GB2210688A - Detection of faults in diesel fuel injectors by obtaining their vibration velocity signatures - Google Patents

Abstract

Faults or wear in diesel engine fuel injectors are detected by measuring the vibrational velocity of an external surface of the injector body during one or more injection cycles. The measured velocity values are then compared with a reference value to give an indication of abnormal vibration characteristic. Preferably, the velocity is measured using a non-contact method, for example, using a laser-doppler vibrometer although a conventional accelerometer may be used. The method enables the detection of such faults as blocked or enlarged valve opening, a broken valve spring or a needle that is stuck. <IMAGE>

Description

DETECTION OF FAULTS IN DIESEL FUEL INJECTORS Most diesel engine operating characteristics, in particular power, fuel consumption and emissions, are influenced directly by the quantity and way in which the fuel is injected into the -combustion chamber by the fuel injection system. Indeed, it has been shown that 16t of all engine stoppages reported during a given five year period could be attributed to defects in fuel pumps and injectors. Despite these findings and the greater operational demands imposed by new engine noise and emission legislation, the majority of diesel engines in service today do not incorporate means for monitoring the operating characteristics or 'health' of their fuel injection equipment.

In general, health monitoring techniques allow a watch to be made over the condition of a running machine. Should the failure of some component commence, an early warning is provided that can initiate corrective action and thus prevent the occurrence of serious and costly damage. Such monitoring is particularly beneficial in the marine environment, where breakdown at sea can cause not only severe financial penalties but often a threat to the safety of the crew.

Insofar as diesel engineering is concerned such techniques have not, however, been widely used. Although the feasibility of several systems has been investigated they are, in general, complex and expensive to operate and none has yet proved commercially acceptable. Generally, therefore, it is still the accepted practice to overhaul and, where necessary, replace fuel system components in accordance with the preset maintenance schedules determined by reference to the average lifespans of the components in question.

Engine manufacturers have demonstrated an unwillingness to design their engines with a view to accommodating the transducers required for monitoring. Consequently, monitoring systems have been fitted after installation of the engine, rather than being considered as an integral part of the engine at the design and production stage.

We have appreciated that the operating characteristics of diesel engine fuel injector equipment can be monitored simply and effectively by measuring the vibrational velocity of the fuel injector body.

In accordance with the invention there is provided a method for detecting faults or wear in fuel injectors comprising measuring the vibrational velocity of an external surface of the injector body -during one or more injection cycles and comparing the measured values with a reference value to give an indication of an abnormal vibration characteristic and, hence, of the presence of a fault or wear in the injector.

Preferably, the vibrational velocity is measured using a remote non-contacting laser-Doppler vibrometer. A method in accordance with the invention will now be described in detail by way of example with reference to the drawings, in which: Figure 1 shows a typical vibration velocity signature measured at the top of a fuel injector; Figure 2 shows part of the vibration velocity signature of Figure 1 on an expanded time scale with: (a) fuel supply connected (b) fuel supply disconnected; Figure 3 is a schematic view of a multi-hole injector nozzle; Figure 4 is a view of the tip of the nozzle of Figure 3 shown on a larger scale; Figure 5 illustrates effect of changes in (a)b and (b) K on the value of peak valve velocity; Figures 6 and 7 shows vibration velocity signatures analogous to Figures 1 and 2 with a 40% reduction in nozzle opening pressure.

We have found that when the vibrational velocity at the top of a diesel engine fuel injector is measured, a vibrational velocity pattern or signature is obtained. Figure 1 shows an example of such a signature demonstrating that a pulse occurs in the injector body vibration response each time an event occurs in the injection cycle.

In the particular case shown in Figure 1, three successive injections, labelled A, B and C are present. As the engine in question operates on a four-stroke cycle at a speed of 1400 rpm, the injections are separated by intervals of 86 milliseconds, each of which is equivalent to two crank revolutions.

When the trace forming Figure 1 is expanded with respect to time (as shown in Figure 2(a)), it is found that within a single injection pulse there are in fact two quite distinct pulses (labelled I and II). Moreover, the time interval between the two pulses I and II is found to match closely the injection period measured under the given engine operating conditions. This suggests that these pulses arise from movement of the fuel injector needle valve during the injection process.

Figures 3 and 4 illustrate schematically the construction of a typical spring-loaded multi-hole injector nozzle 10. Fuel enters the injector 10 through an inlet 12 and passes by means of one or more passages 14 formed in the injector body to a chamber 16 surrounding a needle valve 18. When closed, the valve body 18 is held in place on its seat 20 (shown in Figure 4) by a high rate spring 22. In this position, combustion gases are prevented from entering the fuel system after injection and the leaking of fuel into the cylinder during certain phases of the engine's working cycle is also avoided.In the course of the fuel pump injection stroke, however, hydraulic pressure in the fuel chamber 16 rises to a level sufficient to overcome the preset spring pressure (nozzle opening pressure) and the valve 18 lifts off the seat 20, permitting fuel, which is under high pressure, to pass through hole(s) 24 formed in the nozzle and spray from the nozzle tip. At the end of injection, the pump "spill transient" causes fuel pressure in the chamber 16 to fall rapidly, and the spring 22 returns the valve 18 to its seat 20 so that the fuel injection spray into the combustion chamber is terminated.

