FIELD OF THE INVENTION

The present invention generally relates to internal combustion engines equipped with exhaust after treatment devices. It more specifically relates to the control of an engine equipped with an oxidation catalyst and a NOx treatment device such as a SCR catalyst.

BACKGROUND OF THE INVENTION Current emission control regulations necessitate the use of exhaust after treatment systems in order to reduce the concentration of combustion byproducts and/or products of incomplete combustion.

Spark ignition (i.e., gasoline) engines conventionally use three-way catalytic converters to satisfy emissions regulations. Compression ignition (i.e., diesel) engines, however, are typically equipped with two-way catalytic converters (also referred to as Diesel Oxidation Catalyst - DOC), which may not efficiently reduce nitrogen oxides (NOx). Accordingly, diesel engines may include a reductant-based selective catalytic reduction (SCR) device in order to seek reduction in NOx, often the most abundant and polluting component in exhaust gases. In addition, diesel engines may also include diesel particulate filters (DPF) for particulate matter control.

Urea-based SCR catalysts use gaseous ammonia as the active NOx reducing agent. Typically, an aqueous solution of urea, also known as carbamide ((NH ₂ ) ₂ CO), is carried on board of the vehicle, and an injection system is used to supply it into the exhaust gas stream entering the SCR catalyst where it decomposes into gaseous ammonia (NH ₃ ) and is stored in the catalyst. The NOx contained in the engine exhaust gas entering the catalyst then reacts with the stored ammonia, which produces nitrogen and water. It is worth noting that to attain the conversion efficiency required to achieve the emission limits, the SCR-catalyst must be at least at a predetermined temperature known as light off temperature.

In an engine's exhaust system, the SCR catalyst is conventionally placed downstream of other exhaust after-treatment means, namely the catalytic converter and possibly a particulate filter.

One of the challenges in such exhaust line with serially mounted exhaust after treatment devices is its thermal management. In the above case of an exhaust line comprising, in series, an oxidation catalyst, a particulate filter and a SCR catalyst, the catalytic elements should reach their respective light off temperature as early as possible.

But the SCR catalyst is usually the last treatment component in the exhaust line and the most remote from the engine exhaust valves. In addition, each component in the exhaust line has a thermal inertia, whereby the tem- perature increase of components situated further away from the engine is delayed as compared to those near the engine. Thermal losses in the exhaust pipe itself are also to be taken into account. Fig .1 illustrates a typical temperature trace, in a diesel engine, for each of the diesel oxidation catalyst (DOC), the diesel particulate filter (DPF) and the SCR catalyst. As can be seen, the temperature of the SCR catalyst is heavily dependent on the thermal behavior of the other components in the exhaust line. In order to heat-up the catalysts as soon as possible, the engine is normally operated in a heat-up mode by acting on engine control parameters to heat-up the temperature of engine-out gases or by injecting fuel in the exhaust line that is burned in the DOC. Such heat-up strategy is e.g. described in US 2009/0217645.

The heat-up mode may be triggered from engine start up. Conventionally the thermal management is then carried out by monitoring the current temperature of the SCR catalyst; and the heat-up mode is stopped as soon as the target temperature in the catalyst is reached, e.g. the light off temperature. Unfortunately, under such control conditions the DOC gets very hot, while at the same time the SCR is still relatively cool due to thermal inertia of the components in the exhaust line. As can also be understood from Fig .1 , when the heat-up mode is stopped at, e.g., a light off temperature of 250°C measured in the SCR-catalyst, the thermal inertia in the system leads to an overshoot in the SCR temperature above 300°C. Such high temperatures are not required for NOx conversion and thus reflect a waste of fuel to heat-up the exhaust line.

OBJECT OF THE INVENTION

Hence, there is a need for an improved method of operating an internal combustion eng ine comprising an exhaust system with an exhaust after treatment device such as an SCR-catalyst.

SUMMARY OF THE INVENTION The present invention concerns a method for operating an internal combustion engine comprising an exhaust system as claimed in claim 1 .

The exhaust system comprises first exhaust after treatment means and, downstream thereof, second exhaust after treatment means; and the engine comprises an ECU configured to allow engine operation in at least one of a normal mode and a heat-up mode.

