Power exhaust concepts and divertor designs for Japanese and European DEMO fusion reactors

Conventional divertor designs have been developed for the JA and EU DEMO concepts, based on the ITER divertor. Large power handling of P_sep/R_p = 30–35 MW m⁻¹ is a significant challenge for the JA DEMO, and a long leg divertor (L_div = 1.6 m, i.e. 1.6 times longer than ITER) was proposed as shown in figure 4(a). The inner and outer divertors cover all divertor plasma volume, and compression of the neutral particles and efficient formation of the plasma detachment will be expected particularly near the strike-point. The poloidal angles at the inner and outer targets (θ^div) are 30° and 25°, respectively, and the magnetic flux expansion factor, i.e. f_exp ^div/sinθ^div, is similar (∼12) at the both targets. Design concept of the ITER divertor is simplified for the EU DEMO divertor in order to increase the tritium breeding blanket volume. While the similar long leg divertor with similar θ^div at the outer strike point is planned, the baffles and dome are removed, and the targets cover near the strike-points as shown in figure 4(b), which is rather open geometry compared to closed one such as the ITER and JA DEMO divertors.

Zoom In Zoom Out Reset image size

Figure 4 also shows calculation meshes of SONIC for the JA DEMO divertor and SOLPS5.1 for the EU DEMO divertor. Modelling of various drifts such as ∇B × B and E × B is not incorporated in the former code. The latter code does not activate the drift effects, and the liner is removed in this calculation. Exhaust power (P_out) and particle flux are given at the core-edge boundary. P_out = 250 and 300 MW are applied at the core-edge boundary (r/a = 0.95) for the JA DEMO higher-κ and JA DEMO 2014 reference cases, respectively. The total radiation power in the plasma edge (0.95 < r/a < 1), SOL and divertor (P_rad ^edge + P_rad ^sol + P_rad ^div) was adjusted at a fixed value by feedback of the seeding rate in the iterative calculation, thus P_sep is slightly smaller than P_out because of P_sep = P_out − P_rad ^edge. Reference of the radiation power fraction, i.e. (P_rad ^edge + P_rad ^sol + P_rad ^div)/P_out, was determined to 0.8 for the two cases [11]. On the other hand, for the EU DEMO reference case, P_out = 150 MW is given at r/a = 0.98, thus P_sep is comparable to P_out, and a series of the radiation power scan was performed with increasing the seeding rate. Application of diffusion coefficients on the ion and electron heat fluxes (χ_i, χ_e) and particle flux (D) for the SOL plasma is critical issue to simulate the divertor performance. For the JA DEMO, χ_i = χ_e = 1 m² s⁻¹ and D = 0.3 m² s⁻¹ were the same as the 'standard' values of ITER simulation by SOLPS4.3 [25]. The e-folding length near the outer midplane separatrix (λ_q// ^mid) of the parallel heat flux (q_//) profile (including electron and ion components) corresponded to ∼3.0 mm, which was narrow compared to 3.6 mm in the ITER simulation due to higher T_e ^mid and T_i ^mid in the JA DEMO (370 and 830 eV, respectively). For the EU DEMO, smaller χ = 0.3 m² s⁻¹ (inside separatrix) and 0.18 m² s⁻¹ (in SOL) were given to provide q_// profile with comparable λ_q// ^mid of ∼3 mm.

