Development of pulsed plasma operation scenario and required conditions in JA DEMO

We have developed the pulsed plasma operation scenarios for JA DEMO, a design concept of the steady-state tokamak demonstration reactor, to clarify controls of the current profile and power required for the operation. We compare the scenarios when injecting electron cyclotron waves only and both neutral beam and electron cyclotron waves for external heating and current drive. We demonstrate current profile control that maintains the minimum value of the safety factor above one and avoids creating the local minima in the safety factor profile and power control by argon seeding that maintains the fusion power constant at the desired value and reduces the heat load on the divertor, performing long-time integrated modeling simulations. We clarify the conditions of the heating and current drive system and impurity injection system required for such control. The dependence of power control on argon anomalous transport coefficients is investigated. We have the prospect of maintaining the fusion power of 1 GW for more than two hours, i.e. obtaining the required plasma performance determined using a systems code.


Introduction
The conceptual design activity for a steady-state tokamak demonstration reactor, JA DEMO [1], has been conducted in Japan.The pulsed operation for two hours is also supposed for commissioning and early demonstration of power generation with lower plasma performance than that for the steadystate operation, obtaining hundreds of MWe of the net electric output.Table 1 shows the main parameters of JA DEMO for the steady-state and pulsed operations.Plasma operation parameters are limited by the specifications of engineering components, including the suppliable magnetic flux by the central solenoid (CS) coils, converter voltage of the poloidal field (PF) coils, allowable heat flux to the plasma-facing components, and usable actuators.For JA DEMO, the suppliable flux by the CS coils is approximately 300 Wb [2], and the allowable heat flux to the divertor plates is approximately 10 MW m −2 [3].In reactor design activities, it is necessary to develop plasma operation scenarios that can sustain and control the plasma performance required as design parameters over the required time for constructing a feasible reactor concept and for determining the component designs.Previously, ohmic plasma initiation with PF coil supply voltage similar to ITER [4], flux consumption reduction during plasma current ramp-up phase by electron cyclotron (EC) heating [5], external heating power required for the L-H transition [6], and fusion power control by pellet injection [7] have been studied for JA DEMO.
The previous study on the steady-state plasma operation scenario has evaluated the external input power required for full non-inductive current drive and pointed out the importance of off-axis EC current drive for controlling the internal transport barriers (ITBs) [1].A series of scenario development studies have identified plasma operation consistent with the given specifications of engineering components or identified the specifications of engineering components required for possible plasma operation.In the EU DEMO design activities, identification of the requirements for actuators [8] and the limiter design for first wall protection [9] have been studied based on the developed plasma operation scenarios and control simulations [10,11] including off-normal transient events, developing a modeling tool labeled as 'flight simulator' [12] for these purposes.This paper considers the pulsed plasma operation in JA DEMO.
The main target of control in DEMO pulsed plasma operation is to maintain a constant fusion power, which is affected by the current profile through particle and thermal transport.It would be preferable to control the current profile so that the minimum value of the safety factor does not fall below one to avoid sawtooth oscillations that fluctuate the fusion power.ITBs can be formed when the safety factor profile has local minima away from the magnetic axis.Unintended ITB formation should be avoided in terms of fusion power control and magnetohydrodynamic (MHD) stability.In a plasma of JA DEMO pulsed operation, the current penetration time is estimated to be thousands of seconds; therefore, it would be necessary to actively control the current profile simultaneously with the fusion power over most of the time of a single pulsed discharge.Scenario development studies for a designstage reactor should clarify the selection of the heating and current drive scheme and the specifications of the heating and current drive system to realize such current profile control.In JA DEMO, neutral beam injection (NBI) and EC are supposed to be used from the previous study on steady-state operation [1].In EU DEMO, during the burn phase, the heating and current drive system is used for temperature and MHD controls and is not supposed to be used for the current drive [8].
In JA DEMO, on the other hand, since 30% of the plasma current is required to be driven externally, the heating and current drive system should be capable of controlling the current profile during the burn phase, in addition to MHD control by local current drive.
Intentional impurity injection into the core plasma is a primary approach for reducing the plasma loss power across the separatrix, resulting in the heat load to the divertor plates [13], and is one of the features of DEMO plasmas.On the other hand, impurity seeding causes fuel dilution and radiation cooling, resulting in a reduction of the fusion power.It is necessary to develop impurity injection scenarios that maintain the heat load to the divertor plates below the tolerance and the required fusion power constant.
In this paper, we have developed pulsed plasma operation scenarios for JA DEMO using an integrated modeling code to clarify controls of the current profile and power required for pulsed operation.We compare the scenarios when using EC only and both NBI and EC for external heating and current drive.We discuss the points of current profile control for each heating and current drive scheme and power control to maintain the fusion power and suppress the net plasma loss power by impurity seeding based on long-time integrated modeling simulations.The sensitivities of the impurity density at the separatrix required for maintaining fusion power to impurity anomalous transport coefficients are investigated.The burn time, defined as the time possible to maintain the fusion power of 1 GW, is estimated for each scenario in terms of the magnetic flux consumption.We have obtained the requirements of heating and current drive and impurity injection systems for JA DEMO pulsed operation.

