Insights into fuel start‐up and self‐sufficiency for fusion energy: The case of CFETR

Commercial tritium resources available are too scarce to fully supply the future fusion reactors after International Thermonuclear Experimental Reactor (ITER). Tritium self‐sufficiency, ITER fails to fully validate, was regarded as one of the most important issues needed to be solved in the pathway of achieving fusion energy. After ITER, several concepts of fusion engineering test reactors and fusion demonstration reactors have been proposed worldwide, for example, Chinese Fusion Engineering Test Reactor (CFETR), Fusion Nuclear Science Facility (FNSF), DEMOnstration fusion reactor (DEMO) in European Union and Korea. CFETR is in the engineering design phase and would be hopefully completed around 2020. Tritium resources for the reactor start‐up and tritium self‐sufficiency are two primary issues besides the steady‐state operation for CFETR. The objectives of this work are as follows: (a) to introduce the preliminary fuel cycle concept and available tritium resources for CFETR, (b) to evaluate and discuss the tritium demand for CFETR start‐up (phase I: 200 MW) and the feasibility of DD start‐up, (c) to identify the possible pathways to tritium self‐sufficiency through sensitivity analysis based on the design baseline of CFETR, (d) to evaluate the consequences in case of failing tritium self‐sufficiency, and (e) to identify future R&D needed for tritium self‐sufficiency. It is expected to give insights into the question on how to start the reactor in a more economical way, into the feasibility of tritium self‐sufficiency, and into the question on what will happen in case of failing tritium self‐sufficiency.

for a clean and safe solution for our future long-term energy needs. 5 More and more human and financial resources are being put into the fusion research. 6 As the largest international energy cooperation project including seven members, the International Thermonuclear Experimental Reactor (ITER) located at Cadarache in France is being constructed. It is expected to achieve the first plasma in December 2025 and deuterium-tritium (DT) operation in 2035. 7 Meanwhile, the next fusion reactors were also proposed and supported by many national and local governments based on the roadmap of fusion energy development. Two kinds of strategies were identified. The first is to build a smaller sized fusion experimental facility before building a fusion power plant. China proposed a concept of Chinese Fusion Engineering Test Reactor (CFETR), 8,9 the USA proposed a concept of Fusion Nuclear Science Facility (FNSF). 10 The others are to build a DEMOnstration fusion reactor (DEMO) with over 1500 MW fusion power to produce electricity directly. EU-DEMO, K-DEMO were proposed by Europe and Korea, respectively. 11,12 Deuterium and tritium have been regarded as the fuels for the above-mentioned fusion reactors because DT fusion reaction requires the lowest fusion triple product to achieve energy self-sustaining among various fuels fusion reaction. For a fusion reactor with 1 GW fusion power, about 37.1 kg deuterium and 55.6 kg tritium would be burnt per full power operational year. 13 Deuterium resource is abundant while tritium resource is scarce. For tritium, the only significant national sources are cosmic rays passing through the Earth's atmosphere and possibly accretion from the solar wind. An equilibrium quantity of ~3.5 kg is present from these sources. 14 The commercial tritium consumption is mainly from the artificial generation. Tritium price of Heavy Water Reactor (HWR) production is relative low compared with other tritium production pathways of accelerator and light water reactors. 15 Currently, most of the commercial tritium are from CANadian Deuterium Uranium reactors (CANDU) of Canada and Korea where Tritium Removal Facilities (TRF) are being operated. India has also positive plans to build CANDU reactors and may be the promising country with a high tritium generation rate if operating the TRF in the future. Both our early study and recent studies evaluated the worldwide tritium resources available for fusion reactors and showed the worldwide tritium generated from CANDU reactors may be insufficient for start-up, let alone as burning fuels for next fusion reactors. [15][16][17] Thus, start-up and tritium self-sufficiency is critical important for an economical fusion energy.
ITER is only designed to test tritium breeding using three ports to install six tritium breeding blanket modules. 18,19 Tritium self-sufficiency cannot be fully validated just through the construction of ITER device. Thus, tritium selfsufficiency is one of the key issues remained to be solved in the next fusion reactors. Both of the fusion engineering test reactors and fusion demonstration reactors are being designed to achieve the target of tritium self-sufficiency. 8,[10][11][12] In China, the engineering design phase of CFETR has begun and would be hopefully completed around 2020, which will support the proposal for construction phase according to the recent roadmap. The reactor start-up and tritium self-sufficiency issues should be discussed for CFETR during its engineering design phase considering the uncertainties of tritium resources available and existing risk of failing tritium self-sufficiency. 9,20,21 The objectives of this work are as follows: (a) to introduce the preliminary fuel cycle concept and available tritium resources for CFETR, (b) to evaluate and discuss the tritium demand for CFETR start-up (phase I: 200 MW) and the feasibility of DD start-up, (c) to identify the possible pathways to tritium self-sufficiency through sensitivity analysis based on the design baseline of CFETR, (d) to evaluate the consequences in case of failing tritium self-sufficiency, and (e) to identify the future R&D needed for tritium self-sufficiency. It is expected to give insights into the question on how to start the reactor in a more economical way, into the feasibility of tritium selfsufficiency, and what will happen in case of failing tritium self-sufficiency. The structure of the article is as follows. Section 2 introduces the preliminary fuel cycle concept and model for CFETR. Section 3 describes the fuel start-up for CFETR to answer how to start CFETR more economically. Section 4 discusses the tritium self-sufficiency analysis and future R&D needed to answer what about the feasibility of tritium self-sufficiency, what will happen in case of failing tritium self-sufficiency, and how to do in the future. Finally, conclusions are given in Section 5.

