Development of a multi-physic platform OLYMPE for - PowerPoint Presentation

1. Development of a multi-physic platform OLYMPE for fusion magnets design: progresses update and applicationsCHATS conference, 21st Sept. 2021L Zani1, F. Bonne2, C. Bourcier3, F. Dejoie1, P. Hertout1, C. Hoa2, B. Lacroix1, Q. Le Coz4, F. Nunio5, V. Petit1, A. Torre1, C. Van Wambeke3, S. Varin21 CEA-IRFM, Cadarache, France 2 CEA-IRIG, Grenoble, France 3 CEA-DM2S, Saclay, France 4 Assystem, Pertuis, France 5 CEA-IRFU, Saclay, Franc

2. Outlines2IntroductionOLYMPE general presentation & statusElectroMagnetic-Design (EMD) bimodular loopThermoHydraulic-Design (THD) bimodular loopThermoHydraulic-Cryogenic-Design (THDC) trimodular loopMechanical-Design (MECD) bimodular loopConclusion-perspectives

3. Introduction3Tokamak superconducting magnet system design  multi-scale approach  several stages of analyses detail 1- System scaledownscalePerimeter = reactorcomponents (magnets, VV…)macroscopic models for components2- Magnet scalePerimeter = magnet envelopeMajor sub-elements (conductor, structures)macroscopic models for sub-elements (mechanics, EM…) downscale3- Sub-magnet scalePerimeter = subcomponent (conductor, insulation…)sub-elements details consideredAccurate models of sub-elements (EM, thermohydraulic…), complexity  when scale  : modelling & interfacing systems, calculation ressources…

4. Workflow4OLYMPEThermal Cast3MCryodistribution SimCryogenicsSystem codes-PROCESS-SycomoreThermal loads (NH…)TF pre-designMADMACSGUIPythonThermohydraulics THEAMechanics Cast3M Design criteriaDTMARG, hotspot, mechanical stress…Electromagnetism TRAPSData modelCEA develops an integrated tool which holds variants in scales, physics & fidelity The platform is named OLYMPE (platefOrme muLtiphYsique pour aiMants suPraconductEurs)Strategy: couple several OLYMPE modules in loops to enhance the TF design method efficiency

5. EMD loop: principle5TF pre-designMADMACSElectromagnetism TRAPSData modelElectroMagnetic-Design Logic flow…Step n : an effective field (BEFF) is consideredThe TF design is modified by this BEFFMADMACS outputThe TF design modifies BEFF (current carriers toplogy)TRAPS outputStep n+1 : a new BEFF is considered…loopEMD purpose : issue a BEFF-compliant TF design with regard to detailed analyses

6. EMD loop: principle6BEFF(step n)conductor & WP sizeDesignMADMACSEMTRAPSDBEFF/BEFF <0.1%?noDesign convergedTRAPS time consuming (~10 min per run)Not fully automatizedBEFF(step n+1)yesEMD loop

7. EMD loop: application to DEMO7An application is conducted on the DEMO 2018 WP#3 configuration (pancakes winding, no radial plates) TRAPS calculates the B map with 3D model (separated individual conductors, D-shape)Asymptotic behaviour after few iterationsThe TF self-induced configuration convergesIndependant B offset due to PF-CS creates small oscillations around close designs (1 strand difference)EMD loop is reliable and outputs self-consistent TF designsThe computing time remains importantOwing the fact that other loops should be aggregated (e.g. TH module) efficiency is adressed

8. EMD loop: enhancement8B calculation efficiency is enhanced by considering a full analytical approachB map is set with infinite current carrier having rectangular section (analytical formalism)critical conductor blockB in critical zone is represented by contribution of section blocks which size depend on distance from critical zoneClose neighbouring winding blocks

9. EMD loop: enhancement9Neighbouring legs = monoblocksOther legs : wiresTokamak axisSame approach for TF tokamak blocksBlocks simplicity = reduce the calculation time budgetTF critical legCalculation is reduced to 0.1 s.