The working cycle of the injector is dominated by two specific dynamic events, that is, motion of the valve body 18, while opening and closing away from and toward the valve seat 20, respectively.

Each of these motions gives rise to two impulsive forces which act on the injector body 10. These can be separated as follows: (1) a reactive force which is transmitted to the injector body 10 through the high rate spring 22. This force arises from the sudden changes in velocity experienced by the lower end of the spring 22 as a result of its being coupled, via a thrust spindle, to the valve body 18. The duration of this velocity change, and thus of the resulting reactive force, depends on the time taken for the valve to open or close. This time is typically in the range 0.2-0.4 milliseconds.

(2) a contact force which is developed during impact of the valve body 18 with an end stop 28 and with the valve seat 20, at each end of its range of movement. For a situation where impact occurs between two hard metal surfaces, as is the case in a fuel injector, this force is typically of duration 0.2 milliseconds.

Both contact and reactive forces are always produced during the valve reseating process when the valve body 18 moves back into contact with the seat 20. Under certain conditions of engine speed and load, however, the valve body 18 may not reach the end stop 28 during opening of the valve 10, resulting in an absence of impact.

and thus also of the associated contact force.

Figure 2(b) shows the vibration velocity signature measured at the top of the injector when the fuel supply has been disconnected so as to prevent the valve body 18 lifting. In this case, it will be seen that both pulses I and II are absent. It can therefore be concluded that the pulses I and II identified in Figure 2(a) are a direct consequence of needle valve movement and are the measured response to the reactive and contact forces described above.

Further, measurements suggest that the pulses I and II represent a forced response, as opposed to natural vibrations of the injector body 10, since their associated period is similar to the period of the exciting forces. More importantly, the magnitude of the internal exciting forces, both contact. and reactive, is directly related to the peak velocity attained by the needle valve body 18.

This can be illustrated as follows.

Motion of the needle valve body 18 can be described by the differential equation M.d2x + c.dx + Kx = F(t) (1) dt2 dt where x is needle valve displacement, N is the vibrating mass, C is a damping factor introduced to represent the frictional force between the valve body 18 and its guide and the effect of viscous drag, K is the injector spring stiffness and F(t) represents the forcing term. The system is assumed to be underdamped (ie. c2 < 4KM).

To facilitate solution of Equation (1) for any input force-time history the Green's function method is employed. Adopting this approach, the solution satisfying the generalised conditions that valve displacement and velocity at time to are xO and vO, respectively, is found to be given by

where A = for t iss are the roots of the auxiliary equation + + CA + K = 0 (3) Since the same basic principles are applicable to both the valve opening and the valve closing processes only the former will be considered in the following discussion.

Valve motion during valve open needs only to be considered from the instant it leaves the seat 20 at the onset of injection (which will be the time origin to) to the time it reaches its end stop, designated ti. Incorporating the conditions that both the displacement and the velocity of the valve are zero at to into equation (2) gives

which is best solved through conversion into numerical form and subsequent computer evaluation.If the time interval of interest (ie. tsto to ti) is divided into k steps of length 6, the valve displacement at the nth step is given by

where Sn = Sn-1 + (Fn-1cosss(n-1)@e-&alpha;@(1-n) (6) and Sn = Sn-1 + (Fn-1 sinss(n-1)@e(1-n)) (7) noting that S0 and S0 are both zero. Valve velocity is subsequently evaluated from the algorithm v(n) = (xn+l - Xn-l) / 26 (9) and the peak velocity is simply the maximum value of v(n) which, in general, is the velocity at t=t.

In general terms the forcing function can be expressed as F(t) = PF(t)net - To (9) where PF(t) net is the net pressure force acting on the valve body 18 and Fo is-the initial injector spring force exerted by the spring 22, which is dictated by the required nozzle opening pressure. When the net pressure force exceeds Fo and the valve body 18 begins to lift from its seat 20 a drop in pressure in the fuel chamber 16 occurs as a result of the increase in system volume and the commencement of fuel outflow through the holes 24 in the nozzle.

This pressure drop, however, is accompanied by an increase in the exposed area of the valve due to its differential design. Since -the time taken for the valve to open to its full extent is small (ie.

the interval t-t0 to t-ti is of the order of 0.2 - 0.4 milliseconds) the combination of the two effects described above can, to a first approximation, be assumed to give 8 constant rate of change of pressure force over this period. The net force acting on the needle valve during opening can thus be represented by rust) = at or r(n) = (10) çhere C is a constant.

