Abstract: The performance and particulate emission of a diesel engine are affected by the consumption of lubricating oil。 Most studies on  oil  consumption  mechanism  of  the  cylinder  have       been  done  by  using  the  experimental  method,  however  they  are  very  costly。  Therefore,  it  is  very  necessary  to  study  oil  consumption  mechanism  of  the cylinder  and  obtain  the  accurate  results  by  the  calculation method。  Firstly,  four  main  modes  of  lubricating  oil  consumption  in  cylinder  are  analyzed  and  then  the  oil  consumption  rate  under common  working  conditions  are  calculated  for  the  four  modes  based  on  an  engine。  Then,  the  factors  that  affect  the  lubricating  oil consumption  such  as  working conditions, the  second  ring closed  gap, the elastic force  of the  piston  rings  are also  investigated for  the four modes。 The calculation results show that most of the lubricating oil is consumed by evaporation on the liner surface。 Besides, there are three other findings: (1) The oil evaporation from the liner is determined by the working condition of an engine; (2) The increase of the ring closed gap reduces the oil blow through the top ring end gap but increases blow-by; (3) With the increase of the elastic force of the  ring,  both  the  left  oil  film  thickness  and  the  oil  throw-off  at  the  top  ring  decrease。  The  oil  scraping  of  the  piston  top  edge  is consequently  reduced  while  the  friction  loss  between  the  rings  and  the  liner  increases。  A  neural  network  prediction  model  of  the 

lubricating oil consumption in cylinder is established based on the BP neural network theory, and then the model is trained and validated。 The main piston rings parameters which affect the oil consumption are optimized by using the BP neural network prediction model and 83517

the  prediction  accuracy  of  this  BP  neural  network  is  within  8%,  which  is  acceptable  for  normal  engineering  applications。  The  oil consumption is also measured experimentally。 The relative errors of the calculated and experimental values are less than 10%, verifying the validity of the simulation results。 Applying the established simulation model and the validated BP network model is able to generate numerical results with sufficient accuracy, which significantly reduces experimental work and provides guidance for the optimal design 

of the piston rings diesel engines。 

Key words: diesel engine, lubricating oil consumption in cylinder, simulation, piston rings 

1   Introduction

   The  performance,  service  life,  and  working  costs  of  adiesel engine are affected by the consumption of lubricating oil。  The  lubricating  oil  fled  into  the  cylinder  is  not  easy  to be  burned  at  high  temperatures  and  leads  to  hard  coke formed  on  the  top  surface  of  the  piston,  which  affects  the heat  transfer  of  the  piston  and  exposes  the  diesel  engine piston  to  the  risk  of  crack。  The  wear  of  cylinder  liner increases  with  the  accumulation  of  carbon  on  the  land  of