GB2506660B

GB2506660B - Method of regenerating a filter in an exhaust

Info

Publication number: GB2506660B
Application number: GB1217895.0A
Authority: GB
Inventors: Bromham Jim; Demory Romain; Cstellano Javier
Original assignee: Ford Global Technologies LLC
Current assignee: Ford Global Technologies LLC
Priority date: 2012-10-05
Filing date: 2012-10-05
Publication date: 2019-06-19
Anticipated expiration: 2032-10-05
Also published as: RU2641325C2; CN103711547A; CN103711547B; GB201217895D0; DE102013219770A1; GB2506660A; RU2013144811A

Description

Method of regenerating a filter in an exhaust

The invention relates to a method of regenerating a filter in an exhaust, especially a Diesel Particulate Filter (DPF). DPFs are provided to entrap soot particles and the like, to mitigate emissions from vehicles. The regeneration of such a Diesel Particulate Filter (DPF) is the technique used to burn the accumulated soot in its interior. That process involves increasing the engine exhaust gas temperature during a period of time in order to combust the soot and it is usually performed during a real drive cycle, with continuously changing driving conditions. Typically, a regeneration event is carried out automatically by the vehicle ECU every 500 kilometres or so of driving.

The increase in temperature is achieved by a reduction of the admission air mass flow and/or the use of post injections and/or exhaust fuel injection that provides a hot exhaust gas flow and/or a flow of hydro-carbons (HC) that combust across a catalytic converter in order to provide sufficient heat to combust the soot in the diesel particulate filter and therefore regenerate the filter. A particular temperature needs to be reached to combust the soot in the filter, and the amount of post/exhaust fuel required to do this varies with a number of factors. Moreover, if too much fuel is added and the temperature becomes too high, violent soot combustion can occur that can cause internal damage to the filter, hampering its filtering efficiency. On the other hand, a lower temperature will cause the system to spend excessive time in regeneration mode with an impact on fuel economy and engine oil contamination and, in the extreme case, it can lead to a blocked filter.

Conventionally, the quantity of post/exhaust fuel to use is calculated as a combination of an open loop term (feed-forward) and a closed loop term.

The open loop term can be a direct fuel request based on the engine operating conditions or it can be model based, where the requested fuel is a calculation of the amount of energy the exhaust gas requires to heat up to the requested temperature as in the following energy balance:

(1)

Where mEG is the exhaust gas mass flow, C is the exhaust gas specific heat, ΔΓ is the desired temperature increase across an exotherm catalyst and /\H°Comb is the heat of combustion of fuel. Based on that, a desired fuel flow mFu can be calculated.

The closed loop control is usually a PID based implementation as shown in Fig. 1. The feedback is given by thermocouples before and/or after the particulate filter, the error is the temperature difference between the measured feedback and the target temperature and the final output is a post injection and/or exhaust injection fuel flow.

However, due to the slow nature of the system and the slow response of the closed loop in correcting large deviations, the overall fuel calculation is strongly biased to the open loop (the feed-forward contribution).

Hence the quality of the open loop contribution is critical in the fuelling calculation and, in consequence, the quality of the regeneration. However the open loop fuel calculation does not take into account a number of different variables and parameters which might affect its accuracy, such factors include: 1. Fuel delivery inaccuracies in the injectors or exhaust injector 2. Fuel (in the form of vapour) slip across the catalyst which doesn't combust and so doesn't generate any heat. 3. The feed-forward sensor location might not accurately represent the temperature distribution across the front face of the catalyst/DPF 4. Post injection fuel that is absorbed by the engine oil and doesn't reach the catalyst. 5. Catalyst efficiency loss due to aging 6. Environmental conditions that affect the external heat transfer from the catalyst such as vehicle speed, external temperature and so on

7. Fuel composition affecting the combustion specific heat

Thus taking into account these factors in the open loop part of the calculation will provide a more accurate temperature of regeneration leading to a more efficient regeneration.

