-----------------------------------------------------------
Minutes of the ABP-RLC team meeting of 04.11.2005
present: UD, WH, EM, TP, FR, GR, RT
web site: http://ab-abp-rlc.web.cern.ch/ab-abp-rlc/
------------------------------------------------------------
(1) Progress report on coherent beam-beam simulations (TP)
----------------------------------------------------------
TP presented an overview of her present activites
on coherent beam-beam effects (see slides attached).
She explained that the main objective is to extend
the present understanding and models to include a
large number of bunches and arbitrary collision
schemes. It is expected that this will change the
behaviour of coherent beam-beam modes very strongly.
Applications to beam diagnostics such as tune measurement
procedures and feedback systems are foreseen.
To address these questions she presented three different
models she has developped: a semi-analytical model based
on the analysis of the full one turn map, a simulation
program using rigid bunches and a fully self-consistent
multi-particle simulation.
The latter is presently using a soft Gaussian
approach for the field calculation.
TP presented the advantages and disadvantages of
the three models and their relative merits in her
present studies. She proceeded by showing several
collision scenarios and the consistency of the models
and how she uses them to explain the observations.
In particular, parameter variations from bunch to bunch
and the consequences are investigated since all models
are optimized to allow such variations.
An unresolved question is a Yokoya factor she has
obtained with the multi particle program which is larger
than expected from previous simulations using the
same approximation and which is closer to the theoretical
value.
A future extension could resolve this puzzle when
the present approximation for the field calculation is
replaced by the exact Hybrid Fast Multipole Method (HFMM).
The disadvantage of this method is the largely increased
computing time needed and it is foreseen to improve this by
a parallelization of the code. Since most time is spent in
the field calculations of simultaneous beam-beam interactions
which can all be performed independently,
this parallel mode can make it possible to study multi-bunch
effects presently out of range.
A proposal is submitted to the EPFL (Lausanne) to use
the BlueGene machine where more than 8000 processors are
available.
TP presented first ideas on a possible implementation.
(2) Operational scenarios for LHCb spectrometer magnet (WH)
-----------------------------------------------------------
WH addressed two questions which were recently
brought up concerning the operation of the LHCb
spectrometer magnet and smaller beta* in IP8
for lower bunch intensities (see slides attached).
He first explained the constraints imposed by
the crossing schemes and in particular on the
sign of the crossing angle. The latter is
imposed by the need to avoid additional crossings
and is fixed (negative for beam 1) due to the
change from the outside to the inside aperture
in IP8.
The LHCb spectrometer magnet together with its
compensator magnets introduces a horizontal
crossing angle of +- 135 murad at top energy.
The crossing angle needed for the separation at
the beam-beam encounters is added in the same plane
and must always be negative for beam 1.
Changing the polarity of the LHCb spectrometer
implies a different sign of the crossing angle
produced by the spectrometer and therefore
requires different external crossing angles to
ensure the correct sign of the net crossing
angle. This is foreseen and implemented in the
nominal optics Version 6.5 for beta* = 10 m.
To keep the luminosity above the required
10^32 cm^-2 s^-1 when the bunch intensity
is smaller than nominal, it is requested to
operated with reduced beta* in IP8.
Running at smaller beta* imply different
contributions of long range beam-beam effects and
stricter requirements on the available aperture.
WH showed the necessary crossing angles to
ensure the nominal beam separation. For the
case of beta* = 2 m this can be achieved for both
spectrometer polarities with a maximum crossing
angle of 210 murad.
In the case of beta* = 1 m the necessary crossing
angle of 255 murad for one of the polarities is not
accepted for reasons of limited aperture.
A reduced crossing angle of about 200 murad or below
provides a significantly smaller separation at the
long range beam-beam encounters (about 5 sigma).
Although it is only foreseen for low bunch intensities,
we do not consider this a robust running scenario.
A further request concerned the ramping strategy of
the spectrometer magnet and its operation at full
field already at injection energy was discussed.
