Identificación de la interrelación entre actores y actoras en la cadena de valor de las

a. Start of Injection

Start of injection was defined to be the time that fuel initially appeared on the tip of the injector. Figure 26 shows the sequence of frames before and after SOI for the Yanmar injector. Visual inspection of the figure yields that injection occurred in either frame #112 or #113.

Figure 26 SOI image sequence

Rather than relying on visual observation of each individual combustion run to determine SOI, an image-processing code was used to automate the process. This code is included in Appendix B. Essentially, the code created a small box on the tip of the injector and used pixel averaging to determine the frame in which injection began. Figure

Frame #109 Frame #110 Frame #111

Frame #112 Frame #113 Frame #114

27 plots a typical pixel averaging sequence for a run using the Yanmar injector. It is clear that injection occurred when the slope of the plot began to increase. According to the code, SOI occurred in frame 112 for this particular run. This corresponds well with the visibly image determination shown in Figure 26. Each frame captured by the camera received a time stamp, and the time at SOI was determined from this time stamp.

Figure 27 Start of injection (SOI) processed pixel average

b. Start of Combustion

Differing from the definition of SOI, start of combustion was not as easily defined. Two definitions could be used: 1) start of combustion could be defined to be the instant that the first molecule of fuel initially appeared to combust, or 2) start of combustion could be defined to be the instant that the bulk section of the fuel began to combust. Figure 28 shows the typical ignition sequence for the Yanmar injector. Because the fuel was dyed, the entire injector spray can be seen in the high speed images, the star

0 50 100 150 200 250 -50 0 50 100 150 200 250 300 Image Frame # R e la ti v e P ix e l B ri g h tn e s s 3% ethylene/air/O2 preburn Injection Signal sent at t = 1.4 sec

 = 14.8 kg/m3

pattern shape is the fuel as it sprays out of the injector. The increased intensity of light along the fuel jets in the images relates to the occurrence of combustion.

Figure 28 Start of combustion sequence

Appendix C contains the code used to determine the start of combustion based on the first definition, the first occurrence of combustion. Similar to the determination of SOI, this code found the maximum pixel value for each frame, and determined the frame at which that value significantly increased. This approach proved less robust at yielding consistent results throughout the testing. The laser reflection on the optical window, evidenced as the streak in the top right corner of each image in Figure 28, needed to be removed from the image, and thus any combustion occurring in that area could not be considered in the analysis. Also, the average pixel value varied with a certain amount of noise, and required a thresholding method to determine when the spike occurred, which is

Frame #133 Frame #134 Frame #135

Frame #136 Frame #137 Frame #138

a similar technique as the bulk averaging method. This method was explored as a means to calculate SOC, but was not used in this study.

Appendix D contains the code used to determine the start of combustion based on the second definition, the first occurrence of bulk combustion. This code operated similar to that used to determine SOI, using an averaging approach. The value of each pixel in the frame was averaged and bulk combustion was determined to be the frame when this average began to increase. Figure 29 shows the a typical plot for the average pixel value in each frame. The code determined that this value occurred at Frame 136, which corresponds with the initial appearance of combustion visually in Figure 28.

Figure 29 Start of combustion processed pixel maximum

Compared with the first definition, using the initial onset of combustion to determine SOC, the bulk combustion approach resulted in SOC that always occurred later; however, the bulk combustion approach was used as the primary means of

0 50 100 150 200 250 0 50 100 150 200 250 Image Frame # R e la ti v e P ix e l B ri g h tn e s s 3% ethylene/air/O2 preburn Injection Signal sent at t = 1.4 sec

 = 14.8 kg/m3

determining start of combustion because it corresponds with techniques used by other researches in the field. As mentioned in the introduction, researchers have been using pressure to determine when ignition begins, which itself is a bulk measurement. Also, the bulk averaging of the frame decreases the effect that outliers, quickly combusting fuel droplets, would have on the determination of the start of combustion. The time at start of combustion was determined from the time stamp of the frame where the code indicated bulk combustion began.

c. Alternative Methods for Determining Start of Injection

In addition to using the high speed imaging, two different methods for determining the time of SOI were also explored during this experiment. The high speed imaging is dependent on being able to maintain a clear image through the optical window throughout the test. As mentioned previously, the combustion products from the preburn are primarily water and carbon dioxide. The fuel injection event and subsequent combustion could potentially produce soot which would occlude the window. Once occluded, additional imaging data could not be taken for hours, sufficient time was required to allow the walls to cool, clean the window, and then reheat the walls. Initially, it was thought that the time delay between the trigger signal that initiated the injection sequence and the actual injection of the fuel into the chamber would be a repeatable value that could be used to determine start of injection. This delay was determined by running “cold tests,” runs where the preburn was not initiated, and taking data using the high speed imaging system. The lack of any combustion in these cold runs eliminated the possibility of the optical window being occluded, and the high speed images were consistently clear. Table 4 shows the statistical results for the cold tests. These data show that the time interval between the injection trigger and the start of injection was not consistent. The standard deviation for a given pressure ranges from 1.16 to 3.06 milliseconds. The magnitude of the ID being studied in this project is on the same order as the magnitude of the error using this method, and thus this method would not be effective at determining SOI. The data also shows that the performance of the injector did not significantly increase for a specific pressure.

Table 4 Trigger-SOI delay statistics

Based on the testing apparatus set up for the Yanmar injector, it was thought that the inconsistency in the data resulted from the actuation of the high speed valve which received the injection trigger then opened the valve that delivered the high pressure fuel to the injector. A high speed Kistler pressure gage, similar to the gage used to measure the dynamic pressure in the chamber, was added to the fuel line after the high speed gage and before the Yanmar injector. The cold runs were conducted again in the same manner as described previously, in an attempt to find a constant time delay between the high pressure fuel being delivered to the injector and the actual start of injection. Table 5 shows the results from these runs. It is clear that again, there was no consistent delay. The variation in the data is greater than what was desired for the experiment, and thus this method was not used. These tests determined that the time required for the Yanmar injector to actuate, measured from the arrival of high pressure fluid to the fuel spray leaving the injector tip, was not consistent in each run, and varied on the order of hundreds of microseconds. Pressure (psi) Runs (# total) SOI Delay (msec) Std. Deviation (msec) 3600 7 58.66 2.33 3750 6 58.15 2.83 3850 5 60.16 2.86 4000 5 57.02 2.24 4100 5 59.14 2.91 4250 5 57.62 1.16 4350 5 59.32 3.06 4500 7 59.43 2.00

Table 5 Delivery of pressurized fuel to SOI delay statistics

After this analysis, it was determined that the high speed imaging system produced the most consistent and accurate results with the least amount of variation. This method was used as the primary means of determining the time at which start of injection occurred. It was imperative that the optical window remain clean throughout the combustion events for this data to be taken, and steps were developed to ensure that the window remain clean.