1. Introduction

Anionic radionuclides (e.g., I and Tc) are highly mobile because of their high solubility and low adsorption tendency towards the solid media in the underground environment [1,2]. Consequently, they possess one of the highest risks regarding the long-term safety assessment of the disposal repository [3]. Moreover, they are considered as the radionuclides from the accidents; therefore, studies have been conducted to develop techniques for the immobilization, separation, and recovery of the anionic radionuclides [4‒13]. The representative methods to capture such radionuclides from aqueous media are mostly based on chemical as well as physical adsorptions. In particular, chemical adsorptions mainly rely on the chemical reaction between the transition metals and anionic radionuclides, whereas the physical adsorptions are reliant on the hydrophobic reactions at the interface between the adsorbent and the anionic radionuclides [14]. Thus, the key components for immobilization include the anionic radionuclide, transition metals, and adsorbent with adequate interface properties. The adsorbent materials can be activated carbons, inorganic porous materials, metal organic frameworks, or nanomaterials [15‒18].

In this regard, carbon nanotubes (CNTs) are well-known nanomaterial-based substrates that can be used as potential adsorbents of anionic radionuclides; they are composed of carbon and have large surface areas with high potential for surface modification [19]. Theoretically, pure CNTs are linear and possess only carbon atoms. From the viewpoint of the surface, the CNTs exhibit π-π stacking of electrons; thus, they likely have negative surface charges similar to the anionic radionuclides. Thus, the intermediates having positive charges would be helpful in the effective immobilization using a stepwise approach. Fig. 1 schematically represents the anionic radionuclide immobilization concept. The key components (e.g., the anionic radionuclides, transition metals, and CNTs) as well as the stepwise approach for the anionic radionuclide immobilization are shown in Fig. 1(a) and 1(b), respectively.

Fig. 1

Principle of the immobilization of anionic radionuclides using carbon nanotubes, a nanomaterial-based substrate. (a) describes the components with the charge notations (e.g., + and ‒) needed for the immobilization and (b) shows the step-wise approach for the immobilization. AR and TM represent anionic radionuclides and transition metals, respectively.

Here, we hypothesize that the anionic radionuclides are immobilized successfully onto CNTs with the help of transition metals, and the particles generated during the process can flow through the water [20]. We conducted a particle tracking analysis using computational fluid dynamics (CFD) to separate the particles from the flow and design a sandglass-like separator based on the concept of a cyclone separator. We discuss the morphology-associated factors resulting in the high efficiency of particle separation. We subsequently discuss the effects of the magnetic properties introduced in the sandglass-like separator.

2. Particle tracking based on computational fluid dynamics

We consider the immobilized radionuclides to be particles denser than water and analyze the particle trajectory using ANSYS 18.1, which probabilistically analyze using CFD based on the finite volume method (FVM). Moreover, behaviors of the particles are considered and analyzed based on the fluid flow. Thus, the fluid domain for the simulation is first defined based on the following parameters: the reference pressure, material, temperature, heat transfer, and turbulence model are set as 1 atm, water, 25℃, none, and k‒ε model, respectively. The boundary conditions for the inlet and outlet are as follows: The velocity for the inlet was set to 1 m·s^-1, and the relative pressure for the outlet was 0 Pa. The wall condition was set to no-slip condition, reflecting the actual water flow characteristics. We note that free-slip wall condition is generally used for gaseous flow. We set the thermodynamic properties of the particle and water as follows: the specific heat capacity and density for the particle are 0 J·kg^-1·K^-1 and 2,000 kg·m^-3, respectively, whereas those for water are 4,181.7 J·kg^-1·K^-1 and 997 kg·m^-3, respectively. The sizing parameters used for the particle are as follows: minimum, maximum, and mean diameters, and the standard deviations are 50×10^-6, 500×10^-6, 250×10^-6 and 70×10^-6 m, respectively. We consider the mass based on a normal distribution for particles with an average diameter, and the number of particles used is 10,000 EA. A uniform flow injection is set along the inflow cross section because our purpose of the analysis is to study whether the fluid (i.e., a mixture of water and the particles) can maintain a slow rate in a certain section.

