Conflicting Volumes in PISA

Previous: PISA with no volumes   |   Index   |   Next: ??


Turn all detectors OFF in pisa.kumac, and check for volume overlaps (Here is how). There were 16 cases of overlapping volumes, ranging from 18cm to many mm. It is not clear what happens in Geant when overlapping volumes are encountered, but at best the behavior is unpredictable. This snapshot was taken after the first set of fixes.
One set had to do with the 'flowerpots' that are mounted on the back of the CM iron. They consist of a brass cone (MUPP ans MUPQ), with a layer of lead (MUWS, MUWN) on the outside. The one on the South conflicted with the prototype of the new absorber (ABSS). In the process, the problem appeared to be that the flowerpots were defined too big. The brass was 46cm long, not 50, and the lead on the back was 6.6cm thick, not 7.0. The lead wrap was 1/2" thick, not 1cm. Angles were adjusted. No more overlaps with ABSS (new absorber prototype). Also fixed the dimensions on the North flowerpot.

[Note I took numbers from the engineering drawings (as-built), and that the older envelope drawings are wrong.]

Files affected: phnx.par, magnet.f
The next problem set had to do with the CM disks (AMGU, AMGD) and the vertical pieces (CRRU, CRLU, CRRD, CRLD) that connect it to the horizontal return yokes. The problem volumes where the overlaps occur are shown in red.
I replaced the blocks with a composite consisting of a block and three trapezoids (CRR1,CRR2,CRR3).

Files affected: magnet.f, phnx.par
Next, the beampipe. The picture has the vertical scale expanded greatly. The central piece in green-blue is the BE beampipe filled with vacuum (PIPE, PVAC), outside of that in red-blue a same-diameter stainless steel pipe (PIP2, PVA2), and outside of that a blue-blue stainless steel pipe with a larger diameter (PIP3, PVA3). You can see the small ss pipe sticks into (=overlaps with) the big ss pipe by more than a meter. (Same problem with PIPN, also fixed)

There are still questons about where the real boundaries are. For now, I put the Be-SS weld at +-75cm, and the small-big boundary at 261cm.

Sources: AN107 fig.4 (1998), drawing RD002-0500-0010A (1995)

Files affected: phnx.par, beam_pipe.f
The mother volumes (MUA1, MUA2) holding the Muon N,S lampshades (MEGA, AGEM) were a bit too small, and a bus bar (BUSN) ran into the N lampshade (MEGA). I enlarged MUA1,2 and slid the bus bar to smaller radius.

Files affected: phnx.par, magnet.f


After all this, there are no more overlaps or extrusions greated than a micron in Phenix with all detectors turned off.

Next I turned on each subsystem in turn, and noted the number of overlaps greater than 1mm. Note that these are all likely real errors, since they are too large for errors caused by roundoffs in calculations, which are typically at the 10-7 level.

Subsystem Overlaps
2010
Overlaps
2012
Mean Overlap (cm) Comments
ZDC 16 17 1.10 Zero-degree calorimeter
SVX 0 13 - This is the Silicon Vertex Tracker
FCL 5 7 1.51 This is the FCL (forward calorimeter)
BBC 0 0 - BBC on with track stack used
ITR 7880 7880 4.16 Latest version of Dch and PC1
CRK 100 80 1.30 RICH with CO2 radiator gas
PAD 52 149 0.99 Latest version of PC2/PC3
AER 0 0 0 - This is the AER (aerogel counter)
TFW 0 ? - TOF-West
TRD 0 0 - This is the TEC
TOF 4 18 15.9 Time of Flight
EMC 34 763 2.46 EMCal, H.I. with Cerenkov
HBD 186 951 0.24 Hadron-blind
MUM 129 8 3.32 Muon trackers
MUI 7 8 17.1 Muon identifier
RXN 0   0 - reaction plane detector
MUP 8 ?? 0.10 Pad chamber for Muon Trigger
MPC 0 0 - Muon Piston Calorimeter
RLT 0 - - Relative Luminosity Telescope
All 8421   3.99 All 11 offending subsystems


The raw data are here: stats.txt. In this file you can see the Geant volume names that cause the problems, detector by detector.


Plotting macro: ovplot.C. This macro reads the stats.txt file, and makes plots for each of the detectors. What is shown here is all of those plotted. The dashed line is the sum. If you want to see the induvidual plots, grab the txt file and the macro and run it yourself.





Previous: PISA with no volumes   |   Index   |   Next: ??

Jan 2010 HvH