| Recently funding activities
To close the gaps to CFETR requirement, Chinese government deployed many new R&D projects under the frame of national key research and development plan since 2017 to start the engineering design of CFETR. A groups of projects have been approved with a total funding up to CN¥1 153 160 000 (about US$167 367 199) in 2017. The contents of the projects are development of the proto type N-NBI system, developing the long pulse high-power gyrotron, the integration design and technical study of tritium system for CFETR, study on advanced SSO scenarios with high fraction of bootstrap, integration engineering design of CFETR and research, fabrication and key technical study for CFETR solid test blanket, and so forth. In 2018, Another CN¥340 190 000 are being put into the CFETR research.
Among the above projects, three projects are related with tritium plant with a total funding up to about CN¥253 400 000. The contents are overall design technology of CFETR tritium | 459 NIE Et al.
plant system, tritium recovery, and recycling technology for plasma exhaust gas of CFETR, tritium extraction, and measurement engineering technology for breeding blanket of CFETR. 22

| Preliminary fuel cycle concept
The fuel cycle concept of CFETR is mainly proposed to be based on R&D experiences of ITER fuel cycle. [22][23][24] For ITER, the tritium fueling into the vacuum vessel is burnt partly and most of the tritium will be pumped out accompanied by deuterium, helium, and other impurities. The cyclic utilization of tritium is necessary and the tritium will be extracted from exhausted gas in a tritium plant. To ensure the achievable of target fusion power, equal percent fuels (50:50 mix of DT) were adopted in all fueling systems of ITER. 25 The equal percent fuels required the Isotope Separation System (ISS) essential in the fuel cycle path. Then the cyclic tritium can be re-injected into the vacuum vessel by the fueling system.
For CFETR, a modified fuel cycle concept was proposed to increase fuel cycle efficiency and further decrease tritium inventory in the tritium building. 22 For the inner fuel cycle, the major difference is that the ISS would not be the main path in the fuel cycle. The helium and impurities of fuels pumped out from vacuum vessel would be removed through the Tokamak Exhaust Processing system (TEP), then the mixed fuels would be re-injected into the vacuum vessel directly through the tritium Storage and Delivery System (SDS). The small remained tritium in the exhaust of TEP would be separated using ISS to control tritium emission from safety consideration. In addition to the above, some design optimization was proposed to realize the quick removal, such as the gas chromatography technology. 22 For the outer fuel cycle, a stable Tritium Extraction System (TES) would be used for tritium extraction from the blanket. The extracted tritium should also be processed before fueling into the reactor. A separate ISS was designed to ease the burden of the inner cycling loop because the features of hydrogen isotope separation between the inner and outer fuel cycle are very different, such as about equal percent of DT in exhaust of inner fuel cycle while nearly no D in the outer fuel cycle. 22 Both the inner and the outer fuel cycle of CFETR are shown as simple block diagram in Figure 1.

