During 1997 the HCAL group will construct a full size wedge prototype. This prototype will be instrumented with scintillators and tested with cosmic rays. Finally it will be taken to the H2 test beam at CERN for more extensive testing. Part of the HCAL project is to construct a test beam motion table so the wedge can be scanned through the H2 test beam.
The wedge is assembled from back to front, starting with the stainless steel back plate. The back SS plate is attached to a flat assembly plate, as shown in Fig. 2.15. Shear keys and the alignment/shear pin are positioned into place. (The top of the key is has a 1mm radius and the upper plate keyway is oversized by 0.5mm, to allow an easy fit.) Then the next brass absorber plate is positioned into place onto the alignment pin, then lowered onto the shear keys. The plate is pushed transversely (according to the direction of gravity that this wedge will experience in the completed barrel), to engage the key in the oversized upper keyway. Then the plate is bolted down with M16 bolts. Keys and the alignment pin are installed on this plate, then the next absorber plate is positioned, and so on until the wedge is completed.
Special tooling required for stacking the wedge are the assembly table, shown in Fig. 2.15, and a plate lifting fixture, shown in Fig. 2.16. The plate lifting fixture is designed such that the plate is not deformed during the lifting and assembly process. The plates are required not to be elastically distorted by more than 0.2mm during the lifting process. This constraint arises because the key ways will not fit together if the plates sag too much during lifting.
The HCAL absorber plate lifting fixture is used to facilitate HCAL wedge assembly. Its purpose is to lift individual absorber plates and position them on top of one another so that they may be bolted together. The philosophy behind the assembly procedure is to maintain each individual absorber plate to a high degree of rigidity so that it may be accurately positioned with respect to it's neighbor's shear key. Once the absorber plate lifting fixture is bolted to each absorber plate it acts as a rigid spine, stiffening an otherwise flexible plate. The fixture is bolted to the plate in sixteen locations to ensure a maximum deflection in the plate of less than 80 microns (0.003 inches). This maximum deflection was calculated based on a 20mm minimum cross-section of the copper absorber plate. Stresses induced in the copper plate were equally low, on the order of 1.4 MPa, where the yield stress for copper is about 70 MPa.
Calculations were also done for the stainless steel outer and inner radius absorber plates. On a plate with a 63.5mm cross-section, a maximum deflection of 46 microns (0.0018 inches) was obtained. Stress values in the plate were 0.6 MPa, compared to a yield stress in stainless steel of 210 MPa.
The maximum bolt forces were calculated based on the reaction forces at the points were the plates were fastened. The maximum values obtained were around 2500 N (560 lbs.) Since we will be using the existing M16 threaded holes in the absorber plates to attach to the lifting fixture, this is clearly within the acceptable range.
The lifting fixture itself is constructed of an American Standard I-Beam (15 42.9) (designation 15 42.9 means the I-beam is 15 inches (381 mm) in depth and weighs 42.9 lbs/ft (64 kg/m)) with four similar I-beam outriggers welded to it. Fig. 2.16 shows the plate lifting fixture.
Machined steel plates, 40mm (1.5 inches) thick, are bolted to the outrigger section of I-beam and shimmed flat with respect to each other. Slots are machined in the plates in order for them to accept the bolting pattern of all the absorber plates.
The lifting fixture was designed for rigidity so its deflection under maximum load is only on the order of 25 microns (0.001 inches).
The total deflection of all components in the wedge assembly is about 100 microns (0.004 inches). This requirement is imposed by the assembly procedure and is necessary to position the absorber plates on their shear keys. The plate lifting fixture deflection gives a factor two better positioning tolerance than the 200 microns required in order to ease wedge assembly.
The HCAL barrel wedge load fixture, Fig. 2.17, is designed to simulate the load conditions that an HCAL rail wedge will encounter in its final assembled state. The rail wedge in the HCAL barrel assembly is the wedge that sees the greatest stresses and for that reason it is the one that should be simulated. Since the first prototype wedge is to be a "common" non-rail wedge, provisions for adding a "rail" to it must be accommodated. Bolting and pinning a "rail" to the outer stainless steel absorber plate will be done before the wedge is structurally tested. At the wedge to wedge connection locations in the stainless steel inner and outer radius, the prototype wedge will be attached to the load fixture by a series of mechanical actuators. These actuators will allow loading of the wedge to provide comparable deflections to those obtained in the finite element analysis results.
