1. MHD/Heat Transfer considerations for SiC FCI in DEMO and ITER Sergey Smolentsev DCLL Special Meeting at UCLA April 23-24, 2007

2. Background, I As shown, 5 mm SiC/SiC FCI reduces the MHD pressure drop in the poloidal ducts by ~10 2 in both ITER and OB DEMO. However, at the velocities 5-10 cm/s, the MHD pressure drop is not an issue. Even without the FCI,  P~10 -2 MPa, i.e. very small. That is why the main FCI function is thermal insulation and decoupling hot PbLi from the Fe wall, but not electrical insulation. For the IB DEMO, electrical insulation in the poloidal ducts may be needed since the MHD pressure drop is proportional to B (non-conducting walls) or B 2 (perfectly conducting walls). Thus,  P~10 -1 MPa (without the FCI), which is not negligible.

3. Background, II The FCI design and choice of the SiC material properties depend on: - (A) thermal losses into He; - (B) Max  T across the FCI (or Max thermal stress); - (C) Max PbLi-Fe T (or Max corrosion rate); - (D) Max SiC temperature. In ITER conditions, (A)-(C) are not issues, even in off- normal scenarios. Therefore, there are no special requirements on SiC properties in ITER. No headache! In DEMO conditions, (B) and (C) can be really severe. Current data show that satisfying (B) is hardly possible, unless more complex FCI design or new SiC materials with unique properties are developed and implemented.

4. Summary of MHD effects The following most important MHD phenomena that affect heat transfer have been identified: - formation of high-velocity near-wall jets; - 2-D MHD turbulence; - buoyancy-driven flows (mixed convection). One more effect (which has not been discussed in detail yet) but can be very important: wetting SiC by PbLi. Current approach: decoupling one effect from the others. This allows for qualifying the impact of a particular phenomenon on heat transfer and shows the variation range, e.g. max and min heat losses.

5. Wetting Vs. no wetting Perfect wetting All current MHD/Heat transfer results are based on the assumption that SiC is perfectly wetted by PbLi. No thermal or electrical interface resistance are assumed. Well established MHD models can be used. No wetting or pure wetting At first glance, lack of wetting promises some advantages, such as higher thermal and electrical interface resistance. What may happen: Unpredictable flow behavior with local “hot spots” in the areas where wetting occurs. Quite different (new !) MHD approach should be used. We need to know to what degree SiC will be wetted by PbLi in DEMO-like conditions !

6. Current results for DEMO, I 5 mm FCI, 2 mm gap Nominal PbLi  T=200 K (500-700  C) G=104 (bulk)+1.2(gap)=105.2 kg/s U front =6.4 cm/s U return =3.43 cm/s Q total =(0.55+3.08  1.136)  1  2=8.10 MW (1) laminar and (2) turbulent flow model at  SiC =20, 100 S/m and k SiC =1, 2, 5 W/m-K. Stress on the heat loss and FCI  T !

7. Current results for DEMO, II Laminar, 100 S/m, 1 W/m-KLaminar, 100 S/m, 5 W/m-KLaminar, 100 S/m, 2 W/m-K Laminar, 20 S/m, 2 W/m-KTurbulent, 100 S/m, 2 W/m-KTurbulent, 100 S/m, 1 W/m-K Characterization of the heat loss from PbLi

8. Current results for DEMO, III Case , %  T, K Front duct  T, K 1 st return duct  T, K 2d return duct  T, K total k=0 (all heat from the structure and FCI goes into He) 60.32045730247  SiC =100 S/m k=1 W/m-K laminar 54.9208304225  SiC =100 S/m k=2 W/m-K laminar 49.420114-11202.5  SiC =100 S/m k=5 W/m-K laminar 41.2191-9-33170  SiC =20 S/m k=2 W/m-K laminar 48.420415-11206  SiC =100 S/m k=2 W/m-K turbulent 47.92028-19196.5  SiC =100 S/m k=1 W/m-K turbulent 54.520828.51.5223

9. Current results for DEMO, IV Ideal insulation k SiC =5 W/m-K,  SiC =100 S/m CARACTERIZATION of HEAT LOSSES in DEMO Maximum achievable  =Q PbLi /Q total ~60% (could be slightly higher providing some heat generated in the FCI returns into PbLi). The limit is related to the volumetric fraction of solid (Fe and SiC) in the blanket, since almost all heat generated in the structure goes into He.

