Physics and Engineering Studies of the Advanced Divertor for a Fusion Reactor

1 FIP/3-4Ra Physics and Engineering Studies of the Advanced Divertor for a Fusion Reactor N. Asakura 1, K. Hoshino 1, H. Utoh 1, K. Shinya 2, K. Shimizu 3, S. Tokunaga 1, Y.Someya 1, K. Tobita 1, N. Ohno 4, M. Kobayashi 5, H. Tanaka 5 1 Japan Atomic Energy Agency, Rokkasho, Aomori 039-3212 Japan 2 Toshiba Nuclear Engineering Services Co., Yokohama, 235-8523 Japan 3 Japan Atomic Energy Agency, Naka, Ibaraki 311-0193 Japan 4 Graduate School of Engineering, Nagoya Univ., Nagoya 464-8603 Japan 5 National Institute for Fusion Science, Toki, 509-5292 Japan E-mail contact of main author: asakura.nobuyuki@jaea.go.jp Abstract. A short super-x ertor is proposed as a new option for fusion ertor, where field line length from the ertor null to the outer target was largely increased (more than two times) in a similar size of conventional ertor. Physics and engineering design studies have progressed. Minimal number of the ertor coils (1 or 2) were installed inside the toroidal field coil, i.e. interlink-winding (interlink). Arrangement of the poloidal field coils and their currents were determined, taking into account of the engineering design such as vacuum vessel and the neutron shield structures, and maintenance scenario of the ertor and blankets. Divertor plasma simulation showed that large radiation region is produced between the super-x null and the target, and the plasma temperature becomes low (1-2 ev) both at the inner and outer ertors, i.e. fully detached plasma was obtained efficiently. 1. Introduction Magnetic configurations of advanced ertor were recently proposed in order to handle large exhausted power for the tokamak reactor [1-3], and two concepts of super-x ertor (SXD) and snowflake ertor (SFD) were proposed, which have advantages to increase both the field line length in the ertor and magnetic flux expansion in the ertor. Therefore, the radiation power and detachment in the ertor will be enhanced, comparing to the conventional ertor. These concepts have been demonstrated in tokamak experiments [4-6]. SXD and SFD concepts have been investigated [7] for an ITER-size compact Demo reactor of SlimCS (I p = 16.7 MA, B t = 6.0 T, R p = 5.5 m, a p = 2.1 m, and the fusion power of 3 GW level) [8], as new options of the Demo ertor design. For their applications to the fusion reactor, both the plasma performance and engineering design should be investigated. In Sec. 2, the magnetic equilibriums of SXD and SFD and their ertor coil arrangements for a Demo reactor are compared. In Sec. 3, a conceptual design of the short super-x ertor is proposed for the DEMO application. The ertor plasma characteristics calculated by SONIC simulation are presented in Sec. 4, and physics and engineering studies of the short super-x ertor are summarized in Sec. 5. 2. Plasma equilibriums and coil arrangements of advanced ertors for Demo Plasma equilibrium for SlimCS was calculated using the equilibrium calculation code, TOSCA, by solving the Grad-Shafranov equation with given a plasma current profile. For the

2 FIP/3-4Ra reference plasma equilibrium with the conventional ertor, all eight poloidal field coils (PFC) were arranged outside the toroidal field coils (TFC), and the PFC currents were determined to minimize least square errors at the plasma boundary, corresponding to shaping parameters of the plasma center location (R p = 5.45m, Z p =1.0m), minor radius (a p = 2.05m), upper an lower elongations (κ up = 1.6, κ down = 2.3), upper and lower triangularities (δ up = 0.68, δ down = 0.3), and the location of the ertor null point (R X, Z X ). TOSCA code for the advanced ertor has been developed as is described in Reference 7. Number of PFCs is increased to nine, and arrangements of PFCs and current distributions were investigated in order to produce plasma equilibriums for the SXD and SFD. In order to produce the super-x (SX) null, two parameters, i.e. location of the super-x null (R SX, Z SX ) and a ratio of the poloidal flux at the super-x null to that at