WO1996005420A1

WO1996005420A1 - Method for quasi-feedback lean burn control using a narrow-banded lambda sensor for stoichiometric mixtures

Info

Publication number: WO1996005420A1
Application number: PCT/SE1995/000913
Authority: WO
Inventors: Hans Johansson
Original assignee: Mecel Ab
Priority date: 1994-08-08
Filing date: 1995-08-08
Publication date: 1996-02-22
Also published as: SE9402651L; SE9402651D0

Abstract

The invention relates to a method for regulation of a combustion engine having a narrow banded lambda sensor for stoichiometric mixture ratios of air-fuel (μ = 1,0) arranged in the exhaust system, where the combustion in the combustion engine during its major part of operation is regulated towards lean mixture ratios of air-fuel exceeding μ = 1,1, and preferably μ = 1,2-1,6 or leaner. By forcing the engine during short duration to stoichiometric mixture ratios at several load cases could a base value for the necessary amount of fuel Fgr be determined for maintaining stoichiometric operation. The necessary amount of fuel for lean operation could thereafter be determined by use of a correction factor Fkorr, where Fvirt = Fgr.Fkorr. During continued operation at the predetermined lean mixture ratio is Fvirt used for regulating the amount of fuel supplied, at the load cases where Fvirt have been established, or by an interpolated value of Fvirt in-between the load cases where Fvirt have managed to be established. By the inventive method a quasi-feedback regulation could be obtained for lean burn with an inexpensive and reliable lambda sensor for a richer stoichiometric mixture ratio.

Description

METHOD FOR QUASI-FEEDBACK LEAN BURN CONTROL USING A NARROW- BANDED LAMBDA SENSOR FOR STOICHIOMETRIC MIXTURES

This invention relates to a method in accordance with the preamble of patent claim 1.

PRIOR ART

Several different systems are known regulating the air- fuel mixtures towards very lean mixtures. Approximately 14,5 kg of air (A) is needed for a complete combustion of 1 kg fuel (F). The excess air factor which describes this stoichiometric condition A/F is λ=1.0, where λ is the amount of air supplied divided by the theoretical amount of air needed. A lean mixture of air-fuel (λ > 1 ) have a surplus of air and a richer mixture (λ < 1) contain less air than the theoretical amount of air needed for a complete combustion of the fuel supplied.

An optimal lean regulation of the air- fuel mixture, λ=l,3, is often desirable in order to reduce the specific fuel consumption as well as emissions, but very difficult due to that the engine is very close to the misfire limit when the air- fuel mixture is extremely lean. The maximum power output obtainable from the engine and the smoothest running of the engine is otherwise obtained with a small surplus amount of fuel at λ=0,9, but at this mixtures high levels of HC and CO emissions are obtained as well as a high specific fuel consumption. In order to comply with stricter demands on low level emissions have three way catalysts been developed, which reduce harmful emissions with highest efficiency when λ-=l,0. In order to regulate the engine in such a way mat the lambda value is maintained at this level, most engines have a special developed sensor located in the exhaust system. This sensor is called the lambda sensor, which monitors the residual amounts of air in the exhaust gases, and hence could correct the amount of fuel supplied in a feed-back manner if the lambda value deviates from 1,0. Another problem with combustion of leaner air-fuel mixtures where λ is in the interval 1 ,2- 1 ,5 and in some special circumstances as high as 2,0, is that the combustion could not be supervised by inexpensive lambda sensors. A lambda sensor having a broad-band, which produce an output signal being proportional over a λ-range in the interval 1,0-1,5 is very expensive and have not been implemented to any larger extent in mass-produced engines, due to costs and insufficient reliability. The lambda sensors implemented to an significant extent in Otto engines, regulated at lambda values around 1,0, have a very narrow band with an output signal which is fully saturated at small deviations from λ =1,0., and therefore not suitable for a feed-back regulation when operating at λ values in the interval 1,2-1,5.

