Thermalhydraulic optimization of hypervapotron geometries for first wall applications

Plasma disruptions and Edge Localized Modes (ELMS) may result in transient heat fluxes as high as 5 MW/m2 on portions of the ITER first wall (FW). To accommodate these heat loads, roughly 50% of the first wall will have Enhanced Heat Flux (EHF) panels equipped with water-cooled hypervapotron heat sinks.


INTRODUCTION
Engineers are often faced with a choice in designing actively cooled heatsinks for plasma facing components. Either they can design for the worst case heatloads expected for off normal transient conditions, or they can optimize the design for nominal operating conditions with enough safety margin to survive most expected off-normal events. This study optimizes the geometry and flow parameters of hypervapotron heatsinks for fIrst wall applications under nominal operating conditions. The flow parameters and flow circuit design are just as important as the internal dimensions of the hypervapotron. If multiple hypervapotron fIngers can be flowed in series, it can simplify the coolant routing signifIcantly. This article also describes optimized geometries for off-normal heat loads in two-phase flow and discusses the trade-offs required in selecting the best confIguration for nominal single-phase operation with sufficient safety margin for the off-normal events provided in part by boiling.

HYPEVAPOTRON GEOMETRY
We previously performed extensive CFD heat transfer analysis of a Chinese (CN) fIrst wall mock-up consisting of beryllium armor tiles bonded to a water-cooled CuCrZr hypervapotron heatsink attached to a 316LNG stainless steel strongback. This model became the test bed for our geometry optimization study. Fig. 1    in the literature [4,5]. Experimental results from high heat flux testing of various geometries including widths and side slot variations are also in the literature [6,7].

A. Long Teeth, Deep Channel
The reference geometry consists of 4-mm-deep hypervapotron teeth, 3-mm-wide on a 6-mm-pitch. Slots, 3mm-wide and 4-mm-deep, run down the sides of the teeth the entire length of the back-channel. The back-channel is 43-mm-wide and its height is 5 mm. A cross-section of this geometry appears in Fig. 3 in the upper left.

B. Short Teeth, Shallow Channel
The next case consisted of 2-mm-deep teeth, 3-mm-wide on a 6-mm-pitch. The side slots were 2-mm-deep and 3-mm wide. The back-channel height is reduced to 3 mm to maintain the same average velocity at a reduced mass flow. Fig. 3 shows the geometry dimensions in the upper right.
C. Long Teeth, Shallow Channel The third case consisted of 4-mrn-deep teeth, 3-mm-wide on a 6-mm-pitch. The side slots were 4-mm-deep and 3-mm wide like the reference geometry. However, the back-channel height is reduced to 3 mm for comparison to the previous case.
A cross-section of the geometry is shown in the lower left of Fig. 3.

D. Short Teeth, Deep Channel
The final case consisted of 2-mm-deep teeth, 3-mm-wide on a 6-mm-pitch. The side slots were 2-mm-deep and 3-mm wide. The back-channel height is 5 mrn and is identical to the reference case. Fig. 3 displays the dimensions of this geometry in the lower right.

CFD MODELING RESULTS
The most important results from the CFD simulations are the surface temperature distributions on the beryllium tiles, the velocity profiles in the grooves of the hypervapotron, and the calculated local convective heat transfer coefficient (HTC) in the grooves. The HTCs between the nominal heat flux case and the off-normal case are compared for each geometry as well as the beryllium surface temperature.   The velocity distributions for the four geometries appear in Fig. 7. The average velocity in the back-channel is about 2 m/s. Note that with the short teeth, the velocity in the bottom of the grooves is higher than the long teeth (0.8 mls compared to 0.2 mls) The pressure drops range between 14.5 and 15.6 kPa.  The vapor fractions for the off-nonnal heat loading appear in Fig. 8. The short teeth/shallow channel has more boiling with a vapor fraction of 6.7% compared to 5.1 % for the deep teeth/deep channel. This is a small amount of water vapor that only partially fills the hypervapotron grooves and side slots. No vapor appears under the teeth; so some margin is available before reaching critical heat flux.
2mm/5 mm   A comparison of the local heat transfer coefficients appears in Fig. 9 Figure 9. Comparison of peak heat transfer coefficients IV.

DISCUSSION
The advantage of the hypervapotron is that it relies on boiling heat transfer to provide the safety margin for off normal conditions. The optimum for nominal heating and single phase flow is not the optimum geometry for boiling heat transfer. Therefore, a compromise between the two must be made. Ideally, one would like optimum performance under nominal conditions and sufficient performance under off normal heating to survive any transients.
All the studied geometries have acceptable thermal performance, so the design drivers become the available mass flow and allowable pressure drop. In addition, layout of the cooling circuit and manifolding will impact the final choice of hypervapotron geometries. V.

CONCLUSIONS
The short teeth/shallow back-channel geometry boils in the bottom of the grooves more at 5.0 MW/m 2 case than the long teeth/deep back-channel geometry. This results in a reduction in surface temperature on the beryllium tiles. However, the short teeth/deep back-channel (2 mm 15 rum) provides more water volume for downstream mixing and condensation of the vapor and has the best performance for off-normal conditions. The 0.5 MW/m 2 nominal case shows nearly equivalent temperatures between the cases. While the 213 configuration has the highest convective heat transfer coefficients in localized areas, the 4/3 configuration has the highest average heat transfer coefficient over a larger area, and thus the lower surface temperatures. The short teeth geometry with a shallow back-channel has a 4% increase in pressure drop even with the reduced mass flow.
A big advantage of the short teeth/shallow backchannel design is the performance it provides at half the mass flow. Thus hypervapotron fingers can be fed in parallel with the same total mass flow. This greatly simplifies the water circuit for a forty-finger panel.
Flowing pairs in series requires an additional flow return for each pair in a panel with limited space available.
We will include the effects of hypervapotron width on off normal performance using these same cross sectional geometries in a more detailed future article.