2-4 December 2013
Europe/Berlin timezone

SUMMARY of the 1st XBEAM-XRING Workshop


“Beam Dynamics meets Magnets”



The first XBEAM-XRING workshop took place at the Darmstadtium in Darmstadt, Germany, from December 2th to 4th 2013. The scientific program of the workshop had been set up following suggestions by an International Advisory Committee (IAC) composed of: O. Boine-Frankenheim (TUD/GSI), P. Fabbricatore (INFN) E. Fischer (GSI), H.G. Khodzhibagiyan (JINR), S. Machida (RAL), K. Ohmi (KEK), S. Russenschuck (CERN), F. Schmidt (CERN), C. Spencer (SLAC), F. Zimmermann (CERN). The workshop was sponsored/supported by EUCARD2, GSI, FAIR, HIC4FAIR, and TUD.  The program, the presentations, videos of the sessions and session summaries are available through the indico page:



There were 84 registrants distributed over the following countries, laboratories and companies:


Germany:  GSI 21, Goethe University 7, TUD 6, KIT 2, FZ Juelich 2, FAIR 1

Switzerland:  CERN 22, PSI 3

UK: Cockcroft 1, ASTeC 1, SFTC 6, RAL 3, ISIS 3, and Imperial College 1

JAPAN: KEK 2, J-PARC 1, JAEA 1, University of Japan 1



Austria: MedAustron 1

Belgium: KU Leuven 1

Russia: JINR 1


The workshop lasted for 2 and half days and the program had been developed based on experiences in the projects and institutions of the attendees. The structure of the sub-sessions was as follows:


2nd December

1) Beam dynamics: Nonlinear dynamics, do we care?

         2) Magnets: the Superconducting Ion Synchrotron SIS100 enterprise at GSI
            3) Beam Dynamics & Magnets: the LHC experience and beyond
         4) Discussion + short talks: mainly Magnets


3rd December

1)   Magnets & Beam Dynamics: from 50 years of accelerator magnets to achieving synergy

2)    Magnets: the challenge of the design

3)    Beam Dynamics: The state of affairs at J-PARC, KEKB & small machines

4)    Discussion + short talks: mainly Beam Dynamics

4th December

1)   Beam Dynamics & Magnets: advanced concepts and developments

2)    Joint Discussion and Workshop Summary



The chairs of the discussion periods were: on the 2nd December, Egbert Fischer; the 3rd, Wolfram Fischer; the 4th, Peter Spiller and Stephan Russenschuck


BELOW IS THE LIST OF SOME SUGGESTED TOPICS FOR THE DISCUSSION PERIODS: PROPOSED BY THE IAC (not all these topics were actually covered in the discussion periods)


The process of setting the magnet requirements


(1) When choosing the maximum integrated strength to require, can the beam physicist trust the magnet engineer to provide that strength? Or does the beam physicist ask for 5% more strength than they really need, but the magnet engineer does not know that and adds another 5% to the current the computer model of the magnet predicts. Then the power supply engineer does not trust the magnet engineer and adds another 5% to the power supply current specs. So one ends up with a much stronger and more expensive magnet and power supply system than one need.


(2) How precisely do beam physicists need to know the higher multipoles in a magnet? Before a magnet is fabricated we can only predict the allowed multipoles in a magnet. Prior experience in fabricating and measuring magnets gives one an idea of the typical range of sizes of all their multipoles. But do the beam physicists modeling the beam dynamics as it passes through the lattice use the same multipole values in all the magnets of the same style? That is not a realistic scenario, is it not better to use different multipole values in different magnets of the same style when simulating a beamline of magnets? Then a “tolerance” study can be carried out.


(3) Can lattice designers remember a physical magnet is longer than its effective length? If they are not sure how much longer to allow for the coils then they should consult with their magnet engineers.


(4) When the beam physicist chooses the magnet aperture can they specify if they’ve left room for a beampipe or not?


Designing Magnets


(1) Before starting to design a magnet the magnet engineer needs to know several other characteristics from the beam physicist: does it run in a DC mode or how is it ramped; does it need to run in a bipolar way; will the magnet style be needed in large quantities; will groups of the magnets be strung on one power supply; are there vibrational tolerances; does it matter how much heat the magnet gives off; how precisely will it have to be placed and aligned in the beamline; add your favourite characteristic?


(2) It is not always necessary to do a 3D computer model of a magnet, a 2D model can often produce a magnet that meets the field quality requirements.


Fabricating Magnets


(1) The un-allowed multipoles in any magnet arise from asymmetries created during the fabrication process, usually in the steel core but could also be in the coils. In order to estimate the mechanical tolerances to put on the magnet drawings one can use a set of coefficients developed by Klaus Halbach in 1969, which correlate different types of errors in the placement of poles with the creation of un-allowed multipoles, e.g. a sextupole component in a quadrupole. Magnets can be mechanically measured on a Coordinate Measuring Machine to find the fabrication asymmetries and then magnetically measured so one can evaluate the correlations predicted by Halbach’s perturbation analysis. Knowledge of such correlations can guide the setting of mechanical tolerances on future magnets.


Magnetically Measuring Magnets


(1) What parameters need to be measured on standard magnets? 


(2) When is an integrated strength not sufficient, when is a 3D field map needed? Under what circumstances would one need to make a 3D field map of a solenoid?

 (3) If a dipole were curved would a 3D field map around the ends of the magnet core be useful to the beam physicist?


(4) What order polynomial do you want to use to relate current to integrated strength; this will affect how many different currents to measure the strength at?


(5) How do you calibrate the magnetic measuring apparatus so its field accuracy is reliable? Should all types of apparatus be traceable to an NMR probe?


(6) How precisely do the higher-order multipoles need to be measured?


(7) Does a beam physicist need to know the sizes of multipoles beyond the 2nd allowed harmonic, e.g. in a typical quad do you need to know the 28-pole’s size, in a typical dipole do you need the 14-pole’s size?


(8) Does a beam physicist need to have separate measurements of the normal and skew higher-order multipoles?  How precisely do they need the angles of the poles measured?


(9) If there are for e.g. 5000 quadrupoles of exactly the same style fabricated for a new accelerator, and there are not enough resources to magnetically measure every one, what is the lowest fraction of these quads that would need to be measured so a beam physicist would still be able to commission and operate the beamline?  The rest would be characterized by average values.


Operating magnets in an accelerator beamline


(1) What criteria are used to determine if a magnet needs to sit on a mover that can adjust the magnet’s position in the beamline via motors under remote control?  


(2) The beam physicist should have set a reproducibility tolerance on the integrated strength after the magnet has been turned off and back on. The magnet engineer will have developed a standardization procedure for the magnet to be taken through whenever it is turned on, that will guarantee it produces the same integrated strength at a set current to within that tolerance. If the beam physicist and accelerator operators get impatient and do not allow the magnets to be standardized when they have been off for any reason then the reproducibility tolerance may not be satisfied.


(3) When does online monitoring of the magnetic field value need to be done?



Overall Summary of the Workshop



The workshop has brought together the communities of particle accelerator beam dynamics and of accelerator magnets. The collection of talks has brought several topics to the attention of the audience.  The main conclusions of the meeting are the following.


1)    Magnets with nonlinear components are essential for properly controlling the beams with even simple optics.

2)    Dynamic aperture experiments at LHC confirm that extremely long term behavior of circulating beams can successfully be predicted by computer calculation.

3)    Correct predictions of beam dynamics models still remain challenging to make when there are high intensity beams being modeled (such as those at J-PARC).

4)    In accelerator projects the interplay of the magnets community and beam dynamics community follows a workflow, which is characteristic for each project. There were two different workflows that were identified in the discussions.  For the LHC, for example, the specifications of their magnets came from their beam dynamics modeling and were given to the magnet engineers before they started designing the magnets. In the other type of workflow, the magnet engineers make the design with the best field quality they can achieve; they do not receive any specs from the beam physicists. The values of the multipoles are given to the beam physicists and they put them in their model and see how the beam behaves. If it is not satisfactory then they will have to add corrector magnets and revise the lattice. The FAIR project has proceeded with this latter type of workflow.

5)    Magnet modeling in terms of predicting multipoles that would measured when the magnet had been fabricated was a long debated topic; on advanced topics (such as measuring multipoles at large horizontal positions in quadrupole apertures, one idea was to merge multipoles measured on several circles) the debate remained open.

6)    Use of measured magnet data for sorting out which magnet to place where in  a beamline when there are large numbers of magnets of the same type is a massive procedure, which requires careful long term planning at project level.

7)    For small machines such as a FFAG (Fixed Field Alternating Gradient), the distinction between magnet experts and beam dynamics experts is dissolved, because each community needs to have good knowledge of both topics in order to design a FFAG. .

8)    There were extensive discussions on specific FAIR related issues; the audience did not express any definite opinions on these FAIR issues.