In US,A,4526001 a constant adjustment towards the lean direction is determined for the entire load- and speed range from a short operating time at stoichiometric mixtures. This is unfavourable for engines operating at larger speed- and load ranges, where non-linear effects could occur such as non- uniform replenishment of air into the combustion chamber. Also in US,A,4681077 is a constant adjustment towards the lean direction for the entire load- and speed range determined from a short duration operation at stoichiometric mixtures, where the entire fuel map is corrected by a constant offset value ΔD_ON-

OBJECT OF THE INVENTION

An object of the invention is to obtain an improved feed-back regulation of combustion engines operating at the lean side of λ=l,0, assisted by a conventional narrow-banded lambda sensor with an effective range around 1,0. The invention enables an optimal adjustment to the proper lean amount of fuel for the entire speed- and load range for the engine in concern. In this way could a control system using an inexpensive and mass-produced lambda sensor be implemented in special engines mainly operating at leaner air-fuel mixtures. A Diesel engine being converted for natural gas fuel operation, advantageously used as propulsion source for city-buses or as a power source for stationary electric generators, could for example be equipped with an inexpensive but yet reliable control system.

SHORT DESCRIPTION OF THE INVENTION

The inventive method is characterised by the characterising clause of claim 1.

The regulation is obtained by an intermittent, non-continuos, and at several load- and speed cases activated feed-back of a parameter representative of a by force obtained non normal combustion in the combustion chamber, in a manner hereafter designated as a quasi feed-back. The feed-back signal is the output signal from a lambda sensor used for stoichiometric air-fuel mixtures λ=l,0, and a stoichiometric operation obtained by force of short duration establish the base value for the necessary amount of fuel for the present load- and speed case, which amount of fuel by correction according a given function or conditions is reduced in order to establish the reduced amount of fuel needed for the major part of operation at a predetermined and desired lean mixture.

The establishment of a base value is activated at several load- and speed cases and for load- and speed cases in between are new base values given by interpolation.

Other distinguishing features of the invention is evident from the characterising part of the remaining claims and the following descriptions of a preferred embodiments, which description is made by reference to the figures specified in the following list of figures.

FIGURES

Figure 1, shows a flow-chart for the inventive method, Figure 2, shows a flow-chart for correcting the fuel amount dependent of detected ignition failure, Figure 3, shows the output signal from a conventional narrow-banded lambda sensor, when the A/F ratio is subjected to a rapid change,

Figure 4, shows the output signal from a conventional narrow-banded lambda sensor at λ values ranging from 0,98 to 1,02,

Figure 5, shows schematically an arrangement for regulation of fuel supplied to a combustion engine.

DESCRIPTION OF AN EXEMPLARY EMBODIMENT

In figure 1 is shown a flow-chart for the inventive method. The method is for example used in Diesel engines converted to natural gas fuel, where the λ- value during operation should be maintained at 1,4-1,5 for a optimal fuel usage and minimising of exhausted pollutants. For such converted Diesel engines is a conventional narrow-banded lambda sensor with an effective range around λ=l,0 implemented in the exhaust system.

In step 1 the routine is started, which could be triggered by an interrupt routine at recurrent time intervals. In step 2 a control is made if a first condition is fulfilled which could be attainment of a certain predetermined temperature of the engine, which indicates that the lambda sensor should have reached a sufficiently high temperature in order to produce a reliable output signal, alternatively that the temperature or output signal of the lambda sensor itself is controlled to determine if the proper operating temperature have been reached. The sequence will return to start as long as the lambda sensor has not reached its operating temperature, normally close to 300°C. When the lambda sensor have reached its proper operating temperature, then the routine proceeds to step 3.

In step 3 a control is made if detection of λ- value is of current interest The regulating sequence following a demand of a new or renewed detection is dependent on two conditions.

