Long plasma duration operation analyses with an international multi-machine (tokamaks and stellarators) database

Combined high-fusion performance and long-pulse operation is one of the key integration challenges for fusion energy development in magnetic devices. Addressing these challenges requires an integrated vision of physics and engineering aspects with the purpose of simultaneously increasing time duration and fusion performance. Significant progress has been made in tokamaks and stellarators, including very recent achievement in duration and/or performance. This progress is reviewed by analyzing the experimental data (109 plasma pulses with a total of 3200 data points, i.e. on average 29 data per pulse) provided by ten tokamaks (in alphabetical order: Axially Symmetric Divertor Experiment Upgrade, DIII-D, Experimental Advanced Superconducting Tokamak, Joint European Torus, JT-60 Upgrade, Korea Superconducting Tokamak Advanced Research, tokamak à configuration variable, Tokamak Fusion Test Reactor, Tore Supra, W Environment in Steady-State Tokamak) and two stellarators (Large Helical Device and Wendelstein 7-X) expanding the pioneering work of Kikuchi (Kikuchi M. and Azumi M. 2015 Frontiers in Fusion Research II: Introduction to Modern Tokamak Physics (Springer)). Data have been gathered up to January 2022 and coordination has been provided by the recently created International Energy Agency-International Atomic Energy Agency international Coordination on International Challenges on Long duration OPeration group. By exploiting the multi-machine international database, recent progress in terms of injected energies (e.g. 1730 MJ in L-mode, 425 MJ in H-mode), durations (1056 s in L-mode, 101 s in H-mode), injected powers, and sustained performance will be reviewed. Progress has been made to sustain long-pulse operation in tokamaks and stellarators with superconducting coils, actively cooled components, and/or with metallic walls. The graph of the fusion triple products as a function of duration shows a dramatic reduction of, at least two orders of magnitude when increasing the plasma duration from less than 1 s to 100 s. Indeed, long-pulse operation is usually reached in dominant electron-heating modes at reduced density (current drive optimization) but with low ion temperatures ranging from 1 to 3 keV for discharges above 100 s. Difficulties in extending the duration may arise from coupling high heating powers over long durations and the evolving plasma-wall interaction towards an unstable operational domain. Possible causes limiting the duration and critical issues to be addressed prior to ITER operation and DEMO design are reported and analyzed.

Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

Introduction and the 'grand challenge' of long-pulse operation (LPO)
Controlling fusion plasma for long periods, while gaining experience in steady-state and/or LPO with superconducting magnets, continuous heating and current drive systems, active pumping and fuelling, and active cooling systems that can maintain the plasma-facing components at a stable temperature, is essential for the success of ITER 19 and future fusion reactors.
ITER shall demonstrate the scientific and technological feasibility of fusion energy by combining high-fusion performance burning plasmas (with large values of fusion gain Q, defined as the ratio of fusion power over the absorbed power applied externally to maintain a steady state) with LPO [1,2].The aims of the ITER experiment are to produce and study burning plasma with a larger fraction of non-inductive current and longer durations in different operational regimes, such as (e.g [1-6]): (i) the baseline scenario at plasma current 15 MA and magnetic field 5.3 T, inductively driven burning plasma at Q ⩾ 10 for an extended duration of 300-400 s, (ii) the intermediate or so-called 'hybrid' scenario at plasma current ∼12.5 MA and magnetic field 5.3 T at Q ⩾ 5 for LPO up to 1000 s, (iii) and, ultimately, for steady-state burning plasma conditions in a fully non-inductive current drive regime (∼10 MA/5.3 T) for a total duration up to 3600 s.
Access to LPO will be important to demonstrate the availability and integration of essential fusion reactor technologies and to test components such as the tritium breeding blanket modules to close the tritium cycle in a fusion reactor producing electricity on the grid.
LPO in tokamaks and stellarators means addressing control of stable plasma for a duration much longer than the plasma energy confinement time and approaching plasma-wall interaction (PWI) timescales, where physics processes may still evolve over very long timescales: such as first-wall erosion processes, including material ageing, melting, and dust issues (e.g [7][8][9][10][11][12][13][14]). Figure 1 illustrates the characteristic timescales involved in fusion devices when progressively increasing the time duration of operation [15,16].To operate fusion reactors for a few hours, even in a pulsed mode of operation, the timescales range from 7 to 8 orders of magnitude, i.e. from typically 100 microseconds for magnetohydrodynamics (MHD) phenomena leading to off-normal transient events, such as disruptions and edge localized modes, seconds for energy and particle confinement times, tens of seconds for internal plasma current radial diffusion time to hundreds of seconds for PWI processes (particle retention in the walls, first-wall particle desorption, erosion, and material redeposition) to several hours for first-wall material ageing processes and thermal equilibrium of the whole device.Among the reported challenges, it is worth pointing out that all these processes occurring over various timescales are inter-connected or nonlinearly coupled.For instance, the long-term erosion of first-wall material will lead to the formation of deposits and potentially trigger disruptions (MHD event).To make progress towards achieving LPO, different physics and technology limits set by these various processes with different timescales (as shown in figure 1) must be overcome.In the case of inertially cooled plasma-facing components, pulse length is limited by the amount of energy these components can store for safe operation.However, R&D on actively cooled components offers the possibility of extending pulse duration, limited by the continuous heat exhaust capability of the whole system.It is worth noting that developing the baseline scenario on ITER with a duration of over 100 s will already address many of the challenges of LPO, as all processes that occur over shorter timescales must be mastered and controlled for safe operation.
LPO is often called the 'Grand Challenge' for fusion science [15], requiring pushing the limits of physics and technology integration in a nuclear environment for fusion reactor applications.To address these broad challenges, the development of LPO requires, among other things, a coordinated worldwide effort that encompasses: (1) Experimental programs on existing short-(aimed at physics development or proof-of-principle experiments) and long-pulse facilities for tokamaks and stellarators; (2) Technology R&D programs (e.g.actively cooled plasma unit components, superconducting magnets, linear facilities); (3) Control methods and recovery techniques to maintain a fusion burning plasma within a safe and stable operating domain that can be transferred to ITER and to thermonuclear fusion reactors.
Indeed, the development of fusion energy with stable and safe reactor operation requires: 1.The sustainment and the control of the kinetic and magnetic configuration within the stable boundary domain; 2. Safe and reliable operation of the superconducting magnets; 3. Control of the PWI processes, where all the plasmafacing components are actively cooled and monitored to stay within a safe domain of operation; 4. Efficient fuelling and pumping systems for continuous wave or continuous waveform (CW), control of the fusion fuels, and management of the tritium cycle; 5. Efficient external CW heating and current drive systems for fusion burn access and control; 6. Integrated real-time control for machine protection, safe operation, performance with minimum set of CW diagnostics, CW data acquisition and large data processing tools; 7. CW power supplies, and a heat transfer system for high pulse repetition rate; 8. Comprehensive mastering of the nuclear aspects for operating facilities with high deuterium-tritium 14.1 MeV neutron fluence conditions.
To facilitate the exchange of information at the international level on these overarching challenges, the International Atomic Energy Agency (IAEA) has established a network of experts called CICLOP, which stands for Coordination on International Challenges on Long duration OPeration.The medium/long-term objective of the CICLOP group is to promote activities and to collect and disseminate information on the physics and engineering issues of LPO for tokamaks and stellarator facilities.To further strengthen the synergy between tokamaks and stellarators, the group has set up a high-level 0D multi-machine database that combines physics and technology information for quantifying recent and significant progress on LPO.The group has used the CICLOP database to (i) identify gaps in physics and technology for LPO, and (ii) assess the limiting factors in duration or fusion performance in present experiments.Very recently, this activity has been reported and discussed more broadly within the frame of the International Atomic Energy Agency (IAEA) Technical Meeting devoted to Long-Pulse Operation of Fusion Devices [14-16 November 2022, IAEA Headquarters, Vienna, Austria, https://conferences.iaea.org/event/258/].This paper will report on the development and analysis of the 0-D multi-machine database combining recent results obtained in tokamaks and stellarators.Within this context, it will also provide a high-level summary of the discussions and challenges raised during the aforementioned 2022 IAEA Technical Meetings on Long-Pulse Operation of Fusion Devices [https://conferences.iaea.org/event/258/].After this introductory section that provides the broad context, the paper is divided into three main sections devoted to the: (i) database description and the operational domain towards LPO with the description of the existing limiting factors (section 2); (ii) description of the progress towards the sustainment of fusion performance over long duration by continuing and expanding the pioneering work of Kikuchi and Azumi [17] (section 3); Section 4 will provide a conclusion highlighting the scientific gaps for the development of LPO.

