This article considers the application and refinement of artificial neural network methods for the prediction of NO x emissions from a high-speed direct injection diesel engine over a wide range of engine operating conditions. The relative computational cost and performance of two backpropagation algorithms, Levenberg–Marquardt and Bayesian regularization, for this application are compared, with the Levenberg–Marquardt algorithm demonstrating a significant cost advantage. This work also assesses the performance of two alternative filtering approaches, a p-value test and the Pearson correlation coefficient, for reducing the required number of input variables to the model. The p-value test identified 32 input parameters of significance, whereas the Pearson correlation test highlighted 14 significant parameters while additionally providing a ranking of their relative importance. Finally, the article compares the predictive performance of the models generated by the two filtering methods. Overall, both models show good agreement to the experimental data with the model created using the Pearson correlation test showing improved performance in the low-NO x region.

No abstract

The understanding and prediction of NOx emissions formation mechanisms during engine transients are critical to the monitoring of real driving emissions. While many studies focus on the engine out NOx formation and treatment, few studies consider cyclic transient NOx emissions due to the low time resolution of conventional emission analysers. Increased computational power and substantial quantities of accessible engine testing data have made ANN a suitable tool for the prediction of transient NOx emissions. In this study, the transient predictive ability of artificial neural networks where a large number of engine testing data are available has been studied extensively. Significantly, the proposed transient model is trained from steady-state engine testing data. The trained data with 14 input features are provided with transient signals which are available from most engine testing facilities. With the help of a state-of-art high-speed NOx analyser, the predicted transient NOx emissions are compared with crank-angle resolved NOx measurements taken from a high-speed light duty diesel engine at test conditions both with and without EGR. The results show that the ANN model is capable of predicting transient NOx emissions without training from crank-angle resolved data. Significant differences are captured between the predicted transient and the slow-response NOx emissions (which are consistent with the cycle-resolved transient emissions measurements). A particular strength is found for increasing load steps where the instantaneous NOx emissions predicted by the ANN model are well matched to the fast-NOx analyser measurements. The results of this work indicate that ANN modelling could strongly contribute to the understanding of real driving emissions.

Accurate modelling of the initial transient period of spray development is critical within diesel engines, as it impacts on the amount of vapor penetration and hence the combustion characteristics of the spray. In addition, in multiple injection schemes shorter injections will be mostly, if not totally, within the initial transient period. This paper investigates how two different commercially available computational fluid dynamics (CFD) codes (hereafter noted as Code 1 and Code 2) simulate transient diesel spray atomization, in a non-combusting environment. The case considered for comparison is a single-hole injection of n-dodecane representing the Engine Combustion Network's "Spray A" condition. It was identified that the different break-up models used by the codes (Reitz-Diwakar for Code 1, KH-RT for Code 2) had a significant impact on the transient liquid penetration. From differing initial base setups, Code 1's case was matched as closely as possible to Code 2's case. Despite the nominal equivalence between the two simulations, there existed a discrepancy in liquid length prediction throughout injection between codes. This was caused by differing implementations of the KH-RT model in both codes. Therefore, a new implementation of the KH-RT model was input into Code 1 in order to allow correct matching of the liquid length to experimental data throughout the injection period. Results from the new model are shown and compared to the previous implementation, showing an improved ability to match to experimental data.

It is known that low-temperature combustion (LTC) strategies can help simultaneously reduce nitrogen oxides (NOx) and particulate matter (PM) emissions from diesel engines to very low levels. However, it is also known that LTC may cause emissions of unburned hydrocarbons (UHC) to rise — especially in low load operating conditions. Recent studies indicate that end-of-injection (EOI) processes may support ignition recession back to injector nozzle thereby helping to reduce these emissions. This paper contributes to the physical understanding of this EOIphe-nomenon, combustion recession, using computational fluid dynamics studies at LTC conditions. Simulations are performed on a single-hole injection of n-dodecane under a range of Engine Combustion Network’s “Spray A” conditions. The primary objective of this paper is to assess the ability of a Flamelet Generated Manifold (FGM) combustion model to predict and characterize combustion recession. First, a baseline condition FGM simulation is compared with two other combustion models namely the Well Stirred model (WSR), the Representative Interactive Flamelet model (RIF) using the commercially-available CFD solver, CONVERGE. Further studies were carried out for FGM model alone including: varying ambient temperature conditions and chemical mechanisms. Two chemical kinetics mechanisms with low temperature chemistry for n-dodecane are employed to help to predict the occurrence of combustion recession. All simulations are performed under the Reynolds-Averaged Navier-Stokes (RANS) framework in a grid-converged Lagrangian spray scenario. The simulation of combustion recession is qualitatively validated against experimental data from literature and the efficacy of each model in predicting combustion recession is evaluated. Overall, it was found that the FGM model was able to capture the combustion recession phenomenon well — showing particular strength in predicting distinct auto-ignition events in the near nozzle region.