A staged approach to Indian DEMO

We present a revised strategy for Indian DEMO in the context of new technologies and concepts in fusion research. The central idea behind the new strategy is that the power plant is a reactor-park consisting of multiple, preferably compact, reactors with moderate fusion power (∼1000 MW) with 35%–50% availability for each. The DEMO is a single net electricity producing unit that becomes the basis for replication into multiple units on a commercial scale. One of the key enablers for the revised strategy is the emergence of high-temperature superconductors for high field magnets. For a steady-state burn we show that there exists an optimum regime of plasma β and confinement where the fusion gain is maximum. Thus, we adopt a strategy with moderate confinement regimes and plasma β. This makes current drive a necessity for the reactors. Based on these considerations a four-stage approach to DEMO is proposed. It is argued that an electricity producing pilot plant (PP) with fusion power of 200 MW–300 MW is needed before the DEMO to establish the power performance, tritium breeding and its re-use over sufficiently long pulses. An integrated test facility must precede the pilot to test and qualify the technologies for the pilot stage. The revised approach takes into account realistic assumptions on power balance, current drive efficiency and magnet lifetime-dose; factors that pose constraints in identifying potential reactor configurations. Parameter choices for possible options for the integrated test facility (Fusion Engineering Science and Test), PP and DEMO are presented that can be used to initiate conceptual designs and directed R&D.


Introduction
The DEMO design strategy is evolving worldwide because of the advancement in fusion plasma science, the emergence of new ideas, advanced manufacturing techniques, new materials and as other technologies mature with time.New learning is arising from the experience of constructing complex and first-of-a-kind systems for ITER.These elements are influencing the strategies for the realization of fusion energy that aim for a high-performance, cost-effective DEMO.Almost a decade earlier, the designs of DEMO were aimed at large devices with fusion power of ∼3 GW.Many of these designs were based on the idea of a single step from ITER to DEMO [1][2][3][4][5][6][7].
Lately, this has changed to at least a two-stage program where an intermediate step of a fusion pilot plant (PP) seems necessary to fill in the gaps between ITER and a power plant [8][9][10].These PP concepts are of a few 100 s of MW of fusion power, smaller in size compared to ITER but with the capability to demonstrate electricity production [11][12][13][14][15].
The most important goal of a PP is to reliably demonstrate the electricity production as well as the tritium fuel-cycle.Furthermore, the critical technologies used in a PP should be scalable for a power plant so that ultimately the objectives of net electricity production and tritium self-sufficiency are met.Earlier explorations of PPs were of spherical tokamak (ST) type having copper magnets, which made it possible to have demountable magnets with joints, that would have allowed an ease and efficiency in the replacement of the components.In fact, one could conceive high neutron fluence, compact, volume neutron sources as well to test components in a real reactor environment [16][17][18].The copper magnets are less complicated compared to superconductors for small devices with short pulse operation.However, for the longpulse operation the resistive power dissipation in the copper is known to be quite large.It is so large that a reasonable net-electric power would be obtainable only when the fusion power is large (∼a few GW), adversely affecting the service life of the in-vessel components, magnets and the duty cycle.Thus, it would be considered more efficient to go directly for a DEMO from ITER by using low-temperature superconducting (LTS) magnets.The down side of this is that demountable joints are not possible with LTS and given the constraints arising from the limiting magnetic field and the neutron load from practical considerations of component replacement, one must adopt large enough devices as the only viable solution.This stage is now set to change.The technological advances in the field of high-temperature superconducting (HTS) materials-their ability to have higher current density as well as higher limiting magnetic field compared to LTS and most importantly, their feasibility to make demountable joints have become important in setting new directions for the DEMO strategy.Although the ideas for demountable magnets were proposed by Powell et al in 1980s [19] the work on HTS joints by Hashizume et al in 2002 identified some of the configurations and their issues [20], which was later followed by [21][22][23] to evolve into a STARS concept (stacked tapes assembled in rigid structure) [24] with dramatic reduction in joint resistance by a factor 10 −5 .A 'comb-teeth' joint model was developed by Mangiarotti et al in 2014 [25] and an industrially scalable, experimentally qualified design called VIPER (vacuum pressure impregnated, insulated, partially transposed, extruded, roll-formed) has been reported by Hartwig et al [26].In a recent paper, the feasibility of lowresistance, dry-type demountable joints was demonstrated by Weiss et al [27].
In addition to the demonstration of electricity production and tritium extraction, these PP designs also aimed to perfect key science and technology aspects that are necessary to run a fusion DEMO reactor/power plant.Various configurations were reported by the fusion researchers to address the challenges of high power density operation with high bootstrap fraction as well as current-drive for long pulses, demonstration of advanced plasma scenarios, handling high power exhaust, high-field designs with HTS magnets and joints and novel maintenance strategies [1,11,[28][29][30][31][32][33].Separate facilities were proposed for testing materials and components relevant to DEMO.Component test facility designs were also conceived, where, typically R ∼ 0.8-1.5 m, so that with moderate fusion power (100 MW) a high enough neutron load (1-2 MW m −2 ) can be achieved [34][35][36].Such devices with intermediate targets seem extremely important to establish the performance reliability and act as a decisive step in making correct technology choices for the success of PP.
The Indian strategy was described 15 years ago [3,37] with an intermediate device SST-2 [38] to address various physics and nuclear aspects of the DEMO without demonstration of power extraction.Similar to the other configurations of that time, the DEMO strategy was to have large high power devices (∼3 GW) by using LTS magnets and a conventional aspect ratio of ∼3.The progress in the domestic fusion research has led to new laboratories for the development and testing of technologies that are relevant for DEMO such as RF and neutral beam heating, pellet injection and pumping, divertor, magnet, blanket, fuel cycle, remote handling, D-T neutron generator, and cryogenics while continuing to work on ITER deliverables [39][40][41][42][43][44][45][46][47][48][49].The key drivers for a revision of the strategy are cost reduction, construction-time minimization, modularity, smaller units with shared infrastructure, deeper involvement of industries both, in design and in investment and exploitation of innovations in plasma and fusion science.The new strategy demands short-time-frame outputs which are readily utilizable by not just the science community but a wide range of users in various sectors, such as energy, materials, healthcare, space, industries, etc.On a longer term, given the unique features of fusion reactors that they can produce their own fuel 'onthe-fly' and generate net electricity, the DEMO roadmap must focus on affordability, high return-on-investment and performance reliability.
We envision a scenario of baseload electricity generation sustained by a park of moderately sized fusion reactors.Each reactor unit by itself should generate net electricity and the entire plant should operate in such a way that at least one of the units is always available, while others may be under scheduled maintenance.Significant cost savings can occur as a result of shared infrastructure of land, power, cooling, cryogenics, heating systems, radwaste management etc.Construction of multiple units and facilities for manufacturing, measurement, testing and validation at site can improve industry's willingness for shared construction and eventually operation.A four-stage approach is conceived for the realization of Indian DEMO: (S1) R&D Stage, (S2) an integrated test facility that we refer here as FEST (Fusion Engineering and Science Test) facility, (S3) a fusion PP with engineering gain Q E ⩽ 1 and (S4) DEMO.The word 'DEMO' is used in the sense that it should have all the features of a fusion power plant (Q E > 3), not just a demonstration, but a proof of giving stable net electrical output to the national power grid.For net electric power, the most important factor will be the fusion-gain Q.Given the efficiency of conversion from thermal to electrical (η th ) as close to 33% and the 'wall-plug' efficiency η wp ∼ 50% (ratio of power actually coupled to the plasma and the power drawn from the grid), one must look for a reactor with Q ∼ 20.To achieve this challenging goal of sustained high performance with reliability, it is necessary develop an approach where technologies are systematically taken to their desired maturity level by targetoriented R&D program.