Operational domain for LPO
Significant effort has been made by the CICLOP group to collect and validate a 0D multi-machine database with data coming from both tokamak and stellarator experiments.The database includes experimental data provided by ten tokamaks (Axially Symmetric Divertor Experiment (ASDEX Upgrade) with tungsten plasma-facing components, DIII-D, Experimental Advanced Superconducting Tokamak (EAST), Joint European Torus (JET) with a carbon wall, JET with the ITER-like wall (JET-ILW), JT60-U, Korea Superconducting Tokamak Advanced Research (KSTAR), tokamak à configuration variable (TCV), Tokamak Fusion Test Reactor (TFTR), Tore Supra, W Environment in Steady-State Tokamak (WEST)) and two stellarators (Large Helical Device (LHD) and Wendelstein 7-X (W7-X)).The latest references for these facilities are provided by the overview papers published in Nuclear Fusion following recent IAEA Fusion Energy Conferences [https://iopscience.iop.org/journal/0029-5515/page/Fusion%20Energy%20Conferences],i.e.ASDEX Upgrade [18], DIII-D [19], EAST [20], JT60-U [21] JET [22][23][24], KSTAR [25], TCV [26], TFTR [27], Tore Supra and WEST [28,29], LHD [30], W7-X [31].In addition, annex 1 provides a list of references per facility for the discharges selected in the CICLOP database together with some indication (when available) of the plasma regime.The database comprises a total of 109 pulses and around 3200 data points (on average 29 data per pulse) and is accessible throught an IAEA web page: https://nucleus.iaea.org/sites/fusionportal/ciclop/SitePages/CICLOP-DB.aspx.The pulses have been selected by the scientific contact persons for the respective facility and correspond to the most representative fusion scenarios (cf.annex 1 like ITER baseline scenario, hybrid scenarios, reversed shear scenario, transient hot-ion modes, regimes with internal transport barrier, etc) achieved either in terms of duration or/and fusion performance.In the table of annex 1, an indication is also provided when the discharges have been obtained in fully (or quasi fully non-inductive) conditions (e.g. with zero loop voltage at the plasma edge for tokamaks).For the JT-60U and TFTR data, the pulses are the ones provided by the dataset published by Kikuchi and Azumi [17].The database of LHD is completely open, and the information on the selected discharges is available from the following link: https://doi.org/10.57451/lhd.analyzed-data.
The data provided by the CICLOP's members acting as contact persons for the respective facilities listed above consists of: the facility's name and type (tokamak/stellarator), pulse number, date of the pulse, a published reference, regime of operation with general characteristics (H-mode or L-mode), main fuel species (hydrogen, deuterium, tritium, helium), firstwall and divertor materials, information on the actively cooled components technology, injected additional power Figure 2 is the plot of the total injected energy versus the high-fusion performance duration drawn either for the whole database or a subset of the tokamak database with improved confinement based on the formation and sustainment of an edge pedestal transport barrier, namely the H-mode regime.Figure 3 is the same dataset, where the injected power has been  plotted versus the high-fusion performance duration for the whole database (3-left) and for the H-mode data (3-right).The dashed lines in figure 3 correspond to lines at constant injected energy (the injected power times duration) of either 0.01 GJ, 0.1 GJ, or 1 GJ.Experiments performed with a metallic wall (ASDEX Upgrade, EAST-Metal, JET-ILW, and WEST) have symbols with a black contour line.The graphs indicate that long-pulse regimes (typically above 100 s) are sustained at reduced power, i.e. typically below 5 MW.This is one of the challenging issues when LPO is operated in the H-mode regime, where the power needs to be above a given power threshold to access and sustain the edge transport barrier with good confinement properties (figure 3(right)).Figures 2 and 3 provide an overview of the operational domain in terms of injected energy and power obtained in present tokamaks and stellarators, which is further described below.
The record of injected energy (3.3 GJ/2859 s, nearly 48 min) with 1.2 MW of injected power has been obtained on the superconducting stellarator LHD in helium plasmas with an actively water-cooled stainless steel wall and carbon divertor [32,33] that maintain a constant surface temperature by extracting, in a continuous manner, the heat at the surface of the plasma-facing components.In these experiments, it has been reported that plasma duration and performance were limited by the overheating of some carbon flakes deposited on the plasma-facing components, where the heat could not be properly extracted [32].The flakes detaching from the plasma-facing components could lead to radiative collapses by entering into the hot plasmas and radiating energy.On the stellarator line, W7-X [34] has completed its initial phase of operation from December 2015 up to November 2018, where the energy limit was set by the inertially cooled carbon plasmafacing components.During this phase, plasma durations have been reached up to 100 s by injecting 2 MW of electron cyclotron heating, and were limited by the injected energy that can be accommodated without exceeding a maximum surface temperature set by the inertially cooled system (indeed, without cooling, the surface temperature is continuously increasing with the injected energy) [35].In a preliminary attempt to reduce the surface temperature and the heat load on the components, operation with detached plasmas conditions has been successfully developed, where 30 s long discharges have been sustained at a power level of 5 MW with a reduction of peak heat flux by a factor of 8-10 compared to attached plasmas [36,37].After this initial operational phase, an actively cooled first wall and divertor have been installed during the machine shutdown up to the end of 2021, and scientific exploitation restarted at the end of 2022 with the purpose of significantly increasing the capability to inject high power over long durations [38,39].
EAST has been recently equipped with an actively cooled tungsten divertor and has reached world record injected energy (1.7 GJ/1056 s) in a tokamak with a total radio-frequency (RF) power of 1.65 MW [40,41].In a fully non-inductive regime (without a residual inductively driven ohmic current, where the plasma current is driven by a combination of external current drive sources and the self-generated bootstrap current), the pulse duration was limited by the maximum RF power that could be coupled in LPO (e.g.hot spot formation on the antenna guard limiters or on the divertor).Indeed, it has been recently reported in [41] that 'the limitations of the long-pulse H-mode plasma are from hot spots in the local area of the lower divertor, due to leading edges, and also from sputtering and erosion from the RF-antenna and guard limiter'.To reach this new record, EAST has implemented major upgrades to enhance the machine capabilities towards LPO in terms of diagnostics and control, heating and current drive systems, and with a lower tungsten divertor to enhance particle and power exhaust capabilities with active water-cooled technology.The previous injected energy record for tokamaks was set by Tore Supra with actively cooled carbon fibre composite tiles on limiters, where 1.07 GJ discharges have been sustained for durations of 6 min and 18 s with 3 MW of lower hybrid powers with no sign of saturation of the fuel retention in the carbon wall [42,43].It is worth noting that similar operational limits, as previously discussed for stellarators on LHD, i.e. carbonflake formation with local overheating leading to disruptions, have also been reported on Tore Supra [44].The upgrade to Tore Supra, WEST, involved replacing the actively cooled plasma-facing carbon components with a full tungsten divertor to test the ITER divertor technology, which can withstand the high heat and particle fluxes expected in steady-state operation [28].The WEST experiment also features access to elongated ITER-relevant plasma shapes, improved heating and current drive systems and diagnostic tools, as well as an improved control system [45] to manage the plasma and its interactions with the tungsten walls.In the first phase of operation (2017-2020), the WEST lower divertor has been equipped with a few prototypes of ITER-grade plasma-facing units integrated into the inertially cooled W-coated components.In this configuration, WEST has already obtained 53 s long discharges sustained with 3 MW of injected power [29,46,47].During this phase, the operational space was limited by the absence of active cooling of the lower divertor.A fully actively cooled divertor with an ITER plasma-facing unit was installed in 2021, and operation started in 2022 with the objective to further extend the pulse duration in a reliable manner and address the consequences of high fluence operation on the ageing of the plasma-facing components [29,48,49].
