TECHNICAL FIELD

This application relates to a method of controlling catalytic units operation in a vehicle exhaust system, such as those including Selective Catalytic Reduction (SCR) units where stored ammonia therein converts NOx. It has particular application to exhaust systems which comprise a first SCR unit upstream of a second SCR unit and where each SCR unit has a corresponding urea (reductant) doser unit located upstream thereof. The dosers dose urea which is converted to ammonia (NH3) for the corresponding SCR unit. It has particular application to a method of controlling the reductant (urea) dosing in the said dosers.

BACKGROUND OF THE INVENTION

Particularly in Diesel applications for CN6b and Euro 6d / Euro7 markets there will be a requirement for improved control of emissions and thus this will require more complex after treatment systems in vehicle exhausts than CN6a or Euro 6d-temp.

One typical variant consists of dual SCR catalysts (i.e. two serially located SCR units) combined with two corresponding separate urea/reductant (e.g. Adblue) dosers therefor. In the exhaust system of such system is a first SCR unit located upstream of a second (downstream) SCR unit. The second SCR unit is often referred to as an Under-Floor SCR unit (UF SCR), and the first one is often a combined Diesel particulate filter (DPF) with SCR functionality; often referred to as a SDPF. Hereinafter reference to SDPF can be interpreted as reference and interchangeable to the first (upstream) SCR unit and reference to the UF SCR can be interpreted as reference and interchangeable to the second (downstream) SCR unit Above 450-500°C, most of the NH3 (from urea) is oxidized in the SDPF and can then not convert NOx Adding a second doser at a colder location allows NOx conversion in a second downstream SCR (e.g. UF SCR)

The addition of a second urea injection point increases the complexity of the SCRs control: the second urea doser provides an extra degree of freedom for the control. The second SCR catalyst can be fed with NH3 slipping out of the SDPF and also by direct urea injection with the second doser.

With two independent dosers, the control of two SCRs requires new open-loop and closed-loop strategies. Each SCR has to be controlled to provide a given stored amount of NH3 for NOx conversion. Expensive NOx sensors have to be kept to a minimum number. Dual doser control needs to have commonality with single doser control and there need to be an avoidance of proliferation of controls and calibration methodologies

Such dual doser control can become much more complex than single doser. The 2019 publication titled “Integrated Diesel System Achieving Ultra-Low Urban and Motorway NOx Emissions on the Road” by Demuynck et al, Vienna 2019 - AECC/IAV) describes a dual doser control system. Several issues are linked with this kind of control. The Kalman filter employed breaks the link between the physical/chemical behavior of the actual system and the control, which makes it difficult for the calibrators to fine-tune the control based on physical/chemical observations. The NOx sensor located downstream of SDPF will operate with high levels of NH3 in the gas most of the time (NH3 slip from SDPF). This is due to fact that a SDPF requires high level of NH3 filling to perform at its highest possible efficiency. NOx sensors being cross-sensitive to NH3, the estimation of NH3 concentration and NOx concentration downstream of SDPF are very inaccurate. The signals will typically be less accurate than the delivery accuracy of state-of- the-art dosing systems. Moreover, NOx sensors greatly increase the total cost of an SCR system. Adding this extra sensor should be avoided if not absolutely required. A further problem is the inaccuracy of the NH3 signal (and NOx) between SDPF and SCR, combined with the activation of the second doser makes it very risky to correct the dosing flow of the second doser (closed-loop): a drift of the front doser can wrongly be identified as a drift of the second doser, when closing the loop with the rear NOx sensor. In case of urea slip due to poor mixing / degraded front SCR, this becomes even more obvious that the second closed loop wrongly corrects the second doser.

The added complexity of the above control and its additional NOx sensor are difficult to justify because the second doser is mainly (/only) intended to be used in very high exhaust temperature conditions such as DPF regeneration. In normal conditions, it is actually detrimental to the global NOx performance of the system to control the rear SCR with the second doser instead of overfilling the front SCR to generate extra NH3 : the SDPF performance greatly increases when it is saturated with NH3, which is the case when the front doser is used to control both SCR catalysts.