The first condition is that the engine at the moment is operating at a preferably predetermined operating case, i.e. a combination of at least load and revs (number of revolutions), where a reference value is desired. For example could reference values be desired at specific or close to specific revs and loads. For a converted Diesel engine such specific revs could be revs between 500-2000 rpm at intervals of 100 rpm, and the load levels different loads from the partial load range (20-30% of full load) up to full load (100% of maximum supply of fuel amount) at load intervals of 10%. If the engine is operating near a operating case which is a combination of any of these preferably predetermined revs/loads, then the first condition is fulfilled.

The second condition could be a time function determining if the renewed detection should be activated for the current operating case. This time function could for example control when a reference value latest was determined for the current operating case, and if a certain predetermined rninimum time have elapsed since the latest renewed detection was made at the current operating case, then the second condition is fulfilled. This predetermined minimum time could preferably be as long as a couple of tens of minutes, in order to prevent a renewed detection occurring unnecessary often and occupying computational capacity of the control system, and reduce unnecessary deviation from the ideal λ- value for the combustion in the engine.

If both conditions above are fulfilled then the routine proceeds to step 4 where a change of the current λ-control is initiated from having been regulating towards λ- values in the range 1,4-1,5 to a regulation towards a λ- value of 1,0. As long as the lambda sensor have not detected that the new temporary target value, λ=l ,0, have been reached, the sequence proceeds to step 5 where the amount of fuel is corrected, i.e. a gradual enrichment of fuel is initiated. When the lambda sensor detects that the temporary target value, λ=l ,0, have been reached, then the sequence proceeds to step 6 where the present fuel amount is stored, which amount of fuel F is the amount of fuel needed for the corresponding amount of air obtaining a λ- value of 1 ,0 at the current operating case. This value is denoted in the figure by Fg. , being the base value required for λ=l ,0 regulation.

The sequence proceeds thereafter to step 7 wherein a calculation is made in order to determine the reduced amount of fuel F„rt , reduced in relation to Fp , which reduced amount is required at the present operating conditions in order to be able to regulate towards λ=l,4 alternatively 1,5. This reduced amount of fuel F_w is obtained by extracting an empirically determined correction factor F_to for the present operating conditions from a map or table, for example 30% reduction of fuel, necessary for reaching a wanted λ- value, for example a lambda value about 1,4. After completion of step 7 the sequence returns to the main program and further regulation of fuel is made by using the established value F„π as a regulating parameter in a nan closed loop manner, i.e. an open-loop regulatioa

In figure 3 is the inertia or response for a conventional lambda sensor shown when subjected to a rapid change of the residual amount of oxygen in the exhaust. At the horizontal axis is indicated the number of combustion cycles and at the vertical axis is indicated the output signal from the lambda sensor measured in volt(U). In order to demonstrate the inertia of the sensor is the engine, here an Otto engine, running at 9000 rpm and the ignition system is cut out at combustion cycle 0, where the output signal of the lambda sensor is stable at an output level U-. An increases amount of non- combusted oxygen will thereafter be exhausted into the exhaust system because no combustion will occur due to ignition failure, and not until 30-50 combustion cycles thereafter will the lambda sensor reach a new stable output signal level U₂. The inertia or step-response is almost linear, and will be delayed approximately 30-50 combustion cycles, corresponding to approximately 0,20-0,34 seconds. This step-response of the lambda sensor establish the lower time limit for the regulation of short duration towards λ=l,0.

The step-response shows on the other hand that deviations of very short duration from the ideal lean mixture proportions is needed for the combustion engine in question, before a new base value F is obtained for calculation of an updated F^ for the present operating case, which is used for further regulation.

Figure 4 shows the output signal U from a conventional narrow-banded lambda sensor at different lambda values in the interval 0,98-1,02. Already at very small deviations from λ=l,0 is the output signal fully saturated. This type of lambda sensor could not be used in the lean operating range for closed loop regulation at lambda values exceeding 1,02, and definitely not at λ- values in the interval 1,2-1,5. By using the inventive method with forced regulation of short duration down to λ=l,0, could these narrow-banded lambda sensors be used also for combustion engines operating with very lean mixtures.