As it will be understood by those skilled in the art, the normal mode and a heat-up mode result from a given control of engine combustion to achieve a certain result in the exhaust system . The normal mode may for example correspond to engine settings designed to optimize emissions and fuel con- sumption. The aim of the heat-up mode is to increase the temperature in the exhaust system and hence of the exhaust after treatment means, i.e. to perform the "thermal management" of the exhaust line. When these engine combustion modes (normal / heat-up) are controlled in a coordinate manner with respect to the second exhaust after treatment means, they may be said to be associated therewith. In heat-up mode, any appropriate measures may be taken that result in an increase in the temperature of the exhaust gases arriving at the second exhaust after treatment means (as compared to normal mode), namely by acting on engine settings/control parameters to heat-up the temperature of engine-out gases or by injecting fuel in the exhaust line that is burned in the DOC. According to the present invention, a predicted temperature evolution of the second exhaust after treatment means is regularly determined based on a thermal model taking into account the thermal inertia of the exhaust system and having as input the current temperatures of the first and second exhaust after treatment means. The predicted temperature evolution of the second exhaust after treatment means is indicative of the temperature that the second exhaust after treatment means would reach during a simulated time period ahead of the current time, in case the operating mode was changed. This simulated time period starts at the current time and has a predetermined simulated duration, preferably substantially greater than the time required to perform the simula- tion. The operating mode (heat-up or normal) may then be changed depending on the predicted temperature evolution.

The method according to the present invention hence takes into account the current conditions and the thermal inertia of the treatment elements in the exhaust line to predict future temperatures. This simulation of predicted temperatures is however carried out on the hypothesis of a change in the operating mode. For example, if the engine was operated in heat-up mode to increase the temperature of the second exhaust after treatment means, the thermal model determines the predicted temperature evolution supposing that the heat-up mode is stopped and that engine is operated in normal mode. With this method, it is thus possible to stop the fuel-consuming heat-up mode as soon as possible, and at the same time warrant that the second after treatment means will operate at the required temperature.

In doing so, the obtained predicted temperatures (i.e. temperatures within the predicted temperature evolution) may be compared to an upper tempera- ture threshold and the heat-up mode may be stopped when it is determined that the predicted temperature has reached or exceeds such upper temperature threshold.

Similarly, the present method may be used to determine the point of time at which operation in normal mode should be switched to "heat-up" mode to avoid a drop of temperature below a predetermined temperature. Again, this may be carried out by comparing the obtained predicted temperature to a lower temperature threshold and engine operation may be switched from normal to heat-u p mode, when it is determined that the predicted temperature has reached or dropped below such lower temperature threshold. Such upper and lower threshold temperatures may be determined by calibration. Their selection implies a compromise between fuel consumption and conversion efficiency, but the thresholds should preferably ensure a sufficient overall NOx conversion efficiency; in practice those thresholds may be different but relatively close to the light off temperature. As it will be understood, these upper and lower thresholds allow apprehending the heat-up speed behaviour of the SCR catalyst.

Preferably, the thermal model uses a respective, predetermined steady- state temperature of the exhaust line that is associated with each operating mode. The steady-state temperature may be representative of the steady state temperature of one component of the exhaust l ine, preferably the second exhaust after treatment means. This input is a hypothesis on the long-term temperature of the second exhaust after treatment means and one limit condition for the calculations. For speed and ease of implementation, first-order lag filter models are preferred. The present method proves particularly interesting for the thermal management of an engine's exhaust system comprising an oxidation catalyst and/or a particulate filter as first exhaust after treatment means and, downstream thereof, a SCR-catalyst as second exhaust after treatment means.

These and other preferred embodiments are recited in the appended de- pendent claims 2 to 9. According to another aspect, the present invention also concerns an internal combustion engine comprising an exhaust system with exhaust after treatment means as claimed in claim 1 1 . Preferred variants of this internal combustion engine are recited in the appended dependent claims 12 to 15. It remains to be noted that while the present invention has been developed with the aim of optimizing the thermal control of a SCR-catalyst in an engine exhaust system, the present invention may be applied to the thermal control of other types of exhaust after treatment devices located downstream in the exhaust line and affected by the thermal inertia of the exhaust system and heat produced upstream thereof.

As it will also appear to those skilled in the art, the present method may be implemented with more than two operating modes for the second exhaust after treatment means, e.g. there may be one normal mode and two heat-up modes corresponding to different heating strategies (e.g. a light off heat up mode and a running heat up mode).