SUMMARY OF THE INVENTION

In one aspect is provided a method of control for a vehicle exhaust system, said vehicle exhaust system including a first SCR unit, and a second SCR unit located downstream of said first SCR unit, and including a first urea doser adapted to inject urea reductant upstream of said first SCR unit, and a second urea doser located upstream of said second SCR unit and adapted to inject urea upstream of said second catalytic unit, said method comprising the steps of : providing a model or MAP of said second SCR unit, adapted to provide a demand signal for said second doser; said model or MAP including one or more stored parameters indicative of characteristics of the second SCR unit and/or dosers therefore; said method further including subsequent to operation of the first doser; the steps of a) switching off the operation of the first doser; b) operating said second doser for a flushing period of time; c) after expiry of said flushing period, running the second SCR in closed loop control, d) updating said model or MAP it and doser during said closed loop control.

Step d) may comprise updating said parameter(s).

Said second doser may be operated in open loop control.

Said demand signal for said second doser is dependent on said parameters stored respectively in said models/MAP of said second SCR.

Said flushing period is preferably sufficient such that the first SCR is empty of NH3/urea.

The flushing time is preferably sufficient that additionally the second SCR unit is operating under conditions where there is no effect on NH3 level stored on said second SCR unit of previous operation of the first SCR unit doser.

The flushing time is preferably sufficient for the second SCR unit and doser to be working under steady state conditions.

The method may comprise the additional steps of: i) providing a value for NH3 storage on said first SCR units ii) determining the flow rate of potentially converted NH3 equivalent NOx flowing into the first SCR; iii) integrating the product of the flow rate from ii) multiplied by the NOx conversion efficiency for said first SCR unit over time iv) comparing the value of step iii) and step i) determining the flushing time is finished when the value determines form step iii) substantially reaches said NH3 storage value of step i)

Said value in step 1 may be based on the maximum storage capacity of the first SCR. The method may including additionally; v) providing a value for NH3 storage on said second SCR unit vi) determining the flow rate of NOx flowing out of the first SCR; vii) integrating the product of the flow rate form vi) multiplied by the NOx conversion efficiency of said second SCR unit over time; iv) comparing the value of step vi) and step v) and determining the flushing time is finished when in addition, the value determined from step vii) substantially reaches said NH3 storage value of step v).

The value in step v) may be based on the maximum storage capacity of the first second

The method may include providing a model or MAP of said first SCR unit, said model or MAP including one or more stored parameters indicative of characteristics of the first SCR unit and/or dosers therefore and adapted to provide a demand signal for said first doser and/or to provide an estimate of the NOx flowing out of said first SCR unit. The demand signal for said first doser may be dependent on said parameters stored said models/MAPs of said first SCR.

The method may include subsequent to step d); of; e) turning off operation of said second SCR doser; f) turning on operation of said first SCR doser in closed loop control; g) subsequently updating said stored parameters of said model of said fisrt SCR unit. The method may include providing a second flushing period between steps e) and f). Said exhaust system may have a NOX sensor located downstream of said second SCR unit, said sensor being used to provide feedback control during said closed loop operation(s).

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention is now described by way of example with reference to the accompanying drawings in which: - Figure 1 shows an example of an exhaust system with dual SCR in which embodiments of the invention can be performed;

- Figure 2 shows a control system which may be employed in aspects of the invention;

- Figure 3 above shows how the flushing quantity and time can be calculated; - Figure 4 illustrates how the flushing period may be calculated.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

Figure 1 shows an example of an exhaust system with dual SCR in which embodiments of the invention can be performed. It shows a portion of a vehicle exhaust system having dual doser/SCR after- treatment. Essentially there is a first upstream SCR (e.g. SDPF ) unit 1 and a second SCR downstream 2 therefrom. There is a first urea doser 3 adapted to dose urea/reductant upstream of the first unit and a second urea/reductant doser/injector 4 adapted to inject urea/reductant upstream of the second unit (and downstream of the first unit) and shown in figure 1, where SDPF stands for SCR on DPF. There is also located a NOx sensor 5 as shown. Typically a NOx sensor 7 is required to measure the NOx into the system (into the first SCR unit) and used as an input in control. Prior art system include a further NOx sensor 8 located between first and second SCR unit

In the figure, the “a” represents the NH3 (slip) form the first catalytic unit and “b” represents the NH3 flow provided by the second urea doser. Background to Closed-loop (CL)

NOx conversion efficiency depends on the catalyst temperature. As a result, SCR control generally involves: an open-loop dosing of reducing agent based on maps or based on chemical modelling of the SCR system; closed-loop correction of the dosing with a post- SCR exhaust sensor, typically a NOx sensor. It can also be regarded that the closed loop use of the post SCR exhaust sensor is used generally to be fed back (e.g. via a model) to provide fine control.

