The IAEA DEMO Programme Workshop Series: 2012–2021 report

The IAEA DEMO Programme Workshop Series was initiated to be the reference forum to debate, evaluate and establish the next steps to be taken within the international fusion community to deliver fusion as a reliable source of clean energy. The DEMO programme refers predominantly to a future magnetic confinement Tokamak design concept, after the ITER project, with a goal of delivering electricity to the grid. To enable this, the programme workshop series provides a unique frame where the discussion and analysis of the progress and findings of the various DEMO programmes, and not just the presentation of results being the major goals. More recently, due to the construction development of ITER, the workshop has also provided a forum to bring together the fusion community and industry. This is an important development for DEMO programmes, which will be more reliant on industry given their focus on devices proximate to a commercial fusion power plant. In this paper a summary of all editions of this workshop (2012–2013–2015–2016–2018–2019) up to 2021 are summarized. Topics addressed are listed and findings and open questions pointed out for each edition.


Introduction
The IAEA DEMO Programme Workshop Series arose from the recognition, within the international community, that the beginning of the ITER project marked a transition from a fusion programme dominated by physics to one in which engineering played an increasingly significant role. This was * Author to whom any correspondence should be addressed.
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. particularly evident when considering the devices that would follow ITER, often referred to as 'demonstrators' (DEMO), aimed to demonstrate the performance representative or close to that of a fusion power plant (FPP), focusing predominantly on magnetic confinement TOKAMAK design concepts. DEMO should be considered as one of the steps in between ITER and a commercial FPP based on TOKAMAK design.
As described by the 'EU Fusion Roadmap': '[…] demonstration power plant DEMO, which will deliver hundreds of megawatts of electricity to the grid and operate with a closed fuel-cycle. […]. The ultimate goal is commercial electricity, and so it is critical that DEMO is on this path' [1]. For these devices, engineering, and the issues attending the construction of less experimental devices, such as maintenance and lifetime, are key determinants of cost and feasibility. Several fusion programmes had already produced preliminary designs of DEMO and there was no consensus on the requirements for these next-step devices and no single roadmap for fusion beyond ITER.
This workshop series was initiated to be the reference forum to debate, evaluate and establish the next steps to be taken within the international fusion community to deliver fusion as a reliable source of clean energy. To enable this, it provides a unique frame where the discussion and analysis of the progress and findings of the various DEMO programmes, not only the presentation of results, are the major goals. The ethos was to encourage open debate of the scientific and technological challenges encountered and solved, identification of new challenges and to reach an international consensus on the main issues. It was also foreseen that there was scope to add substantial value through encouraging international cooperation in defining and coordinating the various DEMO programme activities. Thus, the workshops encourage discussion and debate around the major challenges rather than publicizing single achievements, providing a different type of forum for the fusion community.
More recently, as construction of ITER has progressed, the workshop has also provided a forum to bring together the fusion community and industry. This represents an important development for DEMO programmes, which will be more reliant on industry given their focus on devices proximate to FPPs. It is important that the fusion community understands the present status of industrial capability when considering their designs.

Scope of the workshop series
The workshop is organized by the IAEA Scientific Secretariat assisted by a Technical Programme Committee (TPC) led by the TPC chair. Members of the TPC are invited by the Secretariat on the basis of their recognized standing in the fusion community or by recommendation of members of the international community and considering geographical criteria. A TPC member is expected to have a broad and current knowledge of the work being done in their field(s) of expertise, have a leading position and be well connected in their field.
Each workshop contains three topics that are determined by the TPC to be of significant impact to the DEMO and FPP programmes. Each topic has two co-chairs, one a member of the TPC and an external representative recommended by a TPC member and who is co-opted onto the TPC. A key feature of the workshop is that the presentations have to provide an overview of the sub-topic in an international context, highlighting recent developments and new challenges that have been identified. Each speaker has to address the issues raised by considering the subject in the context of DEMO, the work and resources needed to solve the issues, and the analysis of the problem in the context of integrated DEMO design. To ensure that this objective is achieved, the co-chairs guide the generation of the presentation of each of their speakers through iteration and close communication. This ensures that the speakers fully understand their brief and that the presentation serves the purpose of the workshop; in this respect the workshop also differs from a normal conference.
The identification of gaps in the knowledge and the facilities needed to fill these gaps are the major outputs of the workshops. It has become increasingly apparent in the community that relevant supporting facilities are needed but their provision is not often included in the DEMO programmes presented by different states. Therefore, the workshop is the place where these problems can be identified and influence the various DEMO programmes, especially through international cooperation.
A key feature of the workshop is the significant time for debate included in the agenda. For this reason, the number of presentations is limited. The debates are an opportunity for the delegates to raise matters of conflict between different technologies or proposed concepts, challenge perceived orthodoxies and identify gaps in the programmes or propose solutions. They are often wide ranging, and participation is high.
In addition to the three workshop topics, time is allowed for at least three special topics which are stand-alone presentations on specific subjects. These may include, but are not limited to: i. national or international roadmaps or strategies towards the realization of fusion power launched or significantly updated since the last workshop, ii. novel approaches or innovations in any relevant discipline, iii. significant studies on integrated DEMO designs completed since the last workshop.
One of them is usually reserved for the workshop host to present work relating to their own fusion programme.
The workshop concludes with a summary session of each topic presented by one of the topic co-chairs and final remarks by the workshop chair and the scientific secretary. The presentations and a final report provided by the TPC chair is made available to delegates.
The workshop operates on an 18-24 month cycle with meetings in spring and autumn to avoid the main conference season.

Editions
There have been seven workshops to date as shown in table 1, which also includes the topic titles discussed at each one. The workshop has evolved rapidly from its first incarnation at UCLA, with increasing discussion time in the programme as requested by delegates.
This aspect of the workshop has proven popular with attendees and the discussions have been wide-ranging to allow exploration of multidisciplinary areas and points of impact between disciplines in the DEMO design. It is often within the context of discussion that common themes arise between workshops that are not necessarily covered by the specific topics. Some examples of 'thematic' content, as opposed to programme content, are shown in table 2 together with the year(s) in which each appeared. It is interesting to note that issues International practices on regulation of future nuclear fusion power plants, covering safety and security, radwaste management and considerations for safeguards. related to engineering and technology are strongly present in topics selected within the workshop series. This reflects the concern of DEMO programmes about the construction of a device that demonstrates all the technologies needed or a FPP.
One of the frequent themes is the application of technology readiness levels (TRL) to fusion, subject of much debate in the community. Material related to each edition of the workshop can be found at the DEMO Workshop Programme Series website: https://nucleus-new.iaea.org/sites/fusionportal/Pages/ DEMO_landing.aspx.