10. EMD loop: enhancement10A benchmark of analytical approach (2D) with TRAPS (3D) was conducted on JT-60SA TF configuration Difference between the two approaches is < 1%Sufficient for ensuring a convergence close to final state

11. EMD loop: enhancement11BEFF(step n)conductor & WP sizeCICC design subroutineDBEFF/BEFF <0.1%?noDesign convergedBEFF(step n+1)yesThe 2D EM analytical script was integrated into MADMACS structure (enhanced MADMACS+) DesignMADMACS+2D EM analytical subroutineMADMACS+ calculation time < 1 secondEDM+ loop: MADMACS+ convergence is placed before EMD loop (MADMACS & TRAPS)

12. EMD loop: enhancement12The EMD loop+ expected to run much less iterations with TRAPS (to be checked)EMD loop+ full automatization to be carried out BEFF(step n)conductor & WP sizeDesignMADMACSEMTRAPSDBEFF/BEFF <0.1%?noDesign convergedBEFF(step n+1)yesEMD+ loopDesignMADMACS+First step

13. THD loop: principle13Thermo-Hydraulic DesignTF pre-designMADMACSThermohydraulics THEAData modelLogic flow…Step n : an operation temperature (TOP) is considereda TF design is issued from this TOPMADMACS outputThe TF design modifies the TOP (hydraulic properties)TRAPS outputStep n+1 : a new TOP is considered…loopTHD purpose: issue a DTMARG-compliant TF design with regard to detailed analyses

14. THD loop: principle14TOP(step n)conductor & WP sizeDesignMADMACSTHTHEADTOP/TOP <0.1%?noDesign convergedFully automatizedRun time is about 1 min per iteration  ~ 20 min usually for convergenceEM considerations can be introduced with MADMACS+TOP(step n+1)yesTHD loopMADMACS+ output is a single BEFF value while THEA works with a 1D EM map B(x)Initial B(x) is with TRAPS for first step, for next steps B(x) is shifted so that max(B(x))=BEFF.

15. THD loop: application on DEMO15 An application of the THD loop with MADMACS (original version) was initiated on DEMO 2018 TF WP#3 concept (pancake option without radial plates) : parametric exploration with different [DP, TIN]  ref [1][1] Sandra Varin, François Bonne, Christine Hoa, Jean-Marc Poncet, Louis Zani, Benoît Lacroix, Quentin Le Coz, Optimization of the overall Toroidal Field Coil cryomagnetic system at the pre-conceptual design phase of the European DEMO fusion reactor, presented at SOFT 2020, to be published in Fusion Eng DesignA dedicated GUI was developped (Python) for the THD loopPresent objective : conduct similar parametric studies butWith more realistic thermal load on conductor (i.e. considering the load from structures)across a broader [DP, TIN] range = [0.5 – 1.5 bar]*[2.5 – 4.5 K]MADMACS and MADMACS+ modules used to quote the effect of BEFF variation with design across the spanned zone

16. THD loop: application on DEMO16 As result both cases look similar Differences are marginal as Dcost < 1 %This derives from the small difference in conductor size and then in BEFF The BEFF calculation functionality does not change much the global dependance with DP and TIN

17. THCD loop: principle17Cryodistribution SimCryogenicsTF pre-designMADMACSThermohydraulics THEAData modelThermo-Hydraulic & Cryogenic DesignloopLogic flowTHD loop is run across an operation domainTHD loop outputs defined factors of merit THD loop data useful for cryogenic design are transferred to Simcryogenics Simcryogenics is run across the same operation domainSimcryogenics outputs factors of merits of same nature as for THDFactors of merits are combinedObjective: produce system-scale parametric studies considering magnet and cryoplant

18. THCD loop: application to DEMO18In the logic of extension to coupled cryo-magnetic TF system similar as in ref [1] we carried out a parametric exploration with different [DP, TIN] with the same conditions as in THD loop exampleThe purpose is to combine the cost variations of the TF system and cryogenics systems across the [DP, TIN] range.As per TF magnet part we retained the outputs obtained with MADMACS+ module (see previous)As per TF cryogenic part the configuration retained is a 2 bathes structured cold end with a 4.4 K bath that feeds a subcooler bath at Tin – 0.2 K .  The cost of the cryogenic part is divided in two : - Investment cost estimated with the Green formula - Operational cost is estimated with the electricity cost over 20 year exploitation  DPTINCICC

