IAEA INTERNATIONAL ATOMIC ENERGY AGENCY st IAEA Fusion Energy Conference Chengdu, China, 6 - October 6 IAEA-CN-9/EX/- THE PERFORMANCE OF IMPROVED H-MODES AT ASDEX UPGRADE AND PROJECTION TO A.C.C. SIPS, G. TARDINI, C. FOREST, O. GRUBER, P. Mc CARTHY, A. GUDE, L.D. HORTON, V. IGOCHINE, O. KARDAUN, C.F. MAGGI, M. MARASCHEK, V. MERTENS, R. NEU, A. PEETERS, G.V. PEREVERZEV, A. STÄBLER, J. STOBER, W. SUTTROP AND THE TEAM Max-Planck-Institut für Plasmaphysik, EURATOM-Association Boltzmannstrasse D-8578, Garching, Germany. Max-Planck-Institut für Plasmaphysik, EURATOM-Association, D-8578, Germany. The University of Wisconsin, Madison, USA. Dep. of Physics, University College Cork, Association EURATOM-DCU, Cork, Ireland. This is a preprint of a paper intended for presentation at a scientific meeting. Because of the provisional nature of its content and since changes of substance or detail may have to be made before publication, the preprint is made available on the understanding that it will not be cited in the literature or in any way be reproduced in its present form. The views expressed and the statements made remain the responsibility of the named author(s); the views do not necessarily reflect those of the government of the designating Member State(s) or of the designating organization(s). In particular, neither the IAEA nor any other organization or body sponsoring this meeting can be held responsible for any material reproduced in this preprint.
EX/- The performance of improved H-modes at and projection to A.C.C. Sips, G. Tardini, C. Forest, O. Gruber, P. Mc Carthy, A. Gude, L.D. Horton, V. Igochine, O. Kardaun, C.F. Maggi, M. Maraschek, V. Mertens, R. Neu, A. Peeters, G.V. Pereverzev, A. Stäbler, J. Stober, W. Suttrop and the Team Max-Planck-Institut für Plasmaphysik, EURATOM-Association, D-8578, Germany. The University of Wisconsin, Madison, USA. Dep. of Physics, University College Cork, Association EURATOM-DCU, Cork, Ireland. E-mail contact of the main author: ccs@ipp.mpg.de Abstract At, stationary discharges with improved confinement (H 98 (y,) > ) and improved stability (β N >.5) have been developed since 998. New results presented here concentrate on extending the operational range of these improved H-modes at and extrapolating the results to. The performance is optimised for q 95 ranging from to 5, using real time control of β N and ECCD to suppress NTM activity at low β N ~. Discharges are obtained at different values of collisionality, at high density, and with a first wall predominantly covered by tungsten coated carbon tiles. For the extrapolation to, the fusion power is estimated using and kinetic profile shapes as obtained in and <n e >/n GW =.85. The fusion gain that could be obtained is evaluated using different confinement scalings. These ASDEX Upgrade results indicate that improved H-modes are a candidate for an hybrid scenario or could extend operation beyond what is currently foreseen using standard H-modes.. Introduction The principal reference scenario for [] is based on H-mode operation. It provides the basis for achieving s primary goal of operation at Q= at a fusion power P fus ~MW (with Q defined as the ratio of fusion power to external input power). An estimate for the energy confinement in is based on the empirical IPB98(y,) [] scaling. The average density in should be as high as possible, and is chosen to be 85% of the Greenwald density for H-mode operation, n GW = I p [MA]/πa[m] [], where I p is the plasma current, a is the plasma minor radius. The operational space of the inductive reference scenario at 5 MA has been assessed for a range of H 98 (y,) factors to determine the envelope of performance within s capabilities and physical constraints []. These show a strong dependence of Q and P fus with H 98 (y,), average plasma density and impurity concentration. Hence, a significant increase in performance could be achieved with a relatively small improvement in energy confinement. Discharges that achieve H 98 (y,) >, and operate at higher beta compared to the reference scenario, β N > can, when realised in, be used to increase the performance at 5 MA (β Ν = <β>ab T /I p, with <β>, the volume averaged normalised pressure (p) in the tokamak). Alternatively these discharges can be utilised for operation at I p =- MA, while keeping Q=5-. This so called hybrid scenario is designed to achieve long pulse operation with I p driven by a combination of inductive and non-inductive currents. This would maximise the neutron fluence (neutron wall load x pulse length) in for material testing. At [,5,6,7] stationary H-mode discharges have been developed which achieve the required level of confinement and stability for improving s performance at full current (P fus > MW and Q > ) or for hybrid operation at reduced current. These improved H-modes are characterized by a q-profile with low magnetic shear in the centre and q ~, typically obtained by applying heating during the current rise phase of the discharge. Also other experiments (DIII-D [8], JT-6U [9] and JET []) have shown rapid progress in the development of this operational scenario in the past few years.