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention will now be described, by way of example, with reference to the accompanying drawings, in which:

FIG. 1 : is a graph showing temperature curves (vs. time) of a conventional mechanization for an SCR system during an FTP75 (or EPA III) emission cycle; FIG. 2: is a block diagram of the thermal model used in the preferred embodiment of the present method;

FIG. 3: is a combined graph illustrating the temperature simulation principle in accordance with the present method, when starting from heat up mode;

FIG. 4: is a combined graph illustrating the temperature simulation princi- pie in accordance with the present method, when starting from normal mode;

FIG. 5: is a combined graph illustrating current and predicted temperatures in accordance with a preferred embodiment of the present method. DETAILED DESCRIPTION OF A PREFERRED EMBODIMENT

As explained with reference to Fig .1 hereinabove, a simple temperature- based control of the SCR catalyst heat-up mode, wherein heating is stopped when it is measured that the SCR catalyst has reached its light off temperature, is not optimal since it typically leads to a temperature overshoot. This effect, due to the thermal inertia of the components in the exhaust line, implies excessive fuel consumption, while the higher temperature of the SCR catalyst is not even useful in terms of NOx conversion.

The present invention provides an optimized thermal control of exhaust after treatment means such as e.g. an SCR-catalyst in an engine exhaust line, wherein the SCR is located downstream of first exhaust after treatment means.

A preferred embodiment of this method will now be described with respect to Figs. 2 to 5 as applied to a diesel internal combustion engine comprising an exhaust line with first exhaust after treatment means comprising a diesel oxidation catalyst (DOC) contiguous to a diesel particulate filter (DPF) and further downstream, second exhaust after treatment means taking the form of a SCR-catalyst.

In such exhaust line, the temperature of the SCR catalyst is heavily dependent on the thermal behavior of the other components in the exhaust line. Due to important thermal inertia, the heat is progressively transferred from the DOC to the SCR catalyst by the exhaust gas stream.

As already discussed above, a control based solely on the monitoring of the current temperature of the SCR-catalyst leads to an overshoot in temperature, due to thermal inertia of the components. As a matter of fact, in the SCR heat-up mode the DOC and DPF reach high temperatures (and thus accumu- late heat) before the SCR actually reaches the light off temperature, indicated T _LO in Fig .1 . The delayed thermal transfer then causes the temperature overshoot in the SCR-catalyst.

The present inventors have thus observed that due to this thermal inertia of the components located upstream of the SCR catalyst, it is possible to stop the heat-up mode at a certain moment, even if the SCR is still cold, provided that the accumulated heat in the upstream components is sufficient to heat-up the SCR catalyst to its desired operating temperature (e.g. light off). Accordingly, the present method uses a mathematical thermal model to determine whether, by stopping the heat-up mode at a certain point of time (preferably towards the end of the simulation), the heat accumulated in the exhaust line and produced will be sufficient to reach the desired operating temperature of the SCR-catalyst. For this purpose, the engine control unit (ECU) comprises a so-called "real time predictor" designed to estimate the temperature value(s) in the SCR-catalyst during a certain time period of predetermined length ahead of the present time -referred to as simulated time period- if the heat-up mode was stopped (or more generally if the operating mode was changed). The resulting temperature value(s) may thus be used in the thermal control scheme of the SCR-catalyst. Actually, the extrema of the temperatures that the SCR-catalyst may reach during the future time period are of particular interest in this variant, as will be explained further below.

The type of suitable model to describe the thermal behavior of the exhaust line can be apprehended from the following mathematical approach. In fact, for each component (i.e. for the DOC, the DPF and the SCR), there is an energy balance equation that can be expressed as: dT

YYl gas ' C rD gas ' T out = fji gas · Q rD gas ' Ί ^Ί i-n + ΪΥΙ wall ' C rD wall · ^wa ^ where: m _gas is the mass flow rate of the exhaust gas stream; cp _g as is the specific heat of the exhaust gas;

T _out and T _in are the respective outlet and inlet temperatures of the component; m _wa ii is the mass of the wall of the component, c/ its specific heat and T _wa ii its temperature.

This equation can be solved to provide the outlet temperature T _out , assuming that T _out = T _wa ii (perfect gas-wall exchange). This results in a first order filter of T _in with a time constant that is a function of the exhaust flow: mwaU ^{' C} Pwall ^out γ _ γ

out in

gas dt which can further be expressed as dT

τ · out

out in '

dt where τ represents the thermal inertia time constant.

Theoretically, the maximum predicted SCR temperature could then be mathematically solved but this would lead to a very complex form with 3 cascades of differential equations.

Therefore, the proposed solution is that of modeling the temperature of the SCR-catalyst depending on the current temperatures (i.e. the temperatures at the moment the simulation is started) of the components in the exhaust line, i.e. the current/actual temperatures of each of the DOC, the DPF and the SCR.