Divertor operation in low n_e ^sep range and influences of the key parameters such as P_sep, f^* _rad ^div and λ_q// ^mid were recently investigated in the JA DEMO with Ar seeding [27]. A series of n_e ^sep scan was performed with changing fuel gas puff and divertor pumping rates, where Ar puff rate was controlled to keep a given value of P_rad ^edge + P_rad ^sol + P_rad ^div. It is noted that P_sep and f^* _rad ^div were slightly smaller than fixed parameters of P_out and (P_rad ^edge + P_rad ^sol + P_rad ^div)/P_out, respectively. Two reference series for 'JA DEMO high-κ' (P_sep ∼ 235 MW and f^* _rad ^div ∼ 0.78) and 'JA DEMO 2014' (higher P_sep ∼ 283 MW and the same f^* _rad ^div), and extreme cases for the two references with reducing f^* _rad ^div to ∼0.7 were investigated. For the four cases, the total plasma power to the target, i.e. P_sep·(1 − f^* _rad ^div), correspond to 50, 60, 75 and 90 MW, respectively. Distribution of the radiation loss in the inner and outer divertor by Ar seeding and profiles of plasma and heat load at the target were reported, for example, in figure 2 of reference [27]. The large radiation area was seen at the upstream of the inner target, and T_e ^div was decreased to ∼1 eV over a wide area of the target, which we described as 'full-detachment'. In the outer divertor, large radiation peak was also seen at the upstream near the separatrix. On the other hand, it shifted towards the target at the outer flux surfaces. The plasma detachment was produced near the strike-point, which we described as 'partial detachment'. Typical plasma profiles and heat load components at n_e ^sep = 2.0 × 10¹⁹ m⁻³ for the JA DEMO high-κ case are shown in figures 5(a) and (b). The width of the partial detachment for the reference case is relatively large (∼12 cm). Peak q_target (∼5 MW m⁻²) appears at the boundary of the attached region, where T_e ^div and T_i ^div are increased and n_e ^div is decreased significantly, thus the peak q_target is sensitive to their profiles. Profiles of T_e ^div and T_i ^div at the attached region are relatively flat or rather increased towards the outer SOL, but they are still lower than those at the upstream SOL, as shown in figure 5(c), by radiative cooling. Total q_target is evaluated by including surface recombination of the ions (q_t ^rec = n_i ^div C_s ^div E_ion, where n_i ^div, C_s ^div and E_ion are ion density, sound velocity at the divertor sheath and recombination energy, respectively), radiation power load (q_t ^rad) and neutral flux load (q_t ⁿ), in addition to the plasma heat load (q_t ^plasma). The peak q_target is attributed by q_t ^plasma and q_t ^rad in the attached region. The width of the partial detachment corresponds to ∼1 cm of the midplane SOL radius, as shown in figures 5(c) and (d). Decay lengths of the parallel heat flux profiles near the separatrix at X-point (q_//e ^Xp and q_//e ^Xp + q_//i ^Xp) are 2.3 and 2.9 mm (mapping to midplane radius), respectively. Extremely large heat flux in the near-SOL is significantly reduced in the divertor.

Zoom In Zoom Out Reset image size

Since the peak q_target is generally larger than that of the inner q_target, results of the outer q_target are plotted in figure 6(c). Circles and squares show series of reference JA DEMO higher-κ and 2014 cases, respectively, where the peak q_target in figure 5(b) is marked by open circle. The peak q_target is gradually decreased in higher n_e ^sep, compared to that in lower n_e ^sep. The two series of f^* _rad ^div ∼ 0.8 are acceptable in expecting low n_e ^sep (⩾2 × 10¹⁹ m⁻³) to reduce q_target ⩽ 10 MW m⁻². The JA-DEMO higher-κ case has advantages to provide enough operation margin to the recrystallization temperature of the tungsten target. Triangles and diamonds show series of lower f^* _rad ^div (∼0.7) cases. The peak q_target is generally increased with increasing P_sep·(1 − f^* _rad ^div), and it is increased with reducing width of the plasma detachment. Figures 6(a) and (b) show profiles of T_e ^div, T_i ^div, n_e ^div and heat load components at n_e ^sep = 2.0 × 10¹⁹ m⁻³ for the JA DEMO higher-κ case with the lower f^* _rad ^div. The detachment width is decreased to 7 cm, which corresponds to inside of the near-SOL of the q_// ^Xp profile. The peak q_target is significantly increased to 12.5 MW m⁻², where the local T_e ^div increases to 32 eV, while the local n_e ^div is similar at ∼2 × 10²⁰ m⁻³. In the outer flux surfaces (r^div > 20 cm), n_e ^div is decreased to a few 10¹⁹ m⁻³ and T_e ^div and T_i ^div are increased. As a result, higher n_e ^sep operations, i.e. n_e ^sep ⩾ 2.3 × 10¹⁹ and 2.7 × 10¹⁹ m⁻³, are required for the two lower f^* _rad ^div cases.