EX/- The new results presented here concentrate on extending the operational range of improved H-modes at. The performance is optimised for q 95 ranging from to 5, for different values of ν* and at high density. The requirements for operation and control of high beta plasmas are given. The fusion performance in is estimated by scaling the kinetic profiles obtained in, keeping the as obtained in and setting <n e >/n GW =.85 (<n e >, line averaged electron density). The fusion gain that could be obtained is evaluated using different confinement scalings.. at During the current rise phase of improved H-modes at a lower single null divertor configuration is used (formed at t ~.5s). This allows operation at low plasma density (< x 9 m - ) and a control of the impurity content. Additional heating is applied to slow down the current diffusion. By using such a scenario, a low magnetic shear in the centre is obtained at the beginning of the flat top phase of improved H-modes. Contrary to what one would expect on the basis of neoclassical current diffusion, the q-profile remains stationary with low central shear throughout the discharge. MHD modes are thought to be responsible for keeping q fixed. Typically (,) NTMs (combined with higher m/n activity) or fishbone activity in the core are observed in improved H-modes. [] Table I: Overview of discharges used. Ranges are given, values in brackets are average values. dataset # q 95 n e x 9 <n e >/n GW H 98 (y,) β N l i Type I ELMy.9-5.5.6-.5-..5-.6 9 H-modes* (.) (7.) (.6) (.8) Improved.8-5..-.-.97.8-.5 59 H-modes (.) (6.6) (.5) (.) *: This dataset also contains discharges with low magnetic shear in the centre.7-.5 (.89).-.6 (.).75-. (.95).75-. (.89) H-mode operation at is achieved for a wide range of plasma conditions. For the analyses presented in this paper, two H-mode datasets are created with I p =.6-.MA and B T =.6-.T, the values obtained are averaged for the time window selected. The first dataset contains all type I ELMy H-modes with q 95 < 5.5 that are stationary for more than. seconds (grey symbols in Fig. a). A second dataset includes all improved H-mode discharges that are stationary for more than.5 seconds (red symbols in Fig. a), and not a mere selection of the best discharges realized at. The two sets of data are.6.6.. H 98 (y,).. H 98 (y,)...8.8.6 H-modes (Type I) β N Fig. a: H-mode operation at achieves a wide range of H 98 (y,,) vs β N.6 H-modes (Type I)...6.8. <n e >/n GW Fig.b: H-mode operation at over a range of <n e >/n GW
EX/- compared in Fig. and Table I. Recently improved H-modes have also been obtained in discharges q 95 > without preheating []. Also these discharges (marked as red symbols) are included in Fig.. On average improved H-mode discharges achieve higher H 98 (y,) at higher β N compared to standard H-modes. By using early heating the average value of the plasma inductance (l i ) is lower (l i =.89), compared to the averaged l i =.95 for ELMy H- modes. Magnetic measurements are used to evaluate l i (no MSE data before 5), leading to large scatter of the values obtained. Hence, physics based selection criteria, using for example l i <.95 to identify improved H-modes, are not used here. do not exclusively occupy the domain H 98 (y,) > at β N > -. Standard H-modes that achieve high beta may have developed a q-profile with low magnetic shear in the centre, as the pedestal has a significant contribution to the beta achieved [], giving a significant bootstrap current at the edge of the plasma. As a result the type I ELMy H-mode discharges at high beta would be similar to improved H-modes. So far it has not been possible to check all type I ELMy discharges in the dataset. For comparison, H-mode discharges from submitted to the confinement database (DBv5) have an average value of H 98 (y,)=., rather than the average of H 98 (y,)=.8 of the type I ELMy H-modes in the dataset used here, indicating some higher confinement discharges (improved H-modes) are included in the ELMy H-mode dataset. As shown in Fig b, improved H-mode discharges typically operate at low density, <n e >/n GW =.5-.6. at higher density are obtained at by increasing the density after the formation of the q-profile, together with an increase in heating power. The highest values of <n e >/n GW =.85, with good confinement H 98 (y,) ~. are obtained using a configuration at high triangularity, δ=. (see section ). Control of the impurity influxes