Minutes of the 14th LHC Insertions Upgrade Working Group held on 15 May 2008
Present: R. Assmann, F. Butin, F. Cerutti, S. Fartoukh, P. Fessia, J. Kerby, N. Kos, J.-P. Koutchouk, A. Mereghetti, R. Ostojic, H. Prin, E. Todesco, L. Williams
Excused: M. Giovannozzi, F. Zimmermann
Invited: F. Borgnolutti, F. Regis
1. News and approval of the minutes of the last meeting
The minutes of the last meeting are approved without any comments.
Ranko announced considerable advancement in the optics studies. Stéphane will give a dedicated presentation on the 29th of May. Ranko also mentioned that the US collaboration will most likely supply the D1 separation dipoles based on RHIC design, a proposal of which will be presented in the US-LARP review on 19-20 June. In view of the goal to have the conceptual design of the Phase-I insertions finalized for mid-2008, an internal CERN review will be organized beginning of July at CERN.
2. Quadrupole design study for the LHC phase I upgrade (F. Borgnolutti, ppt file)
Frank presented the results of his studies concerning the cable arrangement in the quadrupole cross section that gives maximum critical gradient.
Frank introduced his talk by recalling the constraints of reusing the main dipole inner and outer cables wound in a 2 layer configuration. The calculations presented were done for a 120 mm aperture. Two types of arrangements are considered: the conventional one named "normal grading" that uses one type of cable per layer, and the second one named "special grading" that combines cable type 01 and 02 in the outer layer.
Computations taking into account the mechanical constraints as well as the field quality considerations show that the highest critical gradient that can be obtained for both designs is almost the same, and is 148 T/m. The "special grading" layout offers a larger set of solutions close to the highest critical gradient, but it requires in most of the cases a larger amount of cable 01. On the other hand, the quantity of cable 02 is significantly lower. From the graphs , Franck showed that for a conventional arrangement there is a clear linear dependence between the critical gradient and the amount of cable used. The relationship between the amount of cable 01 and 02 is also linear in this case. The situation is completely different for the "special grading", where the critical gradient reaches a flat top for a certain amount of cable of either type 01 or 02; no obvious dependence can be drawn between them.
Franck concluded with two examples of cross sections of both grading types showing the cable layouts that give 148 T/m. For each of them he gave the field quality in terms of b6 and b10 units and the number of turns of each cable per coil. The contribution of the iron yoke taking into account 37 mm thick stainless steel collars, is in the order of 2 to 3% that lead the maximum critical gradient of 153T/m.
Ranko remarked that b10 has to be reduced as far as possible and that b14 has to be checked. Paolo will confirm the layer jump feasibility.
3. Study of the quadrupole collar structure (F. Regis, ppt file)
Federico completed the previous presentation with the calculations of the collar surrounding the coils: the collar thickness, key positioning and dimensioning, as well as the forces in the system and the stresses induced.
The first approach to define the collar thickness is to scale the two well known magnets: the main quadrupole MQ used in the LHC arcs and the Fermilab MQXB used in the present triplet for the Q2. After computing the Lorentz forces at the coil surface as a function of aperture, the calculation of the collar thickness is made by solving an equation depending on the aperture radius and the coil thickness to get the same displacement as the magnet used for scaling. The analysis demonstrates that the magnetic forces increase with aperture and that using MQ as a reference is more stringent than using MQXB: for a 120 mm aperture quadrupole the collar thickness is 39 mm when scaling MQXB, and 49 mm when scaling MQ.
The azimuthal stress on the pole is obtained by summing up the residual stress after powering (estimated less than 25MPa), the cool down effects and the pre-stress induced by collaring. Federico underlined that the required level of pre-stress at warm seems to be near the Apical creep limit, and that the azimuthal forces slightly increases with the collar thickness.
As the horizontal forces decrease with collar thickness, the key dimensioning was done using the smallest collar thickness and phosphor bronze material to be more conservative. The reaction forces on the keys depend on their number per quadrant and their angular position. Up to an angular separation of 24 degrees, a double key layout is more efficient than a single one. Beyond this angle the sharing between horizontal and vertical reactions does not give an effective contribution. The retained layout that provides a stiffer structure has 2 keys at 15 degrees.
The coil-collar section was modeled with finite elements for 120 and 130 mm apertures to evaluate coil radial displacement and bending effect on each layer, and to check the influence of the collar thickness at each step of the magnet cycle. The computation of the equivalent stress on 35 and 45 mm thick collars shows that the stress distribution is quite similar and that there is no significant difference in the high stress areas (coil and key interfaces). The critical conditions for these regions appear at warm after collaring. The proposal is to use a 35 mm collar for a 120 mm aperture magnet, and 38 mm for 130 mm aperture. Further studies will be made to refine the geometry of the keys and collars.