Summary of workshop 1
This edition aimed to define programs and facilities to solve their scientific and technical issues, and to identify opportunities to make greater progress through international collaboration.
The conclusions identified the complex challenges of the blanket-first wall, arising from the multiplicity of functions it performs in a harsh and complex operating environment. It was strongly suggested to pursue laboratory experiments, and the most important immediate step to facilitate this was to build new multiple-effect laboratory facilities to address key issues (such as thermofluid-MHD behavior of complex geometry blanket designs, the impact of neutron irradiation on properties and performance, high duty-cycle plasma exhaust processing and remote handling and maintenance of blanketfirst wall components).
It was also highlighted that high level availability represents a major challenge, and that reliability, availability, maintainability and inspectability (RAMI) issues will be a key consideration for DEMO. In this regard setting goals for mean time between failures and mean time to replace were more meaningful than setting lifetime goals for materials. It was essential to establish a reliability growth programme of test/fail/repair/improve to achieve higher availability for DEMO.
Participants, experts in the field of plasma power exhaust and impurity control, emphasized that an integrated approach is needed to develop solutions for DEMO, where the challenges are far more demanding than those of ITER. A selfconsistent strategy must take into account the core plasma physics performance; the exhaust of power through the plasma edge, scrape-off layer and divertor; and the choice of materials and heat removal technologies for the plasma-facing components. It was underlined that there was no reliable predictive modeling capability for plasma exhaust and that a combination of improved diagnostics and greater effort through international collaboration was needed to solve these problems.
Considering the magnetic configuration and operating scenario for a next-step fusion nuclear facility, it was clarified that, despite a diversity of views on its mission and design characteristics, there is the need for a deuteriumtritium-fuelled, high neutron-loaded, high heat-flux, tritium self-sufficient next-step facility. The progression and timeline for major facilities leading to commercial fusion is highly intertwined with the strategy for materials development, but the optimum strategy for advancing materials testing and system integration was not clear. The workshop discussions highlighted some themes that hinted at characteristics of a DEMO programme still in its early planning stages. For example, the importance of ITER as a critical element of the DEMO programme was affirmed. Beyond ITER, it was clear that the roadmap needed to include both integrated fusion nuclear facilities and fusion material irradiation facilities. The roadmap as well as the optimum modes of collaboration will be defined by the initiatives that are taken by parties to construct and exploit these large-scale facilities.

Summary of workshop 2
A benchmarking exercise comparing results for a given design problem was shown. The result was the finding of small differences between different fusion design codes and approaches in use. The physics and technology assumptions were specified rather than left to the user, but in practice different groups make different assumptions in their DEMO studies, varying in their degree of optimism about the state of future scientific and technological knowledge. Perhaps the most important critical issue, considering both degree of uncertainty and design impact in combination, was in the plasma exhaust area, where work is needed to improve the understanding and distil this into improved models to be used in system codes. Although stellarator design codes are more configuration-specific than for tokamaks, there were efforts to integrate stellarators into existing system code frameworks, using similar tools to evaluate designs wherever possible.
Concerning plasma power exhaust and impurity control, a consensus emerged on the 'zero-th order' requirements for a DEMO physics solution. Plasma heating power (the sum of auxiliary and alpha-particle self-heating power) must be exhausted by radiation as far as possible in order to reduce peak heat fluxes incident on target surfaces to values compatible with technological limits, about 5-10 MW m −2 . In addition, the plasma temperature at the material surface needs to be below about 5 eV everywhere (a condition known as 'full detachment') in order to minimize power load from recombination and limit erosion and plasma contamination by the target material. As in the first workshop, the need for an integrated solution for the DEMO power exhaust problem was identified. The most urgent of the needs was reliable experimental measurements, with focus on the transport of impurities in the boundary and divertor plasma regions. In addition, the appropriate divertor size and geometry for full detachment needed further investigation. Reliable control of detached conditions must be demonstrated experimentally. The applicability of advanced divertor configurations, such as Super-X, Snowflake and liquid metal to DEMO await a more complete understanding of their benefits and design impacts.
It was clear that the assumptions about DEMO plasma controllability and prospects for control solutions vary across the world community and strongly affect the different perspectives on the design and operating scenario for DEMO. The strong link between core conditions and the requirements for plasma exhaust, already highlighted in topic 2, implied that an integrated core and edge control strategy is necessary. The control requirements for reliable operation of a high-performance plasma close to the core or divertor operating limits will be challenging, so adequate margins must be imposed. The consequences of disruptions and other fast timescale events must be taken into account from the beginning. The commonalities and differences between stellarators and tokamaks were considered. Progress in plasma control sensors and actuators was made as a result of the extensive worldwide R&D effort in support of ITER. However, the configuration and operating environment of a DEMO plasma would be very challenging for many current measurement and heating techniques, thus new solutions will be required.
The reliability requirements for all DEMO systems were again raised with provisions for redundancy and rapid return to service to minimize control-related failures and downtime. The energy efficiency of actuators, especially of heating and current drive (HCD) systems, was identified as being of greater importance for the net energy-producing DEMO plant than in experimental and control-related R&D. Anticipated planning is required to ensure that it is off the critical path for DEMO. Developments of new techniques have to start at an early stage and concepts with breakthrough potential should be tested as soon as possible in order to assess their potential and their impact on DEMO scenario choices.
A report from Japan on a recent assessment of fusion plant safety and security, focusing on issues of decay heat and tritium safety, attracted great interest. The 2011 accident at the Fukushima Daiichi nuclear power plant has drawn public attention to safety and energy policy, and the assessment by the Japan Society of Plasma and Nuclear Fusion Research (JSPF) provided a useful update on fusion safety.

Summary of workshop 3
The participants discussed the ITER project, as well as a set of planned integrated devices intended to take significant steps in fusion nuclear science and technology beyond ITER. The expected accomplishments of ITER and its contributions to DEMO physics, technology and programme planning were presented. The next-step machines currently being studied by several ITER partners would clearly make important advances that go well beyond ITER, but to what extent they could fill gaps toward FPP readiness is yet to be quantified as these plans mature. These machines themselves have readiness gaps, especially for their later stages, for which R&D is necessary in the near term. Plans for filling these gaps needed to be clarified.
The subject of availability appeared again in relation to the in-vessel systems and the need for careful strategic decisionmaking in the early stages of plant configuration development. Reliability requires having a database of materials properties for relevant conditions and designing to be robust with ample margin against damage and synergistic effects in the harsh fusion in-vessel environment. Remote maintenance (RM) was also considered as a key design driver of tokamak architecture, requiring, for example, adequate space for in-vessel piping, the decoupling of primary functional requirements, and the possibility to access auxiliary systems without dismantling.
The engineering problem of DEMO is the design of a truly integrated machine with the competing demands of the constituent parts of the device and their impact on each other, and the effect of uncertainties as they propagate through the incipient integrated fusion machine designs. Participants highlighted that the step beyond ITER will require innovation in the plasma scenario and in-vessel systems, including the divertor hardware configuration, and the materials and technology of plasma-facing components.
Participants considered the relationship between the ITER TBM projects and long-term blanket development needs for DEMO. There is an opportunity for blanket development to take advantage of the facility. However, it was clear that very large extrapolations are required for many significant design parameters, such as tritium breeding rate, neutron dose, coolant and liquid mass flow, and extracted heat. The main contribution of the TBM experiments will be to generate a database that can be used as a benchmark for the validation of modeling tools needed for DEMO design, an area in which international collaboration can be very useful.
The multiple plans for the next-step tokamak machine appear to be very similar in how they would contribute to the basis of future steps. All of them might have the potential to demonstrate net electricity at some stage, but they would not necessarily go far enough in physics and technology to fill the readiness gaps for commercial power plants. Supporting facilities focusing on narrower sets of issues, which might reduce technical risks for the next integration steps, are less prominent in the planning. It is not clear at all whether the emerging programmes are the optimum ones in terms of number and diversity of planned facilities. These circumstances led to discussion of a possible international strategy to improve coverage of DEMO needs that are currently under-addressed, to reduce duplication, and to be more robust against setbacks. Considering the costs of the next steps of fusion, there could be significant advantages in an international strategy for planning and coordination of work.
Discussions at this and previous workshops have raised questions about the nonproliferation aspect that must be considered in the planning of fusion. Further dialog with experts was necessary to understand the technical measures needed and when they should be implemented to be most effective.

