature sensor.

According to the invention, the thermal model uses a nominal inertia corrected by a correction coefficient, the correction coefficient being determined by a learning procedure on the basis of a deviation between modeled and measured outlet temperatures during a transient engine operating phase It will be appreciated that the present method proposes a way to estimate the actual inertia of a component in the exhaust line, which is reflected by the correction coefficient. In the method, the actual inertia of an exhaust treatment device located between two temperature sensors is thus estimated in order to improve the accuracy of the temperature estimation downstream of this component compared to a model simply based on a nominal inertia.

The present method is applicable to any component in the exhaust system and to any kind of engine type. The after treatment device may in particular be a catalyst, for example an oxidation catalyst, diesel oxidation catalyst, three-way catalyst, selective reduction catalyst, or NOx adsorber, or a particulate filter, for example a diesel particulate filter, or gasoline particulate filter, or a combination of the latter. The after treatment devices can be packaged individually or in combination. In other words, as used herein, the term after treatment device also designates the case of two or more after treatment devices assembled as one component. Depending on the embodiment, the temperature sensors can be arranged upstream and downstream of a single after treatment component of the assembly, or upstream and downstream of the assembly of devices.

The learning procedure can be reliably carried out over a transitory period involving a sufficient temperature variation, in particular a temperature increase of at least 80, 100, 120 or 140°C.

In embodiments, the learning procedure is conveniently performed during a monitoring window corresponding to a heat-up phase from a cold engine. Firstly, executing the learning when engine and exhaust components start from cold condition avoid adding inaccuracies caused by the initial state. Furthermore, the cold start ensures a sufficient variation of temperature and excludes exothermal reactions. Indeed, the transient engine operating preferably phase excludes exotherms, which modify the thermal transfer through catalysts, for example.

Accordingly, when starting from a cold engine, the monitoring window is preferably ended when the downstream temperature meets an upper temperature threshold, from which exothermal reactions could occur. In practice, the upper temperature threshold may be in the range of 100 to 140°C.

Although performing the learning procedure from cold start is advantageous, it is not the only option. The learning procedure can also be executed under other operating conditions, in particular where a sufficient temperature variation can be reached. For example, a transient regime with a temperature variation of within the above-mentioned range of 80 to 140°C is considered sufficient. This temperature variation may occur, e.g., over a time period of 60 to 180 s. The monitoring window should however exclude exotherms. Such transient phase may e.g. occur when a vehicle exits an urban area to enter the motor way and drives at a constant higher speed for a few minutes.

The learning procedure may be performed at each engine (cold) start or with a predetermined periodicity, and said correction coefficient used by the temperature model is updated upon completion of the learning procedure. In general, the learning procedure can be performed at any desired moment provided the above-mentioned transient conditions are met.

In embodiments, the correction coefficient corresponds to an inertia ratio reflecting the ratio of the actual inertia over the nominal inertia. The inertia ratio may be estimated from the relative position of the downstream temperature measured with said second sensor between two modeled temperatures determined with the temperature model using respectively an excess inertia and a reduced inertia.

In some cases, where measured temperature is very close to one of the modeled temperatures (with excess or reduced inertia), the inertia ratio can be approximated by linear interpolation.

In embodiments, the inertia ratio is approximated by assimilating thermal inertia to a low-pass filter, whereby for a constant inlet temperature, the inertia ratio is equivalent to a ratio of convergence percentage close to the time constant, the inertia ratio being determined using the relationship: inertia ratio - 1 In ( 1 1 -%conyactual) where %convactuat is the percentage of convergence of the measured signal.

In embodiments, the determined inertia ratio can also be used to perform fault detection in case of abnormal inertia ratio value. Suppose that, e.g., the monitored after treatment device is removed from the exhaust line. During the learning procedure, the inertia ratio would be very low. This can be detected by comparing the inertia ratio, respectively the correction coefficient, to a predetermined threshold value.

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention will now be described, by way of example, with reference to the accompanying drawings, in which: Figure 1: is a graph illustrating the evolution of modeled and measured temperatures vs. time during the monitoring window; Figure 2: is a graph illustrating the convergence ratio of a filtered signal for a normalized constant inlet temperature; Figure 3: is a graph illustrating modeled and measured outlet temperatures during engine operation; Figure 4: is a diagram of an exhaust piping system; and Figure 5: is a flow chart of an embodiment of the present method. 20 DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS Fig. 4 shows a conventional exhaust piping system 10 connected to an internal combustion engine 12, e.g. a multi-cylinder diesel engine, which generates an exhaust gas stream containing pollutants with soot and/or particles. An Engine control unit (ECU -not shown) is signally and operatively connected to a number of sensors and actuators for controlling and monitoring engine operation, as it is known in the art.