Rationale for the staged approach
Fusion energy production has been demonstrated in the past in the Tokamak Fusion Test Reactor [50] and Joint European Torus (JET) [51].The duration is of several seconds due to pulse-length limitations arising from copper magnets.Next generation devices like ITER are aiming to produce about ten times larger fusion energy for several hundred seconds.The next logical step would be to convert the fusion power to electricity and demonstrate baseload delivery to the grid.The staged approach is a natural consequence of the R&D gaps from the present-day status to DEMO and the time and effort it will take to fill them.Opportunities for improved designs and mid-course corrections will arise due to the constantly evolving tokamak plasma science and fusion-enabling technologies.For fusion to be a reliable and competitive source for baseload power, the plant must operate around the clock with each individual unit meeting its performance specifications.Long-pulse operation, sustained high performance, significant tritium breeding and efficient thermal to electrical conversion are the most important gaps between a research reactor and DEMO.Staged approach is an answer to systematically fill these gaps by directed research to develop enabling technologies and to validate them by extensive testing, individually as well as by an integrated test.The staged approach needs to be future ready, i.e. designed for opportunity utilization.The innovations that can potentially reduce cost and risks, reduce construction-time and operational complexity and create enhanced margins of reliability should be readily incorporated as the DEMO design progresses in time.
The high level objectives for a single reactor unit in the park are: 1. Achieve fusion power P f ∼ 1000 MW and gain Q ∼ 20 2. Generation of net electrical power in a range of 200 MW-250 MW with reliable performance 3. Design the reactor for 20 full-power years (FPY) of operational lifetime with availability between 35%-50% 4. Achieve long pulses of several hours duration, with dwelltime consistent with fuel-cycle constraints to meet the baseload power requirements.5. Demonstrate in-service inspection, replacement and repair by remote handling.6. Demonstrate adherence to nuclear quality, safety and regulatory objectives throughout the reactor life-cycle (construction, operation and de-commissioning).
In order to meet the above objectives the key science and technology requirements are: • Long-pulsed (few hours), H-mode operation scenarios using efficient non-inductive current-drive with a significant bootstrap fraction (f bs > 60%).• Maximum coverage by breeding blankets for ensuring fuel self-sufficiency and high-grade heat extraction.• Superconducting magnets with winding-pack current densities J wp ∼ 35 A mm −2 and the ability to operate in peak magnetic fields of order 25 T.
• Advanced control of plasma disruptions; control over plasma parameter profiles, heat deposition and radiation loss, control over bursty events of heat-loads on plasmafacing components.• Efficient fuel-cycle for handling exhaust over long duration, efficient re-processing, isotope separation and storage for the re-use of tritium recovered from the exhaust and the breeding blankets.• Divertor operation with steady state heat loads <10 MW m −2 .• Limiting the average neutron wall-load ⩽1.5 MW m −2 to allow the desired operational lifetime and to minimize the number of maintenance shutdowns.