Summary of workshop 4
Topic selected raised significant uncertainties and development issues for the post-ITER machines. The presentations stimulated lengthy discussions on each topic, and cross-cutting items were identified, such as the impact of plasma scenario on tritium burn-up rate, fuelling efficiency, HCD efficiency and the heating power required to reach L-H transition.
The topic on tritium provided an overview of the required tritium systems and their performances for a DEMO device, including the plasma fusion reaction. It was clear that any DEMO machine (including the stellarator-based option) must be designed for optimal fuelling and pumping efficiency. It was also clear that the approach must be holistic and integrated, considering all the aspects of the tritium fuelling, breeding and processing cycle.
The tritium start-up inventory and tritium availability post-ITER was discussed. The former is strongly dependent on the tritium burn fraction, f b , the tritium fuelling efficiency, η f , and the tritium processing time, t p . For a fusion device generating 3 GW fusion power the start-up inventory can vary from 10 kg to over 30 kg [2]. Decreasing the processing time and increasing the burn-up fraction and fuelling efficiency create margin for the largely unknown tritium breeding ratio (TBR). It was concluded that the R&D goal should be to achieve the product (f b × η f ) > 5% and not less than 2% with a tritium processing time of less than 6 h. It was also recognized that there is a need to reduce the uncertainty in many parameters, both fundamental data and derived quantities.
The need to scale tritium plant by factors of 10 or 20 identified major technology and safety gaps from ITER to DEMO. The workshop also noted that there is no practical external source of tritium for fusion beyond ITER and a strategy must be developed to address this.
The workshop concluded that extrapolations in plasma physics from present day experiments towards DEMO often have limited validity. Some 'global' parameters can be chosen in agreement for both small and big devices, for example, the plasma density as a ratio of the Greenwald limit (n/n GW ) but some more fundamental physics parameters, such as the ion Larmor radius normalized to plasma radius (ρ * ), cannot be made to match simultaneously. With the aim to clarify open issues towards DEMO, an international programme is required to address the problems of scaling from current experiments, including: i. development of new scaling laws that use a more physicsoriented approach (and which also better cover the DEMO relevant parameter range), ii. improvement of the theoretical understanding of underlying physics, iii. increased use of integrated modeling using detailed physics models, iv. benchmarking with the largest available experiments from today (JET, JT-60SA will play a major role), v. final validation on ITER (unfortunately only possible in ∼20 years from now but the ITER programme should be designed to accommodate this).
During the workshop it was emphasized that this may be the right time to launch an international collaborative programme of integrated scenario modeling incorporating benchmarking across numerous experiments. Other areas for future research include confinement modes (QH-and I-), which offer the prospect of edge-localized mode (ELM)-free operation. The development of quantitative predictive modeling of fully detached plasmas and its validation were considered as a priority. The fully detached plasma scenario requires impurity seeding to control the plasma power radiated from the core and the power radiated from the scrape-off layer. However, the impurity seeding also makes the experimental validation difficult.
The uncertainties in the physics basis and models lead to a cautious approach to engineering with large operational margins applied to component design and over-specification of system requirements. It represents a major obstacle to advancing DEMO engineering design and limits the ability to select the most appropriate technology. The key areas should be identified and addressed by the fusion community, making best use of those machines that can provide DEMO-relevant operational parameters. It is important to note that a final confirmation of some of the physics assumptions will only come with ITER. There is a requirement for extensive engineering/ technology facilities for testing of the components on a variety of scales. The specifications of these facilities should be developed and prioritized (according to criteria to be identified).
The uncertainty in the predictive capability of plasma physics has a major impact on the HCD programmes for DEMO. Confidence in the integrated physics models must be achieved to determine the HCD system requirements. A stable scenario is needed before the engineering design activity can begin. Ideally the HCD mix should be established as early as possible to avoid unnecessary resourcing of non-viable systems. The role of HCD systems as actuators for plasma control needs to be considered.
DEMO proposals from the EU, China, India, Japan and Korea all include electron cyclotron (EC) at frequencies between 170 GHz and 300 GHz (driven by B field value), a range that encompasses the ITER EC system but extends to well beyond present availability [3,4]. There is also a requirement to increase the output power to above 1 MW, in order to reduce the number of gyrotron units. The simultaneous increase in frequency and power and continuous wave (cw) operation is a major challenge for gyrotron technology. Present achievable wall plug efficiency is ∼60% at lower frequency and power, but availability needs to increase to ∼98% to meet even the EU-DEMO modest requirements. Ancillary systems such as steering, waveguides and windows compatible with high cw power also need to be developed.
At present, despite the obvious problems of higher energy operation (and hence reliability, cost and efficiency) neutral beam injection (NBI) has also been selected by all five DEMOs. The major challenge to NBI is the poor wall plug efficiency with a gas neutralizer (∼30% at present) and significant R&D will be required to develop the more promising alternative photoneutralizer. This implies the construction of large test facilities (the ITER PRIMA facility may be useful in this context). The management of extraneous power deposition is a general issue for high power NB systems.
No system is yet at the required level of maturity; they all require further R&D for the source, transmission or improved efficiency. The RAMI database for the HCD systems is scarce and a data collecting and analysis methodology needs to be established that all machines contribute to. The requirements of the HCD systems to act as plasma control actuators must be addressed within the context of the HCD mix. HCD systems development is concentrating on efficiency and RAMI but programmes are not adequately addressing the nuclear aspects of the designs such as material activation, shielding etc.
There was considerable discussion of TRLs and applicability to DEMO. A fusion DEMO development programme will require a methodology for systematically tracking technological maturity and progress toward an end product. The causality dilemma between component testing and representative testing facilities is a challenge for fusion and may require either some adaptions of the TRL approach or alternatives involving standards organizations and licensing and regulatory authorities.

