EP4669846A1

EP4669846A1 - METHOD FOR DETECTING PRE-IGNITION EVENTS IN GAS-POWERED ENGINES

Info

Publication number: EP4669846A1
Application number: EP23841595.4A
Authority: EP
Inventors: Jérémy RIVA
Original assignee: Phinia Delphi Luxembourg SARL
Current assignee: Phinia Delphi Luxembourg SARL
Priority date: 2023-02-23
Filing date: 2023-12-22
Publication date: 2025-12-31
Also published as: GB2627472A; GB2627472B; WO2024175244A1; GB202302574D0; CN121079495A

Abstract

Method for detecting pre-ignition events in a gaseous-fueled engine, the method comprising the steps of monitoring a rotation of an engine crankshaft to determine rotational characteristic data of said crankshaft (81), computing a pre-ignition metric representing the real part of a Discrete Fourier Transform applied to said rotational characteristic data within a predetermined analysis window (S2), and concluding to the occurrence of a pre-ignition event if said pre-ignition metric meets a predetermined threshold (S3).

Description

METHOD FOR DETECTING PRE-IGNITION EVENTS IN GASEOUS-FUELED ENGINES

Technical field

[0001 ] The present invention generally relates to a method for detecting pre-ignition events, in particular in the context of gaseous-fueled engines.

Background of the Invention

[0002] In the automotive industry, pre-ignitions refer to uncontrolled self-ignitions of fuel in a cylinder which cause fuel combustion to begin before the cylinder has reached its Top Dead Center (TDC) position. Such combustions typically generate toxic NOx gas and a resistive torque which hinders rotation of the crank shaft and can potentially damage the cylinder chamber. It is highly important to detect pre-ignition events, so that appropriate measures to prevent them from occurring can be taken.

[0003] In traditional gasoline engines, a knock sensor is typically configured to measure high-frequency vibrations of the engine and transmit a corresponding signal to the Engine Control Unit (ECU). The signal is then processed by the ECU to determine the timing of the combustion and the occurrence of pre-ignition or knocking events.

[0004] Unfortunately, knock sensors are unreliable in gaseous-fueled engines. Indeed, such sensors are oftentimes unable to process the shape and the shock induced by an early combustion of a gaseous fuel, thereby leaving pre- ignition events undetected.

[0005] There is thus a need to devise a method able to reliably detect pre-ignition events in gaseous-fueled engines. This need is particularly great in the context of Hydrogen engines, as Hydrogen gas has a high flammability and is thus more susceptible to ignite too early.

Object of the invention

[0006] It is therefore an object of the invention to provide a method able to reliably detect pre-ignition events in gaseous-fueled engines. General Description of the Invention

[0007] The present invention provides a method for detecting pre-ignition events in a gaseous-fueled engine. The method has been particularly developed for hydrogen engines, although it may be employed with other gaseous fuels.

[0008] The present method comprises the steps of:

- monitoring a rotation of an engine crankshaft to determine rotational characteristic data of the crankshaft;

- computing a pre-ignition metric representing the real part of a Discrete Fourier Transform applied to the rotational characteristic data within a predetermined analysis window;

- concluding to the occurrence of a pre-ignition event if the pre-ignition metric meets a predetermined threshold.

[0009] A merit of the present invention is to have found a reliable way of detecting pre-ignition events based on the rotation of the crankshaft in a gaseous fueled spark engine. The present invention can be easily implemented in an internal combustion engine since it uses the signal from the rotation sensor.

[0010] A particular merit of the invention is to have identified that there is a correlation between the occurrence of pre-ignition and the real part of the discrete Fourier transformation applied to the rotational characteristic data. The pre-ignition metric represents this real part. Depending on the implementation, the pre-ignition metric may be the value of the real part, or the value of the real part can be processed by a predetermined function to give a number or indicator that is linked to the amplitude of the real part.