Zoom In Zoom Out Reset image size

At the same time, while the peak q_target is already reduced, low T_e ^div and T_i ^div are required for the year-long divertor operation since the ion fluence is expected to be 50–100 times larger than ITER. Net erosion (Δd) was estimated at the peak q_target in figure 5(a) (T_e ^div ∼ T_i ^div ∼ 20 eV, n_e ^div ∼ 10²⁰ m⁻³) by a simple formula: Δd (mm) = 4.95 × 10⁻¹⁹ R_net⋅Y_i⋅C_i⋅Γ_i⋅t for W-surface, where R_net, Y_i, C_i, Γ_i and t are the ratio of net to gross erosion rates for the tungsten target, sputter yield, impurity concentration, incident ion flux and operation time (year), respectively. R_net and Y_i⋅C_i are critical factors and Y_i is significantly increased under the attached plasma condition. Assuming R_net = 0.1 and using Y_i⋅C_i ∼ 4 × 10⁻⁴ for c_Ar ^div ∼ 0.2% [23] for the relatively low T_e ^div ∼ 20 eV and Γ_i ∼ 10²³ D m⁻² s^-1 for the reference JA DEMO higher-κ and 2014 cases, Δd is estimated to be 2.5 mm after a year-long operation, which is a half of the thickness of the monoblock design (5 mm). Simulation of the plasma material interaction including physics processes of R_net will be necessary for further evaluation. Operation at further high n_e ^sep or sweep of the strike-point location will be required.

Furthermore, effect of smaller λ_q// ^mid on the divertor operation of the JA DEMO divertor was investigated with reducing both χ and D to half values, i.e. χ_e = χ_i = 0.5 m² s⁻¹, D = 0.15 m² s⁻¹ [27]. The radial gradients of T_e ^mid, T_i ^mid and n_e ^mid were increased, and λ_q// ^mid of the q_//e ^Xp + q_//i ^Xp profile was decreased to 2.2–2.6 mm. Both peak q_target and n_e ^sep were increased for each case. Lower boundary of n_e ^sep for q_target ⩽ 10 MW m⁻² was determined to be 2.0 × 10¹⁹ and 2.3 × 10¹⁹ m⁻³ for the JA DEMO higher-κ and 2014 reference cases, respectively, as shown in figure 6(d). The influence was enhanced with reducing f^* _rad ^div; both peak q_target and n_e ^sep were significantly increased for f^* _rad ^div ∼ 0.7. As a result, the divertor operation was difficult in the low n_e ^sep range of 2–3 × 10¹⁹ m⁻³. Therefore, f^* _rad ^div ∼ 0.8 was determined as the reference value of the divertor performance for the JA DEMO 2014 and higher-κ cases.

The divertor plasma performance for the EU DEMO divertor has been investigated from P_sep/R_p = 17 MW m⁻¹ (reference) to 22 MW m⁻¹, using SOLPS5.1. Here, n_e ^sep = 2.9–3.3 × 10¹⁹ m⁻³ was higher than that of JA DEMO, but lower than that of ITER. Reductions in T_e ^div and q_target were investigated with increasing the Ar seeding rate in addition to small Kr or Xe seeding. Results are shown as a function of c_Ar ^SOL in figures 7(a) and (b), where q_target includes q_t ^plasma, q_t ^rec and q_t ⁿ, but not q_t ^rad. These results are also plotted as a function of f^* _rad ^div in figure 7(e). For the reference case (P_sep = 150 MW), appropriate power handling of the divertor (q_target ⩽ 10 MW m⁻²) is achieved at f^* _rad ^div ⩾ 0.67 (c_Ar ^mid ⩾ 0.3%), which is slightly larger than ITER. At the same time, peak T_e ^div is decreased as shown in figure 7(c); the partial detachment is not clearly produced, while the peak T_e ^div at 2–6 cm away from the separatrix is decreased. Further requirement of low T_e ^div (⩽5 eV) for reducing the target erosion to ignoring level is fulfilled at f^* _rad ^div ⩾ 0.78 (c_Ar ^mid ⩾ 0.45%). Appropriate divertor operation is restricted in the large f^* _rad ^div similar to that in the JA DEMO. Furthermore, the plasma detachment (T_e ^div = 1–2 eV) will require to increase f^* _rad ^div > 0.85 (c_Ar ^mid > 0.5%). Wider c_Ar ^mid scan is performed for P_sep = 200 MW. The peak q_target is reduced to less than 10 MW m⁻² with increasing f^* _rad ^div to 0.85 (c_Ar ^SOL ∼ 1.9%), while the peak T_e ^div ∼ 40 eV in the outer region as shown in figure 7(d). Larger f^* _rad ^div ∼ 0.9 (c_Ar ^mid ∼ 2%) is required to achieve appropriate divertor operation with reducing T_e ^div ⩽ 5 eV, and 'full detachment' is produced in f^* _rad ^div ∼ 0.93 (c_Ar ^mid ∼ 2.6%).