SCR catalysts used on passenger cars are usually based on Cu or Fe zeolites, which exhibit a significant NFb storing capacity. This capacity acts as a buffer for NOx conversion and NFb slip. Drift occurs when SCR units e.g. deteriorate with time and thus lose efficiency. Various systems are know where such drift is corrected using a closed loop feedback control wherein a downstream NOX sensor is used in a feedback control loop to correct or adjust the urea command signal for the doser corresponding to that SCR unit. In such systems, as described in e.g. a model/map is typically provided for first and or SCR units which may have various model inputs including e.g. NOx in (e.g. from NOx sensor 7), and NOx out from downstream NOx sensor. The models may have outputs of NH3 flow and stored NH3 mass which is used to determine a urea command signal for the respective doser. Stored map/model parameters such as NOx conversion efficiency is stored in the model and is also used to provide the urea dosing command signal. So in order to compensate for drift, i.e, a change in NOx conversion efficiency, a downstream NOx sensor signal is either input to the model(s) or input to further control blocks to provide corrected urea injection command for the urea injector. A problem with using such control methodology in dual systems with first and second SCR units and respective dosers is that control becomes more complex. This also requires more than one NOx sensor.

A specific problem is drift of the second SCR unit and dosing system. If the NOx conversion efficiency conversion of the second SCR changes, it is difficult to determine this and if not accurately determined, even when the second SCR is turned off and only the first doser is operational, the system will not operate according to requirements. It is difficult and complex to accurately control and requires extra sensor 8 in order to do this. The inventors have determined a method whereby characteristic parameters such as the NOx conversion efficiency of the second unit can be accurately determined for use in the control of dual systems, both when the second SCR unit is used on its own and also when the first SCR unit is used on its own, and further more requires no additional 8 NOx sensor.

With regard to a dual SCR unit/doser system, the second (downstream) doser is not meant to be used in most of normal vehicle operations, while the first doser is meant to be off when the second doser operates. The feed-forward control presented in e.g. the Applicant co-pending application then favoritizes single operation of one or the other doser.

Figure 2 shows a control system which may be employed in aspects of the invention. The bottom portion of the figure represents an exhaust system similar to that of figure 1.

The top portion shows how the control of a doser controller 12 (for doser 3 ) and doser controller 13 (for doser 4) is performed. Essentially there are two models; a model 10 of the first SCR unit and a model 11 of the second SCR unit. A NOx sensor signal from a downstream sensor 5 (see figure 1) is input to both model of the first SCR unit and also the model of the second SCR unit; the signal is important to provide a value for closed loop control. The models have various inputs as is well known in the art and shown in the figure.

It is important to understand that with respect to each SCR model and its respective doser controller , these may be combined or envisaged as one unit.

A further important point is that there may be other inputs not shown such as temperature sensor inputs or (e.g. demand) inputs form the ECU which are not shown.

Model 10 for the first SCR has an input from upstream NOX sensor 7. The outputs form the dosers are desired (demand) urea flow for the respective dosers. The first model 10 determines a NH3 slip form the model and a NOx output and these are input to the second model. The first and second model determine first and second SCR unit NH3 stored (model) which are input to the respective dosers. Inputs to the doser controllers are first and second NH3 stored target respectively. The stored Nh3 targets are basically calibration maps (in the ECU) which in advanced system included “smart” modifiers that we described in previous patents

. Input to the first doser controller is the desired SCR1 NH3 slip, which may be an output of the second SCR doser controller (or model), Input to the second doser controller (or model) may be the NH3 slip fraction demand.

The respective SCR models determine parameters of characteristics for the respective SCR units such as NOX conversion efficiency for each of the first and second SCR units. In addition, the NH3 stored on each SCR unit is determined.