A staged approach to DEMO
In order to meet the above requirements, a staged approach to DEMO is outlined.We define a set of targets consistent with the technological maturity at that particular stage enabling one to take a go/no-go decision for the next stage.This allows a step-by-step increase in investment and efforts based on the achievement of targets.The four stages we consider are shown in figure 1.The arrows that point towards preceding stages represent a flow of specifications or requirements, e.g.before the DEMO can be built, a number of open questions need to be answered at the PP stage.Similarly, the arrows that point towards the succeeding stage represent the outcome from the previous which then becomes a deciding factor in the design of the next stage and technology choices therein.The logic behind the configuration choices discussed in stages 2-4 is shown in the decision-tree in figure 2. The figure shows two bifurcation points: (1) whether HTS magnets with joints can be successfully demonstrated and (2) whether superconducting magnets can operate in presence of high-field.The spherical tokamak configuration has a center post (CP) that is surrounded by a central solenoid (CS).An example of the joints configuration with copper coils is NSTX [52] where, the CP joins the outer arms of the toroidal field (TF) coil.The access to removal and maintenance of CS coils is possible only if the joints of superconducting magnets can be made demountable.Further, from nuclear considerations, such joints will need to satisfy remote handling compatibility.The joints help in conventional tokamak designs as well, for example the ARC tokamak design [32], where the joints enable large parts to be dis-assembled for ease in maintenance.Path A leads to a spherical tokamak option.The possibility of using conductors that can tolerate high field leads to a compact configuration.The paths A1 and B1 lead to compact configurations of both spherical (A1) and large aspect ratio (B1).The path A2 is an ST with low-field operation and B2 route is a low-field large aspect ratio option.For specific design constraints a compact convention A configuration with joints is possible (for example ARC [32]) but is not covered in this decision-tree.
The configurations reported in this paper are derived by using a systems code 'SARAS' which calculates the steadystate values of reactor parameters using 0-D models with 1D profiles for density, temperature, elongation (κ) and  The possibility of high field operation leads to a compact option for both ST and conventional paths which are A1 and B1.The options for having magnet joints leads to a spherical tokamak path both A1 and A2.B1 is a compact large aspect ratio path and B2 is a conventional large aspect ratio route.triangularity (δ) [53].We have considered H-mode as the baseline scenario for the operation and used ITER H-mode scaling for the energy confinement time τ E calculation.The physics inputs to the code are major radius R, magnetic field B t and dimensionless parameters aspect ratio A, elongation κ, triangularity δ, safety factor q, normalized pressure β N , Greenwald fraction f G , bootstrap current fraction f bs , H-mode confinement time multiplier H h , ratio of α-particle confinement time to energy confinement time ρ He , fractions of 9 different types of impurity species f i Z and the profile parameters for density and temperature.The current drive efficiency γ CD is defined as . The engineering inputs are the winding pack current density of the magnets (J TF and J CS in A mm −2 ), conductor size of the magnets (w cond in mm), vacuum vessel (VV) and breeding blanket thickness.We have considered a CS-assisted start-up of the plasma and the bore radius is calculated from the flux requirement for the Ohmic-only/Ohmic-dominated scenarios.The code calculates the machine parameters from the input data.The fusion power is calculated using Bosch-Halle [54] cross-sections self-consistently over the entire plasma volume by taking into account the effects of fuel dilution and the impurity density profiles of a variety of impurity species.The line radiation is calculated by assuming a corona model using the average-ionmodel [55].The power balance requires that the sum of αheating power and the total auxiliary heating power (P α + P aux ) is balanced by the sum of transport power P L across the separatrix and the radiated power from the core P c rad .The P aux will, in general, consist of the power required for current drive P cd and additionally, the power required for heating the plasma P addh to satisfy the power balance.The power balance requires that the sum of α-heating power and the total auxiliary heating power (P α + P aux ) is balanced by the sum of transport power P L across the separatrix and the radiated power from the core P c rad .The P aux will, in general, consist of the power required for current drive P cd and the additionally the power required for heating the plasma P addh to satisfy the power balance.The former depends on the bootstrap fraction.The parameters for the design choices are arrived by scanning this multi-parameter space using constraints.The model equations and assumptions are summarized in appendix.The scans reported in this work are carried out by assuming a γ CD between 0.2-0.35MA•m −2 •MW −1 .This is taken by considering NBI as the source of current drive.However the actual heating and current drive mixture could be different and that needs to be worked out in detail after selecting the configuration.For the electric power estimation we have assumed a wall-plug efficiency (η wp ) of 50% and power conversion efficiency (η th ) of 33%.We have assumed a double-null divertor configuration as the baseline for all the options considered in this paper.The choice of κ and δ and internal inductance l i in this paper are indicative.The maximum value of κ is taken from [56].The optimized values will depend on the PF coil positions and the detailed considerations of vertical stability.