Zoom In Zoom Out Reset image size

As a result, the divertor power handling (q_target ⩽ 10 MW m⁻²) was achieved for the three P_sep cases in the open and shallow divertor geometry, where the partial detachment were not clearly formed. Further requirement of the low T_e ^div (⩽5 eV) could be achieved at large f^* _rad ^div ∼ 0.8 and c_Ar ^mid ∼ 0.5% for the reference (ITER-level P_sep/R_p) case. It would be possible to reduce the peak q_target efficiently by improving the divertor geometry, but this is determined from physics and engineering requirements such as control of the large radiation in the long divertor leg and extending the tritium breeding blanket volume.

Plasma simulations of the JA DEMO divertor were performed also for lower P_sep ∼ 150 MW (P_out = 160 MW) in order to compare detachment characteristics in the closed and open geometries with similar P_sep. Profiles of T_e ^div, T_i ^div, n_e ^div and heat load components at the outer target for n_e ^sep = 1.9 × 10¹⁹ m⁻³ and f^* _rad ^div ∼ 0.67 are shown in figures 8(a) and (b). Both T_e ^div and T_i ^div are reduced to 1–2 eV near the separatrix (∼13 cm), and slightly wider 'partial detachment' is produced. In the attached plasma region, T_e ^div and T_i ^div are increased similar to those for the larger P_sep cases, and the peak q_target appears in the attached plasma region (r^div ∼ 16 cm), where the plasma transport component becomes dominant. Comparison of T_e ^div in the JA and EU DEMO divertor simulations is summarized as a function of f^* _rad ^div in figure 8(c); the peak q_target ⩽ 10 MW m⁻² is achieved for all results in both DEMO divertors. While T_e ^div (and q_target) near the separatrix is significantly reduced in the JA DEMO divertor, T_e ^div at the peak q_target is still high 25–28 eV for f^* _rad ^div ∼ 0.6 and 18–20 eV for f^* _rad ^div ∼ 0.7. On the other hand, the EU DEMO divertor requires f^* _rad ^div ⩾ 0.67 to reduce q_target ⩽ 10 MW m⁻², and further high f^* _rad ^div ⩾ 0.78 to reduce near-separatrix T_e ^div ⩽ 5 eV. This comparison shows benefit of the closed geometry in order to reduce q_target, T_e ^div and T_i ^div near the separatrix. At the same time, low T_e ^div and T_i ^div is required to reduce the target erosion for the year-long operation as was discussed in section 3.2. For the low P_sep case of the JA DEMO, n_e ^div is reduced from ∼1 × 10²⁰ to 2–3 × 10¹⁹ m⁻³ and T_e ^div ∼ T_i ^div ⩽ 20 eV at the peak q_target. Although net erosion of the tungsten target will be smaller than the reference cases, further improvements of the divertor geometry and operation options such as different seeding impurity will be explored in order to extend the partial detachment width and to reduce local T_e ^div and T_i ^div at the attached plasma region. At the same time, feasible values or profile of the diffusion coefficient over the near- and far-SOLs will be demanded in order to determine the divertor operation for the DEMO design.

