Diesel engines have difficulties in cold starting. Cold starting means low temperature intake air that is coming inside the cylinder, low temperature walls, low temperature piston head. All these are making the fuel evaporation difficult. More, the usual way to start a diesel engine is to flooded with fuel. The engine then accelerate to a speed over the idling speed, then he governor kicks in cutting the fuel, the engine decelerate, fuel is injected again and the cycle repeats itself until a stable speed is reached. This cycle is having as a result the exhaust of water vapors, oil particulate, even some fuel droplets remained from the previous cycle.

More, during a normal functioning, there is a succession of firing misfiring cycles, so the emission of the white smoke is favored.

In order to determine the exact causes and the methods to reduce the white-smoke emissions a study was made on a DDC series 50 engine.

The engine is direct injected and uses an electronic unit injector. Due to this fact, it is very difficult to install a needle lift sensor on the unit to allow precise determination of the injection timing and injection duration. To overcome this a strain gauge was installed on each rocker arm of the engine measuring the stress due to the fuel compression. The signal output was calibrated on a test bench specially designed and instrumented for this purpose. For each injector was found a relation between the amount of fuel injected function of injection duration and engine speed. Also the signal from the strain gauge allows the determination of the point where the needle is rising from its seat, so permitting the determination of the injection timing. Tests were made and relations were found for 2 types of fuels JP-8 and DF-2.


1. A fuel injection test bench has been developed. Its function can satisfy the requirement of DDC50 engine fuel delivery mapping and other injection-related test. The tested data are repeatable and are good in accuracy.

2. The fuel delivery rate changes linearly with injection pulse width. The fuel delivery rate changes slightly with cranking speed.

3. The cyclic fuel delivery rate of the JP-8 fuel is slightly less than that of the DF-2 under the same operating conditions. The main reason of this difference may be contributed to the lower viscosity of the JP-8 fuel which causes more leakage through the plunger clearance and other sealing parts than the DF-2 fuel does.

4. The injector-to-injector variation of cyclic fuel delivery becomes larger when the pulse width becomes smaller. When the pulse width is above 10 CA, the variations are less than ± 5% from the average.

5. Regressed equations from the injection data have been obtained for the future engine data analysis. The regression results were satisfactory.

6. The strain gauge measurement of unit injector injection events is feasible. It has the advantage of being simple in installation and operation, less expensive compared to the use of a needle lift sensor or pressure transducer, and reliable result.

7. The dynamic effect on the output signal of the strain gauge during the engine acceleration process has little effect on the accuracy of the needle open pressure calculation.

8. Digitization of the analogue injection signal, whether from plunger force measurement or from needle lift measurement, may cause serious error in both, injection timing and fuel delivery calculation.

9. For fuel delivery purposes only, the start of injection and end of injection points can be arbitrarily defined on the rising edge and the falling edge of the injector pressure profile provided that bench measurement data of the fuel delivery is available.

10. By adopting the plunger cell pressure rising time correlation and center point definition of the rising and falling edges described in this report, the error caused by the digitization of the data can almost be eliminated.

11. Correlation between the nominal pulse width and the measured pulse width has been obtained with the engine speed effect being taken into consideration.

The engine transient behavior during the cold starting period affects the combustion instability and white smoke emissions. The criterion of preset fueling control strategy so far is to start the engine successfully. To get the engine started quickly, large amount of fuel is dumped into the combustion chamber. When engine starts firing, the speed usually is first accelerated up higher above the idling speed, then the governor kicks in and reduce the fuel delivery and eventually stabilizes the engine at idling speed.

The cold starting tests under room temperature on both single cylinder engine and DDC50 engine show that misfiring happens before engine reaches stable idling speed. The data analysis has shown that some misfiring cycles may be caused by later injection timing that is believed to be related to the slow response of the governor to the fast engine speed acceleration.

After the first firing cycle, the engine usually experiences a sharp acceleration due to large amount of fuel dumped. This sharp acceleration caused a speed overshoot above preset idling speed. The misfiring cycles are mostly observed after the engine experiences this overshoot period.

Under cold ambient temperature, misfiring will occur before the engine reaches idling speed. The test results form single cylinder engine show that engine runs at a pattern of firing-acceleration-misfiring-deceleration-firing, and the engine eventually gets started.

If look at the average speed history of the engine cold starting process under different ambient temperatures, the slope of the speed curve against engine cycles gets smaller as the ambient temperature gets lower. It seems that there is a best value of acceleration for the engine to be started smoothly.


The ignition delay itself in nature is a time period for the mixture to obtain heat from the surroundings, and to start chemical reaction to accumulate the radicals to a critical level - a self sustainable ignition nucleus is formed. Not only satisfactory pressure and temperature levels are necessary, the time period when the pressure and temperature are kept at above certain level should be long enough. If the instantaneous engine speed at around the TDC is low, there will be good chance for the autoignition to occur. From the view point of ignition delay, the ignition favorable speed profile should have the characteristics of fast compression to reduce the heat transfer and blow-by losses, then slow speed at around TDC where peak pressure and temperature occur.