2.1 Analysis of the conventional cyclone separator

Studies have shown that cyclone separators are quite efficient for the particle separation from gaseous streams [21-24]. In conventional cyclone separators, the principle of inertia is used to separate the dust from the mixtures of the air and dust, where the relative mass differences in the rotation result in effective separation. However, the effectiveness of the separation for aqueous fluids (i.e., mixtures of water and particles) is not conclusive for the conventional cyclone separator. To verify the separation efficiency for the aqueous fluid, we performed particle tracking simulations for the existing, yet simplified, morphology of a conventional cyclone separator (Fig. 2). It is seen that the aqueous fluid flows through the pipe. The fluid introduced into the inlet section flows in a direction tangential to the wall of the central cylinder, so that the generated centrifugal force guides the flow to reach the separation box located at the bottom. Evidently, the fluid passes through the separation box prior to reaching the outlet. The inlet and outlet are separated by a separation plate and a flow guide to prevent the inlet fluid from going directly to the outlet. This is our basic model for the conventional cyclone separator.

Fig. 2

(a) Basic model of the conventional cyclone separator representing boundary conditions for water inlet and outlet, (b) particle velocity (trajectory), and (c) pressure distributions for 'a'. (d) The model with a flow guide, (e) particle velocity (trajectory), and (f) pressure distributions for 'd'.

Fig. 2(a) shows the boundary conditions and the water inlet and outlet for the basic model. Additionally, the particle tracking trajectory and pressure distribution results are shown in Fig. 2(b) and (c), respectively. The line represents the particle trajectory, whereas its color represents the particle velocity. It is seen that the velocity of the flow was relatively low in the separation box; however, the pressure distribution was the same in the central cylinder and separation box. Thus, the particles are expected to eventually reach the outlet. Fig. 2(d) shows the basic model with a flow guide. We introduce a flow guide to control the pressure field in the central cylinder area, expecting that more particles will reach the separation box compared to the outlet. The velocity is relatively low in the separation box, and the amount of fluid flowing out through it became noticeably less (Fig. 2(e)). We note that the number of particles flowing out through the outlet decreased as well, which supports the need for the flow guide in our model. A relatively small number of particles flowing out through both the separation box and the outlet indicates that a large number of particles rotate and remain in the central cylinder. This is in good accordance with the results of the pressure distribution simulation (Fig. 2(f)). However, it is expected that the particles will eventually escape from the separation box and flow to the outlet with time. Thus, our basic model and the model with the flow guide are not effective in separating the particles from the fluid. This reflects that the effect of the density difference between the water and the particles is not sufficient to separate them in the conventional cyclone separator.

2.2 Effect of the morphology of cyclone separator

From the results obtained for the basic model of the conventional cyclone separator, it is obvious that we cannot achieve and effective separation. Thus, a need for a new design for the cyclone separator arises, by which the particles will be retained in the separation zone. The density difference is still the key principle for the cyclone separator; thus, the fluid must rotate within the separator. However, the particles must not escape from the separation zone through the flow.

We propose a sandglass-like morphology for the separators (Fig. 3). The inlet is installed in the upper part of the separator in a direction vertical to the radius; thus, the inlet fluid rotates in the separator. All three models have a vertical outlet pipe located at the center of the upper part of the separators. Model-I has an outlet pipe that starts from the middle to the top of the entire separator (Fig. 3(a)), whereas model-II has a shorter pipe than that of model-I (Fig. 3(b)). Model-III has a larger lower part as the separation zone of the separator (Fig. 3(c)). When the fluid enters the separator, it flows downward to the bottom and then flows out through the outlet pipe. The results of the particle tracking analysis done for model-I, II, and III are shown in Fig. 3(d), (e), and (f), respectively. The line represents the particle trajectory, whereas its color represents the particle velocity. It is seen that the velocity was relatively low in the separation zone; however, the trajectory showed that about 40% and 50% of the particles escaped from the separation zone in model- II (Fig. 2(e)) and III (Fig. 2(f)), respectively. For model-I, however, we found that almost no particles escaped from the separation zone (i.e., only one out of 10,000 particles flowed out from the outlet). Thus, out of the three models, we conclude that model-I best separates the particles from the mixtures with a reasonable efficiency. We note that the numerical modeling of the fluid for the model-I was performed with 52,562 elements, 10,437 nodes, and a mesh type of tetrahedron. The flow rate at the inlet was set to 1 m·s^-1. The meshed sandglass-like separator and the boundary conditions are shown in Fig. 4.

Fig. 3

Morphology of the proposed cyclone separator: (a) model-I (the vertical outlet pipe is located in the center), (b) model-II (the outlet pipe in the separator is a half size compared to that of model-I), and (c) model-III (the lower part of a separator is larger). The simulation results for the particle tracking for (d) model-I, (e) model-II, and (f) model-III are shown for comparison.

Fig. 4

(a) Meshed sandglass-like separator and (b) the boundary conditions.