Analysis model
The operation scenarios are studied using the integrated modeling code TOPICS [14].The TOPICS code is based on a 1.5-dimensional transport code consisting of the equations of one-dimensional heat, particle, and current transport and two-dimensional free boundary equilibrium and cannot simulate magnetic control.In this study, we solve the time evolutions of the electron and ion temperature profiles, the current profile, the density profile of impurities injected intentionally, and the magnetic equilibrium, prescribing the electron density profile evolution.The temperature profile in the H-mode is given in the region of ρ ⩾ 0.85 by the hyperbolic tangent function [6] and solved in the core region (ρ < 0.85), whereas the temperature profile in the L-mode is solved over the entire plasma region.Here, ρ is the normalized minor radius.As for the turbulent thermal transport model, we use the Bohm-gyroBohm model [15] of the version that was used for validation and verification studies performed using CRONOS [16] and TOPICS and experimental data in JET and JT-60U [17,18].This model considers the formation and sustainment of ITBs by multiplying the Bohm term by a shear function where ω E×B is the E × B shearing rate, γ ITG is the growth rate of ion temperature gradient modes, and s is the magnetic shear.Neoclassical transport coefficients are calculated by the matrix inversion model [19,20].Absorbed NBI heating power and NBI current drive are calculated by the one-dimensional Fokker-Planck model [21].Absorbed heating power and current drive by EC are calculated by ray tracing and solving the adjoint equation for the full relativistic Fokker-Planck equation [22].The density profile of the injected impurities is calculated using IMPACT [23], the impurity transport analysis module of TOPICS.In this study, the impurity transport in the scrape-off layer is not considered, whereas the profile inside the separatrix is solved with the prescribed fraction of the impurity to electron densities at the separatrix.Neoclassical particle transport coefficients for impurities are calculated by NCLASS [24].Anomalous transport coefficients for impurities are prescribed with arbitrary profiles that are constant with time.
The NBI system is assumed to consist of three deuterium beam injectors based on negative-ion sources.The port-through power per injector and beam energy are 33 MW and 1.5 MeV, respectively.The tangency radius of injection is chosen to be 8.5 m to provide current drive in the center to obtain the positive magnetic shear profile, supposing operation with a sufficiently high density such that shinethrough is not a problem.The design of the EC launcher system for JA DEMO has not yet been determined; therefore, we assume that all EC waves are injected from a single common position.The injection position coordinates are the major radius R = 11.5 m, which is 0.58 m away from the outboard side equatorial position of the last closed flux surface, and the height Z = 0.9 m, the same as that of the magnetic axis.
The times at which the L-H transition occurs and at which pedestal growth stops are determined by an L-H transition model [6].In the H-mode, the pedestal density and temperature are assumed to be 0.85 times the Greenwald density limit and 3 keV, respectively; the pedestal is within the MHD stable region [6] which is obtained from the ideal MHD stability code MARG2D [25].The density profile is assumed to be flat at the time when pedestal growth stops, to become a peaked profile 20 s after pedestal growth stops, and then to remain the profile.The electron density at the separatrix is assumed to be 2.5 × 10 19 m −3 .The helium density is determined by a global particle confinement time, τ He = 5τ E , where τ E is the energy confinement time.The density profile of main ions is determined by quasi-neutrality, and the deuteron and triton densities are assumed to be the same.Argon (Ar) is considered as the impurity species injected intentionally to control the net plasma loss power across the separatrix and the fusion power.The fraction of the Ar density to the electron density at the separatrix, f Ar,sep , is set to zero until 66 s, to rise to 0.1% by 67 s, and then to be varied so that the fusion power maintains a desired value (∼1 GW), and that the net plasma loss power maintains below the tolerance (∼250 MW [3,26]).We assume that the anomalous diffusion coefficient for Ar, D Ar,ano , is uniform, and the convection velocity, v Ar,ano , is proportional to the normalized minor radius.As the typical value, D Ar,ano = 2.0 m 2 s −1 and v Ar,ano = −0.25 m s −1 at the separatrix are assumed; these values are comparable to those evaluated in ITER [27].

