[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: [Fwd: RE: question] (fwd)
more info from Yves, the plots are in
http://www.jlab.org/~elke/12GeV-Beam/
called pdist-lin.ps and phist-log.ps
cheers elke
( `,_' )+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=
) `\ -
/ '. | +
| `, Elke-Caroline Aschenauer =
\,_ `-/ -
,&&&&&V Jefferson Lab +
,&&&&&&&&: HALL-D 12C / F381 121-A Atlantic Avenue =
,&&&&&&&&&&; Suite 8 Hampton, VA 23664 -
| |&&&&&&&;\ 12000 Jefferson Ave +
| | :_) _ Newport News, VA 23606 Tel.: 001-757-224-1216 =
| | ;--' | Mail: elke@jlab.org Mobil: 001-757-256-5224 -
'--' `-.--. | +
\_ | |---' Tel.: 001-757-269-5352 =
`-._\__/ Fax.: 001-757-269-6331 -
+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+
---------- Forwarded message ----------
Date: Wed, 16 Jul 2008 10:30:03 -0400
From: yves roblin <roblin@jlab.org>
To: 'Elke-Caroline Aschenauer' <elke@jlab.org>
Subject: RE: [Fwd: RE: question]
Dear Elke,
>-----Original Message-----
>From: Elke-Caroline Aschenauer [mailto:elke@jlab.org]
>Sent: Tuesday, July 15, 2008 9:50 PM
>To: Leigh Harwood
>Cc: roblin@jlab.org; halld-tagger@jlab.org
>Subject: Re: [Fwd: RE: question]
>On Tue, 15 Jul 2008, Leigh Harwood wrote:
>Dear Leigh and Yves,
>first of all let me thank you both for keeping Hall D / GlueX so nicely
informed about these new calculations. It is highy appreciated.
>Actually I have a couple of some questions to the new calcualtiosn, which
indeed look very very promissing.
>I must admit I have a really hard time believing that there no non-gaussian
tails.
Actually, if you fit the distributions, you do see a significant excess in
the tails compared to the Gaussian fit. However, they don't exhibit the
spread that we had in the original dimad simulation.
>That is really suprising at least if I use my experience from HERA, yes
Cebaf is not HERA, but there is synchrotron radiation and yes I know Cebaf
has >less synchrotron radiation compard to HERA ((E2-E2)^4), because of the
lower beam-energy, but CEBAF keeps the radius in the arcs the same with
doubling >the beam energy, this should increase the synchrotron radiation
more than just what you would expect from the energy difference. I also know
the number >of turns is less than in HERA, but again no non gaussian tails
at all. There is misalignement of the magnets which widens your emittanceand
their is >rest gas, which actually inreases because of the increased
synchrotron radiation.
>So all this effects populated tails, apart from the misalignement, this
increases the emittance and if it is really bad can increase the synchrotron
>radiation, because a lot of particles go of center through the quad fields.
>So my question is are all this effects simulated.
Yes. All these effects (misalignment, mispowering, missteering, synch. rad,
multipoles in quads and dipoles) were simulated. What we are doing is
particle tracking. No secondaries are emitted. No beam gas was simulated
either. Estimates showed (and we did run early simulations with Geant two
years ago to check it) that at this energy and our usual vacuum specs, there
will be no noticeable effect.
So, in these simulations the halo will indeed come from chromatic effects
exciting the higher order multipoles (going off center in elements that have
multipole kicks because of the dp/p broadening.). We moved from DIMAD to
ELEGANT two years ago for a number of reasons, one of them being that
ELEGANT is well benchmarked against measurements in rings and trusty
synchrotron radiation algorithms. DIMAD also has a synchrotron rad. Feature
but it is somewhat simpler and less tested.
The original DIMAD simulations were done with what was then our baseline
optics and the 10 degrees hallD.
The newer simulations have a radically different optics (Double bend
achromats from Arc6 and up, 8 degrees hallD) which drastically reduces the
emittance growth. Chromatic effects are under control by the fact that we
have, in the design, compensation coils to adjust the dipoles in the upper
arcs in order to compensate for the average energy loss (In ArCA, you lose
about 1/2 MeV per dipole, for a total of 16MeV at the end of that line)
effectively keeping the orbit around the center of the magnets and reducing
our exposure to multipoles. Another thing we have in the design is a
better orbit control (we plan to be at about 600 microns sigma for the
centroid). That was achieved with more careful coverage of the bpm and
correctors in the machine. We calculated the optimal placement for these and
added new bpms/correctors where needed. Finally, multipoles in the
simulations are included for dipoles and quads, random and systematics.