Summary of workshop 5
The 5th DEMO workshop had two unique features: the inclusion of magnet design and engineering (as opposed to field magnitude) with a special focus on high-temperature superconductors and the involvement of speakers from industrial organizations currently engaged in fusion, namely Toshiba Energy Systems and Solutions Corporation (Magnets) and Assystem (RM). The participation of industry provided a different perspective to the DEMO issue, which was enlightening and thought-provoking. In particular, the 'sporadic' nature of fusion projects, with years between each major build, creates problems for industrial partners due to the lack of continuitythere is no incentive to establish a 'fusion industry'. Other key points regarding the feasibility of present DEMO designs, described in the key points below, were also raised.
The plasma control session focused on the boundary conditions for implementation of diagnostics and actuators on DEMO and elaborated on the limitations for plasma control arising from these. High neutron and gamma fluence, as well as strong fluxes of charge-exchange neutrals, will act on the diagnostic front-end components and sufficient lifetime (reliable operation over a long time) of these control components can only be achieved by mounting them in protected (retracted) locations, which leads to severe limitations in measurement performance.
The development of DEMO control is aiming as much as possible to benefit from the progress on the development of the ITER control system [5]. Beyond ITER first plasma, plasma control systems preparations, it is assumed that important outputs to be used further on DEMO will be integrated control methods for burn phase, ramp-up and rampdown, low/no ELM scenarios, minimization and handling of disruptions.
The development of DEMO control has to be quantitatively assessed via simulations. In this context a new software tool under development was presented in which a core plasma simulation was coupled with a Simulink control tool. In one example, a missing fuel pellet was shown to lead to a significant drop of fusion power, even when strong additional heating was applied to stabilize the situation. In a second simulation, the effect of tungsten impurity particles falling into the plasma was analyzed. It was also found that in the case of tungsten influx of more than 3 mg the resulting core plasma cooling by tungsten radiation cannot be compensated by stopping the Xe injection and by adding auxiliary power, such that the discharge evolves into an H-L back-transition and then disrupts [6].
For large tokamaks with superconducting magnets the main poloidal field and central solenoid coils are mounted at some distance from the plasma with shielding structures in between, such that the effective stabilization speed available from these coils tends to be slower than the characteristic time constant of the vertical instability of highly shaped plasmas. This problem can be solved by the installation of additional fast 'in-vessel' control coils as is the case on KSTAR, EAST and ITER. However, the feasibility of such coils for DEMO still needs to be fully determined. For equilibrium control of long-pulse or steady-state discharges, drifting magnetic measurements may have to be amended by alternative approaches such as reflectometry, polarimetry/interferometry and optical methods.
It was recognized that the plasma control requirements for DEMO will concentrate on maximizing availability, not flexibility, and hence differ significantly from those for present machines and ITER. Nevertheless, significant work such as burn control emulation, disruption control and operation with limited actuators/diagnostics can be carried out on existing machines. It is important to establish a dialogue between DEMO designers and existing machine operators (and ITER) to influence programmes.
The stellarator has several principal advantages compared to the tokamak. First, there is no need for current drive, and resulting from this, current-driven instabilities leading to fast disruptions do not occur in a stellarator. Second, the plasma density limit on a stellarator is much higher than in a tokamak, which enhances the freedom to choose an operating scenario and design point within the multi-dimensional parameter space [7]. Third, equilibrium control is simplified due to the absence of the vertical instability. With regard to power exhaust, stable detachment via open loop impurity injection as well as sustainment of a radiative divertor by resonant magnetic perturbation (RMP) application have been successfully demonstrated on stellarators. Hollow density and temperature operating regimes, which may need to be controlled in a fusion machine, have been identified. Core density collapse events expelling a large fraction of the stored energy within a few hundred microseconds have been observed in a certain plasma regime on the large helical device (LHD), but this regime can be avoided for machine operation.
The RM session considered the impact of RM on tokamak design. The need for high plant availability drives a challenging requirement for rapid maintenance, which is safe, highly reliable and recoverable. These requirements can only be met by an early integration of maintenance into the plant design. This makes the maintenance system a design-driving activity. Existing experimental fusion devices do not have such critical maintenance requirements. Key methods for achieving early integration of maintenance into plant design were identified as: i. developing maintenance-oriented strategies, ii. understanding the maintenance space constraints, iii. developing enabling technologies, and iv. minimizing the cost of the support facilities.
There is an important balance between the plant design and practical maintenance to achieve a feasible DEMO design. For example, all the major DEMO concepts now use a vertical maintenance system to avoid the large vessel and magnet size arising from a horizontal strategy, but this introduces a potential conflict with safety requirements as it is necessary to lift heavy components large distances above containment structures. Thus, development of the maintenance strategy and systems must be underpinned by consideration of the requirements to satisfy the safety regulations for a nuclear facility. In addition to the lifting issue, other safety considerations were identified: managing the contamination hazard, particularly activated and tritiated dust within the vacuum vessel (VV); moving confinement barriers to allow maintenance operations, particularly opening the VV to a hot-cell above the tokamak; and impact between components during accidents, particularly blanket and divertor cassette collisions during a seismic event. Regulator involvement will not occur until the designs are more mature, but regulator and stakeholder requirements can be anticipated, and a qualification and regulatory compliance strategy developed. It was also noted that a robust systems engineering approach is required to record the requirements rationale and option selection criteria so that this information is available once the regulator becomes involved with licensing the plant.
It was emphasized that the stakeholder investment protection requirements to minimize risks will be stringent and their impact on the maintenance strategy can be comparable to safety requirements. For example, damage to DEMO during the maintenance phase may have only limited safety implications if safety functions are not compromised, but it could have major operational impact. The maintenance system will require novel technologies that can pose significant technical risks. Those risks need to be identified and mitigated early in the design process. Rapid and reliable service joining for the plasma-facing components is a key DEMO enabling technology that is required to achieve the maintenance duration requirements and to satisfy the nuclear regulator. The design of the tools and their deployment drives the layout of the blankets, ports and pipes, therefore it is important to understand the feasibility and the tool size and deployment options at this early stage in the design of DEMO. Non-destructive testing (NDT) was identified as a major technical risk to service joining because there are no existing systems available, although ultrasonic inspection (including phased array) and electromagnetic acoustic transducers can provide some of the volumetric data required. ITER and industry provide a good knowledge base, but they are not considering the full range of DEMO service joining requirements, particularly in terms of speed and reliability required.
The key findings of the meeting are summarized as: (a) The design philosophy of DEMOs must adapt to include manufacturability, inspectability, safety and stakeholder requirements and decommissioning (including recycling).
There are significant gaps in these areas between ITER and DEMO requirements, (b) The ITER RM scheme omits key aspects of the DEMO requirements: 1. Servicing and joining 2. Speed and reliability 3. NDT 4. Irradiation and activation Enabling technologies/systems need to be developed early and in parallel with the DEMO design: (c) The plasma control requirements for DEMO will focus on maximizing availability rather than flexibility, differing significantly from those for current machines and ITER. Nevertheless, significant work such as burn control emulation, disruption control, operation with limited actuators/diagnostics can be carried out on existing machines. It is important to establish a dialogue between DEMO designers and existing machine operators (and ITER) to influence programmes, (d) Magnet designs need to consider how to minimize radial build and costs, including the proper grading of the conductor; the use of high temperature superconductor (HTS) is best considered for high-field applications (see also point I above). Experience on ITER with Nb3Sn shows that creating a high demand does not necessarily create a price reduction, (e) Experience from W7X shows that cold testing of the magnets to full load, preferably with the final equipment for power supply and quench protection is advantageous considering that some faults appear only under these conditions.