Comparison of scenarios when using EC only and using NBI and EC
The time evolutions of (a) current, (b) power, and (c) normalized parameters, and the profiles of (d) density, temperature, safety factor, and (e) current density at 1800 s of the developed pulsed plasma operation scenarios are shown for the cases of using EC only and both NBI and EC in figures 1 and 2, respectively.Table 2 shows the obtained values of main parameters at 1800 s for the cases of using EC only and using NBI and EC with the reference values [1] which were obtained from the systems code TPC [28].Here, I p is the plasma current, I OH is the ohmic current, I BS is the bootstrap current, I EC is the EC-driven current, I NBI is the NBI-driven current, P EC is the EC injection power, P NBI is the NBI power, P α is the alpha heating power, P rad is the radiation loss power which is the sum of the loss powers of bremsstrahlung radiation, synchrotron radiation, and line radiation from Ar, P sep is the net plasma loss power across the separatrix, q min is the minimum value of the safety factor, H H is the confinement enhancement factor defined as the energy confinement time normalized by the IPB98(y,2) scaling [29], β N is the normalized beta, l i is the normalized internal inductance, n e is the electron density, T e(i) is the electron (ion) temperature, q is the safety factor, j tot is the total current density, j OH is the ohmic current density, j BS is the bootstrap current density, j EC is the EC-driven current density, j NBI is the NBI-driven current density, P fus = 5P α is the fusion power, P CD = P EC + P NBI , Q = P fus /P CD , f GW is the line-averaged electron density normalized by the Greenwald density limit, f BS = I BS /I p , f CD = (I EC + I NBI )/I p , f rad = P rad /(P CD + P α ), ⟨ f Ar ⟩ is the fraction of the volumeaveraged Ar density to the volume-averaged electron density, and Z eff is the volume-averaged effective charge.Four times the normalized internal inductance, 4l i , is shown as an indicator of the beta limit [30].
The simulations are performed, setting the time at which the divertor configuration is formed with the plasma current of 3.3 MA to be 0 s and maintaining the plasma current for the first five seconds.Scenarios in which the plasma current flat-top phase starts at 65 s are considered.In the plasma current ramp-up phase until 65 s, the electron density is assumed to be 0.6 times the Greenwald density limit, and 8.5 MW of EC is applied to reduce the flux consumption so that the safety factor profile becomes the same as that when EC is not applied by using the method proposed in [5].Intense NBI and/or EC injection start at 66 s.In the cases of using EC only and both NBI and EC, the L-H transition occurs at 70.7 and 70.5 s, pedestal growth stops at 82.4 and 79.6, and the electron density becomes the profiles shown in figures 1(d) and 2(d)  at 102.4 and 99.6 s, respectively.The net plasma loss power, P sep , can be raised more slowly by slowing the central density rise if required for divertor plasma operation and divertor plate protection.
The scenarios shown in figures 1 and 2 were developed to satisfy the following: Q = 13 and P fus ∼ 1 GW are obtained, the maximum value of P sep is less than 250 MW, q min > 1 is maintained, and no strong ITBs are formed such that P fus and β N spike.The main parameters mostly satisfy the required plasma performance shown as the reference values in both the scenarios of figures 1 and 2. Especially in the case of using EC only, f CD is low and does not reach 2/3 times the reference value if P EC is fixed at ∼80 MW to adjust P fus and Q to the reference values.The reference value of f CD is based on the assumption of the use of NBI only.Although a higher f CD is preferable for economic efficiency, a lower f CD Table 2. Obtained values of main parameters at 1800 s for the cases of using EC only and using NBI and EC and the reference values [ The plasma size of JA DEMO is a similar level to that of EU DEMO whose major radius and minor radius are 9 m and 2.9 m, respectively.The essential difference is that JA DEMO is a steady-state reactor concept and EU DEMO is a pulsed reactor; however, many similarities in assumptions and results can be found between the scenarios of the JA DEMO pulsed operation and the EU DEMO plasma operation [10,11].The elongation, triangularity, toroidal field on the axis, volume-averaged electron density, volume-averaged temperature, and burn length are almost identical in both reactors.The plasma current and the minor radius are 1.44 and 1.2 times larger in EU DEMO than in JA DEMO, respectively; therefore, the Greenwald density limit is the same.The main differences between the two concepts are the non-inductive current drive fraction and radiation loss power fraction.In the JA DEMO pulsed plasma operation, the non-inductive current drive fraction ( f BS + f CD ) is 0.66 and 0.75 when using EC only and using NBI and EC, respectively.In EU DEMO, the noninductive current drive fraction is approximately 0.4 because the external current drive is essentially not considered during the burn phase.Therefore, the requirements for the JA DEMO heating and current drive system should be determined by the current drive and current profile control during the burn phase, whereas it is necessary to largely reduce the flux consumption in the current ramp-up phase for the EU DEMO heating and current drive system.The radiation loss power fraction in the core plasma and Z eff are 0.6-0.7 and 2.12, respectively, in EU DEMO, while they are approximately 0.3 and 1.6, respectively, in JA DEMO.EU DEMO has more severe requirements for radiation power control in the core plasma than JA DEMO, and selecting higher-Z impurities, such as xenon, is preferable to reduce Z eff and ensure sufficient burn length and fusion power [31].Because the required radiation power fraction in the core is low in JA DEMO, we mainly consider using Ar, which enables the radiation cooling from the divertor to edge plasma by one impurity species [3], placing importance on controllability.