is of e- particular importance with the All AUG discharges progressive increase in the area of the e- first wall covered by tungsten coated carbon tiles at. The e- tungsten concentration in the core of the plasma shows a steady rise in all e-5 plasmas since, with the increase in surface covered by tungsten to 6 W CONC, in the core e-6 e-7 6 8 Discharge # Fig.: The concentration of tungsten in the centre for # (March ) to # 5 (May 6). m in 6 (9% of the first wall surfaces, not the divertor). This is shown in Fig. The tungsten concentration is measured spectroscopically for the plasma region that has T e ~-kev (central part). Most H-modes, including improved H-modes, have central tungsten concentrations below -. This is considered acceptable for a reactor, with the tungsten concentration at the plasma edge below -5. However, for all H-modes at central heating with ICRH or ECRH is required to avoid impurity accumulation. Operation at the lowest plasma densities (-6x 9 m - ) requires a boronisation prior to the experiments to minimise tungsten influxes. More detailed investigations at, on the role of the q-profile in improved H-modes [], on the documentation of the edge pedestal in these discharges [] and on the role of pedestal in improved confinement discharges for various experiments [] are ongoing to establish a physics basis for extrapolation of these results to. Most of this research is coordinated by the International Tokamak Physics Activity (ITPA). Results obtained at form an essential part of these international collaborations.
EX/-. Operation at q 95 = -5 In recent studies q 95 was varied from to 5.5 by changing the plasma current (as would be the case in ), in contrast to previous studies where the toroidal field was changed at I p =MA. Discharges at q 95 ~5 (.8MA/.T) have similar MHD behaviour, stability and confinement compared to pulses at somewhat lower q 95 [6]. These discharges achieve high values for beta poloidal (~). ASTRA code analysis of the kinetic data, including models for the bootstrap current and beam driven current, estimate a.7 non-inductive current drive fraction (I BS /I p =., I NBCD /I p =.). The non-inductive current contributions alone sustain the central q above. Discharges at q 95 > 5 would allow operation closer to fully non-inductive conditions. However, in H-modes the overall confinement, performance and capability to operate at sufficiently high densities all scale with I p, so that these plasmas are deemed not relevant. I p (MA) 8.6.6 6 divertor shape P NBI (MW) β N β p even n odd n l i H 98 (y,) # 9 P ICH (MW) T T i (kev) e (kev) time (s) 6 Fig.: Improved H-mode at q 95 =.7. D α Discharges at lower q 95 (~) are achieved by increasing the plasma current to.ma. Operating at.t, ICRF H-minority heating with a resonance in the centre is used to allow control of the central impurity content and the density peaking. Fig. shows an example of a discharge at q 95 =.7 (#9). Feedback control of beta (β N ) is used (with P NBI as actuator) to avoid early MHD modes and the formation of an internal transport barrier (which poses a risk of a disruption). In this discharge equal amounts of central and off-axis neutral beam heating (current drive) are used. The tungsten concentration in the centre is.5x -5, for <n e >=6.x 9 m - and <n e >/n GW =.. The discharge reaches β N =.9, stationary for.5s ( energy confinement times), limited by the duration of the additional heating. After seconds, n= activity is observed, dominated by fishbone activity lasting throughout the high power heating phase, indicating that q ~ in the centre. Also (,) NTM activity is observed and no sawteeth are present throughout discharge. It is reported in [] that in improved H- modes without (,) NTM activity, H 98 (y,) is typically higher. With the increase of beta during the pulse, H 98 (y,) continues to rise until it reaches. at β N =.9. Hence, this pulse displays a significant increase in H-mode performance.. Operation over a wide range in normalised collisionality (ν*). has shown [5] that improved H-modes at I p =8kA can operate at high plasma density. This is achieved by carefully adjusting the heating power and the gas puff rate in a high triangularity (δ=.5) plasma configuration near double null, with q 95 =.5- at <n e >/n GW =.88 and H 98 (y,)=.. These discharges are included in Fig b. With the first wall covered by tungsten coated carbon tiles, it is important to achieve H-modes at high edge density (n e,ped ) to minimise the tungsten influxes over a wide operational range. Recent improved H-mode experiments have extended high density operation to a plasma current of MA (q 95 =), achieving <n e >=.x m - (<n e >/n GW =.85-.9). This goes beyond the absolute densities required for while maintaining high performance: H 98 (y,)=., β N =.. The density profile is moderately peaked, with n e,ped ~9.x 9 m - and n e ~