| Mixed fuel burning and cycle model
For CFETR, a mixed fuel burning including DD, DT and D 3 He would appear in the plasma if failing the 50:50 DT mixed fuels. In this case, a steady-state fuel burning with a certain fusion power could still be achieved at least, and the evaluation results are shown in the following part. To evaluate the fuel cycle and tritium self-sufficiency, a mass flow model of fuels for CFETR was developed using the system dynamics platform named STELLA. [26][27][28] The advantage of the model was visual fuels flow compared with early model using programming language. The key formulas in the model are shown as follows.
Based on principle of mass conservation and the mean residence time method, the general system dynamics model for each subsystem in fusion fuel cycle is shown in formula (1). where I i is tritium inventory in i subsystem, T i is the residence time of i subsystem, is the decay constant, P i (t) is permeation rate, f is tritium burn-up fraction in plasma.
The formula (1) is used to calculate the tritium inventory in the plasma (vacuum vessel) shown in Figure 1 considering the factors of total throughput, exhaust, decay, permeation into the PFM and burning. For other subsystems in tritium plant such as TEP, ISS, the burning factor is set as zero.
The fuels (T, 3 He, D) burn-up fraction in plasma f is described in formulas (2) (3) and (4). 29 (1) cycle and tritium balance for CFETR where n is ion density, is ion confinement time, is the reaction rate. It should be noted that in formulas (2) (3) and (4), the f T means the ratio of the amount of tritium consumed in fusion reactions every second to the total tritium we lose every second in the burning plasma (tritium ions escaping from the last closed flux surface of the tokamak and never returning again). However, the tritium burn-up fraction of ITER was defined as the ratio of tritium amount of tritium consumed in fusion reactions every second to the total tritium throughput (fueling) amount every second (the total throughput including five factors: replenishing the burnt tritium, replenishing the particle transport loss, replenishing the ELM-caused particle loss, replenishing the loss because of triggering the ELM frequency for reducing the ELM energy loss via low-field side pellet injection, and replenishing the loss because of controlling the peak power load on the divertor plates by auxiliary neutral gas fueling flux). This means the tritium burn-up fraction definition of ITER has considered the fueling efficiency factor (f T × tritium fueling efficiency). In the following part, the tritium burn-up fraction definition of ITER was used to perform the analysis.
Based on the basic modeling and definition of tritium burn-up fraction, the tritium start-up amount and tritium self-sufficiency definition are described in formulas (5) and (6), respectively. Detailed descriptions were shown in Section 4.
where m start−up is tritium start-up amount (not including the tritium retention in materials in the initial operational phase), t cycle is tritium cycle time, m burn−rate is tritium burn rate, m retention is tritium retention amount in the materials of whole fusion system, R decay is tritium decay rate, m release is tritium release amount into the environment, m disposal is tritium disposal amount along with waste, t burn−time is tritium burn time of fusion reactor in a given period.