Workshop discussion notes

The following scientific secretaries have contributed substantially to this workshop by taking note of all responses to the questions and comments during the discussions:

Christine Claessens, Lucio Fiscarelli, Paul Goergen, Manuel Heilmann,

Frederik Kesting, Oliver Koester, Philipp Schneider, Joschka Wagner,

Raymond Wasef, Christina Widmann


Sessions Notes: Monday 2nd December 2013


Session "Beam Dynamics: Nonlinear dynamics, do we care?”, chaired by Frank Zimmermann

Scientific Secretaries: Frederik Kesting, Christina Widmann


1) FAIR challenges to beam dynamics and to magnets, Dr. Peter Spiller

Slides are at


Summary by scientific secretaries:

- Evaluation: Have we done everything correctly for FAIR in terms of beam dynamics and magnets, since the series production of magnets will starts soon?

- Interconnection of beam dynamics and magnets at the point of field specifications (physical and technical reasoning); specific SIS100 issues, that have to be taken into account

- Problem of field quality specification before series production of magnets has started (uncertainties/random errors)

Questions and Answers

- Question: S. Russenschuck: Is the study of beam dynamics for the SIS100 as relevant as for the LHC?

Answer: Yes. Reference to studies done by G. Franchetti on resonance excitations in SIS100.

-Quality requirements on the magnets have to fit beam dynamics requirements

-Remarks on magnet properties used for sorting

-Purpose of this meeting: responsibilities, communication issues, find criteria that can be applied to become satisfied by a certain design - interesting not only for FAIR, but also for other similar projects



Question: S. Russenschuck: Why not using superconducting cos-theta magnets?

Answer: Because they are fast ramping, everything is okay with our design choice.


Question: O. Boine-Frankenheim: Skeptical comment, that magnet modeling is missing in the presentation, which is crucial.

Answer: With pre-series magnets, the model of the magnet can be corrected, but it's difficult to apply the correction scheme to the whole machine (using corrector magnets). The measurement of the whole series of magnets is important.


Question: (probably) T. Mouille: When and how (in the company?) will the magnet measurement be done?

Answer: Measurements will be done at GSI, especially measurement of the mechanical dimensions to see they meet the mechanical tolerances. People from GSI have been 50 times on the manufacturer's site, to ensure the quality of the first magnet.


Question: S. Sanfilippo: Which conditions do you have on the quality of the magnets?

Answer: There is a poor definition in the requirement specification. They have only the 2D lamination design, the producers are basically out of responsibility.


2) Beam Dynamics in accelerators: nonlinear dynamics and experiments, Dr. Wolfram Fischer


Slides are at https://indico.gsi.de/conferenceTimeTable.py?confId=2352#20131202.detailed

Summary by the scientific secretaries:

- Introduction to instabilities/nonlinearities (in storage rings), chaotic motion and dynamic aperture

- Magnetic field description/multipoles, fringe fields/non-linear fields

- Experimental techniques for beam analysis

- Simulation tools/particle tracking

- Examples of experimental results for different measurement techniques: Poincare surface measurement, amplitude dependent tune measurement, resonance driving terms in RHIC, frequency map analysis, amplitude diffusion measurement, echoes; all of these measurement can be compared to the results from simulations

- Dynamic aperture measurement and simulation for SPS: agreement within 10 percent!

- Remark on the Integrable Optics Test Accelerator (IOTA) at Fermilab

- Importance of interplay between simulation and experiment


Questions and Answers

Question: O. Boine-Frankenheim: What does 'well prepared simulation' mean?

Answer: All relevant effects are included; there should be a good communication between the different working groups to make sure, that the errors schemes used in simulations are representing the actual occurring magnet errors.


Question: S. Russenschuck: Using the experience from all existing machines and the tools available for simulations, is it possible to predict the behavior of the beam so well that we don't have to over specify the magnets anymore?

Answer: Tools are developed to predict with an error of 10-20 percent, but the safety margins also depend on the machine, e.g. the dynamic aperture is not this relevant in a collider.

Comment: F. Schmidt: at the beginning of the LHC a factor of 2 was aimed, nowadays 10% prediction of dynamical aperture is possible, but for this we have to know the magnets.


Question: G. Franchetti: There was a 10-20 percent prediction of the dynamical aperture at the PS and SPS experiment. How long did it take to perform these studies?

Answer (Wolfram Fischer): 5 years of experiments and modeling.

Comment (F. Schmidt): SPS is artificial, as oneself creates the nonlinearities.


Question: C. Spencer: With 2d modeling of magnets an accuracy of 1-2 percent is achieved, so an accuracy of 10-20% does not seem very tough.

Answer: What at the end is measured is the lifetime of the beam with variations larger than 1-2 percent; so knowing the dynamic aperture for 10-20 percent is quite good.

Comment (probably T. Zickler): The accuracy of the magnet as a real world product is lower than (worse than) 1-2 percent as there are other facts influencing the fields.


Question: S. Sanfilippo: Even an exact modeling doesn't enable you to predict every scenario, there are for example dynamic effects on the powering system, so more than (better than) 10 percent prediction is not possible.

Answer: But if there exists a good magnetic model of the machine, the dynamic aperture can be predicted in the timescale of minutes, otherwise not.


Question: L. Deniau:  To check the accuracy of a code you need to compare your simulation results to a measurement of the machine. Unfortunately there are not enough high-energy machines to check codes, furthermore all of them are somehow specific and so are the codes designed for each machine.

Answer: Codes are well tested for many machines and the results of three codes are compared for RHIC. As the same tools are tested for several machines, if there exists a correct model of the magnets, a prediction is possible.

2nd Answer (by O. Boine-Frankenheim): Important is not only the modeling of magnets but also of space charge effects, because the losses also depend on space charge induced resonance excitation.


Comment: P. Spiller: Importance of space charge modeling, since we are not dealing with zero-current machines.


Question: O. Kester: What is the effect of space charge for the dynamical aperture?

Answer (by G. Franchetti): Our beam loss study done at GSI, shows that it is a more complicated business, since resonances can be excited.


Question: S. Russenschuck: How well do we know the effects of each harmonic?

Answer: Simulation studies are possible, since we can switch off certain multipoles, as done for RHIC studies. The effects are always machine specific.


Question: F: Zimmermann: Why does the IOTA experiment have no resonance driving terms/ no resonance excitation?

Answer: You can find invariances of motion, even with non-linear fields.


3) Nonlinear dynamics of the collector ring, Mr Sergey Litvinov


Slides are at https://indico.gsi.de/conferenceTimeTable.py?confId=2352#20131202.detailed

Summary by the scientific secretaries:

- FAIR/Collector ring

- Importance of b5 multipole coefficients for dipole magnets and b6 for quadrupole magnets

- Beam envelope calculation for RIB and anti-proton beams, chromaticity correction

- Isochronous modes / ring as ToF spectrometer

- Frequency spread is directly related to the transverse emittance, in order to reach the necessary mass resolving power of 105 the transverse emittance would have to be limited to 10 mm mrad in both planes. As a result, the transmission of the ions into the ring would be reduced drastically

- Use sextupole magnets for chromaticity correction to increase resolution

- Sextupole  (b2) and octupole (b3) components affect the time resolution, but

can be completely compensated using sextupole and octupole magnets


Questions and answers

Question: S. Russenschuck: Are the fringe fields considered as a local distribution function or an average?

Answer: It's an average over all magnets, which is okay for the isochronous mode.


Question: C. Spencer: General complaint, that accelerator physicist only takes into account the effective length and not the physical length of a magnet. Thus, there is sometimes not enough space in between magnets, such that they influence each other.

Answer (P. Schnizer): Calculations with one magnet next to another were made.


Question: P. Spiller: Was the debunching after injection considered?

Answer: The stochastic cooling is done 1000 turns after injection; in isochronous mode there is no debunching.


Question: L. Deniau: does sextupole magnets or dipole design compensate the 1st order achromatic errors?

No answer...


Session “Magnets: the Superconducting Ion Synchrotron SIS100 enterprise at GSI”, chaired by Ulrich Ratzinger

Scientific secretaries: Oliver Koester, Raymond Wasef

Three talks in this session, no slides posted by Schnizer. Mierau slides can be found at

https://indico.gsi.de/conferenceTimeTable.py?confId=2352#20131202.detailed.  Marusov slides can be found at


Advanced multipoles for accelerator magnets and their measurement SCHNIZER, Pierre

Measuring the SIS100 dipole magnets MIERAU, Anna

Measurement of a dynamic field by rotating coil MARUSOV, Vassily


Schnizer summary by the scientific secretaries:

Pierre Schnizer talked about elliptic harmonics in a toroidal system. In detail he mentioned the measurement of three displaced circular harmonics, which he combined to calculate the elliptic multipoles. Using these he can determine the commonly used harmonics on the circle with a good convergence on the whole ellipse. The combining is done by using artificial weight functions, which lead to good results for the SIS100.