The drawing of Fig.4 illustrates a simple exhaust system 10 comprising an exhaust piping 14 in which an after treatment component 16 is arranged, including e.g. an oxidation catalyst. The after treatment device 16 is arranged such the whole exhaust gas flow must pass through it. This is only an example and, as it will be known to those skilled in the art, the exhaust line may comprise one, two or more after treatment devices selected from the conventional devices such as catalysts, particulate filters, absorbers or SCR systems.

Also to be noted, whereas in the above example component 16 comprises a single oxidation catalyst, in other embodiments two or more after treatment devices can be arranged in a same package, e.g. an oxidation catalyst and particulate filter.

Optimal operation of the exhaust after treatment devices requires regulating the exhaust gas temperature and/or the exhaust after treatment devices in predetermined ranges. Therefore, several temperature sensors are usually installed in the exhaust line, at the most relevant locations.

The following description will, for the sake of exemplification, focus on the thermal monitoring of the oxidation catalyst 16. However the same method can be applied to any components in an engine exhaust line.

Referring again to Fig.1, a first temperature sensor 20, or upstream sensor, is arranged before of the oxidation catalyst 16 having regard to the exhaust gas flow. A second temperature sensor 22, or downstream sensor, is arranged after the oxidation catalyst 16.

In case the component 16 consists of an assembly of two or more after treatments devices, the upstream and downstream sensor can be arranged directly upstream and downstream of the individual after treatment device to be monitored, or as shown in Fig.4 if inertia is to be assessed for the whole component.

In the ECU, the temperature of the oxidation catalyst 16 is estimated by means of a temperature model as a function of the measured upstream temperature (determined by upstream sensor 20), heat losses and thermal inertia of the oxidation catalyst. Conventional temperature models may be based on a low pass filter behavior, e.g. a multi-slice model where the inertia of each slice is modeled as a low-pass filter. Such temperature models are known in the art and will not be further detailed herein (see e.g. E P1242724 or EP2661546).

The temperature estimation may be used for various reasons: absence of sensor, control strategies, on-board diagnostic. The thermal model calculates (estimates) a temperature that is referred to as outlet temperature or modeled temperature.

As regards more specifically thermal inertia, the model uses a so-called nominal inertia, i.e. a default value of inertia reflecting the "nominal" thermal inertia, which is initially set into the ECU to reflect the inertia for the part reference (e.g. manufacturer data or measured/calibrated value). The nominal inertia is thus a parameter of the thermal model that characterizes the inertia of the modeled device.

In the present method, a correction coefficient that is determined by way of a learning procedure is applied to the nominal inertia. The learning procedure is performed during a transient engine operating phase (preferably without exotherms) and the correction coefficient is determined on the basis of measured and modeled temperatures.

In other words, the present method proposes a way to estimate the actual inertia of a component in the exhaust line, which is reflected by the correction coefficient.

In the method, the actual inertia of an exhaust treatment device (e.g. catalyst, particulate filter...) located between two temperature sensors is thus estimated in order to improve the accuracy of the temperature estimation downstream of this component compared to a model based on a nominal inertia (i.e. not corrected).

The correction coefficient for the thermal inertia is conveniently determined during the heat-up phase after a cold start. The learning procedure is hence executed from the initial temperature (inlet, outlet and component at the same temperature) until reaching a predetermined outlet temperature. The end of the monitoring window for the learning procedure is typically an outlet temperature of the oxidation catalyst, measured by sensor 22, inferior to temperatures where exothermal reactions could occur in case of catalyst. In practice, the upper temperature limit may be between 100 and 140°C.

It may be noted that within this monitoring window, the inlet temperature can be very dynamic and calculating the inertia of the after treatment device by using the measured upstream and downstream temperatures, as such, is not simple because of the temperature distribution inside the device.

Accordingly, the present method proposes using the temperature model for the after treatment device and calculating two outlet temperature estimations: one for an inertia lower than nominal, referred to as 'reduced inertia'; and one for an inertia higher than nominal, referred to as 'excess inertia'.

