Details

Autor(en) / Beteiligte

• Lea, C J

Titel

Second-Moment Closure Computations of In-Cylinder Flows in Idealised Reciprocating Engines

Ort / Verlag

ProQuest Dissertations & Theses

Erscheinungsjahr

1994

Quelle

ProQuest Dissertations & Theses A&I

Beschreibungen/Notizen

• Computations of turbulent flow in the cylinder space of idealised reciprocating engines have been made to assess the performance of a second-moment Reynolds stress closure when applied to internal combustion engine flow modelling. Comparisons are made against experimental measurements and K-ε model simulations.The computed geometries are all axisymmetric and only motored engines have been considered to remove any uncertainties in predictive performance caused by a combustion model. The successful implementation of the Reynolds stress closure into an existing computer code has been verified by computing flow through a sudden pipe expansion.Five test cases are used to assess the turbulence models. Two of these cases are steady-state, consisting of flow through a valve/cylinder assembly with the piston removed and using differing valve-seat angles. The remaining cases are all transient model engine flows, including both piston and valve motion to give a four stroke cycle with compression. Two cases have a flat-top piston, at compression ratios of 3.5 and 6.7, the third consists of a deep piston bowl geometry, also at a compression ratio of 6.7. The utility of this latter test case for assessing turbulence model performance is limited by uncertainties in the experimental measurements. For all the test cases, it is found to be important to use a higher-order convection discretisation scheme which is free from numerical diffusion, when assessing turbulence model performance.The steady-state test cases, having much in common with intake stroke flows of reciprocating engines, reveal that the Reynolds stress and K- ε turbulence models give an overall similar level of performance in the region immediately downstream from the valve. A similar result is found for the transient reciprocating cases in the first half of the intake stroke, in which both models are in good agreement with measured mean velocities.For the transient test cases, it is found in the latter half of the intake stroke that the Reynolds stress closure predicts the main vortical structure to be substantially more vigorous than that given by the K- ε model. However, for flat piston geometries, the rate of decay of this vortex at the start of the compression stroke is greatest for the Reynolds stress model. At top-dead-centre of compression, the measured tendency for the mean flow to decay to a state of uni-axial compression with flat-top piston geometries, is only captured by the Reynolds stress closure. The K-ε model predicts the spurious presence of a mean vortical motion occupying the whole of the in-cylinder space. These differences in turbulence model behaviour and performance occur because only the Reynolds stress closure is able to model streamline curvature and Reynolds stress transport phenomena.Because these flows do not qualify as being rapidly-distorted, the Cε3- related term in the e-equation is found to have little influence on the predicted mean flows and Reynolds stresses. At top-dead-centre of compression for the flat-top piston cases, the reported near-homogeneity in the turbulence field is captured solely by the Reynolds stress closure, but only at the lower compression ratio.It is concluded that the Reynolds stress closure shows an overall advantage in predictive performance over a K-ε turbulence model when applied to in-cylinder flows. This advantage is offset by an increase in required computational resource. It is recommended that additional test cases be examined to further quantify these benefits and drawbacks.