Quads have skew terms too. These values were derived from part measurements
of the existing magnets and/or estimates of what we expect in the other
dipoles (estimates were done numerically via magnet modeling or tracking and
analytical approaches. They were confronted with realistic construction
errors and specs were given to the magnet designers for the construction of
the new magnets.). In short, a lot of studies went into these whereas the
first go around (with DIMAD) was not as thorough.
Furthermore, due to the nature of the optics, some of the effects, such as
the sextupole terms are set to largely cancel in the arcs because of the
symmetries in the placement of the dipoles and quads.
I don't doubt that the beam will have some tails in real life, but those are
things that will not come from the design itself, but rather things that we
are already dealing with today (for example dirty beam at injector scraping
somewhere and generating secondaries, etc..). What these tracking
simulations do is validate the design to make sure it does not induce halo
by itself because of ill-chosen optics.
>Please don't misunderstand my mail I don't want to be negative on the
results, but woud like to understand a bit better how the simualtions are
done. I >think it would be also nice to have the energy spectrum going
together with this new spatial distributions of the beam.
Attached are two plots of the p distribution at the radiator (dp/p actually.
Deviation from the nominal p0 energy dp/p=(p-p0)/p0). Second plot is in
logscale. In green are Gaussian fits. You see an excess of energy to the
left (more negatives dp/p) which is what I would expect because of the synch
rad.
Both DIMAD and ELEGANT give the proper synch rad broadening in terms of dp/p
and they also agree with analytical calculations. However, when you start
looking 6 or more order of magnitude down (which is where those tails are)
there are differences. If you consider the process of synch rad itself, the
way its simulated is :
1) compute probability of emission of photons, compute how many of them
2) compute energy of photon emitted using energy prob. Distribution.
3) update electron to do the kinematic balance.
DIMAD and ELEGANT have different algos. So far, I have found that it would
appear DIMAD emits harder photons than ELEGANT, meaning it is not using the
same energy distribution.
The real one in log scale goes is almost flat, climbing slightly to the
critical energy Ec and then drops quickly. Some simulation codes use a
constant rectangle (keeping the area the same) to do that. If there is an
overestimate of the hard photons, you end up having some electrons kicked
further out in the quads/dipoles and experiencing harder multipoles,
generating halo. The truth is probably somewhere in between what DIMAD and
ELEGANT gives us.
Yves
>Thanks and have a nice evening
>elke
> Date: Tue, 15 Jul 2008 21:04:51 -0400
> From: Leigh Harwood <harwood@jlab.org>
> To: elke Aschenauer <elke@jlab.org>
> Subject: [Fwd: RE: question]
>
> Elke,
>
> I asked Yves to calculate the number of particles outside the diamond
> (+/- 5mm) and outside the inned edge of the frame (said to be +/-
> 7.5mm) The results are below.
>
> I'm not sure I believe them!
>
> L
>
> -------- Original Message --------
> Subject: RE: question
> Date: Tue, 15 Jul 2008 11:31:29 -0400
> From: yves roblin <roblin@jlab.org>
> To: 'Leigh Harwood' <harwood@jlab.org>
> References: <487C2CD2.7080605@jlab.org>
> <000801c8e688$34bdbdd0$9e393970$@org> <487CC0D1.9000806@jlab.org>
>
>
>
> Leigh,
>,
>
>
> Here are the numbers :
>
>
>
>
> DIMAD
> ELEGANT
>
> Past 5mm
> 35150 1504
>
> Past 7.5mm
> 10583 0
>
>
>
>
>
> At first glance, looking at the plot it did not look that it would be
> that much in DIMAD. However,
>
> I checked in the file and there are many bins and they are in the
> dozens or hundreds counts..
>
>
>
> The counts were calculated for each plot independently (not aligned
> together).
>
>
>
> Yves
>
>
>
>
( `,_' )+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=
) `\ -
/ '. | +
| `, Elke-Caroline Aschenauer =
\,_ `-/ -
,&&&&&V Jefferson Lab +
,&&&&&&&&: HALL-D 12C / F381 121-A Atlantic Avenue =
,&&&&&&&&&&; Suite 8 Hampton, VA 23664 -
| |&&&&&&&;\ 12000 Jefferson Ave +
| | :_) _ Newport News, VA 23606 Tel.: 001-757-224-1216 =
| | ;--' | Mail: elke@jlab.org Mobil: 001-757-256-5224 -
'--' `-.--. | +
\_ | |---' Tel.: 001-757-269-5352 =
`-._\__/ Fax.: 001-757-269-6331 -
+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+=-+