[0011] In particular, it has been found that pre-ignition events tend to be indicated by negative values of the real part of the DFT. Accordingly, the occurrence of a pre-ignition event is preferably determined if a value of the real part, respectively the pre-ignition metric, is lower than a predetermined threshold TRIM having a negative value.

[0012] The predetermined threshold TRIM is preferably mapped vs. engine speed and/or engine load. [0013] The analysis window should have a sufficient width to include a representative number of measurement points (samples), e.g. at least 10, preferably at least 20 points, e.g. about 30 or 40. More generally, the analysis window may include a number of measurement points corresponding to any integer between 20 and 40. It may encompass one or more combustion events. The width of the analysis window may be adapted by the skilled person depending on the desired application. When using a toothed flywheel for crankshaft sensing, the number of measurement points conventionally depends on the number of teeth over the crank angle range corresponding to the analysis window.

[0014] In order to allow cylinder by cylinder detection, the analysis window advantageously begins when a cylinder under analysis it at or about its firing TDC position and ends when the next cylinder -in the engine firing order- is at or about its respective firing TDC position. The analysis window therefore typically at least partially comprises the power stroke of the cylinder under analysis.

[0015] The width of the analysis window will thus vary depending on the number of engine cylinders. In general, the analysis window spans a duration equal to 2 rotations of the crankshaft divided by the number of cylinders in the engine. That is the analysis window represents 180° for a 4-cylinder engine, 120° for a 6-cylinder engine, etc.

[0016] In embodiments, the rotation rate of the crank shaft is monitored by means of a flywheel having a plurality of teeth, the flywheel being rotationally coupled with the crankshaft, and a sensor configured to detect its proximity to a tooth of the flywheel.

[0017] In embodiments, the rotation rate is sampled for every tooth passing by the sensor of the flywheel, and/or wherein the time period between two adjacent teeth is sampled for every tooth passing by the sensor of the flywheel.

[0018] In embodiments, the analysis window includes at least 10 teeth, preferably at least 18, 19 or 20 teeth, and/or has a circumference equal to the width of 60 teeth. [0019] In general, the pre-ignition metric is computed for sets of rotational characteristic data included in the analysis window defined for each cylinder, whereby pre-ignition events are detected on a cylinder-by-cylinder basis.

[0020] In general, the rotational characteristic data may include or more of angular position, angular velocity, rotational rate, engine load, time lapsed for a predetermined angular displacement.

[0021 ] However, in practice the method may be easily implemented on sets of rotational characteristic data comprising or consisting of a collection of time values corresponding to the time interval between teeth over the respective analysis window. Alternatively, it may be the collection of engine speed values over the respective analysis window.

[0022] In embodiments, a cylinder-associated counter is incremented each time a pre-ignition event is detected for the respective cylinder, and a pre-ignition flag is raised when the cylinder-associated counter exceeds reaches or exceeds a predetermined threshold Tc.

[0023] According to another aspect, the invention relates to a method for operating an internal combustion engine comprising a plurality of cylinders fed with gaseous fuel, in particular hydrogen, wherein pre-ignition events are detected by implementing the method herein disclosed; and wherein one or more injection control parameters is/are adapted when one or more pre- ignition events are detected for a given cylinder.

[0024] In embodiments, the one or more injection control parameters is/are adapted when the pre-ignition flag is raised.

[0025] For example, the injection control parameters to be adapted include fuel quantity, injecting timing, spark timing, etc. Those skilled in the art may adapt one or more of these parameters in a known manner in order to avoid pre- ignition in the respective cylinder.

Brief Description of the Drawings

[0026] Further details and advantages of the present invention will be apparent from the following detailed description of not limiting embodiments with reference to the attached drawing, wherein: Fig. 1 a is a schematic view of an engine having a crank shaft rotation sensor;

Fig. 1 b is a representation of the output signal of the crank shaft rotation sensor;

Fig. 2 is a flowchart of the method according to the invention;

Fig. 3 is a plot of a mapping of rotation rate of the crank against time;

Fig. 4a is a plot of the pre-ignition metric against time;

Fig. 4b is a plot showing pre-ignition events detected from the plot of figure 4a.