Depending on the NOx conversion efficiency, and other inputs, a urea command signal is determined which is sent to the respective doser. The first SCR model will also determine a NH3 slip value, that is NH3 going from the first SCR unit doser to the second, which is input to the second SCR model. So examples of the invention use some known techniques of processing of the NOx sensor signal employs known features of the embedded control strategy, such as an SCR catalyst model which determines NOx conversion efficiency, as well as outputs e/g. stored NH3 on the SCR unit and the (converted NH3) flow through the unit (slip) model. Such models are known in the art; a particularly preferred SCR catalyst model and NOx conversion efficiency model are described in European patent application no. 07253 090.0, which is incorporated herein by reference.

Description of Invention Basic Example

In examples of the invention, after the operation of the first doser is terminated i.e. switched off, the system is run in open loop (where only the second SCR unit/doser is operational) (i.e. feed forward control) for a period referred to hereinafter as “flushing time”.

A sufficient “flushing time” is provided such where NOx from the engine flows through the system to purge the NH3 stored in the first catalyst. Only thereafter is the system switched to closed loop control (with only second doser operational) and operational parameters of the second SCR (e.g. model parameters) such a NOx conversion efficiency are then updated. This is done in the closed loop control which uses feedback e.g. of the downstream NOx sensor 5.

Thus the first SCR is flushed for sufficient time to purge the NH3 stored on the first SCR. Any subsequent operation of the second SCR unit means that there is no NH3 slip from the first SCR unit. This means the dual system is effectively working as a single SCR system which allows the second SCR model to be updated and any drift corrected.

Refined Example

Again before measurement of NOx out is used to update the second SCR model parameters (NOx conversion efficiency), the flushing time is such that preferably as well as flushing such that the first SCR is depleted of NH3 stored thereon, the total flushing time is such that the operation of the second SCR unit and doser reaches a stable steady state and/or where the NH3 on the second SCR is not overfilled or underfilled. This is because there may otherwise be errors transient errors as a result of switching) or simply because the front doser may have been drifting while the second doser is OK) where the second SCR may be overfilled (or underfilled) even when the first SCR is depleted of NH3. So it is ensured that control of the second SCR reaches a steady balanced state before the appropriate measurements are taken to update second SCR model parameters.

To make an analogy, consider that “red” urea is dosed with the front doser, and that the doser is faulty (injecting too much but it is not known). Both SCRs are then filled with “red” NH3.

When the front doser is switched off and the second activated, and starts dosing “blue” urea, without error (it is not still known if it is without error) , before activating the closed loop on the second doser/SCR, it is needed to make sure all the red urea has been replaced by blue urea or by no urea at all (in the first SCR), and we know that the red urea will only disappear from both SCRs thanks to NOx or reduction of NH3 storage capacity=f(T) in both SCRs. It is not required for catalysts to actually be empty of NH3, but it is required that all the red NH3 is replaced by blue NH3, because we the CL to correct blue urea (/NH3) only and not the red. So in such a refined embodiment the flushing time may be such that there is also no NH3 stored on the second SCR unit which has come from the first SCR unit. How this is done will be explained in example later.

In a preferred example as will be explained, the flushing time is minimized, which also minimize the time there is no closed loop control (open loop only).

After, the flushing period, the system is switched from open loop to closed loop control with just the second SCR unit and doser operational, and any errors/drift in the second SCR system can be determined and characteristic parameters, e.g. the NH3 conversion efficiency, in the second model can be updated. After this when switching to using just the first SCR system /doser, the second SCR model is fully accurate and the NOx sensor downstream can be used with the accurate second SCR model to update the model parameters (e.g. NOX conversion efficiency) of the first model, as will be explained hereinafter with respect to “Further Subsequent Refinements”.

Further Subsequent Refinements

When the rear doser has been off sufficiently long (again after a flushing period) the methodology employed in subsequent closed loop control of the front SCR system assumes that any drift detected by the rear NOx sensor can be imputed to either the front doser or the front catalyst (SDPF) because the rear SCR is supposed to degrade much less than the front SCR (SDPF). The reason is that aging mainly consists in thermal degradation or poisons, which first impact SDPF. Thus subsequent to the methodology described in the examples, after a sufficient time which may include a second flushing period, the system is switched to operation of the first doser only in closed loop control where the downstream NOx sensor is used in feed back control and the operational control parameters (e.g. stored parameters) in the first model, such as NH3 efficiency of the first SCR unit) 10 are updated. During the second flushing period the system is run in open loop (feed forward control) and the flushing time may be that required to deplete NH3 on the second SCR unit.

