Speaker
Description
The Fermilab Main Injector Complex operates with 208 quadrupoles magnets of the IQ type, nominally 84” (IQB) 128 each, 100” (IQC) 32 each, and 116” (IQD) 48 each. The 84” quads were originally all reused Main Ring quads, longer quads are new construction. In constructing the Main Ring quadrupoles a layer of pre-impregnated glass tape was applied to the bare coils and cured, coils were wrapped with an additional layer of dry glass tape and were assembled into the cores. The whole magnet was then vacuum impregnated with epoxy. This resulted in a robust mechanical structure. The IQB have failed over the last 50+ years at a manageable rate, where failure is defined as developing a ground fault sufficiently severe to require replacement in the tunnel. When a magnet fails, it can be rebuilt.
The new magnets used the same basic construction approach, attempting to improve reliability for the Main Injector, more of the manifolding was encapsulated by epoxy loaded with glass beads and a thin layer of G 10 was added between the coils and the core as additional ground insulation. Both changes have proved detrimental. Burying the braze joints makes them much more time-consuming to repair, requiring a magnet change rather than the in-situ repair often possible when the joints are exposed. Even worse, the Imperfect vacuum impregnation of both sides of the impervious G 10 sheets turned them into slip planes that concentrated the stresses from differential thermal expansion of coils and cores and caused cracks in the epoxy insulation at the ends that became ground faults. With a high failure rate of the newly constructed magnets in the first two years of operation, an improvement project was launched, relying on vacuum impregnating insulated coils and installing them in existing cores secured only in the middle to allow differential expansion without a build-up of shear stress. The first spares were not tested exhaustively until January 2022, when an IQC failed in the tunnel. Two of the new spares were installed and failed within days as the coils were inadequately secured against the lateral magnetic forces under power and the ends moved up to a millimeter and more at full excitation, concentrated in a few thin spots in the copper tubing where it joined the ceramic insulators leading to work-hardening. A new scheme was devised to secure the coils against lateral motion while maintaining the core geometry during assembly and allowing removal of coils without damaging the insulation. As of December 2024, a few IQC magnets have been rebuilt with this approach. Measurement of coil position as a function of excitation has shown that the end motion has been controlled. One magnet has been subjected to test runs at repetition rate of 32 cycles per minute, consisting of a ramp up at approximately the 120/0.3 GeV rate considered for ACE¬¬ MIRT but a slower ramp down due to power supply constraints. The longest test ran 72 hours. No failures appeared and when the coil position was monitored during a slow cycle after all the fast ramps, no motion was evident. From this we conclude that the new configuration is sound, but the custom fit-up of each magnet will limit the rebuilding rate unless improvements are made.