Description of Preferred Embodiments

[0027] The schematic diagram of Figure 1 a illustrates an internal combustion engine 100 comprising a plurality of cylinders (not shown) comprising spark plugs for firing the cylinders in a predetermined order. A crank shaft rotation sensor 102 is configured to monitor a rotational characteristic (step S1 in figure 2) of its crankshaft (not represented). The crankshaft is mechanically coupled to respective pistons arranged in the plurality of cylinders. The engine 100 and the crank shaft rotation sensor 102 may be of conventional design and will therefore only be briefly described.

[0028] The rotation sensor 102 comprises a flywheel 1 , which is fixed to the crank shaft to rotate therewith, as well as an inductive sensor 1 .2. A plurality of identical teeth 1.3 is spread around the flywheel 1 , spanning its entire circumference except for a gap 1 .4 of which the width is approximatively equal to the width of two teeth 1.3. The inductive sensor 1.2 is arranged radially in close proximity to the flywheel 1 , such that its magnetic field varies as the flywheel 1 rotates due to the distance between the tip and the bottom of a tooth 1.3.

[0029] Figure 1 b represents the analog signal emitted by the inductive sensor 1 .2. In particular, tooth pattern 1 .5 represents the variation in the magnetic field of the inductive sensor 1 .2 as a single tooth of the flywheel 1 traverses the space directly in front of the inductive sensor 1.2. Moreover, for each complete rotation of the flywheel 1 , the inductive sensor 1.2 emits a unique signal pattern 1 .6 characteristic of the geometry of the gap 1 .4.

RECTIFIED SHEET (RULE 91) ISA/EP [0030] This analog signal is sent to a control unit, for example the ECU, and processed to determine rotational characteristic data of the crankshaft, such as angular position, angular velocity, rotational rate, engine load, time lapsed for a predetermined angular displacement, etc. In particular, the rotation rate of the crankshaft may be computed from the angular spacing of the teeth 1 .3 and the time period between two adjacent teeth 1.3. The rotation rate and the time period between two adjacent teeth can thus be sampled once per tooth pattern 1 .5 and stored.

[0031 ] The plot shown on figure 3 represents the rotation rate 2 of the crankshaft in revolutions per minute (RPM) in function of time, on an interval comprising a pre-ign ition event. Prior to the event of pre-ign ition , the rotation rate 2 repeats the bell-like pattern 2a. When a pre-ign ition occurs in a cylinder of the engine, a resistive torque with direction opposite to the rotation of the crank shaft is generated, which results in a sudden drop in rotation rate. The characteristic shape of the pre-ignition event is indicated as pattern 2b. Following the preignition event, the rotation rate 2 of the crankshaft fluctuates with a greater amplitude before returning to its normal state, i.e. following pattern 2a.

[0032] The inventive method uses the rotational characteristic data of the crankshaft, here as determined by the rotation sensor 102, to determine the occurrence of pre-ignition events. As used herein and according to conventional meaning, pre-ignition (or preignition) refers to such event, in a spark engine, wherein the air/fuel mixture in the cylinder ignites before the spark plug fires. Pre-ignition is initiated by an ignition source other than the spark, such as hot spots in the combustion chamber, a spark plug that runs too hot for the application, or carbonaceous deposits in the combustion chamber heated to incandescence by previous engine combustion events. Pre-ignition is thus an autoignition of the mixture before the spark.

[0033] Pre-ignition is a technically different phenomenon from engine knocking, which occurs after the spark. Specifically, knocking occurs when combustion of some of the air/fuel mixture in the cylinder does not result from propagation of the flame front ignited by the spark plug, but when one or more pockets of air/fuel mixture explode outside the envelope of the normal combustion front. [0034] Pre-ignition is also to be distinguished from so-called misfires, where no combustion occurs at all.

