LHCb 2001-065 EDR
Report on the Engineering Design Review for the modules of the LHCb Electromagnetic Calorimeter
Present: Ph.Bloch (referee), A.Golutvin, H.J.Hilke, V.Kochetkov, J.Lefrançois,
The engineering design of the ECAL modules has been successfully reviewed Thursday, March 15, 2001 by an external referee, Philippe Bloch, CERN/EP. A number of questions have been raised (see Appendix) concerning the general dimensions of the ECAL system, the detailed design of shashlik modules, the stacking procedure of modules, the cooling of photomultipliers, and safety aspects. The questions have been answered to the satisfaction of the referee, and details of the discussion are summarized in the following.
The LHCb electromagnetic calorimeter (ECAL) uses the “shashlik” technology. It is built from individual modules that are made from lead absorber plates interspaced with scintillator tiles as active material. Wavelength-shifting (WLS) fibers penetrate the lead/scintillator stack through holes, and are readout at the back of the sampling structure by photomultipliers. The ECAL structure is segmented into three sections with one type of module per section. All three types of module have an identical square size of 121.2 mm, but differ by the number of readout cells. The ECAL section closest to the beam pipe consists of 167 modules containing 9 cells each, the middle section has 448 modules containing 4 cells each, and the 2688 outer-section modules are made from a single cell. The detector is built in two separate halves from individual modules that are positioned in layers on two movable platforms and fixed to a surrounding frame.
A lot of experience in the design and construction of shashlik modules was gained in the past, since the design of the modules and the overall structure of the LHCb ECAL are very similar to the one of the HERA-B electromagnetic calorimeter. Therefore the review was focusing on potential problems that could arise from differences in the design of the two systems. Dedicated engineering design reviews on the overall integration of the calorimeter system and on its support structures are foreseen at a later stage and are not part of this review. However, questions related to these items were also addressed as far as they could affect the design of the modules.
3.1 General dimensions and access
On the general dimensions of the ECAL, the clearance of the modules at the front (including the calibration system) and at the rear (including space for cabling) was questioned. Precise engineering drawings of the front-cover and the end-cap design were presented. In front of the lead/scintillator stack an adequate space is allocated for hosting the splitters and connectors of the calibration fibers, and to form the WLS fiber loops. At the back of the modules, the detailed space requirements for routing of the signal cables, of the HV cables and of the cooling tubes were reviewed, showing that space has been allocated with a proper safety margin. Since the final choice for a photomultiplier tube has not yet been taken, the longitudinal space attributed to the photomultiplier and its base is chosen according to the longest tube amongst the present candidates.
The issue of access to the photomultipliers in case of a short circuit was briefly discussed. Since in average about 16 tubes are fed with one HV cable, a short circuit on one tube would affect a group of 16. Such a problem could be cured on a relatively short timescale since a short access (of order one day) would be sufficient to access the back of the out-sliding ECAL and to replace photomultiplier tubes.
3.2 Cooling of the photomultiplier base
The cooling of the photomultiplier bases, especially for the central modules containing nine cells, was discussed in detail. The maximum power consumption per resistive divider is estimated to be 2.5 W. The end-cap of each module, that covers the photomultiplier and its base, consists of an aluminum cover. It has been demonstrated that enough space is available in the cable area to include cooling tubes of 1 cm diameter that will allow injecting air to the back cover of each individual photomultiplier tube. Taking into account the thermo-conductivity coefficient for aluminum and an injected airflow of 3.4 m/s, the increase in temperature will be 14° C. Between the cable area and the aluminum cover, a free area of 4 cm is foreseen for the heated air to escape. A short-term temperature stabilization of ~1° C is sufficient for calibration purposes. The referee suggested looking into a liquid cooling system. However it was agreed that a detailed study of the cooling system would be performed on a mockup for the EDR on integration of the calorimeter system, which could show that air-cooling is sufficient.
3.3 Design of the shashlik module
Details on the mechanical tolerances on the sub-components of the shashlik module were requested, and have been presented. The tolerance of ±0.02 mm on the thickness of the scintillator plates is achieved by molding technology. The required tolerance of (+0 -0.05) mm on the 2 mm thick lead converter plates is within standards of industry. The tight tolerance of 0.02 mm on the relative position of the holes in the lead plates has been achieved previously using the punching technology. The clearance for fiber insertion in the holes is 0.1 mm.
In order to avoid aluminizing the fiber ends, the fibers are forming a loop at the front of the module. This loop could affect the uniformity of response, which needs to be monitored at production. The fibers are bent by pulling them over a bar with appropriate diameter and by heating them with dry air. The quality control procedure foresees measurements of the light attenuation in the loops on a certain fraction of the fibers. A systematic quality control should be performed on the pre-production to establish the failure rate and to determine the production procedure. From module 0 production it is shown that the rms of the light attenuation variation in the loops is 2.5 % at most, and that the fiber-to-fiber variation is about 3 to 4 % depending on the fiber type under consideration. From Monte Carlo calculations of the light collection in the scintillator tiles, a 5 % rms in fiber response leads to 0.7 % rms in the light collection for minimum ionizing particles traveling close to the fiber axes. For inclined tracks at 300 mrad, this is reduced to 0.24 %.