It is worth noting that figures 2(right) and 3(right) indicate that the tokamak database with an H-mode edge is more limited in particular in conditions with metallic walls, i.e. with contributions from the ASDEX Upgrade with an inertially cooled W wall and divertor [50], DIII-D [51] and KSTAR [25] with an inertially cooled carbon wall, EAST when equipped with an actively cooled W divertor, and JET equipped with the so-called 'ITER-like wall', consisting of a beryllium wall and a W divertor but inertially cooled.Indeed, long-pulse H-mode operation requires not only additional powers to be coupled above the L to H power threshold but also to be sustained reliably over long durations.Within this context, EAST [52] and KSTAR [25] have sustained H-mode operation for durations up to 100 s with injected energy up to 425 MJ (as in the end of January 2022).This is major progress towards the development of sustained H-mode regimes for the ITER baseline of operation.
Significant effort should be made to develop LPO at high CW injected power, which is extremely challenging in practice since it requires the coupling of high power in a continuous and reliable manner in configurations where the energy could be extracted continuously without exceeding local overheating of some plasma-facing components.To characterize and compare the heat exhaust capability of the various facilities, a simple proxy has been defined as the ratio of the plasma heating power, P (including alpha heating for burning plasma conditions like ITER), normalized to the whole plasma surface, S, P/S in MW m −2 .Figure 4 shows this ratio P/S calculated for the CICLOP database versus the high-fusion performance duration.This figure highlights the challenge in terms of the heat exhaust capability towards LPO.The highest values of P/S ∼0.3 MW m −2 have been obtained on the ASDEX Upgrade with a tungsten wall, but it is limited in duration by the machine's technical capability (no active cooling and normal conducting magnet).LPO with duration above 100 s has been reached with P/S ⩽ 0.05 MW m −2 and at reduced power below 5 MW.There is an important gap (a factor 4 is P/S) to be filled in the coming years between the presently achieved values and the ITER targets.Indeed, for ITER, the expected ratio of P/S is in the range of 0.2 MW m −2 , to be sustained from 100 s up to 3000 s (when adding the externally injected power to the self-alpha heating power produced by the D-T fusion reactions, P is in the range of 150 MW).The CICLOP group has elaborated, in collaboration with the EUROfusion Operation Network [53], a table summarizing, for both tokamaks and stellarators, the potential limits that should be overcome when reaching high-performance LPO.The objective is to identify, for the pulses in the database, what are the physics and machine limits that prevent higher fusion performance and/or higher duration, as already discussed.The identified limits are summarized in the table below for both the machine/engineering limits (section A) and the plasma physics limits (section B): To progress in LPO development, discharges should remain in the stable operational domain, i.e. away from the identified plasma physics and engineering limits.This objective should be achieved by developing integrated control (continuous control) and event handling (asynchronous control) methodology that should steer the plasma pulses away from the limits and maintain the discharge in the stable operating domain, even with the occurrence of unforeseen events (event handling or an exceptional handling strategy) (e.g [54,55]).This is a scientific challenge for the coming years and, in addition, the development should be carried out using algorithms or methodology that could be transferred to the ITER plasma control system to optimize the operational time on ITER with a limited number of operational pulses in a nuclear and complex environment [55][56][57].
Within this context, facilities that do not have long-pulse capabilities with active cooling systems and superconducting magnets (AUG, DIII-D, TCV, JET in the CICLOP database of operating facilities) should continue to play a leading role by proposing proof-of-principle experiments, particularly for model validation within (or close to) the operational limits and extrapolation (e.g [50,58]), by developing physics experiments that provide the scientific basis for extending the duration, and by validating controllers.
For instance, the DIII-D and EAST research programme has been further developed due to a joint collaboration coordinating research towards long-pulse high-performance operation by firstly developing the scenario physics basis on DIII-D over short durations and, secondly, by adapting and extending the scenario to LPO with a metallic first wall on EAST by taking care, in particular, of new aspects related to core-edge integration and a different heating and current drive mix [59,60].The progress in the joint development of a high poloidal-beta tokamak has recently been reviewed in [61].Indeed, different routes towards steady-state conditions could be explored in short-pulse facilities in a more flexible manner at reduced operational risks before the extrapolation to long-pulse facilities.Inertially cooled components offer a wider and safer operational space, for instance, by avoiding the risks of water leak, leading to a long-shutdown time of the facility, due to local overheating and damage of components operating at high heat flux.Within this context, AUG [18,50], DIII-D [19,51,62], and TCV [26,63] are actively exploring different access conditions to high normalized pressure (β N as defined in annex 2) with different magnetic and/or divertor configurations by combining a high fraction of self-generated bootstrap current with an externally driven current, as recently reported at the 2022 IAEA TM on Long-Pulse Operation of Fusion Devices [14-16 November 2022, IAEA Headquarters, Vienna, Austria, https://conferences.iaea.org/event/258/].
In a steady-state tokamak reactor, the fusion gain (which is proportional to the toroidal beta) and the bootstrap current fraction (which is proportional to the poloidal beta) have to be maximized simultaneously.Therefore, to fulfil both the requirements of continuous operation and high fusion gain, it is necessary to operate at high values of β N (cf.for instance, the review [64]).To illustrate this requirement, β N has been plotted versus the duration of the highperformance phase in figure 5 for the tokamak CICLOP database (figure 5(left)) and for the regimes which are fully (or quasi-) non-inductive (figure 5(right)), as listed in annex 1. High β N values (approaching 4) have been obtained on DIII-D and the ASDEX Upgrade in non-inductive conditions, but the discharges are limited in duration due to the technical limits of the facilities (no active cooling and no superconducting magnets).