[0035] It will be appreciated that the present inventors have found that with engines running on gaseous fuel, such as hydrogen, pre-ignition cannot be determined by means of a knock sensor. It has been observed that on hydrogen engines, pre-ignition causes a peak in in-cylinder pressure that often does not comprise the pressure oscillations that exist in case of knocking.

[0036] Remarkably, the present inventors have found that the occurrence of pre- ignition can be determined by processing the rotational characteristic data of the crankshaft through discrete Fourier transformation. In particular, the present inventors have found that there is a correlation between the occurrence of pre-ignition and the real part of the discrete Fourier transformation.

[0037] An embodiment of the present method will now be explained in detail with reference to the flowchart of Fig.2. Step S1 corresponds to the constant monitoring of the rotation of the engine crankshaft to determine rotational characteristic data thereof. This is preferably done with the conventional rotation sensor 102. As described above, such sensor typically generates a sinusoidal sensor signal that allows calculating rotational characteristic of the crank shaft. In particular, it allows calculating the time interval between two consecutive teeth, and hence also the angular speed. In the present embodiment, the time intervals are computed and stored; they represent the rotational characteristic data used for pre-ignition determination.

[0038] Next, in Step S2, a transformation function is applied to the acquired set of rotational characteristic data over a predetermined analysis window. As indicated, the set of rotational characteristic data is built as a set of values representing the time interval between adjacent teeth of the sensor wheel.

[0039] The analysis window is generally set to encompass the ignition/combustion of a given cylinder, without overlapping ignition/combustion in another cylinder. Preferably, for a given cylinder, the analysis widow may be set to begin at or about the TDC ending the compression stroke of said cylinder, i.e. the firing TDC; and may end when the next cylinder (in firing order) has reached its respective compression TDC position (firing TDC). As a result, the analysis window at least partially comprises the power stroke of the cylinder under analysis.

[0040] The transformation function is configured to determine a value corresponding to the real part of the DFT of the set of rotational characteristic data.

[0041 ] In the present method, this value of the real part defines the pre-ignition metric and is thus the output of the transformation function step S2.

[0042] The present process can be implemented using conventional DFT equations. The DFT equation can generally be expressed as: « = S;o¹x„ - e-^|2’'^,"^/M [Eq.1 ]

And its real part as: where x_n is the rotation rate at sample n, N is the total number of samples, and k is the current frequency with k e [0, /V — 1] . The rotation rate is sampled for every tooth passing by the inductive sensor 1.2 during the analysis window.

[0043] Accordingly, the transformation function is advantageously configured to implement Eq.2. The parameter (k) is inferior or equal to the number of samples of the analysis window, preferably comprised between 1 and 4.

[0044] In the following step S3, the pre-ignition metric is compared to a predetermined threshold TRIM. Where the pre-ignition metric meets the threshold, it is concluded that pre-ignition occurred. Preferably, the threshold TRIM is a negative number that is calibrated (e.g. on a bench or an engine dynamometer). Threshold TRIM may be fixed, but is advantageously a function of engine speed and/or engine load (torque). Hence, TRIM can be mapped against speed and load.

[0045] The comparison step is illustrated by diamond S3. Where the pre-ignition metric (i.e. the real part) is inferior to the threshold TRIM, this means that pre- ignition occurred in the corresponding cylinder. Accordingly, a pre-ignition counter (associated with the respective cylinder) is incremented at Step S4.

[0046] At step S5 the pre-ignition counter for the respective cylinder is then compared to a threshold Tc. If the pre-ignition counter exceeds the threshold Tc, then a pre-ignition flag is raised for the corresponding cylinder. The pre- ignition flag is indication for the ECU that pre-ignition tends to occur in the associated cylinder. From there, the ECU may apply corrective actions to avoid the occurrence of pre-ignition. For example, the ECU may adapt one or more injection control parameters, such as, e.g., fuel quantity, injection timing, spark timing, etc. This parameter adaption step is symbolized by step S6.

[0047] Where the thresholds are not met, i.e. where the pre-ignition metric is superior to threshold TRIM or where the counter is inferior to threshold Tc, the detection of pre-ignition based on the current data set is finished. The method is applied to the next set of rotational characteristic data acquired for the next cylinder, i.e. Step S2 for the next analysis window.