5 EX/-.x m -. At these high densities in the core the electron and ion temperatures are equal T i =T e =kev. The results presented here can be extrapolated to using the normalised collisionality (ν* ~ a 7 <n e > κ /ε W, with W the total stored energy). Devices smaller than, like, operate at densities relevant for exhaust (<n e >/n GW =.85-.9) at relatively high collisionality, ν*/ν* ~. Hence, it is important to document the confinement improvement over IPB98(y,) scaling in versus ν*/ν* for standard H-modes and improved H-modes covering a range of electron densities from.x m - to.x m - (Fig. ). The highest values of H 98 (y,) are obtained at relevant ν*. Improved H-mode discharges operate typically close to collisionality values and occupy the upper envelope of the operating space in Fig..6. H 98 (y,). β N.8.6 H-modes (Type I) H-modes (Type I) 5 5 ν*/ν* 5 q 95 Fig. : H 98 (y,,) vs. collisionality, normalised to the collisionality (ν /ν ). Fig. 5: The beta (β N ) achieved in H-mode discharges in ADSEX Upgrade for q 95 =-5.5. 5. Operation at high beta: Limits and control of NTMs High beta, β N ~, can be achieved in nearly all experimental conditions explored so far for improved H-modes Fig. 5 shows that the maximum values for β N obtained have no dependence on q 95. The maximum values of β N achieved are in most cases near the no wall beta limit (β N =l i ) over the range of q 95 explored. The beta limit in (improved) H-modes manifests itself as a (,) NTM [6]. The improved H-modes at low q 95 have fishbone activity at high beta (β N ~.5-). These discharges are prone to develop large (,) NTM modes, strongly deteriorating confinement, during the first few seconds when β N is increased above. A characteristic of (,) NTM activity at low q 95 <.5 is that the impact on confinement is generally stronger compared to improved H-mode discharges at higher q 95 [5]. In the example given before at q 95 =.7 (Fig. ) the (,) NTM mode is seen on the MHD measurements from.7 to. seconds. This NTM behaviour is reproducible using the same discharge set-up and in some cases grows in amplitude leading to a (,) NTM. For two similar discharges at.ma/.t (#69 and #7) the conditions of #9 were repeated. Co-ECCD was applied at MW in #69 at ρ pol =.77, while in #7 no ECCD was applied. In both pulses the (,) NTM started spontaneously (without sawteeth trigger) at around second, growing in size until. s. The (,) NTM is stabilised rapidly in #69 when ECCD is applied from. s. (Fig. 6). Fig. 7 shows how the kinetic profiles recover when the NTM is stabilised. These profiles are obtained from fits to various
6 EX/- diagnostic measurements available at. After the (,) NTM stabilisation, the discharge continues and is subsequently heated to reach β N =.5, and H 98 (y,)=.. Without ECCD (#7), the (,) NTM remains, deteriorates the plasma confinement, leading to a (,) NTM that grows and locks. When this occurs, the plasma protection triggers an impurity gas injection to induce a mild disruption... # 69 (MW ECRH) # 7 β N H 98 (y,) P ECRH (MW).8.6...8 Time (s), activity disruption (a) ne_fit (X 9 m - ) Ti_Fit (kev ) (c)....6.8. rho_pol # 69 ECRH deposition 8 6 t=.9s t=.s NTM With ECRH t=.9s t=.s # 69 ECRH deposition NTM With ECRH....6.8. rho_pol (b) Ti_Fit (kev ) (d) t=.9s t=.s NTM With ECRH....6.8. rho_pol # 69 ECRH deposition P ECRH =.5MW, rho_pol=.77, I CD =9.kA, j CD =.5MA/m Fig.6: Two similar discharges at.ma. #69 uses ECRH to stabilise the (,) NTM. Fig. 7: #69: MW ECRH is applied at ρ pol =.77 to stabilise the (,) NTM. Te, Ti and n e are given before (red) and after NTM stabilisation (green). 