| Tritium start-up demand
The tritium start-up amount is mainly influenced by the tritium burn-up fraction and the cycle time for a fusion reactor with a certain fusion power. Regarding the CFETR phase I, the fusion power is 200 MW which corresponds a tritium burning rate as 3.5 × 10 −4 g/s. Tritium burn-up fraction is the ratio of tritium burning rate to tritium fueling rate. According to R&D experiences of ITER, 30 the tritium fueling rate can be affected by many factors to maintain the operation of a fusion reactor. To maintain the equal share of deuterium and tritium in plasma, equal percent fuel was adopted in all fueling systems of ITER. Thus, the tritium fueling will contribute to all fueling requirements: substitute the burnt tritium, substitute the particle transport loss, substitute the Edge Localized Modes (ELM) loss, and so on. For ITER, the maximum tritium fueling rate is 100 Pam 3 s −1 , 31 and the tritium burning rate is 0.35 Pam 3 s −1 . Thus, a conservative 0.35% of burn-up fraction was regarded to be achieved in ITER under the condition 50:50 mix of DT. In an optimistic scenario, the tritium burn-up fraction can reach about 1% if only tritium fueling is used for replenishing the burnt tritium and particle transport loss and deuterium fueling for other fueling requirements, for example, triggering the ELM frequency for reducing the ELM energy loss, controlling the peak power load on the divertor plates, 30 as they do not contribute too much to the fueling for the core plasma burning. For tritium cycle time, early tritium cycle modeling studies always assumed 24 hours which referred to Tritium Systems Test Assembly (TSTA) results at Los Alamos National Lab (LANL) in 1986. [32][33][34] In recent studies, an ambitious goal for tritium recycle has been set as 1 hour for ITER tritium plant design. 35,36 Due to the comparable but different tritium fuel cycle of CFETR, the state-of-art prediction for CFETR was 2-6 hours. 37 Those predictions of tritium burn-up fraction and tritium cycle time will be validated in future ITER experiments.
To give a broad discussion, a sensitivity analysis of tritium start-up amount for CFETR was performed. The results are shown in Figure 2. They are based on the baseline parameters of fusion power (200 MW), tritium cycle time (4 hours), and tritium burn-up fraction (1%) using formula (5). It is clear that the tritium start-up amount requirement is about 500-1500 g if tritium cycle time and tritium burn-up fraction are controlled as 2-6 hours and 1%, respectively. From the engineering design viewpoint, the tritium start-up amount should be larger than the calculated value as some of tritium would be retained in the materials of vacuum vessel, tritium plant, and other pipes even in the initial operational phase. The research on dynamic initial tritium retention in the whole system is being conducted using COMSOL Multiphysics and EcosimPro platform. 38,39 (3)

| Feasibility of DD start-up
Under the pessimistic scenario of commercial tritium resources available, it seems no way to purchase enough tritium from CANDU reactors abroad for CFETR tritium start-up. 15 A solution named DD start-up that fusion reactor could be started up only fueling with initial deuterium was proposed. 40 Several studies evaluated the feasibility of DD start-up future fusion reactors. 27,34,41 Regarding CFETR phase I, dynamic ion density and fusion power in its initial start-up phase was estimated based on the parameters in Table 1. 8,42,43 The results of the estimation are shown in Figure 3. It can be seen that (a) D and T ion density will reach steady state after operation for a certain time without initial tritium supply. For CFETR, it needs about 2-3 years to reach steady-state ion density. The time depends on Tritium Breeding Ratio (TBR), tritium cycle time and fusion energy, etc.; (b) The ratio of D and T ion density depends on TBR when the ion density reaches steady state. The higher TBR is (before tritium selfsufficiency), the nearer a ratio of deuterium and tritium ion density approximates to 1.0. When the achievable TBR is higher than the required TBR of tritium self-sufficiency, the ratio is equal to 1.0; (c) A steady-state fusion power could be achieved even without initial tritium resources. And the steady-state fusion power at last depends on TBR (irrespective of whether tritium self-sufficiency is reached or not). Full power could be reached under the steady-state power operation if CFETR could achieve tritium self-sufficiency. On the other hand, if failing tritium self-sufficiency, for example, TBR = 0.99, full power could never be reached.
It is clearly visible from Figure 3B that the DD start-up would cause a fusion power loss for the fusion reactor in the initial phase, and that the fusion power loss corresponds to a benefits loss. On the other hand, purchasing the tritium for start-up would also require about US$(1.25-3.75) × 10 7 according to the tritium start-up amount of 500-1500 g and the current tritium price of US$25 million/kg. 15 The question which way is more economical one was also evaluated. For the TBR = 1.17, the benefits loss due to fusion power loss would be about US$3.71 × 10 7 regarding the ratio of electric power and fusion power as 47% and the grid purchase electricity price in China as US$0.057 per kW·h. 44 For the condition with higher TBR, the benefits loss would be decreased as the higher TBR would achieve full power earlier. Thus, both of the two ways for CFETR start-up are almost equivalent economical.
Both considering the factors of economy and tritium resources available, the DD start-up solution for CFETR or other fusion reactors would be more reasonably as long as the tritium price could not be greatly decreased in the future.

