GB2439566A

GB2439566A - Cold adaptive fuelling

Info

Publication number: GB2439566A
Application number: GB0612704A
Authority: GB
Inventors: Nick Dashwood Crisp
Original assignee: Ford Global Technologies LLC
Current assignee: Ford Global Technologies LLC
Priority date: 2006-06-28
Filing date: 2006-06-28
Publication date: 2008-01-02
Also published as: DE102007028380A1; GB0612704D0

Abstract

A method of operating an internal combustion engine comprising the steps of referencing a fuelling map to determine a nominal amount of fuel to be delivered to the engine and modifying the nominal amount during periods of closed loop feed back control based on a signal from an exhaust gas oxygen sensor, in order to achieve a target air fuel ratio characterised by storing within a further adaptive correction map, the modifications applied to the nominal amounts of fuel delivered during periods of closed loop feedback control when the engine coolant temperature is within a predetermined range, and referencing both the fuelling map and the further adaptive correction map during subsequent open and closed loop operations of the engine in order to better achieve the target air fuel ratio when the engine coolant temperature is within the predetermined range.

Description

-1 -2439566 Cold Adaptive fuelling This invention relates to engine

control and more particularly to adaptive feed back control of the engine in different operating conditions.

Engines are controlled by engine control units (ECUs) which alter the fuel delivered to each cylinder and the timing of the spark based on the engine speed and the engine load, as determined from the mass air flow (MAF), manifold absolute pressure (MAP) or the throttle angle position.

The ECU includes a map for every given engine speed and engine load level, there is stored a fuel quantity and spark ignition angle required for the engine to run properly.

Running properly is dependent on many factors, the principle being the air fuel ratio supplied to the engine. By monitoring the exhaust gas oxygen content, which is the residuals of the cylinder combustion, it is possible to determine how well the engine is burning the fuel introduced to it. The admitted air fuel ratio (lambda A) of the charge is usually defined by reference to a stoichiometric mixture (A = 1). Stoichiometry is the chemical ratio of reagents required to achieve complete oxidation of the fuel.

Complete combustion is difficult and often undesirable to achieve whereas imperfect combustion leads to undesirable emissions such as unburned hydrocarbons and NOx. For this reason modern vehicles employ exhaust aftertreatment devices such as three way catalysts which ensure that tail pipe gases are not excessively harmful to the environment. In order for these devices to function at their optimum efficiency, the feedgas supplied by the engine must have an air fuel ratio within a suitable range. With this in mind, the mapping of the engine is crucial to ensure the correct feedgas air fuel ratio.

In order to ensure the engine is operating at the desired air fuel ratio (AFR), the oxygen content of the exhaust is measured using an exhaust gas oxygen (EGO) sensor. EGO sensors may be of the heated or unheated type.

Using data from such EGO sensors the ECU can determine if the engine is achieving the desired target of exhaust air fuel ratio. This is necessary because estimation of the mass of fuel needed per combustion stroke can vary in dependence upon many factors, including, for example, air humidity or variations oxygen concentration.

Using feedback control from EGO sensors, the ECU can compensate and correct the fuel map in order to achieve the desired target for that given engine load and engine speed.

Such operation is known as closed loop control because the fuel introduced into the engine is varied based on the oxygen detected in the output gases of the engine.

Closed loop control is not always possible as it is usually employed when the target air fuel ratio is approximately stoichiometrjc. This is because the EGO sensor employed is typically a binary sensor. These are also referred to as narrow band sensors whose output voltage appears to change very quickly (over a narrow band of AFRs) in the region of a stoichiometrjc fuel mixture. When the mixture is far removed from stoichiometry, or lambda is not approximately equal to one, the sensor output voltage remains constant.