As part of our overall quality control, each wedge will be tested for structural soundness at the manufacturing site. The wedges will be loaded by the load test fixture, and deformations measured.
The wedge will be placed in the H2 test beam at CERN. In order to move different parts of the wedge into the beam, a motion table is required. The table will also be capable of azimuthal motion so that the beam can be placed in each projective tower of the wedge. The 2 degrees of motion are achieved by: motion via a vertical axis fixed to the floor; j motion via rotation about a horizontal axis that is part of the motion table. The angle of rotation j is ± 20o. Ultimately, the HCAL group will build 2 prototype wedges. The motion table will be designed to be able to support both of these wedges of the HCAL barrel. In addition, the table will be able to accommodate a 40 degree sector prototype of the HE endcap hadron calorimeter. It will also accommodate prototypes of the ECAL barrel and endcap calorimeters. Thus a full 40 degree sector of the CMS central calorimeter can be held on the motion table. Fig. 2.18 shows the layout in the H2 beam line. The position of the CMS prototype calorimeters are shown in 2 configurations: one where the beam is traveling at h=3, and the highest h edge of the calorimeter is just out of the beam; and the other position where the beam is at =0, and the lowest h edge of the calorimeter is just outside the beam. Fig. 2.19 shows the isometric view of the motion table loaded with 2 HB wedges, the HE prototype, and ECAL EB and EE prototypes. Also shown are the muon chamber and the muon return flux iron.
The platform also rotates around the vertical axis y. The rotation angle is 90o. The rotation around the y axis (90o) will take 30 minutes maximum and the rotation around the j axis will take 15 minutes maximum. Accuracy of positioning around both axis is ±1 mm. Both motions will be controlled by computer. The turntable will have computer-readable position indicators.
The internal structure of the wedge cannot be measured after assembly. We will test the rigidity of each wedge after construction to verify its mechanical integrity. We will use the wedge test loading fixture at the absorber factory, installing each wedge into the load fixture and applying reference loads. Then we will measure the deflections of the wedge. This allows us to test the internal structure of the wedge.
The dimensions of the wedge are critical. We will insert scintillator tray templates into each slot to verify go/nogo. Each wedge will be surveyed to assure that its dimensions are within tolerance. Finally the entire barrel half-rings will be assembled at the factory to verify that the completed structure is within tolerance. The support cradle and spider tooling will be reused for final assembly at the CMS assembly hall. The half barrels will be assembled in the nominal configuration with the symmetry axis horizontal. The half barrels will be surveyed, and any corrective machining or shimming will be done. Finally the half barrels will be disassembled and the individual wedges and tooling shipped to CERN.
Each HCAL half-barrel consists of 18 wedges, bolted together at the inner and outer radii. The bolted-together stainless steel front plate and back plate polygons act as structural members to minimize the deflections. The half-barrels rest on rails attached to the vacuum tank inner wall, at 3 and 9 o'clock.
The wedges are attached together by brackets which are bolted to the back face of the SS outer radius plate. The bracket is designed to take and shear forces between the wedges. At the inner radius, the wedges are bolted together through slots machined in the SS inner plate.
Fig. 2.20 shows the detail of the outer stainless steel plate. The notches for the cross-bolting of wedges are also shown as well as the bolting together of two wedges Fig. 2.21 shows a detailed view of the inner stainless steel plate. The wedges are bolted together using M20 bolts between adjacent wedges at the inner plate, brackets between adjacent wedges at the outer plate.
The HCAL half-barrels reside inside the CMS vacuum tank, resting on rails welded to the inner stainless steel wall of the vacuum tank. The half-barrel is much more rigid than the vacuum tank inner wall. Thus, a non-uniform load distribution would be expected on the rails, with most of the load being taken at the minimum and maximum z extent of the half-barrel. To make the half-barrel more compliant, a load distribution system has been designed.
The weight of the half barrel is carried to the vacuum tank through the wedges at the 3 o'clock and 9 o'clock positions. These wedges have a support structure built into them to mate with the rail structure on the vacuum tank.