[0048] The present method thus permits a cylinder-by-cylinder detection of pre- ignition events based on the signal of the crankshaft rotation sensor. The so- acquired rotational characteristic data are processed according to the present method, over an analysis window that corresponds to a given cylinder. The rotational characteristic data are thus recorded and processed by the transformation function as data sets corresponding to the respective analysis windows.

[0049] In this embodiment, the set of rotational characteristic data consist of a collection of time values that corresponds to the time interval between two consecutive teeth. Alternatively, the set of rotational characteristic data may consist of a collection of values representing the angular speed of the wheel at each tooth, which is inversely proportional to the time interval.

[0050] The width of the analysis window for a four-cylinder engine having a 1 -3-4-2 firing order is shown in Fig.3. As alluded above, the width of the analysis window, noted W, preferably spans from the firing TDC, noted TDC^F, for the cylinder under analysis to the TDC^F of the next cylinder in firing order. Referring to Fig.3, the analysis window Wei for detecting pre-ignition in cylinder C1 spans from TDC^Fi to TDC^F3. Each dot in the plot corresponds to one measurement point. As explained before, in Fig.3 the Y-axis is the speed but the same events can be observed with a Y-axis corresponding to time intervals between consecutive teeth.

Exemplary embodiments:

[0051 ] Figs. 3 and 4 were obtained on a hydrogen combustion engine with four cylinders and featured with a rotation sensor 102 comprising a flywheel with 58 teeth and a tooth gap having a width corresponding to the width of two teeth (so-called 58X). Hence 60 teeth are theoretically able to fit around the circumference of the flywheel. For ease of explanation, the flywheel will be considered to have 60 teeth in the section below.

[0052] The rotation rate of the crankshaft and/or the time period between two adjacent teeth are monitored as previously detailed. As explained, the analysis window should cover a crank angle spanning from the firing TDC of the cylinder under analysis and to the firing TDC for the next TDC^F, such that the analysis window at least partially comprises the power stroke.

[0053] Hence, for a four-cylinder engine, the analysis window corresponds to half a rotation of the crankshaft, whereby the analysis window would encompass 30 sample/measurement points corresponding to 30 teeth of the flywheel.

[0054] It may be noted that in practice, the set of rotational characteristic data for the cylinder under analysis may simply comprise the collection of time interval values acquired by the sensor that corresponds to the analysis window. Once the engine is started and the ECU has achieved synchronization of the rotation sensor 102, the ECU will know the angular position corresponding to each measurement point. Hence for building the set of rotational characteristic data for a given cylinder, it is sufficient to take the 30 measurement points that follow the firing TDC of that cylinder.

[0055] Fig.4a shows a plot of the pre-ignition metric mapped against time for a four- cylinder engine, as represented by curve 10. It should be noted that it has been verified on an engine dynamometer that no misfire or knocking events occurred within the domain of the mapping. [0056] As it can be seen, the pre-ignition metric generally oscillates with a relatively small amplitude (i.e. lower than 5), but occasionally spikes to comparatively much greater values, as shown e.g. at region 10a. As mentioned previously, such spikes have been found to be characteristic of pre-ignition events and are detected by comparing the value of the pre-ignition metric to a negative threshold TRIM represented on Fig.4a by line 12. This has been confirmed by simultaneously performing in cylinder pressure measurements.

[0057] Whenever the pre-ignition metric meets or exceeds the threshold TRIM within a given analysis window, the algorithm concludes that a pre-ignition event has occurred in its associated cylinder and increments a corresponding counter. In a preferred embodiment, said counter may be stored as a mapping, as shown on Fig.4b by plot 14, with a sudden change in value 14a indicating the occurrence of a pre-ignition event.

[0058] For a three-cylinder engine with the same rotation sensor, the analysis window preferably corresponds to two-thirds of a rotation of a crankshaft. In other words, the DFT analysis is performed for an interval starting from the TDC and ending at 40 teeth after the TDC and thus having 40 sample points.