Current profile control and requirements for heating and current drive system
Figure 3 shows the profiles of (a) EC-driven current density and (b) ohmic current density at 85, 200, 300, 600, 1100, and 1800 s in the case of using EC only.The EC-driven current density profiles shown in figure 3(a) are obtained by combining 16 EC injection conditions shown in table 3 with adequate balances of injection power, where f EC is the EC wave frequency and θ tor(pol) is the injection angle in the toroidal (poloidal) direction whose definition is shown in figure 4. When the density and temperature profiles largely vary by a sudden increase in external heating power and the L-H transition, the Shafranov shift and Doppler effect vary the position in ρ at which the current is driven by EC of a certain injection condition with time.In such a case, the EC-driven current profile control requires precise diagnostics and prediction of the current profile and the changes in frequency, injection angle, and power in fine steps with short time intervals.In the scenario of figure 1, to avoid this complicated control, the heating is carried out by combining EC#1-8 so that the current is not driven from the ramp-up phase until a few seconds after pedestal growth stops (85 s).After 85 s, the current drive is performed by combining EC#9-16.While the ohmic current penetrates to a certain extent (until 200 s), the current is driven by EC at ρ ∼ 0.6, where the driving position hardly changes because changes in density and temperature are small.Figure 5  shows the profiles of safety factor and total current density at 65, 85, and 105 s.From before starting the intense EC injection (65 s) to after forming the peaked density profile (105 s), the total current profile inside ρ = 0.5 is not largely changed, and the positive magnetic shear profile is maintained.The EC current drive efficiency, i.e. the driven current per injection power, is higher when the current is driven closer to the plasma center where the electron temperature is higher.After 200 s, the driven current profile should be shifted towards the magnetic axis to increase the current drive efficiency.The EC-driven current is necessary to be modest in the region of ρ ∼ 0.1-0.4,where the bootstrap current is large, resulting in j EC at 300 s with two peaks at the center and at ρ ∼ 0.55.Depending on the ohmic current penetration, the EC-driven current profile is adjusted with time to avoid q min < 1 in the plasma center region and to avoid creating local minima on the safety factor profile, i.e. to avoid ITB formation.In the center region, the driven current is reduced as the ohmic current increases.The peak of the driven current at ρ ∼ 0.45-0.55 is shifted towards the center to increase f CD as the ohmic current profile narrows.
Figure 6 shows the time evolution of the injection power (a) from 15 to 85 s with the conditions of EC#1-8 and (b) from 85 to 1800 s with the conditions of EC#9-16.To obtain the time evolution of the current profile discussed above, i.e. to maintain q min > 1 and avoid the ITB formation, the injection power for each EC condition was manually adjusted and determined to be the waveform shown in figure 6.
Figure 7 shows the profiles of (a) NBI-driven current density, EC-driven current density, and (b) ohmic current density at 85, 200, 300, 800, 1000, and 1800 s in the case of using both NBI and EC.The EC injection conditions shown in table 3 are used also in this case.For the same reasons as the case of using EC only, EC#1-6 are used to heat the plasma without the current drive until 85 s, and EC#14 and 15 are used to drive the current in the region around ρ = 0.55 from 85 s to 200 s.After 200 s, EC drives the current in the region around ρ ∼ 0.4-0.5,where the NBI-driven current is small, with a combination of EC#11-14.To utilize the high current drive efficiency, the NBI power is increased as high as possible; however, decreasing the NBI power is required depending on the ohmic current penetration to maintain q min > 1, and the EC power is increased so that P CD = 80 MW.In the previous study [32], a scenario was developed using three NBIs with different vertical injection angles which are set within the limits allowed by the current port design.In this case, the possible NBI-driven current profile was so narrow in the center region that q min could not be maintained above one when significant NBI power was injected to utilize the high current drive efficiency after the ohmic current penetrated sufficiently to the center.The previous scenario was required to match the case of using EC only, eventually, to maintain q min > 1.To obtain sufficient off-axis injection and a wider NBI-driven current profile, we used three types of NBI whose ports are tilted downward with different angles, fixing the tangency radius of injection of 8.5 m, as shown in figure 8, for the scenario of figure 2. These ports locate different toroidal angular positions from each other and do not interfere with other structures around them including the toroidal field and PF coils.As significant NBI power can be sustained using these NBIs shown in figure 8, a higher f CD is obtained than that in the case of using EC only, as shown in table 2, maintaining q min > 1.  shows the time evolution of the injection power of (a) NBI#1 and 2 and (b) EC#1-6 from 15 to 85 s and (c) NBI#1-3 and (d) EC#11-15 from 85 to 1800 s.In the same manner as the case of using EC only, the time evolution of the injection power of each condition of NBI and EC was determined by manual adjustment to maintain q min > 1 and avoid the ITB formation.