Fusion performance and long-pulse duration
In this section, the fusion performance is characterized by the fusion triple product, as in the pioneering work of [17], nT io τ E , where n is the fuel density in the plasma core, T io is the fuel ion temperature in the core, and τ E is the volumeaveraged plasma energy confinement time.As indicative values, it has been shown that in deuterium-tritium (DT) mix plasmas [65], and for an optimal range of ion temperature (14 keV), that the minimum value of triple product to be reached is nTτ E is 4.6 × 10 20 m −3 keV s (or 0.73 atm s) to achieve the fusion amplification Q = 1 (scientific breakeven) and 29 × 10 20 m −3 keV s (or 4.6 atm s) for ignition (Q = ∞), where the external heating source could be switched off.Optimizing the fusion triple product will maximize the fusion amplification factor, Q, despite the fact that there is no direct proportionality between the two quantities [65].For ITER scenarios, as defined in Green et al 2023 [4] and recalled in the introduction of this paper, the indicative values of the triple product are in the following ranges: The fusion triple product (in pressure time second, atm s) has been calculated for the CICLOP database and plotted as a function of the duration of the high-fusion performance phase, as illustrated in figure 6(A).The figure indicates a significant reduction of fusion performance by two orders of magnitude when increasing the duration from ∼1 s to 100 s, confirming previous analysis [17,66].It is clear that machines of different sizes will achieve different values of triple product, and it is only on ITER that a high-fusion triple product and duration could be obtained simultaneously.Nevertheless, even for a given facility without long-pulse capabilities (like active cooling, superconducting magnets), where high-fusion performance has been reached for short durations and attempts have been made to extend duration up to the maximum technological limits (as in the ASDEX Upgrade, DIII-D, JET, and JT-60U), the extension in duration is obtained at the expense of the fusion performance (cf. the individual points for these facilities).In fact, as will be further detailed below, LPO is usually obtained at reduced power (cf. the previous section), reduced plasma current, and in regimes of low density and high electron temperature to optimize the non-inductive current drive effects with, therefore, low fusion performance or triple product.Within this context, figure 6(B) is the same figure as figure 6(A), where the fusion triple product is plotted as a function of the duration of the high-performance phase, but only for the fully non-inductive regimes, as listed in the table of annex 1.It confirms that, whereas the ASDEX Upgrade, DIII-D, JET, and JT-60U have obtained fully (or quasi-) noninductive regimes, the operation could not be prolonged due to the respective technical limits of the facilities.Only EAST, KSTAR, WEST, LHD, W7-X, and the near future JT-60SA will explore, over long durations, high-performance fully noninductive regimes prior to ITER operation.
Future facilities (such as ITER) will have to face the challenges of the technological and the scientific capabilities to combine high-fusion performance of burning plasma with Q larger than one for durations well above the confinement time with an actively cooled wall and superconducting magnets.The scientific challenge will be to find an optimal path where plasma conditions will be maintained in a stable manner.Within the context of ITER scenario preparation, JT-60SA, as a satellite facility to ITER, will have the capability to explore LPO up to 100 s with actively cooled plasma-facing components to cope with heat flux of 10 MW m −2 , and high injected power (40 MW/100 s) together with advanced integrated plasma control for developing an ITER relevant scenario with detached operation by seeding extrinsic impurities like neon or argon [67,68].
The core ion pressures [atm] and ion temperatures have been plotted separately versus the high-fusion performance duration (figure 7) with a view to better disentangle some of the reasons for the reduction in fusion performance with duration.Indeed, the figures indicate that LPO in superconducting tokamaks or stellarators (EAST, LHD, Tore Supra, WEST, and W7-X) are achieved in a domain with dominant electron heating (using mainly RF heating schemes like lower hybrid heating and current drive, electron cyclotron heating and current drive, and minority ion cyclotron resonance heating) but at reduced density to maximize core electron heating and the non-inductive current drive effect generated by external power sources.Consequently, for durations typically above 10 s, these regimes are obtained at reduced ion temperature (⩽3 keV), where the electron and ion fluids are decoupled.In the quest of increasing the fusion performance over long durations, KSTAR has recently sustained a regime with high T io ∼ 10 keV for a duration of 20 s, as shown in figure 7, with low loop voltage to reduce the primary flux consumption [69].Nevertheless, this regime is also obtained at reduced density (consequently, without a significant increase in the core ion pressure and fusion triple product), where a high fraction of fast ions is reported to be an essential ingredient to improve the thermal performance of the hot plasmas.A route for LPO at higher density and higher core pressure remains to be found in these pioneering experiments.On stellarators, when operated with dominant electron heating and reduced density, a similar low level of ion temperature is reported (typically below 1.5 keV in W7-X [70,71]).This is partly attributed to the lack of ion heating and reduced energy collisional exchange power between the electron and ion fluids that scales like the square of the electron density.
