Promoting excellence in mobility engineering

  1. FISITA Store
  2. Technical Papers

Gasoline Engine Part Load Performance: Cylinder Pressure Curves Prediction Using Neural Networks to Reduce the Dependency on Testing
F2018/F2018-PTE-107

Authors

Sreekanth Rayavalasa
Renault Nissan Technology & Business Centre, India

Praveen Prasath, Rangarajan Srinivasan, Anand Gurupatham, Stéphane Ruby

Abstract

Part load cylinder pressure curves – pressure ‘P’ as a function of crankshaft angle ‘?’ over 720 degrees – are essential to evaluate thermal boundary conditions of engines at part load, especially for Engine Thermal Management (ETM) studies to evaluate transient ThermoHydraulic behaviour of coolant, oil and liner temperatures. However, part load cylinder pressure curves are barely available at early stages of engine development, as first prototypes are usually dedicated to full load performance evaluations. Thus, the aim of this study is to predict part load cylinder pressure curves using only full load Test measurements, which will reduce the need for prototyping and physical testing. The study is carried out on a 3-Cylinder Naturally Aspirated gasoline engine with port injection using GT-SUITE 1D simulation software (2). To begin with, engine full load performance model is calibrated within ±5%. Then part load engine performance prediction is carried out using Neural Networks and Controllers yielding ±15% of correlation with test. Validation of coolant and oil temperature on ETM Models are within ±10ºC compared to test data and ±4ºC in the most relevant comparison zones.

Add to basket