Questions and Answers

Q: Giuliano Franchetti: Can you really measure three sets of harmonics and combine them to get a valid result for one set of harmonics? And does it really work? FermiLab guys couldn't make it work.

A.: The paper with its calculations is a first start. It should be challenged and maybe improved.


Q: Stephan Russenschuck: Parts of the ellipse is not covered by measured data and why should not three sets of harmonics be used instead of one?

A.: Final accuracy is better and there is the possibility to extrapolate.


Q: Stephan Russenschuck: What about the area in the ellipse, which is not covered by the measurements? Can we extrapolate in this unknown area?

A.: It is possible and the results show, that the accuracy is good there, too.


Q: Marco Buzio: Is there a threshold of a/b of the ellipse for a good application?

A: No, you have to try and compare with measurements


Q: Cherrill Spencer: Why do not use the stretched wire to obtain the data on the border of the ellipse?

A.: This method is not accurate enough.


Summary of Mierau talk by scientific secretaries:

Anna Mierau explained the measurements of the SIS100 and its high requirements. She described the difficulties with the anti-cryostat and the low temperature.

Questions and Answers:

Q: Daniel Schoerling: Could the magnet get shimmed?

A.: It would take a lot of time to disassemble, probably about a month.


Q: Stéphane Sanfilippo: Which system is used as a crosscheck calibration, reference system?

A.: Main system is the rotating coil; the stretched wire is used as crosscheck method.


Q: Stephan Russenschuck: Why don't you use more displaced measurements to combine them? In y-direction e.g.?

A.: No space left because of the small aperture.


Q: How to measure the position displacement in the cryostat?

A.: Move the whole cryostat.


Summary of Marusov talk by scientific secretaries:

The talk of Vassily Marusov related to the measurements of dynamic fields with the rotating coil. He presented his approach of expanding the common circular harmonics by Fourier series in time. The following matrices showed his theoretical work for either fast or slow ramp rates to the rotating speed.

Questions and Answers

Question: What are the assumptions?

A.: Uniform rotation needed, especially between encoder ticks.


Question: Is this approach checked with simulated data?

A.: It will be tested with measurements.


Further details of the discussion are found in the video of the talk in the Marusov slot in  the indico page.


Session: “Beam Dynamics and Magnets: the LHC experience and beyond”, chaired by Pasquale Fabbricatore

Scientific secretaries: Raymond Wasef, Paul Goergen


The slides for this session can be found here:



Talk: The LHC dynamic aperture experiment (F. Schmidt)


Frank Schmidt reported on the benchmarking experiment of the dynamic aperture of LHC. The talk introduces the theory of the dynamic aperture, and details on the method for measuring the DA are given. The method is based on kicking a small beam and finding at which point in the phase space it gets lost. The talk shows an excellent agreement of measurement and simulations.


Questions and Answers

Q: S. Machida: What parameters do you scan with your simulations?

Answer: Phase space and seeds.


Q. S. Machida: Also the multipoles?

Answer: We have misalignment and multipole data, continuously adapted to the machine state.


Q. S. Russenschuck: Do you really need the factor 2 margin?

Answer: Given the quality of sorting, alignment, etc. it now seems not necessary, probably 20-30% are enough.




Talk: Sorting of the LHC magnets and lessons learned (M. Giovannozzi)

Summary of talk by scientific secretary:

During the talk, Giovannozzi emphasizes that at LHC, due to the delays in installation, there was a big pool of magnets available for sorting, which made accurate sorting possible.


Comment (L. Deniau): The behaviour of a magnet in terms of dynamic component was not used for the sorting but for operation.


Q. S. Russenschuck: Is the factor 2 gain on the b3 specific to the LHC, or it could be scaled to other machines?

Answer: It depends on many parameters (phase advance per cell, errors, etc.). Before sorting you should have an idea, depending on the random errors you expect, if the sorting will be worth it or not.


Q. S. Russenschuck: Would you recommend to GSI to store their magnets in a warehouse or on a field intermediately (in the meantime)?

Answer: Yes, depending on the random variation they expect. For low random variation it makes little sense.


Comment: Giuliano Franchetti: Yes for LHC this made sense because of previous studies on random fluctuations of magnets had shown it would bring a benefit.

A. Giovannozzi: Values came from simulation, from there the threshold was set, sorting was done according to the threshold.


Talk: Establishing C^3, the coherence between beam physics requirements, magnet manufacture (S. Russenschuck)

Russenschuck proposes a comprehensive view on the development of the magnetic description of magnets, and at the end of his talk proposes a new formulation of the magnetic methods raising the question if this approach would not be more suitable for the beam dynamics codes.


Questions and Answers

 Comment (S. Gilardoni): The experience in the CERN PS proved that good agreement between magnet people and beam dynamics people can lead to very good results. But one should remember that achieving these results took very long and it was difficult because of the lack of knowledge about the magnets. For the (very old) PS, we still think that we need magnetic measurements, so it should be done for every future machine.

(No further questions because of delay)

Talk: Field quality and warm-cold correlations: the experience of LHC dipoles and quadrupoles (S. Sanfilippo)


Stéphane reported on the experience made at CERN for the LHC magnets.

Details of the warm-cold correlations are given: for more detail the video of the talk is available in the indico page.


The talk was moved into the second afternoon session due to delay.

Questions and Answers

Question: What was the current of the test? Is the extrapolation to nominal current possible?

Answer: 10A, so very low compared to the nominal one, but the extrapolation was made using measurements at 5kA (at injection).


Q. S. Russenschuck: In retrospective: would you have done more or could you have done with less cold measurements?

Answer: In 2003 we wanted more cold measurements. In the end time and pressure led to less measurements. We need to pay the price: more offline tests for cycle, more field modeling, especially for special magnets.


Q. Cherrill Spencer: In discussion about a high fraction of measurements needing post processing. Don't you think that there were, maybe, a lot of errors due to the measurement operators? Many of them were new and had to be trained on site?

Answer: Because of the complexity and the size of the magnets, it is difficult to have high quality and reproducible measurement.


Question: Do you expect any problems when you go to nominal energy 7TeV?

Answer: No, everything has been designed for nominal energy.


Comment (M. Giovannozzi): The hysteresis effect will be better but the saturation will be worse, so effects may be worse or better.


Session of Monday 2 December 2013, from 16:55 to 18:30.

Topic: discussion + short talks: mainly magnets

Chaired by Mr. Egbert Fischer Scientific secretaries: Lucio Fiscarelli, Christina Widmann


Slides for Mr. Mouille’s talk available at


Presentation given by Mr. Thomas MOUILLE: ISIS TS2 Kicker Magnet Transient Modeling

Summary by the scientific secretaries:

The presenter starts by giving the main magnet characteristics.


Then he passed to describe the model for eddy currents (Transient Magneto-Dynamic Simulation) in order to investigate the field quality and the possible hot spots. The model is based on 1/8 of the real geometry. Hex mesh and layering are applied. The core is modeled with Fe-Si 270-50 laminations with anisotropic (packed) permeability and anisotropic conductivity. The coil support structure is modeled as a non-magnetic and isotropic conductivity material (S/S 316L) and the coils as Biot‑Savart conductors.


The standard current drive (cycle) without compensation produces a field affected by eddy currents as measured in 2007. A special cycle with compensation is introduced to fix the problem.


A static solver is used (Tosca Magnetic).


Comparisons with measurements are shown for two cases (with and without current compensation). The effects of eddy currents are simulated with a good accuracy. The simulated field quality is also well modeled and in this case measurements show some limitations.


Comparison of various simulation set-ups is presented. In particular, results for transient versus static solver and anisotropic versus isotropic permeability are reported. The computation times of transient solver are much longer and the results not always justify the choice of fine transient simulation. The same argument is valid for anisotropy of permeability: isotropic model is often enough.


Questions and Answers:


Q. Russenschuck: Why did you say that those measurements were not good?

A. Mouille: Putting together all the information one can understand that the measurements are affected by some errors.


Q. Russenschuck: They are field maps?

A. Mouille: No, they are several measurements taken at several positions by a static coil.


Q. E. Fischer: Have you made any comparison of different codes?

A. Mouille: No.


A. De Gersem: Also if you use several codes it is not easy to say which one is better.  There is no clear winner.

A. E. Fischer: Some codes seem to be not suitable for some specific application e.g. eddy currents.





Mr. Egbert Fischer starts the discussion by pointing out the problem of communication level between beam dynamics and magnet people. He gives the example of the SIS100 dipole magnet. Then he proposes to go through the list of discussion points already present on the web site.


The first point is the multiplication of safety factors as proposed by Mr. Russenschuck.


Taylor: someone should check there is only one safety factor applied and not an accumulation of them.


E. Fischer: as in many other problems, an optimization on several parameters is more difficult than on one parameter. This seems to be the case.


Spencer: this is a common problem in SLAC, for resistive magnets at least. In particular for the power supply maximum current. It is a question of trusting what other people do.