Physics considerations for the staged approach
One of the important observation from the parameter-space scans is that increasing the confinement does not necessarily improve the fusion gain in a reactor.The variation of P f and Q as a function of the confinement factor (H h ) is shown in figure 3 for a representative case of R = 3 m, A = 1.9 and ρ He = 5.The curve of Q is flat at higher H h because of the constant auxiliary power.The fusion power does not change with H h if there is no helium content in the plasma.At lower H h , the ITER confinement time scaling results in a higher τ E and a Figure 3.The variation of fusion power P f , P L (a) and gain Q (b) as a function of confinement factor with (red) and without helium dilution (blue) for a representative case of R = 3 m, A = 1.9, κ = 2.2, δ = 0.3, Bt = 3.6, q = 6.5, β N = 2.3, ρ He = 5. higher loss power.The power balance consideration forces to have a higher heating power to compensate the loss power at lower H h .After a certain H h value the heating power is decided by the current drive power which is a constant.Thus, P aux is a constant and so is Q.The reduction in the Q at high H h is due to the increased helium dilution that is taken in the model.The helium confinement time is taken as a constant factor multiplying (typically taken as 5, unless explicitly stated otherwise) τ E .The τ E increases with H h and this results in a higher fuel dilution and a consequent reduction in the fusion power which reduces the Q.We find that it is difficult to obtain a higher Q at lower P f due to the requirement of a minimum auxiliary power to reach the desired P f .It appears that moderate confinement regimes are interesting for a reactor.
The effect of ρ He in fusion performance is shown in figure 4 for a constant H h of 1.2.It is evident that as ρ He increases, the helium fraction increases ( f He , figure 4(d)) and this results in an increase in the auxiliary power to maintain the same plasma β.Similarly the reduction in the fuel ion density reduces the fusion power and the combined effect of both reduces the fusion gain.
The variation of plasma performance as a function of normalized-β is shown in figure 5 for H h = 1.2 and ρ He = 5.As the β N increases the fusion power increases, the fusion power and the bootstrap fraction increases that reduces the current drive power (figure 5(c)).However, the heating power required to achieve the β increases which, effectively reduces the Q.The limiting β N calculated from [56] for this parameter set was 6.4 and the scans are done well below the β max N .We consider ITER H-mode confinement-time scaling for the baseline, where the confinement time (τ E ) depends on the transport power flowing across the separatrix, τ E ∝ P −0.69 L .Since the stored energy W ∝ β and P L = W/τ E , we have P L ∝ β 3 .In general the P L dependence on β remains strong no matter which scaling law is adopted [12].For ITER-confinement time scaling, this implies high-β configurations will have high P L and consequently high auxiliary heating power from the power balance considerations, thereby reducing Q.It is therefore crucial to explore access to advanced confinement regimes where the confinement time degradation is lesser than that predicted by the ITER H-mode scaling.In our parameter scans, we have chosen q and β N as input parameters, so both β and β p decrease as β N is decreased.For low values of β p the fraction of bootstrap current f bs is small, which results in greater power requirements for current-drive.In short, while high β N operation can reduce Q due to heating requirements, low β N operation tends to reduce Q due to increased currentdrive requirements.So an optimum β N needs to be found.
The constraints arising from the optimum β, H h and ρ He show a moderate β, moderate confinement regimes are suitable for reactor configurations.Jakobs et al have reported that in a burn equilibrium with helium, the high confinement regimes perhaps are not permissible either due to the β-limit or from the limitations in achieving Q [57].Our calculations show that the Q starts degrading from its peak value much before the β-limit is reached for a configuration due to helium dilution.
The non-inductive current drive is the most important consideration for long-pulse reactors.Currently, the experiments with LHCD, ECRH and NBI have generated a certain database, showing efficiency γ CD up to 0.35 MA•m −2 •MW −1 [58][59][60][61].However, for reactor level plasmas, the LHCD may not be suitable due to high densities and temperature and may drive current only in the outer region.The coreregion current drive remains an open issue.An efficiency of 0.35-0.5 MA•m −2 •MW −1 seems adequate since Q remains unaltered at higher γ CD which is shown in figure 6 for a representative case of R = 5 m, A = 1.9.The reduction in Q at higher γ CD is due to the increase in the auxiliary heating power as opposed to the current-drive.Although NBI can create such high γ CD , the core current drive is an issue due to the geometric constraint arising from the requirement of a minimum tangency radius.This forces one to go for higher aspect ratio which reduces the f bs and hence enhances the requirement of current-drive power.Thus the ways to improve γ CD with wave-heating methods remain a challenge.Besides the γ CD , the overall efficiency of the current drive, which is the product of wall-plug efficiency η wp and γ CD is also an important factor in deciding the net electricity production (Q eng ).For ITER-NBI the η wp is about 0.27 [62] and for DEMO this product (γ CD η wp ) needs to be around 0.17-0.20 MA•m −2 •MW −1 to minimize the re-circulating power and to achieve a significant Q eng .This requires ways to improve the η wp .
The α-heated burning plasma regime is one of the key aspect that cannot be studied in the present-day devices.It can partly be studied in JET but a full manifestation will be possible only in ITER.The generation of α needs to be balanced by the removal of helium ash to control the fuel dilution and consequently fusion power and the quality of the H-mode.Since the helium ion transport in burning plasmas is not fully understood, one of the key assumptions is taking the helium particle confinement time as a constant factor multiplied by the energy confinement time.This directly impacts the estimate of helium ash accumulation in the core, fuel dilution, fusion power and such important parameters while choosing the design choices.