In addition, to be able to easily use a first order filter model for each component, a further input parameter to the thermal model preferably is a prede- termined temperature that is selected by way of hypothesis as a steady state temperature (hereinafter noted T _SCR future) for the SCR and associated with a given engine operating mode (heat-up or normal). This temperature T _SCR future is a hypothetical temperature that the SCR-catalyst would reach in either the "normal" or the "heat-up" mode under regular, steady state (or long-term) driving conditions. Although named "hypothetical", such temperature may be determined by calibration; it is only hypothetical in the sense that driving conditions may lead to a different value in fine. Hence, the present inventors have found a way to avoid complex, resource-consuming calculations by using first-order lag filter models together with the long-term temperature hypothesis {T _S C _R future), which renders the use of the present real-time predictor possible at industrial scale (i.e. using conven- tional engine computing resources). The reason for this approach is that since it is never possible to know the engine's behavior in the future (due to a sudden driver acceleration for example), it is preferable to keep the model as simple as possible, which can be achieved by means of the long-term SCR temperature together with the first-order lag filters and using realistic main inertia time constants for an efficient implementation.

As mentioned above, the long-term temperature value (TscR Mure) may be determined by calibration, e.g. by using average steady state conditions of an FTP emission cycle or any other predetermined cycle. Alternatively, the long- term temperature value may be determined from past average driving condi- tions that hence take into account the actual driving behavior of the driver. Those skilled in the art may further devise other ways of determining a suitable long-term temperature value that provides a hypothesis on the future, steady state SCR temperature.

Turning now to Fig.2, box 10 contains: - a first order lag filter model for the DOC, indicated 12;

- a first order lag filter model for the DOC, indicated 14;

- a first order lag filter model for the SCR, indicated 16;

- and a simplified thermal loss model for the exhaust pipe in between the DPF and the DOF, indicated Losses. As explained, the thermal model 10 has four inputs: the current temperature of each component T _DOC feedback, T _DPFfeedback and T _SC Rfeedback- These temperatures may be measured or estimated by conventional methods. The fourth signal required for the models is the long-term temperature of the SCR, i.e. TscRfuture- As indicated above, this temperature depends on the operating mode of the SCR-catalyst. For example, it may be 280°C in heat-up mode or 220°C in normal mode.

However, while T _SC R uture \ the predetermined temperature associated with the operating mode (box 18), the thermal model in box 10 starts with the temperature T _DOC future- Well, since the DOC and DPF may be considered to be at the same temperature, T _DOC future may be considered to be equal to T _SCR future plus the thermal losses ("Losses" in Fig.2) in the exhaust pipe.

The resulting temperature, as calculated by the final model 16, is noted

T SCR Pred■

For each simulation phase, each sub-model may then be initialized with the measured temperature of the modeled component, ensuring that the stored energy of each component is taken into account. This is of advantage since the DOC and/or DPF can internally generate heat via catalytic combustion of hydrocarbons and carbon monoxide. Finally, a supervisor (box 20) achieves a scheduling function that controls the thermal model 10 depending on the current operating mode and starts the simulations. The simulation is carried out during a preferably short lapse of time (a few seconds) in order to determine the possible evolution of the temperature in a predetermined time period following the starting point of simulation and extending during a comparatively longer simulated time period (e.g . 50 to 1 00 times -or more- the duration of the simulation). Accordingly, during the simulation period a plurality of predicted temperatures are preferably calculated that are indicative of the estimated temperatures that the SCR will take in the future. However in a preferred embodiment, as will be explained below, only the extrema are memorized for decision making.

For example, the simulation may be carried out during two seconds in order to simulate the temperature evolution of the SCR-catalyst during the following 200 seconds. The time period of 200 s taken into account for the simulation has been selected because for a standard internal combustion engine (3 to 6 cyl inders) of an automobile, this time period is considered appropriate, having regard to the time constants of thermal inertia.

The thermal control of the SCR-catalyst may then be performed in the engine ECU by comparing the predicted temperature values to a threshold to decide whether or not to switch from one operating mode to the other (from heat-up to normal or vice-versa).

Fig .3 is a principle diagram illustrating the simulation and decision making in accordance with the present method. On the left of the relative time t=0 (also indicated "now"), the current operating mode is "heat up". At t=0, a simulation is started in accordance with the thermal model of Fig .2 that simu lates the temperature of the SCR catalyst over a time period of predetermined length "simulated time period", here e.g. 200 s.

Still in Fig .3, the dashed l ine indicates the so-called "maximum" decision threshold ( T _T H_MAX) ^or deciding whether or not the heat-up mode may be ended, while ensuring that the SCR will subsequently operate in a desired temperature range. At t=now, the temperatures correspond to the current, real temperatures of each component (DOC, DPF and SCR). On the right of the t=now line, all temperatures are simulated , supposing that the eng ine operating mode is switched from "heat up" to "normal" and assuming that the SCR will reach a steady state temperature TscR Mure i (the long term temperature associated with the normal mode).

