Recent Upgrade of g4lbnf: Horn B & C, Conceptual Design revision - - PowerPoint PPT Presentation

January 10 2017 Realistic Horns in G4LBNF 1

Recent Upgrade of g4lbnf: Horn B & C, Conceptual Design revision 2

Paul Lebrun ND/Fermilab.

January 10 2017 Realistic Horns in G4LBNF 2

This a corrected version of the talk written a week ago, presented on Friday January 6

January 10 2017 Realistic Horns in G4LBNF 3

Outline

✔

Horn B & C, revision 2 => Coded up. ==> geantino plots

✔

(Horn A, small improvements to the code. )

✔

Comparison, Horn “ideal” (as coming out of the genetic algorithm based optimization

• f Laura F. et al) vs “Realistic”, detailed G4 implementation of all “relevant” volumes

found in the NX-9 CAD model, and corresponding changes to the magnetic field map.

✔

Next steps:

–

Retune a bit (spacing between horns, horn current..)

–

small changes in horn length??? Is this worth trying?

–

Study the effect of field maps upgrades in the Inner/Outer conductor regions, and other non-unformities

–

Realistic Beam window, Rev X. ?

January 10 2017 Realistic Horns in G4LBNF 4

… Few programming notes..

✔

Now using Geant4 v 4.10.2.p02 .. The change I made to the code caused crashed using the previous release, due to bug in the G4 geometry navigation code, fixed months ago...

✔

So, I am using ROOT 6.x, and a newer compiler, as I haven't found a matching DK2Nu with the old old ROOT and this new compiler.

– This might break the old style nutrino N-tuples code. My

advice: use the Dk2NU package from now on.

✔

I haven't not yet committed the changes.. Bigger change then just G4. Laura F. has upgraded our neutrino analysis script(s), but... is it in use??

January 10 2017 Realistic Horns in G4LBNF 5

… Coding the realistic geometry..

✔

No changes in the methodology.. Fired up the Citrix server, TeamCenter, get to the model #, run Nx9, and do the measurements that are needed.. TeamCenter Doc# entered in the code.

✔

Enter the geometry, mostly “hardcoded” in the LBNEVolumePlacements class. One by one, trying to check every number twice...

– Curves ==> Polygonal shape.

✔

Conceptual Design Rev1 no longer avialable at run time: if needed, you'll need to re0build g4lbne from a previous release (v3r4p4 ?.. )

✔

(a good thing a few of us will stay around for the next ~30 years).

January 10 2017 Realistic Horns in G4LBNF 6

Nx9 screen captures... (Horn B)

I/O trans. Outer Conductor Inner Conductor Current Eq section 1 m. I/O Insulator

January 10 2017 Realistic Horns in G4LBNF 7

..found wire frame a bit more convenient..

Nx9 has good interactive feature to explore the model...Not all features are implemented in the G4 model: simplified some curve, beveled edges for flanges. Also, drain holes not implemented Hanger rings Welds

January 10 2017 Realistic Horns in G4LBNF 8

..comparison with Geantinos plots..

Nx9 cut view G4lbnf geantinos R/Z plot Spider hanger added.. (NuMI)

January 10 2017 Realistic Horns in G4LBNF 9

For sake of completion, HornA.

Nx9 cut view G4lbnf geantinos R/Z plot Spider hanger added.. (NuMI) “tapered geometry”

January 10 2017 Realistic Horns in G4LBNF 10

Note: The upstream Inner/Outer transition, and associated flanges, are crudely represented in G4blnf, simply because they upstream of the starting point of the target. The 4 titanium tubes that support the downstream end of the target, and provide Helium gas for cooling are not shown in the previous slide, but are implemented in the both the NX9 and G4lbnf models.

January 10 2017 Realistic Horns in G4LBNF 11

Horn C.

G4lbnf geantinos R/Z plot Spider hanger added.. (NuMI) Current equalizer sections, upstream

January 10 2017 Realistic Horns in G4LBNF 12

Neutrino (in neutrino mode) spectrum, far detector.

Idealistic vs Realistic

Simply replace the Idealistic (as taken of the “run26_8079_NumiTarget” from Laura). All horns are either “ideal” shape (polygonal, from the optimizer) or Realistic (Nx9 models) . All realistic horns are revision 2.

