THE PLASMA SCENARIOS FOR THE SPHERICAL TOKAMAK FOR ENERGY PRODUCTION (STEP) AND THEIR TECHNICAL IMPLICATIONS H. MEYER FOR THE STEP TEAM United Kingdom Atomic Energy Authority Abingdon, Oxon, United Kingdom Email: [email protected] Abstract The Spherical Tokamak for Energy Production (STEP) [1] is a fast-paced UK national programme within UKAEA that aims to construct the first prototype tritium self-sufficient magnetic confinement fusion power plant by 2040 delivering Pel ≥ 100 MW to the grid. To achieve this ambitious timeline high technical risks need to be balanced against the overall cost and schedule of the STEP prototype power plant (SPP). The spherical tokamak (ST) concept has great potential to provide a compact, cost-effective solution due to the combination of high natural elongation, κ~3, high normalised β, βN~4 − 5, moderate toroidal field, Bt(Rgeo) = 3.2 T and low aspect ratio A < 2 since fusion power scales as Pfus ∝ 1/A (κβNBt)⁴. SPP design has followed the NASA process based on concept maturity levels (CML)[2] and Technology Readiness Level (TRL). Each concept is assessed with respect to a set of measures of effectiveness with a successive down selection from 40+ concepts to the baseline concept (CML5) with major radius Rgeo ≈ 3.6 m, A = 1.8, plasma current Ip ≈ 21 MA, auxiliary power Paux ≃ 150 MW and 1.5 GW ≤ Pfus ≤ 1.8 GW. Central to the SPP design are the plasma scenarios that have been developed in the last 4 years. Four baseline fully non-inductive (NI) burning plasma flat-top operating points (FTOPs) with Q > 10 are dominated by electromagnetic turbulence arising from kinetic ballooning (KBM) and micro tearing (MTM) modes [3,4] rather than the electrostatic turbulence commonly observed in present day conventional aspect ratio devices. Neither well-validated predictive reduced or surrogate models nor reliable converged nonlinear gyrokinetic simulations of KBM and MTM turbulence are readily available leading to a substantial uncertainty in the plasma confinement and therefore the device size used in the design process [5]. This contribution will summarise the design process, give an overview of the baseline concept and discuss the key decisions that have led to this design. The focus will be on plasma scenario design for DT operation [6], the solutions to the key plasma challenges and their technical implications for the tokamak and wider power plant design. This is the first time an integrated ST power plant plasma scenario and engineering design have been studied in such detail. 1. INTRODUCTION The STEP programme aims to deliver the first magnetic confinement fusion power plant targeting 2040 for the completion of the build for phase 1 of four operational phases. The key requirements are that the design should be based on the spherical tokamak (ST), the plant must be tritium self-sufficient and should produce about Pel ≈ 100 MW net electricity. To achieve this a more technological affine culture is adopted in favour for device cost. To explore the concept design space in an integrated way the NASA based concept maturity level (CML) process [2,7] was followed. In the initial phase about 46 whole plant design points were generated with the system code PROCESS [8], the most promising undergoing a further refinement and analysis. Iterations with a fast 1.5 D plasma workflow (see Figure 1) were used to refine the process run for about 15 concepts. Using the plant optimiser BLUEMIRA, simplified 3D tokamak designs were achieved that served as starting points for dedicated design sprints to assess quickly potential showstoppers and choose between different technologies for the most promising concepts. This