TY - JOUR
T1 - Optical Investigation of the Diesel Spray Characteristics and Spray Geometry Prediction Model by Artificial Neural Network
AU - Cheng, Qiang
AU - Ahmad, Zeeshan
AU - Grahn, Viljam
AU - Hyvönen, Jari
AU - Kaario, Ossi
AU - Larmi, Martti
N1 - Publisher Copyright:
© 2023 SAE International. All rights reserved.
PY - 2023/4/11
Y1 - 2023/4/11
N2 - Spray evolution in diesel engines plays a crucial role in fuel-air mixing, ignition behavior, combustion characteristics, and emissions. There is a variety of phenomenological spray models and computational fluid dynamics (CFD) simulations have been applied to characterize the spray evolution and fuel-air mixing. However, most studies were focused on the spray phenomenon under a limited range of injection and ambient conditions. Especially, the prediction of spray geometry in multi-hole injectors remains a great challenge due to the lack of understanding of the complicated flow dynamics. To overcome the challenges, a series of spray experiments were carried out in a constant volume spray chamber (CVSC) coupled with high-speed Mie-scattering imaging to obtain the spray characteristics at various injection and ambient conditions. Based on the data set, the spray geometry (e.g., penetration, cone angle, spray tip velocity, area), shot-to-shot probability, and plume-to-plume variation were estimated. Furthermore, the artificial neural network (ANN) is introduced to predict the key parameters of the spray geometry to avoid the prediction errors of the existing mathematical models, and the optimal model is determined to facilitate future prediction of the spray geometry of the fuel based on the data set for algorithm training. The quantitative validation results showed that the ANN model is capable of predicting spray performance with acceptable accuracy.
AB - Spray evolution in diesel engines plays a crucial role in fuel-air mixing, ignition behavior, combustion characteristics, and emissions. There is a variety of phenomenological spray models and computational fluid dynamics (CFD) simulations have been applied to characterize the spray evolution and fuel-air mixing. However, most studies were focused on the spray phenomenon under a limited range of injection and ambient conditions. Especially, the prediction of spray geometry in multi-hole injectors remains a great challenge due to the lack of understanding of the complicated flow dynamics. To overcome the challenges, a series of spray experiments were carried out in a constant volume spray chamber (CVSC) coupled with high-speed Mie-scattering imaging to obtain the spray characteristics at various injection and ambient conditions. Based on the data set, the spray geometry (e.g., penetration, cone angle, spray tip velocity, area), shot-to-shot probability, and plume-to-plume variation were estimated. Furthermore, the artificial neural network (ANN) is introduced to predict the key parameters of the spray geometry to avoid the prediction errors of the existing mathematical models, and the optimal model is determined to facilitate future prediction of the spray geometry of the fuel based on the data set for algorithm training. The quantitative validation results showed that the ANN model is capable of predicting spray performance with acceptable accuracy.
UR - http://www.scopus.com/inward/record.url?scp=85160706287&partnerID=8YFLogxK
U2 - 10.4271/2023-01-0302
DO - 10.4271/2023-01-0302
M3 - Conference article
AN - SCOPUS:85160706287
SN - 0148-7191
JO - SAE Technical Papers
JF - SAE Technical Papers
T2 - SAE World Congress Experience
Y2 - 18 April 2023 through 20 April 2023
ER -