Parametric study of the cost estimate for ultra precision RF components

The purpose of this thesis was to find out how the shape accuracy and surface roughness of an RF disk affect the machining time of a disk and the cost of the accelerating structures. The disks in question are part of a future Compact Linear Collider (CLIC) project. This thesis was made because the price of a single disk is a major cost driver of the CLIC project. The cost of a disk is composed mainly of the time consuming ultra precise machining with expensive diamond tools. The surface roughness Ra 25 nm and shape accuracy 5 μm requirements are the reason for the need of ultra precision. The total number of disks to be produced for CLIC at 3 TeV center-of-mass energy is roughly 4.1 million, thus the project is a typical example of mass production where learning curves can be applied. Based on the theoretical calculations, changing the surface roughness Ra from 25 nm to 100 nm decreases the total ultra precise machining from 644 minutes to 311 minutes with current tooling and machining parameters. The effect of shape accuracy to the cost proved to be hard to estimate. Going from shape accuracy 5 μm to 20 μm will cause cost savings in tooling and machinery. However, because the connection between shape accuracy and surface roughness is not exactly known at micro- and nanometre scale and the wear of the replacement for a single crystal controlled waviness diamond tool is not exactly known in series production, the cost effect of shape accuracy relaxation could not be estimated. Changing the ultra precise machines to cheaper precise machines will only have a marginal effect on the cost of the accelerating structure. In conclusion, it can be said that after applying the learning curves, the ultra precise machining times drop to a level at which the project is viable. Regardless of the learning percent used, the change of surface roughness Ra from 25 nm 100 nm cuts the ultra precise machining time roughly to half. This means that the total cost of the accelerating structures reduces 6–16 percent depending on the number of production lines and the applied learning percentages, which will probably be in the order of 90 to 95 percent. To verify the results of this thesis, machining tests with similar tools and machining parameters should be done.