Modelling and simulation of media flow in non-circular channels by abrasive flow machining

Abrasive Flow Machining (AFM) is a polishing method that allows efficient surface finishing and polishing operations in difficult-to-access selected external surfaces and complex internal channels. Currently, application of AFM is largely guided by best practices, which is iterative, time-consuming and expensive. Computational fluid dynamics (CFD) simulation is a potential method to predict the process outcome through simulation of the process parameters. However, the difficulty lies in the calibration of the simulation to meet actual process parameters. The two main objectives of this project are (1) to evaluate flow speed, media pressure, material removal and surface roughness during AFM of a square (20mm x 20mm), straight (260mm) channel and (2) to conduct CFD simulation to acquire wall shear stress and media pressure and validate them against experimental results. The experiment is split between three sub-trials using (1) a transparent workpiece to observe the flow, (2) a set of tool steel workpiece with pressure sensors attachments to measure the pressure and (3) a set of carbon-steel workpiece to measure the material removal and roughness. The simulation utilizes ANSYS FLUENT software using an existing Carreau Yasuda model for media modelling. The findings are summarised into three categories: Across the 20mm by 20mm cross-section, across the flow profile of 260mm and the simulation findings. Firstly, across the 20mm by 20mm cross-section, the flow speed near the middle region of the wall is approximately 30% higher than the flow speed near the corner. This is based on direct observation of flow using a transparent workpiece. Material removal near the middle region of the wall was found to be approximately 45% higher than the material removal near the corner. This is based on an average measure of material removal of the middle region (10mm from corner) and region near to corner (3.33mm from corner) along the channel profile at different heights. Secondly, across the flow profile of 260mm, the pressure gradient was found to be unaffected by extrusion pressure. This is based on the pressure drop along the flow profile for three different extrusion pressures. Surface roughness also improved from approximately Ra of 3μm to Ra of 0.3μm at all measured locations after 13 cycles and remain steady thereafter with 90% of the roughness drop occur during the first 3 cycles. Thirdly, CFD simulation found that the selected simulated parameters (values not shown due to confidentially in ARTC) of the Carreau Yasuda model (Low µ0; Medium λ; Medium n) result was found to be within the range of the experimental results. The pressure drop was shown to be similar as well. This shows that the selected parameters for the modelling of the AFM media (MV-36) using Carreau Yasuda model has met the objective of simulation validation. The assumption for a no-slip condition for the simulation did not allow any correlation to the experimental wall slip velocity result as the simulated wall slip velocity is zero. The simulated flow profile results could not be used for any correlation as there was no experimental flow profile made. Simulated wall shear stress was verified to have the same trend to the experimental wall slip velocity. This is inferred from a cross-section of one side of the wall channel. Simulated path line shows that there are a lesser particles travelling along the entrance of the square channel relative to the channel profile. This imply that media flow is lesser at the entrance which is a wrong simulated result as understood from real life experiences, the edge of the channel will always more material removal, or a filleting. The simulated results might be due to the unaccountability of the Barus effect.Bachelor of Engineering (Mechanical Engineering