An uncertainty analysis of a cumulative injected fuel mass measured with a vehicle driving condition

A system that can measure the amount of fuel injected by fuel injectors on a level identical to that of an actual vehicle driving condition was constructed on a laboratory scale without driving a vehicle on a road. In order to inject fuel under conditions identical to those when driving a vehicle, various vehicles' sensor signals on an identical level to the driving condition were input to the ECU of the device. When the various vehicles' sensor signals are supplied to the ECU, the ECU drives the injector in a manner equivalent to the vehicle running condition. The vehicle performance parameters under the vehicle driving conditions were computer simulated using GT-Suite®. Vehicle specifications including the engine part load data, a vehicle automatic transmission shift map, the K-factor and torque ratio data of the torque converter, and the transmission gear ratio were used as input data for the GT-Power® simulation. In this study, FTP-75 served as the vehicle driving mode. The vehicle performance parameters simulated by GT-Suite® were the throttle opening angle, intake air flow rate, and the engine speed. Various sensor signals based on the vehicle performance parameters were generated by the FPGA board and real-time OS. When these various sensor signals generated by the FPGA board are supplied to the ECU, the injectors driven by the ECU inject fuel into each collecting chamber. A load cell and a dynamic strain amplifier were used to measure the cumulative amount of fuel injected at 0.05 second intervals after the driving of the injectors started. The cumulative amount of fuel injected while driving the ECU in the Phase I section (0-512 seconds) of the FTP- 75 mode was measured and the experiment was repeated 20 times. The average and standard deviation of the cumulative amount of injected fuel were determined at each of the measurement time steps during Phase I. Moreover, an uncertainty analysis of the experimental data was conducted using the t-distribution error function. The cumulative injected fuel quantities measured at 0.05 second intervals were distributed within 2% of the mean value at a 95% confidence level.