According to the present invention there is provided a method of regenerating a particulate filter during use of an engine comprising an exhaust, the method comprising: carrying out a first regeneration event, which comprises the steps of: (a) estimating a fuel flow rate required to be added to the exhaust in order to achieve a target temperature; (b) adding fuel at said fuel flow rate into the exhaust; (c) combusting at least a portion of said fuel in said exhaust in order to increase the temperature in the exhaust; (d) monitoring the temperature in the exhaust and, comparing it to the target temperature to determine an error value, wherein the error value is a measure of the error of the open loop fuel; and (e) storing the error value determined in step (d) in a memory device; carrying out a second regeneration event, which comprises the steps of:(f) carrying out step (a) using the stored error value to refine the estimate of the flow rate of fuel which is required to be added to the exhaust to achieve the target temperature;

Thus rather than focusing on the complexities of the open loop calculation (so as to determine a more complex and representative version of equation (1)), the inventors of the present invention have considered that the complexity of the resulting calculations would be difficult and time consuming to model, and in any case, would not allow for individual engine and exhaust variations, aging, or operating conditions.

Instead embodiments of the present invention do not require further complex modelling on further various parameters but instead uses information learned in a previous regeneration event to refine the open loop terms later in subsequent regenerations.

The estimate in step (a) may be based on the current engine operating conditions (open loop fuel).

The temperature monitored in step (d) may be the particulate filter temperature.

Step (f) may include determining a fuel correction factor which is used to refine the estimate of the flow rate of fuel.

In step (d), the error value may be derived from a closed loop component.

The closed loop control, no matter how small its authority (the amount of fuel it is allowed to add) will always try to control the measured temperature closer to the target temperature regardless of the quality of the open loop calculation. Thus, the close loop control activity is a measure of the quality of the open loop calculation. A small quantity of fuel requested by the closed loop means that the open loop calculation is very accurate. However, if the closed loop is contributing with a large quantity, the open loop calculation is poor.

Embodiments of the invention may store the information that, in real time, the closed loop is offering about the performance of the system and the quality of the open loop, and as a result, a correction to the open loop can be permanently stored to enhance its performance in future events.

The memory may be a non-volatile memory such as an electronic control module non-volatile memory.

The data may be stored for a period of time being at least 1 day, normally much longer, such as at least one month or at least one year. For preferred embodiments, the data is stored indefinitely or until overwritten by data obtained from a subsequent regeneration event.

The particulate filter is normally a diesel particulate filter and accordingly the engine is normally a diesel engine. However, for certain petrol engines, a similar filter may be provided in the exhaust and the method according to the present invention also used.

Steps (a) to (f) are normally started in sequence but may continue concurrently for a period of time, such as 5 to 10 minutes.

The method may include conducting a stability measurement before step (f), optionally before step (e), and especially whilst steps (a) - (d) are occurring concurrently, that is determining that the engine is being operated in a relatively stable manner. This may be done before step (e) so that only particularly useful data is stored, but in any case, for such embodiments, is done before the data is used, so that only the more stable data is used in step (f). To conduct a stability measurement, a moving average and moving standard deviation of exhaust gas flow may be calculated whilst steps (a) to (d) are occurring concurrently. If the moving standard deviation remains below a calibrated threshold during a calibrated period of time, the data may be stored in step (e) and may be used in step (f). If the moving standard deviation exceeds the threshold, the calculation may be reset and started again with the same conditions. Thus for such embodiments, the data used in step (f) is calculated as the ratio between the average total fuel over the average open loop fuel used during that interval.

The target temperature may be a function of the amount of soot stored in the filter.

The data used in step (f) may be added to an open loop element of the calculation to increase the quality of the estimation of the fuel flow rate required so the resulting temperature will be closer to the target temperature, ceteris paribus.

Once the stability condition is satisfied and the fuel correction calculated, the average engine speed and load and their standard deviations may be used to calculate the shape of the correction. If using two variables to characterize the engine operating point (for example engine speed and load), the correction map may be calculated as a bivariate binomial distribution, centred in the average speed and load points and with their corresponding standard deviations along the axis.