process delivered a preferred concept in early 2023 and culminated recently in the final internal review at the concept stage (CML5). The different concepts are compared using a set of predefined measures of effectiveness that provide an integrated assessments of all aspects of the power plant. The plasma scenario is central to the plant and often drives the design. During the first four years of the STEP programme considerable progress has been made in understanding the scenario design space for STEP [6,9], although predictive capability in several key areas is still missing. In these areas special attention has been given to parameter sensitivities to achieve enough margin in the design to cope with these uncertainties. The plasma scenario design, its challenges and critical design impacts are discussed in this paper. After introducing the plasma design process and key scenario assumptions in section 2, the core transport is discussed in section 3. The exhaust (section 4) is a particular challenge in a compact tokamak as well as the active control (section 5) needed to achieve the required high, κ, and normalised beta, βN = β(%)aBt/Ip = 2μ0 ⟨p⟩a/(BtIp) (a: minor radius, Bt: toroidal field, Ip: plasma current, ⟨p⟩: volume-averaged pressure) . High κ and βN are required to compensate for the lower Bt due to the limited space in the central part of an ST, as the fusion power scales with Pfus ∝ 1/A (κβNBt)⁴ [10]. The FTOP must be fully non-inductive (NI) in an ST and probably any commercially attractive magnetic confinement fusion power plant. This requires an efficient heating and current drive technique (section 6) and in the ST also a largely non-inductive current ramp phases (section 7). 2. DESIGN PROCESS AND FLAT TOP OPERATING POINTS The plasma design process [6] is shown schematically in Figure 1. It is centred around the JINTRAC integrated scenario modelling suite [11] and the 1.5 D core transport module JETTO that allows integrated scenario modelling with different levels of fidelity in the modules. A fast workflow (Figure 1 left) was used for quick FTOP optimisation and the mapping out of the scenario design space [9]. Parallel to the integrated scenario modelling, higher fidelity simulations are carried out for the exhaust, core transport, pedestal physics and the HCD to inform or develop lower fidelity models in JINTRAC/JETTO as well as separate assessments of plasma stability, control and fast particle confinement. In addition, higher fidelity JINTRAC modelling is used where necessary to calibrate the simplified assumptions around the scenario. Currently JETTO is used as an assumption integrator as there is no predictive core transport model for STEP available (see section 3). Profiles consistent with the sources and sinks are generated (see Figure 2) with a gyro-Bohm (gB) transport model with dominant electron transport and ions close to neoclassical, motivated by observations in current STs [12] as well as gyro-kinetic (GK) simulations. The gB transport coefficients are scaled to match the pre-defined βN yielding 1.5 ≤ Pfus ≤ 1.8. Nevertheless, using “predictive” models developed and benchmarked on current STs (MAST and NSTX) the gB transport assumption has been partially verified, although the models mimic transport driven by micro-tearing modes (MTM) whilst GK modelling suggests that transport from kinetic ballooning modes (KBM) is dominant [4] (see section 4). The simple predictive modelling was also used to study some sensitivities (see Figure 3). The concept design sprints have led to the selection of fully NI FTOPs with positive triangularity and 4 ≤ βN ≤ 4.4, marginally above the no-wall