To further increase the core pressure, tokamaks and stellarators should be operated simultaneously at a higher ion temperature and density due to, for example, density profile optimization.A higher ion temperature could be reached by directly injecting higher powers coupled to the ion fluid or, like on ITER or future reactors, by increasing the electron temperature and density to enhance the collisional exchange between the electron and ion fluids to provide a broad ion-heating source.Indeed, ion heating will result from the collisional power transfer from the electrons (externally heated and/or self-heated by the fusion-born alpha particles generated by deuteriumtritium reactions) to the ions that depend on plasma density and electron (T e ) to ion (T i ) temperature difference as the energy exchange power scales as n 2 × (T e − T i )/Te 3/2 .A promising 30 s LPO has been achieved in W7-X at high density in detached divertor regimes and led to a sustained high-fusion triple product, as illustrated in figure 7(top) [36,37].Stellarator configurations at high density for longduration operation are sustained without the need for an external non-inductive current drive (that is usually optimized at reduced density in tokamaks) since the magnetic configurations are set by the external superconducting magnetic field configuration.In tokamaks, operation at high density (close to the density limit) requires operation with a large fraction of self-generated bootstrap current to minimize the need for an external current drive.
Figures 8 and 9 are similar graphs to figure 6 but where a subset of the CICLOP's database has been used by selecting either facilities with a metallic first wall/divertor or experiments performed with a deuterium-tritium mix fuel.Figure 8 indicates that significant progress with metallic wall operation in support with ITER and DEMO has been made on the ASDEX Upgrade, EAST equipped with the tungsten divertor, JET-ILW and WEST in its first phase of operation, but the database is still relatively limited.Within this context, and to expand the expertise and know-how for the preparation of ITER operation, facilities will be upgraded to be equipped with an actively cooled tungsten divertor on WEST for phase II of operation beyond 2022 [29], KSTAR [25], and over a longer term on JT60-SA [49,67] and W7-X [31] in a second phase of operation after an initial exploitation for both facilities within the wide operational space provided by the carbon plasma-facing components.In addition, new facilities like the Divertor Tokamak Test (DTT), which is being built in Frascati (Italy) [72,73], will explore long-pulse and high-power operation with an actively cooled tungsten divertor, where applied powers will be increased in a phased approach up to 45 MW [74] to investigate energy and particle exhaust challenges for ITER and future fusion reactors.
The last graph, figure 9, is a summary of the fusion performance versus high-fusion duration for deuterium-tritium A B Figure 6.(A) The fusion triple product, nTτ E , in atm s versus the duration of the high-performance phase [s].Experiments performed with a metallic wall (ASDEX Upgrade, EAST-Metal, JET-ILW, and WEST) have symbols with a black contour line.The horizontal dashed blue line corresponds to the fusion triple product of 0.7 atm s, which is indication of the minimum value to get Q = 1 at the optimum ion temperature, as described in [65].None of these experiments are at this optimum value.The energy confinement time τ E is in the following ranges for each facility: ASDEX Upgrade: 0.04-0.08s; DIII-D: 0.07-0.5 s; EAST: 0.04-0.1 s; JET: 0.2-1.1 s; JT-60 U: 0.2-1.1 s; KSTAR: 0.06-0.14s; TFTR: 0.13-0.3s; Tore Supra and WEST: 0.04-0.09s; LHD: 0.13-0.17s; W7-X: 0.1-0.2s; ITER: 2.5-3.4 s. (B) The fusion triple product, nTτ E , in atm s versus the duration of the high-performance phase [s] for fully (or quasi-) non-inductive regimes only.mix plasmas obtained in TFTR from 1993 to 1997 [27,75], JET during the deuterium-tritium experiment 1 (DTE1) with the C-wall/divertor in 1997 [76][77][78][79], and DTE2 with the ITERlike wall in 2021 [22,23,80,81].The TFTR experiments are characterized by high-fusion performance hot-ion modes called 'super-shot', where record ion temperatures 20-40 keV (cf.also figure 7) have been reached transiently (less than 1 s) to optimize the fusion power up to 10 MW.These regimes were obtained in a limiter tokamak (i.e. with no divertor), but with intense wall conditioning by injecting lithium pellets into the torus that has resulted in improvements in deuteriumtritium fusion power production [82].In a transient ELM-free regime, JET during DTE1 has obtained record fusion power (16.1 MW) and high triple product values (∼1-1.4atm s) in a transient manner (less than 1 s) with high ion temperature.ELMy H-mode regimes were sustained during 5 s with a fusion triple product of ∼0.3 atm s at a level of 4 MW of fusion power but were limited by the heating power duration at a level of 24 MW [77].Following the installation of the ITER-like wall at JET in 2011, the upgrade of the neutral beam system and diagnostics, significant effort has been made by the JET team to develop stationary and reproducible high-fusion scenarios in deuterium for a duration of 5 s that have transferred in deuterium-tritium plasmas, as shown in figure 9 with values of nTτ E ranging from 0.45 to 0.50 atm s, sustained for a duration of 4.3-5.6 s [22,23].Finally, neon-seeded radiative scenarios with (semi-)detached plasmas were tested for the first time in DT plasmas, where the benefit of neon injection in proof-ofprinciple experiments is shown in figure 9 with the increase in the fusion triple product compared to similar reference pulses but without neon seeding.This integrated ITER-relevant scenario that could be extended up to 10 s with reduced heat flux on the divertor requires optimization in DD and DT and will be further developed during the third DT campaign called DTE3 [83,84] that is taking place in 2023.To conclude, the sustainment of burning plasmas during LPO will be one of the main and unique scientific objectives of the ITER programme and will fill a major gap in the development of fusion power production.