Zoom In Zoom Out Reset image size

Consequently, owing to the different power exhaust concept of the main plasma, divertor performances with P_sep/R_p = 17 and 30–35 MW m⁻¹ for the EU and JA DEMOs, respectively, were numerically demonstrated. Some common views were achieved, (i) the peak q_target can be reduced to less than 10 MW m⁻² in the long leg divertor (L_div = 1.6 m) by increasing f^* _rad ^div larger than ITER-level (0.5–0.6), (ii) closed divertor geometry is efficient both to produce the plasma detachment and to reduce peak T_e ^div and T_i ^div, and (iii) adequate reduction of T_e ^div and T_i ^div is required by increasing f^* _rad ^div ⩾ 0.8. It was noted that net erosion rate under the partially detachment condition may become a critical lifetime issue of the DEMO divertor. Further optimization of the divertor geometry and operation parameters (f^* _rad ^div, n_e ^sep, seeding impurity species, etc) as well as determination of the appropriate physics parameters such as diffusion coefficients are necessary in both divertors. At the same time, benchmark study between updated versions of both SONIC (JA) and SOLPS (EU) codes in the same geometry and input parameters is required to discuss physics models incorporated.

Water-cooled divertor target design of the ITER technology, i.e. W-monoblock, Cu-alloy (CuCrZr) cooling pipe and Cu-interlayer, is a common baseline concept of JA and EU DEMO divertors. Recent status of the JA and EU divertor designs is summarized in table 2, which is updated from 2015–2016 designs [9], and details of each design concept are explained in section 4.2. Common design issues and requirements were identified during the BA DDA. Number of the divertor cassette (48) is now the same for both DEMOs, and three adjacent divertor cassettes are replaced through one maintenance port for the maintenance. Total weight of one JA divertor cassette (∼22 ton) is 2.7 times heavier than EU divertor cassette (8.3 ton) since the former covers most of the divertor plasma below the X-point, similar to the ITER divertor.

Neutron irradiation on the W-monoblock and Cu-alloy heat sink is significantly increased in the both DEMOs due to increasing operation time as well as the neutron flux. The maximum irradiation doses for a full power year (fpy) become larger than those for the full operation period of the ITER divertor, which are 0.54 and 2.5 displacements per atom rate (dpa), respectively [28]. Therefore, the engineering design and technology of the divertor component should consider reductions of irradiation damage and degradation of material properties until the planned replacement. Reduction in the thermal conductivity of W will be acceptable up to several dpa (∼3 years), which is comparable to replacement of the tritium breading blankets. Selection of Cu-ally (CuCrZr or Cu-base composites) for the heat sink material is owing to the excellent thermal conductivity. Design constrains under the neutron irradiation condition are summarized in table 3. Design constrains of the power handling will be firstly determined by mechanical property of Cu-alloy at high temperature (>280 °C) due to radiation induced softening and creep [29, 30]. Corrosion on the inner wall of the pipe above 200 °C is another common design issue. At the same time, radiation induced hardening at the low temperature also starts from the low irradiation dose (∼0.2 dpa). Therefore, the coolant temperature (T_cool) for the high heat flux component is increased to higher than that of ITER (∼70 °C), and the selection of T_cool (JA: 200 °C, EU: 130 °C) would be a critical lifetime issue. Comprehensive database of these properties, design criteria and their improvement are required to determine the life time of the power handling unit.