the separatrix (f SX = [ψ SX -ψ ax ]/[ ψ sep -ψ ax ], where f SX < 1 for the separatrix extending outboard) are introduced. Figure 1 shows an example of the SXD magnetic configuration with R SX = 7.0 m and f SX = 0.95. One ertor coil is added and three ertor coils play an important role for control of the SXD configuration: the direction of No. 8 poloidal field coil, PFC(8), current is reversed to produce the SX null, and currents of PFC(7) and PFC(9) are largely increased to produce the ertor and SX nulls at their input locations. An image of the ertor geometry is also shown, while it is overlapped on the SlimCS sector structure. SOL field lines corresponding to the separatrix distances (Δr mid ) of 1, 3, 5 and 10 cm at the outer midplane are shown, and opening of the ertor baffle is designed such that the SOL field lines for Δr mid < 3 and 5 cm are connected to the inner and outer targets, respectively. It was found that large ertor coil current such as more than 100 MA-turns (MAT) is required for the care of Fig. 1. Plasma equilibrium for super-x ertor with R SX = 6.0 m, f SX = 0.95. All PFCs are installed outside TFC. PFC current distribution is shown (negative and positive values correspond to the forward and reversal directions of the plasma current). installation outside TFC. It is necessary to install some of them inside the TFC, i.e. interlinkwinding (interlink coil), in order to reduce the coil currents appropriate for engineering design comparable to the plasma current (16.7 MA). At the same time, the ertor size such as leg length and sub-ertor volume is restricted comparable to that of the conventional ertor since remote replacement of the ertor cassette is a crucial for Demo design. Examples of the SXD and SFD were shown in Fig. 2 (a) and (b), respectively. Here, structures (ertor cassette, neutron shield and vessel) are not considered firstly. For the SXD case, location of the SX-null is shifted closer to the ertor null, R SX = 6.0 m as shown in Fig. 2(a), and the outer ertor target is located at R = 6.7 m outer the SX null. We call short super-x ertor. Since the flux expansion in the SXD is increased from the ertor null to the SX-null, ertor opening is extended wider to connect the SOL flux surfaces (Δr mid < 5cm) to the target. The large radiation loss and resulting reduction in the plasma temperature will be enhanced in a ertor size comparable to the conventional one. For the SFD, a single ertor null is produced by merging two magnetic nulls. Flux expansion and connection length are increased both at the inner and outer ertors, particularly near the ertor null, expecting to reduce T e, and to enhance n e and P rad

3 FIP/3-4Ra Fig. 2 Plasma equilibriums for (a) short super-x ertor with R SX = 6.0 m, f SX = 0.95. and (b) snowfrake ertor. Locations of four PFCs are arranged inside TFC to obtain high triangularity plasma shape with δ up (0.68) and δ down (0.3). Current distributions are shown. near the ertor null. Thus, reduction in the peak heat flux will be expected in a compact ertor, compared to the SXD. TABLE I summarizes the magnetic and target geometry parameters for the SlimCS reference and long-leg ertors, and SXD and SFD. For the Demo ertor, small f exp is a disadvantage since the ertor coil is located further away from the magnetic null compared to ITER (f exp = 5-6). In a simple ertor model, longer L // reduces the ertor plasma temperature (T ), and the peak q target is decreased with increasing the wet area (A wet ), where mid A wet = 2πR λ q f exp /sin(θ ), R is the major radius of the strike-point, and a mid characteristic width of the heat flux profile at the midplane (λ q ) is assumed to be 5 mm. For the conventional ertor in SlimCS, L // is increased 1.5 times longer than ITER, thus θ is reduced to 18 (25 in ITER) in order to increase A wet. For the short-sxd concept, while the bottom of the ertor is similar to the SlimCS reference ertor, poloidal length of the separatrix from the ertor null to the outer target, L p, extends comparable to the long-leg ertor (2.5 m). On the other