Structural behavior of electrical post-insulators for the MITICA Beam Source Mock-Up SOFT 2024

Previous R&D activities conducted on the MITICA Beam Source (BS) identified the need for an experimental validation of the 1 MV High Voltage (HV) holding in vacuum of its accelerator, using a full size Mock-up that reproduces the external geometry of MITICA BS relevant for HV holding. This paper describes the analyses and tests of a critical structural component of this Mock-Up, referred to as “post-insulators”. This component is made of PolyEtherEtherKetone (PEEK), an engineering thermoplastic, and has a double-function of electrical insulator between the stages of the accelerator and of mechanical support carrying the cantilevered structure of the Mock-Up. Being critical components withstanding high loads, a Design by Experiment approach has been followed to complete the Design by Code and Design by Analysis of the PEEK insulators conducted before the construction of the experiment.

In the analyses, the stress–strain pattern of the PEEK post-insulators is simulated using two FEM models, so as to evaluate the behavior during operation and to support the interpretation of the results obtained during subsequent tests. A first model uses accurate geometry, including screws, threaded inserts and bolts connections used to fasten the PEEK post-insulator to the stainless-steel flanges. The second model is a simplified version, where insulator and flanges are connected directly with bonded contact type.

Experimental tests are performed on four post-insulator samples, using a uni-axial hydraulic test system and a particular test set-up, in order to establish the quasi-static mechanical behavior of the component under a precise combination of tensile-bending stress.

The test results are positive, however, the numerically-derived stiffness is overestimated compared to the experimental data. Such discrepancy has consequences in terms of total deformations, equivalent stress and natural frequencies of the MITICA Mock-Up BS. These differences have been quantified by calibrating the numerical models of the tensile test with the experimental results, and the Mock-Up structural simulations have been re-ran with the real insulator stiffness. Lastly, an explanation for this discrepancy is investigated by developing an accurate sub model of the M14 screw, threaded insert and PEEK bulk material, showing how modeling the screw-threaded insert connections with bonded connections overestimates the stiffness of the model compared to the real-case scenario.