19. THCD loop: application to DEMO19As result on the TF cryomagnetic system we obtain cost increase Cost combination claims for an optimum…cost increase

20. THCD loop: application to DEMO20Result on the overall cryomagnetic TF systemThe overall system cost spans across a broad range (545 ~ 900 M€)An absolute minimum is found at ~ [TIN = 3.4 K DP =0.6 bar] minimum cost is ~ 545 M€Compared to DEMO nominal conditions [TIN = 4.5 K DP = 1 bar] cost difference is ~ 140 M€

21. THCD loop: application to DEMO21Discussion: is the minimum shallow or steep ? A 3% decision flexibility margin is considered to identify the boundaries of optimized operation domainOn DP dimension, the minimum is rather shallow and any value of DP between 0.5 and 1.2 bar seem appropriate Conversly minimum is steeper across TIN dimension, claiming for a TIN between 3 K and 3.7 KAs per cost merits, the overall TF cryomagnetic system optimum pushes for low temperature operation PerspectivesRefine cost merit functionsConsider technological choices (cryo…)Extend to the whole cryomagnet system (CS, PF)

22. MECD loop: principle22MECanical DesignTF pre-designMADMACSMechanics Cast3M Data modelloopLogic flow…Step n : a couple of stress values (sCASING sJACKET) is considereda TF design is issued from this (sC sJ) (macroscopic analysis)MADMACS outputThe TF design develops local (s*C s*J) (detailed analysis)CAst3M output Comparison with ASME limits (DsC DsJ) definedStep n+1 : a new (sC sJ) is considered…Objective: issue a stress-compliant TF design with regard to detailed mechanical analysis

23. MECD loop: principle23[sCAS sJACK](step n)Jacket and casing sizeDesignMADMACSMECACast3MDs = s*- sASME < 1 MPa ?noDesign convergedCast3M time consuming (~2 hours per run)can be improved (friction considered, GPS method not optimized…)Data exchanges in loop not yet automatized[s*CAS s*JACK]yesMECD loop[sCAS sJACK](step n+1)Ds

24. MECD loop: application to DEMO24The first application attempt is conducted on DEMO 2015 TF WP#3 conceptDesignMADMACSsCAS= 680 MPasJACK=622 MPaCast3M modelFeatureValue(mm)Jacket thickness11Cable side48.3TF nose thickness511TF thickness1175NB: MADMACS Excel version is used for first attempt to ensure a stable optimisation with respect to both s

25. MECD loop: application to DEMO25Detailed analysis of DEMO 2015 load caseLocationMembrane stress (MPa)Margin to ASME (MPa)Casing616+ 50Jacket 568+ 100Jacket stress mapCasing stress mapsCAS= 680 + 50 = 730 MPasJACK=622 +100 = 722 MPaDesignMADMACSFeatureValue(mm)D (mm)Jacket thickness8.7-2.3Cable side48.30%TF nose thickness480-31TF thickness1100-75new Cast3M run to be conductedMECD loop operationality to be assessedConsider membrane + bending criterionThinner jacketTF coil smaller radial buitcritical pathsNB: critical paths are automatized

26. Conclusion - Perspectives26In the framework of OLYMPE development several multi-physic design loops were built using the different modules of the platformElectromagnetic Design loop was established with a two-stages approachAn integration ni MADMACS was carried out for analytical part (moderate fidelity)An hybrid loop is combined with TRAPS (high fidelity)Thermo-hydraulic Design loop was established A parametric study [DP, TIN] was conducted on DEMO configurationIn this application considering BEFF variation with design has margina impact on resultsThermo-hydraulic Cryogenic Design loop was established The parametric study [DP, TIN] was extended conducted on DEMO configurationA global optimum for cryomagnetic system was found, claiming for low temperature operation Mechanical Design loop was established A first try on DEMO 2015 was attempted to be consolidatedAssess all loops convergence along their actual maturity (EMD, THCD, MECD…)Develop data exchange automatization and GUIsRun several application cases to establish their convergence stabilityImprove time computation budgetExtend to all tokamak systems (CS, PF)perspectives

27. Thank you for your attentionQuestions ?