Using techniques according to all the above examples, only one post SCR unit NOx sensor need to be used. So according to examples methodology robustly correct independent drifts of the individual dosers and SCR catalysts using closed-loop control based using the above techniques.

An important point is that the drifts of the SCR units/doers units in a dual system can be performed independently . The term “flushing” as used hereinafter to describe when NOX is allowed to flow through the system for a while so as to deplete the NH3 stored in the first SCR (so there is subsequently no slip of NH3 to the second doser) and optionally additionally to ensure the (subsequent) operation second SCR /doser system reaches a steady stable state where e.g. none of the NH3 on the second SCR unit has come from slip (the first SCR unit).

So a further advantage of correcting the second SCR in this warm operation is that the corrected model of the second SCR can then be assumed to be correct when the closed loop operates on the front doser and front SCR (SDPF). This is important because the closed loop control only relies on the rear NOx sensor, which then measures the total conversion efficiency and NH3 slip of the complete SDPF+SCR system. Being able to correct the second SCR individually is then a major advantage for a robust control.

When operating at higher temperature, such as DPF regen, the front doser is shut- off. When the front doser has been off long enough (again see “flushing” below), any impact of the front (first SCR and doser) on the rear SCR catalyst of a drift from the front doser or SCR (SDPF) is assumed to have been “erased” and the closed loop control then fully impute potential tailpipe NOx/NH3 deviations to the second doser and/or second SCR.

Example of Flushing time Calculation

Example 1

As explained above, when alternating from one doser to the other, the closed loop control will not be accurate and so preferably disabled for some time (flushing time), to make sure that a potential drift of one doser (/SCR catalyst) is not considered by the closed loop as a drift of the doser currently being operated.

In methods according to the invention, the control then introduces a feature which determines an adequate flushing time and (preferred) keeps this to a minimum; this also minimizes the time when the closed-loop is de-activated. This flushing period is based on the embedded chemical model of each SCR catalyst. During deactivation of the closed loop control during the flushing period there is preferably open loop control, though this is not essential. The figure 3 above shows how the flushing quantity and time can be calculated. The figure is broadly divided into a top portion and bottom portion.. In the top portion is shown the principle of how the first SCR unit flushing quantity P T Scr 2 scr flush nh3 qty nvv designated with reference numeral 21 is determined that is the quantity of stored NH3 which needs to be depleted from the first SCR unit during a flushing period.

The bottom portion shows how the flushing quantity for the second unit P T Scr 2 scr 2 flush nh3 qty nvv (22) is determined, that is quantity of stored NH3 which should be preferably also be needed to be depleted from the second SCR unit during a flushing period. p t scr capacity corr and p t scr2 capacity corr are respective NH3 storage capacities of the SCR unit (and this may be multiplied by a calibratable safety margin). In any operating condition, regardless of any drift in the system, the control can safely assume that any of the SCR catalyst has not stored more than their respective p t scr capacity corr and p t scr 2 capacity corr mg of NH3.

The calculation is enabled when the first doser is commanded off. At first calculation step, P T Scr 2 scr flush nh3 qty nvv and

P T Scr2 scr2 flush nh3 qty nvv are initialized to their respective p t scr capacity corr and p t scr 2 capacity corr mg of NH3.

P T Scr nh3 flow stoich (designated by reference numeral 25) and P T ' Scr 2 nh3 flow stoich (designated by reference numeral 26) represent the NH3 equivalent of inlet NOx mass flows in each respective catalyst/catalytic unit. The first value 25 may be obtained from the upstream NOx sensor. The second may be obtained from the first SCR model 10 (NOx-out model see figure 2); this may be calculated on the basis of NOx in (to the first SCR and the assumed modelled NH3 conversion efficiency of the first SCR).

For each SCR unit a NOx conversion efficiency table is provided, 23 and 24 respectively for the first and second SCR unit. They each have input from the respective current catalyst temperature. The output is the corresponding maximum possible efficiency of each catalyst. Considering that the SCRs are full of NH3 (65535mg = max of the range) is required because the catalysts are potentially filled with NH3 if the doser is stuck open or drifting.

