Internal Combustion Engine Friction Testing and Virtual Modeling
MetadataVis full innførsel
A virtual test bench capable of performing integrity analysis of internal engine componentshas previously been developed in FEDEM. It does however not includeany damping and friction effects. As a result, engine performance is overestimatedand shows little dependency on the inertia and mass of engine components. In thisthesis, the friction and damping effects related to the piston assembly, connectingrod, and crankshaft of an engine were identified. A customized engine test rig wasdesigned, and motored testing performed from 3000 rpm to 9000 rpm to measurethe friction torque of a partial assembly of a Honda CRF 250 R engine. A virtualtest bench representing the physical testing was developed and tuned in FEDEM. Problems were encountered with a damaged torque transducer, possibly affectingthe accuracy of the measurements, and resulting in less experimental data thandesired. Engine oil was found to squeeze past the piston rings as a consequenceof testing without cylinder pressure, possibly affecting lubrication conditions. Testresults showed that the dominating piston assembly lubrication regime changes fromboundary, via mixed, to hydrodynamic with increasing engine speed. This causedthe measured friction torque to go from 1 Nm (3000 rpm), via a peak of 1.6 Nm (5000rpm), to 1.5 Nm (9000 rpm). The engine friction s sensitivity to oil temperaturewas noted. Engine friction and oil shearing produced enough heat to keep the oiltemperature at 87 °C at 4000 rpm, exemplifying the need for cooling systems, evenin motored testing. Measurements taken before and after engine break-in clearlyshowed the importance of break-in on the engine s friction losses. The virtual test bench was successfully able to recreate the measurements fromphysical engine testing. Simulation with reduced connecting rod mass did not producethe expected friction reduction, and it was suspected that a weakness in themodeling approach was responsible. The friction modeling formulation in FEDEMdid not have the capability to capture the changes in lubrication regime for thepiston assembly. It did not support a variable coefficient of friction, but requiredthe Coulomb coefficient to have a constant value. Compensating for this weaknessby adding additional viscous damping resulted in a less realistic model behavior. It is concluded that more test data and improved friction modeling in FEDEM is requiredto obtain a virtual test bench accurate enough to predict real engine behavior,when inertia and mass properties of critical engine components are changed.