A NEW METHOD FOR CHARACTERIZING THE KNOCK RESISTANCE FOR GASEOUS FUELS

The globalization of the gas market and the drive towards sustainability are increasing the diversity of the supply of gases to the natural gas infrastructure. As a rule, "new" gases have different chemical compositions than the gases traditionally distributed. For the optimal and safe operation of gas engines, it is of great importance to characterize the knock resistance of gases accurately. Knock phenomena are caused by autoignition of unburned fuel mixture, the so-called end gas, in the cylinder before the mixture is completely consumed by the propagating flame. Mild engine knock increases pollutant emissions, while severe knock can cause engine failure or physically damage the engine, and should thus be avoided. Rather than rely on the empirical methods using gas mixtures and “standard” engines traditionally employed for this purpose, we have developed a method to characterize the knock resistance of gases based on the combustion properties of the fuel mixtures. The core of the method described in this paper is the computation of the autoignition process during the compression and burn periods of the engine cycle. The chemical mechanism used to compute autoignition has been tested against experimentally determined autoignition delay times for a broad range of gaseous fuels measured in our Rapid Compression Machine (RCM) at conditions relevant to spark-ignited gas engines. In addition to the effects on autoignition itself, the effects of fuel composition on the in-cylinder pressure and temperature conditions relevant for knocking, such as changes in the combustion rate are also incorporated in the method. Comparison between predicted and measured pressure profiles shows that the model accurately predicts the changes in the in-cylinder pressure when varying the fuel composition in our high-speed medium Brake Mean Effective Pressure (BMEP) engine. To rank the knock resistance of different gases for the lean-burn, spark-ignited gas engine used in this study the recently developed propane-based scale (Propane Knock Index, PKI) is extended to consider pipeline gases ‎34]. In this scale, the knock resistance for a given gaseous fuel mixture is expressed as an equivalent fraction of propane in methane under the same engine conditions. We demonstrate the veracity of the knock model by comparing the predictions of autoignition of the end gas using the model for a wide variety of fuel compositions, expressed as the computed PKI, with the Knock-Limited Spark Timing (KLST) measured in the DNV GL engine for the following mixtures: Dutch Natural Gas (DNG), DNG/C2H6/C3H8, DNG/H2, DNG/C2H6/C3H8/H2, CH4/C2H6, CH4/C3H8, CH4/i-C4H10, CH4/n-C4H10, CH4/C2H6/C3H8 and CH4/C3H8/N2. The predicted ranking of knock resistance of these mixtures is shown to be in excellent agreement with the KLST measurements. Comparison of the verified PKIs calculated for the range of gases studied with methane numbers calculated using the AVL and the MWM methods points to shortcomings in these yardsticks in predicting knock resistance for the gas engine used in this study. Since the DNV GL method is based on the physical and chemical processes that govern knock, it can be adapted in a straightforward manner to new engines and fuels. Being a fundamentally correct approach, the predictive power of the DNV GL method makes this methodology well suited to serve as the basis for a standard. Additionally, the method provides a valuable tool for gas engine manufacturers to define knock-free gas engine performance ratings for today’s and tomorrow’s gaseous fuels