Furthermore, closed loop operation can only be utilised when the sensor is operational. The sensor is typically not operational immediately after engine start because of a finite time taken for the component to heat to operating temperature, and the risk of cracking from water vapour in the exhaust. The fuelling map is a table or 3D graph, the surface of which indicates a desired value of lambda (air fuel ratio) for each engine operating point, usually as a function of engine speed and load. Each target air fuel ratio will mean the ECU has to supply a specific quantity of fuel for the amount of air it has measured.

Variations between engines due to, for example, engine wear, fluctuations in oil and fuel pressure, tolerances between components, friction and many others lead to differences between the target AFR and the measured AFR as a result of the fuel amount supplied being incorrect. Another critical error is the variation in lost & transient fuel due to for example, valve deposits, carbon build-up, and ring tolerances. Further errors are introduced due to air charge drift effects such as valve lash and assembly tolerances.

Closed loop feedback control, when applicable, enables the engine to correct the amount of fuel supplied in order that the AFR measured from the EGO sensor(s) is acceptably close to the target AFR within the fuel map.

At the same time, the ECU utilises an adaptive learning function which monitors the error created between target and measured AFR. In learning what compensation is required for a given engine speed and load, the ECU creates an additional correction map or develops a suitable mathematical algorithm to alter the delivered fuelling and minimise the error. The map may be a table of values each corresponding to a specific engine speed and load, while an algorithm would be a mathematical function or operator applied to the nominal fuel quantity within a preset region of the fuelling map.

The intended result of this adaptive learning function is to reduce the correction of the delivered fuel quantity provided by closed loop feedback. This will mean that at times when the engine cannot be operated under closed loop control, such as when the EGO sensor is not yet enabled, or in regions where the engine is operating excessively rich, such as under power, or lean such as at low load, the corrections learned by the adaptive learning function enable the engine to operate closer to or at the intended AFR.

Adaptive learning stores arid updates the corrections it learns in order to minimise target error. This means that as the engine characteristics vary over time, the corrections will continue to change. This is of particular importance as the engine changes as it ages. The adaptive learning function automatically accounts for long term and also short term changes such as oil degradation or a temporary fuelling change due to a tank-full of a different fuel blend. Long term changes may include cylinder wall and piston ring wear, bearing wear and gradual deposit of carbon inside the exhaust manifold causing turbulence and increased back pressure. Carbon deposits on the valves & pistons also have a measurable effect on transient & lost fuel). Another example is sensor and actuator drift.

Experimentation has shown that the same adaptive learning correction is not suitable for all running conditions as compensations that have been learned and stored under certain operating Conditions are not always applicable to others.

With a view to mitigating the foregoing disadvantage, the present invention provides a method of operating an internal combustion engine as set forth in claim 1 of the appended claims.

Preferably the adaptive learning correction is continuously modified as a function of coolant temperature.

It is preferable if the adaptive learning correction is further modified as a function of any of mass air flow, manifold absolute pressure, exhaust gas oxygen content, ambient pressure, fuel pressure, engine oil temperature, engine oil pressure and ambient temperature.

It is further preferable if when the engine coolant temperature is within the predetermined range, the fuel delivered to the engine is of a different type compared to the fuel delivered when the coolant temperature is outside that range.

Advantageously the temperature range may be below -10 C, between -10 C and 15 C, between 15 C and 40 C, between 40 C and 70 C and above 70 C.

The invention will now be described further, by way of example, with reference to the accompanying drawing in which: Figures 1 and 2 show graphical representations of the fuel maps and adaptive learning corrections of different embodiments of the present invention.

The present invention improves over the prior art by recognising that the same adaptive learning function of conventional ECUs fail to distinguish between hot and cold operating conditions which have a substantially different impact on the fuel map.

Traditionally the adaptive learning correction gathers the correction values to be stored in a table when the engine is hot, i.e. the coolant and oil temperatures are stable at normal operating temperature. Applying these correction values to cold running is not ideal because the values are not able to take into account the short term differences between hot and cold running engines. These short term changes can have a significant impact on the engine's volumetric efficiency, for example valve stem growth and tappet clearances. The contribution to the correction values created by long term changes such as engine wear will of course be constant regardless of whether the engine is running cold or not.

