Compensation failures proceed

[Bruce Yee]

Dear Didier,

Follow the papers of J.Biarrotte ([1] Phys. Rev. ST Accel. Beams 11, 072803 (2008)) and [2]BEAM DYNAMICS STUDIES FOR THE FAULT TOLERANCE
ASSESSMENT OF THE PDS-XADS LINAC DESIGN, in Proceedings of the EPAC 2004, 2004., the first reference appears in TraceWin Manual)

I want to confirm the proceeding for the Compensation failure.

If I understand correctly, the local compensation can be studied in two steps:

1) Steady-state (behavior before the failure and after the compensation)
2) Transient behavior.

An example of steady-state is the "cavity_failure_recovery" in the example folder.
And for the transient behavior is the "RFcavity_transient" in the example folder.

The transient behavior gives you the time necessary to apply for the compensation before a certain limit of beam loss. right?

For the total of beam loss in the transient study, Do I need to simulate the full lattice?

Finally, Can TraceWin simulate transient analysis for magnets?

Thanks for the help and support,

Bruce

[Bruce Yee]

Dear Didier,

About the question "For the total of beam loss in the transient study, Do I need to simulate the full lattice?".
After re-read well the paper, it is clear that I need to make the full lattice simulation.


Best regards,

Bruce

[Didier]

Dear Bruce,

1) Steady-state (behavior before the failure and after the compensation)
2) Transient behavior.

Yes

The transient behavior gives you the time necessary to apply for the compensation before a certain limit of beam loss. right?

Yes

The transient behavior gives you the time necessary to apply for the compensation before a certain limit of beam loss. right?

This was our approach, but you can look at many other parameters. The emittance for example can also be a limit setting this time limit.

For the total of beam loss in the transient study, Do I need to simulate the full lattice?

Yes, otherwise how to see the losses !

Finally, Can TraceWin simulate transient analysis for magnets?

Nothing has been impleted concerning magnets, but why not in the future

Regards,

Didier

[Bruce Yee]

Dear Didier,

Thank you very much for your confirmation. I am very interested to implement the transient effect for the magnets.
Is any way to help or collaborate for that goal?

Best regards,

Bruce

[Didier]

Dear Bruce,

Developing something similar with magnets should not be as complicated as with cavities. If you have a numerical model showing the behavior of a quad turning off or starting, I could try to implement it. The question is more, what exactly do you want to simulate.

Regards,

Didier

[Bruce Yee]

Dear Didier,

Thank you very much. I want to simulate how the magnetic field decay inside the magnet and estimate its effect in the beam.

Changing of topic. I would like some help to apply for the cavity compensation when I simulated the transient effect of an RF cavity.
So far, I combined the two examples "cavity_failure_recovery" and "RFcavity_transient" (I attached some pictures).

The error starts at 1ms and "applied correction " 1.5 ms after the failures (2.5ms)

However, I did not succeed. Could you give me some advice?

Best regards,

Bruce

[Didier]

Dear Bruce,

Next please send me you project files, because I'm not sure to was able to rebuilt exactly what you did. Anyway, secondly I don' know also what meaning "yo didn't succeed".
So I made simulation without any correction and I have the following result concerning the output energy according to time.

Image1.png (12.28 KiB) Viewed 3666 times

Now I add correction like you starting at 2.5 ms and I get following result

Image2.png (13.27 KiB) Viewed 3666 times

That seems better but not perfect, probably 3 cavities not enougth to compensate correctly the failure, so, I incease the field marging, increasing the "MargeField" parameter in "spoke_25MV_nominal.cav" file from 0.7 (70%) to 4 and I get new result better.

Image3.png (13.23 KiB) Viewed 3666 times

Now, I don't know if from your point of view this is a success. It's probably not the right correction scheme, more cavities should maybe be used, but I do see an improvement.

Regards,

Didier

[Bruce Yee]

Dear Didier,

Thanks for your answers.

Apologies, when I mean did not succeed, I mean we did not get the full compensation. However, you are right the energy increased.

Example.png (80.53 KiB) Viewed 3665 times

I implemented to my work

Example2.PNG (60.94 KiB) Viewed 3665 times

However, the compensation was not achieved, even when the case is compensated for the so-called "steady-state".
Maybe is as you suggested, I should increase the "MargeField" parameter.
Let me try that, If I have some problems I will send you the files.

Best regards,

Bruce

[Bruce Yee]

Dear Didier,


Thank you very much for your help. Follows the previous discussion. My goal is to develop a correction scheme in the stead-state and tested if the results are compatible for the same condition at a given time in the transient-state.

For example, using the example file "RF cavity transient", I tried to implement a correction scheme similar to "cavity_failure_recovery".

For the "RF cavity transient", I reproduced the same failure conditions at 8 ms (phase ~-160 deg and same nominal voltage), as is shown in the next figure

fig1.png (329.21 KiB) Viewed 3643 times

.

Then, I simulated the failure in the steady-state to develop a compensation scheme as is shown in the next figure.

fig2.png (620.23 KiB) Viewed 3643 times

.

The compensation produced higher energy than baseline.

Then, I went back and added the element to make the compensation in Transient studies. I will implement the correction 2.5 ms after the beginning of the simulations (1.5 ms later than the failure). See figure

fig3.png (274.23 KiB) Viewed 3643 times

.

Something interesting is that in this case, the scheme produces an energy compensation lower than the baseline. This can be explained by the "Margefield" in the cavity parameters files. (By that way this Margefield parameter is the available field that can be used for compensation, you mentioned the last reply that can be used a value of 4, but this means that the nominal voltage of the cavity is multiplied by 4? Could you explain a little bit more about it? And how this value could be compared with the 5th parameter of the command "ADJUST 20 -3 0 0 3.5 0.1" used to adjust the cavity fields )

I looked in the time of 8ms, to see if the results are similar to the compensation scheme in steady-state (Fig2)

fig5.png (666.94 KiB) Viewed 3643 times

The cavities did not get as much voltage as the steady-state, even, I used the same command to adjust (perhaps it only is adjusted by the values of the cavity parameters files). However, the most interesting point is the synchronous phase for the first cavities does not change (Why?). For the steady-state all the phases of the cavities used for the matching changed. Indeed, it seems that the synchronous phase of the other cavities did not do the same. Maybe you can give some hints.
I attached the input files.

Thanks for the help and support,

Bruce

[Didier]

Dear bruce,

A lot of things. I tried to answer some of them, expected that solve the others
First your steady-state ddin't reach the right energy because, your number of optimization step is two low, increase to 500, the "Maximum number of iteration" in "Matching" page -> "Second step with error".
For the transient studies, decrease htis number to 10 ou 20, otherwise, that seem become instable and I don't understand why for the moment, I was still looking about that point.

Yes, 4 means 400% of nominal voltage marging and the 5th parameter of the command ADJUST is in transient mode not an absolute limit but a variation limit for each step of optimisation. In other words don't use it to control the maximum voltage. The parameter form cavity file is the good one to control.

So using '20' for the number of optimisation and 1.7 and MargeField=1.7 I got following figures for output energy and phase synchrone of first cavity. As yo can see, on my side, it changed. I've also changed a little bit the trransient time parameters to accelerate computation.

transient.png (12.05 KiB) Viewed 3639 times

One last point, I'm sorry if I've lost the understanding of this whole process, but these studies and developments have been done in 2007 and I've never come back on them and I don't remember all the details anymore.

Regards,

Didier

output_energy.png (16.38 KiB) Viewed 3639 times

sync_phase_first_cav.png (17.61 KiB) Viewed 3639 times