Compensation failures proceed
Re: Compensation failures proceed
Dear Bruce,
About your first remark, to clarify...
Considering your case:
End time(s) : 0.01s
Beam dynamics : 1e5s
> Transient simulation time = 1000 steadystate one.
if you include correction:
start time: 0.0025s
step : 0.0001 s
> 25 corrections
So you have to add to Transient simulation time: 25 x number_of_iteration x steadystate time
Now, you have also to add the transient cavity simulation (integration step = 1e7) : 100000 step.
The general scheme is as follows
Your doubts :
(1) Yes it's the same except a very important point. When you adjust the RF phase of field of the cavity, it reacts with its filling time and the following cavity is influenced by the resulting beam variation + its own adjustments. It is therefore very complex to how the whole system will react.In fact I'm not sure that this is how it should be done in reality, but when these developments were made, I didn't have time to develop a specific procedure so I used the one I had at hand, the steady state one.
A last more general remark, at the end of this study made for MYRRHJA, we all considered, despite the encouraging results, that it was better to stop the machine for a few seconds to finally restart it with the steadystate results.
Indeed, it seemed too complex to implement on a real machine and above all it did not solve the problem of longitudinal acceptance. Indeed, it is not enough to find the right phase and energy of the beam by adjusting cavities, it is also necessary to do it while minimizing the reduction of the acceptance.
(2) Below, what I observe with your example, I do not find what you indicate
(3) I think it's impossible
Regards,
Didier
About your first remark, to clarify...
Considering your case:
End time(s) : 0.01s
Beam dynamics : 1e5s
> Transient simulation time = 1000 steadystate one.
if you include correction:
start time: 0.0025s
step : 0.0001 s
> 25 corrections
So you have to add to Transient simulation time: 25 x number_of_iteration x steadystate time
Now, you have also to add the transient cavity simulation (integration step = 1e7) : 100000 step.
The general scheme is as follows
Your doubts :
(1) Yes it's the same except a very important point. When you adjust the RF phase of field of the cavity, it reacts with its filling time and the following cavity is influenced by the resulting beam variation + its own adjustments. It is therefore very complex to how the whole system will react.In fact I'm not sure that this is how it should be done in reality, but when these developments were made, I didn't have time to develop a specific procedure so I used the one I had at hand, the steady state one.
A last more general remark, at the end of this study made for MYRRHJA, we all considered, despite the encouraging results, that it was better to stop the machine for a few seconds to finally restart it with the steadystate results.
Indeed, it seemed too complex to implement on a real machine and above all it did not solve the problem of longitudinal acceptance. Indeed, it is not enough to find the right phase and energy of the beam by adjusting cavities, it is also necessary to do it while minimizing the reduction of the acceptance.
(2) Below, what I observe with your example, I do not find what you indicate
(3) I think it's impossible
Regards,
Didier
Re: Compensation failures proceed
Dear Didier,
Thank you very much for your patient and time to explain me in detail.
I was thinking that "step" parameters in the correction part, means after the 'start_time" value, the correction is implemented in every "step" time.
I said so, because for our case. After 2.5 ms, every 0.1 ms, the program make a small pause and it seems that is computed the adjustments.
About the filling time, that's the reason that I start to apply the correction after 2.5 ms, because the phase and voltage present more less a constant behavior (a kind of steadystate). By that way, Are the corrections is done with respect to the last "adjusted" configuration or with respect to the original settings?
About the MYRRHA studies, I was thinking that the faulttolerance scheme was a kind of "temporally" solution in case of a cavity fails, right?
Or the pretend to operate "normally" with that scheme.
Finally, about the simulations different, I was using a shorter time, perhaps is that, but I run a longer simulations and I got the same results. However, my point was when I got 1000 interactions ( )
The results looks worst.
Anyway, thanks a lot for the discussion.
Best regards,
Bruce
Thank you very much for your patient and time to explain me in detail.
I was thinking that "step" parameters in the correction part, means after the 'start_time" value, the correction is implemented in every "step" time.
I said so, because for our case. After 2.5 ms, every 0.1 ms, the program make a small pause and it seems that is computed the adjustments.
About the filling time, that's the reason that I start to apply the correction after 2.5 ms, because the phase and voltage present more less a constant behavior (a kind of steadystate). By that way, Are the corrections is done with respect to the last "adjusted" configuration or with respect to the original settings?