Sanfilippo: some margin is always needed because real objects are different from designed ones. Then a certain margin is also useful because sometimes a project will be upgraded and one cannot change everything.


Russenschuck: sometimes for certain magnets, for example solenoids, where some assumptions are taken then the requirements are set to be very strict. Then it is very expensive to build and to measure such a magnet. We need two weeks to map a single magnet and the data may be never seen by anybody. This is a cost. Simulations can help to avoid useless measurements. If there are good simulations, once the model is validated only a check of the magnet-by-magnet reproducibility is needed.


Buzio: In the lifetime of the PS Booster the energy has been multiplied by a factor of almost three with the same magnets. The quality deteriorates somehow, but this can be shielded away.


Franchetti: How many multipole components from the beam dynamics side are necessary to design a magnet. In the prototype of GSI a table with 29 orders of multipoles exists. In the measurement, up to which order of the multipole does this make sense? What are the experiences in other laboratories? What is the maximum order in the LHC magnets?


Answer (several): 11 and 15 in circular coordinates.


Comment (??): Up to order 30 might be needed in the case of elliptical magnets.


Comment (??): The question is, what kind of coordinate system you use if you have for example a rectangular aperture. Beam Physics does not care about the right representation of the multipoles; you need just to know a kick at a certain position.


Russenschuck: In the LHC measuring with a round aperture measuring with a round, long probe there is not really a discussion, also if you look at the studies, even if you look at the attribute in the physical meaning of each multipole. In smaller magnets with rectangular aperture it's different.


Franchetti: Someone turned on one of the octupoles in LHC to see if there is a special effect and nothing happened. So apparently LHC was also overdesigned.


Deniau: the value of these measurements go up to 10^13; above 15 I would not trust much measurement; the harmonic cell was used to find the magnetic axis of the dipole by canceling these harmonics. Canceling the first order did this but then it was found that it was better to cancel 3, three harmonics above 10 were finally used. The problem was the coils that were applied to harmonic order 13. So harmonic 13 and 14 you cannot trust. So finally we went back to just canceling the c10. The other problem is the theta. When you measure something you don't know where is. How you do the alignment of the theta to the frame of the magnet is something that is most of the time not discussed. We have a difference between warm and cold measurement and don't know which one goes right. We did top energy measurements and did the theta on the load line, which is a stable cycle for us, and then we use this shift of s and y permanently along the magnet and recalibrated the measurement for all the magnets. This is also true for the … and for normalizing the harmonics.


Franchetti: So all you used for the sorting was this recalibration?

Schmidt: We looked at the geometry, at b2, b3....


Carsten Omet: It was also shown that you should not trust everything above order 11, so this should also not be included in the beam dynamics.


Russenschuck: And don't scale them with the factor r_0^(n-1), that is worse than using nothing. But if you have a rectangular or elliptic magnet or a different aspect ratio, of course you can project that onto any set of functions.  We tried that at CERN with polynomials and others. After a while, we wondered if you gain much. In circular multipoles it is just a set of 10 numbers or 20, if you also take the skew components. That is a number you can handle. But as soon as you go to sets of 20 times 20 numbers that gets hard. Why is it not possible just to use a field map or a set of single and double layers? After some post-processing you could directly give that to your tracking code as a function of the spatial coordinates and get away from the harmonics and close to what we actually measure.


Schnizer: We measure something like fluxes. Only after we apply orbit corrections we measure fluxes...


Franchetti: Who knows if there are codes that can swallow potentials or fields and that can make long term tracking for the dynamic aperture from usual standard dynamics?


(Background): In plasma physics, for Tokamaks.


Schmidt: It was said that there is kind of a limit, where you can rely on the numbers of high order multipoles. So magnet people should not give the numbers of the higher orders to the beam dynamics people. On the other hand, if you do simulations and you find that the dynamic aperture is large, to be able to measure it, you need the higher orders. If the dynamic aperture is small, this plays no role. For instance at the LHC at top energy there was a correction system for the higher orders, but when it came to the beam-beam-effects, it played no role at all, as the dynamic aperture was small. Only the lower order resonances were of importance, so you cannot generally answer that.


Comment (??): The corrector magnets have also to be considered. The correction is trivial if you know the B3 of the neighboring dipoles. You start looking at field maps then the correction mechanism is fully covered.


Schnitzer: If we abandon circular multipoles we should reconsider how to build magnets. We are building magnets, which are basically some coefficients of a Taylor series. We could also use any other series.


Sanfilippo: You don't have to abandon the multipoles and the rotating coils. Taking the case of the LHC if there was not made any measurement before, at the companies, even up to b10 or b11, it would have ended in a catastrophe. All the magnets that were coming in at the beginning were completely off for the b3, the b5, b6 and b10 for some reason. Thanks to early measurements of the multipoles up to high orders the cross-section could be corrected and the catastrophe avoided. So we should not just ask ourselves what is the best you can do, but it should also be checked as early as possible.


C. Spencer: The difference is if you have a room temperature magnet, the fields are weaker so the signal you get out of your rotating coil is a low number of volts. For superconducting magnets you have a lot of field and you get a better signal, so you can go to these higher multipoles.


(???): Do you have pole shims to direct some fields?


C. Spencer: Rarely. Depending on what you are talking to and how strong the signal is that you get out of your rotating coil, you can go to higher multipoles or you can't.


(???): At least you can increase the number of turns in your (measuring) coils. And there is a lot of electronics to improve the reading of your measured signal. There is a lot of progress in the electronics, even for low currents.


(???): There is a solution if you modulate your current. Then you modulate the field of your magnets and by doing a Fourier transform you can amplify your signal.


Spencer: See talk of R. Tomas and myself about the magnets of the ATF2. Up to which order of multipoles did you consider in your simulations, Rogelio?


R. Tomas: That were relevant up to the order you provided. Up to the sixth order, which are 12 poles (in the quadrupoles). Above that nothing matters.


Pine: Coming back to the comment that someone turned on some octupoles at the LHC. The LHC is one of the accelerators, where you have very a good magnet model of the magnets and the machine works very well. So you can do the inverse study and you could bring errors in and see what it does with the beam, so you can define the tolerances. You could do experiments while the machine works.


Spencer: That is also something that should be done in beam dynamics. There should be a scattered error applied in the simulations. Magnets (of the same style) come out of fabrication with very different errors.


Pine: Has this anyone done or do you just say that it works? You could use that to see how bad it can make it and that could go into the design.


Schnitzer: In the end you need an extra 10 % margin.


Zimmermann: At HERA we just went to the control room and changed the sign of all decapoles. It did not make a difference.


Schmidt: Nobody looked at the chromaticity.


Russenschuck: We had recently magnets to measure with a good field zone with a rectangular aperture with certain accuracy. Where does this come from and how does this relate to multipoles? One of the colleagues panicked because he thought he would have to measure also in the last corner.


Schnitzer: People would say you have a certain area with a field variation of 10-4. We measured not at the mid-plane for SIS18 but with elliptical coordinates, the field quality variation was 10-4 not 2*10-4. Often things are specified using the methods of today.


(???): There was a flip coil used because the fields had to be measured between the poles, so there was no chance to measure with rotating coils. And of course this was with a search coil. If you have a dipole with a curved path you have the same representation in the measurement and the simulation in this way.


Schmidt: To make a machine bad also in simulations is not the problem. It is the way we have to do that anyway. The LHC is too linear, so we have to make it nonlinear.


Sessions Notes: Tuesday 3nd December


Session “Magnets & Beam Dynamics: from 50 years of accelerator magnets to achieving synergy”, chaired by, Yukiyoshi, OHNISHI

Session secretaries: Joschka Wagner, Philipp Schneider


All talks given on Tuesday can be found here on the web here:



Speaker: Thomas Taylor (CERN) “Magnets in accelerators: a Historical overview”


The talk of Thomas Taylor reviews his personal experience from industry to laboratory (CERN). The talk reviews the early history of magnets and discusses of the types of magnets (conventional, permanent, superconducting). His talk discusses superconductivity in LHC, and example of beam/magnet interface at LEP.


Questions and Answers


1. Questioner: Dr. Stéphane Sanfilippo (Paul Scherrer Institut, Villingen):

„I have two questions. Two subjects we discuss nowadays. The first one is that you mentioned about the permanent magnets, you may know I am sure you know that for electron machines for example the ESRF in Grenoble they planned a big part of the magnets as permanent magnets. What is the position of CERN respect for the future machine like PEAK. And we need thousand of magnets to do different things but we considered an option also to use for a small part permanent magnets “

Answer: Tom Taylor

„It is considered. Anywhere were you do not have to vary the field and it has not to be that high it is a fantastic thing.“


Q. Stéphane Sanfilippo (Paul Scherrer Institut, Villingen)

„And the second question will be related to the philosophy of the LHC that you mentioned, you know so that there is a lot of companies now developing more and more of magnetic systems characterized in a more precise way, also the magnets. And there is this discussion, do we have to change this philosophy to give the responsibility of the magnetic, lets say design and magnetic measurements to the company? What do you think about this all? I mean not for a gigantic project like the LHC“


Answer: Tom Taylor

„You have to be careful. It is whenever, I know many occasions where people have thought they could go and buy a magnet I mean (“off the shelf”). There have been laboratories, especially for the experiments, they think they can go and buy a magnet.