For certain embodiments a fuel correction factor is determined in step (f) based on the following equation:

Alternatively, the fuel that would have been required during the drive cycle can be calculated regardless of the closed loop fuel and based on the equation:

where mFu(Error) can be calculated using the same expression as to calculate the original open loop fuel

An embodiment of the present invention will now be described, by way of example only, with reference to the accompanying figures, in which:

Fig. 1 is a known PID system;

Fig. 2is a Bivariate normal distribution mapping parameter used in one example of the present invention;

Fig. 3 is a Bivariate correction map for one example of the present invention;

Figs. 4 to 7 are Simulink based models for procedures used in examples of the present invention;

Fig. 8 is a an example of Matlab code used for one example of the present invention/;

Fig. 9 is a graph showing the recorded vehicle speed, engine speed, engine indicated torque and calculated exhaust gas flow recorded for a particular vehicle operation, used in an example of the present invention;

Fig. 10 is a graph showing a pre-DPF target and real temperature being compared and the error between them as used in an example of the present invention;

Fig. 11 is a graph showing variables when processing a stability condition used in examples of the present invention;

Fig. 12 is a graph showing an output model produced for embodiments of the present invention.

During regeneration of a diesel particulate filter, fuel is entered into the exhaust line which combusts and raises the temperature sufficiently in order to burn off the soot accumulated on the diesel particulate filter.

Choosing the optimum amount of fuel and in particular the fuel flow rate to achieve the desired temperature is beneficial to prevent over heating or under heating of the soot and associated problems. Thus the fuel flow rate required is based on a model (open loop) and also includes a feedback element (closed loop).

For closed loop control, it is common to use a proportional-integral-derivative (PID) controller such as shown in Figure 1. A PID controller calculates an error value as the difference between a measured process variable (in this case, the actual temperature in the exhaust) and a desired target (temperature), and attempts to minimize the error by adjusting the process control inputs. A particular advantage of embodiments of the present invention is that the feedback, gained from the PID system comparing target and actual temperature in the exhaust, is stored and used later in subsequent regenerations.

The inventors of the present invention have noted that there are numerous reasons why the open loop calculation may not provide an optimum amount of fuel flow for the required temperature. The model may not be sophisticated enough to take into account all parameters. Moreover, individual components of a particular car may be inhibited or less efficient, which will affect the temperature required, and the use of the car may also affect the regeneration required e.g. it is used in a colder or warmer climate.

The normal feedback loop of a particulate filter temperature is slow and unwieldy. Embodiments of the invention store data from the feedback so that subsequent regenerations start from a closer fuel flow rate to reach the target temperature.

During a regeneration event, some information regarding the quality the regeneration can be extracted from the closed loop fuelling feedback. From the definition of a PID, the closed loop contribution is:

(2)

Equation (1) can be rewritten as:

(3)

Under steady state conditions and assuming that the temperature target is met, the fuel calculated by the PID, mFu (CL} , is a measure of the error of the open loop fuel to achieve the target. A fuel correction factor can be defined as:

(4)

The fuel correction factor represents the fuel required to correct for various disturbances and it required no calibration other than the P, I and D gains tuning. Multiplying the correction factor times the open loop fuel; a corrected open loop fuel can be obtained.

The above was defined under steady state conditions, which are difficult to maintain during normal driving cycles. The condition of steady state depends on the physical properties of the exhaust system and the exhaust gas flow for a particular engine operating point. To decide whether the engine operating conditions are steady enough, a stability check is preferably undertaken.

The exhaust gas flow is calculated as the adding of the engine admission gas flow (measured with a MAF sensor) and the overall fuel used, including post/exhaust injections.

To conduct a stability measurement, a moving average and moving standard deviation of the exhaust gas flow may be calculated. If the moving standard deviation remains below a calibrated threshold during a calibrated period of time, the command to perform the learning as in equation (4) is activated. The fuel correction is thus calculated as the ratio between the average total fuel over the average open loop fuel used during that interval. On the other hand, if the moving standard deviation exceeds the threshold, the calculation is reset and started again with the same conditions. The interval (/15/2) is the time interval where the stability condition has been satisfied.

Once the stability condition is satisfied and the fuel correction calculated, the average engine speed and load and their standard deviations are used to calculate the shape of the correction. If using two variables to characterize the engine operating point (as engine speed and load), the correction map will be calculated as a bivariate binomial distribution, centred in the average speed and load points and with their corresponding standard deviations along the axis.