limit, βN_NW ~ 3.9 and well below the ideal wall limit βN_IW ~ 6. To access 2nd stability for ballooning modes and avoid low m/n neoclassical tearing modes, the safety factor, q, profile is optimised to be largely monotonic with q_min > 2. The maximisation of κ and βN_IW, whilst maintaining good divertor control, requires a broad current profile with low internal inductance 0.25 ≤ li(3) ≤ 0.3 and off axis current drive using microwaves in the electron cyclotron frequency range. The magnetic field and plasma density were optimised to allow HCD access with both electron cyclotron (EC) and electron Bernstein (EB) waves [13]. Full wave modelling confirms that a scenario with pure EB HCD is not viable as the Doppler shift of the resonance at fusion-relevant temperatures prevents access to the inner 40% of the minor radius and core current drive is needed to prevent the formation of a current hole in a NI plasma. In all cases the entire HCD power is used also for current drive. The energy confinement time, τE = Wpl / Pheat (Wpl: stored energy, Pheat: total heating power), in these calculations is an assumption, which can be evaluated by comparing it against the value, τE,scal, derived from common empirical scaling laws. The average, ⟨H⟩ = (Hscal + Hscal*)/2 between H-factors with, Hscal = τE / τE,scal, and without Hscal*, subtracting the core radiation is used to account for the uncertainty in correct treatment of the core radiation that is an instant loss for the power going to the electrons. All but one FTOP are optimised to minimise ⟨HCD⟩ [14], which drives the solution to high Greenwald fraction, fGW [9]. The value of ⟨Hscal⟩ itself is not very significant for STEP as none of the confinement scaling laws are likely to be applicable to the STEP parameter range. The data sets are dominated by plasmas with different transport characteristics (section 3) even for scaling laws developed specifically for STs [15,16]. These are derived from sparse data sets of highly rotating neutral beam heated ST plasmas. The rotational shear will strongly suppress the KBM transport considered important for STEP. Hence, they imply a far better confinement than has been assumed for STEP giving 0.47 ≤ ⟨HST1⟩ ≤ 0.57 for [15] and 0.18 ≤ ⟨HST2⟩ ≤ 0.31 for [16]. The impurities (Xe, Ar and He) are treated self-consistently assuming neoclassical transport for the high-Z impurities and for He the same transport as for D and T. The Ar concentration is set to cAr(ρ = 1) ≈ 0.5% at the separatrix, the Xe concentration is adjusted to get a radiation fraction of frad ≈ 0.7 and the He source is calculated from the fusion reaction. Table 1 Typical parameters for FTOPs for SPP-001 using different assumptions and HCD mixes (numbers in brackets reflect runs without Ar as impurity): - Major Radius Rgeo [m]: 3.60 - Aspect ratio A: 1.8 - Toroidal field Bt(Rgeo) [T]: 3.2 - Triangularity/Elongation δ/κ: 0.5 / 3 - EC-HD: Ip = 20.5 (20.6) MA, Pfus = 1.53 (1.60) GW, Pel = 67 (95) MW, PEC = 150 MW, PEB = 0 MW, Prad = 311 (321) MW, Q = 10.2 (10.7), βN = 4.4, fBS = 0.88, fGW = 0.95 (0.94), li(3) = 0.28 (0.29), ηCD_EC = 0.014 A/W, ηCD_EB = N/A, Zeff = 2.7 (2.2), Psep/Rgeo = 36 (37) MW/m, (H98+H98*)/2 = 1.43 (1.42), Psep/PL2H_M = 2.1, Psep/PL2H_T = 1.4 - EB-CC: Ip = 22.8 (22.6) MA, Pfus = 1.56 (1.61) GW, Pel = 122 (142) MW, PEC = 9 MW, PEB = 120 MW, Prad = 296 (292) MW, Q = 12.0 (12.5), βN = 3.9, fBS = 0.81, fGW = 0.96 (0.97), li(3) = 0.23, ηCD_EC = 0.017 A/W, ηCD_EB = 0.033 A/W, Zeff = 2.3 (1.8), Psep/Rgeo = 40 (40) MW/m, (H98+H98*)/2 = 1.24 (1.23), Psep/PL2H_M = 2.1, Psep/PL2H_T = 1.5 - EC-LD: Ip = 25.1 (24.8) MA, Pfus = 1.51 (1.62) GW, Pel = 34 (84) MW, PEC = 160 MW, PEB = 0 MW, Prad = 308 (327) MW, Q = 9.4 (10.1), βN = 5.1, fBS = 0.81 (0.82), fGW = 0.61 (0.57), li(3) = 0.30 (0.29), ηCD_EC = 0.026 A/W, ηCD_EB = N/A, Zeff = 4.6 (4.3), Psep/Rgeo = 40 (41) MW/m, (H98+H98*)/2 = 1.75 (1.78), Psep/PL2H_M = 2.7, Psep/PL2H_T = 1.1 - EB-HQ: Ip = 21.3 (21.2) MA, Pfus = 1.62 (1.68) GW, Pel = 302 (324) MW, PEC = 6.3 MW, PEB = 48.2 MW, Prad = 252 (242) MW, Q = 29.8 (30.8), βN = 4.3, fBS = 0.90, fGW = 0.98, li(3) = 0.26 (0.25), ηCD_EC = 0.020 A/W, ηCD_EB = 0.035 A/W, Zeff = 2.4 (1.8), Psep/Rgeo = 35 (36) MW/m, (H98+H98*)/2 = 1.42 (1.38), Psep/PL2H_M = 1.9, Psep/PL2H_T = 1.4 The sensitivity to key assumptions has been evaluated using a simple “predictive” model. The electron transport is calculated using the Rechester-Rosenbluth (RR) model [19] with a closure by Wong et.al. [20] developed for NSTX. Ions are modelled with gB like transport slightly above neoclassical. In this set-up the NI solution shows a strong nonlinear performance increase with HCD power. The Q ≈ 10 solution in all cases requires roughly Ip ≈ 20 MA. 3. CORE TRANSPORT The transport in STEP is governed by electromagnetic turbulence rather than the usual electrostatic turbulence encountered in CT. Linear gyrokinetic (GK) simulations show that in the high βe = 2μ0pe / Bt² STEP plasmas MTM are unstable on the ion and electron scales as well as KBMs over a broad spectral range [4,21]. The transport characteristics of these modes are not captured in commonly used reduced models such as TGLF or QUALIKIZ-NN. The growth rates of both, the electron scale MTM and the KBM are very sensitive to the equilibrium parameters such as plasma flow shear and βe'. Nevertheless, recent local non-linear GK (NLGK) simulations suggest that a KBM hybrid mode dominates the transport in all channels, whilst the electron transport by MTMs is negligible. The NLGK simulations for STEP parameters are very difficult and currently only MTM only (by setting δB∥ = 0 to suppress KBMs) or KBM-only (by not resolving the MTMs radially) simulations have been performed. Figure 4 shows the total heat flux Qtot from KBM hybrid turbulence as function of βe without flow shear and the normalised flow shear (a/cs)γE. Increasing βe initially increases Qtot as the drive is increased, but later βe' stabilisation sets in leading to strong reduction. Without flow shear the heat flux Qtot ~ 10³ MW/m² would be well above the assumed value for the FTOP, but already the introduction of flow shear of the order of the diamagnetic value gives a comparable Qtot. Table 2: Total particle and heat fluxes on the q=3.5 surface in EC-HD of KBM-hybrid turbulence (with flow shear γE = 0.1 cs/a): - Γtot [10²⁰ m⁻² s⁻¹]: CGYRO = 1.37, GENE = 0.52, JETTO = 0.2 - Qtot [MW m⁻²]: CGYRO = 2.27, GENE = 1.95, JETTO = 1.4 - Qi [MW m⁻²]: CGYRO = 0.51, GENE = 0.16 - Qe [MW m⁻²]: CGYRO = 1.76, GENE = 1.79 4. EQUILIBRIUM AND EXHAUST DESIGN To handle the plasma exhaust in such a compact device a double null (DN) divertor solution is proposed with all divertor legs having a non-standard configuration to reduce the required impurity seeding needed for fully detached divertor operation by increasing the connection length and overall flux expansion. Divertor performance and optimal magnetic configuration have been assessed using SOLPS-ITER with and without plasma drifts included. Adequate performance is achieved with an outer leg strike-point extended to large major radius and an inner leg approaching an X-divertor (IXD) where an additional X-point is created close to the inner target. To realise the IXD two pairs of high field side coils (S1, S2) with opposing currents have been implemented in the design. Together