Brief introduction to CFETR TEP system

TEP is one of the most important systems in the inner fuel cycle process of CFETR, and its main design goal is to treat the plasma exhaust gas from the vacuum vessel and recover most of the tritium in the gas. Similar to ITER's TEP system [17], the TEP system of CFETR also uses a three-stage process to treat plasma exhaust gas: front-end processing, impurity processing, and final cleanup processing [18]. The equipment of the front-end processing system is placed in a separate glove box served by a Glove box Detritiation System (GDS), and the other two systems share one glove box with another GDS. Before being released into the environment, the remaining waste gas is temporarily stored in the decay tank. Power distribution cabinets (PDCs) are specially equipped to supply power. All the aforementioned subsystems and components of the TEP system are placed in a dedicated process room (Fig. 1).

Fig. 1Full-screen View

(Color online) TEP process room. (a) Schematic of TEP process room from top view; (b) 3D virtual geometric scene built by out tritium technology team.

During normal operation, the process room is under negative pressure (–100 Pa), provided by the HVAC (heating, ventilation, and air conditioning) system. If contamination is detected in the air, control actions are implemented by the Detritiation Systems (DSs). The ventilation of the affected room is isolated by the detritiation system from the HVAC, and the following measures are undertaken depending on the contamination levels: a) all personnel are evacuated, b) the source of contamination is identified and isolated, and c) the detritiation system is initiated and the depression in the affected room or area is maintained through the DS by a pressure controller and a flow damper.

Description of postulated accident

The postulated initiating incident is the instantaneous failure of the tritium process line in the front-end processing system at the maximum design flow rate. Because the pressure of the process line is higher than that of the glove box, process gas is released into the glove box. The release causes the tritium concentration to increase rapidly in the glove box. If the glove box system also fails, tritium sprays into the process room along the breach, resulting in the tritium concentration in the process room increasing rapidly.

According to the latest CFETR design, the tritium concentration in the process room equal to or higher than 1 DAC (3.5 × 10⁵ Bq/m³, alarm set point) would trigger the contamination alarm, resulting in operator evacuation procedures being initiated. In the event of the tritium release exceeding the threshold of 1 × 10⁸ Bq/m³ (isolation set point), the failed tritium process line would be isolated according to the instructions of the control system, and the DS would automatically isolate the affected sector from the HVAC system and activate the detritiation and depression functions. The DS would purge the contaminated process room by air to decontaminate the room atmosphere, and the design flow rate is 9000 m³/h. All the aforementioned parameters (i.e.,alarm set point, isolation set point, and design flow rate) are provided by the designer of CFETR tritium plant. The leak route of tritium is as follows: process line >> glove box >> process room >> DS >> environment, and the time sequence of the accident is summarized in Table 1.

Notably, there is a time interval between the moment of the exceeded tritium concentration being detected and the process line isolation and DS initiation. The delay time is designed to be 300 s. Thus, during this 300-s delay, tritium concentration is continuously increasing in the glove box and process room until the DS starts to purge air.

Governing equations

The aforementioned accident involves three-level confinement systems: the first barrier is the TEP process pipeline (broken), the second barrier is the related glove box system (failure), and the third barrier is the process room. The leaked tritium transport in the process room, and in the glove box, is essentially the mass transfer issue of the fluid in the flow process. The process complies with the basic equations of fluid control, which are the equations for the conservation of mass, momentum, and energy.

All the released tritium is assumed to be in the form of T₂. Tritium adsorption, desorption, and isotope exchange reactions are not considered in this work. The reason for their exclusion is that the humidity and oxygen content in the glove box and room are designed to be low. The temperature is maintained constant at 295 K during the accident, and no effective heat source is present in the process room. Therefore, considering energy transport is unnecessary. The gaseous fluid velocity in the glove box and process room is smaller than the local sound velocity, and the Mach number (Ma) is well below 0.3 in over 90% of the space. Thus, density changes in that low-velocity gaseous flow can be generally ignored, which means the gaseous fluid is incompressible.