hand, magnetic field lines curve outboard, which increases θ at the target. Therefore, the target inclination, θ target, is slightly increased from 30 to 35, and A wet ~1.3 m 2 is comparable to that for the long-leg ertor. In the SXD, f exp and L // are enhanced near the separatrix (Δr mid < 1 cm) as shown in Fig. 3. Larger f exp of 8-11 is produced through the ertor leg, and L // becomes 57 m, which is 1.5 and 1.3 times longer than the reference and long-leg ertors, respectively. As a result, short SXD increases f exp along the Geometry parameters Conventional reference Conventional long-leg Short SXD (f SX =0.95) SFD f exp (at outer target) 4.0 2.4 4.2 1.6 R (at target, m) 5.8 6.4 6.6 6.4 θ (poloidal fieldline angle at target) 18 18 45 20 φ (fieldline angle at target) 1.8 2.6 4.8 6.1 A wet for λ mid q = 5 mm (m 2 ) 2.4 1.5 1.3 0.9 L p (from. null to target, m) 1.72 2.48 2.57 2.0 L // (from. null to target, m) 36 44 57 68 TABLE I: Magnetic and target geometry parameters of the outer ertor for SlimCS conventional reference ertor and long-leg ertor, SXD with f SX = 0.95, and SFD.

4 FIP/3-4Ra ertor leg and L // longer than the long-leg ertor in the similar size of the reference ertor. Control of the plasma detachment in the ertor is the most important requirement of the ertor physics concept for Demo. Increment of the connection length in the ertor (L // ) is an important advantage in order to decrease the plasma temperature (T ), i.e. T is decreased with q // 10/7 /(n u ) 2 (L // ) 4/7 in a simple point model, where q // and n u are parallel heat flux and upstream plasma density, respectively. The advanced ertor designs are interested in control of large radiation and detachment. For the SFD, both f exp and L // are increased significantly near the ertor null. While f exp becomes less than that for the conventional ertor in L p > 0.8 m, L // of 68 m is 1.8 and 1.5 times larger than the reference and longleg conventional ertors, respectively. From the viewpoint of the connection length, the compact ertor design is expected. On the other hand, f exp of 1.6 and A wet of 0.9 m 2 at the target are smaller than the conventional and SX ertors. The SFD concept will be appropriate for compact ertor. The SFD magnetic configuration requires ertor and CS coil currents larger than short-sxd and wider baffle opening. Control for the SF-null and the plasma shaping must be developed. Larger distance between the inner and outer baffles (ertor baffle opening) is a disadvantage for particle and impurity retention in the ertor chamber. 3. Short Super-X ertor for Demo Fig. 3. Connection length (at Δr mid = 1 mm) from the ertor null and flux expansion as a function of poloidal length in the conventional, short-sx and SF ertors. Further investigation of the coil arrangement will be required to takes into account of structures such as ertor cassette, neutron shield and vacuum vessel, maintenance ports for the ertor and blanket. At the same time, effects of the advanced ertor on the plasma detachment and impurity transport must be determined by the SONIC simulation in order to determine the appropriate ertor geometry for the Demo reactor design. It is important to determine locations of the ertor coils and the current distribution, appropriate to produce the SX or SF null and the plasma shaping. Minimum number of the interlink-coil are necessary, and short-sxd is preferable to design 1-2 interlink coils with smaller coil currents and to control the ertor-null and plasma shaping such as κ and δ, comparing to SFD. 3. 1. Proposal of short super-x ertor for Demo A short super-x ertor is proposed as a new option for Demo ertor, which is shown in Fig. 4. Number of interlink ertor coils is decreased from three to two, i.e. PFC(8) and PFC(9), where TFC size is Fig. 4 Arrangement of short SXD and PFCs, where two interlink ertor coils are incorporated. Plasma equilibrium with f SX = 0.99 and current distribution for SlimCS Demo reactor. Structures image of blanket modules, vacuum vessel, neutron shield and maintenance ports are also shown.

