Background of the invention It is common practice to operate three-way catalytic converters using oscillating feed gas air-fuel ratio (AFR).

It is also possible to adjust the frequency and amplitude of these oscillations to account for variations in the oxygen storage capacity of the catalyst with (for example) temperature and ageing effects, as described in US 5,678,402.

The mean value of feed gas AFR may be adjusted away from exact stoichiometry by a small rich or lean bias, to shift the catalyst efficiency in favour of reducing or oxidising reactions at chosen engine conditions in order to optimise overall emissions performance.

Summary of the invention According to one aspect of the present invention, there is provided an engine management system for regulating the quantity of fuel supplied to the engine cylinders of an engine fitted with a three-way catalytic converter, the system being operative in its normal mode of operation to regulate the air to fuel ratio (AFR) such that emissions of hydrocarbons, carbon monoxide and oxides of nitrogen are minimised by the action of the catalytic converter and having a second mode of operation in which the feed gases to the catalytic converter have excess oxygen, the management

system being operative to switch periodically to the second mode of operation and to remain in said second mode only for sufficient time to expose up to the entire volume of the catalytic converter to an oxidising environment.

It has been found that during operation, there is noticeable degradation in the performance of a catalytic converter, especially if the fuel contains sulphur. The basis of the present invention is that due to the oxygen storage properties of the catalytic converter, during normal operation, despite oscillating input AFR, much of the catalyst (towards the rear) is not exposed to oxygen. In this case, the catalyst efficiencies deteriorate over time, but by periodically introducing surplus oxygen, so exposing an increased volume of catalytic converter to oxidising conditions, catalyst efficiencies on returning to the normal fuelling regime can be considerably improved.

The engine management system of the preferred embodiment of the invention, has a normal mode of operation the system is operative to cause the air to fuel ratio (AFR) to oscillate at a first frequency between values that are respectively richer and weaker than stoichiometry and a mode of operation for producing an excess of oxygen in the feed gases to the catalytic converter during which the system is further operative to modulate the AFR oscillations such that the AFR bias measured over several cycles of the first frequency is modulated with an amplitude lower than the AFR oscillations and at a second lower frequency.

This preferred embodiment of the invention allows the oxygenation of the catalyst to be achieved in a convenient manner that does not affect the driveability of the vehicle in which the engine is fitted. There are alternative ways in which such oxygenation of the catalyst can be carried out, for example by switching on a bypass or additional air supply, by deactivating the fuel to one or more engine

cylinders or by switching to an open loop lean burn mode for a time.

Switching to the second mode of operation may be initiated by sensing or predicting that the conversion efficiency of the catalytic converter has degenerated.

The return to normal operation can be carried out either after a time predicted to be sufficient to fully oxygenate a desired volume of catalytic converter or until excess oxygen is sensed by an EGO sensor at the downstream end of the catalytic converter, at which point the entire volume of the catalytic converter would have been exposed to oxygen.

The invention differs from the switching that is normally carried out between lean burn and near stoichiometric modes in that the average excess air used in the present invention is too short to achieve any noticeable improvement in fuel economy and lasts only for long enough to oxygenate the catalytic converter. Indeed, it is important to revert to normal fuelling at the earliest opportunity as the excess oxygen mode risks high NOx emissions.

Brief description of the drawings The invention will now be described further, by way of example, with reference to the accompanying drawing, in which : Figure 1 is a block diagram of a engine fitted with a three-way catalyst, EGO sensors and a management system that regulates the quantity of fuel supplied to the cylinders, Figure 2 is a chart showing the measured emissions at the tail pipe of an engine operated with a conventional management system,

Figure 3 is a chart similar to that of Figure 2 showing the effect of superimposing on the AFR oscillations a low amplitude and low frequency perturbation to vary the bias periodically, in accordance with the invention, and Figure 4 is a bar chart showing the reduction in hydrocarbons, NOx and carbon monoxide achieved by the present invention.

Detailed description of the preferred embodiments In Figure 1, an engine 10 has an inlet manifold 12 into which ambient air enters by way of a mass air flow meter 16.

The intake air flow is controlled by a main butterfly throttle 14 and fuel is added to the intake air through a fuel injector 18. The exhaust air flows through a down pipe 20 to a catalytic converter 22 that comprises two matrices 26 and 28 arranged in a casing 24. The matrices or bricks 26 and 28 consist of a ceramic honeycomb carrying particles of a three-way catalyst. Such a catalytic converter can store surplus oxygen present in the exhaust gases when the engine burns a lean mixture and use this stored oxygen to oxidise carbon monoxide to carbon dioxide and unburned hydrocarbons to carbon dioxide and water when the engine is subsequently operated with a rich mixture. A three-way catalyst can also store NOx and reduce it to nitrogen by reacting it with hydrocarbons and carbon monoxide when the engine is operating with a rich mixture.

In order to minimise all three pollutants, the engine air to fuel ratio (AFR) is oscillated about stoichiometry (k = 1). This is achieved by means an electronic control unit (ECU) 36 that receives signals from two exhaust gas oxygen (EGO) sensors 32 and 34 located upstream and downstream of the catalytic converter 22, respectively, and from the mass air flow meter 16, calculates the appropriate injection quantity on the basis of a stored algorithm and

controls the fuel injectors 18 to deliver the calculated quantity of fuel to the engine cylinders.