with the equilibrium optimisation, 2D scrape-off-layer (SOL) multifluid calculations have been performed to assess the most optimal divertor configurations with respect to neutral particle trapping, pumping efficiency, access to detachment and field line angle. Table 3: Inner divertor performance parameters: - IXD: q⊥,T = 1.25 MW m⁻², αinc = 2.0°, Neutral compression = 87, Main chamber recycling flux penetration = 15.2% - IPT: q⊥,T = 2.72 MW m⁻², αinc = 6.3°, Neutral compression = 23, Main chamber recycling flux penetration = 17.5% - IVT: q⊥,T = 1.85 MW m⁻², αinc = 3.9°, Neutral compression = 93, Main chamber recycling flux penetration = 2.6% - IHT: q⊥,T = 0.86 MW m⁻², αinc = 4.1°, Neutral compression = 23, Main chamber recycling flux penetration = 32.1% The prime function of the DN configuration is to shield the inner targets from high heat and particle loads. At the high plasma current in SPP the expected scrape-off layer decay length λSOL ~ 1 − 2 mm is very narrow, requiring very high accuracy for the vertical control (ΔZ ≲ ±5 mm) around the DN configuration to stay close to DN. 5. PLASMA STABILITY AND CONTROL Operating above the no-wall β limit, βN_NW ≈ 3.9 means that resistive wall modes (RWM) can be unstable [24] with 0.1 < Cβ = (βN - βN_NW)/(βN_IW - βN_NW) < 0.6 (βN_IW: ideal wall limit). Active stabilisation with 8 mid-plane in-vessel coils behind the blanket is foreseen. Figure 7 shows a conceptual drawing of the envisaged 3D coil set for controlling n = 1, 2 RWM, edge localised modes (ELMs) and correcting the error field (EFC). Core MHD such as NTM or TAE are stable in the FTOPs due to high curvature and thermal ion damping. Large type-I/III ELMs will likely cause problems for divertor lifetime and need suppression via n = 3 or 4 resonant magnetic perturbations (RMPs). Two separate coil sets comprised of 32 coils are being designed: an in-vessel (ELM-iv) resistive set and an ex-vessel (ELM-ev) superconducting set. In the FTOP for an n = 3 field, I_ELM-iv ≈ 50 kAt and I_ELM-ev ≈ 90 kAt are needed. Controlling the vertical stability at κ ≈ 3 requires active in-vessel coils behind and passive toroidally conductive structures inside the blankets. Low li and a broad pressure profile benefit both vertical stability and RWM control. Disruptions represent a major threat due to high plasma current and stored energy. Multiple shattered pellet injection (SPI) with D2 and Ar+D2 is planned to minimise runaway electron (RE) generation, control current quench time, and mitigate thermal quench heat loads. RE beam currents of 10 MA ≤ I_RE ≤ 15 MA are likely, requiring active RE mitigation. 6. HEATING AND CURRENT DRIVE Fundamental to the scenario design is an efficient heating and current drive solution. All major HCD techniques have been assessed, leading to a microwave-only HCD system. The STEP magnetic field and density allow both ECCD and EBCD. Currently PHCD ≤ 300 MW injected (336 MW installed) at ~8 frequencies between 90 GHz and 210 GHz are proposed. Coupled JETTO/Gray simulations of the EC-HD scenario show that the required current profile can be matched with O1 and O2 heating at 160, 170, 180, 200, and 210 GHz requiring PHCD = 160 MW. GENRAY/CQL3D calculations show normalised current drive efficiency ζCD = 32.7 ηCD[A/W] ne[10²⁰ m⁻³] R[m] / Te[keV] can be improved by a factor of 2 to 3 for EBCD due to the Ohkawa current and trapped particle populations in ST geometry. A 1.8 MW EBCD system will be implemented on MAST-U to validate the modelling. 