Arrangement of heat removal units and the coolant pipes in the divertor cassette for the JA DEMO, and cooling water temperature and flow velocity are shown in figures 9(a) and (b). Heat sink, i.e. CuCrZr or reduced activation ferritic martensitic (RAFM) steel (F82H) pipes, was determined based on neutronics calculation and simulation result of the heat load profile. CuCrZr heat sink units were incorporated to the inner and outer targets, which can broadly cover the high heat load region of 0.8 m near the strike-point. The neutronics calculation showed that the maximum dose was ∼1 and ∼1.5 dpa for the inner and outer targets, respectively, for year-long steady-state operation as shown in figure 8 of reference [11], thus the replacement will be expected every 1–2 years, i.e. more frequently compared to that of the blankets. Pressurized water with relatively high inlet T_cool of 200 °C (5 MPa) was used to minimize the irradiation embrittlement. Upper temperature limit for the CuCrZr pipe was provided to be 350 °C, whereas influence of thermal softening is somehow fostered above 280 °C–300 °C [32]. T_cool increases to 231 °C, and it is well below the critical heat flux (CHF) of 28 MW m⁻². Arrangement of the cooling water route for inner and outer targets was recently revised from series to parallel to supply the coolant with comparable velocity of 11–12 m s⁻¹ in order to reduce the pressure drop at the inner target with incorporating smaller number of the W-monoblock units, i.e. 32 and 43 for the inner and outer units. This is also common design in JA and EU DEMO divertors.

Zoom In Zoom Out Reset image size

The W-monoblock units with F82H pipe were arranged at higher neutron irradiation region such as baffles, dome and reflectors, where the dpa rate on the F82H pipe was increased up to 5–6, but q_target was lower than a few MW m⁻². High T_cool water (290 °C, 15 MPa) similar to the blankets [33] was used for the electricity generation by turbine system similar to a pressurized water fission reactor. In the heat exhaust evaluation of the cooling water routes as shown in figure 9(b), the total thermal power to the divertor was assumed to be 380 MW including 40%–70% margin, which was larger than the simulation result to the divertor chamber (∼270 and ∼220 MW for the JA DEMO 2014 and higher-κ cases, respectively). In addition, nuclear heating of the target units (totally 118 MW) was considered. Since maximum heat load to the divertor baffles, dome and reflectors was 1–1.5 MW m⁻² in steady-state operation, the cooling water route to the reflectors and dome was arranged by a series connection due to benefit in remote maintenance, i.e. reducing cooling pipe branches and joints.

The water-cooled target concept is selected for the EU DEMO as a baseline concept [32, 34–36], which can take advantages of the thermal conductivity and the thermohydraulic properties compared to helium gas. The divertor configuration has been revised recently to the one shown in figures 9(c) and (d). Both the inner and outer baffles were removed, and the breeding blanket was extended instead, in order to increase the tritium breeding volume. At the same time, the total weight was reduced to ITER-level (8.3 tons). Length of the vertical target was comparable to the JA divertor target (0.8 m). The coolant route was divided to the inner and outer targets. The inlet temperature was recently reduced from 150 °C to 130 °C, and the flow velocity were increased to 16 m s⁻¹ in order to increase the CHF to 48 MW m⁻². On the other hand, because of removing divertor baffles, the maximum dpa rates on the W armor and CuCrZr pipe of the target were increased to 2.0 and 7.2 dpa fpy⁻¹, respectively. Here, these dpa rates on the W armor and CuCrZr pipe are comparable to those on the divertor baffle for the JA DEMO divertor, where the dpa rate for CuCrZr is slightly larger than that for F82H ferritic steal. It is noted that the dose for the CuCrZr pipe at the outer strike point is expected to reach 2.8–3.7 dpa fpy⁻¹ for the EU-DEMO divertor [37, 38], which is larger than ∼1.2 dpa fpy⁻¹ for the JA DEMO [11], due to influence of the open and shallow divertor geometry. Reduction in the thermal conductivity of CuCrZr and Cu-interlayer is anticipated at larger than 10 dpa for multi-year operation. Degradation of the mechanical property of heat sink and joint/interlayer caused by radiation induced softening and/or hardening at the lower dpa can be minimized by application of new W and Cu/Cu-alloy target concepts as shown in section 4.4.

