Update on wound-truss carbon fiber structures: Optimization, tests and simulations

1. Update on wound-truss carbon fiber structures: Optimization, tests and simulations Claudio BORTOLIN (CERN) Enrico DA RIVA (CERN) Corrado GARGIULO (CERN) Manuel GOMEZ MARZOA (CERN) WG4 Meeting - 16thOctober 2012 WG4 Meeting - 16th October 2012

2. Contents • D08 prototype: test data post-processing • Prototype optimization • Design parameters • Stave calculations • Cooling fluid calculations • D08 prototype: simulation • D09 prototype: preliminary tests with water • General conclusion WG4 Meeting - 16th October 2012

3. Data post-processing • D08 water tests -> Stave thermal resistance calculation • From Silicon to water, two thermal resistances can be defined: • Water tests allow recalculation of HTC: RtConv RtCond-Stave Water Silicon Pipe inner wall Laminar & Gnielinski WG4 Meeting - 16th October 2012

4. Two-phase C4F10 tests: overview • Inlet vapor quality: • Superheating at stave outlet: T = const x = const • Mass flow rate calculation: 1 p [bar] where L is latent heat [kJ kg-1]: 3 3’ 2 4 Qstave[W] • Usually: h [kJ kg-1] WG4 Meeting - 16th October 2012

5. D08: water vs. C4F10 @0.3 W cm-2 Evaporative cooling system performs as good as single-phase water WG4 Meeting - 16th October 2012

6. D08: water vs. C4F10 @0.5 W cm-2 • Results independent of the mass flow rates. • Controlling the vapor quality at the inlet/outlet is very important. • Almost subcooled liquid at the stave inlet! WG4 Meeting - 16th October 2012

7. D08: C4F10 tests discussion • Two extreme cases: • Low vapor quality at the stave entrance: subcooled liquid entering stave? • ꜛ m, ꜛ HTC, but ꜛΔp. Since pOut = constant, ꜛpInlet, ꜛTsat-Inlet, ꜛΔTFluid • Low vapor quality at the stave entrance: saturated liquid entering stave? • Mass flow rate too low: superheated vapor at stave outlet WG4 Meeting - 16th October 2012

8. Data post-processing • D08 C4F10 tests: HTC prediction and flow characterization • Pool boiling or convective boiling? HTC predicted through Rt 0.3 W cm-2 0.5 W cm-2 *Assumedsmooth pipe. Forcorrelations, powerisassumedto be distributeduniformlyaroundthecooling pipe. Used Cooper (1984) and Liu-Winterton (1991) correlationsfor HTC. WG4 Meeting - 16th October 2012

9. Design parameters • Preliminary calculations: • Heat transfer path: • Si-glue • Glue-wrapping CF • Wrapping CF-CF sleeve • Heat distribution within CF sleeve • CF sleeve-cooling pipe • Cooling pipe-fluid 5 4 6 2 3 1 • Materials: • Carbon Fiber: • Wrap fiber: K13D-2U CF: kFiber ~ 450 W m-1 K-1 ; kTransv ~ 1.2 W m-1 K-1 • Carbon Paper: kFiber ~ 550 W m-1 K-1 ; kTransv ~ 1.2 W m-1 K-1 • Glue: k ~ 1 W m-1 K-1 • Pipe (Polyimide): k= 0.12 W m-1 K-1 , OD = 1.5 mm, wall 35 µm thick • Silicon: k = 150 W m-1 K-1 WG4 Meeting - 16th October 2012

10. Stave calculations • 0. D08 prototype • Decrease stave width to 15.5 mm • Increase CF width to 1.75 mm • Increase CF width to 1.90 mm • Increase CF thickness to 100 µm • Decrease number of fibers to 40 • Increase wrap angle to 40 deg • Increase wrap angle to 60 deg • Increase glue thickness to 200 µm and decrease its th. cond. to 0.5 W m-1K-1 • Increase pipe outer diameter to 2.5 mm • New prototype (D09) Strategy: estimate the ΔT that every part of the structure will introduce. WG4 Meeting - 16th October 2012

11. Stave calculations: conclusion • 0. D08 prototype • Decrease stave width to 15.5 mm • Increase CF width to 1.75 mm • Increase CF width to 1.90 mm • Increase CF thickness to 100 µm • Decrease number of fibers to 40 • Increase wrap angle to 40 deg • Increase wrap angle to 60 deg • Increase glue thickness to 200 µm and decrease its th. cond. to 0.5 W m-1K-1 • Increase pipe outer diameter to 2.5 mm • New prototype (D09) WG4 Meeting - 16th October 2012