Paolo mentioned that the FE model has to be updated for the final aperture in order to take into account the quench heaters, ground insulation and coil protection sheets. He particularly insisted that the new cable insulation scheme has to be taken into account and the results reevaluated according to measured coil E-modulus.
4. Preliminary study of the inner triplet beam screen (N. Kos, ppt file)
Nicolaas summarized the main requirements which have to be taken into account for the design of the beam screens. He then listed a series of proposals that essentially derive from the current beam screen design and materials: magnets from Q1 to Q3 could be equipped with the same cross section made out of 2 mm P506 stainless steel which is estimated to be stable to contain the forces induced by the eddy currents after quench. The inner surface will be copper coated to homogenize the temperature and give low impedance. The outer surface could be equipped with 4 cooling tubes to evacuate each 1 W/m heat loads. The Q1 absorber would be placed inside the beam screen.
Based on the above requirements, Nicolaas presented a race track cross section scaled from the existing beam screen design with two extra cooling pipes. To standardize the orientation inside the cold bore and avoid the horizontal and vertical positioning, the flat surfaces are turned by 45 degrees. The race track design optimizes the aperture requirements, but the past experience in forming the beam screen shows that the control of the curvature is an important issue. Nicolaas reported that the companies who produced the arc beam screens are not interested in manufacturing the small quantities for the triplet upgrade. The option for simplifying the manufacturing process is to approximate the round shape with facets obtained by locally pressing the edges. An octagonal cross section was envisaged with the drawback that it reduces the clear aperture in both horizontal and vertical planes. This led to a shape called "optimized octagonal design" where the horizontal and vertical facets are split in two to increase the aperture in both planes. This shape gives more stiffness and is more resistant to the quench forces.
Two possibilities were envisaged for the absorber design inside Q1. The first one consists in an 8 mm thick stainless steel pipe inserted in the beam screen. This solution has the drawback that it partially masks the pumping holes that have to be increased, and requires machining many pumping holes through the absorber. Nicolaas drew attention to the total beam screen weight that becomes 240 kg. The second solution consists in four copper machined sections fixed onto the beam screen surface. In addition to the ease of manufacturing , the weight is reduced to 170 kg for a 10 m long beam screen, which is still considerable.
Nicolaas recalled the formulas from the Vacuum Technical Note 01-13, EDMS 350449 for defining the radial beam screen size at room temperature and the half aperture at operating conditions depending on the cold bore diameter and tolerances, the thermal contraction coefficients and all other parameters to make the assembly feasible. He presented a table that summarizes the potential gain in aperture showing that the main contributor to the dispersion is the cold bore inner diameter tolerance. A possible method to minimize this factor is to measure the cold bore inner diameters at reception and use the measurements instead of nominal tolerances to finalize the beam screen dimensions. This requires that all cold bores are delivered before finalizing the BS design. The question was raised whether a heat treatment is necessary to fix the geometry after assembly. It was also mentioned that the beam screen can be straightened after mounting.
Nicolaas presented a table summarizing the heat load along the triplet that has to be completed after agreement on the beam screen cross-section and the energy deposition calculations. The 3 K temperature difference between the liquid Helium circulating in the capillaries and the beam screen inner surface measured in the current arc configuration has to be re-estimated due the substantial mass distribution in the new cross section. Ranko remarked that the operating temperature considered for the beam screen is 5-20 K with a possibility to go up to 40-60 K (see presentations made at the LIUWG#3 and LIUWG#5), which will not affect the beam screen geometry but has to be confirmed by thermal studies.
An interconnection study is presently on the way to define how to connect the four capillaries in the tight space around the beam pipe. This has to be completed by the RF contact junction between the magnets and particularly at the level of the Q1 absorber.
SF commented that the so-called "optimized octagonal design" leads to a loss of aperture quickly estimated to n1~1 (~ 3mm) w.r.t. to a race-track shape beam-screen oriented according to a specific crossing-scheme chosen for the IR (presently vertical in IR1 and horizontal in IR5). On the other hand it has the obvious advantage to standardize the interconnect type in both IRs and will allow, at moderate cost, to change a posteriori the crossing-plane in IR1 and IR5. Indeed, such an option for a race-track shape b.s. leads to a loss of mechanical aperture of the order of 5 mm (i.e. outer diameter of the b.s. capillaries) when the beam xing-plane is no longer oriented according to the major axis of the b.s. In any case a detailed aperture study is needed to balance the benefits and draw-backs of the new shape proposed for the beam-screen.
=> ACTION: ABP
5. A.O.B.
S. Fartoukh, R. Ostojic and H. Prin