beam-ion effect

[tiansk]

Hi
Recently, I am trying to make a C++ programme for for beam-ion effect simulation. As the first step, I actually follow the approaches used in Elegant. The main steps can be descried as follows (take the weak-strong mode for example). The index k is used to specify the interaction point and the index j is used to specify the bunch index


1) latticeInterActionPoint.ionLineDensity[k] = corssSectionEI * latticeInterActionPoint.vacuumPressure[k]
/latticeInterActionPoint.temperature[k]/Boltzmann * beamVec[j].electronNumPerBunch;

2) latticeInterActionPoint.ionNumber[k] = latticeInterActionPoint.ionLineDensity[k] *
latticeInterActionPoint.interactionLength[k];

3) beamVec[j].WSRMSCal(latticeInterActionPoint,k);

4) latticeInterActionPoint.IonGenerator(beamVec[j].rmsRx,beamVec[j].rmsRy,beamVec[j].xAver,beamVec[j].yAver,k);

5) latticeInterActionPoint.IonsUpdate(k); //to get the accumulated ions which affect the beam thereafter
6) latticeInterActionPoint.IonRMSCal(k);

7) beamVec[j].WSIonBunchInteraction(latticeInterActionPoint,k);

8) latticeInterActionPoint.IonTransferDueToBunch(beamVec[j].bunchGap,k);

10) beamVec[j].BunchTransferDueToIon(latticeInterActionPoint,k);

The Bastine complex error function is used to get the space charge force between ions and beam in this weak-strong simulation.

As to the "strong-strong" simulation, two times of "weak-strong" is used to get the forces of ions and the bunch particles. Since this process is still not self-consistent, I would rather call it as a "quasi-strong-strong" model.

The source files of the developed code can be found with the following links
https://github.com/ChaoLiIHEP/FASTION


Sorry for such a long introduction, let go into the the question part.

The attached files shows 3 sub figures, the a) and b) are the beam centre oscillation as functions of turns with and w/o synchrotron radiation damping respectively, the weak-strong model is applied; c) is obtained with "quasi-strong-strong" simulation and the synchrotron radiation is turns on. Other condition as filling pattern, beam current, and lattice are the same.



As seen, I am surprised that the "strong-strong" does not give me any bunch oscillation information. Compared with the weak-strong simulation, almost nothing useful is given.

Since I am really new in this field and do not have too much experience on the beam-ion interaction simulation, I am really wandering what would be the case for the "strong-strong" results? Should be similar to the subfig b), but with a much smaller oscillation amplitude ?

Sorry for bother you too much, since right now, I did no know how to benchmark it, helpfully, we can discuss more in the coming days and more details can be explore.

Best regards
yours Chao and Saike Tian

[tiansk]

Hi,

Have spend a lot of time to debug the code (during the last several weeks), and suddenly I have realized the problem in the beam-ion effect simulation with the "quesi-strong-strong' model.

It is discussed by several researchers that the ions distribution will not with a Gaussian type, where as in the original model in simulation, we still assume that the ions follow the Gaussian type, and the rms values of the ions are used in Bastine formula to get the E force from ion to electron particles. Usually the accumulated ions will not be lost till reach the beam pipe, where the pipe size is much larger that the rms beam size. In this case, the rms size of ion distribution is also large, which gives very tiny effect on electron particles.

At this moment, I have made some test code running, where the ion loss criteria is changed to several rms beam size (in my simulation it is chosen as 4 rms beam size, beyond which the ions are erased). The results of the "quasi-strong-strong" model right now give bunch oscillation information.


I think I am getting a feeling why new fitting process of ion density is now adopted in elegant. However still since the non-self-consistent process, I am worrying that the, even with a better fitting curves, criticism of the non-self-consisten will be still there. It probably have to go further to a ion "Poisson solver", like PIC approach.

If you have time, please make comment on my understanding discussed above. Will help a lot to get a clear physical picture.

By the way, I am also very curious on the method you used for beam-ion effect code benchmark, would you mind let me know your approaches, that will help a lot.

yours sincerely Chao

[michael_borland]

Chao,

Sorry for not responding to your earlier message. The strong-strong simulation is very challenging and it is fair to say that we are not satisfied with the results we are getting in elegant using the new fitting-based methods. In addition to being quite slow, they seem to predict instability when we don't observe an instability in the actual machine. Noise is suspected, but it is hard to suppress even when the number of ions is very large.

Our concern about a poisson solver is that it will have similar noise issues. My colleague Ryan Lindberg points out that the bi/tri-gaussian/lorentzian fitting functions we are using should filter noise fairly effectively, whereas a Poisson solver might not be as good unless the grid is sparse.

We are gearing up for some ion instability experiments in APS in the next few months. We hope to inject Argon gas into a short section of chamber and will use the experimental results for code benchmarking.

--Michael

[chaoli]

Dear Michael,

Great thanks for your kindly reply. It is very helpful.

As you mentioned:
"In addition to being quite slow, they seem to predict instability when we don't observe an instability in the actual machine"

It is somehow a mysterious. The beam ion effect that we can predict clearly in simulation but can not find it in experiment.
It is probably some other mechanims those can suppress the beam-ion effect a little bit.

There would be another points that I am still a little bit confusing, would you mind guide me a little bit on the question:
how will the beam and accumulated ion will perform when the beam-ion and beam-wakefield interactions are considered
simuteanously?


It will be very interesting for me to follow the results of your coming experiments. Would be a very important result if
gaps between models used in simualation and experiment can be filled.

Thanks again.
Yours sincerely Chao

[michael_borland]

Chao,

My colleague Joe Calvey has been studying the question of how ion instabilities are affected by things like chromaticity, wakefields, multiple-ionization, and other effects, but has yet to publish his conclusions. Our gas injection experiment is going forward, so we expect publication in early 2020.

--Michael