For each SCR unit, the respective values of P T Scr nh3 Jlow stoich 25 and P T Scr2 nh3 Jlow stoich 26 are multiplied by the respective NH3 conversion efficiencies of the SCR units, to determine the amount of NH3 depleted from the respective SCR units (30,31). The conversion efficiencies are designated P T SCR NOX CONV EFF APM and P T SCR2 NOX CONV EFF APM for first and second SCR units respectively. The potentially converted NH3 -equivalent NOx are then decumulated from P T Scr 2 scr flush nh3 qty nvv and P T Scr2 scr 2 flush nh3 qty nvv. by the blocks generally shown as 30 and 31 respectively.

P T Scr 2 scr flush nh3 qty nvv and P T Scr2 scr 2 flush nh3 qty nvv, which are an indication of the flushing state of each SCR (in terms of noisy impact on the other doser / SCR drift) is continuously limited by p t scr capacity corr and p t scr 2 capacity cor r, which greatly reduces the flushing period in case of temperature increase.

In one example the flushing period may end when the NH3 stored on the first SCR unit is calculated to be substantially zero or less.

In a preferred embodiment when both P T Scr 2 scr flush nh3 qty nvv and P T Scr2 scr 2 flush nh3 qty nvv reach 0, the complete SCR system (front doser + SDPF + rear doser + SCR) is assumed to have been “cleaned” (/flushed) from any impact of front doser and/or first SCR (SDPF) drifts. The closed-loop on the second doser and second SCR can then be activated.

Flow Chart

The flow chart of figure 4 illustrates also the how the flushing period may be calculated and mirrors essentially the process of figure 3.

The process is initiated or starts at step block SI. The methodology proceeds to block S2 where whether it is determined if the first doser is shut off. i.e. P T Scr urea mass flow = 0.

If not the process returns to step S2 and continually monitors whether the doser is shut off.

When the doser is shut off the process moves to implement two sub processes/routines shown in blocks 100 and 101. In the first block 100, in step S3 the maximum NH3 quantity in the first SCR (is initialized to be set to maximum corrected capacity. P T Scr 2 scr flush nh3 qty nvv = p t scr capacity corr . In enhanced embodiments the NH3 stored may be provided more accurately by calculation e.g. from the model/operational history

In the next step S4 the potentially converted NH3 equivalent NOx is calculated by multiplying the parameters P T Scr nh3 flow stoich x P T SCR NOX CONV EFF APM. The parameters are defined and determined as described above with reference to figure 3.

In the next step S5 the potentially converted NH3 equivalent NOx is subtracted from the current maximum NH3 quantity to give a corrected NH3 capacity (corrected value of the NH3 stored on the SCR unit) .

In the next step S6 it is determined if the result is lower than the corrected SCR capacity. If so the maximum NH3 capacity is set to the result at step S7. If not at step S8 the maximum NH3 quantity is set is set to the corrected SCR capacity. After steps S7 or S8 the process proceeds to step S9

There is a similar process shown in the right hand side box with respect to the second SCR. In step the second routine, in step S 10 the maximum NH3 quantity in the second

SCR is initialized to be set to maximum corrected capacity.

P T Scr 2 scr 2 flush nh3 qty nvv = p t scr 2 capacity corr

In the next step Sll the potentially converted NH3 equivalent NOx is calculated form multiplying the parameters P T Scr 2 nh3 flow stoich x

P T SCR2 NOX CONV EFF APM. In the next step S 12 the potentially converted NH3 equivalent NOx is subtracted from the current maximum NH3 quantity to give a corrected SCR NH3 stored (capacity). In the next step S 13 it is determined if the result is lower than the corrected second SCR capacity. If so at step S14 the maximum NH3 capacity for the second SCR is set to the result of step S13 . If not at step S15 the maximum NH3 quantity is set is set to the corrected second SCR capacity. After steps S14 or S15 the process proceeds to step S9

At step S9 it is determined if the maximum NH3 quantity for the first SCR and for the second SCR is equal to zero. If so the process proceeds to step S16 where it is concluded that the SCR system has had sufficient flushing time. If not the process returns to S2

Some additional advantages are that same control and calibration can be applied to a single doser application. The methods does not require an extra (expensive) NOx sensor between both SCR catalysts and provides robust corrections of individual components (dosers and SCR catalysts