The HCAL half barrel rests on the two rails welded to the inner wall of the vacuum tank. The forces and deflections on the rails are of great interest to us. To understand this system, we made a finite element model of the rails, the vacuum tank, and the vacuum tank attachment to the muon support iron. The model is shown in Fig. 2.22. We modeled ½ of the vacuum tank taking advantage of its mirror symmetry about the midplane. We modeled the worst case scenario as if the vacuum tank rails were as far apart as the vacuum tank drawing tolerances allow. In this case the HCAL applies the maximum torque on the rails. The rails are split at the midplane. A coarse model of the vacuum tank and rails was formed. In this model, the 2 rail halves (on either side of the midplane) are not fixed at the midplane. The rail welds were not modeled in detail. Rather they were treated as two strips, one on top of the rail, and one below. The calorimeter barrel halves had no contact during the deformation. The vacuum tank and rail deflections were calculated using the coarse model. The results are shown in Fig. 2.23 and Fig. 2.24.
To understand the effects on the stresses due to the muon system support, 2 models were created: one with ribs connecting the central barrel wheel to the vacuum tank (See chapter 9 of the Magnet TDR), and the other with no ribs and constraints in the contacts between the ribs and the outer shell. The details of the central barrel wheel attachment had no effect on the rail and inner vacuum tank stresses. This is as expected. The flange disk connecting the outer and the inner vacuum tank shells isolates the inner shell from distortions of the outer shell.
A refined submodel was created with a detailed description of the welds, Fig. 2.25 . The refined submodel was used to calculate stresses on the rails, welds, and inner shell of the vacuum tank. Fig. 2. 26 shows stresses in the inner shell of the vacuum tank, which are small. Fig. 2.27 shows stresses on the rail material. The stresses are again small. Fig. 2.28 shows the rail deflection within the length of the half-barrel. This deflection must be accommodated by the plunger/Belleville washer system.
The results of the FEA study are:
a) There was no effect of the details of the muon system attachment.
b) The inner shell deflections, Fig. 2.23 and Fig. 2.24, were within +- 10mm.
c) The total rail deflection over the length of the half-barrel was 2.4mm, but the relative deflection was only 0.6mm. The relative deflection is important for the calculation of the stroke of the suspension plungers. Fig. 2.28 shows how the rail cross section moves as a result of the calorimeter load.
d) The stresses in the rail material (SS 304) were well below the limit of elasticity, sy = 245 MPa (35ksi). As shown in Fig. 2.27, the maximum rail stress was 97 MPa (14ksi).
e) The maximum stresses in the rail welds are acceptable, as shown in Fig. 2.29 and Fig. 2.30 . Ignoring the stress concentration at the upper radius of the rail (important only for fatigue issues which are irrelevant here), the stress in the weld is about 90 MPa (13 ksi). This is less than ½ of the maximum allowed stress of 210 MPa (30 ksi). Thus our safety factor on this weld is 2.3.
f) The stresses in the inner shell are shown in Fig. 2. 26. They are well within allowable limits.
As pointed out earlier, the inner vacuum tank/rail is a critical
component of the HCAL support. Therefore an independent FEA analysis
was performed to check the results presented above. There was
good consistency between the two analyses, giving us confidence
that these results are correct.
The rails deflect by 2.4mm as the HCAL half barrel is installed into the vacuum tank. Of this, 1.8mm is an overall drop in the vacuum tank height, and 0.6mm is due to local deformation. The weld stresses are acceptable.
The calorimeter half barrel is guided and supported by the rails welded to the vacuum tank inner shell. The rails have a relative deflection of 0.6mm under the weight of the half barrel. The half barrel is provided with a layer of compliance to compensate for the rail deflection and rail non-flatness. For this purpose, each side of the half barrel has a row of spring preloaded plungers. See Fig. 2.31.
The barrel is suspended on disk springs that exert force on the plungers. These plungers transmit the load to sliding pads connected to the plungers through hemispherical surfaces. The pads are made of bronze and have a Teflon coating on the sliding surface.
