Choose timezone
Your profile timezone:
Review and analysis of the aborted collaring of CC11
→
Europe/Zurich
Minutes of meeting
Participants:
A. Devred, A. Milanese, D. Pulikowski, D. Tommasini, J.L. Rudeiros Fernandez, F. Savary, G. Spigo
Distribution:
Participants, L. Bottura
Foreword:
Arnaud introduces the meeting. An incident occurred during the first stages of the collaring process of the collared coils assembly CC11. It is planned to use this collared coils assembly CC11 as 2nd aperture of the magnet S5. The magnet S5 being critical for the project, Luca and Arnaud expressed the need of assessing the risk of having damaged the coils during the incident.
Contributions
Background and introduction [F. Savary]
Frédéric describes briefly the incident, and provides information as to the reference documentation (NCR, MTF).
The collaring attempt took place soon after the restart of the activities on the CERN site, after the lockdown period. All the personnel needed for this activity was present both on the CERN side, and on the GE side.
Nothing abnormal was observed as to the components, and the shimming plan (and also later as to the shims, once fabricated and ready for assembly with the coils). All had been checked, as usual.
Collaring process and monitoring - Case of CC11 compared to previous CCs [J.L. Rudeiros Fernandez]
José Luis describes the collaring process, the parameters that are monitored, and what triggered the suspicion of unusual behavior, possibly alarming, during the alignment steps (tooling closed on shims) prior to collaring (tooling closed without shims between the top and bottom parts).
Giancarlo recalls issues which were encountered at the beginning of the series production of the LHC main dipoles at Alstom Belfort. The total thickness of ground insulation located between the cold bore tube and the coil ID nearby the collar nose was excessive. Also, the collaring shims of the inner layer were slipping during collaring towards the cold bore tube, in such a way the c.b.t. could be squeezed and flattened locally.
In the case of the 11T dipole, the cold bore tube centring supports are made of polymer material and therefore are flexible enough in order not to push excessively on the cold bore tube.
It is visible that the case of CC11 is special, when compared to the previous assemblies, as the tooling is open by additional 70 microns at the pressure of 45 bar. However, this is a small amount, as it means a difference of 35 microns per coil. Note that for slides 8 and 9, a box plot representation would show most of the points inside the box, as there is small variation of the gap along the tooling.
The tooling and the process followed for CC11 were the same as applied for the previous assemblies. Only the number of sections where the gap measurements are done was increased after the first 4 collaring in the beginning of the production.
The pressure applied to get the tooling opened by 0.2 mm, 45 bar, corresponds to about 50% of the usual collaring pressure.
Observations at dismounting [F. Savary]
It is clarified that, at the location of the damages observed on the GI layer 4 (wrinkles/folds), it has been possible to pass the nominal go-nogo gauge of 49.70 mm.
It is asked to check the whether the longitudinal marks observed along the bottom edge of the shims correspond to any of the other features observed like the local reduction of the c.b.t. ID, or the wrinkles/folds area. It is clarified that the marks are present on almost all the length of the shims.
The titanium poles are 240-mm long segments, and the collar packs are 65-mm long (not 100 mm as said during the discussion). The points of initiation of the wrinkles/folds, actually the locations where the ground insulation layers 4 and 5 are swapped, seem to be corresponding to the length of a collar pack (if it would be 100 mm). The cut in the GI layer (located on the layer-jump side) is close to the wrinkles/folds in the longitudinal direction. It is likely a consequence of the wrinkles/folds (even if these wrinkles/folds are located on the side opposite to the layer-jump). In the area of the wrinkles/folds there is extra thickness, which is provoking over pressure, pulling and tearing off the GI at its vicinity.
The cut in the GI layer is not on the same leg of the coil as for the wrinkles/folds. This was checked after the meeting.
It is asked whether one can measure the titanium pole at the location of the defects observed (metrology), and yes, it is possible.
According to Davide, there is no clear indication that the coil has been damaged during the incident. Giancarlo agrees on this.
Davide did a life demonstration in order to show that the suggested mechanism for the formation of wrinkles/folds is very reasonable over a length of about 20 cm (it would not be the case on a much shorter segment).