Conventionally, the mixture strength is controlled between two fixed AFR's that lie typically up to 4% below and above stoichiometry (i. e. X = 1.04 and k = 0.96), in other words the amplitude of the AFR oscillations is fixed.

The frequency of the oscillations of the AFR is typically lHz to 2Hz, the frequency being chosen in dependence upon the various delays that occur around the control loop. The mean AFR is conventionally also set at a fixed value which need not be X = 1 but may differ from stoichiometry by a fixed amount, termed the bias. Conventionally, the standard or base'bias may be slightly rich of stoichiometry for optimal emissions performance.

The present invention is predicated upon the experimental discovery that emissions performance of a three-way catalytic converter can be improved considerably by intermittently switching from the normal mode of operation to a second mode in which a small lean bias shift (typically 1% lean of stoichiometry) is superimposed onto the standard AFR oscillations. This tends to intermittently purge the catalyst with AFR's that are on average slightly lean of stoichiometry. It has been found that these periodic perturbations of the bias regenerate the catalyst and allow improved catalyst efficiencies for HC, CO and particularly NOx for a period after each purge. This degree of emissions improvement cannot be obtained using conventional closed loop bias and waveform optimisation. Experimentation has suggested that best results are achieved by the perturbations of the bias causing relative short lean excursions followed by relatively long periods with bias at conventional (slightly rich) values.

The experimental evidence suggests that during standard closed loop bias operation conversion is gradually inhibited

over a time scale of the order of seconds. This phenomenon has been identified in different circumstances, on different platforms, and in both dynamometer aged and vehicle aged catalyst hardware. The catalyst can then be regenerated by purging some or all of the catalyst volume with lean gas.

It is important to control the frequency and amplitude of the bias shift applied so as to purge the catalyst without breaking through significantly at the tailpipe, as this will cause emissions penalties. The frequency and amplitude of the bias shift required will therefore depend on the total oxygen storage of the catalyst as well as the exhaust mass flow rate. Catalyst oxygen storage will vary with catalyst mileage. The oxygen sensors situated downstream of catalysts can provide a signal as to when the entire catalyst has been purged and also a direct method for measuring oxygen storage as the catalyst ages. A model incorporating catalyst oxygen capacity and level of stored oxygen such as that disclosed in US 5,678,402 may also be utilised.

As stated above, tests have shown that the improvement in emissions persists for several seconds and thus the period between bias excursions can be much longer than the lean excursions themselves. The graph in Figure 2 shows how the NOx content of the exhaust gases (plotted on the Y axis) varies with time (plotted on the X axis) in the absence of intermittent lean bias while Figure 3 is a similar graph with intermittent lean bias.

As previously mentioned, it is the lean excursions of the bias that have been found to be important and they need not last as long as the periods of base fuelling. This asymmetry of the required bias excursions, coupled with the low amplitude (around 0.01 lambda) means that the method is not intrusive for driveability or other factors.

Figures 2 and 3 attached graphs show the magnitude of the effect on a 1.6 litre engine made by the Applicants and known as the Sigma engine. This platform has been used for much of the development of this system although the same effect has been seen on various platforms.

Figure 2 shows base tailpipe Nitric Oxide (NOx) emissions for the Sigma engine over a portion of a dynamometer (dyno) based simulation. Catalyst hardware was dyno aged to 50 K miles equivalent. Emissions were optimised using the standard closed loop calibratable parameters, i. e. bias and peak-to peak lambse. NOx emissions were measured with a fast NOx meter. The tailpipe NOx emission trace shows high frequency switches between low tailpipe NOx emissions and high tailpipe NOx emissions. These are in phase with standard closed loop lambda switches ; i. e. rich lambda gives good conversion efficiency. Feed gas NOx emissions are stable on these time scales with only a slight gentle increase as the vehicle speed rises due to increased engine load.

Figure 3 shows the same conditions with the bias switched lean periodically for a short period. The frequency of the lean switches and their duration will depend upon the exhaust mass flow. As the catalyst has been exposed to the standard bias for some time, the tailpipe levels begin to rise as NOx conversion is inhibited. Purging the catalyst with a lean AFR is followed by a period of very high catalyst efficiencies as the bias switches back to rich. The ideal frequency of lean purge will depend on how quickly the catalyst loses efficiency after a purge and the NOx penalty (if any) from each purge.

The improvement in conversion efficiency is most noticeable for NOx emissions. However, as shown by the bar chart in Figure 4, improvements in emissions of HC and CO are also seen. This bar chart shows the quantities of all

three gases measured on the extra-urban portion of the European drive-cycle.

Emissions improvements vary considerably with the ageing condition of the catalyst system. Green catalysts can show little or no breakthrough after light-off under which circumstances no improvement is possible. However, catalysts aged to 50,000 miles can show significant breakthrough. With current and future emissions legislation set to require durability up to 100,000 miles, this system should provide a cost effective mechanism for meeting future legal emissions targets.