At the end of the monitoring window, the measured temperature downstream of 10 the after treatment device is compared to the two estimated temperatures and the relative position of the measured outlet temperature is converted to an inertia ratio.

Referring to Fig.1, the monitoring window starts at to and ends at ti. In this example, the monitoring window has a duration of 120 s. The graph of Fig.1 shows the evolution of: - the actual outlet temperature, line 30, i.e. the temperature measured by the temperature sensor downstream of the component being monitored; - the nominal temperature, line 32, i.e. the outlet temperature of the temperature model using the nominal inertia value without applying a correction coefficient; - the modeled temperature with excess inertia, line 34; - the modeled temperature with reduced inertia, line 36. The bottom graph in Fig.1 shows the evolution of the correction coefficient. In the following, for the sake of exemplification, the reduced inertia is 30% lower than the nominal inertia, whereas excess inertia is 30% higher than the nominal inertia.

As can be seen in Fig.1, at the end of the monitoring window the measured temperature Tempactuar is closer to the value of the estimated temperature with reduced inertia. The modeled temperature value with reduced inertia is noted TempLo whereas the modeled temperature value with excess inertia is noted TempHI In the present method, the inertia ratio (ratio of actual inertia / nominal inertia) is calculated to reflect the relative position of the measured temperature between 5 the two modeled temperatures Tempe, and Temp,* at the end of the monitoring window.

This inertia ratio can only be approximated. Indeed, for a dynamic input, there is no direct relationship between the relative position of the measured temperature at a specific time and the inertia ratio, in particular because: inlet temperature is not constant; temperature distribution inside the catalyst/component is unknown; water condensation influences the temperature rise. Approximation is thus required.

One possible approach to approximate the inertia ratio is to simplify the system based on the hypothesis that: inertia is assimilated to a simple low-pass filter, the inlet temperature is considered constant and the calculation is executed close to the time constant (1 TAU).

In this case, one can write: Ratioraw = (T emPactuat T emPrei) AT emPlo errIPHO (eq. 1) With: Ratio: relative position of the measured temperature between the 2 modeled temperatures, Tempactuai. measured temperature, Tempm: temperature modeled with high inertia, Tempe° temperature modeled with low inertia.

Eq. (1) can be rewritten as: TemPactuat -Tempm + Ratioraw * (Tempi, -Tempm) (eq. 2) Based on the hypothesis that inertia corresponds to a low pass filter, a given time constant T corresponds to a given convergence rate. For a same input, the output is equivalent to the convergence rate. Fig.2 illustrates the principle of a convergence rate (line 40) for a normalized signal (line 42) In this case, the relative position is equivalent to a ratio of convergence percentage which can be converted to an inertia ratio: %convact",il = %convlli + Ration:a, * (%conyto -%convKi) (eq.3) with %convadual: percentage of convergence of the measured signal; %convEll: percentage of convergence of the temperature modeled with high inertia (+30%); %cont/Lo: percentage of convergence of the temperature modeled with low inertia (-30%).

For a low-pass filter, one can write: %cony = (1 Edt) (eq-expH With: %cony: percentage of convergence 1dt: elapsed time T: time constant of the low-pass filter Equation 3 can then be re-written as: (1 1 1 I + Ratio",, * ((1 exp(Eat) exp(El j (.1 exp(Edl if cHi cLo tlii Eta) exp( tuctuat (eq.5) With: Tama,: time constant corresponding to the actual inertia = x* nominal, x: ratio of actual inertia / nominal inertia, TN!: time constant corresponding to the high inertia = 1.3* T nominal, no: time constant corresponding to the low inertia = 0.7* r nominal. We then have: (1-exp)) (1 exp Edt Ratioraw * ((1 exp( at)) (rt c no mina! (1.3' c nominal) °-7"-nominal) (1 Lac)) (eq. 6) exp(1-An) nominat If we consider that the calculation of the ratio will always be executed after a time Eat close to the time constant of the nominal system, -1, then Tnominal %cony exp(n) A expG)-actual -1 = (1) + Ratiora, * ((1 1 expG 1)) lect) (1-exPb) %convactuat can be calculated from Ratioraw. . The actual inertia ratio x can be calculated: x - \ (eq. 8) In( 1 r-Aconvactuad In equation 8, term 'x' corresponds to the ratio of the actual inertia over the nominal inertia.

