RFQ CAD Model Beam Dynamics Studies Simon Jolly 3 rd August 2011

The State Of Play… Comsol/Matlab code to create field maps from Inventor SAT-files is complete: –Inventor model now models arbitrary number of cells (up to 1000) and changes dynamically based on spreadsheet. –Code dynamically identifies end flanges and grounds them if they’re present. –For some reason, native Inventor files won’t import properly: end flanges are missing… Inventor models have been built for 6 models, featuring combinations of: –Standard and final FETS matching section. –With/without lead out section. –Full model with lead out and end flanges. 03/08/11Simon Jolly, Imperial College London2

CAD Models: Matching Sections 03/08/11Simon Jolly, Imperial College London3

CAD Models: Lead Out/End Flanges 03/08/11Simon Jolly, Imperial College London4

Comsol Meshing Import CAD model and select single quadrant: take advantage of RFQ symmetry. Optimum meshes different for different regions: –Vane tips: triangular (extremely fine auto). –“Inner Beam Box”: 2mm x 2mm, swept rectangular (0.25mm x 0.25mm x 32 slices). –“Outer Beam Box”: 10mm x 10mm, tetrahedral (extremely fine auto). –“Air Bag”: 15mm x 15mm, tetrahedral (normal auto). Model vanes as “terminals”: only interested in surface fields. If end flange is present, model as ground plane. 03/08/11Simon Jolly, Imperial College London5 Vane tips Air Bag Inner Beam Box Outer Beam Box

Nov’10 UKNF Results 03/08/11Simon Jolly, Imperial College London6

Scott’s Matching Section, No Lead- Out 03/08/11Simon Jolly, Imperial College London7

Scott’s Matching Section, Lead-Out 03/08/11Simon Jolly, Imperial College London8

Standard Matching Section, Lead-Out 03/08/11Simon Jolly, Imperial College London9

Scott’s Matching Section, End Flanges 03/08/11Simon Jolly, Imperial College London10

Standard Matching Section, End Flanges 03/08/11Simon Jolly, Imperial College London11

Preliminary Conclusions Well, some good, some very bad… Good: –No significant difference between beam transmission with/without lead-out section. –Scott’s matching section shows slightly better transmission than the standard! Not yet sure why… Bad: –Significant beam losses when end flanges are included! –This is not as bad as is seems: turns out I started the beam in the wrong place… Reran simulations using Alan’s setWBemittance function and correctly aligned field map. 03/08/11Simon Jolly, Imperial College London12

Scott’s Matching Section, Lead-Out 03/08/11Simon Jolly, Imperial College London13

Scott’s Matching Section, Lead-Out (New) 03/08/11Simon Jolly, Imperial College London14

Scott’s Matching Section, End Flanges 03/08/11Simon Jolly, Imperial College London15

Scott’s Matching Section, End Flanges (New) 03/08/11Simon Jolly, Imperial College London16

Standard Matching Section, Lead-Out 03/08/11Simon Jolly, Imperial College London17

Standard Matching Section, Lead-Out (New) 03/08/11Simon Jolly, Imperial College London18

Standard Matching Section, End Flanges 03/08/11Simon Jolly, Imperial College London19

Standard Matching Section, End Flanges (New) 03/08/11Simon Jolly, Imperial College London20

Scott’s Matching Section, End Flanges (New) 03/08/11Simon Jolly, Imperial College London21

Results Hurrah! Proper transmission for full field map using final matching section, lead out section and end flanges. Virtually no difference between with/without end flanges. Current seems to be lower for “Lead-Out” models: –> 92% for previous, 91.7% for newer. –Turns out using “setWBemittance” gives slightly bigger beam than 10,000 particle input file I was using before. –Obviously sensitive to input conditions! Also measured effect of 10 micron and 100 micron tolerance on “Lead-Out” model: –Fixes transverse parameters to nearest 10/100 microns in spreadsheet. –Can we set the machining tolerance? 03/08/11Simon Jolly, Imperial College London22

Standard Matching Section, Lead-Out (New) 03/08/11Simon Jolly, Imperial College London23

Lead-Out Section, 10 micron Tolerance 03/08/11Simon Jolly, Imperial College London24

Lead-Out Section, 100 micron Tolerance 03/08/11Simon Jolly, Imperial College London25

Conclusions Possible to generate “arbitrary” models in Inventor: takes a few minutes to update model (but need to test in 2012…). Nice to be able to generate models in Comsol very easily without any “coaxing”: full run takes 6 hours, but can model individual cells if only a few change. Lots of interesting results from simulations: –Virtually no difference in transmission or energy spread when we change model. –Slight increase in emittance: Input: eps_x = pi mm mrad; eps_y = pi mm mrad. Without end flanges or lead out section: eps_x = pi mm mrad; eps_y = pi mm mrad. With half cell but no lead out section: eps_x = pi mm mrad; eps_y = pi mm mrad. With lead out but no end flanges: eps_x = pi mm mrad; eps_y = pi mm mrad. With lead out but no end flanges: eps_x = pi mm mrad; eps_y = pi mm mrad. Looks like tolerance is less than 100 microns, but 10 microns is okay. Not yet sure how realistic the model is… 03/08/11Simon Jolly, Imperial College London26

For Next Time… Jürgen’s results show that the field leaks out into the end flange: need to start beam 1-2cm back from matching section to include these effects (should be small). Run beam backwards from matching section using 2D space charge and 60mA current, calculate trajectories and produce 3D bunch with correct londitudinal distribution that can be started at any point (use Matlab interpolation). Check acceptance for all models using zero beam current: not perfect but gives upper limit. Include “map3D_remove” GPT element and particle removal map using CAD model. No need for Stephen Brooks to repeat RFQ transmission simulations… 03/08/11Simon Jolly, Imperial College London27