| Definition of tritium self-sufficiency and main design parameters of CFETR
The tritium self-sufficiency is defined if the achieved TBR is not less than the required TBR, or if the tritium production is not less than the tritium consumption. The tritium production depends on the global blanket breeding design. The tritium consumption includes three parts. The first is tritium burning in the plasma. The second is tritium decay which depends on tritium inventory (flow and retention) in the whole fusion system. The third is tritium nonradioactivity loss, for example, part of tritium retained in the fusion materials and finally regarded as radioactive waste without recycle. The definition of tritium self-sufficiency and the required TBR was shown and calculated using the formula (6).
To evaluate the feasibility of tritium self-sufficiency for CFETR, some baseline designed parameters were shown in Table 2. The tritium burn-up fraction and tritium cycle time have been discussed in the previous part. The burning availability of 30%-50% is the design target of CFETR. Regarding the tritium retention in fusion materials, a global simulation should be performed exactly. The maximum tritium retention amount in Plasma Faced Materials (PFM: tungsten) would be higher than 1 kg based on previous evaluation when reaching steady tritium concentrations. 45 However, based on ITER experiences, a safety limit of 1 kg was set from the public safety as well from a reasonable operational domain consideration. 46 To ensure public safety, it is self-evident to set the same safety limit for CFETR. It is not exactly known how much tritium will be retained in the blanket materials and tritium plant materials of CFETR. Based on ITER experiences, it seems that the tritium retention in tritium plant materials would be less than 1 kg. For tritium retention in blankets, a very rough estimate has been performed for ARIES advanced power plants, and the results showed that the maximum tritium retention amount was about 2 kg for ceramic breeder blankets. 47 Thus, referred to the data above, the maximum tritium inventory in the whole system of CFETR would be about 4.5-5.5 kg, which is slightly higher than ITER. The tritium inventory corresponds to a decay loss rate of 246-301 g/a.
Other factor of nonradioactivity loss can be determined only after finishing the detailed engineering design of detritiation systems and treatment methods of tritiated waste. If assuming that the lifetime of these components is 5 years and detritiation techniques could achieve 95% of tritium removal and recycling before the final disposal, 48 the nonradioactivity loss rate would be about 40 g/a. Another tritium nonradioactivity pathway is the normal emission into the air, and the emission rate is about 0.6 g/a based on ITER experiences. 46