Power control by impurity seeding
Figure 10 shows the time evolutions of P EC , P NBI , P α , P sep , P rad , f Ar,sep , and ⟨ f Ar ⟩ for the cases of (a) using EC only and (b) using NBI and EC.To suppress the heat flux to the divertor plates below the tolerance of 10 MW m −2 , P sep must be kept below approximately 250 MW by injecting impurities into the core plasma [3,26].The core electron density shown in figures 1(d) and 2(d) was determined so that P fus ∼ 1 GW and P sep < 250 MW are compatible.As the current penetration time is thousands of seconds, P fus does not become steady-state while the current profile is changing, even if P CD and density are constant.In this study, as well as suppressing P sep below 250 MW, Ar injection is also used for control such that P fus = 1.04 GW (= QP CD when Q = 13 and P CD = 80 MW) is maintained.If f Ar,sep is kept constant, P fus increases with time due to current penetration and control of the externally driven current profile.After P fus increases to 1.04 GW (P α = 208 MW), P fus is maintained by increasing f Ar,sep gradually.Although f Ar,sep = 0.1% at the beginning of intense external heating in the scenarios shown, the L-H transition succeeds even when f Ar,sep ∼ 0.4% [6].If P fus and P sep are required to be increased more gradually, Ar seeding can be started at a higher f Ar,sep .At 1800 s, f Ar,sep required to maintain P fus = 1.04 GW is 0.4%-0.45%.Compared to reference values, higher f rad can be obtained despite lower ⟨ f Ar ⟩, i.e. lower fuel dilution as seen in table 2. Eventually, P sep becomes sufficiently below the acceptable value in the scenarios shown.By increasing P fus more gradually, it would be possible to decrease the Ar density fraction and core electron density, i.e. to increase f CD .For the same electron density and P CD , higher f Ar,sep is required when using NBI and EC than when using EC only due to the contribution of NBI to ion heating.In the scenario when using NBI and EC, 30% of the NBI power is absorbed by the ions from 100 to 1800 s; the ion heating power by NBI is 17 and 8.6 MW at 150 and 1800 s, respectively.
Figure 11 shows the dependence of f Ar,sep required for control of P fus = 1.04 GW at 1800 s on D Ar,ano and v Ar,ano at the separatrix in the case of using EC only.When v Ar,ano = 0 m s −1 , the required f Ar,sep decreases with increasing D Ar,ano .For the cases of v Ar,ano = −0.125 and −0.25 m s −1 , the required f Ar,sep is a convex upward function of D Ar,ano .When v Ar,ano = −0.5 m s −1 , the required f Ar,sep increases with increasing D Ar,ano .For v Ar,ano ̸ = 0 m s −1 , the Ar density profile peaks at the center and is flattened with increasing D Ar,ano .If the electron density profile and f Ar,sep are fixed, the fusion power increases by the flattening of the Ar density profile because it increases the fuel ion density fraction and decreases the radiation loss power.Increasing f Ar,sep is needed to reduce the excess fusion power.As D Ar,ano increases further, the average charge state of Ar in the pedestal region decreases due to increased Ar transport from the separatrix.This increases the radiation loss power and decreases the fusion power; therefore, f Ar,sep is needed to be decreased.In this study, a uniform D Ar,ano profile and a linear v Ar,ano profile were assumed.The profiles of the anomalous transport coefficients can influence the f Ar,sep required to reduce P sep and maintain P fus .Understanding impurity transport is important for accurate power control and impurity injection system design.

