Date of Award
Master of Science
Ryan B. Wicker
Additive Manufacturing (AM), or layer-by-layer part fabrication, has played a tremendous role in the maker culture by allowing ideas to be materialized with limited resources or knowledge in manufacturing. Various cutting edge AM technologies exist today that are used to create end-use parts; however, these technologies are still new and the processes have not gone through the rigorous evaluation process that traditional manufacturing (i.e. milling, stamping, casting) methods have been through. As a result, several important questions arise when looking to adapt AM technology, including control of the manufacturing process, effect of manufacturing process on part properties, level of variance between the virtual input and the fabricated object, part defects and identification of defects, and understanding destructive and non-destructive testing etc. Because the process is new and parts fabricated from AM technologies vary widely, each part upon fabrication is unique and hence, certification and qualification of the fabricated part is important. This Thesis tries to answer these important questions by developing and implementing a monitoring system and incorporating virtual metrology leading into part qualification.
Due to the potential to qualify a part during every stage of fabrication, layerwise monitoring has become an area of interest in the field of AM. Spatial measurement and qualification of every stage of fabrication, including each layer while a part is under fabrication, were demonstrated through this research by the application of a virtual measurement system with a powder bed fusion AM technology. Specifically, the focus of this research was to study the metrology for AM fabrication systems, focusing on powder bed fusion technology, using an in situ optical measurement device that included an infrared (IR) camera. The utilization of an IR camera during fabrication allowed the acquisition of several data sets (e.g. thermal data, surface area, porous areas, etc.), which were acquired, analyzed, and compared with experimental measurements. Development of a MATLAB (The MathWorks, Inc., Natick, MA) code and incorporation of a novel continuous data acquisition system were the ultimate goals of the project.
Aiming towards finding the level of geometric variance upon fabrication, the first task was to develop a basic MATLAB code to detect the object within IR thermographs captured during operation of a powder bed fusion electron beam melting (EBM) fabrication system. An algorithm was developed upon selection of a region of interest, and the segmented images were used for the remainder of the process. Surface areas of the fabricated regions were acquired with pixel measurement upon detection of objects. Pixel area (virtual area) was converted into a standard scale of measurement and compared with the experimental measurement system. A 0.98% difference of fabricated area was found, between the geometric data obtained for simple circular features using the acquired IR images and measurements taken from the fabricated parts. Hence, a virtual measurement system was introduced using an IR camera.
Detection and quantification of defects (i.e. porosity) was another important factor of part qualification and the next development in this research using IR imaging. As AM systems provide an advantage to fabricate complex functional parts, which include support structures, metrology for complicated structures entails the applicability of the optical measurement system. A bracket, having support structure and variation of geometry with height change, was fabricated and an algorithm was developed for the extraction of an object surface area from the support structureâ??s surface area. Porous areas were also detected as child boundaries and quantified within the IR images captured during fabrication of a part with intentionally seeded defects. Quantified areas were compared with CT scan data and showed a 52-69% difference existed between intentional defects. Nonetheless, results showed that the IR images could be stacked post-fabrication to represent the part that has been fabricated and help identify any defects within the part.
One potential advantage of the IR camera is the ability to acquire thermal data, which can be used to determine part quality related to microstructure and mechanical properties. A MATLAB code was developed to store the thermal history of the fabricated object from image analysis. The average processing temperature obtained from image analysis was found to be ~1250K for Ti-6Al-4V with a maximum deviation of 7.8K (~0.6%) that was due to round-off errors during pixel conversion to temperature data. Several factors such as mean radiant temperature, calibration of IR camera, atmospheric condition, have potential to affect the result. An uncertainty analysis was not performed but can be calculated in future research to determine the specific contribution of errors on temperature measurements. Microstructure was obtained at three different heights of a fabricated bracket. Alpha and beta phase for Ti6Al4V microstructure was quantified using a MATLAB code (alpha phase varied within 57%-71%). Cooling rate was calculated at the respective heights (within 5-17K/s based on height) and correlated to the resulting microstructure. A linear relationship between microstructure and cooling rate was obtained which has the potential to predict about the microstructure based on cooling rate in future.
To acquire temperatures of each process step and cooling curves during fabrication, continuous thermal data acquisition was required. An ad hoc device, compatible with the Arcam S12 system for continuous data acquisition, was designed and integrated with the Arcam EBM fabrication system to acquire the thermal history of all process steps for fabrication. The view port available for the IR image acquisition system was used and a block was designed that fit the existing view port. The view port for the shutter mechanism was sealed and a ZnSe glass (for optimal infrared transmission) and Kapton film were incorporated in the design between the IR camera and build platform.
The aforementioned issues (i.e. monitoring of process conditions, defect detection) are important across all AM technologies. To demonstrate the developed algorithms, a powder bed technology was selected and evaluated with the final goal to demonstrate the feasibility in using the developed detection method and integrated metrology toward another AM system. A binder jetting system was utilized and images were captured upon binder deposition. The same algorithms developed in this Thesis were applied for object detection and area quantification. Pixel measurement were found to provide 3.6% difference (higher) comparing with ideal value and the machine is suspected to deposit binder on a bigger surface area to compensate for the shrinkage after fabrication. Accurate pixel measurement and correlation with shrinkage can provide an indication towards percent amount of bigger surface area entails to be considered during binder deposition. In future, the metrology can be applied to control binder deposition to obtain parts with specific dimensions based on CAD geometry while compensating for part variation (i.e. shrinkage) effects inherent to the manufacturing process.
Received from ProQuest
Ridwan, Shakerur, "Process Monitoring In Additive Manufacturing Aimed Toward Part Qualification" (2015). Open Access Theses & Dissertations. 1134.