The inertia ratio determined by the learning procedure is then updated in the 20 thermal model associated with the component 16 so that the outlet temperature is estimated by using the nominal inertia corrected by the learned value of correction coefficient. a

Based on the same example as Fig.1, the graph of Fig.3 shows (for an engine operating period) the modeled outlet temperature with nominal inertia -line 32, the measured outlet temperature -line 30, and the modeled temperature with nominal inertia modified by the updated correction coefficient -line 38.

Fig.5 is a flow chart that briefly summarizes the inventive method applied to an after treatment device. The learning procedure starts with a cold engine start at 100. During the learning procedure, which encompasses a heat-up phase up to a predetermined limit temperature Turn, the upstream and downstream temperatures, Temp up and TeMPactual are measured (box 102) by means of the sensors. The thermal model associated with the after treatment device is used to compute (box 104) the two temperatures values with excess and reduced inertia: TempFri and TempLo.

As long as the measured downstream temperature Temn -,-actual is below Turn, the learning is continued (test box 106 = No). When Tempactual reaches the limit value Tun, the monitoring window is closed The inertia ratio is calculated (box 108) based on the values obtained at the end of the monitoring window.

Then the thermal model is initialized at 110. The correction coefficient (cc) used in the temperature model for correcting inertia is updated based on the determined inertia ratio. The modeled temperature downstream (Tout) of the part is initialized to the measured temperature TeMPactual. The improved modeled temperature can then be used for control and/or diagnostic strategies.

Claims

Claims 1. A method of operating an exhaust piping system of an internal combustion engine, said exhaust piping system comprising at least one exhaust treatment device and a control unit receiving temperature signals from a pair of temperature sensors arranged respectively upstream and downstream of said exhaust treatment device, wherein said control unit runs a temperature model to estimate a temperature of said exhaust treatment device, said temperature model taking into account the thermal inertia of said exhaust treatment device and receiving as input an inlet temperature determined by the upstream temperature sensor; characterized in that said thermal model uses a nominal inertia corrected by a correction coefficient, said correction coefficient being determined by a learning procedure on the basis of a deviation between modeled and measured outlet temperatures during a transient engine operating phase.
2. The method as claimed in claim 1, wherein said learning procedure is performed during a monitoring window corresponding to a heat-up phase from a cold engine.
3. The method as claimed in claim 2, wherein said monitoring window ends when the downstream temperature meets an upper temperature threshold, in particular in the range of 100 to 140°C.
4. The method as claimed in claim 1, wherein said learning procedure is performed during a monitoring window corresponding to a transient engine operating phase with a temperature increase of at least 80°C, preferably between 100 and 120°C.
5. The method as claimed in any one of the preceding claims, wherein said transient engine operating phase excludes exotherms.
6. The method as claimed in any one of the preceding claims, wherein said correction coefficient corresponds to an inertia ratio reflecting a ratio of actual inertia over nominal inertia.
7. The method as claimed in any one of the preceding claims, wherein said inertia ratio is estimated from the relative position of the downstream temperature measured with said second sensor between two modeled temperatures determined with the temperature model using respectively an excess inertia and a reduced inertia.
8. The method as claimed in claim 7, wherein said inertia ratio is approximated by linear interpolation between the two modeled temperatures corresponding to the excess inertia and reduced inertia, respectively.
9. The method as claimed in claim 7, wherein said inertia ratio is approximated by assimilating thermal inertia to a low-pass filter, whereby for a constant inlet temperature, the inertia ratio is equivalent to a ratio of convergence percentage close to the time constant, the inertia ratio being determined using the relationship: inertia ratio -In ( I. 1 -%convactua) where %convactuaf is the percentage of convergence of the measured signal.
The method as claimed in any one of the preceding claims, wherein said learning procedure is performed at each engine cold start or with a predetermined periodicity, and said correction coefficient used by said temperature model is updated upon completion of said learning procedure.
11 The method as claimed in any one of the preceding claims, wherein said exhaust after treatment device is a catalyst or a particulate filter.
12 The method as claimed in any one of the preceding claims, wherein the determined inertia ratio, respectively the correction coefficient, is used to perform fault detection, in particular by comparison to a predetermined threshold value.
13. An internal combustion engine comprising an exhaust piping system with at least one exhaust treatment device and a control unit receiving temperature signals from a pair of temperature sensor arranged respectively upstream and downstream of said exhaust treatment device, wherein said control unit is configured to implement the method as claimed in any one of claims 1 to 12.