Estimation of burn time
We estimate the burn time, defined as the time possible to maintain the fusion power of 1 GW, using the scenarios until 1800 s in terms of magnetic flux consumption.The suppliable flux by the CS coils is 318 Wb.The flux consumption for plasma initiation and current ramp-up until I p = 3.3 MA, i.e. before 0 s in the scenarios of figures 1 and 2, is assumed to be 65 Wb [5].The current ramp-up from 3.3 to 12.3 MA consumes 147.5 Wb of magnetic flux in the scenarios.Therefore, the usable flux in the flat-top phase (after 65 s) is estimated to be 105.5 Wb. Figure 12 shows the time evolutions of the flux consumption in the flat-top phase, Ψ(t) − Ψ(65 s), and of the change rate of the flux consumption, dΨ /dt, for the cases of using EC only and using NBI and EC; here, t is the time.In the case of using EC only, the flux consumed until 1800 s is 28.2 Wb, and the change rate of the flux at 1800 s is 0.0126 Wb s −1 .After 1800 s, assuming a constant change rate of the flux, further operation for 6135 s [= (105.5 − 28.2)/0.0126] is possible.Considering that P fus reaches a sufficient value at 100 s in the scenario of figure 1, the burn time is estimated to be 2.18 h (= 6135 s + 1700 s).In the case of using NBI and EC, the flux consumed until 1800 s is 20.6 Wb, and the change rate of the flux at 1800 s is 0.008 50 Wb s −1 .In the same manner as the estimation for the case of using EC only, the burn time is estimated to be 3.25 h.In both cases, the estimated burn time could be expected to be an underestimate because the change rate of the flux consumption still shows a decreasing trend at 1800 s.The CS coils for JA DEMO have been designed supposing two-hour pulsed operation.We have obtained the prospect that the two-hour operation is sufficiently possible in both the cases of using EC only and using NBI and EC.