Engineering considerations for the staged approach
The rise of HTS magnets is one of the key technology drivers for the staged approach.The feasibility of demountable magnets with remote handling compatibility could play the role of disruptive technology when it comes to fusion.For a compact reactor, both high field magnets as well as high current density of the winding pack (J wp ) can influence the machine size.The non-scalable neutron mean-free-path makes arbitrary size reduction in the inboard side impossible.The only way to compensate it is to have magnets with higher J wp .In this article we use the term J wp loosely, i.e. it is not reserved for the TF magnets alone, it is representative of conductor, its insulation, its support structure, etc in making of the entire magnet crosssection.So it is used interchangeably with the corresponding current density in the CS coil as well.
The constraints arising on the size of the machine from the magnet current density and the maximum achievable magnetic field for a reactor of 200 MW fusion power is shown in figure 7. The product of major radius, R and the field at R is kept constant in the scan.The contours of the total magnetic field (dotted lines) experienced by the magnet and the minimum inboard thickness, R − a (solid lines) for different J wp and conductor size (w cond ) are shown.The minimum inboard thickness is estimated for a shielding thickness for 1 FPY operation by assuming a neutron decadal length of 13 cm along with the VV (no breeding is assumed).This is the average value taken over the entire blanket region for purpose of [11] and seems reasonable for materials like tungsten-carbide.The intersection of the contours is marked by the open circles.For a limiting magnetic field of 28 T, the minimum R − a is 1200 mm.As the conductor size increases for the same B, the R − a increases and the required current density reduces.Thus, to increase the compactness of the machine higher J wp and smaller w cond are required.Thus the achievable engineering current density is of crucial importance in high-field operation, since it is going to decide the trade-off margin for inboard shielding and therefore the operational life of the TF/CS [53].
At the magnet, the conservative limit of the acceptable neutron dose is taken to be 10 22 n m −2 s −1 .The equivalent dose to insulation can be taken as 5 × 10 6 Gy.This equivalence will vary a bit with the material and the neutron spectrum.This limits the requirement of shield blanket thickness in the inboard size and consequently the minimum machine size.Although a higher B field allows a smaller configuration, the neutron mean-free-path makes it impossible to have very small reactors.In our calculations we have considered shielding materials such as tungsten carbide and the shielding materials developed by India for ITER [63].The advanced materials with better shielding can reduce the machine size further.
Another important consideration is the divertor heat handling.We consider a double null configuration for the baseline.Let us define a parameter P db = 5P SOL /(4π Ra) where P SOL is the transport power across the seperatrix.We have taken an ad-hoc total poloidal wetting length as a/5, including both, upper and lower divertor zones.Since the advanced divertor configurations can have a larger wetted area, the P db is the higher limit of the heat flux.The target is to design configurations that limit the P db below 10 MW m −2 such that the actual heat load will be further low.For the heat removal, options of water-cooled, helium-cooled and supercritical carbon dioxide (sCO 2 ) divertor need be explored.In the reactor concept, a double null divertor lead to a reduction in the total blanket area and optimization of tritium breeding with double null divertor is a consideration to be explored.
Constraints arising from the materials in the neutron environment pose significant challenges in the availability of the reactor as well as the efficiency of heat extraction.The structural materials such as reduced activation (RAFM) steels are tested in fission neutron facilities and ion-irradiation experiments up to 20 dpa [64][65][66][67].Development and qualification of materials that are capable of withstanding displacement damage up to 50 dpa is required for minimizing the downtime due to the maintenance breaks of the reactor.This poses constraints on the average neutron-wall load on the first-wall.In our calculations we limit the average neutron wall-load to 1 MW m −2 .A higher neutron tolerant material will minimize the downtime of the machine.Similarly a higher temperature window compared to RAFM (maximum allowed temperature is 550 • C [64]) will allow efficient heat extraction from the blanket.
One of the important aspects of the compact reactor is the choice of the maintenance scheme that decides the machine availability.The large variation of the curvature across the radial distance makes ITER-like maintenance schemes difficult [68].This requires innovative maintenance schemes where either the entire sector of the blanket is removed through the vertical port or the modular poloidal sectors are removed through the different ports.The impacts the design choice of the blanket.Besides, it is also required to have either additional space between the blanket and the VV or take the TF coils far from the plasma [69] which will take the PF coils further away from the plasma and can influence plasma control.
We have taken into account these considerations in choosing the numerical examples of configurations at each stage.
Some of the technology challenges can be addressed in Stage-1 where the focus is on the directed R&D.The challenges related to (1) system integration, (2) long-pulse, high performance operation at I p = 5 MA with non-inductive current drive and (3) operation and control of plasma with P f ∼ 25 MW including physics of α and plasma-wall interactions could be addressed in Stage-2.The description of different stages is discussed in the next section.