In figure 5 is shown schematically an arrangement for controlling delivery of fuel to a combustion engine 31. An electronic control unit 32 controls that the right amount of fuel is delivered to the combustion engine 31 proportionally to the amount of air drawn into the cylinders. With the fuel dosing device 30 is the right proportions of the air-fuel mixtures delivered to the engine via the inlet manifold 36. A lambda sensor 34 is arranged in the exhaust system of the engine upstream of a catalytic converter, as seen in the direction of exhaust flow. The arrangement is of the type used in Otto-engines regulated towards a stoichiometric mixture, λ=l,0. Dependent of the residual amount of oxygen detected in the exhaust is the amount of fuel corrected in a closed-loop manner, resulting in that a stoichiometric condition, λ=l,0, is maintained. For combustion engines operating at lean mixtures having a λ-value in the interval 1 ,3- 1 ,5 or in some cases higher, is the major difference found in the type of catalytic reactor used. Otto engines regulated towards λ= 1 ,0 have a specially designed three-way catalytic reactor which have maximum degree of efficiency if λ is maintained at a value of 1,0, and being able to oxidise CO and HC as well as reducing NOχ at the same time.

In figure 2 is shown a flow-chart for a misfire dependent correction of the value F^, obtained from the routine shown in figure 1. Regulation towards very lean mixture ratios of air-fuel leads to increased risk for misfiring, demanding a misfire control and a subsequent correction of F^ towards richer mixture ratios if misfire occurs. Misfire correction starts in step 8 and could be initiated at each ignition event In step 9 detection of a misfire is performed, which preferably could be obtained by supervision of the ionisation-current in the spark plug gap using a type of device shown in EP3.188180 or SE.C.457831. No ionisation current is developed during the so-called post- ionisation phase in the spark plug gap if a misfire occurs.

If a misfire is detected then the misfire correcting sequence will proceed to step 10 wherein a correction variable F_fa is increased by a correcting step ΔF_fet for each misfire event detected. F_fe, is reset to zero value at each start of the engine, but Ffet or a predetermined share of F_fe could also be stored in a non- volatile memory to be used for the next start of the engine. ΔF_fe, is preferably a predetermined fixed correction step which increases the amount of fuel some percent or share of percent of the fuel supplied to the engine.

The misfire correcting sequence proceed thereafter to step 11 wherein F^ is corrected in the enrichment direction as; Fvirt = F, - ¥*„, + Ffe .

If no misfire is detected in step 9, then the sequence will proceed to step 11 without increasing the correction variable F_fa .

In a further developed embodiment, not shown, could Ffe, be resetting stepwise to a zero value. Return to leaner air-fuel mixtures could then preferably be initiated by using smaller return steps than the correcting step ΔFfe, initiated in the rich direction after a misfire. Return take place preferably after a predetermined number of combustion cycles or a predetermined lapse of time after the latest detected misfire, for example a couple of hundred cycles or a couple of seconds. The regulating system will then continuously strive to return to the optimal lean ratio of air-fuel mixture, and an occasional misfire will not lead to any long term operation at non optimal lean ratios of air-fuel mixtures.

Fvi_rt could as previously described be interpolated for each load and revs between the preferably predetermined operating cases where a reference value have managed to be established by the forced regulation down to a λ- value at 1 ,0. In an alternative embodiment could the operating cases where establishment of reference values Fvj, are made, be selected automatically as soon as any operating case have reached a stable condition, i.e. the engine is not subjected to a transient load- or speed case. For a smooth regulation could all operating cases where reference values Fvi,, are established be situated more closely together in operating ranges used more frequently. For a Diesel engine converted to natural gas fuel operation, which could be used as a power source for stationary electric generators, almost 90% of the operating time take place at essentially constant speed with moderate changes in load. For these type of engines could the operating cases where reference values are established be located more closely together in the normal operating load- and speed range.