The crossing angle in such a case would be as large
as +-2.1 mrad. While for one of the two polarities
this can be accepted, it is excluded for the other
polarity since it would require an external crossing
angle larger that -+2.2 mrad.
We should therefore like to draw the following
conclusions:
1. The LHCb spectrometer magnet cannot be operated
at full field for one of its polarities.
2. We strongly suggest to drop the case beta* = 1m
as an option to run with lower bunch intensities.
(3) Update on the impedance of the "sausage" (EM)
-------------------------------------------------
Last RLC Section meeting, EM presented first results on the calculation
via HFSS simulation of the impedance of the "sausage"
(round->round->elliptical connector).
EM had shown a table with the resonance frequencies of different modes
and the respective quality factors (see minutes of the last meeting).
This time EM has given an estimation of the R_s factors (~10 Ohm) and
power loss per mode for the first 10 modes. The power loss can be
calculated pessimistically, if the frequency of the mode is a multiple
of the bunch frequency, and values show peaks of about 14 W in the range
4-7 of the mode numbers. The power loss evaluated with HFSS considering
the real frequencies of modes in the structure only show peaks up to
3.5 mW.
FR says that there is a proposal for an experiment close to ATLAS, or
to CMS or to both, for which a 8m long movable wall should go very
close to the beam. A statement is awaited from our group saying whether
that could potentially be a killer. A post-doc from Daresbury will soon
start to work on this issue.
FR will not be able to attend the TOTEM review committee and therefore
has given the name of EM to replace him and give recommendations from
the impedance point of view.
Concerning the progress on HFSS simulations for the impedance of the
LHC collimators, EM has shown plots in which there is a very good
agreement between the analytical formula for the longitudinal impedance
(Z_||/L) and the value computed with HFSS (FR argues that, as both
results of analytical estimation and of simulation lie in the range
01-10 Ohm/m, it would be better to display the results on a scale which
in not logarithmic and spans over more than 6 orders of magnitude, EM
replies he has only used the same scale used by Tsutsui in his paper).
There is already indication that some minor discrepancies should be
corrected by re-launching the simulations with a finer mesh and higher
precision.
On the discussion of the previous meeting about what impedance one
effectively measures when using one or two wires, EM shows results
from Tsutsui (CERN-SL-Note-2002-034-AP). Tsutsui's formula clearly
shows that the measured quantity reduces to the longitudinal impedance
if the measurement is carried out with one single centered wire (even
correctly scaled by the Yokoya factor accounting for the non-roundness
of the chamber). With one displaced wire, one sees zero in the
horizontal plane and the vertical impedance (some of its dipolar and
quadrupolar components). With two wires, one sees the sum of horizontal
and vertical dipolar impedances. Measuring with wires displaced at
different distances, one should observe a parabolic increase of the
"transverse" contribution to the measurement.
This explains why there are different results when measuring with one
or two wires, but it still does not explain why from measurements there
appears to be a negative real part of the horizontal impedance. This
might be due to the different materials in the two planes (good
conductor in the horizontal plane and ferrite in the vertical plane),
or to the finite length, or...
GR says that preliminary results from HEADTAIL simulations on the TMCI
thresholds in the SPS show a clear increase of the threshold with the
injection energy. Going from 26 to 40 GeV, the TMCI threshold jumps
from ~0.6 10^11 p/bunch to 1.5. At 60 GeV injection energy the
threshold reaches 1.9. These thresholds have been evaluated without
taking space charge into account. Space charge can raise the threshold
at 26 GeV up to about 0.8, and surprisingly enough it still seems to
affect the threshold even at 40 and 60 GeV. Simulations were done
keeping the same longitudinal emittance in all cases and re-matching
the bunch to bucket each time. Also, the transverse normalized
emittance was kept constant, which translates in smaller beam sizes for
higher injection energies.
EM observed after the meeting that the increase of the TMCI threshold
is perfectly consistent with his formula, which has a proportional to
eta dependence.
Posted on the web: slides by TP-WH, EM-AG (sausage), and EM (Tsutsui).
Web site: http://ab-abp-rlc.web.cern.ch/ab-abp-rlc/