About the MYRRHA studies, I was thinking that the faulttolerance scheme was a kind of "temporally" solution in case of a cavity fails, right?
Or the pretend to operate "normally" with that scheme.
Finally, about the simulations different, I was using a shorter time, perhaps is that, but I run a longer simulations and I got the same results. However, my point was when I got 1000 interactions ( )
The results looks worst.
Anyway, thanks a lot for the discussion.
Best regards,
Bruce
Re: Compensation failures proceed
Dear Bruce,
Otherwise, I come back a little to what I was saying, the results of the steady state are finally relevant. I tried to put them directly in algorithm and that works. The problem obviously comes from the correction calculations which do not find the steady state as they should. In particular, I noticed that the power limit imposed on the cavity (30kW) was one of the causes. In fact, I think the incapability of the cavity to reach the new field or phase make instability of the ssytem. But that said, by increasing the power to 200kW with 1000 iterations I have that good following result: Using that:
Regards,
Didier
Totally agree, that's what I said in the previous post!I was thinking that "step" parameters in the correction part, means after the 'start_time" value, the correction is implemented in every "step" time.
I said so, because for our case. After 2.5 ms, every 0.1 ms, the program make a small pause and it seems that is computed the adjustments.
Yes I agree, but when you make a correction, it is not applied immediately for the same reasons.*/About the filling time, that's the reason that I start to apply the correction after 2.5 ms, because the phase and voltage present more less a constant behavior (a kind of steadystate). By that way, Are the corrections is done with respect to the last "adjusted" configuration or with respect to the original settings?
Otherwise, I come back a little to what I was saying, the results of the steady state are finally relevant. I tried to put them directly in algorithm and that works. The problem obviously comes from the correction calculations which do not find the steady state as they should. In particular, I noticed that the power limit imposed on the cavity (30kW) was one of the causes. In fact, I think the incapability of the cavity to reach the new field or phase make instability of the ssytem. But that said, by increasing the power to 200kW with 1000 iterations I have that good following result: Using that:
Regards,
Didier
Re: Compensation failures proceed
Dear Didier,
Sorry I forgot to submit my reply. Thanks for the explanation.
Best regards,
Bruce
Sorry I forgot to submit my reply. Thanks for the explanation.
Best regards,
Bruce
Re: Compensation failures proceed
Dear Didier,
Do you have an example for Magnet compensation?
In the chart file, the "Gradient" button only plot the gradient for the linac. I am wondering if is possible to also plot the gradient for the beam.
I said so because in analogy with the "Synch.phase" button in the Cavity failures, It would be useful see how the gradient dropped to zero in the magnets failures.
One last thing, in the file "Supra_2" from the Examples folder, does not run. It said that you can not used the options "Matching with Family & Twiss commands" and "Match using Diagnostic" at the same time.
Thanks for your help and time.
Best regards,
Bruce
Do you have an example for Magnet compensation?
In the chart file, the "Gradient" button only plot the gradient for the linac. I am wondering if is possible to also plot the gradient for the beam.
I said so because in analogy with the "Synch.phase" button in the Cavity failures, It would be useful see how the gradient dropped to zero in the magnets failures.
One last thing, in the file "Supra_2" from the Examples folder, does not run. It said that you can not used the options "Matching with Family & Twiss commands" and "Match using Diagnostic" at the same time.
Thanks for your help and time.
Best regards,
Bruce
Re: Compensation failures proceed
Dear Bruce,
There is no transient calculation for gradients in quadrupoles so far, so I don't quite understand what you call gradient for the beam
I fixed the problem about the combination of Matching with Family & Twiss commands" and "Match using Diagnostic
Regards,
Didier
There is no transient calculation for gradients in quadrupoles so far, so I don't quite understand what you call gradient for the beam
I fixed the problem about the combination of Matching with Family & Twiss commands" and "Match using Diagnostic
Regards,
Didier
Re: Compensation failures proceed
Dear Didier,
Thanks for your information.
About the gradient, I mean in the steadystate case. For example if you see the next figure From the chart window you can plot the "accelerating gradient" for the "Linac", red curve, and for the "Beam ", blue curve.
In this example, you can see the accelerating gradient dropped for the last cavities.
