13
FETS RFQ Beam Dynamics Simulations for RFQSIM, CST and Comsol Field Maps Simon Jolly 2 nd June 2010

FETS RFQ Beam Dynamics Simulations for RFQSIM, CST and Comsol Field Maps Simon Jolly 2 nd June 2010

Embed Size (px)

Citation preview

Page 1: FETS RFQ Beam Dynamics Simulations for RFQSIM, CST and Comsol Field Maps Simon Jolly 2 nd June 2010

FETS RFQ Beam Dynamics Simulations for RFQSIM, CST and Comsol Field Maps

Simon Jolly2nd June 2010

Page 2: FETS RFQ Beam Dynamics Simulations for RFQSIM, CST and Comsol Field Maps Simon Jolly 2 nd June 2010

RFQ Field Map Comparisons

• We now have 3 consistent methods of producing field maps for the RFQ:– RFQSIM (from coefficients).– CST (optimised field from August 2009).– Comsol (just finished in time for IPAC’10).

• IPAC’10 RFQ beam dynamics paper: “Integrated Design Method And Beam Dynamics Simulations For The Fets Radiofrequency Quadrupole”

• Needed to compare beam dynamics simulations for all 3 field mapping methods.

• Simulations in GPT to compare CST and Comsol with 2 types of RFQSIM field (approximated and full).

• Vary current between 0-120mA and measure transmission and final energy spread.

02/06/10 Simon Jolly, Imperial College 2

Page 3: FETS RFQ Beam Dynamics Simulations for RFQSIM, CST and Comsol Field Maps Simon Jolly 2 nd June 2010

Input Conditions• Current variation from 0-120mA; bunch length: 1RF

period; SCtree3D space charge simulates bunched beam.• Using same input distribution as for previous

publications: – xmax = ymax = 2.2mm. – x’max = y’max = 90mrad. rms = rms = 0.25 mm mrad.

02/06/10 Simon Jolly, Imperial College 3

Page 4: FETS RFQ Beam Dynamics Simulations for RFQSIM, CST and Comsol Field Maps Simon Jolly 2 nd June 2010

Field Map Differences

• Differences between RFQSIM approximated and full fields at 5% level:– Smooth variation of coefficients between

cells.– Full Bessel functions rather than truncated

series.• CST uses maximum mesh density (4,700

points) with 6 RFQ sections (matching section, 2x500mm, 3x1m).

• Comsol uses same vane model but not yet using tangential boundaries.

• All field maps use 0.5mm point spacing (RFQSIM field maps match CST and Comsol, but different from previous simulations).

02/06/10 Simon Jolly, Imperial College 4

Page 5: FETS RFQ Beam Dynamics Simulations for RFQSIM, CST and Comsol Field Maps Simon Jolly 2 nd June 2010

Results: RFQSIM 2008

02/06/10 Simon Jolly, Imperial College 5

Transmission Energy Spread (60mA)

Page 6: FETS RFQ Beam Dynamics Simulations for RFQSIM, CST and Comsol Field Maps Simon Jolly 2 nd June 2010

Results: RFQSIM 2010 (Simple)

02/06/10 Simon Jolly, Imperial College 6

Transmission Energy Spread (60mA)

Page 7: FETS RFQ Beam Dynamics Simulations for RFQSIM, CST and Comsol Field Maps Simon Jolly 2 nd June 2010

Results: RFQSIM 2010 (Full)

02/06/10 Simon Jolly, Imperial College 7

Transmission Energy Spread (60mA)

Page 8: FETS RFQ Beam Dynamics Simulations for RFQSIM, CST and Comsol Field Maps Simon Jolly 2 nd June 2010

Results: CST

02/06/10 Simon Jolly, Imperial College 8

Transmission Energy Spread (60mA)

Page 9: FETS RFQ Beam Dynamics Simulations for RFQSIM, CST and Comsol Field Maps Simon Jolly 2 nd June 2010

Results: CST

02/06/10 Simon Jolly, Imperial College 9

Transmission Energy Spread (60mA)

Page 10: FETS RFQ Beam Dynamics Simulations for RFQSIM, CST and Comsol Field Maps Simon Jolly 2 nd June 2010

Preliminary Conclusions

• RFQSIM simple field gives very similar results to old simulations: this is good! Differences probably due to GPT interpolation since point spacing is different.

• RFQSIM simple and full fields also give very similar transmission results:– 92% transmission at 60 mA for both.– Full field expansion gives slightly higher final

energy.

• CST and Comsol also give very similar results, both for transmission and energy. But…

02/06/10 Simon Jolly, Imperial College 10

Page 11: FETS RFQ Beam Dynamics Simulations for RFQSIM, CST and Comsol Field Maps Simon Jolly 2 nd June 2010

Poor CAD Model Transmission

• CST and Comsol give significantly poorer transmission than RFQSIM for higher beam currents.

• Why the difference? Poor meshing or real RFQ properties?

• Try increasing the field strength by up to 30% to see if we can recover transmision…

02/06/10 Simon Jolly, Imperial College 11

Page 12: FETS RFQ Beam Dynamics Simulations for RFQSIM, CST and Comsol Field Maps Simon Jolly 2 nd June 2010

Increased Field Strength

02/06/10 Simon Jolly, Imperial College 12

• 10% increase in field strength recovers transmission: we’re back in business!

• Is such an increase feasible in reality? Does it compare to known RFQ’s?

• Not yet sure of the origin of this difference: might be mesh-based, might be real.

Page 13: FETS RFQ Beam Dynamics Simulations for RFQSIM, CST and Comsol Field Maps Simon Jolly 2 nd June 2010

Conclusions• Lots of results that make sense (a turn-up for the books!):

– Simple and Full RFQSIM field maps show virtually no difference: slightly better RF capture and acceleration from Full field gives higher final energy.

– CST and Comsol give very similar results: looks like we’re producing the same map through the same method.

• Clear differences between CAD-based methods and RFQSIM:– Field strength nominally correct, since no extra transverse

losses.– Longitudinal fields give problems: poor RF capture and

acceleration.– But we can recover transmission by increasing the field

strength: maybe field is closer to reality?– Longitudinal vane curvature certainly more subtle than

transverse: need better mesh longitudinally.• Perhaps try a single Comsol simulation with very high mesh

density (200mm RFQ sections) and see if transmission improves at 60mA.

• Should we start optimising on “realistic” beam?

02/06/10 Simon Jolly, Imperial College 13