Conclusions
High-fusion performance and LPO are major integration challenges between physics and technology for tokamaks and stellarators on the path towards fusion power-plant development.
To address these challenges, a fully integrated vision of the physics and engineering aspects is required to simultaneously increase the time duration and performance of the plasma discharges.As reviewed in this paper and illustrated through the exploitation of a multi-machine 0D database set up by the CICLOP group, significant progress has been made recently towards extending the plasma duration and/or performance.Nevertheless, gaps between the present results and the need to prepare and optimize ITER operation and beyond have been identified by either following the analysis of the multimachine CICLOP database or highlighted during discussions at the 2022 IAEA Technical Meeting devoted to Long-Pulse Operation of Fusion Devices [14-16 November 2022, IAEA Headquarters, Vienna, Austria, https://conferences.iaea.org/event/258/].The major programme gaps that have been identified are summarized below: 1. LPO is usually performed at reduced power level (below 5 MW), and there is a need to operate at higher power levels in regimes that are compatible with the power handling capacity of the facility.As described in section 2, the power handling capacity of the facility is limited both by the machine/engineering (e.g.limit in energy or temperature for the PFC) and physics limits (e.g.transient events like disruption or ELMs, first-wall erosion).2. LPO is usually performed in attached divertor conditions, which is not the operating regime for ITER and the reactor.There is a need to sustain a fully detached (or semi-detached) divertor regime for long-pulse duration and address compatibility with the core plasma performance.In addition, the ageing of plasma-facing materials (divertor, first walls, etc) shall be addressed in long-pulse detached regimes, as well as in accidental attached conditions.The long-term impact of the materials' evolution on plasma operation also needs to be assessed.3.There is a need to further develop operational procedures (e.g.plasma wall conditioning) that are not only valid on a given facility but that can also be transferred to ITER in a safe and reliable manner within the context of nuclear operation.4.There is a need to further develop and test control methods and plasma performance recovery techniques to maintain the plasma within a stable operational domain (e.g.control in case of loss of divertor detachment that may lead to local overheating of the plasma-facing components).
The developed methods should be transferable to ITER and beyond.5. Finally, LPO should be viewed within the context of operating a nuclear facility with a large level of neutron flux and fluence.
The planned and foreseen upgrades in existing facilities (heating and current drive upgrades, divertor and first-wall modifications), together with the operation of new facilities like JT-60SA and the DTT in support with ITER and reactor design, will provide ideal platforms to address, in a coordinated manner, the scientific challenges required to efficiently address the gaps listed above towards safe and sustained LPO of ITER and fusion powerplants.(

