Minutes of the 7th LHC Insertions Upgrade Working Group held on 13th December 2007
Present: F. Bordry, O. Brüning, F. Cerutti, S. Fartoukh, M. Giovannozzi, K.H. Meß, R. Ostojic, V. Parma, E. Todesco, R. Van Weelderen, E. Wildner
Excused: J.-P. Koutchouk, H. Mainaud Durand
Invited:
1. News and approval of the minutes of the last meeting
The minutes of the last meeting are approved without any comments.
2. Powering strategy for the new inner triplets (F. Bordry, ppt file)
In its present version, the powering scheme of the LHC inner triplets is the same in the four experimental insertions of the machine. As the MQXB and MQXA magnets, used in Q2 and Q1/Q3 quadrupoles have different nominal currents (I_nom=11390 A and 6450 A, respectively), powering in series of the whole inner triplet is not possible. The most economical option was chosen, consisting of two nested PC loops, the outer loop of 8kA powering all magnets in series, and an inner loop supplying an additional 6 kA to Q2. At a later stage, a bipolar 600 A trim was added to Q1. The PC controls include a dedicated decoupling logic such that an independent adjustment of the quadrupole gradients in Q1, Q2 and Q3 is possible. This is in particular useful for beam-based measurements such as K-modulation or (a*,b*) measurement and adjustment. The inner triplet converters are standard LHC converters but specific protection devices are implemented, such as free wheel diodes in each loop, which are needed due to the very different time constants of the three circuits. Finally, the installation of the PC's in the tunnel was found to be a real headache as the appropriately shielded slots around the IP's are very limited, in particular in point 5. This aspect is far from being a detail and has to be taken into account from the start for any upgrade of the LHC insertions.
For Phase I, the four quadrupoles in the triplet, namely Q1, Q2a, Q2b and Q3, could be a priori with the same operating current. The nested PC scheme is therefore no longer necessary in this configuration. Several options are possible: the simplest (scheme A) would consist of a single 13kA power converter, powering in series the four triplet quadrupoles (estimate made for 130 mm aperture quadrupole with an operating gradient of 122 T/m, see E. Todesco et al. in PAC'07 Proc.). This scheme would be very economical in terms of cost and tunnel space, and beneficial in terms of control and stability requirements (partial self-compensation from magnet to magnet for tune modulation caused by PC ripple). On the other hand, it is not very attractive in terms of optics flexibility. At the other extreme, a configuration with four identical PC's could be envisaged to power independently Q1, Q2a, Q2b and Q3. Intermediate schemes with three 13 kA PC's (Q2a and Q2b powered in series) or two 13 KA PC's (i.e. two independent circuits for Q2a/b and Q1/3) with gradually some advantages and draw-backs in terms of cost, space, controlability and optics flexibility, are also possible.
Frédérick's presentation triggered several comments and questions.
RO asked about the additional space that would be needed for the new 13 kA PC(s) in Points 1 and 5. FB and KHM replied that the relevant scaling parameter for the volume of the power converter is its voltage, therefore depending on the inductance of each circuit.
OB asked about the additional space which would be needed to accommodate the power converters of a cold D1. FB replied that the present PC for two D1 magnets (MBXW) is located on the surface (left/right powering in series in IR1 and IR5). Therefore, the option of a cold D1 in IR1 and IR5 would increase the complexity of integration of the new inner triplets and their powering equipment.
OB and SF showed some reluctance for scheme A arguing against the total lack of optics flexibility in this configuration (strictly the same current in Q1, Q2 and Q3) and, in particular, the impossibility of performing K-modulation for each quadrupole separately. FB suggested the possibility of equipping each quadrupole with bipolar trims (e.g. with 600 A PC's as is presently the case for Q1, scheme A-bis), which should not add too much complexity to the system. OB and SF felt that this could indeed be a good compromise. SF remarked that this scheme would exclude the option where Q1, Q2a, Q2b and Q3 would have exactly the same length but with a net reduction of the gradient in Q2 (this solution would indeed be very economical in terms of spares and production cost, while a bit less performing than an optimized symmetric triplet option).
3. Quench protection and other related issues (K-H. Mess, ppt file)
KHM first reminded that quench protection must be considered from the beginning as an integral part of magnet and string design, and that the energy deposited in the magnets after a quench should be minimized by all means. In this respect, he insisted that in view of the considerably larger stored energy of the new triplets, the magnets should be decoupled using diodes or thyristors, and an energy extraction system provided (these elements are not part of the present triplets). Quench heaters will also be required to distribute the resistance and reduce the voltage during quench.
KHM reviewed several strategies for protecting the new inner triplet, and then focused on the powering scheme A-bis consisting of one 13 kA power supply and four 600 A bipolar trims for each quadrupole, which seems the most appropriate in terms of cost and tunnel occupancy. In this configuration, two possible options for protecting the magnets can be envisaged: one requiring cold diodes, cold busbars, and DFBX connected via water cooled cables to the power converters, and the other based on an HTS link between the magnets and the PC's. The latter solution seems very attractive in all aspects, in particular as the busbars and DFBX boxes with all their complications would not be needed.
Finally, Karl-Hubert listed a number of practical points which should be kept in mind when designing and building the magnets, their protection devices and instrumentation. In particular, he reminded that any quench protection scheme will be made easier if the voltage withstand of the coils is made as high as possible.
The presentation triggered several comments and questions.
In the configuration A-bis for the triplet powering scheme, OB asked if a mixing of Nb3Sn and NbTi magnets in the inner triplet could complicate the quench protection system of the string. KHM replies that a priori not.
SF asked KHM and FB if the powering and/or the quench protection scheme could have an impact on the choice of the orientation of the quadrupoles (mirror symmetry w.r.t. the IP, lead-end of Q1 on the non-IP side and facing the lead-end of Q2a, conversely for Q2b and Q3). Indeed, SF further commented that this specific choice was made in order to partially or fully compensate the impact of systematic multipole errors over the IR. He also asked if any related constraints exist concerning the positioning of the triplet correctors, more precisely, if it would be preferable to attach them on the connection side or on the lyra side of the triplet magnets. To both questions KHM and FB replied that they do not see a priori any such constraint which could come from the magnet powering or quench protection scheme.
VP asked KHM an estimate of the time which would be needed to develop and produce the HTS links. KHM replied two to three years.
ET asked KHM a cost estimate of the two quench protection options proposed above. KHM replied that it is premature to give a figure. He mentioned, however, that, to first order the development and production cost of the HTS links is comparable to the cost of the DFBX boxes.
4. A.O.B. and follow up of actions
Next meeting scheduled for 24 January 2008: heat deposition studies in the new inner triplet (E. Wildner).
S. Fartoukh and R. Ostojic