January 10 2017 Realistic Horns in G4LBNF 13

The target is always “engineered”, see Cory Crowley's DUNE DocDb note 1923-v2. All Horns are at Rev 2 (Team Center document #F10068454, #F10071359, #F10071767) respectively. These Ideal horns were obtained via the Laura's optimization procedure, based on a genetic algorithm. However, the Horn A IC downstream radius that was found to be too small to accommodate for the NuMI style target. We had to increasse it from 21 mm to 33 mm. This new “ideal” horn A is the basis for the results shown in this talk. (Early version presented at the Beam Simulation meeting on Friday Dec. 16 2016 had that radius at 21 mm). As shown later, this is a critical design parameter for Horn A, for the neutrino spectrum at 4 to 5 GeV. We now compute the ratios, Realistic (a.k.a. engineered) over Idealistic.

DUNE-doc-1923-v2

January 10 2017 Realistic Horns in G4LBNF 14

Neutrino (in neutrino mode) spectrum, far detector, ratio

Realistic (a.k.a engineered) / Idealistic

Big effect in the 4 to 6 GeV region, as anticipated by Laura.! We have a bit more focusing power in HornA, as seen later. As pointed by Mary B., not good news for systematics, if we expect new physics in this energy range.

Loss due realistic sim”: absorption Gain due to increase

• f focusing power

In realistic sim.

January 10 2017 Realistic Horns in G4LBNF 15

Details needed! One Horn is “Ideal”, two “Real”

Here, the black square curve is the same as the previous slide. We then study the case (colored curve) where we compute the ratio of flux, between the partial realistic scenario, where only one horn is idealized. The denominator is always “All Horn ideal”. Horn A shape and field integral seems to matter most for E ~ 3.5 GeV/c. The trailing edge of the focusing peak (5 GeV) is gone when Ideal Horn A is used.

January 10 2017 Realistic Horns in G4LBNF 16

An other way to compare.. : One horns “realistic”, two “idealized”

Indeed, the focusing difference at 3.75 and 5. GeV is due to HornA shape difference, most likely at small radius, downstream section of the IC & IO transition. Is this worth addressing ? Overall, as we added material with the realistic version, we loose neutrinos at low energy, in the focusing peak, and gain quite a a tiny bit at high energy (5 Gev).

January 10 2017 Realistic Horns in G4LBNF 17

Impact on the Near to Far ratio.

This is the ratio of ratios previously shown, Near over Far. As expected, Horn A shape change induced a signficant effect at 2.5 GeV. And, as shown in previous slides, a bigger “oscillation” at 4.5 GeV/c

January 10 2017 Realistic Horns in G4LBNF 18

Probable Root Cause: Focusing changes: Effective magnetic length of horns.. Ideal vs realistic..

In the ideal version, the I/O transitions region are much simplified, they are nearly vertical

• lines. We (Cory C. ? ) could compromise and reduce a bit the length of the IC, to match

the same B.dl that enter (or leaving) the I/O transition.. Or, more simply, adjust to the distance between Horn A and Horn B. Then study the sensitivity of the Near/Far versus these parameters.

Horn B Horn A Geantino plot.

January 10 2017 Realistic Horns in G4LBNF 19

Here, went back the the simplified cylindrical target that fits into the strongly tapered Horn

• A. We vary the radius of this tapered section, from 21 mm to 33 mm. (The target had a

radius of 14 mm, which is very thicvk, but it should not affect our conclusion) The gain at 5 GeV can indeed be very strong..

January 10 2017 Realistic Horns in G4LBNF 20

Here, we vary the distance between Horn B and Horn A. The “nominal” distance (via G4UI data card) is set to 2900., where the start of horn A defined as the position of the start of the IC. The size of the induced “oscillation” in the spectrum is about 1 % per 5 cm. Since the mechanical/survey tolerance will be at least 10 time better, we are O.K.

January 10 2017 Realistic Horns in G4LBNF 21

Here, we vary the longitudinal distance between the target and Horn A. The denominator corresponds always to the “nominal” position of the target, whose most upstream fins is aligned with respect to the field region of the horn. Positive Dz corresponds to moving the

• 2m. Long target into the horn. If so, high energy pions, produced at relatively small angles

are no loger focused, and one loose yield for neutrino energy of 5 GeV.

January 10 2017 Realistic Horns in G4LBNF 22

Effect of downstream target support..

<~ 1 % change across the spectrum! Perhaps sensible: In the focusing peak, pions at a bigger radii than the bulk of the support, which at a Radius of ~ 40 mm. Not affected.. At higher energy production of pions that can be focused by HornC do compensate the absorption. And, of course, no magnetic effects.