If we look at the engine speed history curves at different ambient temperatures, as in figures 1 - 5, it were the firing cycles that dominated the average engine acceleration and drew the average speed history profile. This fact can be seen more clearly at lower ambient temperatures(figure 4 and 5). It is as if the engine would have started the same way with or without those misfired cylinders. The fuel injected in the misfired cycles was just wasted and contributed nothing to the whole starting process except generated white smoke. This average engine acceleration profile become slower and slower as ambient temperature become lower and lower. This is the balanced result between the physical and chemical restraints , as represented by the firing - misfiring boundary, and the performance of the control device.

Figure 1. Instantaneous engine speed at cold start, 21 oC ambient temperature

Figure 2. Instantaneous engine speed at cold start, 5 oC ambient temperature

Figure 3. Instantaneous engine speed at cold start, 0 oC ambient temperature

Figure 4. Instantaneous engine speed at cold start, -5 oC ambient temperature

Figure 5. Instantaneous engine speed at cold start, -10 oC ambient temperature

For the traditional cold starting method, the fueling control strategy is usually designed to dump large amount of fuel during cranking period, and then reduce the fuel delivery when engine reaches preset idling speed. The over fueling was believed to have more chance to start the engine quickly. When engine starts firing, with large amount of fuel dumped into the combustion chamber, the engine speed usually is first accelerated up high above the idling speed, then the governor kicks in and reduce the fuel delivery and eventually stabilizes the engine at idling speed. This type of starting speed history pattern can be called overshooting pattern. The fast acceleration of engine speed during this overshooting period may easily cause the engine to run into the misfiring zone. Figure 6 shows a firing-misfiring boundary line.

Figure 6. Firing-Misfiring boundary line

Under low ambient temperatures, the up speed limit the engine can run without suffer misfiring is restricted by the firing/misfiring boundary line, and the up speed limit gets lower and lower as ambient temperature gets lower. The attempt to accelerate the engine quickly to reach idling speed will certainly be frustrated by unstable combustion: misfiring and white smoke. With the traditional method of cold start, an engine that eventually reaches a stable idling speed, after experiencing speed surge and misfiring, is the result of the gradual expansion of the firing zone by the firing cycles which, unnecessarily interrupted by the misfiring cycles, warms up the combustion chamber.

Both analysis and experiment have shown that the cold start combustion instability is strongly affected by ambient temperature. With the lower down in ambient temperature, the injection timing should become more and more advanced in order to start the engine. Advancing the injection timing, However, has a limit, because too early injection timing may cause the fuel being injected on the cold combustion chamber walls when cylinder pressure is too low. It can then be concluded that, under very low or marginal ambient temperatures, the firing zone will be very small regarding to both available injection timing and engine speed ranges. The firing zone will only cover an area of a low speed range with very early injection timing. By the traditional cold starting method, large amount of fuel is dumped into combustion chamber first, when the engine fires, it will be accelerated too fast by the early stage firing cycles, and the engine will immediately run into the misfiring zone. In such a situation, a slow acceleration profile of engine speed by the firing cycles would provide the succeeding cycles with not only gradually warmed up cylinder temperature, which itself will in turn provide a larger firing zone for next cycle, but long enough time around TDC for autoignition development, and hence may start the engine successfully in a shorter time meanwhile eliminate the misfiring.

For the question that whether or not there is the relation between the cranking speed and cold startability, the answer is that there exist a relation between the two. The relation, however, can not be expressed by a simple curve, but a zone, the firing zone. If the cranking speed is too high or too low, the engine will be in the misfiring zone and can not be started

If the ambient temperature condition is not marginal, however, the misfiring problem may be more possibly solved by better organized fueling control. This includes control on both injection timing and fuel delivery amount, and the control algorithm should reach the level of cycle-to-cycle basis until idling speed is reached because misfiring happens during this transient period.

The quantity of firing - misfiring boundary lines calculated by many of ignition delay equations, which are from the bomb or vessel test, usually do not match the real engine data. The analytical and experimental results both show that the injection timing plays a key role in the determination of the boundary line of firing - misfiring zones. In general, to have a practical ignition delay equation able to predict the misfiring of a real engine, the equation should include parameters of temperature, pressure, injection timing, cetane number, and perhaps the engine speed as well.

The misfiring boundary line, for a real engine, can be obtained through experimental work which may be conducted under well controlled engine operating conditions. In the future investigation, it is suggested to establish the relation which can be used under real engine condition to predict the misfiring, and then the modified cold start control strategy based on the newly derived relation can be tested on the real engine.


We are currently issued the key level 2 by DDC to access the engine control unit DDECIII, We are allowed to log the pulse width data and other parameters from the DDEC. The most important parameter, the BOI, however, is not allowed to read it from the DDEC by us. All the data are time based.

1. The first stage of the test is to reduce the fueling during cold cranking period. The reduced fueling level will be based on current calibration table used by DDEC. 4 fueling reduction levels are proposed, 10%, 20%, 30%, and 40%. The test will be conducted at

-5 and -10 oC.

2. The second stage is to change the BOI according to the results from first stage since the demand for BOI change may be reduced because of the slower acceleration as a result of reduced fueling.














Copyright ©2011 College of Engineering, Wayne State University
Wayne State University does not endorse external sites.