Abstract
The main purpose of this investigation was to explore the potential benefits of structural functional surfaces using facilities available within the University. The potential benefits were demonstrated by applying functional surfaces to a set of particular engineering applications. The thesis mainly concentrated on improving the frictional performance of a surface structure for hydrodynamic bearing application. This thesis has also included some preliminary investigation into drag-reducing riblet structures but this chapter mainly discusses the development of a novel experimental apparatus which is needed for precise boundary layer profile measurements and also to obtain the actual surface drag for each sample. To be able to assess these surfaces experimentally, they first, have to be manufactured. So, an extensive literature review of current manufacturing technologies was carried out. Each manufacturing method was ranked in its ability to cost-effectively produce surfaces with accuracy and repeatability also being considered. It was concluded that rolling, currently, has the best ability to structure large surface areas with the lowest costs associated. Other manufacturing methods, such as laser surface texturing, provide excellent repeatability and accuracy as well as the ability to create complex surface structures but is incredibly time-consuming for large surface areas. It was suggested that a hybrid of multiple manufacturing technologies would be incredibly useful for structuring surfaces. By combining rolling with more elaborate surface texturing methods (i.e. use a method such as LST to texture the roller surface), it is possible to amplify the productivity of less efficient methods, substantially. Before any journal components were textured, it was decided to test a batch of ground components. These components were finished with an abrasive tape process. The process parameters were varied for each sample and by doing this, a set of components with different roughness characteristics should have been obtained. The components were measured for 2D roughness parameters, 3D roughness parameters and surface energy. The components were tested on a tribometer apparatus in order to obtain a coefficient of friction (COF) for each sample. Correlation coefficients were then generated for the different surface measurements against COF, so that any strong correlations or trends could be identified. The idea was to try and obtain a reliable performance indicator (PI) so that frictional losses could be identified. It was found that the roughness parameters Sc (core void voume), Ssc (mean summit curvature) and Rku/Sku (profile/surface kurtosis) showed promise in the ability to predict the performance of a surface. The next stage was to texture the surface of the journal component. This would done by the application of the type III texturing grinding process, described by Stepien (Surface Engineering, 24: 219-225), to the cylindrical grinding process. Some initial components were manufactured and the textures generated were found to be of an ellipsoidal shape. In order to guarantee the benefits of such surfaces, the configuration of the surface pattern has to be optimised. A python script was developed during this investigation in order to automate a full modelling process. The computational fluid dynamics (CFD) modelling used a full 3D Navier-Stokes approximation. This script was used in conjunction with the Taguchi optimisation technique and a best surface configuration was found, resulting in a maximum surface drag reduction of 16.6% at a 3μm clearance. Further grinding trials were performed and the input parameters of the process were designed so that surface patterns were close to the recommendations of the optimisation process. The performance of the textured samples was impressive, with a maximum reduction in COF of 18.4% seen against a non-textured component with similar average roughness (Sa) value. Again, all components were measured for the aforementioned roughness parameters and surface energy. Sku continued to predict the best-performing component, showing promise as PI for both non-textured and textured samples.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have