Summary of workshop 6
The workshop focused on the three major topics shown in table 1. For the first time, the meeting specifically included topics related to the effects of irradiation on materials and its impact on the lifetime and reliability of manufactured components. These topics combined to include discussion on the role and expectations of the regulator with respect to both plasma behavior and component lifetime. The interdependence of plasma stability and robustness of in-vessel components was highlighted. The session on plasma transients, defined as a deviation from the steady-state burn phase, included effects in stellarators and divided such events into two types: 'foreseeable', such as plasma ramp-up and ramp-down and 'un-planned', such as might result from failure of a control component. It was noted that the steady state may also show temporal variations of fusion power, for example, due to sawteeth and ELMs, and it would be essential to ensure compatibility of plasma-facing components with such events. This consideration also extends to the divertor, where the divertor plasma may become detached, another fault condition, and subsequently re-attached. It should be noted that with present DEMO designs operation close to detachment is necessary for performance. ELMs may not be acceptable in a FPP due to the high power loading they present to the PFCs. It was pointed out that stellarators also exhibit ELMy H-modes. The development of plasma scenarios other than H-mode may provide a solution and, since they are passive, are preferable to RMPbased solutions.
Sudden stored energy loss from the plasma, such as triggered by malfunction of the control system or by radiative losses from dust entering the plasma, needs to be quantified in terms of frequency and magnitude. Although the stellarator does not suffer disruptions, it is still subject to these events that occur over a fraction of the energy confinement time.
It was concluded that transients should be expected, and that design of any DEMO should proceed on this assumption. Mitigation of disruptions and ELMs may not be a feasible option and the plasma operating scenario must be designed and analyzed for robustness against these events. This requires the development of a plasma 'flight simulator', a plasma code capable of predicting all effects of operation including off-normal events. For failure of the control system this must be implemented at the single-component level. This code will enable design of the plasma scenario and analysis of its robustness against unforeseen events. It will also enable predictive control of the plasma to be developed. Such a code must be validated, and this will need a structured programmatic approach for experiments and modeling (in competition with more academic physics studies).
More emphasis is required on the investigation of component failure modes, in order to achieve the reliability required of a DEMO. This will require dedicated test facilities and in silico techniques to be developed. A systematic analysis of the contribution that ITER can make to these problems should be undertaken.
The session on irradiation of materials and components covered a wide range of issues, from the neutron irradiation effects on candidate materials to manufacturing and inspection of components. Multi-discipline discussions between experts are needed, such as those hosted by the IAEA DEMO Workshop series. There is still a high level of uncertainty in the requirements and performance of various joining techniques, particularly under neutron irradiation. Therefore, a well-defined and planned programme of neutron irradiations, under relevant conditions (yet to be defined), needs to be planned and then performed, as well as the accumulation of a comprehensive and consistent database to inform on materials performance criteria, rules and validation. It was emphasized that testing must include joins, both of dissimilar materials and identical materials and for a range of joining techniques. The severe lack of data for functional materials, such as coatings, and materials used in diagnostics was noted. Moreover, the need for neutron irradiation data, obtained under relevant conditions was emphasized and concerns were raised over the current lack of a fast-neutron spectrum reactor, considering that the current mixed spectrum reactors are approaching end of life and would disappear.
The development of design rules defining irradiation limits is essential to ensure structural integrity, but the challenge is that it must be done in situations where there is no (or very little) experience with the actual reactor environment. Irradiation effects are considered in some design codes (e.g. RCC-MRx), but since existing codes were developed for fuel cladding materials and stainless steel, a variety of R&D efforts are needed to adopt or develop design criteria for fusion incore components. Other fusion-specific design rules for brittle fracture and cyclic softening need to be developed. Reduced activation ferritic martensitic (RAFM) steel is the only (and likely the only) material that can withstand the qualifications required for structural DEMO design work. The current irradiation database of RAFM steels consists of statistically low-quality fission irradiation data. Since there are currently no fusion neutron sources that can provide large irradiation volumes, a probabilistic approach with Bayesian estimation has the potential to resolve this ambiguity.
Mixed spectrum, fast spectrum, and 14 MeV fusion neutron sources are needed to support DEMO material/component development and DEMO qualification. However, the currently operating mixed-spectrum reactors, our 'workhorses' in irradiation database maintenance, are expected to systematically disappear in the near future. Furthermore, the lack of international fast-spectrum reactors useful for irradiating W materials is a disturbing problem. It was noted that currently planned mixed-spectrum and fast-spectrum reactors may fill the gap, but that community efforts are needed to realize the potential of these facilities for experiments on fusion materials and components. With regard to fusion neutron sources, there is great promise and development, but they may not be consistent with DEMO reactors, and a robust modeling program is needed to convert the data.
Modeling can identify important processes and mechanisms that contribute to radiation-induced property changes. In other words, the effects of alloy chemistry and microstructure can be largely explained by models from the atomic to the mesoscale. Micromechanical models can also provide guidance on probable failure mechanisms. However, specimen size/volume effects and mechanisms determined by 'weakest links' need to be considered. Microstructure-mechanical property correlations-heavily guided by experiments-can provide some predictive capability for anticipating component failure conditions (temperature, stress, time/effects etc.). However, modeling alone may not be sufficient to replace data from experiments in representative irradiation environments, particularly in regulatory space. In such situations, a combination of modeling and fission/ion irradiation can help guide the use of fusion neutron sources and the screening and selection of candidate materials.
While the ITER project would help to inform some of the issues of manufacturing for DEMO, for example, issues arising from the large number of welds within the VV, there is need of further, collaborative work to solve specific DEMO issues, such as tritium permeation, thermal and stress management and remotely deployable NDT and inspection techniques.
Given their limited market, the cost of the advanced materials that may be adopted for DEMOs raised an additional concern, and international consensus and cooperation may be needed to control this.
The third session concerned the materials engineering of plasma-facing components. It was underlined that significant progress had been made by R&D in these areas. An example was the development of 'smart', self-passivating tungsten alloys and fiber-reinforced tungsten, which potentially offer significant performance advantages for both first wall and divertor applications [8]. These are now progressing towards industrial-scale production. Despite these advances, the session emphasized the need for certainty in the plasma edge conditions, which will determine the performance requirements of the first wall and divertor.
It was noted that the water-cooled divertor design derived from ITER (and now current in all DEMO approaches) severely limited the operational conditions and the lifetime due to the known behavior of the CuCrZr heat sink. It was suggested that alternative materials or approaches would be necessary for an economic FPP. Although, liquid metal first wall and divertor options are under development, there are large gaps in understanding the behavior of these systems in DEMO, in particular the interaction with the plasma edge and the interaction of the liquid metal with the supporting structure under neutron irradiation.

Extended summary of workshop 7
The 7th DEMO workshop took place in virtual mode due to the covid-19 pandemic and covered only one topic: the international practices on regulation of future nuclear FPPs, covering safety and security, radwaste management and considerations for safeguards. The following subtopics were presented and discussed:

Tritium and its safety issues (P. W. Humrickhouse, ORNL, USA)
The fusion of deuterium and tritium is achievable at lower temperatures than other prospective fusion reactions. For this reason, fusion energy research efforts worldwide have long focused on D-T fusion machines and the D-T fusion fuel cycle, despite a variety of challenges associated with it. Many of these are directly related to the fact that tritium, an isotope of hydrogen containing two neutrons, is radioactive-it undergoes a weak (18.6 keV maximum) β decay with a 12.3 year half-life. The low energy of this decay poses no external radiation exposure hazard; absorption and especially inhalation are the primary exposure pathways. In gaseous form (e.g. T 2 or HT), most inhaled tritium is simply exhaled with minimal absorption. However, as an isotope of hydrogen, it is readily incorporated into water, organic molecules and other materials; in oxidized (water) form (e.g. T 2 O or HTO), it is readily absorbed and consequently 10 000 times more hazardous. Following absorption, its biological half-life is approximately 10 days.
Each D-T fusion reaction consumes one tritium atom and produces 17.6 MeV of energy; any D-T machine, then, will consume 55.8 kg of tritium per GW-y of fusion power produced. Because of its relatively short half-life, there exists no natural supply, and consequently future devices must simultaneously breed tritium in a lithium-bearing blanket at least at the same rate, and quickly recover this bred tritium for reuse as fuel. This breeding/extraction process is referred to as the outer fuel cycle (see figure 1). The inner fuel cycle (IFC) provides deuterium and tritium to the plasma, exhausts unburned fuel, ash and other (e.g. plasma enhancement) gases and separates the former from the latter for re-use as fuel. While the rate of tritium consumption and breeding is already significant, presently achievable fueling efficiency (η f ) and burn fraction ( f b ) are low (η f f b ≈ 0.09% [9]), and so the throughput in the IFC must be correspondingly higher, in which case tritium inventories in the processing systems can become quite large (10 s of kg) and pose a significant potential radiological hazard. Improvements in fuelling efficiency, burn fraction and processing technology are therefore needed in order to improve the situation.
Perhaps the most basic strategy for confinement of radionuclides-surrounding them with a sealed, solid barrier as with fission reactor fuel cladding-is incompletely effective for tritium, which has a unique ability to diffuse directly through solid materials (and particularly metals) at significant rates. The rate of this permeation typically increases exponentially with temperature (the activation energy and constant of proportionality, the permeability, vary by material), and is therefore increasingly problematic in high-temperature blanket coolants, which are otherwise desirable in order to achieve high thermal efficiency in power conversion. Modeling of plant-wide permeation and transport is an important part of fusion machine safety analysis, and experimental determination of material properties such as permeability (and others related to tritium transport in fluids and at interfaces) is essential to inform these.
Without additional mitigation, rates of permeation through large networks of machine system piping may exceed allowable releases to the environment and resultant radiation exposure to individuals [10]. Exposure limits for members of the U.S. public are prescribed by multiple parts of the U.S. Code of Federal Regulations (CFR). 10 CFR 20.1301 limits dose from routine releases (from all sources) to 1 mSv yr −1 ; doses from air pollutants are limited in 40 CFR 61-0.1 mSv yr −1 , and 40 CFR 141.16 limits doses from drinking water to 0.04 mSv yr −1 , the practical implication of which is a drinking water activity limit for tritium of 740 Bq l −1 . Doses from off-normal events (accidents) are limited to 250 mSv per event by regulation, though doses above 10 mSv per event would require public evacuation. The desire to avoid such a requirement in any scenario has led to a 10 mSv per event limit in Department of Energy standards in the U.S. [11].
A variety of mitigation schemes are being researched in order to reduce permeation rates and plant inventories of tritium. A significant driver of plant inventories in an ITER-like fuel cycle scheme is the need to separate exhausted, unburned tritium and deuterium; since a 50/50 mix of these is what ultimately what must be injected as fuel, direct internal recycling schemes (see figure 1) aim to bypass this isotope separation step and simply remove ash and other gases from the D-T mix, significantly reducing the necessary tritium inventory in the process [12]. Continuous pumping technologies including continuous cryopumping and super permeable membranes are being investigated that would support this in future devices. Limiting permeation will likely be accomplished via a combination of efficient extraction systems and permeation barriers at various scales, from thin oxide coatings on component surfaces, to actively de-tritiated heat transfer equipment vaults.
Finally, while mitigations along the lines of those described above may successfully reduce tritium releases to the environment below levels mandated by regulation, smaller releases on a regular basis will nevertheless result in measurable increases in the amount of tritium in the environment, in air, water and organic material. Though the resultant doses and concomitant health risk may be negligible, such changes to the environment may nevertheless prove to be a significant obstacle to social acceptance of fusion energy, which emphasizes the importance of reducing these to the maximum extent reasonably achievable.

Fission and fusion licensing approaches (J. Herb, GRS gGmbH, Germany)
GRS together with KIT performed a study which included the comparison of fission and current fusion licensing approaches [13]. GRS also did a review of the safety concept for fusion device concepts and transferability of the German nuclear fission regulation to potential FPPs [14] together with IPP, KIT and the Institute of Applied Ecology. Based on these studies the following findings and results were obtained.
A screening of open literature about the regulatory approaches, focused on facilities which include tritium, showed that no country has a dedicated, comprehensive, fusion-specific regulatory framework for the whole lifecycle from siting to decommissioning of fusion facilities. The safety requirements applied to fusion facilities are based primarily on experience with activities related to fission facilities. Some countries have regulated fusion facilities (e.g. UK, France, Germany) with or without tritium.
The regulations for radiation facilities and radiation protection apply and form the basis for licensing of fusion facilities. It was found that there seems to be a gap for licensing of larger fusion facilities or FPPs, because they have higher radiological hazard potential compared to typical radiation facilities, but a lower radiological hazard potential compared to fission power plants.
There are issues with the application of the existing graded approaches in nuclear regulations because they is linked to the thermal power of fission devices and do not credit the less radio-toxic inventory of fusion devices. It was found that the graded approach seems to be more easily applicable to new technologies like fusion than the prescriptive approach.
The licensing of the ITER and JET demonstrate that existing regulations for facilities like research and/or commercial reactors (basic nuclear installation, 'INB') or radiation generating facilities ('other non-nuclear licensed sites') can be applied to fusion facilities. For both facilities, more goalorientated licensing frameworks are applied.
Also, different aspects were identified which should be considered in the development of future fusion regulations. Different aspects of the fission legal and regulation framework were found which should also be considered in a framework for fusion regulation. But there are other aspects which are fusion specific, based on the different technology of fusion physics and technology which also needs to be included in fusion-specific regulation.
Different regulations, safety standards and safety guides issued by the European Commission and the IAEA were screened. Based on these, a recommendation was made as to how to implement a legal and regulatory framework specifically for fusion facilities. It was recommended that the legal framework should be based on Council Directive 2013/59/Euratom-European Basic Safety Standards and Council Directive 2009/71/Euratom (amended by Directive 2014/87/Euratom)-Nuclear Safety Directive. Also, it was recommended that the regulatory framework should include general safety requirements for siting, leadership and management, safety assessment and decommissioning of facilities. It should include development of a safety concept for FPPs based on: the specification of safety objectives; application of the defense in depth concept, the concept of multi-level confinement of radioactive inventory; protection against internal and external hazards; application of a graded approach; a system for operating experience feedback; consideration of fusion-specific safety aspects; identification and development of codes and standards; and specification of the interface between safety, security and safeguards.

ITER licensing.
ITER is the first fusion machine to require a construction license decree under the French nuclear facilities regulations (nuclear basic installation order). The decree was passed in 2012 after the IRSN conducted an in-depth assessment of the safety and radiation protection measures adopted by ITER Organisation (IO). ITER will operate with up to 4 kg of tritium. It is classified as a nuclear basic installation (NBI) because of its radioactive inventory since 10 16 Bq (27 g of tritium) are enough to be classified according to the French legislation.
Concerning the radiological hazard potential of a fusion machine, even if this hazard in the event of an accident may be lower compared to a fission reactor, this is not the case for normal operation and maintenance. Gaseous tritium releases during operation and maintenance for ITER are higher than those for all pressurized water reactor (PWR) in France (58 reactors), see table 3. ITER is licensed according to the French NBI order which applies to all nuclear facilities on French soil, but its specificities are taken into account by the safety expertise needed for the licensing process. Even if plasma discharges are intermittent on ITER, it cannot be considered a radiation generating facility (similar to a medical center which uses ionizing radiation) for radiation exposure hazard because of the long duration of plasma pulses (7 min).
In France, a goal-setting approach that focuses on performance and outcomes is always preferred for every NBI whenever possible and as long as specific limits imposed by safety regulations are respected (shielding and evacuation doses in case of an accident, doses for worker etc.). For ITER, a balance between prescriptive and goal-setting approach has been chosen.

Differences between ITER and DEMO.
Conceptual design of the tokamak is the same but future demonstration machines will mainly differ from ITER by seeking to attain tritium self-sufficiency and achieve significantly longer operating times. These differences will have a significant impact on design and a direct influence on safety.

Self-sufficiency in tritium.
Obtaining selfsufficiency in tritium production is essential for the industrialization of fusion machines. Therefore, in DEMO, a complete tritium breeding blanket will be installed in the VV, not like in ITER where individual experimental TBMs are expected to be installed. Tritium produced by the breeding blanket has to be maximized, taking into account a limit on the surface of the VV covered by the breeding blanket.
The lower the minimum achieved TBR, the easier it will be to achieve reaction self-sufficiency. The minimum TBR can be achieved by maximizing the burn-up fraction and by minimizing the time for recycling the tritium.