Discussion on the feasibility of JA DEMO plasma operation and future work
In the scenarios shown in figures 1 and 2, the current profile was controlled by combining ECs of different frequencies in 10 GHz steps at almost the same injection angle (∼40 • ), as shown in table 3. From a cost perspective, it is desirable to use a common vacuum window for the EC system, which requires current profile control by combining ECs of different frequencies in approximately 30 GHz steps.Figure 13 shows the current profiles driven by the 1 MW injection of EC#10-13 and EC#15-16 shown in table 3 and similar profiles driven by different injection conditions.We can obtain a current profile equivalent to that driven by EC injection with a combination of frequency f EC and injection angle θ tor by selecting a combination of lower frequency and smaller injection angle.Current profile control in the scenarios of figures 1 and 2 could be achieved by combining ECs with frequencies of 160, 190, and 220 GHz.
In this study, current profile control was examined by changing the balance of injection power of the EC conditions listed in table 3 with time.Since the installable total EC power is limited in terms of ensuring the blanket space to provide a sufficient tritium breeding ratio, current profile control should be achieved by changing the frequency and injection angle of each gyrotron with time instead of adjusting the balance of injection power between fixed EC conditions.The EC system for JA DEMO would be required to include variable frequency gyrotron and remote steering of the injection angle.To avoid complicated current profile control from the current ramp-up to the completion of the L-H transition, EC focused its role on plasma heating using EC#1-8 without the current drive.The difference in the injection angles between EC#1-8 and EC#9-16 would be beyond the range of injection angles expected to be able to be varied by remote steering systems.Fewer ports for EC injection are preferable for increasing blanket space, and it is reasonable to use EC conditions with similar injection angles before and after the L-H transition.It is important to develop a current profile control method that can respond to fast plasma profile changes and Shafranov shift for allowing the use of similar EC conditions before and after the L-H transition and for maximizing the burn time and tritium breeding ratio.
In this study, the balance of the injection power between the conditions of NBIs and ECs shown in figures 6 and 9 was manually determined to maintain q min > 1 and avoid the ITB formation.Therefore, the current profile was not optimized in the scenarios shown in figures 1 and 2. The current profile should be optimized to maximize the non-inductive current drive fraction while maintaining q min > 1 and avoiding the ITB formation.As changes in the externally driven current profile vary the ohmic and bootstrap current profiles, control of the total current profile by the external drive is complicated.It is important to develop methods for evaluating the optimum current profile and determining the commands to the actuators to obtain the optimum profile, which can respond to fast plasma changes including the L-H transition, by evolving the methods proposed previously such as seen in [33][34][35].Small amplitude MHD instabilities may be possible to be used actively for control.If the amplitude is small, the sawtooth oscillation might be useful to control the current profile in the center and even to selectively exhaust the helium ash [36].In that case, the fine, severe control of q min by external current drive, as was performed in the scenarios shown, is no longer necessary.
We have studied fusion power control by Ar injection, focusing on suppressing the gradual power increase due to current diffusion.The time evolution of the Ar density fraction at the separatrix is an input parameter, adjusted to maintain P fus = 1.04 GW, in the simulations and can be interpreted as a result of transport in the core plasma and scrape-off layer and an Ar gas puff rate.The waveforms of f Ar,sep obtained in this study provide information on the requirements for the Ar injection system for JA DEMO.The fusion power is maintained almost constant to the target value in the simulations as shown in figure 10.This can be achieved because perturbation that requires active control and simulation of pellet injection are not considered.The total heating power is set to be almost constant, and the gradual increase in the fusion power due to current diffusion is suppressed by the gradual increase in Ar density in the scenarios shown.In actual control, the fusion power cannot be maintained at a strictly constant value, and the fast variation of the fusion power should be avoided.To avoid and respond to the fast variation, establishing active burn control mechanisms, including heating and current drive, fuelling, and impurity injection systems, is necessary.
In addition to intentionally injected impurities, thermalized helium generated by the fusion reaction and tungsten (W) from the first wall exist in the core plasma.These impurities can affect the reactor feasibility and can change the aspect of power control by Ar seeding.The impact of W concentration on the ITER plasma operation has been studied [37][38][39].The EU DEMO plasma control simulation was performed when 1 mg of W was injected into the core [10].We estimate the effect of W concentration on plasma performance in JA DEMO pulsed operation by simple calculations which use the electron density and temperature profiles at 1800 s of the scenario when using EC only and assume that the W density profile is proportional to the electron density profile and that Ar density is zero.Figure 14 shows the dependence of P rad and Z eff on the W density fraction, n W /n e .In the scenario when using EC only, P rad = 82.0MW and Z eff = 1.59 at 1800 s.The radiation loss power can be adjusted by reducing the Ar density fraction when n W /n e < 4.5 × 10 −5 .When n W /n e < 4.5 × 10 −5 , Z eff is below 1.59, and the fusion power above 1 GW can be obtained.However, the n W /n e limit becomes lower than 4.5 × 10 −5 because the Ar density fraction at the separatrix cannot be zero due to Ar seeding in the divertor plasma.To assess the W concentration limit, simulations consistent with the divertor plasma are necessary, as well as modeling of W source and transport; these are future work.Magnetic control has not been considered in the simulations.We have assumed the flux consumed in the plasma initiation and current ramp-up phases before the formation of the divertor configuration and a constant plasma shape.The current values are largely imbalanced between the CS coils because the plasma shape is not optimized, and the limit for PF coil supply voltage is not imposed.To evaluate the burn time more accurately, magnetic control simulations should be performed from the beginning of the discharge, considering the plasma shape evolution, the specifications of the PF coil system, and magnetic diagnostics.Our future work includes developing modeling codes and scenarios by combining magnetic control simulation.
We have developed the scenarios which should be the target for JA DEMO operation so that they achieve plasma performance close to the results of the systems code.These scenarios confirm the possibility of obtaining the required plasma performance and are useful for studying the control and actuator specifications required for such operations.On the other hand, the scenarios developed are based on various assumptions.To ensure that performance sufficient as DEMO can be obtained in actual JA DEMO operations, it is important to clarify the dependence of plasma performance on various parameters with wide ranges, as has been studied for the ITER operation [39][40][41].In this study, the time evolution of the electron density profile was assumed.The peaking factor of the density profile in the burn phase is n e0 /⟨n e ⟩ = 1.57and n e2 /⟨n e ⟩ = 1.47,where n e0(2) is the electron density at ρ = 0(0.2) and ⟨n e ⟩ is the volume-averaged electron density.With the conditions at 1800 s of the scenario when using EC only, the peaking factor is calculated with the three scaling formulae proposed by Angioni et al [42] to be n e2 /⟨n e ⟩ = 1.57, 0.99, and 1.67.The edge density limit is calculated using the scaling law proposed by Giacomin et al [43] to be 3.1 × 10 20 m −3 which corresponds to 4.7 times the Greenwald density limit.Therefore, various electron density profiles appear to be possible.The pedestal density and temperature were fixed at 0.85 times the Greenwald density limit and 3 keV, respectively; however, there are various combinations of MHD stable pedestal density and temperature, which are evaluated in [6].The scenarios should also be considered where H H is lower than 1.2 obtained in the scenarios shown.The peaking factor of the density profile, pedestal density, pedestal temperature, and possible value of H H affect the plasma performance and, therefore, the injection power for the current drive and the Ar density fraction required for power control.Our future work also includes the prediction of the density profile and investigation of the dependence of the plasma performance on particle and thermal transport models.