10. Current results for DEMO, V Heat losses are more pronaunced in the return ducts Turbulent heat losses are higher than laminar Heat losses slightly decrease as  SiC decreases Ideal thermal insulation:  =Q PbLi / Qtotal =60.3% k=1 W/m-K:  =55%. If k<1 W/m-K, there is almost no effect of turbulence and near-wall jets on the total heat loss k<<1 W/m-K: high temperature spike in SiC Goal: k=0.5-1 W/m-K Summary of the heat loss analysis

11. Current results for DEMO, VI Case Front duct front wall Front duct side wall 1 st return front wall 1 st return side wall 2d return front wall 2d return side wall  SiC =100 S/m k=1 W/m-K laminar  T FCI =150 K T int =480  C 700 580 225 480 270 485 230 480 235 475  SiC =100 S/m k=2 W/m-K laminar 100 495 300 660 130 520 150 510 140 510 135 500  SiC =100 S/m k=5 W/m-K laminar 130 490 500 610 190 510 205 500 200 500 190 490  SiC =20 S/m k=2 W/m-K laminar 200 495 240 580 200 515 210 515 200 515 200 515  SiC =100 S/m k=2 W/m-K turbulent 220 495 220 570 210 515 220 515 215 515 215 515  SiC =100 S/m k=1 W/m-K turbulent 240 495 260 560 245 505 250 505 245 505 245 505  T across the FCI and the interface temperature

12. Current results for DEMO, VII Reduction of k to ~1 W/m-K is desirable from the point of view of reduction of heat losses. Smaller k also results in lower T int. The negative effect is, however, a significant increase in  T FCI. Very low k (<<1) is also not acceptable because of the temperature spike in the SiC. PbLi flow has a very strong effect on  T FCI.. Therefore, adjusting  SiC or the FCI thickness is an effective tool of reducing the thermal stress in the FCI. The present parametric study shows how variations of  affect  T FCI. However, even in the best case scenario, the  T FCI and T int seem to be unacceptably high. New design solutions or new SiC material capable of standing up to ~250 K across the 5 mm FCI are needed. Summary of the analysis for  T FCI and T int

13. STRATEGICAL SUGGESTIONS Variant 1. Keep the same design (including one- layer SiC/SiC FCI) and wait for new materials with unique properties. Variant 2. Keep essentially the same blanket design but redesign the FCI (e.g. nested FCI). Variant 3. Redesign both the blanket and the FCI. Variant 4. Give up the idea of high-efficiency blanket by reducing the exit PbLi temperature to ~ 500  C. Less problematic options are then possible, e.g. “sandwich FCI”. (topic for discussion)

14. Possible design changes Use “nested” FCI instead of present one-layer FCI (S. Malang) Reconfigurate the PbLi flow, starting it from the back (C. Wong) Reduce the radial depth of the front channel (increase velocity). One more return channel will likely be needed (N. Morley) Increase heat transfer coefficient in He, where the interface temperature is too high, by reducing the He channel size or pumping more He

15. Questions to material people Is k~1 W/m-K achievable? Is  ~1-100 S/m achievable? It appears that we know what would happen with k under the neutron flux. What would happen with  and how fast? What is the effect of T on  ? Is there any documented information on wetting SiC by PbLi. If no-wetting occurs how would it look like in the blanket conditions? What is the maximum allowable  T (or stress) for the existing SiC composites? How this maximum stress could be extrapolated to future materials?