However, if there is a failure in the magnets, the gradient is set to zero, only one curve appears for the gradient (red, I guess is for the "Linac" case ?).
Changing of topic, I found something peculiar for the case of magnet failure (gradient is dropped to zero). You can see the case in the next figure. The accelerating field and the synchronous phase of the cavities downstream of the faulty magnet dropped to zero(top center, right plots). If I understand well, after the faulty cavity these plots said not energy gain. However, the energy plot said that beam model has almost the same energy as the linac model.
Indeed, see the next figure By looking the density distribution the longitudinal plane looks unaffected. The model ends with almost the same energy as the ideal case.
Thus, What does mean the zero synchronous phase and Eacc?
I attached the files, if you see in more detail.
Thanks for your help and support,
Bruce
Thanks for your information.
About the gradient, I mean in the steadystate case. For example if you see the next figure From the chart window you can plot the "accelerating gradient" for the "Linac", red curve, and for the "Beam ", blue curve.
In this example, you can see the accelerating gradient dropped for the last cavities.
However, if there is a failure in the magnets, the gradient is set to zero, only one curve appears for the gradient (red, I guess is for the "Linac" case ?).
Changing of topic, I found something peculiar for the case of magnet failure (gradient is dropped to zero). You can see the case in the next figure. The accelerating field and the synchronous phase of the cavities downstream of the faulty magnet dropped to zero(top center, right plots). If I understand well, after the faulty cavity these plots said not energy gain. However, the energy plot said that beam model has almost the same energy as the linac model.
Indeed, see the next figure By looking the density distribution the longitudinal plane looks unaffected. The model ends with almost the same energy as the ideal case.
Thus, What does mean the zero synchronous phase and Eacc?
I attached the files, if you see in more detail.
Thanks for your help and support,
Bruce
 Attachments

 LastCryMag.zip
 (10.81 KiB) Downloaded 307 times
Re: Compensation failures proceed
Dear Bruce,
I fixed error in envelope simulation when 100% error is applied to quad. I fixed also a displaying problem about field map aperture.
I added a new curve to the magnetic gradient to have a similar behavior to the electrical gradient, but be careful, it is not quite the same thing. For the longitudinal aspect the beam curves are calculated using the beam as a probe. For quads it is much more trivial and obtained by using the applied errors read. Probably it is transparent most of the time, but keep it in mind.
Regards,
Didier
I fixed error in envelope simulation when 100% error is applied to quad. I fixed also a displaying problem about field map aperture.
I added a new curve to the magnetic gradient to have a similar behavior to the electrical gradient, but be careful, it is not quite the same thing. For the longitudinal aspect the beam curves are calculated using the beam as a probe. For quads it is much more trivial and obtained by using the applied errors read. Probably it is transparent most of the time, but keep it in mind.
Regards,
Didier
Re: Compensation failures proceed
Dear Didier,
Thank you very much for your help. I will keep on mind your comment. If I find something interesting, I will inform you.
Best regards,
Bruce
Thank you very much for your help. I will keep on mind your comment. If I find something interesting, I will inform you.
Best regards,
Bruce
Re: Compensation failures proceed
Dear Didier,
I am simulating errors in the superconducting solenoids ((Magnetic field dropped to zero)).
The envelope simulations run normally, but all the particles are lost downstream that faulty solenoid, when I use a beam distribution. As you can see, the left plot is the envelope and the right one is the beam distribution.
I checked the beam distribution in the 3 planes, However, the beam is well contained in the apertures; thus, I did not understand why all the particle got lost.
I attached the input files .
Finally, I have a question about how to compute the Abs phase present in the chart.
Also, how do you compute the RF input field phase of the Field maps ?
Thanks for your help and support,
Bruce
I am simulating errors in the superconducting solenoids ((Magnetic field dropped to zero)).
The envelope simulations run normally, but all the particles are lost downstream that faulty solenoid, when I use a beam distribution. As you can see, the left plot is the envelope and the right one is the beam distribution.
I checked the beam distribution in the 3 planes, However, the beam is well contained in the apertures; thus, I did not understand why all the particle got lost.
I attached the input files .
Finally, I have a question about how to compute the Abs phase present in the chart.
Also, how do you compute the RF input field phase of the Field maps ?
Thanks for your help and support,
Bruce
 Attachments

 Abs. Phase.PNG (23.83 KiB) Viewed 4829 times