Cassette body design for the DEMO divertor has been developed both for EU and JA to incorporate the power exhaust units and coolant pipes, which will be consistent with reducing the fast neutron flux to protect the vacuum vesse (<2.75 dpa) and replacement of the power exhaust units. Number of the main coolant pipes is minimized to 4 for the both designs, and inlet and outlet are located at the outboard cassette. The cassette structures and cooling water condition are different, while the cassette structure can be made by welding of similar RAFM steel (Eurofer97 or F82H). EU cassette body is composed of a box structure with internal ribs, and it is cooled with another cooling loop with the inlet T_cool of 180 °C, which ensures the entire structure sufficient fracture toughness at the neutron damage (<6 dpa) [39, 40]. The dome ('liner') and reflectors are simplified from those in ITER as shown in figure 9(c). Since the exhaust slot is designed to be at the bottom of the cassette, the liner mainly has a role in reducing the neutron flux to the cassette and the bottom area of the vacuum vessel. The same coolant of the cassette body is used by a series connection as shown in figure 9(d). The interface between feeding pipes and fixation systems, and the design integration are on-going. JA cassette consists of thick (25 cm) plate structures with two lines of cylindrical coolant puddles in order to reduce the fast neutron and γ-ray fluxes. The exhaust slot is located at the outboard bottom in order to reduce the neutron flux to the vacuum vessel. The high T_cool water (290 °C, 15 MPa) is provided from the inboard and outboard manifolds of the cassette through the side routes of the cassette body to the puddles as shown in figure 9(b), and it finally exits to a manifold near the exhaust slot. The coolant velocity of 1.5 m s⁻¹ is enough to remove the total nuclear heat of 0.7 MW in one cassette body, corresponding to totally 32 MW for 48 cassettes.

Analysis of the heat transport and stress on the power exhaust unit of the JA DEMO target was carried out with three-dimensional modelling, considering the divertor heat load components in the partial detachment. The peak q_target of 10 MW m⁻²-level on the flat target consists of combination of the plasma (along the magnetic field line) and radiation (perpendicular to the target) loadings. Figure 10(a) illustrates geometry of the fish scale W-monoblock target and heat flux components. The monoblock size in the toroidal direction, the gap width, the height between the adjacent surfaces, and the inner diameter of the coolant pipe were 25 mm, 0.5 mm, 0.4 mm, and 12 mm, respectively, which were similar dimension of the ITER monoblock target. Since the total incident angle of the magnetic field line to the fish scale target surface (2.4°) is smaller than ITER, the plasma-wetted area corresponds to 63% of the downstream side of the target. Two cases of the heat load profiles were given for heat transport analysis: the maximum W-surface temperatures correspond to ∼1200 °C (case-1), and ∼1400 °C (case-2), which appear near the downstream edge. The former and the latter correspond to just below and well above the critical temperature of W-recrystallization, and the peak q_target at the plasma-wetted area are 13.5 and 15.2 MW m⁻², respectively, as shown in figures 10(b) and (c). It is noted that these values correspond to the peak q_target = 9.1 and 10.8 MW m⁻² for the flat target. Nuclear heat in the W-monoblock is included, which is less than 10% of the total heat load. The maximum temperatures of the CuCrZr pipe, 351 °C and 365 °C, respectively, appear at the upper surface. Mechanical toughness of the cooling pipe is also near critical against thermal softening, thus, stress and elasto-plastic analysis were performed for the CuCrZr pipe as a structure material.

Zoom In Zoom Out Reset image size

The heat fluxes up to 18 and 22 MW m⁻² are widely distributed on the water side of the cooling pipe for case-1 and case-2, respectively, as shown in figures 10(d) and (e). These values are well below the CHF (28 MW m⁻²): the heat fluxes from the pipe to the coolant correspond to 64% and 79% of the CHF, which are acceptable for the heat removal. As a result, the target design with T_cool of 200 °C can handle the peak q_taret of 10 MW m⁻² level including the nuclear heat, while less than 9 MW m⁻² is preferable to minimize the surface recrystallization.