Stage-1: directed R&D
The Stage-1 is the target-oriented technology R&D.The thrust areas are:

Stage-2: fusion engineering science and test (FEST) facility
The integrated test facility called here as FEST, is the most important turning point on the DEMO roadmap.Its mission is to confirm integrated performance of fusion enabling technologies and demonstrate low-power levels of nuclear fusion, but for long duration pulses.The performance of the key technologies that have been developed in stage-1 will decide their qualification for being deployed in the PP.The FEST may initially derive its targets from the PP design, but only after actual experiments will one know the best that could be achieved and correspondingly feedback on the PP design.The FEST brings realism to PP, hence the mission of FEST are: • To validate the magnet technology choice (i.e.performance of magnets with joints, the limiting magnetic field for their performance, maximum current-density that could be accomplished, etc) • To demonstrate ∼30 MW of fusion power in noninductively driven D-T plasmas with f bs ∼ 50% • To test blanket modules of selected breeder/coolant combinations so that the blanket design for the PP can be selected.
Ideas to increase the efficiency γ cd and η wp aiming for steadystate plasma currents up to 5-7 MA with varying degree of bootstrap fraction will lead to a sound basis for the next step.Confinement optimization in presence of helium, stability of high bootstrap current equilibria at moderate to high β and role of energetic particles will open a new front.A number of activities can be interesting in this stage such as experiments to simulate α-heating and kinetic instabilities, trace helium experiments to simulate fuel-dilution and impurity confinement scaling.Predictive performance simulations along with advanced diagnostics are to be used in this stage.
In the technology front it opens up new opportunities in terms of pulsed D-T operation for testing modules of different breeding blanket concepts, ELM control and vertical stability with control coils positioned so as to mimic reactor scenario, tests on disruption control etc.In the blanket testing front it can be used as a multi-concept testing facility where test blanket modules of multiple concepts such as solid breeder with Be multiplier, solid breeder with Pb-Li hybrid concept, a pure Pb-Li based concept, advanced molten salt concepts etc can be tested and qualified for the PP.The measurement of tritium production rate and the demonstration of its extraction will form the selection criteria for the concept apart from its thermal performance.It will also act as a test bed for fuel cycle concepts that will be used in pilot.Post-irradiation experiments on the magnet material samples, insulation, plasmafacing and armor materials will open an exciting frontier and give inputs for the next stage.The FEST will be crucial in deciding the maintenance scheme for different blanket concepts, its compatibility with remote handling, especially the magnet joints.The FEST will be decisive in whether the magnet joints are RH-maintainable to use in the PPs and whether an ST route to DEMO can be adopted or not.In this context FEST device will be unique; it will confirm the design and technology choices adopted and it will work as a facility do exciting discovery science.
A tokamak R = 2.25 m is a good representative of the device options that may be developed for the FEST device.While choosing this configuration, our main focus was to keep the plasma current ∼5-7 MA and fusion power ranging from 25-75 MW.To achieve a fusion gain Q ∼ 0.5 is quite challenging given the existing current-drive efficiencies.Two configurations of interest are shown in table 1.For naming a specific configuration we use a convention of writing letter 'R' for reactor, followed by its major radius in centimeters.The option ST-R225 is an ST configuration and R225 is a conventional tokamak.If the indicated J wp values can be achieved and the HTS magnet joints are successful, then ST-R225 can be chosen.We would like to point out here that ST options require comparatively higher β values to attain the same fusion power as compared to conventional options.As a result, while the ST may have higher f bs and consequently lower current drive power requirement, they will necessarily have larger transport losses.It will then depend on the fine details of other parameters to obtain a reasonable Q.Whichever option is finally chosen, the scientific program of FEST needs to be driven by the primary requirements from the PP.To that extent, there has to be a feedback between stage-2 and stage-3, where PP is able to adjust its design based on the actual outcome from stage-2.Nevertheless, the FEST must extensively test the science and technology that will become the foundation for the PP.As an example, long-pulse high performance operation with modest fusion power must be demonstrated.