You have to know, you can do it but you still have to have somebody who is buying the magnet who knows how to make a magnet.“


Q.: Stéphane Sanfilippo:

„So what will be the criteria? The complexity of the magnet, the reliability of the company?“

Answer: Tom Taylor

„You need to have somebody in the institute who knows enough about magnets to be able to follow the construction. The magnets we have made for the LHC for experiments, two magnets [...] and I can tell you that we had to send a technician from CERN. One time he spent 6 months. We had to send tooling made at CERN.“


2.Questioner: Giuliano Franchetti (GSI Darmstadt)

„I wanted to ask you one thing on your conclusion. You marked two points in red. The safety factors, apply sensible safety factors. Is it a question?“

Answer: Tom Taylor

„No, it is not listed as question. This is an important thing“


Speaker: Cherrill Spencer (SLAC National Accelerator Laboratory, Menlo Park California, USA) “Procuring the ATF2 magnets: new and recycled”


There is no summary?

Questions and Answers.

1.Questioner: Dr. Stéphane Sanfilippo (Paul Scherrer Institut, Villingen)

„Coming back to this collaboration (Japan, USA...). Would you advice or propose for a long series of magnets, of course if it is possible, the same measuring system in the case for all the dipole measurements? In posing the quality insurance that all the people who are making different magnets will use the same measuring machine.“

Answer: Spencer

„Well if you are procuring 10,000 magnets you can't use the same measuring apparatus for all of them; if you have a set of magnets of the same style you can arrange for them to be measured by measuring apparatus of the same design.. “


Q: Stéphane Sanfilippo

„In the case of LHC for example you make magnetic measurements that one in three companies impose the same one, the same magnetic system to measure the same thing.“

Answer: Spencer

„I would force them to have very similar systems, but we weren’t realizing how important the skew and the normal components were, not just the overall multipoles.

So yes. A lot of oversight is needed. The magnet engineer, as R. Tomas saying, somebody back at the home-base, that magnet engineer needs to know about magnetic measurements and need to be in charge of this, or someone from your magnet measuring group.“


2. Questioner: Stephan Russenschuck (CERN)

„I don’t know if I understood correctly what you said about the skew sextupole. It seemed to depend on the powering scheme, or the configuration of the magnets.“

Answer: Spencer

„It is the positioning of the 4 iron poles that matters. The iron poles (in a quadrupole), if they are not symmetrically placed then you will get a sextupole component, it could be skew or normal.“


Q: Stephan Russenschuck (CERN)

„But you said after a while the measurements changed for the third batch and the fourth batch.“

Answer: Spencer

„Yes that was the Chinese technicians, they were told they had to rush along and hurry up because we needed the magnets. They were not taking much care as they had in the first place.“


Q: Stephan Russenschuck (CERN)

„I see. I thought it was about powering during the measurements and they did different clarity conventions and so on“

Answer: Spencer

„Yes, and you see that the sextupole got bigger as time went on, that was one thing. The measurement people were rushing and they were not paying as much attention as they had. So it was in the end a measurement problem. Regarding the general increase in the sextupole component- that was caused by rushing in the fabrication; the strange none agreement of the last ten magnets, that was a rotating coil problem. Not knowing which way it was rotating, thinking they knew what they were doing but they didn't.“


Q: Stephan Russenschuck (CERN)

„Because we had recently the discussion whether or not we should measure the magnets in their final configuration in the machine, as we always said in the past. We should come through that. We should measure it in the convention that we have for the measurement frame. And then we say what the optical function is later, then the responsible people for the configuration management have to do their transformation.“

Answer: Spencer




Speaker: Rogelio Tomas (CERN) “The ATF2 Story”

There is no summary for this talk.

Questions and Answers

1. Questioner (during talk at min07:05): Frank Zimmermann (CERN)

„Perhaps I missed something. Why were the tolerances not correctly calculated? “

A: „I didn't say they were not correctly calculated. I said it was very difficult to understand what the tolerances are (looking at the design report).“


2.Questioner: Wolfram Fischer (Brookhaven National Laboratory)

„What was the reason for this low current in one of the (sextupole) coils? Are they all in series? “

 Answer: Cherrill Spencer

„We thought maybe there was a short, a turn to turn short, so there were fewer amp/turns there than we needed. But they didn't have any time to actually establish that; they only had three sextupoles, so they switched that sextupole with another one that runs at very low strength. We didn't have a spare sextupole so they didn’t have the possibility in the time available so they found a different solution which worked.“


3.Questioner: Frank Zimmermann (CERN)

„Maybe a question to Cherrill. Cherrill showed this stabilized table, which I think maybe it has the two quadrupoles and two short sextupoles. 

 Answer: Cherrill Spencer

„They are on the same table, the final doublet and the 2 sextupoles. “


Frank Zimmermann (CERN)

„ And the beam size monitor was not on the same table?“

Answer: Cherrill Spencer

„It was four meters away where the Interaction Point would be“


Frank Zimmermann (CERN)

„Is that also on a similar table, the beam size monitor?“


Answer: Rogelio Tomas

„[It is not on a similar table, but] it is established with the similar system, that gives the similar stabilization“

Answer: Cherrill Spencer

„It is not on the honeycomb table. You need them (magnets and beam size monitor) to move up and down together. As there is this coherence in the floor. Across the whole four meters apparently the floor behaves similarly.“


4.Questioner: Peter Spiller (GSI)

„A very similar question. How stable is your building? It typically moves on the ground about centimeters.“

Answer: Rogelio Tomas

„This has been measured. There is moving [...] “


Q: Peter Spiller (GSI)

„How do you repeat your beam line alignment? After a few days your alignment is destroyed“

Answer: Rogelio Tomas

„This is done. Exactly, there is a continuous monitoring of the beam size and position, we estimate how it is detuning and detecting. In the future linear colliders will not operate with continuous [?..]“


Answer: Cherrill Spencer

„A new special floor was put down. We had to wait while it shrank before installing any beamline equipment and had to get air conditioning to regulate the temperature. All these things had to be taken into account.“


5.Questioner: Oleksiy Dolinskyy (GSI)

„What is the main limitation for the beam size? Were the constraints not enough?“

Answer: Rogelio Tomas

„ […] All what remains are the chromatic aberrations, and a high intensity effect […]”


Session “Magnets: the challenge of the design”, chaired by Cherrill Spencer

Scientific secretaries: Joschka Wagner, Philipp Schneider


Slides for all the talks in this session can be found at



Sebastian Schöps (TU Darmstadt): “Approaches for the quantification of uncertainties in stochastic magnet design


There is no summary for this talk.

Questions and Answers

(16:30 Min.)


Q. Marco Buzio (CERN): What are the units on the vertical axis?

A. Sebastian Schöps (TU Darmstadt): That should be Tesla-to-Tesla, right? So Tesla.


(16:39 Min.)


Q. Pierre Schnizer (GSI): One, two, three, four. You're speaking about 60 ppm, 30 ppm here, yeah? 60 ppm to 20 ppm....

<Unknown>: Well, take the relative, a factor of five. In between four to six....


A. Sebastian Schöps (TU Darmstadt): I will conclude and we will go back, is that okay?


Q. Rogelio Tomas (CERN): But you need the radius? What is the radius?

A. Sebastian Schöps (TU Darmstadt): I don't know the radius by heart. I can give you the radius, if you're interested, so I mean it's not about the particular multipole, right? I just want to tell you how you can do it to give reliable bounds. So I know, that many might think: "Well, I've known as a rule of thumb". Yes, of course. You could have the rule of thumb, but if you do the rule of thumb, the rule of thumb, the rule of thumb, you end up with constants. This is a method that can give you reliable bounds. That's the method I want to show here.



(20:06 Min.)


Q. David Scherling: I have one question. I guess you used this method <not understandable> before insurance plan. I'm just wondering how you plan or if you or someone else has an idea how it's reflected the quality insurance plan of the magnet. If you for example buy the steel, and...


Pierre Schnizer (GSI) (heckling): Very simple. This is a university-study and it's not...


Q. David Scherling (continuing): And also in the drawings. There is always the question how much tolerance you allow. And this could give you the answer.

For tolerances.

A. Sebastian Schöps (TU Darmstadt): Yes, exactly.


(20:43 Min.)

A. Pierre Schnizer (GSI) (heckling): The magnet we are building it’s done in house. All the studies that we prepared for that. And the tolerances that you need for the field quality is above what you can reach with final blank.