| Required TBR for tritium self-sufficiency
According to the preliminary estimation, tritium decay and nonradioactivity loss rate is about 287-342 g/a. Thus, the net tritium breeding rate should be no less than the value to achieve tritium self-sufficiency. For CFETR phase I with a fusion power of 200 MW and burning availability of 30%-50%, the tritium burning rate will be about 3336-5560 g/a. Comparing the tritium generation with the consumption, the required TBR for tritium self-sufficiency should be higher than 1.1 under the condition that the burning availability is 30%, the tritium burn-up fraction 1%, the tritium cycle time 6 hours and the tritium retention amount 4 kg.
To further discuss the required TBR variation interval, the sensitivity analysis was performed over tritium cycle time (2-6 hours), tritium burn-up fraction (0.5%-2%), tritium retention amount (1-4 kg), and burning availability (30%-50%) refer to Table 2. The results are shown in Figure 4. From the analysis, the required TBR of tritium self-sufficiency is in the interval of 1.07-1.13 based on CFETR design parameters. On the other hand, if some designed targets failed, for example, burning availability, which is a key issue for the future fusion reactors, the required TBR would be increased significantly. The required TBR would be higher than 1.3 if only achieving burning availability of 10%. Beyond the burning availability, the tritium retention would also be a significant parameter for tritium self-sufficiency. For the current condition, the tritium retention contributes most to the total inventory. The decay factor of tritium retention contributes most to the required TBR. Lots of previous studies also showed concerns about a possible quite tritium retention which would bring great challenges for both the tritium selfsufficiency and the personal safety. [49][50][51][52][53] For CFETR, more R&D should be performed to validate tritium retention data and to develop new techniques with lower tritium permeation and retention in materials.

| Achievable TBR of blankets
Three options of blankets have been designed for CFETR, including a helium gas cooled tritium breeding blanket, a water cooled blanket and a LiPb liquid metal coolant blanket. 8 Tritium breeding is an important function of the blankets along with energy extraction and radiation shielding. Neutronics studies have been performed for three kinds of blankets. All calculations showed that the TBR would be about 1.1-1.2. [54][55][56][57] Since these calculations do lack of detailed engineering modeling, for example, ignoring the ports effect, the calculated TBR would drop greatly if ports and heterogeneity effects would be considered in detail in 3-D engineering modeling. In our works, the TBR was estimated using neutronics modeling based on blanket design of CFETR. The models are shown in Figure 5 and are defined as 1-D sphere model, 2-D column model, simplified and complicated 3-D model, complicated 3-D model with different port numbers. The TBR results of DD and DT fusion neutrons are shown in Table 3. The results clearly show that the TBR decreases with the complexity of blanket models. If we keep 16 ports for the CFETR machine, the TBR DT would decrease to about 0.99. From this point, potential risk of achievable TBR DT < 1.0 cannot be eliminated. As for the real numbers of ports, these are not yet finalized for CFETR. Referred to ITER, about 44 ports (18 upper ports, 17 equatorial ports and nine lower ports) are needed for remote handling operations, diagnostic systems, heating and vacuum systems. 46 Although there may be less ports for CFETR, it still may be the biggest challenge for the required TBR for tritium self-sufficiency.
Besides the determination of the TBR loss, there are also considerable uncertainties in the calculation results: due to the uncertainties of modeling, methods, and nuclear data, up to 10%-20% of maximum uncertainties exist in the calculations. 13,47,58 Therefore, both considering the potential tritium loss and the uncertainties, the achievable TBR of a global blanket may be lower than the required TBR for tritium selfsufficiency. From this follows that the risk of failing tritium self-sufficiency of CFETR could not be eliminated.

| Consequences in case of failing tritium self-sufficiency
There is no doubt that full power could be reached if tritium self-sufficiency is reached. However, according to the analysis above, the risk of failing the tritium self-sufficiency could not be eliminated. The consequences of failing tritium self-sufficiency were evaluated and are shown in Figure 6. CFETR could still achieve steady-state operation with a certain fusion power if failing tritium self-sufficiency. The direct effect is that the fusion power of steady state would be lower than the designed value. For the scenario of TBR = 1.08, the The fusion power loss corresponds to a benefit loss for a fusion reactor compared with the tritium self-sufficiency state. For phase I, the discussion of benefit loss is not so important, as cost-benefit is not the foremost concern. But for CFETR phase II with a fusion power over 1000 MW for DEMO validation, the cost-benefit would be a more important issue. For CFETR phase II, the cost-benefit evaluation for full year operation was performed and is shown in Figure 7. In case of failing tritium self-sufficiency, the costs of purchasing tritium to keep full power and benefits loss of losing energy are almost equivalent if the TBR is higher than 0.97. Below this value, costs of purchasing tritium to keep full power operation would be more than the benefits loss of the losing energy. Thus, if tritium self-sufficiency is failed, it is more economical to do nothing than to purchase tritium produced by CANDU reactors to compensate fusion power to full power.