Stage-3: PP
The PP is the most important step towards DEMO.The purpose of the PP is to generate a complete experience of successful construction, commissioning and operation with electricity production with engineering gain, Q eng < 1.The Q eng is defined here as the ratio of gross electrical power produced and the electrical power drawn from the grid for auxiliary heating and non-inductive current-drive.The design of PP would be based on the FEST outcome.The up-scaling of parameters from FEST to PP would be in plasma current and fusion power.The PP will need to be equipped with blanket modules for tritium-breeding and high-grade heat extraction.Almost all the inner surface area, excepting the mandatory ports will need to be covered with the blanket modules.The space constraints on the inboard-side may not allow tritium-breeding zone, hence it would have only shielding modules.This would mean that the PP will need tritium fuel supply from another source to make up for the T-breeding shortfall.
A critical element of the PP is the maintenance scheme.Installation/removal of the blanket and shielding modules define the spatial constraints for remote handling.This and the other maintenance considerations play a crucial role in the sizing of the VV and correspondingly the cross-section and length of the equatorial port duct.This in-turn impacts the NB tangency radius, which will make it necessary to optimize the shielding geometry and trade-off the dose-rate in the dedicated NB cell.This will need to be finalized for repair and replacement of modules that are not directly accessible from the VV ports.There could be a way to remove the entire blanket sector by lifting it through the vertical port.This has immediate implications that the blanket must be physically away from the VV with adequate clearance.Another option could be to attach the blanket modules to the VV inner wall, in which case each module must be installed and maintained from its front-side.This would be a scheme similar to that of ITER.We find that for the PP having R ∼ 3 m, it is quite challenging in terms of space clearance to install/remove modules from the front-side because of the relatively sharp plasma surface curvature in compact configurations.If one can make HTS magnets with joints, a significant advantage is foreseen in maintenance schemes where accessible area can be enhanced.
The aim of the PP is to demonstrate electricity generation.So as the PP operation gradually transits into long pulse fusion power operation the stage would get set for power conversion.
Here, the Rankine cycle, most commonly used in fission reactors could be adapted.The pulsed nature of the fusion power needs to be evened out by a buffer energy storage system, for example, of the kind used in solar power arrays.Such an intermediate stage isolates the dynamic nature of the fusion power pulse from the steady state steam generation that is required for the power plant [70,71].We note here that the length of the pulse will ultimately get decided by the exhaustreprocessing speeds that can be achieved and the site-storage limits on tritium.If one considers 20 pulses per day and about 10 days of continuous power conversion demonstration, the tritium volume reprocessed would be about 10 kg.For a twoday re-processing cycle, this implies the exhaust processing volume of about 1.5 kg of tritium.
Therefore the PP specifications should be such that it demonstrates the following: (i) Stable high-confinement operation with a significant bootstrap fraction (ii) Moderate fusion power and gain (P f ∼ 200-300 MW and Q ∼ 3-5) (iii) α-heating (iv) Efficient non-inductive current-drive and auxiliary heating (v) High heat-flux handling with novel radiative divertor configurations (vi) Breeding blankets on the outboard side (vii) Shielding blankets on the inboard side (viii) Fuel cycle with the use of tritium bred in the blankets (ix) Remote handling of the in-vessel components (x) Nuclear and engineering safety (xi) Radwaste management using RH and hotcell The PP design options will be improvised based on the outcome of the Stage-2.For PP, we consider configurations of major radius about 3-3.5 m.We have given 3 possible configurations for about 1 FPY operation in this range and one configuration with a larger major radius.
The PP options are: (1) ST configuration with R = 3 m (ST-R300) with demountable HTS magnets and a J wp of 45 MA m −2 , (2) another ST configuration with R = 3.5 m with a lower J wp and higher fusion power and Q (ST-R350), (3) a conventional aspect ratio compact device with R = 4.4 m (R440) using only LTS magnets, and (4) a conventional aspect ratio configuration with R = 3 m (R300) with LTS/HTS magnets.The possible parameter choices for these configurations are shown in table 2. These configurations are only indicative and the actual values will depend on the detailed optimization of the TF coil and the shield blanket design.Depending on the choice, the operational life of the TF/CS magnet varies from 0.6 years to about 15 years of full-poweroperation (assuming a limiting dose of 10 22 n m −2 ).This implies that the integrated D-T fusion exposure time will have to be within the stated limits.At this stage, the sustained exposure of materials to neutron wall load >1 MW m −2 , high heat exhaust with radiative divertor, control of impurity radiation from the core and attaining Q ∼ 4-5 can be the targets.
Although the demonstration of tritium self-sufficiency is not the mandate of PP, it is expected to demonstrate an overall tritium breeding ratio of about 0.8 and exceeding 1.1 when calculated only over the breeding zone.The PP is expect to demonstrate the use of the bred tritium as a part of the closed fuel cycle.
The PP will also generate experience in radwaste management.A hotcell will be needed to house the components with a facility to isolate and remotely examine the components or their coupons.Quantification of radwaste and its storage will allow a sound basis for projecting these estimates for DEMO.