7. PLASMA INITIATION, RAMP-UP AND RAMP-DOWN Space restrictions in the central column do not permit a large central solenoid. A small resistive solenoid delivering ψCS ≈ 10 Vs is foreseen for breakdown, burn-through, and inductive ramp-up to Ip ≲ 2 MA. DYON simulations show solenoid-only breakdown and burn-through can be achieved with 9 V ≲ Vloop ≲ 15 V within ΔtBD ≲ 0.5 s and prefill pressure p0 = 1 mPa. For oxygen levels above cO > 2%, EC assistance of 0.2–0.6 MW is essential. A full-bore diverted Ip ~ 2 MA plasma can be obtained ohmically within ΔtIR = 3 − 10 s consuming ΔψCS ≈ 4 Vs. Non-inductive ramp-up is split into two phases: (1) generating the full plasma current at low density using a linear power ramp up to Paux ≈ 170 MW of ECCD, and (2) increasing density to fusion conditions. Transition from high Te ≲ 40 keV to Ti ≈ 20 keV occurs as confinement time exceeds equilibration time. Ramp-down reduces fusion power within 200 s by transitioning from DT to D, then ramping down HCD power over ΔtRD ≲ 1200 s. 8. CONCLUSIONS In the first four years of the STEP programme significant progress has been made in understanding the required plasma conditions for a fusion power plant prototype based on a spherical tokamak. In the absence of a full scenario predictive model, key sensitivities have been assessed and the design adapted to reach 1.5 GW ≤ Pfus ≤ 1.8 GW and Pel ~ 100 MW. Core transport, pedestal performance, vertical stability, and RWM control remain key focus areas, but current findings substantially increase confidence that a viable plasma scenario for STEP exists. REFERENCES [1] H. Wilson et al., Commercialising Fusion Energy (IOP Publishing, 2020), pp. 8. [2] R. R. Wessen et al., Root (2013). [3] B. S. Patel et al., Nuclear Fusion 62, 016009 (2021). [4] D. Kennedy, 29th IAEA FEC, London, UK (2023). [5] F. Casson, 29th IAEA FEC, London, UK (2023). [6] H. Meyer, 48th EPS Conference on Plasma Physics (2022). [7] R. Wessen et al., AIAA SPACE 2009 (2009). [8] S. I. Muldrew et al., Fusion Eng. Des. 154, 111530 (2020). [9] E. C. Tholerus et al., to be submitted to Nuclear Fusion (2023). [10] J. E. Menard et al., Nuclear Fusion 56, 106023 (2016). [11] M. Romanelli et al., Plasma Fusion Res. 9, 3403023 (2014). [12] S. M. Kaye et al., Plasma Phys. Control. Fusion 63, 123001 (2021). [13] S. Freethy, 29th IAEA FEC, London, UK (2023). [14] ITER Physics Expert Group, Nucl. Fusion 39, 2175 (1999). [15] P. F. Buxton et al., Plasma Phys. Control. Fusion 61, 35006 (2019). [16] G. S. Kurskiev et al., Nuclear Fusion 62, 016011 (2021). [17] Y. R. Martin et al., J. Phys.: Conf. Ser. 123, 12033 (2008). [18] T. Takizuka et al., Plasma Phys. Control. Fusion 46, A227 (2004). [19] A. B. Rechester, M. N. Rosenbluth, Phys. Rev. Lett. 40, 38 (1978). [20] K. L. Wong et al., Phys. Rev. Lett. 99, 135003 (2007). [21] B. S. Patel et al., Nuclear Fusion 62, 016009 (2022). [22] R. T. Osawa et al., Nuclear Fusion 63, 076032 (2023). [23] J. Karhunen, 29th IAEA FEC, London, UK (2023). [24] G. Xia et al., Nuclear Fusion 63, 026021 (2023). [25] D. A. Ryan et al., Plasma Phys. Control. Fusion 61, 095010 (2019). [26] A. Prokopyszyn, 29th IAEA FEC, London, UK (2023). [27] M. Komm et al., Nuclear Fusion (2023). [28] M. Lennholm, 29th IAEA FEC, London, UK (2023). [29] A. Fil, 29th IAEA FEC, London, UK (2023). [30] M. Henderson, 29th IAEA FEC, London, UK (2023). [31] S. Freethy et al., 48th EPS Conf. Plasma Phys. (2022). [32] B. Biswas et al., Nuclear Fusion (2023). [33] T. Wilson, 21st Joint Workshop on ECE and ECRH (2022). [34] H. Webster, 29th IAEA FEC, London, UK (2023). [35] V. F. Shevchenko et al., Nuclear Fusion 50, 22004 (2010). [36] H.-T. Kim et al., Nuclear Fusion 62, 126012 (2022). [37] A. Kallenbach et al., Plasma Phys. Control. Fusion 58, 045013 (2016). [38] J. Mitchell et al., Fusion Eng. Des. 192, 113777 (2023).