Conclusion
We have developed the pulsed plasma operation scenarios for JA DEMO when using EC only and using NBI and EC for heating and current drive, using the integrated modeling code TOPICS.We have discussed the points of current profile and power control under the situation where the required plasma performance, which was obtained from the systems code TPC, is almost achieved.Although the use of NBI is more advantageous in terms of burn time, we have the prospect of a burn time of more than two hours even when using only EC, which has relatively fewer long-term research and development issues, for early demonstration of fusion power generation.
Based on the long-time simulations, we have examined the current profile control which maintains the minimum value of the safety factor above one and avoids creating local minima in the safety factor profile away from the magnetic axis.The externally driven current profiles are necessary to be made fine adjustments with time depending on the ohmic current penetration over most of the time of a single pulsed discharge.NBI is advantageous for obtaining a high external current drive fraction, while EC is advantageous for flexible current profile control.We have clarified the heating and current drive conditions required for the current profile control for JA DEMO.The EC injection conditions required for the control are frequencies of 160-220 GHz and the injection angle in the toroidal direction of approximately 40 • .To obtain a wide NBI-driven current profile, it is effective to tilt the beamlines downwards, avoiding interfering with surrounding structures such as the toroidal field and PF coils.An important issue is the development of control methods which can evaluate the optimum current profile and determine the commands to the actuators to achieve the optimum profile, even when the plasma profile rapidly changes.
We have demonstrated power control by Ar seeding, which maintains the net plasma loss power across the separatrix below 250 MW and the fusion power at 1.04 GW.Even if the external heating and current drive power and electron density are held constant, the fusion power is increased by the current profile change, through thermal transport.In this study, the fusion power was maintained constant by increasing the Ar density with time.The fraction of the Ar density to electron density at the separatrix is required to be varied from 0.1 to ∼0.4%-0.45%for such control.The required Ar density fraction is higher when using NBI and EC than that for the case of EC only because of the contribution of NBI to the ion heating.The sensitivity of the Ar density fraction to the Ar anomalous transport coefficients was investigated with the diffusion coefficient ranging from 0.5 to 4.0 m 2 s −1 and the convection velocity ranging from −0.5 to 0 m s −1 .The required Ar density fraction and sensitivity obtained will be utilized to determine the specifications of the impurity injection system into the core plasma.Unintentionally injected impurities such as thermalized helium and tungsten also affect the power balance.Understanding impurity transport is important for accurate power control and impurity injection system design.

Figure 1 .
Figure 1.Pulsed plasma operation scenario when using EC only.The time evolutions of (a) current, (b) power, and (c) normalized parameters, and the profiles of (d) density, temperature, safety factor, and (e) current density.

Figure 2 .
Figure 2. Pulsed plasma operation scenario using NBI and EC.The time evolutions of (a) current, (b) power, and (c) normalized parameters, and the profiles of (d) density, temperature, safety factor, and (e) current density.

Figure 3 .
Figure 3. Profiles of (a) EC-driven and (b) ohmic current densities in the case of using EC only.

Figure 4 .
Figure 4. Definitions of EC injection angles in (a) the poloidal plane and (b) the horizontal plane viewed from above.The point P indicates the departure point of EC waves in ray tracing simulation.

Figure 5 .
Figure 5. Profiles of safety factor and total current density at 65, 85, and 105 s in the case of using EC only.

Figure 7 .
Figure 7. Profiles of (a) NBI-driven, EC-driven, and (b) ohmic current densities in the case of using NBI and EC.

Figure 8 .
Figure 8. Beamlines and NBI ports tilted downward used for the case of using NBI and EC.

Figure 10 .
Figure 10.Time evolutions of P EC , P NBI , Pα, Psep, P rad , f Ar,sep , and ⟨ f Ar ⟩ for the cases of using (a) EC only and (b) NBI and EC.

Figure 11 .
Figure 11.Dependence of f Ar,sep required for control of P fus = 1.04 GW at 1800 s on D Ar,ano and v Ar,ano at the separatrix in the case of using EC only.

Figure 12 .
Figure 12.Time evolutions of the flux consumption in the flat-top phase and the change rate of the flux consumption for the cases of using EC only and using NBI and EC.

Figure 13 .
Figure 13.Dashed lines show the current profiles driven by the 1 MW injection of EC#10-13 and EC#15-16 shown in table 3. Solid lines show the current profiles similar to those drawn by dashed lines, driven by the 1 MW injection of different injection conditions.

Figure 14 .
Figure 14.Dependence of the radiation loss power and effective charge on the tungsten density fraction.

Table 3 .
EC injection conditions required for current profile control in JA DEMO pulsed operation.Here, f EC is the EC wave frequency in GHz, θ tor(pol) is the injection angle in the toroidal (poloidal) direction in degrees as shown in figure4.