Gas turbines are comprised of many parts, which are often expensive and required tosurvive a harsh environment for significant periods (with or without reconditioning). Todifferentiate between parts, and facilitate keeping accurate historical records, they areoften given a unique identification number. However, manually recording and trackingthese is difficult. This has led to increased adoption of machine readable codes to helpreduce or eliminate many of the issues currently faced (mostly human error). The harshenvironment of a gas turbine means that typical methods of applying machine readablecodes, such as printed adhesive labels, are simply not durable enough. Direct part marking(DPM) is necessary to ensure the desired longevity of the code over the part's useful life.The research presented in this thesis was approached in two main phases. Firstly, theauthor sought to investigate the technical solutions available for the elements requiredof a part tracking system (encoding, marking and scanning). This included identifyingthe characteristics of each and their compatibility with one other (across elements). Inconjunction with Alstom, criteria were identified that were used as a basis for comparisonso that the preferred technical solutions could be determined. The outcome of this processwas enhanced by the author developing a number of industrial contacts experienced inimplementing part tracking systems.The second phase related to the legibility of the codes. The harsh environment of agas turbine results in surface degradation that may in turn reduce the legibility of anymachine readable codes present. To better understand why read failures occur, the author_rst looked to the scanning process. Data Matrix symbols (marked via dot peen) requirethe scanner to capture an image for processing. Image capture is typically achieved usinga charge-coupled device (CCD), each pixel of which induces a charge proportional to theincident illumination. This illumination is received via reflection from the surface of thepart and hence the Data Matrix marked on it. Several surface features were identified thatgovern the way in which the part surface will reflect light back to the scanner: surfaceroughness, dot geometry and surface colour. These parameters are important becausethey link the degradation mechanisms occurring { broadly categorised as deposition,erosion or corrosion { with the scanning process. Whilst the degradation mechanismsare distinctly different in their behaviour, their effect on surface reflectivity is commonin that they can all be characterised via the surface parameters identified. This wasdeduced theoretically and so the author completed tests (utilising shot blasting to changethe surface roughness and oxidation to change its colour, independently) to show thatthese surface parameters do indeed change with the introduction of surface degradationand that there is a commensurate change in symbol legibility.Based on the learning derived with respect to Data Matrix legibility, the author hasproposed a framework for developing a tool referred to as a Risk Matrix System. Thistool is intended to enhance the application of part tracking to gas turbine engines byenabling symbol durability to be assessed based on the expected operating conditions.The research presented is the first step in fully understanding the issues that affect thelegibility of symbols applied to gas turbine parts. The author's main contribution tolearning has been the identification of knowledge from various other sources applicable tothis situation and to present it in a coherent and complete manner. From this foundation,others will be able to pursue relevant issues further; the author has made a number ofrecommendations to this effect.
展开▼