Example

Fig. 2 shows a bivariate normal distribution using the following parameters. SPEED (average) = 2000 rpm SPEED (standard deviation) = 200 rpm TORQUE (average) = 80 Nm TORQUE (standard deviation) = 30 Nm

These correction map values are confined between 0 and 1. The Figure 2 map is finally multiplied by 'fuel_cor' to calculate the final correction map in fuel units. That means that the correction will be fully applied at the centre of the distribution (the average speed and load of the period considered stable enough) and then blended out according to the standard deviations of the speed and load.

The process above described would be continuously repeated over the course of a drive cycle in regeneration mode. Every time that the stability conditions are met, a new learning is performed and added to the existing correction map. At the end of the regeneration, the correction map is stored within the NVM (Non Volatile Memory) of the engine control unit and will be used and updated in the next regeneration event. Figure 3 shows a map after four learning loops.

In an alternative embodiment of the invention, an alternative to the fuel correction calculation is used, which avoids the use of the closed loop control data. Rather, the open loop error of a previous regeneration event is used to calculate the fuel required during the drive cycle instead of the closed loop fuel. Thus, equation (4) can be expressed as:

(5)

Where mPu(Errorj can be calculated using the same expression as to calculate the original open loop fuel

(6)

This alternative embodiment has the benefit of being independent of the performance of the closed loop control. The PID can introduce some noise that could affect the learning. And conversely, the PID gains do not have to be compromised with the constraints of the learning algorithm. On the other hand, various inaccuracies may apply to this calculation, so the system may require more learning iterations to achieve the same performance as with a correct learn from the closed loop.

Figures 4 to 7 show Simulink based model descriptions and the terms are defined in the Appendix herewith. In Figure 4, the inputs and outputs are defined and the final calculation of the accumulative correction matrix "correction" is performed. This map represents the accumulative effect of all the possible inaccuracies of the system as listed in the appendix.

Figure 5 shows the Block "Conditions" where a stability check is performed in the block "statistics" and the fuel ratio as in equation 4 is calculated in the block "fuel ratio". A reset of the block "statistics" is performed when one or more of the following conditions is satisfied: • Regeneration is not active • Stability condition is not met • Successful learn

In Figure 6 the rolling average and rolling standard deviation of exhaust gas flow (EGF), engine speed (n) and engine load (tqi_sp) are calculated. The input "Rst" resets all six calculations as explained above. The exhaust gas flow calculations are used for the assessment of the cycle stability. The engine speed and load calculation are used for the calculation of the correction matrix. However, different combinations could be implemented to perform the stability check and/or the final correction (for example, stability could be assessed by checking both engine speed and load and the correction can be calculated based on exhaust gas flow).

In Figure 7 the ratio of the open loop over overall fuel is calculated. However, as the quality of the data increases toward the end of the successful cycle (as the temperature approaches its target) the average can be weighted towards the end of the cycle. The curve "Weighted Average" depends on the average exhaust gas flow as the speed of the signal approaching its target depends on the time constant of the system, and ultimately on the exhaust gas flow.

In Fig. 8 the embedded Matlab code performs the calculation of the bivariate normal distribution based on engine speed and engine load statistical data through the latest learn cycle. The distribution is centred on the average speed and load and its standard deviations correspond to the standard deviations of the speed and load.

Example

Feeding a real driving cycle into the Simulink model described in the previous section:

Figure 9 shows the recorded vehicle speed, engine speed, engine indicated torque and calculated exhaust gas flow are depicted. The first part of the cycle has a more transient operation mode, while the latter is of a steadier nature. It would be expected for the learning stability check to be fulfilled more often in the final part of the cycle.

In Figure 10, the pre-DPF target and real temperature are compared, the error between them (and used by the closed loop control) is depicted below and finally, the comparison between open loop fuel and total fuel that will be used for the fuel correction calculation.

The stability condition can be seen at work in Figure 11. The top line is the rolling average exhaust gas flow 20 and the lower line 22 is the rolling standard deviation. Another line 24 is the threshold of the maximum allowed standard deviation. If the rolling standard deviation exceeds the maximum allowed standard deviation, the calculation is reset as well as the timer 26. When the standard deviation remains under the threshold and the timer reaches the calibrated waiting time 28, a 'learn' is performed and the calculation is also reset.