6. Extrapolation to The prediction of the fusion performance in presented here is based on an extrapolation using the ASTRA transport code starting from experimental profiles. Results are calculated for 68 discharges, conventional and improved H-modes with q 95 =.5-.8, H 98 (y,)=.95-.65 and β N =.7-.5. For these discharges the density and temperature profiles are obtained from a fit to the data of several diagnostics. The thermal beta calculated from the kinetic profiles ( ) is typically.8-.9 β N. Some assumptions are made to scale experiments to [6]. A similar method has been used for DIII-D discharges [7]. The choice is, of course, not unique. The toroidal field (5.T) and the equilibrium boundary are taken from the design []. The plasma current in is chosen to match the q 95 used in, n e (x 9 m - ) - (a) 8 6 Density using AUG n e -profile shape n e n D +n T #9 <n> =.85n GW..5..5. r min (m) Fig 8: The density and temperature profiles in scaled from #9. In these conditions P fus =7 MW, Q=. (b) T i,e (kev) - 5 5 5 X Temperatures using AUG T i -profile shape Temperatures using AUG T e -profile shape #9 = AUG T i = T e..5..5. r min (m) and varies from I p =.MA down to I p =9.MA for the range of q 95 =.5-.8. The density profile shape is kept and its value is adjusted to achieve <n e >=.85n GW in (Fig 8a). At ASDEX Upgrade H-modes obtain good confinement over a range of <n e >/n GW (see Fig.). As a result choosing <n e >=.85n GW is valid, especially considering
7 EX/- uses a plasma configuration at δ~.5. Electron and ion temperature profiles are assumed to be the same in. The shape of the temperature profiles is taken to be equal to the ion temperature profile in when T e <T i or taken equal to the electron temperature profile when T e T i. Usually, the temperature gradient lengths are quite close to each other in H-mode plasmas (Fig 8b), some differences are observed at the lowest plasma densities (at high T i ). The scaling factor for the temperature profiles is determined by the β N;th value obtained in the discharges. It is assumed that similar β N;th values can be achieved, although the normalised Larmor radius ρ* is substantially lower in, which could make discharges more susceptible to NTM activity [5]. The deuterium and tritium concentrations are assumed to be equal. Impurity concentrations are taken from the design: Be %, Ar. %, He. %. As a result, the volume averaged Z eff is approximately.65 in all simulations, and the dilution of the tritium and deuterium fuel is ~.8 as shown in Fig 8a. Using these assumptions, the fusion power in can be calculated directly from the scaled kinetic profiles. An overview of the results is given in Fig. 9. The maximum fusion power obtained ranges from MW (at =.6) at I p =9. MA to 7MW (at =.8, #99, Fig. ) at I p =. MA in. An estimate for the required input power (P aux ) and Q depend on the energy confinement. In Fig. 9 the IPB98(y,) scaling is used, assuming the same H 98 (y,) values are obtained in and. P fus (MW) 8 6 = - predictions at <n>/n GW =.85 5 MA. Open symbols P aux > 7 MW Closed symbols P aux 7 MW 5 I p (MA) Fig. 9: The prediction of the fusion power in by scaling the kinetic profiles data from ASDEX Upgrade. The power requirements (P aux ) to sustain the β N,th use IBP98(y,) scaling. These extrapolations indicate that operation at high beta results in significant fusion power for I p =9.5MA to I p =MA. The bootstrap current fraction at these plasma currents is f BS ~. at β N ~, while Q ranges from 6-5. Alternatively, the results at low q 95 ~ (I p =-5 MA) indicate that the fusion power could be more than doubled compared to the reference scenario, moreover these discharge would obtain ignition (Q= ) in. For the highest fusion power obtained at q 95 ~., the pedestal conditions n e,ped =6.8x 9 m - and T e,ped =T i,ped =5.7 kev are within the design parameters. The fusion gain and input power (P aux ) required have been calculated for three different energy confinement scalings [,8,9]. For each scaling the respective H-values obtained at ASDEX Upgrade are taken. The results are shown in Fig.. For the IPB98(y,) scaling operation at high beta would require, in some cases, a value for P aux which is above the maximum planned for the first stage in (7MW). These are indicted by the open circles in Figs. 9,. This is a feature of the IPB98(y,) scaling law which has a strong beta degradation (Bτ E ~β.9 ), and maximises Q at the lowest possible beta. Especially for high q 95 (I p <MA), the IPB98(y,) scaling predicts discharges at β N >.5 to be inaccessible at <n e >=.85n GW even for H 98 (y,)=.-.5. Far more optimistic values for P aux and Q are obtained using a GyroBohm scaling [8] (Bτ E ~β ). The scaling proposed by Cordey [9] (Bτ E ~β ) would predict higher Q at high beta compared to the IPB98(y,) scaling.