Claims

1. Method for regulating the amount of fuel supplied to an engine, said engine having a narrow banded lambda sensor arranged in the exhaust system of the engine enabling detection of the excess air factor λ, said narrow-banded lambda sensor generating an output signal having a distinct transition characteristic at a stoichiometric ratio of mixture λ= 1,0 and at less amount of deviation, at or before λ<0,9 or λ>l.l, have an essentially fully saturated output signal c h ar a c t e ri z e d i n that the method comprises;

-that the combustion in the combustion chambers is regulated during its major part of operation towards a predetermined lean air-fuel mixture essentially above λ=l .1 and preferably in the interval λ= 1,2- 1,6 or leaner,

-that the combustion during intermittently occurring operations of short duration is forced towards a air-fuel mixture in the order of λ=l ,0 for a number of operating cases in the operating range of the engine, only after the lambda sensor having reached its proper operating temperature, which proper operating temperature could be detected either by the engine having reached a predetermined temperature or by direct control of the temperature or output signal of the lambda sensor, -that during the intermittently occurring operations of short duration is established a base value ¥*, for the necessary amount of fuel for operating the engine at λ= 1,0 at the current operating case, and a virtual amount of fuel Fv_m is established for each of said operating cases, and for intermediate operating cases are F_vi,, established by interpolation for use during the operation of the engine at the lean air-fuel mixture,

-that the regulation returns to regulation towards the predetermined lean air-fuel mixture, after establishment of the base value F_v, and the virtual amount of fuel F^ established is used for further operation of the engine, where Fv_-t = F, • F-_nr, , and where Ft,- is a correction factor for the current operating case giving the reduction of the fuel amount necessary for continuing the operation of the engine at the predetermined lean air-fuel mixture

2. Method according claim l c h a r a c t e ri z e d i n that

-the combustion engine during its major part of operation is supplied with fuel proportionally to the amount of air supplied according a predetermined fuel map dependent of at least load and speed through a non-closed-loop regulation towards a lean air-fuel mixture ratio, and matched to the combustion engine and the fuel intended for the engine, and that the fuel map is updated with the virtual amount of fuel established for each load case of the fuel map occupied by the engine.

3. Method according claim lcharacterized in that

-the correction factor F^ constitute an empirically determined correction value for the present operating case obtaining the predetermined lean air-fuel mixture, which correction factor is stored in a map having the parameters, being at least speed and load, for the present operating case as entry conditions.

4. Method according claim lcharacterized in that

-for the operating range of the combustion engine are predetermined a number of operating cases with predetermined intervals between the operating cases in the operating range, where the forced regulation of short duration towards λ=l ,0 in each operating case is automatically activated when a predetermined time period have elapsed since a previously performed update of Fvi_r, was made for each operating case.

5. Method according claim 4characterized in that between the predetermined operating cases is an interpolation made of F^ for the operating cases in-betweea

6. Method according claim lcharacterized in that the forced regulation of short duration towards λ=l ,0 in time corresponds to at least the step-response of the present lambda sensor, which step-response is the response time necessary for the lambda sensor to reach a new stable output signal when subjected to a jumping change of the mixture ratio.

7. Method according claim όcharacterized in that the step-response and the operating time of short duration corresponds in order of size of at least 30-50 cycles of combustion engine cycles, or at least 0,2-0,4 seconds when measured in time.

8. Method according any of preceding claims characterized in that when a misfire is detected in the engine, is the virtually determined fuel amount F^ enriched by a predetermined incremental fuel amount ΔF_fe for each misfire detected, and that FV_B, preferably automatically returns to the originally determined Fvi_n with an incremental fuel amount in the lean direction being less than ΔF_fe, , when misfire have ceased and according a predetermined time- or engine dependent function.