Abstract
(Englisch)
|
The activities LVV-ETH during the period from 1/2001 to 12/2001 are computational and theoretical and they are very closely linked to experimental work performed by the University of Valencia (UPV, Spain) and by Iveco MF. Our main goals are to develop improved computational capabilities for the simulation of 1) fuel injection in cold high-pressure non-reacting environments for the investigation of spray structure, 2) fuel injection, mixture formation, ignition and combustion in heavy-duty Diesel truck engines, 3) phenomenological zero-dimensional thermodynamic models for the prediction of soot and NOx formation. The first goal is closely linked to the UPV-Injection Test Rig (ITR) experiments, whereas the second is linked to the Iveco-1l engine experiments at Iveco and UPV. The third goal is closely linked to both UPV experiments as well as to available data at ETH-LVV from experiments in similar heavy-duty truck engines. It is worth noting that experimental data were not available for validation and tuning of the computational models during the first half of the project; we received the first data from measurements in both the UPV-ITR as well as the Iveco-1l engine during July 2001, and we only performed some very preliminary tests of the models in experimental conditions. Until now, almost all our activity has concentrated in performing resolution- as well as model parameter variation-tests to identify the sensitivity of our computational tools for future validation and tuning with experimental data. As was discussed in the 12-month-report, we have been using two CFD codes for our simulations in this project. Our earlier simulations (during the first 12 months) were all non-reactive and were primarily performed with the KIVA code, which has the advantage that it is an open source code; however, we have observed several numerical problems in its performance (not very robust, limited increase in spatial resolution, and high levels of numerical diffusion). Therefore we are also employing StarCD, a commercial CFD code, which is not open source. Still, one has the flexibility to alter the values of model parameters in the break-up, atomization, auto-ignition as well as combustion sub-models. During the last few months we have used StarCD in the simulation of both the UPV Injection Test Rig (ITR) and investigated the spatial and temporal resolution sensitivity of the code in an environment of 700 K and 50bar; we then investigated the performance of the atomization and breakup-models in an ambient environment of 300K and 50bar, which is comparable to the experimental setup. Regarding the main objective of this project, which is the efficiency of the engine combustion process in the Iveco-1l engine, we have performed a systematic evaluation of the auto-ignition and combustion models in our codes and we have identified areas for improvement. Our goal during the next 6 months is to use experimental data from UPV to validate and calibrate the auto-ignition and combustion models in the code; besides the pressure and temperature curves, which are less sensitive, we will concentrate on matching the heat release rates as much as possible. For the development of the phenomenological model for soot and NOx prediction, we have measured, using endoscopic techniques, soot concentration evolution over crank-angle for approximately 80 different engine operating conditions, in a 1-cylinder 2.1l displacement engine, equipped with a conventional common-rail system. A first version of a phenomenological model using 2 equations, one for the creation and one for the oxidation of soot, has been tested against these experimental data and we found that, at the current preliminary stage of development, it reproduces qualitative trends, however, it is still far from yielding quantitative agreement with experiments.
|