Residual stress and strain is also evaluated with repeating the high heat flux, where the surface temperature changes between base-temperature of 207 °C and peak value at each location as shown in figure 10(c). Here, zero strain is assumed at the brazing process (950 °C) as the initial condition of the analysis history. Result of elasto-plastic stress analysis in the extreme condition of the case-2 is summarized. Thermal stress of the W-monoblock is increased particularly at the upper and the side inner surfaces: Tresca-stress corresponded to 680 MPa. At the same time, some peaks appear at the upper inner surface and side outer surfaces of the CuCrZr pipe: Tresca-stress corresponds to 280 and 305 MPa, respectively. Figure 10(f) shows that elastic plus plastic strain, i.e. mechanical strain, in the pipe axis (z)-direction (ε_Z ) is increased at the inner (expansion) and outer (compression) of the CuCrZr pipe. The expansion is seen in the heat sink under the monoblock with increasing the heat flux. Thus, the compression of the heat sink is caused between the monoblocks. Stress (σ_Z )–strain(ε_Z ) trace at the maximum stress location shows similar trajectory during 15 heat load cycles in figure 10(g), but the maximum compression stress is gradually increasing. Change in ε_Z was ∼0.25%, which may not be critical for life time issue. Experimental study of fatigue will be required at the operation condition.

Several technologies for the high heat flux target have been developed in EU [34] to survive under DEMO-relevant neutron irradiation condition. All are based on W as a baseline armour material, and Cu-alloy pipe with swirl tape to increase the heat transfer at the pipe wall. Material properties of CuCrZr and Cu-interlayer are anticipated to limit the performance of the target heat sink, e.g. irradiation creep above 350 °C and irradiation embrittlement below 250 °C [32]. Thus, it is reinforced by various kinds of Cu-W composite materials and novel interlayer materials as shown in figure 11 [41].

• (a)  
  ITER-like W-monoblock with a CuCrZr cooling pipe is a baseline concept also in the DEMO divertor, but the monoblock size is modified to decrease the cross section width to 23 mm, instead of 28 mm in ITER, while the axis thickness of 12 mm is the same, in order to reduce thermal stresses. The armor thickness of 8 mm to the cooling pipe is comparable.
• (b)  
  Thermal break concept (developed in CCFC) is based on ITER W-monoblock target. The concept features cut-outs in the Cu-interlayer in the area of the highest heat flux to achieve a more uniform distribution of the thermal flux around the cooling pipe circumference.
• (c)  
  Composite pipe concept (developed in IPP) is based on a W wire-reinforced Cu composite pipe, which expects better strength of the cooling pipe, in particular, at high temperature (>350 °C).
• (d)  
  Functionally graded W/Cu (FGM) interlayer concept (developed in CEA) aims replacing thick Cu interlayer, which is expected to be fully embrittled by fast neutron due to segregation of transmuted He at grain boundaries. Graded thin (30 μm) and think (500 μm) W–Cu interlayer is used to improve joining strength.
• (e)  
  Flat-tile concept with a composite heat sink block (developed in IPP): W particle-reinforced Cu composite block is expected to enhance the mechanical resilience of the irradiated heat sink against structure failure. The use of the W–Cu composite is supposed to reduce the thermal expansion mismatch between the armor tile and the heat sink block.
• (f)  
  In addition, He cooling target using multi-jet pipe by W and W-laminate (developed in KIT) is also developed as an option [36].

Zoom In Zoom Out Reset image size

Recently, small-scale mock-ups of each concept have been fabricated and tested by means of hydrogen neutral beam in the GLADIS under the high power density at 20 MW m⁻² up to 500 cycles (130 °C coolant, which is expected operation temperature). A good production quality and reliable high-heat-flux fatigue performance was demonstrated. All mock-ups showed intact structural integrity and stable heat removal capacity over the entire loading cycles. Modest roughening of the W-armour surface and swelling of the blocks due to inelastic deformation was found. The metallographic examination revealed that the upper half of the armour blocks were completely recrystalized, but no discernable cracks were found in any of the tested armor (290 blocks). Furthermore, other ITER-like mock-ups were tested at 25 MW m⁻² up to 500 pulses (20 °C coolant) for overload tests. Also under the increased power density, the mock-ups exhibited intact structural integrity and stable heat removal capacity, even though there was a single fine crack (6 mm in depth) initiated from the armor surface which underwent a pronounced inelastic deformation leading to overall severe roughening. The mock-ups even withstood the overload up to 32 MW m⁻² (5 pulses) without any detrimental impact nor melting. The future R&D will be focused on an upscaling trial towards a medium-scale manufacturing (40 cm length) using the W wire-reinforced composite pipe.