AbstractAs powertrains and IC engines continue to grow in complexity, many vehicle manufacturers (OEMs) are turning to simulation in an effort to reduce design validation and calibration costs. Ultimately, their aim is to complete this process entirely within the virtual domain, without the need for any physical testing. Practical simulation techniques for the prediction of knock in spark ignition (SI) engines rely on empirical ignition delay correlations (IDCs). These IDCs are used to approximate the complex ignition delay characteristics of real and surrogate fuel compositions with respect to temperature, pressure and mixture composition. Over the last 40 years, a large number of IDCs have been put forward in the literature, spanning a broad range of fuels, operating conditions and calibration methods. However, the applicability of these tools has yet to be verified at the high brake mean effective pressure (BMEP) operating conditions relevant to highly boosted, downsized engines. Here, the applicability of 16 gasoline-relevant IDCs for predicting knock onset at high loads (BMEP > 30bar) has been investigated by comparing the knock predictions from each IDC against experimentally measured knock onset times. Firstly, a detailed investigation into cylinder pressure data processing techniques was performed to determine which knock detection and angle of knock onset (aKO) measurement methods were most appropriate at high loads. A method based on the maximum amplitude pressure oscillation (MAPO) during knock-free operation best estimated cycle classifications, whilst Shahlari’s Signal Energy Ratio technique  most accurately predicted knock onset. To the author’s knowledge, this is the first time that such a comprehensive study on the accuracy of these techniques at such high loads has been conducted. Importantly, these findings represent a valuable framework to inform other researchers in the field of knocking combustion on which techniques are needed to extract accurate and relevant information from measured cylinder pressure records. Secondly, the data processing techniques derived were applied to experimental data collected across a wide range of high BMEP operating conditions (up to a maximum of 32 bar) using a 1.6 litre, 4-cylinder SI engine. Trapped charge composition and temperature were predicted using a calibrated 1D model of the engine, whilst the temperature of a hypothetical hotspot in the unburned zone was estimated separately by assuming adiabatic compression from a point after intake valve closing and by mapping γ (the ratio of specific heat capacities) as a function of temperature. This revealed that none of the IDCs tested performed well at conditions relevant to modern, downsized engines. The IDC that achieved the best overall balance between aKO accuracy and cycle-classification agreement was the “cool-flame” correlation for iso-octane proposed by Ma . However, this had an unacceptably high average aKO error of ±3.5° compared to the ±2°CA limit observed within the literature, and its average cycle-classification accuracy was below 60%. The main reason for this relatively modest accuracy was a large number of false-positive cycle classifications, which mainly occurred in slow or late burning cycles. Further work should therefore focus on methods to reduce the number of false positive classifications obtained with this correlation, which could be achieved using empirical correlations to describe the latest point in the cycle for which knock would be permitted to occur in terms other measureable combustion parameters. Overall, this research has generated a unique insight into combustion at very high loads, as well as an extensive dataset that can be used for future research to improve the accuracy of empirical knock modelling techniques. Furthermore, this work has demonstrated that for the purposes of virtual spark timing calibration and the avoidance of knock, the current crop of practical simulation tools is not accurate enough at the conditions relevant to modern SI engines and has provided a better understanding of their limitations. These findings represent a major contribution to the field from both a research perspective and for industrial applications.
|Date of Award||4 Jun 2018|
|Supervisor||Christian Brace (Supervisor) & Sam Akehurst (Advisor)|
- Ignition delay
Predicting Abnormal Combustion Phenomena In Highly Booted Spark Ignition Engines:
Giles, K. (Author). 4 Jun 2018
Student thesis: Doctoral Thesis › PhD