3. Separation efficiency test for the sandglass- like separator

Fig. 5 shows the detailed design of the sandglass-like separator proposed from the simulation results. The blueprint of the separator includes the drawings of the top, bottom, and body views with precise dimensions. The height of the separator is 150.0 mm, whereas the outer diameter at the top and bottom is each 84.8 mm, and the inner diameter at the middle is 41.8 mm. The top cover of the inlet and outlet was made of stainless steel to provide strength of the assembly. We used transparent acrylic materials for the body part to monitor the flow of the particles. The gaskets are placed between the covers and separator to prevent any fluid leakage.

Fig. 5

Detailed design of the sandglass-like separator. The drawing of (a) the top cover with the inlet and outlet installed, (b) the bottom plate, and (c) the body part made of a transparent acrylic material (all dimensions are in mm).

Fig. 6 shows the design and actual experimental setup for the validation of the separation efficiency of the sandglass- like separator. The separation system includes a water tank, pump, pump controller, flow meter, and separator. The fluid, which is a mixture of water and particles, flows with a constant flow rate through the pump and enters the separator. Inside the separator, the particles are trapped in, whereas only water flows out from the outlet and then back to the water tank (reservoir). The flow meter measures the amount of fluid, while the pump controller controls the flux according to the applied voltage. Finally, a flux of 3 L·min-1 was used, which corresponds to a flow rate of 1 m·s^-1.

Fig. 6

(a) Design and (b) image of the experimental setup for the validation of separation efficiency of the proposed sandglass-like separator.

Fig. 7 shows the separation efficiency test for the proposed sandglass-like separator. We used iron (Aldrich, ≥ 99%, fine powder) as the representative particles for our test because of its magnetic properties and ability to work as an intermediate in our hypothesis (Fig. 1). Iron is a transition metal that belongs to the first transition series in the periodic table and is the most common element on Earth by mass. Fe and Fe-associated materials are used as absorbents for anionic radionuclides [7,8,18,25]. Since Fe shows a strong ferromagnetic property at room temperatures, it can provide a higher separation efficiency, which we validate using the magnet. The size of the iron particles used in this test ranged from below 45 to approximately 200 μm. To remove the initial air in the separator setup, we filled it with water flowing at a constant rate (e.g., 1 m·s^-1); then, we introduced 6 g of the iron particles into the reservoir after the system stabilized (Fig. 7(a)). When the iron particles entered the sandglass-like separator, they rotated down through the flow (Fig. 7(b)) and reached the bottom of the separator (Fig. 7(c)). Then, the majority of the particles were trapped at the bottom of the separation zone, whereas some continue rotating through the flow while it continues (Fig. 7(d)). However, we did not observe any obvious particles escaping through the outlet pipe. As such, the results are in good agreement with our simulation results.

Fig. 7

Experimental process for the separation test in the sandglass-like separator (a) pre-injection, (b) particle injection, (c) the particles flowing within the separator, and (d) the particles trapped in the separation zone.

After 5 min of operation, we collected the particles from the separation zone, dried, and weighed them in addition to collecting the fluid from the outlet to confirm the separation efficiency. The particles obtained from the separation zone weighed 5.94 g, whereas those from the fluid was ca. 0.06 g. This reveals that the separation efficiency of the sandglasslike separator was approximately 99%, and it can be further enhanced by the effect of the magnetic property. We installed a magnet at the outer bottom part of the separator and conducted the test. As shown in Fig. 8, more particles were collected at the bottom part of the separation zone using the magnet (Fig. 8(a)), as compared to those without it (Fig. 8(b)). The efficiency is enhanced to ca. 100% with the effect of the magnet in addition to the sandglass-like design of the separator.

Fig. 8

Particles collected at the bottom part of the separation zone of the sandglass-like separator (a) with and (b) without using the magnet.

4. Conclusion

A sandglass-like separator, based on the concept of a conventional cyclone separator, for the aqueous fluid was proposed using a particle tracking analysis based on CFD. It was seen that only one out of 10,000 particles escaped from the separation zone through the flow of the separator. In comparison, a conventional cyclone separator showed unsatisfactory results of the separation of the particles from the aqueous fluid. The manufactured sandglass-like separator in this study showed an efficiency of 99%. With an additional magnetic effect, the efficiency reached ca. 100%. The sandglass-like separator can be used for the effective separation and recovery of the particles including the immobilized anionic radionuclides. Applicability to the field of separation and filtration for the removal of foreign substances without pressure disturbances in the pipe needs to be investigated.