The cold running period is crucial for reducing cumulative emissions as it coincides with the period prior to catalyst light-off. Emissions during this period contribute enormously to the total car emissions during the average drive cycle as once the catalyst is up to operating temperature, its conversion efficiency can be maintained at around 99%. With this in mind, any method of increasing the accuracy of the fuel map to reduce emissions prior to catalyst light-off is beneficial.

For this reason it is advantageous to provide a separate adaptive learning function within the ECU that monitors engine operating performance through feedback from the EGO sensor(s) and applies a specific cold operating correction.

During the initial period after an engine is first started, the EGO sensor is typically disabled. The reason for this is that the sensors require heating in order to work, and the physical conditions within a cold exhaust system combined with the heating of the sensor can lead to cracking of the sensor element and premature ageing. This may be as a result of condensed water in the exhaust dripping on to the hot sensor. For this reason the sensor is not enabled until it is deemed safe to do so, in order to maintain the longevity of the sensor.

The delay in enabling the sensor after engine start can vary between 5 and 30 seconds from engine start. The exact time for a given engine is dependent on the nature of the engine configuration, the sensor and other factors such as the distance of the sensor from the cylinder, and the shape of the exhaust manifold and or pipe work.

The fuelling during the catalyst light off period is substantially leaned out in order to minimise the tailpipe emissions. This is difficult since lean operation tends to reduce torque and increase the propensity of the engine to stall, run roughly or misfire. These effects are detectable by the driver and therefore undesirable. With such potential downfalls, it is important to control the degree of lean operation as carefully as possible in order to strike the right balance between emissions and driver satisfaction.

This emphasises the need to apply a more accurate correction tailored to open loop cold operating conditions.

For ease of programming and to reduce the memory requirement for simpler ECUs, in its broadest form the present invention suggests one additional adaptive correction map in order to account for cold engine operation. In such an embodiment, an appropriate cut off temperature below which the cold adaptive function would be employed is 30 C. This can be seen graphically by referring to figure 1.

In figure 1, the main 9 by 4 grid represents a portion of the normal or "hot" adaptive corrective table. The values in each grid square are learned during periods where closed loop feedback is enabled. Each grid square relates to a specific location on the fuelling map, having a value of air quantity given by the MAP or MAF value (application specific) and in this case, the engine speed in RPM. Each grid square provides the perturbation to the main fuel map for that value of air mass (or pressure) and engine speed.

The perturbation may be a function of injector duty cycle or injector opening duration.

The regions A, B, C and D of figure 1, are areas of the map relating to acceleration, cruise, idle/light cruise and deceleration respectively.

When the engine is cold and operating in the lower left region of the table, the perturbation to the main fuel map is provided by the smaller 3 by 3 grid. When operating outside the area of this "cold" correction map, the standard or "hot" correction map is applied. The "cold" map is of smaller size because the engine only operates for a brief period at cold conditions, so the opportunity for acquiring relevant data is small.

This area of the "cold" correction map contains regions 31, Cl and Dl which relate to cold cruise, cold idle and cold deceleration respectively.

The engine is deemed to be cold when the coolant temperature is in a predetermined temperature range. The important thing to note is that the corrections values stored in the "cold" correction table are created when operating under closed loop control whilst engine coolant is within the range specific to that "cold" correction table.

This threshold can vary but ideally temperature ranges of less than -10 C, between -10 C and 15 C, between 15 C and 40 C, between 40 C and 70 C, and above 70 C should be considered. These are important engine operating temperature ranges in which the engine will benefit from an individually calibrated, rather than a generic, adaptive correction function.

The multiple ranges given above suggest the possibility of having 5 specific adaptive "cold" correction maps. With such information available from 5 distinct tables, (see figure 2) it is probable that interpolation between the different correction tables could produce a correction that appears to vary as a function of engine coolant temperature, although it would only have 5 distinct calibrated values.