January 10 2017 Realistic Horns in G4LBNF 23

While we hope that the above 3-horn configuration will be the selected one, we still have to support the CDR Baseline, i.e., the good old NuMI horn and target, adapted for 1.2 MW

• pration.

Andm since we haven't run this configuration in many months, it felt into disrepair... Note: we can't simply take the old CDR flux plot, as the Hadronic interaction modeling changed on us. ==> Re-furbish the broken stuff. Two distinct projects:

• a. Port an old version (v3r2p6) to Geant4.10.2p2. Upgrade it to support the current

capabilities of LBNEAnalysis, based on the current Dk2Nu. Abandoned!

• b. Extract the 1.2 MW NuMI target code, circa ~2014, from the above version, and

re-implemented, optionally, of course in the current version of g4lbnf.

January 10 2017 Realistic Horns in G4LBNF 24

Comparison with CDR Ref 2015.

This is the neutrino flux, neutrino mode, current optimized and engineered design, compared to the NuMI design, a.ka. Reference CDR 2015 . The same version of Geant3, v4.10.2p2 is used, along with the Dk2Nu library for the blue dot and magenta triangle The continuous green curve is taken from the CDR Reference Web site.

January 10 2017 Realistic Horns in G4LBNF 25

The difference between CDR 2015 reference design and the current implementation of the very same design were anticipated: (I) different version of Geant4 (v4.9.6 vs v4.10.2.p02), so possibly difference in hadronic interaction (ii) small changes to the implementation of geometry of NuMI horn1, such that different target could be inserted. (iii) The Ref CDR 2015 is for the inclusive νµ flux, for the other three curves, it is only the νµ flux from π+ decay. The v3r2p6 version presented a month ago was found to be antecedent to CDR 2015... By possibly lots of changes. We do not recommand to use this version, ever... Such older version are to be deprecated.

January 10 2017 Realistic Horns in G4LBNF 26

Next Steps...

✔

We do need to retune a bit, if the shape of the focusing tail matters? By this, I mean moving Horn B with respect to Horn A, (easy), or reducing the length of the IC by a few percent? (need to redo the FEA analysis.... Probably not !... )

✔

Study the field maps, and corrections due to horn deformations, I/O transition regions, etc.. (Colorado, UTA & IIT )

✔

Back to v3r2p6 problem: Lesson Learned: If we bother to “git tag”, perhaps, we should also git commit a bunch of reference plots corresponding to this tag in …/g4lbne/BeamStudies.. Tedious, I know... Only neccesary for major releases...

January 10 2017 Realistic Horns in G4LBNF 27

So, what should be my 1rst priority?

✔

Who knows..

✔

Meanwhile, I did some work on the “spectrometer”.

– Should we opt (in a month or two, w'll know) for the ex-situ option, how

to cross-check or tie-in, the results from the replica to the real chase?

–

Proposal (Alberto M., mostly..) :

• Install a small, relatively thin “ionization detector” at the back of

the HornC, where, in the replica, the full spectrometer will be installed.

• The very same detector would be installed in the Chase.
• Concept: the flux of ionizing particles is good tracer of the

charge pi/kaon/muon flux, in the chase. Only if well calibrated to the real flux of these neutrino progenitors.

January 10 2017 Realistic Horns in G4LBNF 28

So, here is what I did in g4lbnf: Usual Chase content (target, 3 horns, details don't matter) Re-check so-called “tracking planes”, recording every (as the 50 MeV threshold is removed ) ionizing particles that goes through them. ==> ability to determine the ionizing particle rate, per PoT. Imagine now a ultra fined grained detector, basicaly digital, (not calorimetric!), such as Silicon Pixel planes used at LHC (pixel size 50 by 100 microns) . Take an active surface

• f 5 cm2. If we have ~ 5 e5 proton per rf bucket during the ~ 100 microsecond horn pulse,

that is, we do slow extraction with the the minimum workable beam intensity in MI, and extract for ~ 1 second, then The probability to get a pixel “on” is of the order of 1, for radial distance greater 40 cm. With plenty statistics, one valid measurement pe MI ramp.

January 10 2017 Realistic Horns in G4LBNF 29

Profile of the probability for a pixel hit vs X

Assuming the instantaneous beam intensity expressed in the previous slide...Rate will be dominated by soft e.m., from π0 decays radiating away on the horns. But, π0 are good tracers

• f π+