5 FIP/3-4Ra extended to the lower, and the interlink coils are installed below vacuum vessel inside TFC. For the SX null location of R SX = 5.9 m and Z SX = -5.1 m, they correspond to -23 MAT and +20 MAT, respectively. The currents of PFC(8) and PFC(9) become larger than those in Fig. 2(a). The interlink coils are made of Nb 3 Al superconductors with 1.5 m width, and the currents slightly larger than the plasma current will be acceptable with developments of engineering issues as described in Sec. 3.2. The conceptual design takes into account of structures such as ertor cassette, neutron shield (60 cm in thickness) and vacuum vessel (20 cm in thickness). Coil current of the external PFC(7) is large in order to extend magnetic field lines outboard. Since the maintenance scenario so call the multi-module segment will be applied in the next reactor design rather than the sector [9], the coil arrangement also considers maintenance ports for the ertor from horizontal port and blanket from the vertical port. It is also found that the total PFC number is reduced to 8 with only one interlink of 19 MAT, while current of the external ertor coil is -59MAT and the plasma shaping control is restricted. We investigate the short-sxd with two interlink coils. Figure 5 (a) and (b) show surface of the short SXD and the plasma equilibriums for f SX = 0.95 and 0.99, respectively. The bottom plate location of the outer ertor (Z = -5.0 m) is the same as the outer V- shaped corner of the reference ertor, and the corner is shifted from R = 6.1 m to 6.9 m in the short SXD. At the same time, the outer ertor target and baffle are extended outward to introduce the magnetic flux surface at the midplane radius of r mid = 5 cm for f SX = 0.99 as shown in Fig. 5 (b). The short SXD has an advantage to change f exp and L //, in particular, near the SX null. With increasing f SX from 0.95 to 0.99, f exp is also increased from 6.5 to 8.3, while θ is increased from 47 to 57. Here, these values are slightly different from the previous study since locations of the interlink ertor coils are far from the mid ertor null and the target. A wet corresponding to λ q = 5 mm increases slightly from 1.9 to 2.1 m 2, which are 1.2-1.3 times larger than the long-leg ertor (1.6 m 2 ). Figure 5(c) shows L // and f exp just outer the separatrix at r mid = 1 mm as a function of L p. For the case of f SX = 0.99, f exp is increased to 20 near the SX-null (L p = 1.5 2.0 m) and L // becomes larger though the ertor to the target. Consequently, L // for the SXD becomes 67-91 m (1.75-2.3 times) longer than that for the long-leg ertor (L // = 45 m). Effect of the long L // in the ertor is important rather than the wet area at the target under the condition of the ertor detachment. The ertor plasma performance is summarized in Sec. 4. 3. 2. Engineering study of short super-x ertor for Demo Fig. 5 Plasma equilibriums in the short super-x ertor for (a) f SX = 0.95, (b) f SX = 0.99. SOL magnetic field lines of midplane radii (r mid ) of 0, 1, 3, 5, 10 cm are shown. (c) Connection length along the field line from the ertor null and flux expansion as a function of poloidal length. Requirement of the interlink superconductor coil is investigated under the maximum toroidal and vertical magnetic fields of B t = 10 T and B v = 6T. Nb 3 Al superconductor is preferable for

6 FIP/3-4Ra the interlink ertor coil than Nb 3 Sn, which is used such as ITER, because of less load ratio less than 50% of allowable stress (500MPa) and preferable for React and Wind process [10]. In the application to the Demo design, small diameter of the superconductor filament is required such as 1 µm, which is obtained in Nb 3 Sn superconductor, in order to reduce the AC loss by a factor of 1/36. Figure 6 shows an acceptable cross-section design of the superconductor coil: width of 21.7 mm and the maximum current is 4.6 KA, which consists of the superconductor strand (diameter of 0.83 mm), SS conduit, insulator and supercritical helium flow channel (diameter of 5 mm) at the center. Cabling