The governing equations for the work gas flow process simulation are expressed as Eq.1 and Eq.2. The Navel-Stokes of Reynolds average (RANS) and k-ε turbulence model [19] equations are given as Eq.3-Eq.6. $\nabla \cdot u=0,$ (1) $\frac{\partial \text{u}}{\partial t}+\text{u}\cdot \nabla \text{u}=f-\frac{1}{\rho }\nabla p+\mu {\nabla }^2\text{u},$ (2) $\rho \frac{\partial k}{\partial t}+\rho \text{u}\cdot \nabla k=\nabla \cdot \left(\left(\mu +\frac{{\mu }_T}{{\sigma }_k}\right)\nabla k\right)+P_K-\rho \epsilon ,$ (3) $\rho \frac{\partial \epsilon }{\partial t}+\rho \text{u}\cdot \nabla \epsilon =\nabla \cdot \left(\left(\mu +\frac{{\mu }_T}{{\sigma }_{\epsilon }}\right)\nabla \epsilon \right)+C_{\epsilon 1}\frac{\epsilon }{k}{P}_K-C_{\epsilon 2}\rho \frac{{\epsilon }^2}{k},$ (4) $P_K={\mu }_T\left(\nabla u:\left(\nabla u+{\left(\nabla u\right)}^T\right)-\frac{2}{3}{\left(\nabla \cdot u\right)}^2\right)-\frac{2}{3}\rho k\nabla \cdot u,$ (5) ${\mu }_T=\rho C_{\mu }\frac{k^2}{\epsilon },$ (6) where velocity vector u and pressure p are to be determined in this study. ρ (kg/m³) represents density and μ (Pa·s) represent dynamic viscosity. f (N)represents force. k (m²/s²) and ε (m²/s³) represent turbulent kinetic energy and turbulent kinetic dissipation rate, respectively. $C_{\mu }$, $C_{\epsilon 1}$, $C_{\epsilon 2}$, ${\sigma }_k$, and ${\sigma }_{\epsilon }$ are the coefficients of the turbulence model and are listed in Table 2.

During the tritium transport process, there are fluid component transport processes because the leaked tritium is mixed with nitrogen in the glove box and air in the process room. When solving this type of problem, COMSOL predicts the local mass fraction of each component by solving the component transfer equation. The tritium leakage during the whole accident can be predicted to be less than 10% of the glove box volume, and this ratio is even smaller than that of the process room. Therefore, considering tritium a dilute substance relative to the background gas (nitrogen or air) is reasonable. The transport process of tritium is described by the convection equation and diffusion equations, Eq.7 and Eq.8, respectively. $\frac{\partial c_i}{\partial t}+\nabla \cdot J_i+u\cdot \nabla c_i=R_i,$ (7) $J_i=-D_i\nabla c_i,$ (8) where c_i (Bq/m³) represents the concentration, to be determined in this study. J_i (Bq/m²/s) represents the diffusion flux vector. R_i (Bq/m³/s) represents the reaction term, and is set to 0. D_i (m²/s) represents the diffusion coefficient, and the values were obtained from the literature [8], that of tritium against nitrogen is 5.65 × 10^-6 m²/s and that of tritium against air is 7.41 × 10^{- 6} m²/s.

Notably, in the process room, only the transport behavior of tritium in the air is considered, ignoring nitrogen leaking from the glove box into the process room. We made this assumption because a) our research object is tritium, b) there is a substantial difference in volume between the glove box (2.62 m³) and the room (242.09 m³), and c) 80% of the air is nitrogen; thus, even if the nitrogen in the whole glove box enters the air in the process room, there is little influence on tritium transport in the air. In other words, only one gas species is considered in Eq.7 and Eq.8, and the subscript i only refers to tritium.

Geometric model

The information from the China Academy of Engineering Physics (responsible for CFETR tritium plant design) on CFETR TEP system indicates that a three-dimensional geometrical model is built using the geometric kernel of COMSOL (Fig. 2). The inside of the TEP process room is a complete rectangular parallelepiped (10.83 m × 6 m × 4 m), filled with air and controlled at a negative pressure of 100 Pa below atmospheric pressure. In the process room, there are three glove boxes serving (1) the front-end processing, (2) the impurity processing and final cleanup processing, and (3) the gamma decay tank. Additionally, two PDCs stand by a wall: PDC-1 and PDC-2.