5.3.2.2.
Operating times. The average operating time with plasma must be increased (in comparison to ITER) for the industrialization of fusion machines. This leads to: (a) an increase of the number of displacement par atoms (dpa) due to 14 MeV neutrons, and so to a faster deterioration of the initial properties of the materials in the structures surrounding the plasma. DEMO will then use materials capable of withstanding intense neutron bombardment; (b) a probable incompatibility with the maintenance operating time expected for ITER. The need to reduce maintenance operation time may have major design impacts compared to ITER design (i.e. automated maintenance with large and heavy vehicles, very large penetrations) which have to be taken into account in the safety analysis.

Residual heat removal.
Residual heat removal is not a safety function in ITER since the temperature rises slowly in the case of a loss of cooling systems. But residual heat is much higher in DEMO due to longer operating time and greater fusion power. In the case of a loss of cooling systems the temperature of the first wall may increase and if an air ingress is experienced, WO 3 (as a highly volatile, radioactive aerosol) may be created. This event is excluded in ITER but has to be analyzed in DEMO from the early design stage.

5.3.3.2.
Ionizing radiation exposure risk. Radiation exposure risk due to the 14 MeV neutrons is high in ITER but it will be higher for DEMO. ITER dose targets are: average 2.5 mSv yr −1 and maximum 10 mSv yr −1 . To have comparable dose targets in DEMO, low neutron activation materials will be needed. Also, activation of all kinds of TBM has to be taken into account at an early design stage. Finally, for DEMO higher doses rates for blanket modules and the divertor cassette during cask transfer and maintenance will be expected.
It appears that DEMO design must undergo a process of optimization for the doses received by operators from the early design stage.

Accidents to consider in DEMO.
Preliminary studies showed that the types of accidents to consider for DEMO as a design basis and design extension condition (DEC) are identical to ITER. However, likelihood and consequences may be very different because of the greater radioactive inventory and energy involved.
For example, two design basis accidents (DBAs) for ITER (loss of coolant accident, LOCA, and loss of vacuum accident, LOVA, in the VV) lead to a tritium and dust explosion. Tritium (1 kg) and dust (1000 kg) limits in the VV are imposed by the regulator (those values have been provided by the ITER Organization). If we have comparable limits for DEMO, how will those limits be respected with the increased operation time for DEMO? Possible trials are: to reduce dust production and tritium absorbed; to lower dust cleaning time or collecting dust during plasma pulses; higher operation temperature to limit tritium absorption. Or it may be demonstrated that radiological consequences with increased tritium and dust quantities are acceptable (additional design measures/physical barriers).
Increased tritium inventory in DEMO means increased potential dose released during normal operation and accidents. DEMO design should include an optimization of release paths from the early design stage.
Permanent plasma discharges are considered in order to increase the operating time in DEMO. This means a more sophisticated plasma control system and malfunctions of the plasma control system may be more frequent and the consequences may be different/more severe.
Increased magnetic energy of coils in DEMO (TC 10 GJ versus 2.28 GJ for ITER): an electric arc may lead to a loss of integrity of the VV (first confinement barrier). This accident has to be studied (for ITER, this event does not damage the VV).
Increased He inventory in the magnetic field coils cooling system. A leak of liquid He may increase the room/equipment/VV pressure and affect their integrity.
DEMO breeding blanket design may be different from the TBMs that will be tested on ITER. A leak of TBM cooling systems must be studied.
Extreme events must be considered at early conceptual design stage.
DEMO wastes may be different/more tritiated than ITER ones. Wastes management constraints based on the general policy of the country hosting the machine must be considered at early conceptual design stage.

Fission and fusion experimental facilities cooled by liquid metal (V. Kriventsev, N. Virgili and L. Gaertner, NENP-NPTDS, IAEA)
The identification of existing experimental infrastructures, as well as of new experimental facilities, based on the recognized R&D needs of Member States with fast-reactor programmes is considered a priority. The IAEA actively promotes the harmonization of these efforts at the international level. As part of these efforts, the IAEA maintains an online catalog of the experimental facilities designed in support of the development and deployment of liquid metal-cooled fast-neutron systems (LMFNS). The latest edition of the catalogue includes 192 facilities, half of them is devoted to the sodium-cooled fastreactors and another half to systems cooled by heavy liquid metals, such as lead and lead-bismuth eutectic (LBE). Fourteen facilities can use both sodium and heavy metal coolants. In addition, eleven facilities included in the LMFNS catalogue are used or can be used to study fusion systems cooled by lead-lithium eclectic alloys. General purpose facilities and laboratories (i.e. creep laboratories, metallographic laboratories, etc) as well as research reactors are not included in the LMFNS catalogue as they are considered beyond its scope. The facilities are structured in the LMFNS catalogue by their applications, namely, by the coolant of the target systems (either sodium or heavy liquid metals), the maximal facility power or zero-power facility for Verification, Validation and Qualification (V&V&Q) and licensing purposes; DBAs and DECs, as well by the primary research topics, such as thermal hydraulics, coolant chemistry, structural materials, systems and components, instrumentation, and in-service inspection and repair. The final node for every catalogue item is a facility profile, which is provided by the vendor or developer and based on the predefined template developed by the IAEA.
For each facility included in the LMFNS catalogue, the following data are outlined in the facility profile: Several facilities included in the LMFNS catalogue were designed or have been already used to investigate the behavior of fusion systems cooled by liquid metals, such as lead-lithium eutectic. Flow circulation in the reactor loops, electromagnetic pumps, heat exchangers and other components can be studied. While many data obtained for liquid metal-cooled facilities can be directly used for assessing thermal hydraulics in the circuits of the fusion device, the behavior of the coolant and structural material under intensive irradiation by high energy 14 MeV neutrons remains a challenge to be addressed in future research.
Corrosion of structural materials in lead alloys is considered as one of the main challenges for the design of a heavy liquid metal-cooled reactor as it directly influences lifetime limits and circuit design. Efforts have been devoted to short/medium-term corrosion experiments in stagnant and flowing LBE. Few experiments have been carried out in pure lead and new experimental campaigns are planned. It is expected that the interaction between structural materials and coolants will be also a challenging issue for the future fusion energy systems. The oxygen control limits, operational temperatures, maximal possible coolant flow velocities and degradation mechanisms induced by the high-energy neutrons have already been studied for fast-neutron reactors. While additional efforts are required to resolve these technological challenges, the extended experience can be used to identify, plan and design fusion experimental facilities for fusion systems (https://nucleus.iaea.org/sites/lmfns/Pages/default.aspx).