Q. Sebastian Schöps (TU Darmstadt): Can you explain me? I didn't understand.

(21:03 Min.)

Pierre Schnizer (GSI): A magnet. We know the geometry, tolerances we want to reach. These tolerances are below what you can reach with final Blanchard grinding which is the limit of what we can do on the magnets geometry. We did a lot of studies on the steel, so we know what values the steel has to fulfill and we're going to measure the steel.


Q. <Unknown>: So you measure every piece of steel?

A. Pierre Schnizer (GSI): You measure random samples, you can't measure every...


(21:31 Min.)

Q. Sebastian Schöps (TU Darmstadt): You use the Monte Carlo method I guess, for that?

A. and Q. Pierre Schnizer (GSI): No. We know what the steel has to do. That the steel is only important for the injection and the perturbation and therefore we would like to ask you what you have did on the hysteresis effect, because it was unfortunately not shown here in <...>. Because the collocation measures all this changes the parameters in the sample you use.  (Further details of this discussion are available in the video of the talk in the indico page for this talk).


(21:58 Min.)

A.  Sebastian Schöps (TU Darmstadt): First Question: We neglect hysteresis because it much more difficult to make a study, a numerical study, but main point: If you have an in-house code, that can do hysteresis and you have reliable tolerances on hysteresis, than you can just put this into the model and it will work. Again: I was not talking about a particular magnet here. I'm telling a method. I mean that’s making the difference. I don't claim that I build this or compute the magnet. It’s theoretical. Second: You can use other models for the material, it doesn't matter.


(22:49 Min.)


Q. Marco Buzio (CERN): I have a comment. I think this method is extremely interesting but I think also that the results are a bit conservative. Because: in reality the stochastic variation of the permeability applies to every volume of the iron. Whereas here, if you don't want to touch your model, you consider a similar curve. Would that be investigable already <...>?


A. Sebastian Schöps (TU Darmstadt): Yes, Definitely. I mean this kind of expansion that I showed you, I did this for the curve but you can do this also in space somehow. And than you can also consider different material, anisotropy for example that is changing somehow in this material, this can be done. That's part of the problem, we didn't do it, but it's definitely possible. You will need additional random variables again, right? Because you cannot go with three maybe, but it’s definitely, I mean, it's competitive, this can be done.



(23:48 Min.)


Vasily Marusov (GSI): <...>


Sebastian Schöps (TU Darmstadt): Yes, sure. Linearization as effect of parameters.


Vasily Marusov (GSI): <...>


Sebastian Schöps (TU Darmstadt): You can increase...


Vasily Marusov (GSI): <...>


Sebastian Schöps (TU Darmstadt): It's linearization, that’s right.


Vasily Marusov (GSI): <...>


Cherrill Spencer (SLAC): You have to have dinner together, because Marusov already going out with Russenschuck for lunch....



(25:15 Min.)


<Unknown>: You said you used three random variables to represent the whole distribution.

A. Sebastian Schöps (TU Darmstadt): Not the radius, the material curve.


Q. <Unknown>: Oh, sorry, yes, material curve.

A. Sebastian Schöps (TU Darmstadt): Dangerous topic!


Q. <Unknown>: How can you get to this number?

A. Sebastian Schöps (TU Darmstadt): The three?

A. <Unknown>: Yes. Just by trial?

Q. Sebastian Schöps (TU Darmstadt): Well, educated trial. You can do discretization and can go up with higher order here and you can see how fast your eigenvalues decrease. And at some point you say, I neglect the higher orders, as you always do. And you get quite reliable results. You could have easy done with one more, but there is no difference at all. So it’s quite sufficient. But again: the method scale is not very fortune with numbers of random variables, we tried to limit this.


Q. <Unknown>: I guess it depends on the curvature of that?

A. Sebastian Schöps (TU Darmstadt): Yes, exactly. So we tried different material curves, and tree or four seem to be a very reasonable number. Actually you see, already the first guy here is quite rough but a good approximation. And they are more or less higher modes if you add them.




Kei Sugita:  “Special design aspects of the superconducting corrector magnets in SIS100”


No summary of this talk.

Questions and Answers

Q. Peter Spiller (GSI): You're talking about the multipole correctors at the end of the arc, or which correctors? Or this is sextupoles, or...?

A. Kei Sugita (GSI): Yes...in general.


Q. Peter Spiller (GSI): In general I wouldn't agree.


- both laughing -


A. Kei Sugita (GSI): Afterwards there came requirements, but at the beginning the requirements from the equidistant <???> accelerator...


Q. Peter Spiller (GSI): For a multipole corrector at the end of the arc, that's a certain....

A. Kei Sugita (GSI): Yes. - Nodding -



Q. Peter Spiller (GSI): Since you are completely ready and fine with the dipole and the allocation of designers seem to be correct. You are ... you ready ... s-dipoles and quadrupoles.



A. Giuliano Franchetti (GSI): I wanted just to say one thing. It is interesting this factorial: 1/n! In all beam dynamics all the codes likes Sixtracks, MADX, etc., they all the codes use this convention. So that basically the expansion of the non-linear fit is not done within the convention of a magnet dipole. But this cannot be changed. It's a tradition that goes on since years, years and years and if I want to compare my own results with other people, nonlinear beam dynamics people use their convention on “multipole”. 


Q. Kei Sugita (GSI): Yea, that's why I....

A. Peter Spiller (GSI): We should recognize this topic for the final discussion.


- Heckling, laughter -


A. Cherrill Spencer (SLAC): ...and then the magnetic measurement people; I know, when my magnetic measurement people measured a sextupole for me we made sure to know whether they have a two in their formula or not, so we spent some time to discuss this. It's all just a question of communication and talking to the right people. Taking them out to lunch and dinner.


- Laughter -



Q. Stéphane Sanfilippo (PSI): Just to comment to your little bit provocative sentence about dipoles, quadrupoles and correctors. I'm talking under the control of Massimo, but for the case of LHC for this pole pieces: we had one of the reasons what we taking cross section of the quadrupoles and the dipoles, because of the capacity of this pole pieces to correct. So it's not true that we design the dipoles and we say ok the corrector should be that, in that case we have made a new report and we created some properties from the socket because the capacity of the correctors was fixed in <pica? > and we had to move two times the cross section of the dipoles and the quadrupoles. So it’s a more complex I would say discussion for the correctors than for the quadrupoles.

A. Kei Sugita (GSI): OK, maybe after the prototype.... <not understandable>


(15:23 Min.)


Q. Oleksiy Dolinskyy (GSI): Just one Question. You have so many correctors, dipoles, multipoles and sextupoles in one direction. Is the expectation on the fields to couple, all these coils.... some influence on the field quality?

(15:41 Min.)

A.  Kei Sugita (GSI): Yes, the...


Q. Dolinskyy (GSI): If you change the current in one coil,...

A. Kei Sugita (GSI): Yes one program that is monitoring ... between coils. But normally the cross speaking has not inter on this. At the end we will get. But there is more.... < not understandable> Second problem is the superconducting filament that is magnetized. I checked this value but it’s very low in this case.


(16:18 Min.)


Q. Cherrill Spencer (SLAC): You had a 3m long magnet...

A. Kei Sugita (GSI): ...a dipole...


Q. Cherrill Spencer (SLAC): ...a dipole. To me you only... When I have anything over one meter, I find I could ignore the ends. The ends are such a small fraction of the integrated field, that I'm not worried about making even a 3D model.


A. Giuliano Franchetti (GSI): But you have a single pass beamline..

A. Cherrill Spencer (SLAC): Well, I designed magnets for various types of beamlines.. But anyway. With a 3 meter long dipole...


(16:57 Min.)

A. Pierre Schnizer (GSI): We checked it. It's because the coil is so small, this was a compromise to make between field quality and actual losses on this magnet. So we had to have a small coil. And than it's the high current density we have in the coil and the coil is very near to the aperture and so on and so on. We checked it very carefully and the ends are significant.


Q. Cherrill Spencer (SLAC): And this has to do with how the coil actually is formed?


A. Pierre Schnizer (GSI): Well, basically it's a current density effect in the first place.


Q. Cherrill Spencer (SLAC): Okay.


(17:38 Min.)


Q. Ben Pine (STFC): Any works on high intensity machines there are the corrector magnets to avoid at power upgrade to which the design <...> availability and performance. At first I massively agree with you that the necessary of corrector magnets in SIS has been a challenge. <...> In fact in any of these machines there are never enough corrector magnets. You are always working at the limits.


Kei Sugita (GSI): - nodding -


(18:18 Min.)


Q. Tom Taylor (CERN): I'm going to comment is what you said about the design effort. You're absolutely right. We had the same problem at LHC. The thing is, the cost of the dipoles and quadrupoles, the money is so much more. So the political leader is interested in these elements. Whereas he's not interested in for example matching quadrupoles in the case of LHC, correctors, all that stuff. Having been in time with that we are in see we can tell you something quite fit. <...> the dipole people always said they will give us people (resources) when they're finished - but they never finished.


