Figure 1: Collaring test of the final cross-section of the D1 in KEK. Yoking press is in the background
A key element of the High Luminosity LHC interaction region is the superconducting separation dipole D1, which will replace the resistive dipole presently installed in the LHC. Thanks to the superconductive technology of Nb-Ti, this magnet will increase the field from the present 1.28 T to 5.6 T, thus reducing the overall length from 25 m to 7 m (A first layout for the High Luminosity Upgrade) and providing at the same time a 30% stronger kick. This large saving in terms of space along the beam line is crucial to allow the installation of longer triplet and of the crab cavities, which are among the main pillars of the luminosity upgrade.
The D1 upgrade was one of the first HL LHC magnets to be studied in collaboration with KEK. Activities started in 2011, led by T. Nakamoto, with the contribution of a young scientist (Q. Xu, now in IHEP, Bejing) focusing on the conceptual design. After the selection of the triplet aperture of 150 mm, KEK has developed an engineering design in 2013-2014: the construction of a short model has been carried out during 2015. The magnet is not trivial due to its large aperture, the relatively high field for Nb-Ti technology, the tight requirements on field quality, and the constraints imposed on the mechanical structure. Since one needs to maximize the iron dimension to limit saturation and fringe fields outside the magnet, the KEK team opted for a concept where the collars are thin spacers and forces are kept by the massive iron: this is the same mechanical structure that was used for the triplet magnet MQXA presently installed in the LHC. Since forces are kept by the iron, the assembly of laminations around the coils and the control of the compression given under the press (see Fig. 1) is critical and entailed in the iteration in the design.
The first model has been completed in March 2016, and test has been done in April and June. The test started at 4.5 K (see Fig. 2) since the test station underwent an upgrade and some additional cross-checks were required. At 1.9 K, the magnet had a first quench at 9.5 kA, i.e. ~60% of the short sample, and approached the 12 kA nominal current (75% of short sample) after ~10 quenches. After a thermal cycle, the magnet showed good memory (i.e. it started from the same values reached before the warm-up).
Figure 2: Training of the first D1 short model in KEK (Image credit: M. Sugano)
The magnet eventually reached 500 A more than nominal current in the second run, but exhibited an erratic behaviour around these current values. At the same time, the strain gauges gave a clear indication of a lack of support in the coil already at around 8 kA (see Fig. 3). This suggested not to push the training towards the goal of ultimate current, take all the information about field quality and quench protection, and possibly have a second assembly with an increase of prestress (a similar step is being done for the first short model triplet). “We will soon have a review in KEK to present the test result and define the next steps”, says A. Musso, who follows the collaboration from the CERN side. “A second model is foreseen in the HL LHC planning for the year 2016-2017, and the construction of the long prototype will take place from 2018.”
Figure 3. Stress variation during the energization of the magnet, with flattening at 8 kA indicating coil unloading (Image credit: M. Sugano)