Fig. 2Full-screen View

Three-dimensional geometry model of glove box and process room and locations of monitors and leak holes. (a) Interested part of glove box 1; (b) Process room.

There are two simulation domains of interest in this study:

(1) Simulation Domain 1 (SD-1) in the simulation is the whole 3.82 m × 0.667 m × 1.2 m rectangular body in GB-1, filled with nitrogen and controlled at negative pressure of 200 Pa below atmospheric pressure. There is one cylindrical buffer tank (Φ 0.3 m × 0.55 m) and four cylindrical permeators (Φ 0.4 m × 0.8 m), all of which are placed on the ground vertically. The tritium in the TEP process line injects into the glove box through the assumed leak hole (Hole-1), and this hole is set at the top of the second permeator with Φ 0.04 m in the accident.

(2) Simulation Domain 2 (SD-2) in the simulation is the process room space. The tritium in the glove box diffuses into the process room through Hole-2, with Φ 0.04 m located at the front view surface of GB-1 in the accident. The coordinates of Hole-2 are set at (4.178, 1.500, 1.500) after analyzing the possible glove damage position due to long-time service. The DS in the process room is assumed to have a supply duct and an exhaust duct in the left and right walls and to operate in the once-through mode with a fixed fluid flow rate.

Nine monitors, from GB-M1 to GB-M9, are set in SD-1 for tritium concentration monitoring. Fifteen monitors, from PR-M1 to PR-M15, are set in SD-2. Table 3 shows the coordinates of tritium leak holes and tritium monitors and the center point of supply and exhaust ducts in the process room.

The basic properties of the working gas nitrogen in the glove box and those of the air in the process room use the parameters provided by the COMSOL material library. The software can automatically set nitrogen density (1.15 kg/m³) and viscosity coefficient (1.75 × 10^-5 Pa·s) according to the working environment of the glove box (295 K, 1 atm - 200 Pa = 101125 Pa). According to the working environment of the process room (295 K, 1 atm - 100 pa = 101225 pa), the air density (1.20 kg/m³) and viscosity coefficient (1.82 × 10^-5 Pa·s) are automatically set as well. The diffusion coefficients of tritium are obtained from the literature [8], and all the parameters have been introduced in Sect. 2.1.

Boundary conditions and initial values

In the tritium release stage of the accident, the boundary condition at Hole-1 is set as a velocity boundary, and the magnitude is given [20] by Eq.9. $v=\sqrt{\frac{2k}{k1-1}}RT\left[1-{\left(\frac{P_2}{P_1}\right)}^{\frac{k-1}{k}}\right]$ (9) where $v$ represents the leakage velocity, m·s^-1. k represents the gas adiabatic index, which is 1.4. R represents the gas constant, which is 8.314. T represents the temperature, 300 K. P₁ represents the pressure in the tritium process line, 0.25 MPa; P₂ represents the pressure in the glove box, 1 atm–200 Pa. According to the aforementioned values and Eq.9, the breach velocity is approximately 63.4 m·s^-1.

The walls shared by SD-1 and SD-2 are set as an interial wall and a thin impermeable barrier. The boundary conditions of the other walls are set to wall no slip. Because the DS is not activated in the release stage, the supply vent boundary is set to the wall boundary, and the exhaust vent boundary condition is the pressure boundary with a relative pressure of –100 Pa. The initial tritium concentrations in SD-1 and SD-2 are both set to 0 Bq/m³ (Table 4).

In the tritium removal stage, because the tritium stops leaking from the holes, the boundary condition at this point is reset to the wall boundary. The boundary condition at the supply vent is reset to the flow rate boundary, and according to the design of the DS, the flow rate is given as 9000 m³·s^-1. The walls shared by SD-1 (glove box) and SD-2 (process room), including the one with Hole-2 on it, remain as an interial wall and thin impermeable barrier. The exhaust vent maintains the pressure boundary at a relative pressure of -100 Pa.