12. Cooling fluid calculations • Convection pipe wall-fluid: • Several HTCs considered • Stave height = 5 mm, fiber wrap angle = 23 deg • Heat transfer area inner pipe wall only below 180-η angle at sleeve. HTC variation Increase pipe section Decrease pipe section WG4 Meeting - 16th October 2012

13. Stave optimization • Stave: • Improvements without increasing mat. budget: • Reduce stave thickness to 15 mm • Increase wrap angle (eventually, use HC Plate) • Improvements increasing mat. budget: • Increase CF width • Increase carbon fiber thickness • Cooling fluid: • Increasing pipe diameter reduces ΔTPipe-Ref , but increases mat. budget (espec. single phase) • Little improvement when increasing HTC pipe-fluid • Further considerations: • Single-phase or Two-Phase? • Fluid? • For better understanding heat transfer mechanisms, CFD simulation was updated • D08 prototype modeled and simulated. WG4 Meeting - 16th October 2012

14. D08 prototype simulation: model • UPDATED CFD Model: • Symmetric (only a sector of stave modeled) • Imposed HTC fluid-wall at the pipe inner wall. HTC = 1700 W m-2 K-1 • Results shown for Twater = 15 °C and q’ = 0.3 W cm-2 • Natural convection with ambient air is included HTC = 10 W m-2 K-1 (also included top part) • Thermal conductivity CF: anisotropic: solver did not cope with such different values • Th. Cond used: different, but analytical calculations show no influence in transversally to CF Geometry views of the model. • Carbon Fiber: • Wrap fiber: K13D-2U CF: kFiber ~ 450 W m-1 K-1 ; kTransv ~ 40W m-1 K-1 • Carbon Paper: kFiber ~ 550 W m-1 K-1 ; kTransv ~ 40W m-1 K-1 WG4 Meeting - 16th October 2012

15. D08 prototype simulation: results Temperature of the silicon: WG4 Meeting - 16th October 2012

16. D09 prototype: design parameters • Comments: • 54 fibers (thermal “bridges”): p ~ 3 mm • Gluing defects found near the connections (inlet/outlet stave end). • All the results for these tests are preliminary. D09 prototype geometrical parameters. WG4 Meeting - 16th October 2012

17. D09 prototype: gluing defects • Gluing defects found near the connections (inlet/outlet stave end) • Heater is there detached from the silicon • Hotspot appeared after the 10 L h-1 , 0.3 W cm-2 test. Inlet/Outlet First hotspot because gluing defect. 10 L h-1, 0.3 W cm-2 case. Second hotspot appears (heater detaches from silicon). 5 L h-1, 0.3 W cm-2 case. Hotspot location 5 L h-1, 0.3 W cm-2 case. WG4 Meeting - 16th October 2012

18. D08 vs D09 water test@0.3 W cm-2 WG4 Meeting - 16th October 2012

19. D08 vs D09 water test@0.5 W cm-2 • Conclusion: • The gluing defects make difficult to analyze how the D09 performs vs D08. • Preliminary results suggest the behavior is quite similar in both, with D08 performing slightly better. WG4 Meeting - 16th October 2012

20. Water tests: results WG4 Meeting - 16th October 2012

21. General conclusion • Post-processing data allows to understand the thermal resistance of the prototype. • Through analytical calculations + simulations, an optimal design can be reached. • It is very important to quantify the influence of the manufacturing process (CF wrapping, gluing…) on the final result. • Guarantee nice gluing: increasing glue thickness a low impact thermally. • Gluing in vacuum: prevent from having voids in the glue layer. • The solution with a high conductivity plate is attractive if the material budget from the D08-like structures keeps increasing. • Tests with squeezed pipes will require a low pressure/leak-less system WG4 Meeting - 16th October 2012

22. Update on wound-truss carbon fiber structures: Optimization, tests and simulations Claudio BORTOLIN (CERN) Enrico DA RIVA (CERN) Corrado GARGIULO (CERN) Manuel GOMEZ MARZOA (CERN) WG4 Meeting - 16thOctober 2012 WG4 Meeting - 16th October 2012