The output of the model is shown in Figure 12. This graph shows the fuel correction calculated as in equation (4) and each one of the learns means that the stability was satisfactory during a period of time. As expected, during the later part, the learns occur more often, and they are all similar. It should be noted that these corrections are just calculated but not fed back to the real system. That is why the error persists after the correction is calculated and why the corrections quantities don’t converge to 1.

Thus, embodiments of the present invention continuously improve the regeneration performance of the future regenerations based on historical data.

Embodiments of the invention provide an adaptive system that corrects the open loop fuelling calculation based of the performance of the system on the previous or current regeneration.

Appendix - Simulink based model description OUTPUTS:

INPUTS:

CALIBRATABLES:

Claims

Claims

1. A method of regenerating a particulate filter during use of an engine comprising an exhaust, the method comprising: carrying out a first regeneration event, which comprises the steps of: (a) estimating a fuel flow rate required to be added to the exhaust in order to achieve a target temperature; (b) adding fuel at said fuel flow rate into the exhaust; (c) combusting at least a portion of said fuel in said exhaust in order to increase the temperature in the exhaust; (d) monitoring the temperature in the exhaust and, comparing it to the target temperature to determine an error value, wherein the error value is a measure of the error of the open loop fuel; and (e) storing the error value determined in step (d) in a memory device; carrying out a second regeneration event, which comprises the steps of: (f) carrying out step (a) using the stored error value to refine the estimate of the flow rate of fuel which is required to be added to the exhaust to achieve the target temperature.
2. The method as claimed in claim 1, wherein the estimate in step (a) is based on an open loop calculation using parameters based on the current engine operating conditions. 3. The method as claimed in claim 1 or 2, wherein step (f) includes determining a fuel correction factor which is used to refine the estimate of the flow rate of fuel. 4. The method as claimed in any of the preceding claims, wherein, in step (d), the error value is derived from a closed loop component. 5. The method as claimed in any of claims 1 to 3, wherein, in step (d), the error value is derived from an error value obtained after carrying out open loop control. 6. The method as claimed in any of the preceding claims, wherein the memory is a non-volatile memory. 7. The method as claimed in any of the preceding claims, wherein the error value is stored for a period of time being at least one day. 8. The method as claimed in any of the preceding claims, wherein the particulate filter is a diesel particulate filter and the engine is a diesel engine.
9. The method as claimed in any of the preceding claims, wherein the temperature monitored in step (d) is the particulate filter temperature. 10. The method as claimed in any of the preceding claims, including conducting a stability measurement before step (f) and during at least step (c). 11. The method as claimed in claim 10, wherein the stability check comprises monitoring a moving average and moving standard deviation of exhaust gas flow during step (c) and, if the moving standard deviation remains below a calibrated threshold during a calibrated period of time, the error value is stored in step (e) and used in step (f) and, if the moving standard deviation exceeds said threshold, the error value is not used in step (f). 12. The method as claimed in any of the preceding claims, wherein the error value used in step (f) is added to an open loop element of the calculation to increase the quality of the estimation of the fuel flow rate required.
13. The method as claimed in any of the preceding claims, wherein the average engine speed and load and their standard deviations are used to calculate the shape of the correction and where two variables are used to characterize the engine operating point, the correction map is calculated as a bivariate binomial distribution, centred in the average speed and load points and with their corresponding standard deviations along the axis. 14. The method as claimed in claim 3 or any preceding claim dependent on claim 3, wherein the fuel correction factor is determined in step (f) based on the following equation:

wherein fuel_cor is the fuel correction factor,

is the flow value of the open loop fuel to achieve the target temperature,

is the flow value of the

closed loop fuel to achieve the target temperature, fi and t2 represent different temperatures.
15. The method as claimed in claim 5 or any preceding claim dependent on claim 5, wherein the fuel correction factor is determined in step (f) based on the following equation:

wherein fuel_cor is the fuel correction factor,

is the flow value of the open loop fuel to achieve the target temperature,

is the measure of the error of the open loop fuel in the first regeneration event, fi and t2 represent different temperatures.