Using the model described above the effect of common injector faults on the dynamic behaviour of the valve can be predicted. For the majority of cases a fault will alter the rate of change of the pressure force in the nozzle and can be simply modelled as producing a change in the value of the constant cr. A nozzle with blocked spray holes 24, for example, will experience additional resistance to fuel outflow and thus produces a higher value of than a healthy unit. An injector with a blown spray tip, that is, in which the holes 24 have become enlarged, on the other hand, will have an increased flow area and reduced flow resistance and thus produces a lower value of than normal.

Other faults, for example a broken injector spring 22, may produce a change in K, the spring stiffness, in addition to a change in the value of Ch. In general, however, all faults will produce a different value of peak velocity. The predicted changes are illustrated in Figure 5 which shows the percentage changes in peak velocity magnitude which accompany changes in the value of and L Predicted changes may be for either one, or both, dynamic situations and include the case of zero peak velocity (ie. no valve motion), which will occur, for example, when the needle valve is either stuck closed or stuck open.

Since injector faults produce a change in the peak value of velocity attained by the valve body 18 during opening and closing they will also produce a concurrent change in the magnitude of the internal forces that excite the injector housing into vibration.

This is then reflected by a change in the measured vibration velocity response of the injector body. Figure 6 illustrates this, showing the vibration response of an injector 10 whose nozzle opening pressure has been reduced by 40%. In these circumstances, pulse magnitudes are visibly smaller than those found for an injector operating under the standard pressure setting, as depicted in Figure 1. Moreover, a study of Figure 7 (showing one of the pulses of Figure 6 on sn expanded time scale) demonstrates the structure of these pulses in more detail, and suggests that the magnitude of the internal forces has been reduced to such an extent that only one of the two pulses associated with movement of the valve body 18 can be readily identified.This suggests that in practice a reduced nozzle opening pressure affects the rate of change of pressure force (ie.Q).

From the above, we have appreciated that a simple method for detecting a faulted injector entails monitoring only the energy associated with the measured vibration response of the injector body during the injection period. It is to be expected that, for a given injector unit, the vibrational energy value will exhibit cycle-tocycle variations even under supposedly stable engine conditions.

Moreover, in a multi-cylinder engine nominally identical units will exhibit variations over the same engine cycle. Given this, it is proposed that an averaged value of vibrational energy is used as an indicator for detecting faults.

Specifically, each injector is tested periodically, to determine an averaged value of vibrational energy and compare this to a baseline or reference value stored in computer memory. Any change which occurs outside acceptable tolerance limits will then indicate that corrective action, for example, the replacement of components, is necessary.

The baseline or reference energy value and its associated tolerance limits can be determined by reference to information originating from several different sources. Ideally, they should be experimentally obtained at "pass-off" when the engine leaves the manufacturerts production line. Subsequent monitoring would then compare the measured values with what should be "best-case" baseline data. This approach is only possible for newly commissioned power units. For engines already in service, however, these baseline reference values could be collated following an engine overhaul, when each injector has been removed, individually tested and is deemed to be in a healthy state.

Alternatively, in place of particular reference or baseline data, a simple comparison can be made instead between the vibrational energy of a number of similar injectors and any major discrepancy taken as an indication that a fault is present. It is assumed, at any one time, that the majority of injectors in a multicylinder engine will be operating correctly and a gross deviation from the norm will be sufficient to highlight a faulty unit.

It should also be noted that, since the peak velocity of the valve depends ultimately on the way in which hydraulic pressure is developed in the injection chamber with time, a change in the monitored energy may also be caused by faults in the pump and/or leaks in the fuel pipelines associated with the injector. However, the technique would still pinpoint the particular pipeline or pumping element at fault.

Measurement of the vibration velocity signatures of the injector body is preferably effected using a non-contact method.

For example, a laser-Doppler vibrometer can be used to measure the vibrational velocity of the top of each injector. This type of measuring equipment has the advantage that it can be used remotely, avoiding the need for substantial engine 'down time' associated with the use of standard transducers. Furthermore, it is portable and can easily be used to measure the vibration velocity signatures of several injectors successively rather than attaching separate sensors to the several injectors simultaneously.

It is not, however, essential to use a non-contacting transducer, although it is preferable to do so. Provided that a time-resolved signature from the injector body can be obtained, any transducer, for example, a conventional accelerometer can be used.

Claims (7)

1. A method for detecting faults or wear in fuel injectors comprising measuring the vibrational velocity of an external surface of the injector body during one or more injection cycles and comparing the measured values with a reference value to give an indication of an abnormal vibration characteristic and, hence, of the presence of a fault or wear in the injector.

2. A method according to claim 1 in which the reference value is derived from previous measurements obtained from the same injector.

3. A method according to claim 1 in which the reference value is derived from measurements made from a plurality of injectors similar to that in which faults are to be detected.

4. A method according to any preceding claim in which the vibrational velocity is measured over a plurality of injection cycles and the measurements averaged prior to comparison.

5. A method according to any preceding claim in which the vibrational velocity is measured using a remote, non-contacting sensor.

6. A method according to claim 5 in which the vibrational velocity is measured using a laser-Doppler vibrometer.

7. A method for detecting faults or wear in fuel injectors substantially as hereinbefore described.