Stage-4: DEMO
The DEMO is a fusion power plant, envisioned to be delivering net electricity to the grid.In terms of its size, power and operational characteristics, the DEMO becomes a working proof that all the systems perform reliably, thereby becoming a basis for replication into multiple units on a commercial scale.After extensive parameter scans, we conclude that given the tokamak physics and other constraints a 250 MWe net-electric fusion power plant will need to produce about 1250 MW of fusion power.
For DEMO we have considered a range of major radii from 4.5 m to 8 m range for different aspect ratios.We have identified three representative configurations for further elucidation of the concept.Three DEMO options, ST-R480, ST-R540 and R770 are shown in table 3. The ST-R480 and ST-R540 are ST options with HTS magnets and joints.The option ST-R540 is the baseline for ST-DEMO with a winding pack current density of 29 MA m −2 and the peak field at the center-post (sum of both CS and CP fields) was limited to 23 T. A higher J wp of 36 MA m −2 allows a reduction of the major radius to 4.8 m with an increase in the neutron wall load to 1.2 MW m −2 compared to 0.88 MW m −2 for the ST-R540.If the major radius is further lowered, the neutron wall-load increases beyond 1.5 MW m −2 and the shielding space becomes too thin to provide for an operational lifetime of 20 FPY.For DEMO configurations it is found that, there arises an upper limit on J wp from the type of the superconductor and the highest field it can be subjected to without significantly affecting its critical current density and the temperature margin.For our case here, we have taken a value 30 T as the limiting total field.The component of the magnetic field perpendicular to the direction of the current may be smaller.However, such detail is not taken into account in this work.We find that increasing the J wp beyond 40 MA m −2 increases the peak field at the center-post beyond 30 T, which rules out designs where this limit is breached.
The option R770 is a conventional tokamak with A = 3 and LTS magnets.This configuration is essentially the same as that reported in [3], but the parameters have been tuned along with accurate fusion power calculations.Here the P f has been limited to 1500 MW to keep neutron wall load ∼1 MW m −2 .The choice of the DEMO option will be decided by the results from the PP.The choices for blanket and power conversion will be the same as in the PP upon its successful demonstration.Practical experience on PP will lead to a better availability of the DEMO.If we consider the reactor park with 3 identical units, the 35% availability gives 250 MWe around the year.But as one increases the availability with experience and innovation the number of days in which 2 units can be operated in overlapping mode can be increased.The number of days in 1 year in which 500 MWe is produced as a function of availability is shown in figure 8.If the availability is increased to 49% the number of days of generating 500 MW of net electric power can be up to 180 days in an year.The availability will be a major parameter that will decide the return on investment and electricity costs.The major role of PP will therefore be to be a tuning ground for continual improvement in availability.
A birds-eye-view of the configurations for FEST, PP and DEMO presented in this work is shown in figure 9.The aspect ratio of each configuration is shown in parenthesis.An increase in J wp leads to a size reduction independent of aspect ratio for a similar fusion power.A similar view of the physics parameters of these configurations is shown in figure 10.
A high level view of the tentative time line for the four stages can be given as follows: From the start of the programme at T 0 the Stage-1 may need about 5 years for magnet joints.The activities such as current drive, confinement improvisation, disruption control, tritium technologies, etc will form a continuum background to support all the remaining three stages.The Stage-2 (FEST) will involve extensive design work of ST and non-ST options.This will need about T 0 + 15 years to the start of operation.For generating decisive input for PPs at least 5-7 years of operation of FEST will be needed.The Stage-3 (PP) will need T 0 + 20 for the start of construction and T 0 + 30 for the start of operation.The Stage-4 will be decided from the outcome of the Stage-3.

Conclusions
A four-stage approach starting from R&D to the realization of Indian DEMO is presented:(1) directed R&D for the exploitation of novel designs, emerging technologies and their component level qualification, (2) an integrated test facility called FEST for the performance demonstration of the systems made from the qualified components, (3) a PP to demonstrate the gross electric power generation using the selected technologies,to establish performance reliability and to eliminate technical risks, and (4) DEMO to produce net electricity of about 250 MWe by up-scaling the technologies used in the PP.
The revised strategy envisions a reactor-park of multiple, moderately powered, compact DEMO reactors with 35%-50% availability for each reactor unit so that an uninterrupted baseload power to the grid is ensured.
In terms of the R&D directions we find that the pursuit of high β plasmas does not necessarily results in a high fusion gain.Similarly the path of increasing energy confinement time becomes counter-productive due to the the increased helium ash content.For ST-based DEMO, the improvement caused by higher J wp beyond 40 MA m −2 is rendered ineffective since the field the conductor crosses the limiting value.
So far, the R&D experiments in fusion had limited the industry contribution to only component development and manufacturing.The path to achieve commercial power plant will require industrial participation in the entire development cycle including design.To build a commercial fusion power plant in partnership mode with industries, it is necessary that a PP be built where mitigation of technical risks and cost basis is deeply matured.The early involvement of the industries in the FEST will serve the purpose of evoking their interest and training in niche areas.FEST can also become an international collaboration platform for discovery science.

Figure 1 .
Figure 1.Schematic of the staged approach consisting four stages.The arrows indicate top-down flow of requirements and bottom-up flow of outcomes.

Figure 2 .
Figure 2. The decision-tree for choosing various configurations.The possibility of high field operation leads to a compact option for both ST and conventional paths which are A1 and B1.The options for having magnet joints leads to a spherical tokamak path both A1 and A2.B1 is a compact large aspect ratio path and B2 is a conventional large aspect ratio route.

Figure 6 .
Figure 6.The variation of Q with γ CD for 1250 MW fusion power with R = 5, A = 1.9.

Figure 7 .
Figure 7. Contours of total magnetic field (dotted lines) and the minimum inboard radius .

Figure 8 .
Figure 8. No. of 500 MWe days in a year as a function of availability for a park with 3 identical reactor units.

Figure 9 .
Figure 9. birds-eye-view of the configurations for FEST, PP and DEMO in Jwp − R-plane.The aspect ratio of each configuration is shown in parenthesis.

Figure 10 .
Figure 10.Physics parameters of the configurations along with ITER inductive scenario baseline from [72].The bootstrap fraction is shown for ITER non-inductive scenario.The ρ * and ν * are the normalized Larmor radius and collisionality.

Table 1 .
Parameters for FEST options, option ST-R225 is ST with center post and demountable joints, R225 is a conventional aspect ratio one with LTS/HTS magnets with no joints.The β N only takes into account the thermal β, fast particle β is excluded in the confinement calculations.

Table 2 .
Parameters for pilot plant options, ST-R300 and ST-R350 are two ST options with centerpost and demountable joints.The R440 and R300 are the conventional aspect ratio configurations with LTS/HTS magnets.The β N only takes into account the thermal β.

Table 3 .
Parameters for three options developed for DEMO.Option ST-R480 and ST-R540 are spherical tokamaks, option R770 is conventional tokamak.The β N only takes into account the thermal β.