- Laughter -


Kei Sugita (GSI): - nodding -


(19:00 Min.)

A. Ben Pine (STFC): I think there is a related problem in LINAC-design. I've got friends who do LINAC design. What they do with the LINAC, they design it pretty long with diagnostics and correctors in between LINACs (accelerator sections) and then they come to build it and they say, well they know it should work, but it never worked then. So they use diagnostics to know where it fits.


Q. Tom Taylor (CERN): If you come late in the project you don't have any money. If you come earlier you can come and commit your money. If you are late there is no money and no people available to do the work.


Kei Sugita (GSI): - nodding -


Session “Beam Dynamics: The state of affairs at J-PARC, KEKB and small machines”, Chaired by Christopher Prior

Scientific secretaries: Manuel Heilmann, Christine Claessens

Tuesday 3.12.13   14:00 - 16:00



1. Yukiyoshi Onishi

2. Susumu Igarashi

3. Hiroyuki Harada

4. Ulrich Dorda

5. Ben Pine

Slides for these 5 talks are available at https://indico.gsi.de/conferenceTimeTable.py?confId=2352#20131203.detailed


Questions and Answers for the 5 above talks (numbered according to speaker number)


1a)     Cherrill Spencer

Question: How do you workout the 35 microns movement in the radius? Did you do that by modeling?

Answer: This is mono calculation. We measure the error field by using a rotating coil for the prototype. Then we can estimate the error field. ...

Question: There is no iron in this?

Answer: No iron.

Question: Is it still very expensive?

Answer: Yeas, it is expensive.


1b)     Rogelio Tomas

Question: Do you foresee beam-based methods to setup your correctors?

Answer: Yes, but before that we use rotating coil etc.


2a)     Peter Spiller

Question: Observations disagree from simulations. So your machine model is not perfectly describing the observations?

Answer: Yes.

Question: How do you generate the set values used for the correction?

Answer: By measuring.


2b) ?

Question: Concerning the losses: Chromaticity by this 5 ms...

Answer: Yes chromaticity is very sensitive, so we have optimized that.

Question: Do you optimize it while you are the dumping the beam?

Answer: Yes we tried everything to optimize it.



Question: Do you observe losses because of this excitation at 0.2?

Answer: Yes.



Question: Do you make the magnets or does a company make them?

Answer: Made by a company (Hitachi).

Question: Did they design then for you?

Answer: Designed by our magnets team at JPARC.



Question: Why do you need to change the linac RFQ?

Answer: For optics reasons.


3d) Hiroyuki Harada

Question: Your simulations reproduce all the experimental results. Did you make simulations before experiments and make comparison before and after?

Answer: We did it, but only possible after including everything.

Question: But to make some requirements to the magnets you should simulate something like them, and you say specify the optics.

Answer: I don’t think we can say by simulation we set up the requirements. We ask the magnet people, what is the best thing they can do...(laughing)



Question: The errors of the welding of the vacuum chamber, is that coming from the permeability of the welding?

(Answer: Thomas is probably more competent.)

Thomas: I am not a welding specialist. You cannot simulate the welding effects.



Question: Let me go back to the changing the magnets center, the measurement there. Was that a dynamic measurement or did you do a series of static measurements.

Answer: static.



Question: You said you want a good field and you are especially interested in the higher order multipole harmonics because you are tracking through this, but why a limit of 10^-4. If we would have found it's only 10-^2, could you live with it or not? And how do you verify that?

Answer: Going from the electrostatic to the magnetic septum if you have too big errors there, you would get problems. The beam study calculated this very carefully. This is how the requirements come up. 10^-3 will have to be possible during the ramp. With 10^-4 we know we are on the safe side. I would rather invest more money now and therefore save just a single day in commissioning than find out we are in trouble later.



Question: 1. When you put your losses,.. Is this a measured tune or a ramp Tune?

Answer:  This is a program tune. A program pointing one plane and a linear ramp in the other plane.

(Question: When you put your losses. ... If you increase the density you increase the space charge.... ?)

Discussion about which plots in the talk are measurements and which are simulations.


Tuesday afternoon session

Chaired by Wolfram FISCHER, scientific secretaries: Paul Goergen and Frederik Kesting


Talk of Ingo Hofmann “Comparison of quadrupole and solenoids dynamics”

Summary written by scientific secretaries:

The talk starts with an overview of the history of strong focusing, beginning with Courant and Snyder. After quadrupoles, other options for focusing are discussed. For focusing simultaneously in both planes a magnetic horn can be used. Another alternative is a solenoid. A formula for comparing a quadrupole doublet with a solenoid in terms of focusing strength is introduced. Emphasis is put on the fact that the fringe fields are dominant in the focusing properties of a solenoid. Ingo

Hoffman proceeds to discuss the use of solenoids for focusing in laser acceleration. In laser acceleration, due to the high energy spread, chromatic effects are dominant. However they can be of use in energy selection. For a short solenoid the chromatic behaviour is less dominant. There on the other hand the spherical aberration plays a strong role. However in a recent paper (B. Biswas in Ref. Sci. Instr. 2013) it was shown that for a pancake solenoid the spherical aberration is again low and comparable to a long solenoid (of inverse aspect ratio).


 Questions and Answers


Q1 (Zickler): The field quality of a solenoid: how is it defined, how is it presented, what are important parameters?

Answer (Hoffman): As a beam physicist: we always assume field is axial symmetric. But to go more into detail: We had an experiment with laser-accelerated protons. Shortly after the target they were to be focused by a solenoid. But we could not get any focusing. After a while it turned out that this was due to distortion of the fields due to the current leads.


Q2 (Russenschuck): You need to take into account for these effects from the leads or from elliptic deformation. How do you feed this information to your tracking codes?

Answer: For this specific case we had a student who used the full 3D field maps. He was able to explain the focusing problem, but unfortunately things did not improve to the extent we had hoped. We suspect this is due to the fact that it is a pulsed solenoid, which is especially hard to model.


Q3 (Simone Gilardoni): Did you also look at solenoid doublets?

Answer: So far we have not but we should try this.




The session starts with one slide by Giuliano Franchetti.


Giuliano Franchetti:  On the 17th of January 2014 the first SIS100 Dipole will arrive and will be measured. We will then have to decide if it is OK. What criteria can we use? Firstly we could measure, take the numbers and put them into the beam dynamics. So there are already simulations with an assumption for the random errors. The last simulation used numbers supposedly representing what we expect from the company. So we are proposing the criteria to decide if we can accept the magnet is if the magnet is within 2 sigma of the assumed random error. Otherwise we would have to ask where the error comes from. Either we suspend the series and try to find the reasons for the discrepancy. Or we can accept the series, measure all components for the first three or four magnets and work out a compensation scheme.


Carsten Omet: We will anyway measure the components for the first magnets; independent from whether the very first is good or bad.


Russenschuck: When you only have one magnet you cannot say anything.


Stéphane Sanfilippo: We can for instance compare to the SIS18 reference magnet.


Giuliano Franchetti: Asks the community how it was done at other labs


Massimo Giovannozzi: There are many sources for random errors. For LHC first there were measurements at the company, warm. Then we had to wait for a cold test, we would see additional random errors from waiting, magnetization, etc. In the end: the random errors were taken as a linear sum of all the components. They have to be taken together.


Giuliano Franchetti: Currently we only have random errors from the model so it would be interesting to know whether for other people there were discrepancies from their models?


Counter question (Chris Edmonds?): Why do you only order one magnet and then accept all the others?


Carsten Omet: Because our series is only 100 magnets.


Stéphane Sanfilippo: For LHC one source of error was that we had different types of cable. For you have only one type of cable. So your main error comes from the geometry.


Pierre Schnizer: It's first geometry and then hysteresis.


Comment on LHC: No Magnets were rejected at LHC. ``Bad'' magnets were thoroughly inspected to spot manufacturing problems in time. Discussion, result: usually bad magnets become spares


Discussion: people discuss about the validity of the model: Conclusion: have to treat superconducting iron dominated magnet like a normal conducting magnet, then main source of error is geometric. Maybe also the hysteresis.


Giuliano Franchetti: One proposal is: Measure the magnet, and then do DA studies with the data of said magnet.


Wolfram Fischer: If the magnet falls into the distribution of errors expected from simulation you can accept is. Otherwise you have to check: how far outside?


Pierre Schnizer: To make a fast decision we could have a list of what harmonics the machine can tolerate.


Giuliano Franchetti: We did this already. For the random errors expected.


Wolfram Fischer: If there is an error found, one harmonic way too high. There are tools to quickly patch the series?


Pierre Schnizer: For certain variations in the laminations we reserved the rights to change things. If there are still problems after 2nd or 3rd magnets it will be more difficult to change.