pattern of the superconducting strand is described as (((2SC+1Cu)x3)x4)x4. The total width and turns of the interlink coil are expected to be 1.6 m and 5436, the maximum current of which corresponds to 25 MA. Fig. 6 cross-section design of Nb3Al superconductor coil. Figure 7 shows assembly image of Nb 3 Al superconductor (SC) coil. The interlink SC can be inserted though space between TF coils below the allowable bending strain of Nb3Al, which is 0.4% (bending radius of 2 m). The conductor is wound by rotating a CS bobbin near the midplane. After insertion of the insulator, the interlink coil is implicated in the support on the TFC. Stress analysis under the magnetic field shows that the maximum Tresca stress on CS coil of less than 250MPa is dominant in the toroidal direction, which is acceptable in 25 MAT level. On the other hand, electromagnetic-force on IL- PFC(8) becomes 500-600 MN downward since the average B r at the coil is 0.67T, although that on IL-PFC(9) is small due to the average B r of zero. The load on TFCs should be reduced by changing the IL-coil location or by installing additional support of the IL-coil. Arrangement and fabrication studies, and feedthrough design will be future issues. 4. SONIC simulation for a short super-x ertor 4.1. Calculation parameters and mesh arrangement The power handling scenario of the large exhaust power of P out = 500 MW into the coreedge boundary (at r/a = 0.95) has been simulated in the SlimCS conventional ertor [11], using SONIC code, where 92 % of Pout (P rad /P out = 0.92) is radiated in the edge, SOL and ertor by impurity seeding. SONIC simulation for the short SXD started in calculation meshes for f sx = 0.99, with the same plasma parameters for the conventional ertor, i.e. P out = 500 MW, n i = 7x10 19 m -3 at the core-edge boundary, and diffusion coefficients of χ i = χ e = 1 m 2 s -1, D = 0.3 m 2 s -1 are the same as ITER simulation [12]. The calculation mesh is shown in Fig. 8. The plasma transport in SOL and ertor is calculated in the flux surfaces of r mid < 2.9 cm, and it is also calculated in the private flux region. Pumping slot of the ertor cassette is located at the bottom due to extension of the outer ertor, while the height of the ertor cassette is comparable to the conventional ertor. Exhaust slot in the outer ertor is designed near the SX-null for the first case. 4.2 Radiation distribution in the ertor Fig. 7 winding image of interlink superconductor coils

7 FIP/3-4Ra The calculation result for the same radiation fraction of P rad /P out = 0.92 was obtained with the smaller seeding rate of 1.1x10 21 s -1 compared to the conventional ertor (1.43 x10 21 s -1 ). Plasma density was increased widely over both the inner and outer target plates, and large radiation power from Ar impurity was seen at the upstream of the outer targets. The large radiation region is further upstream in the outer ertor. Radiation powers in the inner and outer ertors are P -in -out rad =143 MW, P rad = 210 MW for the short SXD are larger than those in the conventional ertor, i.e. 127 MW and 179 MW respectively, while P SOL rad = 57 MW is smaller than 109 MW. Retention of the impurity ions in the short SXD is efficient more than that of the conventional ertor. Distributions of T e, n i and radiation power density in the outer ertor are shown in Fig. 9. The large radiation region changed in the flux surfaces, i.e. it is seen above the SX null (further upstream of the target) near the separatrix, while it is above the target in the outer flux surfaces. Recycling fluxes of impurity and particle are enhanced along the long field lines near the SX null, and T e is decreased to 1-2 ev. Consequently, the large radiation region is further upstream and the plasma detachment is produced near the SX null in the sort SXD. Figure 10 shows profiles of n e, T e, T i and the heat load components at the outer ertor target. Low T e and T i (~1 ev), i.e. full detachment, are observed widely over the target. The total heat load (q target ) is evaluated including radiation