The presence of a burr along the bottom edge of the loading plate is presented as the most likely cause of the incident. Because of the burr, the space given to the titanium pole is smaller than expected, and the pole cannot move to its position at the same force.
It is suggested to verify the presence (or absence) of marks on the layers that cover GI layer 2, which is marked by the burr at the bottom of the pole.
Arnaud suggests to put the burr in the FE model. Davide suggests to lift the pole by 0.2 mm. The discussion is suspended to continue with the next topic, the metrology of the coils.
Coil metrology [D. Pulikowski]
The two coils, 24 and 25, were remeasured after the aborted collaring. For each coil, 2 standard measurements were made. It appears that the coils are slightly smaller than in their virgin state, which is expected. José Luis confirms that from the metrology data and the E-modulus measurements on ten-stacks, we can determine the average stress seen by the coils during the aborted collaring (see next contribution).
The maximum deformations observed are given in the presentation, also their longitudinal position along the coil.
Arnaud says that he does not see anything alarming as to the metrology data shown.
Estimation of stresses in coils [J.L. Rudeiros Fernandez]
The FE modeling has limitations inherent to the convergence process, which can imply penetration of contact surfaces of the order of 30 to 40 µm, which is the range of displacements that we are discussing, thus resulting in inaccuracies. In additions, it is difficult to compute peak stresses with FE modelling. Experimental methods would likely be more effective.
It is clarified that, on slide 12, the radial pressure should be divided by 2.
A reasonable estimate of the stress seen by the coil during the incident can be determined from the following:
- At 45 bar, the gap measured between the top and bottom parts of the tooling is about 0.2 mm. Because at this stage we have the 0.2 mm thick tooling shim, this means a total gap of 0.4 mm for the tooling without shim. Hence, a gap of 0.2 mm per coil quadrant;
- Data from Fuji paper available from the practice collaring carried out prior to the series production (coils CR1 and CR2) for excess values of 0.1 mm (which would correspond to 0.3 mm – 0.2 mm, as in the present case) give an estimate of the average peak stress in the mid-plane;
- Metrology data of the coils before/after the incident (which are showing a global reduction of the coil size in both radial and azimuthal directions), and E-modulus measurements carried out on coil segments (CR03) giving an estimate of the plastic deformation induced by a compressive load like during collaring.
The above tells us that the maximum average stress seen by the coil during the incident is in the range of 40-60 MPa, well below the allowable limit of 150 MPa imposed at collaring. It remains difficult to give an estimate of the peak stress seen by the coil.
One should note that a relatively low pressure can imply large permanent deformation of the coil in the range of what has been observed (meaning "no surprise" on that side).
Executive summary
Three types of defects were observed during / after the incident:
- Cold bore tube squeezed at 3.2 m from the connection side. It has been verified that this deformation was entirely in the elastic domain;
- GI layers 4 of coil 25, and 5 of coil 24, interleaved at circa 2.5 m from the connection side, with wrinkles/folds;
- Burr (from the fibreglass wrap at the ID of the coil) along the bottom edge of the loading plate.
The defect type 1 is not correlated to the defects type 2 and 3. There could be some correlation between the defects type 2 and 3.
The experts present during the meeting did not express concerns that the coil had been possibly overstressed. Nevertheless, it was asked to measure the height of the burr, and to determine a conservative estimate of the local over-stress along the internal fibre of the coil (at the close vicinity of the burr, where the loading pole has pushed on the shims, and therefore on the coil) in order to confirm this opinion. The local stress can then be estimated based on an extra shim thickness corresponding to the height of the burr. This would be a conservative estimate.
The contractor GE is preparing an RCA. It will be attached to the NCR. The NCR will be completed with the additional measurements results (in particular of the burr); it will propose that the disposition be "concession (use-as-is) for assembly in the magnet S". Then, the NCR will be submitted to the GL, DH and PL for final approval. In the meantime, one can proceed with the repair work (including deburring), and reconstruction activities including collaring. However, we shall wait the formal approval of the DH as to the utilisation of CC11 for the construction of the magnet S5.
There are minutes attached to this event.
Show them.