The equivalents to Bi, Cl and Dl are shown here are B5, C5 and D5 as the fifth (coldest) adaptive correction map obscures the fourth, third, second and first.

It is conceivable with sufficient processing power to implement a correction fuelling table that varies continuously as a function of coolant temperature, with better temperature resolution than that provided by the above embodiment.

A further embodiment is intended for use with vehicles operating dual fuels. These employ additional fuel tanks for running an internal combustion engine on a choice of petroleum and an alternative fuel. The alternatives include LPG and ethanol, for example Brazilian ElOO.

These latter fuels are more sustainable as they do not rely on depleting fossil fuel reserves. Ethanol for example, has the disadvantage of remaining liquid at low engine operating temperatures meaning it is more difficult to use in colder climates before the engine is warm. Some vehicles are therefore provided with a small tank of petroleum to allow the engine to warm up to a point where the main fuel is vaporised and therefore usable.

Such engines typically employ an adaptive correction applicable to when the engine is running on the primary fuel -10 - (alcohol etc.). It is not appropriate to apply this adaptive correction to the starter fuel (petroleum) since the burning qualities, injection equipment and the air fuel ratios of each fuel are different.

The present invention is advantageous in such arrangements because it provides an additional adaptive learning function which applies a correction to the fuelling delivered to the engine when the engine is operating on a secondary fuel. The corrective value is learned when the engine is operating under closed loop feedback using the secondary fuel, this usually coincides with the engine coolant temperature being within a predetermined range such as described with reference to single fuel embodiments above.

Claims

-11 -Claims 1. A method of operating an internal combustion engine

comprising the steps of; referencing a fuelling map to determine a nominal amount of fuel to be delivered to the engine and modifying the nominal amount during periods of closed loop feed back control based on a signal from an exhaust gas oxygen sensor, in order to achieve a target air fuel ratio, storing within an adaptive correction map the modifications applied to the nominal amounts of fuel delivered during periods of closed loop feedback control, referencing both the fuelling map and the adaptive correction map during subsequent open and closed loop operations of the engine in order to better achieve the target air fuel ratio, characterised by storing within a further adaptive correction map, the modifications applied to the nominal amounts of fuel delivered during periods of closed loop feedback control when the engine coolant temperature is within a predetermined range, and referencing both the fuelling map and the further adaptive correction map during subsequent open and closed loop operations of the engine in order to better achieve the target air fuel ratio when the engine coolant temperature is within the predetermined range.

-12 -
2. A method as claimed in claim 1, wherein the modified value of the fuel quantity is stored as an algorithm.

3. A method as claimed in claim 1 or 2, wherein the adaptive learning correction is continuously modified as a function of coolant temperature.

4. A method as claimed in any preceding claim, wherein the adaptive learning correction is based on feedback from an exhaust gas oxygen sensor.

5. A method as claimed in any preceding claim, wherein the adaptive learning correction is further modified as a function of any of mass air flow, manifold absolute pressure, exhaust gas oxygen content, ambient pressure, fuel pressure, engine oil temperature, engine oil pressure and ambient temperature.

6. A method as claimed in any preceding claim, wherein when the engine coolant temperature is within the predetermined range, the fuel delivered to the engine is a different type of fuel from when the coolant temperature is outside that range.

7. A method as claimed in any preceding claim, wherein the temperature range is below -10 C.

8. A method as claimed in claims 1 to 6, wherein the temperature range is between -10 C and 15 C.

9. A method as claimed in claims 1 to 6, wherein the temperature range is between 15 C and 40 C.

-13 - 10. A method as claimed in claims 1 to 6, wherein the temperature range is between 40 C and 70 C.

11. A method as claimed in claims 1 to 7, wherein the temperature range is above 70 C.

12. A method of controlling an internal combustion engine substantially as herein described with reference to and as illustrated in the accompanying drawing.