Tritium release stage

4.1.1

Contours of tritium concentration

Using the powerful visualization function of COMSOL, Fig. 3 shows the concentration contour maps to observe the three-dimensional evolution of tritium transport. Approximately 0.01 s after the instantaneous failure of the tritium process line, tritium reaches the top of the glove box; the shape of tritium diffusion during this time is similar to a flower bud. Next, this “flower” blooms, and the outer petal symmetrically covers the whole y direction of the glove-box top at 0.026 s. As the tops of the outer petals increase in size and reach Hole-2 at 0.078 s, the symmetry is broken. The “flower” blooming continues, but the tops of the outer petals are destroyed.

Fig. 3Full-screen View

(Color online) Contours of tritium concentration in the glove box

The contours of the tritium concentration in the process room are similar to those in the glove box. As shown in Fig. 4, approximately 2.30 s after the accident, the symmetrical outer petal of the tritium “flower” reaches the nearest vertical wall of the process room, and as the top of the outer petals increases in size along the wall, the outer petal symmetrically arrives at the bottom of the wall at 23.67 s. Next, the “flower” continues to move along the wall and the floor, and as the dispersion proceeds, the vortex phenomenon becomes more pronounced.

Fig. 4Full-screen View

(Color online) Contours of tritium concentration in the process room

As shown in Figs. 3 and 4, tritium migrates more quickly along the wall than in space, making it easier to understand the movement of the top petals.

4.1.2

Slices of tritium concentration

Figures 5 and 6 show the slices of tritium concentration. As shown in the first picture of Fig. 5, at the moment of the breach, tritium is quickly ejected from Hole-1 (in the pipeline) and released to the top of the glove box within 0.01 s, and from this point of view, it resembles a burning candle. The “smoke” generated from this “burning candle" moves approximately symmetrically to the left and right sides and forms an umbrella-shaped "smoke" distribution under the limiting top wall of the glove box. As the release time increases, the tritium concentration increases, and the maximum velocity remains to be observed at the leak hole or, specifically, on the center line of jet flow. At the end of the tritium release stage (300 s), the tritium concentration of the whole glove box reaches 2.36~3.13 × 10¹⁵ Bq/m³.

Fig. 5Full-screen View

(Color online) Glove box tritium concentration distribution of ZX slice with y = 1.800 m

Fig. 6Full-screen View

(Color online) Tritium concentration of YZ slice with x = 4.178 m (the same x-coordinate as Hole-2)

As shown in Fig. 6, the tritium in the glove box migrates more quickly along the wall than in space. The tritium escapes from the glove box into the process room mainly via three routes: 1) tritium is ejected from Hole-1 into the top of the glove box and migrates along the -y direction and then down into Hole-2; 2) tritium is injected into the top of the glove box and migrates along the +y direction and other directions, and then down into the bottom of the glove box and up into Hole-2; and 3) under the gradient drive of concentration and velocity, tritium in the space of the glove box moves into Hole-2.

4.1.3

Concentration of monitoring points

The variation in tritium concentration at monitoring points might provide more quantitative information from the perspective of safety. Figure 7a shows the variation in tritium concentration from 0 to 300 s at nine monitoring points in the glove box. The conclusions drawn from Fig. 7a are as follows:

Fig. 7Full-screen View

(Color online) Tritium concentration of monitoring points in tritium release stage. (a) In the glove box; (b) In the process room.

1. The trend of tritium concentration is similar at all monitoring points. There is a rapid climb from 0 to 1 × 10^14~15 Bq/m³ during the first 2.5 s, and then tritium concentrations at all monitoring points increase slowly. After approximately 60 s, the tritium concentration basically tends to a constant value. Two groups are formed: 1.63~1.73 × 10¹⁵ Bq/m³ and approximately 2.57~2.63 × 10¹⁵ Bq/m³.

2. At the initial stage, the tritium concentration in GB-M6 is the highest, followed by GB-M5/ GB-M7. At less than 0.06 s, the concentration in GB-M6 and GB-M5/GB-M7 exceeded 10¹⁰ Bq/m³. These phenomena are easily understood because GB-M6 is at the top of Hole-1, and GB-M5/GB-M7 is on the wall very close to Hole-1. For GB-M5 and GB-M7, two spatial symmetry points of the pipeline breach, the tritium concentration basically shows a uniform step, and the former is closer to Hole-2 than the latter one, making the concentration at the former a little higher than the latter in a certain period. A similar phenomenon was observed in [10]: the concentration at the monitoring points under the open space condition is slightly larger than that under the closed space condition because the closed boundary suppresses tritium diffusion.