8 EX/- (a) IPB98 scaling (b) GyroBohm scaling (c) Cordey5 scaling P aux (MW) P aux = 7 MW Q =..5..5 - Q =..5..5 - Q =..5..5 - Fig. : Input power requirements (P aux ) to sustain the beta in as obtained in for three different energy confinement scaling laws. 8. Conclusions At, H-modes with low magnetic shear in the centre achieve H 98 (y,)> and β N =-. The central tungsten concentration can be kept at acceptable levels (< - ) in these experiments at. at q 95 =. achieve H 98 (y,)=. at β N =.9, with fishbone activity in the core keeping the q-profile stationary. ECCD can be used to stabilise (,) NTM activity during the low beta (β N ~) phase of these discharges. Operation at high density with <n e >=.x m - (<n e >/n GW =.85-.9) is demonstrated. However, the highest H 98 (y,) values are achieved at relevant ν*. The kinetic profile shapes at are scaled to (using ASTRA), setting <n e >=.85n GW and keeping. This predicts high fusion power for improved H-mode discharges at q 95 =. (P fus =7 MW, Q= ). In these conditions, the density and temperature at the edge are within design parameters. At lower I p in (9.5MA-MA), significant fusion power can be achieved (P fus MW, Q=6-5). Using the IPB98(y,) scaling, the auxiliary power requirements at high β N >.5 and at I p <MA may exceed the maximum P aux planned for the first stage of (7 MW). References [] Physics Basis 999 Nucl. Fusion 9 7 [] Greenwald M. et al 988 Nucl. Fusion 8 99 [] Shimada M. et al Nucl. Fusion 5 [] Gruber O. et al 999 Phys. Rev. Lett. 8 787 [5] Sips A. C. C. et al Plasma Phys. Control. Fusion B69 [6] Stäbler A. et al 5 Nucl. Fusion 5 67 [7] Gruber O. et al 5 Plasma Phys. Control. Fusion 7 B5 [8] Luce T. C. et al 5, Nucl. Fusion 5 S86 [9] Ide S. et al 5, Nucl. Fusion 5 S8 [] Joffrin E. et al 5, Nucl. Fusion 5 66 [] Pereverzev G. V. et al 99 Report IPP 5/ [] Stober J. et al 6 this Conference, Physics studies of the improved H-mode scenario in ASDEX Upgrade, (EX/P-7) [] Maggi C. F. et al 6 this Conference, Characteristics of the H-mode Pedestal in Improved Confinement Scenarios in, DIII-D, JET and JT-6U, (IT/P-6) [] Suttrop W. et al 6 this Conference, Studies of the edge pedestal and behaviour of Edge Localised Modes in improved H-mode in, (EX/8-5) [5] Zohm H. et al 6 this Conference, Control of MHD Instabilities by ECCD: Results and Implications for, (EX/-Rb) [6] Tardini G. et al 6 rd EPS Conference, Rome 9- June 6, P. [7] Luce T. C. et al Phys. Plasmas 67 [8] Petty G. et al Fusion Sci. Technol. [9] Cordey J. G. et al 5 Nucl. Fusion 5 78