| Future R&D needed for tritium selfsufficiency
According to the evaluation above, reaching the one coming close to tritium self-sufficiency is necessary for an economic fusion energy. There are still big challenges to achieve tritium self-sufficiency, especially the uncertainties between simulation and experimental results. In the future, more attentions should be paid on the R&D issues, such as blanket TBR, tritium burn-up fraction, tritium cycle time and retention in materials during the engineering design phase of CFETR or the next fusion reactors.
For the achievable TBR of blanket, firstly, the coverage rate of ports should be as small as reasonably possible to increase the global TBR as the ports are indispensable to the fueling, heating and so on. Good design should avoid the "buckets effect" for a whole fusion complex. Secondly, the 3D neutronics model should be upgraded from existing ~10 3 cells to ~ 10 5 cells level, which could significantly control the uncertainties of achievable TBR. And, essential experimental validations with DT neutron source should be performed to decrease the uncertainties in neutronics database, for example, tritium breeding experiments under module scale and fusion neutron environment.
Regarding the required TBR, the following related to parameters should be paid more attention: To fill the gaps for CFETR tritium self-sufficiency, in reality, R&D works related with tritium self-sufficiency were deployed and are ongoing with the support by the Ministry of Science and Technology (MOST) of China. An overview of R&D including the scopes of blanket, tritium plant, materials (tritium retention), F I G U R E 7 Cost comparisons of direct loss power and purchase tritium to reach full power for CFETR phase II in case of failing tritium self-sufficiency and plasma (tritium burn-up fraction) is shown in Figure 8. In the early years, most of the attentions were paid to the ITER TBM. Since 2011, the general design group for magnetic confinement fusion reactor was founded, which is also the original CFETR project team. Several projects were approved to support the concept design of CFETR. Eyes were also focused to the blanket design and tritium breeding experiment. Since December 2017, a new series of research projects were approved by Chinese government to start the engineering design of CFETR. At this stage, more attentions have been paid to tritium plant, materials and tritium burn-up fraction in addition to blanket. These researches aimed to balance the achievable global TBR of blanket with the required TBR for tritium self-sufficiency. It is self-evident Chinese government would make a stable budget support to fill the gaps for tritium self-sufficiency of CFETR. And it has long since been the holy grail of the fusion reactor to produce energy with economical fuels input.

| CONCLUSIONS
Tritium resources for the reactor start-up and tritium selfsufficiency during the operation is a key issue for CFETR. From this work, we make the following conclusions.

1.
The tritium start-up amount for CFETR phase I is about 500-1500 g which mainly depends on the tritium burn-up fraction and tritium cycle time. The reasonable value of tritium burn-up fraction and tritium cycle time are 1% and 2-6 hours, respectively. 2. DD start-up seems to be a more feasible choice than to purchase kilograms of tritium from market for CFETR start-up, since the available commercial tritium resources are too scarce to fully supply fusion reactors after ITER consumption, for example, CFETR, EU-DEMO, K-DEMO, FNSF, and Japanese DEMO. 3. The blanket achievable TBR should be higher than 1.1 (evaluation result) of required TBR for tritium self-sufficiency of CFETR phase I. Risk of failing tritium self-sufficiency still cannot be eliminated according to the calculation results due to the inevitable tritium ports loss and the considerable calculation uncertainties. 4. In case of failing tritium self-sufficiency of CFETR, the consequences are the loss of fusion power. Taking TBR = 0.99 as an example, the steady-state achievable fusion power is about 33% of full power. It is more economic competitiveness to do nothing than purchase tritium to compensate fusion power to full power. Totally, it still could be acceptable even if slightly failing tritium self-sufficiency for CFETR.