Fig. 2.32 shows details of the support system. The rail support system consists of the barrel rail (1) welded to the outer SS of the wedge (not shown). The rail houses bronze bushings (2). Plungers (3) have on their lower end the sliding pads (4) with hemispherical backs. This hemispherical back allows the pad to follow any distortions in the vacuum tank rails. Disk springs (5) are mounted on the top end of the plungers. A retainer ring (6) limits the plunger stroke downward, and prevents the plunger from falling out of the bushing. The disk springs limit the plunger stroke upwards. The disk springs push against the top bar (7). The top bar is connected to the barrel rail by bolts (8) and spacers (9).
The barrel weight is transmitted through the barrel rail (1), bolts (8), top bar (7), springs (5), plunger (3), and pad (4) to the vacuum tank rail (10) which is welded to the inner shell of the vacuum tank (not shown).
The size and number of disk springs determine the plunger stroke and the rigidity of the spring set. We have considered a variety of spring combinations and have chosen the one shown in Fig. 2.31. The optimal spring set consists of commercially available springs with OD 70mm and spring thickness of 5mm.
The maximum load on the spring set is 133.8 kN (3104 lbs.) (springs are flat). The maximum load on one plunger is 75.8 kN (.7104 lbs.) The stiffness of the set is 22.5 kN/mm (12.8104 lb/in.) The stroke after the dead weight of the calorimeter is applied is 2.6mm which is sufficient to cover the deflection and non-flatness of the rail.
We will calculate the specific pressure for the load =2F=21.7104 lb=151.1 kN. The doubled load will compensate for the uneven pressure on the pad as a result of the friction in the hemispherical pad support; and compensate for the rail non-flatness. The specific pressure is s = 2F/A , where A is the sliding pad area and equals 70110 mm2.
s = 15400/(70110) = 19.6 Mpa. (4)
CERN has performed a test for us using a 0.3mm thick Teflon pad for the sliding surface. These test results show that the material will stand s = 46.6 MPa which is more than twice our requirement of 19.6 MPa.
We will use the same material and gluing technique as used in the CERN test. In addition, we plan to perform our own test to study the complete rail support design. We will check the disk springs (maximum load and stiffness), the functionality of the hemispherical supports, and the coefficients of static and dynamic friction.
The rail test fixture is shown in Fig. 2.33 Shown are the base, and the sliding rail, constrained on the sides by rollers. The rail sits on the sliding surface of the base, and is loaded by two plungers. The rail is driven by the hydraulic cylinder. A force gauge will measure the friction force and the pad load.
The barrel rails, shown in Fig. 2.34, are welded to the outer SS 304 plate of the calorimeter wedges at 3 o'clock and 9 o'clock. A 3D finite element model was created in order to study the stresses in the rails and welds. The results are shown in Fig. 2.35. The maximum stresses are approximately 140 MPa (20 ksi).
The barrel support system (as shown in Fig. 2.31 including plungers, pads, support bars, and disk springs) is a heavily loaded system. This loading is due to both the high weight of the calorimeter and the confined space for the support system.
We have studied stresses in the barrel support using a finite element model, shown in Fig. 2.36. Results are shown in Fig. 2.37. The maximum stresses in the top bar, area "a", are caused locally by the bolt head as the bolt bends. These stresses can be ignored because area "a" is small and the local contact stress can be much higher than the limit of elasticity, sy = 245 MPa (35 ksi) for SS 304. The stresses in other areas are less than 140 MPa (20 ksi). The stresses on the bottom plate, Fig. 2.38, less than 112 MPa (16 ksi).
The load capacity of a M30 bolt is F= syA , where sy = 280 MPa(40 ksi) for rolled thread, and A is 561 mm2, (0.833 in2) the tensile stress area for M30 bolts. The load capacity is F = 148.5 kN (33,312 lbs.), a satisfactory capacity, since our nominal load is 75.8 kN (1.7104 lbs).
Table of Contents
2.8.6 Prototype Wedge 1172.9 WEDGE CONSTRUCTION QUALITY CONTROL 1252.10 BARREL DESIGN 1252.10.1 Overview 1252.10.2 Wedge layout 1252.10.3 Bolt patterns 1252.11 BARREL MOUNTS (RAILS) 1282.11.1 Cryostat/rail FEA 1282.11.2 Forces and deflections 1392.11.3 Calorimeter support system 1392.11.4 Calorimeter support FEA 143