From the description in the TDR it is not clear how the shashlik module is made light tight. The lead-scintillator stack is wrapped in black paper before steel straps are welded to the four sides of the module. Since the straps are not as wide as the module, the paper around the corners of the module is not specially protected. From experience of HERA-B only very few modules had their paper cover damaged at installation.
In contrast to HERA-B a preshower detector is placed in front of the LHCb ECAL, and the overall performance of the system could be affected by dead material in between the preshower and the first scintillator of the shashlik stack. Furthermore it was questioned, whether the distance between the preshower and the first scintillator could influence the energy resolution. The effect of dead material in front of ECAL was studied with Monte Carlo. If no correction is applied, then the energy loss in a 5 mm thick iron plate in front of the shashlik stack increases the relative error on the energy measured in ECAL by less than 1 % for electrons of more than 10 GeV. However, if this material is taken into account when calibrating the overall system, no sizeable degradation of the energy resolution is observed. This is confirmed by test-beam data with 20 GeV electrons where no degradation of the ECAL resolution was observed due to the 12 mm lead absorber of the preshower detector. As far as the distance between preshower and ECAL is concerned, all detector positions as defined in the technical design were taken into account in test-beam measurements and Monte Carlo simulations. However no detailed studies of the energy resolution as function of this distance have yet been performed. Even though the effect is expected to be negligible or very small, such studies will be undertaken in the near future. Should the distance between preshower and ECAL become an issue, then the space allocated for connecting the calibration fiber in front of the module could be reduced, without affecting the module design.
3.4 Stacking of modules
The stacking of modules to build up the LHCb ECAL wall has big similarity with the HERA-B calorimeter construction. In both cases the pressure on a module that sits at the bottom of the calorimeter is about 2 kg/cm2. The major difference in the design is due to the opening of the LHCb detector into two halves that can slide out perpendicular to the beam pipe. In order to minimize the amount of material in between the detector halves, a steel strip that is fixed and stretched to the side frame surrounds every two rows of modules. It was agreed that the thickness of these strips should be minimized to preserve the transverse uniformity of the energy calibration. This optimization will be part of the engineering design of the support structures.
3.5 Safety aspects
For the modules closest to the beam pipe the radiation dose is expected to be about 250 krad per year. For maintenance reasons the induced activity of a module after a typical run should be estimated. At HERA-B 250 krad per year have been measured for the modules located in the middle region of the detector and no induced activity has been observed. In order to define final safety measures, a dedicated Monte Carlo calculation for the exact geometry and material composition of the LHCb modules will be performed.
3.6 Transport and handling
The way to transport and handle the 27 kg heavy modules was questioned. Previous experience has shown that no special tooling is required to handle the modules that can be carried by a single person during installation.
Appendix: Questions by the referee
There is no indication of clearances in the drawing that I have. Is there sufficient clearance on both sides of the calorimeter? For example, is there a detailed study of the front of ECAL, with the calibration fibers, their protection, etc? Similarly about the back (see also my remarks on cooling).
Did you study the effect of the distance between the Preshower and the ECAL?
I see no indication of tolerances in the documents that I have. What is the tolerance on the lead thickness, on the scintillator thickness? Is it readily obtained? I do not see any "geometrical" tests mentioned in the note LHCb 2000/044, similarly for the transverse dimensions. What is the clearance to insert the fibers?
2) Fibers tests:
Do you test the fibers /measure their response uniformity after having bent them (for the outer and middle section) and before insertion in the towers? How? Same question for the mirrored fibers (inner section).
3) Tower finishing:
I have not understood the finished state of the tower. Is there any pocket or tape above the welded stainless steel strips? How are the corners protected? How is it made light tight?
4) Was there also a Preshower in HERA-B? In the proposal, I have not seen any experimental results of a combined Preshower/shashlik test. In RD36 we have found effects depending on the way the shashlik tower front was engineered.
Was this stainless steel band of 250 microns also used in HERA-B?
(More generally, I have some questions on the ECAL stacking, but this may be simply a proven technique with HERA-B?)
I do not particularly like the air-cooling for the PM's, mostly necessary in the central part. Air-cooling is made even more difficult in the presence of the cables. Has there been any experimental work to investigate its efficiency? Is there a risk to need water-cooling, which would require space and modifications of the module in the back?
Which lead are you using? Has there been any estimate of the induced activity of a module after a run (for maintenance)?
I read in a note that the 30Kg module can be handled "manually". I doubt that this is possible. Is there any tooling foreseen to handle the module? Can this influence the final design?