3. GB-M1 and GB-M3, and GB-M2 and GB-M4, are almost two groups of spatial symmetry points and far from Hole-1 and Hole-2. The variation in tritium concentration shows the uniform changing pattern for the symmetrical points of each group. GB-M8 and GB-M9 are perfect symmetry points, and the change lines for the concentration at these points almost coincide.

4. Notably, there is a concentration inversion. In the first 4 s, concentrations at GB-M6 and GB-M5/GB-M7 are higher than those at the other six monitor points. However, after 23 s, the concentrations at GB-M6 and GB-M5/GB-M7 are obviously lower than those at the other monitor points. The possible reason for this finding is that tritium moves more quickly along the wall than in space. In the first 4 s, the amount of tritium that arrives at GB-M6 and GB-M5/GB-M7 is more than the amount of tritium that escapes into the other monitoring points in the process room (see the cases at 0.10 s and 1.02 s in Figures 5 and 6). As the tritium concentration increases and the space distribution becomes uniform for the nine monitoring points (see the cases at 7.80 s and 20.91 s in Figures 5 and 6), more tritium escapes into the process room along the wall after 23 s compared with the other monitoring points.

Figure 7b shows the variation in tritium concentration from 0 to 300 s at 15 monitoring points in the process room. The conclusions drawn from Fig. 7b are as follows:

1. The trend of tritium concentration is similar at all monitoring points in the process room. The concentration climbs quickly, grows slowly, and then tends to a constant value. Different monitoring points have different change times. The space in the process room is much larger than that in the glove box, and monitoring points are set to be distributed in each representative area. Hence, concentrations at different monitoring points show a relatively dispersed distribution compared with those in the glove box. There is a special monitoring point called PR-M11 between the two glove boxes, and the complex flow field leads to a reduced concentration increase, and this point seems “safer” with the lower tritium concentration because its location between the two glove boxes prevented some tritium from flowing through.

2. The tritium concentration at PR-M2 is the highest, followed by PR-M1, through the whole tritium release period because it is in the direct route of the tritium that escaped from Hole-2. At the initial stage, the next highest points are PR-M3, PR-M4, and PR-M5, whose concentrations exceed the alarm setpoint (3 × 10⁵ Bq/m³) within 1.31 s. Due to tritium moving more quickly along the wall than in the space, once the tritium arrives at PR-M5 (on the top wall), which is farther from Hole-2 than PR-M3 and PR-M4, the concentration in PR-M5 increases faster than that of PR-M3 and PR-M4. PR-M7 (on the wall) and PR-M6, and PR-M10 (on the wall) and PR-M8, are two groups showing similar trends as PR-M5 and PR-M3/PR-M4. PR-M12, PR-M13, and PR-M15 are far along the route of tritium migration, and the concentrations at these monitors increase later.

3. From a safety perspective, the monitoring point with the fastest growth rate of tritium concentration should be selected as a reference, i.e., PR-M2. T₁, T₂, and T₃ in Table 1 are within 0.05 s after the accident. Notably, the tritium continues getting released because of the 300-s time interval between the moment of the exceeded tritium detection and process line isolation, as well as DS activation. Therefore, the tritium concentration continues increasing in the glove box and the process room. At the end of the tritium release stage (300 s), the tritium concentration in the process room reaches 1.02 ~ 7.05 × 10¹⁴ Bq/m³. Next, the DS in the process room starts to bring in the air. Notably, the highest concentration range in the process room is approximately 7 times larger than the lowest, and the concentration distribution in the process room is more dispersed than that in the glove box (1.63~2.63 × 10¹⁵ Bq/m³). This result is observed because the space of the process room is much bigger than that of glove box, monitoring points are scattered among the space, and there are complex obstacles in tritium transport.