EU-DEMO: radwaste classification and implications for disposal (M. R. Gilbert, UAKEA, UK)
The generation of radioactive waste from fusion machines has often been viewed as not being an issue. As fusion begins to consider the near-commercial devices that will succeed ITER in the next few decades, there has been renewed interest in assessing the volumes and classification of waste likely to originate from next-step concepts. Historical environmental impact assessments for fusion focused on the recycling possibilities of fusion materials and concluded that since recycling was theoretically possible there was no waste or disposal issue expected for future FPPs. This perspective has since faded in view of disposal because the latter can be viewed as the simplest, safest, and thus default option. However, within the last few years in Europe, waste predictions considering a scenario where all fusion machine components will be sent for disposal, demonstrate that EU-DEMO may generate significant volumes and masses of radwaste [19][20][21].
The target within Europe has been to achieve near-100% low-level waste (LLW) or better within 100 years of the end of operation, but this, according to the latest predictions, is challenging, or even impossible, to meet using the current criteria for classifying radioactive waste. While recycling-in-principle only considers the γ-dose from in-vessel (and VV) materials, waste disposal repositories generally consider the whole radiological output, which includes β-emissions. Predictions based on the disposal limits specified by different European radwaste repositories (including the UK) show that a large fraction of in-vessel EU-DEMO components may remain at intermediate-level waste (ILW) activity levels for hundreds or thousands of years. For example, figure 2 shows, for a typical EU-DEMO concept, the simulation prediction of the time evolution in mass of waste that would be ILW at the point of disposal, demonstrating that 10 000 of tonnes of waste would be ILW even after 100 years after the end of life.
Most long-lived waste issues foreseen in fusion materials originate from minor elemental components and impurities. For the novel low-activation steels being developed for invessel structural application there are several minor elements, such as nitrogen and niobium, which will generate problem long-lived radionuclides. Even with alternative compositions/grades or tighter impurity control it seems unlikely that the issues for these key alloys can be completely mitigated. Even some functional materials, such as tungsten and beryllium can contain low levels of impurities that cause a disposal problem, which, again, cannot be entirely avoided unless it is possible to spend billions of £/$/€ just on materials.
On the other hand, exploring beyond Europe, we find that some international repositories would view fusion in-vessel waste more favorably, while some countries/regions are not limiting the possibilities of fusion based on achieving acceptable (low-level) radwaste. Both observations suggest that radwaste from DEMO could be reclassified to allow use of existing LLW repositories based on a reassessment of the risks and limits associated with disposal, or alternatively, a fusion-specific near-surface repository could be constructed that would accept most (all) fusion waste without the need for expensive deep-geological disposal (the default for ILW in general).
However, regardless of the relative severity of the waste or the cost of its disposal, the volumes of waste generated from EU-DEMO would be very large, perhaps 50 000 tonnes or more, just from the core and bioshield of a DEMO-sized power plant. Disposing of such large volumes is unlikely to create a favorable message around the low environmental impact of fusion and so the possibility/feasibility of recycling needs to be revisited. As before, assessments based on dose rate show that EU-DEMO could be largely recyclable within 100 years. Realizing this recyclability for fusion materials requires significant development of technologies and industries capable of accepting the material, and there needs to be clear messaging from governments and regulators to promote the development of recycling, including defining acceptance criteria to allow material to be cleared for reuse. 5.6. IAEA activities in safeguards, radwaste, security and safety (S. M. Gonzalez de Vicente, P. Calle Vives and J. Whitlock, IAEA) 5.6.1. IAEA safeguards: an overview. IAEA safeguards are technical means implemented to verify that a State is honoring its international obligations, under the Nuclear Non-Proliferation Treaty (NPT), to use nuclear material and technology for peaceful purposes. States have different obligations depending on their respective status under the NPT: non-nuclear weapons States (NNWS) conclude a 'comprehensive safeguards agreement' (CSA) with the IAEA; nuclear weapons States conclude a 'voluntary offer agreement' with the IAEA; and States not party to the NPT conclude an 'itemspecific agreement' with the IAEA.
Most States are NNWS and therefore operate under a CSA, obligating them to accept IAEA safeguards on all source (natural U, DU andTh) or special fissionable material (Pu-239, U-233 and U-235) in all peaceful activities in the State, and to ensure that these safeguards will be effectively applied. Some States also voluntarily report holdings of Am and Np. Many States also have an additional protocol (AP) to their safeguards agreement in force, which allows the IAEA broader access to relevant information and locations in the State. These AP measures significantly increase the IAEA's ability to verify the absence of covert nuclear activities in the State, in addition to declared activities. IAEA safeguards measures under a CSA and AP include design verification, nuclear material verification (including inspections), and 'Complementary Access' access to investigate other nuclear material, locations or activities of interest in the State.
The speaker offered a number of ideas about how nuclear fusion activities might be relevant to IAEA safeguards, including the verification of any nuclear material used in fusion activities (e.g. DU shielding) as well as the verification of the absence of undeclared nuclear material or activities.

Overview of IAEA activities in radwaste and safety in
fusion. IAEA is currently carrying out activities in the areas of radwaste management as well as safety and regulation for fusion applications. The goal of both activities is to gather member states' experiences in these areas. This information is essential for the future development of recommendations on adequate safety and regulatory frameworks.
The contribution from international experts and institutions to these activities are of paramount importance for two main reasons: to incorporate all existing knowledge and to come to a technical consensus on open questions when possible. Participation in these activities is still possible and welcomed by the IAEA.
In the area of radioactive waste management, a review paper has been published recently in the Nuclear Fusion Journal giving an overview on the management of radioactive waste from fusion facilities: ITER, demonstration machines and power plants [22], and subsequent preparation of an IAEA TECDOC is in progress.
Besides, in the field of safety, work is being done towards the development of an IAEA FPP safety framework. The first steps consist of gathering information among Member States on current practices and plans, followed by the development of high-level design and regulatory principles for demonstration and commercial FPPs. This work will be summarized in technical documents as follows: i. a publication to capture Member States practices and plans for the regulation of fusion facilities including FPPs, ii. a publication to capture Member States practices on design safety approach and safety analysis for FPPs, iii. a publication that lists and explains general principles for the design safety and regulation of FPP. The document could be expanded to cover other areas of the FPP lifecycle.

Conclusions
As a final outcome of this edition of the workshop a number of open issues to be followed up by the international fusion community are listed: 1. A strong technical basis supporting the safety/licensing justification needs to be defined. Safety culture needs to be built into the fusion community. It should be taken into account that national regulations will apply and that they can be different for each country. 2. Prescriptive versus goal-oriented approaches to licensing are possible. Fusion may require a hybrid approach. 3. Pre-conceptual design activities need to consider licensing issues and identify them with the purpose of taking action at earliest possible stage. 4. In the 'ITER to DEMO comparison' it is important to better understand hazard severity and potential licensing obstacles. Also, the regulation applicable to radiological/medical facilities needs to be analyzed from the point of view of applicability to fusion devices and understanding the limits of this approach. 5. The timeline to develop a licensing framework for fusion also needs to be established: both sequential and parallel development to the design of the first-step facilities are possible, but the optimum approach needs to be defined. 6. Environmental tritium emissions from FPPs are an outstanding issue. Low release levels need to be ensured. 7. The control and minimization of tritium inventory is an issue driven by tritium migration and site inventory. R&D efforts to clarify and minimize those are of highest priority. 8. Liquid metal solutions need to be explored more in detail. Certainly, replacing solids with liquids has potential advantages. Power fluids play a significant role in potential accidents and their mitigation. 9. Choice of coolants for fusion devices is a complex balance of thermal performance, solid-liquid compatibility, nuclear/chemical/toxicity factors, reliability, thermal conversion performance etc. All factors needs to be taken into account and extensive R&D is still needed. 10. The standard fusion target of 'LLW for shallow land burial after 50-100 years' is challenged by the practical reality of material choices and impurities in commercially available materials. Fusion-specific materials like fusion RAFM steels (EUROFER97, F82H) have been developed to fulfill that requirement. 11. Fusion will generate significant radwaste volume (LLW mostly and ILW). How to address this issue need to be studied, including the recycling option.
under the support and coordination of the International Atomic Energy Agency.

Disclaimer
The views and opinions expressed herein do not necessarily reflect those of the International Atomic Energy Agency or other agencies or institutions.