Comment by Stéphane Sanfilippo: Sometimes variation in B-H-Curve.


Pierre Schnizer: They will be made from one batch of steel.


Giuliano Franchetti: What about sorting?


Massimo Giovannozzi: You have to do simulations to see if it helps you.


Oliver Boine-Frankenheim: Question to the magnet people: Given random errors in machining from suppliers, if you put them into simulation, how confident can you predict multipole fluctuations from these random errors? (no clear answer)


Wolfram Fischer: What should the sorting criteria be?


Giuliano Franchetti: Optimally strength of resonances excited. In LHC they used b2 b3 and the aperture. b3 because of working point near 2/3.


Wolfram Fischer: Also think about the size of the sorting pool..


Carsten Omet: There are different families, so we have very few magnets to sort. 6 per quadrupole type, half of the dipoles … only thing to sensibly sort are dipoles.


Some more discussion on whether cross talk between magnets is to be expected, it seems to be a matter of discussion whether 1.9T is close to saturation or not. GSI does not plan shielding between magnets.


Sessions Notes: Wednesday 4th December 2014


Session "Advanced Concepts and Development", chaired by Thomas Weiland

Scientific Secretaries: Christina Widmann, Oliver Koester, and Joschka Wagner



1) Speaker Jean-Baptiste Lagrange,FFAG - the working marriage of magnets and dynamics”


Summary of talk written by scientific secretaries

- New concept with unconventional magnets, usually the same people doing the beam dynamics and magnet design

- Focus of the talk on scaling FFAGs, i.e. FFAGS with nonlinear fields to keep a constant tune

- Equation of motion independent of the momentum p leads to two conditions: similarity of the reference trajectories and invariance of the focusing strength

- Strait scaling cell prototype was tested

- Not the multipole components of the fields are analyzed, but the magnetic field measurement is compared to the analytical / theoretical field lope (?)

- Possible application for the linear cell: nuSTORM, a muon storage ring (racetrack shape) for neutrino experiments


Questions and Answers

Q.  G. Franchetti: As the multipole components of the fields are not measured, how is the field quality of the magnets compared and how are tolerances defined?

Answer J.-B. Lagrange: It is just done by error studies, no generalized way; the tolerances also depend on the purpose of the machine: with just 100 turns the accuracy is not that important, with 1000s of turns (e.g. PAMELA)  high accuracy is necessary


Comment Shinji, MACHIDA (STFC/RAL): Not a good idea just to rely on the accuracy of model, the concept of multipoles could still be helpful to determine e.g. the dynamic aperture or the chromaticity analogous to synchrotrons.

Answer J.-B. Lagrange: see next talk, where the FFAG magnets are described with multipoles; but as long as you have an exact magnetic model why not stick to that instead of expressing it in term of multipoles


Q. --> C. Spencer: What did you measure the magnets with?

Answer J-B. Lagrange: it was measured in 3d with a hall probe

Comment C. Spencer: it took a long time?!?


2) Speaker Suzie Sheey:Helical Coil Magnets: Advantages and Challenges from the PAMELA medical FFAG study”


Summary of talk written by scientific secretaries

- PAMELA is a design study for a proton non-scaling FFAG for medical applications

- Strong magnetic fields (1T -4T) are necessary, furthermore the beam moves in radial direction so a large region of good field quality is required; the only possible choice are superconducting helical coils

- As the magnets are quite short (50cm long) they consist almost only of end fields, so no hard edge approximation can be applied for the beam dynamics

- The multipole analysis cannot be done at one point in the center but the fields along the whole magnet have to be considered, so a polynomial fit was used

- To get fast field maps, rather than using field maps exported from TOSCA, instead an in-house Biot-Savart-solver for generating the magnetic fields was developed; it was directly included in the tracking code

- What was learned in the project is that the beam dynamic people and the magnet designers have to work close together during the whole process


Questions and Answers

Q. --> S. Russenschuck: Are these superconducting magnets already built and tested?

Answer: No.

Q. S. Russenschuck: Did you consider that there is an enormous peak field enhancement between the layers of the helix?

Answer S. Sheehy: The peak field on the wire has been studied in detail; the values were in the margin.


Q. --> S. Sanfilippo: did you already make one coil consisting of four layers?

Answer S. Sheehy: No, unfortunately no coil could be built to check that the expectations are fulfilled.


Q. --> G. Franchetti: Do you expect some problem with the accuracy of the angle (of the helical coils)?

Answer S. Sheehy: Error studies were performed on how well the different field components can be achieved when errors in the construction and the coil winding; the magnets are really insensitive to small construction errors, but there might be some systematic errors


Q. --> B. Pine: In the machine, is there space for corrector magnets? And can they correct for the magnetic end field errors of the helical coils?

Answer S. Sheehy: There is space for the corrector magnets left; by changing the parameters of the helical coils after building a first prototype it should be possible to correct the dynamics.


3) Speaker Martin Droba “Beam dynamics in the Figure-8 magnetic field”


Summary of Droba talk written by scientific secretaries


- Project: energy 150 keV per nucleon, 3d fields are considered including guiding fields -> interested in longitudinal magnetic fields

- As longitudinal fields do not compensate for the drift, a figure 8 magnets has to be used

- Magnet consists not only of longitudinal fields, but also some twists that are complicated for beam dynamics

- In real figure 8 magnets, correction coils have to be implemented

- Looking at the beam dynamics during many turns a surface around the beam is generated

- With two counter propagating beams there are regions where the orbits are separated and regions with collisions

- Scaled experiments at room temperature to proof simulations of the injection


Questions and Answers

1. Questioner: Ingo Hofmann (TU Darmstadt / GSI Helmholtzzentrum für Schwerionenforschung GmbH (GSI)):

"My Question is this. You have shown simulations of single particles. Now in all fusion machines there are collisions between the particles and with your low energy you may also have collisions but then these invariance, which come along with the magnetic surfaces, they will break. Is there a problem for you?"

Answer: Martin Droba (IAP, Goethe University Frankfurt)

"This could be a problem because the beam will be a breeding (producing secondaries) beam. But it depends on in which timescale we are thinking. So this problem we can do also in our simulation, like when I am changing this magnetic momentum for example by kicking and do simulations and then we will see if there is an issue or not.


2. Questioner: Ben Pine (STFC):

"What's it for? It is a really interesting machine. You have two counter wise ion beams. So what happens?"

Answer: Martin Droba (IAP, Goethe University Frankfurt)

"The origin idea is to do proton boron fusion, proton on boron. You are then producing three alphas delivering the energy. They are also in the stellarator. They should also be confined and then you have no neutrons and this is very good for a University"


3.Questioner: Frank Zimmermann (CERN):

"How do you use the energy of the ions?"

Answer: Martin Droba (IAP, Goethe University Frankfurt)

"At first it is only for studying, let me say to measure the cross sections or for example are there some differences between a thermal plasma and the beam. And colliding"


-Q: P. Spiller: Are you approaching the fusion earlier than ITER?

Answer: No, but in 2014 Wendelstein will be finished and start operation...



Session “Joint Discussion and Workshop Summary” chaired by P. Spiller, S. Russenschuck

Scientific secretaries: Christine Claessens, Manuel Heilmann


Wednesday 3.12.2013


Discussion subjects:

- Is there a demand for another workshop on this topic and when and in what frame should it take place?


- Magnet fabrication: Steel quality is important and has to be carefully watched. Best to fabricate it yourself but often not possible.

MedAustron made good experiences with steel from industry. Others have made very bad experiences. Sorting can compensate steel quality related errors. Material aging is an issue. Magnets should not be stored too long.


- Corrector magnets are important but often there is little space and little money. Corrector system has to be lay out from the beginning. Can depend of the size of the facilities available.


- Simulations cannot completely replace prototyping and measurements. CERN made a lot of prototypes but others may not have the money.


- Magnet quality has to be specified. As good as possible is not a good requirement. Beam dynamics people have to work with the achievements of the Magnet people. Magnet production is a collaborative work and needs a lot of exchange.


-J-PARC experience is interesting: Simulations could not predict the necessary corrector magnets. Magnets had to be adapted during operation. Simulations cannot reproduce everything.


- Financial situation is tight. Therefore few prototypes are produced and money for correctors is short.


- Early error studies are helpful but not always possible to make correct predictions with simulations. Even though correction tools are available, machine details are unknown and necessary tools not predictable.


- Modeling possibilities should be promoted.


- Something unexpected happens always.


- Young accelerator physicists should learn about the problems in formal trainings.  At conferences everything sounds perfect.


- Administrative work increases. Standard processes have to be implemented and followed strictly to guaranty quality but also to avoid responsibility. Interaction with companies that build magnets should be close so they get feedbacks and can improve their work.


- Measurements are essential, but what accuracy is really necessary.


- One should be careful to not let politics and administration discourage people. Motivated employees/colleagues are important and work better. Companies can be motivated by allowing them to advertise with their work for the scientific community.