[0059] For a five-cylinder engine with the same rotation sensor, the analysis window corresponds to two-fifth of a rotation of a crankshaft. In other words, the DFT analysis is performed for an interval starting from the TDC and ending at 24 teeth after the TDC and thus having 24 sample points.

[0060] For a six-cylinder engine with the same rotation sensor, the analysis window corresponds to one third of a rotation of a crankshaft. In other words, the DFT analysis is performed for an interval starting from the TDC and ending at 20 teeth after the TDC and thus having 20 sample points.

Claims

1. A method for detecting pre-ignition events in a gaseous-fueled engine, the method comprising the steps of: monitoring a rotation of an engine crankshaft to determine rotational characteristic data of said crankshaft (S1 ); computing a pre-ignition metric representing the real part of a Discrete Fourier Transform applied to said rotational characteristic data within a predetermined analysis window (S2); concluding to the occurrence of a pre-ignition event if said pre-ignition metric meets a predetermined threshold (S3).

2. The method according to claim 1 , wherein the occurrence of a pre-ignition event is determined if a value of said pre-ignition metric is lower than a predetermined threshold TRIM having a negative value.

3. The method according to any of the preceding claims, wherein the predetermined threshold TRIM is mapped vs. engine speed and/or engine load.

4. The method according to claim 1 or 2, wherein the analysis window begins when a cylinder under analysis it at or about its firing TDC position and ends when the next cylinder in a firing order of the engine is at or about its firing TDC position.

5. The method according to any of the preceding claims, wherein the analysis window spans a duration corresponding to 2 rotations of the crankshaft divided by the number of cylinders in the engine.

6. The method according to any of the preceding claims, wherein the rotation rate of the crank shaft is monitored by means of a flywheel having a plurality of teeth, the flywheel being rotationally coupled with the crankshaft, and a sensor configured to detect its proximity to a tooth of the flywheel.

7. The method according to the claim 6, wherein rotation rate is sampled for every tooth passing by the sensor of the flywheel, and/or wherein the time period between two adjacent teeth is sampled for every tooth passing by the sensor of the flywheel.

8. The method according to claim 6 or 7, wherein an analysis window includes at least 10 teeth, preferably at least 18, 19 or 20 teeth, and/or has a circumference equal to the width of 60 teeth.

9. The method according to any of the preceding claims, wherein the pre-ignition metric is computed using a real-valued transformation function implementing the equation: wherein parameter (k) is inferior or equal to the number of samples of the analysis window, preferably comprised between 1 and 4.

10. The method according to any of the preceding claims, wherein the pre-ignition metric is computed for sets of rotational characteristic data included in the analysis window defined for each cylinder, whereby pre-ignition events are detected on a cylinder-by-cylinder basis.

11 . The method according to any of the preceding claims, wherein the rotational characteristic data of said crankshaft include angular position, angular velocity, rotational rate, engine load and /or time lapsed for a predetermined angular displacement.

12. The method according to any of the preceding claims, wherein each set of rotational characteristic data consists of a collection of time values corresponding to the time interval between teeth over the respective analysis window.

13. The method according to any of the preceding claims, wherein a cylinder- associated counter is incremented each time a pre-ignition event is detected for the respective cylinder, and a pre-ignition flag is raised when the cylinder- associated counter reaches or exceeds a predetermined threshold Tc.

14. The method according to any of the preceding claims, wherein said gaseous- fueled engine is a hydrogen internal combustion engine.

15. Method for operating an internal combustion engine comprising a plurality of cylinders fed with gaseous fuel, in particular hydrogen, wherein pre-ignition events are detected by implementing the method according to any one of the preceding claims; and wherein one or more injection control parameters is/are adapted based on said detection of pre-ignition events.

16. The method according to claim 15 when depending on claim 13, wherein the one or more injection control parameters is/are adapted when the pre-ignition flag is raised.