CFD and PIV investigations of pipe discontinuities

• Main Author: Li, H.
• Published: Queen's University Belfast 2008
• Subjects: 620.106
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.484975

The flow in the intake and exhaust of a typical two- or four-stroke engine is highly unsteady in nature. The unsteady flow can be used to enhance the engines performance, improve its fuel economy and lower its emissions. In order to optimise an engine system it is normal to employ computer simulation. This is only effective if the simulation methods employed are accurate. The aim of the research described in this thesis was to employ experimental techniques to gain a better understanding of the types of unsteady flow found in engine ducts especially at discontinuities such as: sudden expansions, sudden contractions and sudden contractions with an orifice. In each case the unsteady flow was represented by a pressure wave transmitted through the discontinuity. At various instances, Particle Image Velocimetry (PIV) was used to acquire velocity vector maps within each discontinuity so that the flow development could be studied as the pressure wave passes through. In addition, pressure histories were recorded at a number of locations upstream and downstream of each discontinuity so that the transmission and reflection of the incident wave could be observed. The experimental results were compared to a number of computational fluid dynamic (CFO) models. These included a traditional 10 unsteady gas dynamic analysis and a range of CFD simulations incorporating various turbulence models and differencing schemes. The ID model was only used to simulate the pipe discharging into atmosphere as it poorly predicted the rate of wave propagation. Of the five CFO turbulence models investigated the RNG k-s model was found to predict wave transmission and reflection most accurately for all of the discontinuities. When the predicted flow structure within discontinuities was compared to PIV experimental results it was found that the Crank- Nicholson temporal differencing scheme with the RNG k-s model predicted flow phenomenon with the greatest reliability.