Numerical investigation of steady and transient flow in an automotive cyclone

This study is concerned with computational fluid dynamics (CFD) simulations of flow in automotive cyclone particle separators which are becoming an integral part of automotive crankcase oil droplet filtration. The basic idea behind a cyclone separator is that a contaminated gas enters tangentially at the top and spirals down the conical section, and in the process flings oil particles towards the cyclone walls where they get collected upon impact. Cyclone performance is characterized by two performance parameters: pressure drop across the system and particle separation efficiency where separation efficiency defined as the fraction of particles collected by the cyclone, over those entering it. Traditionally, empirical efficiency prediction models have been used in the design of industrial size cyclones. These models however have proven to be inadequate for automotive size cyclones, especially for smaller particles, and physical testing is still a costly alternative. Since limited studies have been found in the literature investigating automotive cyclone separators, this research focused on CFD simulations to characterize the flow and performance of a standard Stairmand-type reverse-flow automotive cyclone. Steady simulations were performed at a standard flowrate of 47 l/min and the flow in the cyclone was fully characterized. Steady simulations, however, were found to never fully converge, with pressure, velocity and vorticity results exhibiting small oscillations as the solution was iterated further. Transient simulations showed the presence of a periodic main vortex precession that resulted in periodic fluctuations of the flow parameters. The frequency of the precession at a flow rate of 47 l/min was found to be about 50 Hz. This precession was observed at all of the flowrates investigated (between 23 and 71 l/min) and the frequency was found to increase with engine speed. However, despite the existence of this periodic flow feature, results of both steady and transient flow simulations were found to correlate well with experimental testing results. A systematic investigation of the FluentRTM solver settings and the mesh used showed that several parameters affected the accuracy of the CFD predictions and recommendations were proposed that provide the best balance between accuracy and computational time (including the use of RSM turbulence model and non-equilibrium wall treatment). Finally, sensitivity, optimization and reliability based design optimization (RBDO) studies were performed on the cyclone with the multi-objective goal of increasing efficiency and reducing pressure drop. These studies showed that the cyclone diameter and outlet diameter have the largest effect on performance. Furthermore, it was found that while other studies treat the CFD solver as a black box and ignore the error introduced by the solver, this is not appropriate as our optimization investigation of the cyclone geometrical parameters found for instance that the gains in the performance of the cyclone are well within the uncertainty limits of the CFD solver. Model error must hence be accounted for in future optimization and reliability studies of flow in a cyclone separator.