Experimental and computational analysis of oil flow in cooling galleries of diesel engine pistons

• Main Author: Lahr, Jens
• Published: Birmingham City University 2016
• Subjects: 621.43
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.705651

This thesis describes in details the experimental and numerical investigations conducted to determine the filling and flow behaviour inside dynamic operating piston cooling galleries. An experimental test rig was built to replicate the reciprocating motion of internal combus- tion engine pistons to allow for varying engine speed, stroke length, oil flow rate and piston size. Two transparent models were produced, representing cooling galleries for small sized engine pistons found in passenger vehicles and large sized engine pistons found in heavy goods vehicles and earth moving equipment. High speed image processing was undertake to capture the flow behaviour inside the galleries during the piston cycle. An oil mixture was used to replicate the properties of engine oil at engine operating conditions. The flow inside the gallery was recorded from various view positions to capture the flow throughout the gallery. The flow domains representing the investigated gallery shapes were generated and computational fluid dynamic (CFD) studies of the two-phase flow behaviour were performed. The studies of gallery filling and in-gallery flow behaviour were undertaken for the same parametric conditions as defined in the experiments. The flow behaviour and filling of both studies, experimental and numerical, are compared and discussed. The results of the experimental and numerical studies compared well in terms of the identified directions of the main bulk oil flow within the small and large gallery and for the investigated crank speed and flow rate conditions. Both galleries showed that the flow in the gallery from inlet to outlet was mainly driven by the oil jet entering the gallery. The continuous entering jet forced a flow of the oil into the gallery branches. Strong turbulence in the direct vicinity of the inlet occurred as air and oil mixing was significant. It was found that the size of the turbulence region depended on the flow rate and engine speed, as well as the direction of movement of the gallery. It also sustained the presence of a large amount of medium-sized air bubbles due to mixing effects. Although the CFD did not predict the fine detail of the turbulent mixing, it did capture the underlying main flow characteristics, including short circuiting at the inlet. In the mid-gallery section the formation of large air bubbles took place, which could span across the gallery height. The flow behaviour was still driven by the inflow, but also controlled by the gallery cross-sectional shape. At the gallery outlet the flow was predominantly inertia driven as a result of the gallery movement with the oil exiting mainly during the upward stroke, allowing formation of large air bubbles. Distinctly different flows were encountered within the large and small gallery. The large gallery volume showed more unstructured or chaotic flow behaviour, especially in the mid- gallery section, as a result of the complex cross-sectional shape. In the small gallery volume the bulk flow was more controlled and wall-guided due to the limited space and regular cross-sectional shape. It was also found that the overall gallery filling for both galleries varied only by approximately 2% during the crank cycle. In contrast the variation of gallery section fillings of up to 30% and 50% for the large and small gallery respectively were determined, highlighting the effects of air movement within the galleries. The experimental results showed that an increase in flow rate, reflected by an increase in jet exit velocity, led to strong air entrainment of micro-scale bubbles into the oil clearly visible inside the gallery, while an increase in engine speed led to significantly lower formation of micro-scale air bubbles in the gallery. In contrast the CFD struggled to capture such fine details, unless the mesh density reached a very fine level, resulting in unsustainably long simulation times.