Disclaimers
This report was prepared as an account of work sponsored by an agency of the United States Government.Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights.Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof.The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.
ITER is a Nuclear Facility INB-174.The views and opinions expressed herein do not necessarily reflect those of the ITER Organization.

Figure 1 .
Figure 1.Characteristic timescales involved in fusion devices of typically ITER size.The exact value for the various timescales depends on the exact machine size, plasma parameters, and first-wall technology.
[MW], injected energy [MJ], plasma duration [s], duration of the highperformance phase [s], core ion temperature T i0 [keV] and density n i0 [10 20 m −3 ], the plasma energy confinement time τ E [s], plasma current [MA], toroidal field on axis [T], safety factor, major radius [m], minor radius a [m], elongation, and triangularity.For this reference paper, the data has been collected up to January 2022.The descriptions and definitions of the variables are provided in annex 2.

Figure 2 .
Figure 2. (left) Injected energy versus high-performance duration for the whole database; (right) for tokamak H-mode data only.Experiments performed with a metallic wall (ASDEX Upgrade, EAST-Metal, JET-ILW, and WEST) have symbols with a black contour line.

Figure 3 .
Figure 3. (left) Injected power versus high-performance duration for the whole database; (right) for tokamak H-mode data only.The dashed lines correspond to lines at constant injected energy (the injected power times duration) of either 0.01 GJ (black), 0.1 GJ (blue), or 1 GJ (red).Experiments performed with a metallic wall (ASDEX Upgrade, EAST-Metal, JET-ILW, and WEST) have symbols with a black contour line.

Figure 4 .
Figure 4. Plasma heating power normalized to the plasma surface (P/S in MW m 2 ) vs the high-performance duration.Experiments performed with a metallic wall (ASDEX Upgrade, EAST-Metal, JET-ILW, and WEST) have symbols with a black contour line.

Figure 5 .
Figure 5. (left) Normalized pressure, β N , versus high-performance duration for the CICLOP tokamak database; (right) a similar graph for the discharges in fully (or quasi-) non-inductive conditions.

Figure 7 .
Figure 7. (top) Ion plasma pressure, n T io , in atm and (bottom) core ion temperature, T io , in keV versus the duration of the high-performance phase [s].Experiments performed with a metallic wall (ASDEX Upgrade, EAST-Metal, JET-ILW, and WEST) have symbols with a black contour line.

Figure 8 .
Figure 8.The fusion triple product, nTτ E , in atm s versus the duration of the high-performance phase [s] for operating machines with a metallic wall/divertor.

Figure 9 .
Figure 9.The fusion triple product, nTτ E , in atm s versus the duration of the high-performance phase [s] for deuterium-tritium experiments.
Grant Agreement No. 101052200-EUROfusion).Views and opinions expressed are, however, those of the author(s) only and do not necessarily reflect those of the European Union or the European Commission.Neither the European Union nor the European Commission can be held responsible for them.This work was supported in part by the US DOE under Contracts DE-FC02-04ER54698 and DE-AC52-07NA27344.
Annex 1.The pulse list, associated references, and operation regime per facility.