Tritium removal stage

In the tritium removal stage, when the DS is initiated, clean air is gradually blew into the process room and destroys the original tritium distribution pattern in the tritium release stage. The final tritium concentration distribution in the tritium release stage (Sect. 3.1) is the initial tritium distribution in the tritium removal stage. Zero seconds in Sect. 3.2 is the start time of the tritium removal stage, the same as the end of the tritium release stage.

4.2.1

Slices of tritium concentration

As the clean air enters the process room, the tritium concentration in the area facing or near the supply vent decreases first. As shown in Fig. 8, there is a low concentration region at the bottom right corner at 0.20 s; thus, clean air arrives at x = 4.178 m (Hole-1) from x = 0 m (Supply duct) in 0.2 s, and this region remains as the lowest tritium concentration. Tritium concentration continues to decrease as the amount of penetrated air increases. After 30 s, the concentration distributions are similar or the same (particularly after 500 s) except for the range of the legends. In other words, a relatively stable tritium concentration distribution has been formed, and a new tritium concentration distribution pattern has been established.

Fig. 8Full-screen View

(Color online) Tritium concentration of YZ slice with x = 4.178 m (in the tritium removal stage)

4.2.2

Concentration of monitoring points

Figure 9 shows the tritium concentration at monitoring points in the tritium removal stage. Driven by the gradient between SD-1 and SD-2, tritium escaped from the glove box through Hole-2, resulting in a decrease in tritium concentration in the glove box. There is a short period (approximately 20 s) of irregular decline, and then the tritium concentrations at all monitoring points in the glove box become nearly a set of log-linear graphs. There are two groups with similar slopes but different tritium concentrations on the line: one for GB-M6 and GB-M5/GB-M7 and the other for the other monitoring points, the same as the grouping in the tritium release stage.

Fig. 9Full-screen View

(Color online) Tritium concentration of monitoring points in tritium removal stage. (a) In the glove box; (b) In the process room.

Tritium in the process room is driven out of the room directly by the purge air, while tritium continues to escape from the glove box into the process room through Hole-2. Notably, the former is dominant, and in Fig. 9b, the tritium concentration shows a decreasing trend. Similar to the glove box, there is a short period (approximately 30 s) of irregular decline in the process room, and then the tritium concentrations at all monitoring points become a set of nearly log-linear graphs. The slope of the line is related to the tritium removal rate, and the greater the slope, the faster the tritium is removed. PR-M2 has obvious advantages in tritium removal rate because the primary flow direction in this location is along the x+ direction, moving to other spaces or the exhaust vent without back-mixing (Fig. 10). PR-M15 has the lowest tritium removal rate, and this result is easy to understand with the streamline in the process room shown in Fig. 10. In this location, the primary flow is characterized spatially as a "spiral flow," PR-M15 seems to be at the center of the spiral, and most of the tritium cannot move to the exhaust vent directly.

Fig. 10Full-screen View

(Color online) Tritium concentration streamline in process room in tritium removal stage. (a) Visual angle 1; (b) Visual angle 2.

The time required for the tritium concentration to fall to the isolation threshold is 1941 s, and the time needed for it to fall to the alarm set point is approximately 2667 s (equivalent to T4 in Table 1). The DS continues purge the process room for a longer period of time. When the DS continues working for 3000 s, the maximum concentration among the monitoring points in the process room decreases to 2.61 × 10⁴ Bq/m³. At 3600 s, the maximum concentration decreases to 243.5 Bq/m³, less than 1/1000 of the alarm setpoint and 1/100 of the detection limit for general tritium air monitors. Notably, once the DS in the room stops working, the tritium in the glove box will continue to leak, increasing the tritium concentration in the room. However, 1 h of continuous purging decreases the tritium in the glove box and process room to 1.44 × 10³ Bq and 4.70 × 10⁴ Bq, respectively (Fig. 11). The ratio of glove box space size to the process room is ~ 1/100 (2.6192 m³/242.09 m³); thus, even if all the residual tritium in the glove box entered the room, the effect on tritium concentration in the room would be slight.

Fig. 11Full-screen View

Total tritium in tritium removal stage. (a) Tritium in SD-1; (b) Tritium in SD-2