power load (q rad target ) and neutral load (q n0 target ), in addition to the plasma heat load plus surface recombination, pl i.e. q target = γ n d C sd T d plus q rec target = n d C sd E ion, where γ, C sd, n d, T d, E ion are sheath transmission, plasma sound velocity, density and temperature at the ertor sheath, recombination energy, respectively. The peak q target of 7.7 MWm -2 is seen near the separatrix (r ~4 cm). In the full detachment ertor, q rec target (3.0 MWm -2 n0 ) and q target (2.1 MWm -2 pl ) are larger than q target (1.8 MWm -2 ) near the separatrix. q rad target (0.8 MWm -2 ) is small since the large radiation region is further upstream. On the other hand, q rad target is dominant at the outer flux surfaces (r > 10 cm) and q target is less than 4.5 MWm -2. In the short SXD, the radiation power is increased in the ertor and the detachment front can be maintained at further upstream of the target, compared to the conventional ertor. It is fond that the short SXD will be enough to reduce the outer q target less than the engineering design level of 10 MWm -2. On the other hand, the peak q target at the inner ertor is larger than Fig. 8 Divertor geometry and calculation mesh (lower half) for SlimCS short super-x ertor. Locations of Ar impurity and fuel gas injections and exhaust route are shown. Pumping slot of the ertor cassette is located at the bottom. Fig. 9 Enlarged distributions of (a) ion density, (b) electron temperature, (c) Ar radiation power density in the outer short SXD.

8 FIP/3-4Ra the heat load, which is attributed to the larger plasma and neutral fluxes while T e and T i are low. Investigation to control the radiation region and detachment front at further upstream is necessary. 5. Summary Advanced ertor study will provide new options for the Demo reactor design. Short super-x ertor concept was applied in the outer ertor, which can be incorporated in a conventional ertor cassette. Arrangement of PFCs was investigated under engineering restrictions. Installation of two interlink ertor coils was required and the maximum coil current was 20 MAT level for the plasma current of 16.7MA (SlimCS). Plasma equilibrium study determined the basic ertor geometry and advantages. Connection length and flux expansion in the ertor are increased 1.5 2.0 times longer and 2-3 times larger than the similar size of the conventional ertor. Plasma performance and power exhaust were investigated by SONIC simulation. In the short SXD, impurity retention in the ertor was improved compared to the conventional ertor, and full detachment (T e = T i ~1 ev) was produced, where the detachment front can be maintained at further upstream of the outer target. It is fond that the short SXD is enough to reduce the outer q target less than the engineering design level of 10 MW/m 2. Design of the inner ertor target and exhaust slots should be improved to control the radiation region and reduced the peak heat load. Acknowledgments This work was carried out within the framework of the Broader Approach DEMO Design Activity, and was partly supported by Grant-in-Aid for Scientific Researche (No. 25420899). References Fig. 10 (a) Profiles of n e, T e and T i, (b) the heat load components at outer ertor target. [1] M. Kotschenreuther, et al. Phys. Plasmas 14, 072502 (2007). [2] D. Ryutov, et al. Phys. Plas., 14, 064502 (2007). [3] D. Ryutov, et al. Phys. Plas., 15, 092501 (2008). [4] F. Piras, et al., Plasma Phys. Control. Fusion 51, 055009 (2009). [5] M.V. Umansky, et. al., Nucl. Fusion (2011). [6] V.A. Soukhanovskii, R.E. Bell, A. Diallo, J. Nucl. Mater. 438 S96 (2013). [7] N. Asakura, et al. Trans. Fus. Sci. Tech. 63, 70 (2013). [8] K. Tobita, et al., Nucl. Fusion 49, 075029 (2009). [9] H. Utoh, et al., Evaluation of remote maintenance schemes by plasma equilibrium analysis in Tokamak DEMO reactor, Fusion Eng. Des. 89 (2014) In Press. [10] H. Utoh, et al. Fusion Eng. Des. 89 (2014) 2456. [11] N. Asakura, et al. Nucl. Fusion, 53, 123013 (2013). [12] A. Kukushkin, et al, J. Nucl. Mater. 438, S203 (2013).