19/11/2004 1 ITPA-2004-11 Coster Code-Code Code-Code comparisons D. P. Coster*, X. Bonnin** *Max Planck Institute for Plasma Physics, EURATOM Association, Garching, Germany ** Max Planck Institute for Plasma Physics, EURATOM Association, Greifswald, Germany; CEA- France; Paris-XIII
Goal Three major edge codes in wide-spread use, and in use for JET discharges Two of these with developer knowledge at JET To compare the edge codes for a series of cases with increasing physics complexity, to understand the origin of the differences that arise, and to validate the codes Started with EDGE2D-NIMBUS, SOLPS (B2-EIRENE) and UEDGE Lack of UEDGE expertise at JET caused UEDGE to drop out (welcome to join back, do need developer level expertise to be available, though) Progression D, no drifts [completed] D, drifts [started] D+C, no drifts [next] D+C, drifts [last]
SOLPS/EDGE2D-NIMBUS comparison, first results Starting point for the comparisons Pure D Same transport coefficients Same boundary conditions Same feedback gas puff to control separatrix density Strong differences 3 cases EDGE2D-NIMBUS SOLPS, fluid neutrals SOLPS, kinetic neutrals Suggested we had a lot of work ahead
Some observations Comparisons facilitated by Being able to operate the codes on the same grid Being to plot the results from both codes on the same graph, easily Did this by porting the JET TRAN file facility to SOLPS Not necessarily the best long term solution But could be done quickly Having Developer-level expertise for both codes available at the same site for extended periods of time Having sufficient computer resources available
SOLPS/EDGE2D-NIMBUS comparison, flux limiters First big difference traced to different assumptions about kinetic parallel electron and ion energy flux limiters SOLPS had been using flux limiters for both electrons and ions EDGE2D-NIMBUS had the electron flux limiters coded, but not used in these runs Having the same flux limiter choice in both codes brought the two closer together
SOLPS/EDGE2D-NIMBUS comparison, additional differences Also explained a feature that had been seen on the poloidal ion temperature profile --- eliminated when ion thermal flux limiter switched off Feature also decreased when the gas puff was switched from localised (SOLPS) to distributed (EDGE2D-NIMBUS) An additional difference in the starting cases was traced to M=1 vs M>=1 at target
SOLPS/EDGE2D-NIMBUS comparison, final 5e18 Final version of the pure D, no drifts comparison 5e18 upstream density Agreement is very good! Though differences remain Not thought worth spending too much more effort chasing the remaining differences Particularly as Some of the difference thought to come from neutrals The drift cases were calling
SOLPS/EDGE2D-NIMBUS comparison, final 1e19 Final version of the pure D, no drifts comparison 1e19 upstream density Slightly worse comparison than the lower density case Suspect this is due to different neutral treatment See next slide Agreement is still good
SOLPS/EDGE2D-NIMBUS comparison, final 1.5e19 Final version of the pure D, no drifts comparison 1.5e19 upstream density Starting to see differences that come (probably) from differences in the neutrals code (NIMBUS vs EIRENE) Is an effort to produce EDGE2D- NIMBUS To see some of the effects, SOLPS cases were done dropping the energy per ionisation
SOLPS/EDGE2D-NIMBUS comparison, drifts 5e18 With drifts 5e18 case PRELIMINARY!
SOLPS/EDGE2D-NIMBUS comparison, drifts 1e19 With drifts 1e19 case PRELIMINARY!
SOLPS/EDGE2D-NIMBUS comparison, drifts/poloidal velocities Poloidal profiles of the velocity for 5e18 and 1e19 cases Work has been on hold for some months JET not in operation, therefore no secondments Xavier shifting jobs, David busy with other things High priority for 2005
19/11/2004 13 ITPA-2004-11 Coster Code-Code MDSPLUS used to store SOLPS runs SOLPS [JET, IPP, ] New postprocessing routine b2md Communicates via TCP/IP solps-mdsplus. aug.ipp.mpg.de (MDSPLUS server) [Garching] Linux server 1GB RAM 2 cpus Using MR-AFS to provide storage 11062 shots 307 GB data In the Garching DMZ Analysis program (jetdsp, cview, radisplay, idl, matlab) [Fusion Lab] Any of the standard MDSPLUS experimental plotting programs can be used Data can be also be accessed via IDL, MATLAB
Web site contains MySQL index 19/11/2004 14 ITPA-2004-11 Coster Code-Code
Code and Experiment can be plotted with the same tool CIII emission from experiment and code plotted with cview 1 10 20 AUG data 8 10 19 AB:10817/+solps/diag/\top.AUG.DIA:CIII_4650@0.000000 AE:*/*/*/AA_raw & time=aa_time_raw & rhv=rhv+1.08@3.7488 AG:*/*/*/AC_raw & time=aa_time_raw & rhv=0.4+rhv@3.74885 6 10 19 4 10 19 2 10 19 0 cview-prof (gdc) v2.41 - User: dpc - Mon Oct 25 11:25:38 2004 1.00 1.10 1.20 1.30 s coord (m) 19/11/2004 15 ITPA-2004-11 Coster Code-Code