As can be seen, the simulation reveals that T _S CR will reach T _T H_MAX- Therefore, the engine operating mode can be switched to normal from that moment on.

The inverse situation will now be explained with respect to Fig .4. At t=0 (now) the eng ine is currently operated in the normal mode and the current temperatures are those indicated by the points on the vertical line. On the right of the vertical line t=now, the simulated temperatures are represented for the future, simulated time period, with the hypothesis of a switch to the "heat up" mode and a long term temperature T _S CR future 2- As can be seen, if the heat up mode was entered at t=now, the SCR temperature would continue dropping down to a minimum and level off towards a steady state temperature. Detecting the minimum of the predicted temperatures allows deciding when to switch from normal to heat up. A proper selection of the minimum threshold value will avoid a sensible temperature drop of the SCR catalyst. In Fig.4, the simulated SCR temperature drops down to the minimum threshold level indicated T _{TH MIN} , SO that the mode is switched to "heat up" again.

Example: A practical embodiment of this method will now be described with reference to Fig.5. In this example, the operating mode is "heat-up", i.e. measures are currently being taken by the ECU to generate more heat in the exhaust gas stream so as to accelerate the heating of the SCR. The horizontal axis indicates the current time (in seconds) and the vertical axis indicates the temperature. The current (real) temperature of the SCR (i.e. at the time indicated on the axis) is indicated as T _SCR feedb _a ck-

Every 2s, the supervisor 20 triggers the start of a simulation to determine the temperature evolution of the SCR during a simulated time period (here 200 s) in case the mode was switched to normal . This simulation uses, as explained with respect to Fig.2, the current temperatures of the DOC, DPF and SCR and a long-term temperature TscRjutur* (respectively T _DOC future) .

In Fig .5, the grey window indicates one sample simulation started at t=138 s and ending at t=140 s. The predicted temperature evolution of the SCR drawn in this window however corresponds to a simulated time period (tp) of 200 s . As can be seen , the predicted temperature varies from the current temperature (T-i) to about the long-term temperature (T _S s) while passing through a maximum T ₂ . From this sample simulation it can be seen that at 138s, considering the current temperatures of the components, the SCR may attain within the next 200 s a temperature of up to T ₂ if the SCR operating mode was switched from heat-up to normal as a result of this simulation step (say at 140s), this even if the current temperature of the SCR is T-,. In th is example, the difference between the maximum SCR temperature and the current SCR temperature is high because the DOC temperature is actually also very high and thermal energy will then be transferred from the DOC to the SCR.

Hence, the fact that the predicted temperature reaches or exceeds a threshold value, here e.g. T ₂ , allows to decide the switch from heat-up to normal mode, although the SCR temperature is still low at the beginning of the simulation.

For practical reasons, the extrema of the predicted temperature during the simulation time frame (tp) are extracted and named T _{SCR p} redMAx and T _{SCR pr} edMiN- For the simulation step between 138 and 140s, the T _{SCR pr} edMA ^2 and

When operating in SCR heat-up mode, the test for switching from heat-up to normal mode may then be whether T _{SCR p} redMAx reaches or exceeds a predetermined threshold value, which would indicate that considering the current temperatures in the exhaust line, the SCR will attain in the future this threshold temperature if the heat-up mode was stopped.

The minimum T _SCR predMiN is of interest in the other case (as explained with respect to Fig.4) where the ECU is currently not taking any heat-up measure to accelerate the heating of the SCR-catalyst, and thus operating in "normal" mode. In such case, the question arises as to when the "heat-up" mode should be entered again, in order to avoid a drop of SCR temperature below a temperature efficient for NOx conversion (e.g. light off). In this connection, those skilled in the art will understand that due to the thermal inertia of the exhaust system, starting the heat-up mode does not prevent the SCR catalyst from falling below a predetermined temperature. Here the critical aspect is to avoid a too important drop of SCR temperature, which would happen if the heat-up mode was entered too late.

Accordingly, the simulation performed by the thermal model of Fig.2 provides the minimum temperature T _{SCR pr} edMiN, which corresponds to the minimum temperature that would be reached within the simulated time period represent- ing a time period of 200 s ahead of the present time and if the SCR operating mode was then switched to "heat-up". Again, T _{SCR p} red MiN may be compared to a lower temperature threshold, so that when T _{SCR p} red MiN reaches or drops below that threshold, the heat-up mode is triggered.