In this paper, the artificial neural networks (ANN) based deep learning (DL) techniques were developed to solve the neutron diffusion problems for the continuous neutron flux distribution without domain discretization in advance. Due to its mesh-free property, the DL solution can easily be extended to complicated geometries. Two specific realizations of DL methods with different boundary treatments are developed and compared for accuracy and efficiency, including the boundary independent method (BIM) and boundary dependent method (BDM). The performance comparison on analytic benchmark indicates BDM being the preferred DL method. Novel constructions of trial function are proposed to generalize the application of BDM. For a more in-depth understanding of the BDM on diffusion problems, the influence of important hyper-parameters is further investigated. Numerical results indicate that the accuracy of BDM can reach hundreds of times higher than that of BIM on diffusion problems. This work can provide a new perspective for applying the DL method to nuclear reactor calculations.

In many research works, the fluid medium in electroosmosis is considered to be a Newtonian fluid, while the polymer solutions and biological fluids used in biomedical fields mostly belong to the non-Newtonian category. Based on the finite volume method (FVM), the electroosmotic flow (EOF) of viscoelastic fluids in near-neutral (pH = 7.5) solution considering four ions (K⁺, Cl⁻, H⁺, OH⁻) is numerically studied, as well as the viscoelastic fluids’ flow characteristics in a microchannel described by the Linear Phan-Thien–Tanner (LPTT) constitutive model under different conditions, including the electrical double layer (EDL) thickness, the Weissenberg number (Wi), the viscosity ratio and the polymer extensibility parameters. When the EDL does not overlap, the velocity profiles for both Newtonian and viscoelastic fluids are plug-like and increase sharply near the charged wall. Compared with Newtonian fluid at Wi = 3, the viscoelastic fluid velocity increases by 5 times and 9 times, respectively, under the EDL conditions of kH = 15 and kH = 250, indicating the shear thinning behavior of LPTT fluid. Shear stress obviously depends on the viscosity ratio and different Wi number conditions. The EOF is also enhanced by the increase (decrease) in polymer extensibility parameters (viscosity ratio). When the extensibility parameters are large, the contribution to velocity is gradually weakened.

Computational accuracy and resource consumption are two sides of the mesh-based neutron transport calculation, whose balance is a common concern in engineering application. To overcome the inflexibility of the multiblock (MB) refinement technique and the complexity of the adaptive-mesh-refinement (AMR) technique, this paper presents a MB-AMR–based neutron transport lattice Boltzmann method (LBM) for the first time, which is a development and in-depth study for the current nonuniform mesh technique. The neutron transport problems are solved using the LBM with finite Boltzmann scheme, and the mesh configuration is adaptively adjusted using the MB-AMR technique. The MB-AMR technique combines the simplicity of the MB technique and the flexibility of the AMR technique and overcomes their shortcomings. By using invariant blocks, the complicated tree structure used in the traditional AMR technique is eliminated. By adjusting the mesh configuration according to the calculation results adaptively, the inflexible of the MB technique is overcome. By using the finite Boltzmann scheme instead of the traditional LBM, the implementation is further simplified and the interface treatment between different blocks can be solved as inner nodes using streaming process. Based on the above advantages and the simplicity of the LBM, the difficulty of the AMR technique in neutron transport calculation has been greatly reduced. To verify the accuracy and flexibility of the proposed MB-AMR–based neutron transport LBM, five benchmark problems are simulated. Results show that the proposed neutron transport LBM can simulate the multigroup transient and steady-state neutron transport problems accurately and that the MB-AMR technique can adaptively adjust the mesh configuration flexibly and simply. This paper may provide some alternative perspectives to realize the nonuniform mesh–based neutron transport solution and a powerful technique for large-scale engineering.

Based on the fluid hydrodynamic lubrication theory, a mathematical model of liquid film lubrication was established for the waviness hydrodynamic mechanical seal of a reactor coolant pump. The influences of the waviness amplitude and pressure on the sealing performance were investigated by the numerical simulation. The numerical results showed that the leakage rate increases linearly as the pressure and waviness amplitude increase under the force balance condition. The minimum liquid film thickness decreases first and then increase as the pressure increases. Stationary slide rings with three different waviness amplitudes were processed using the pre-deformation method and tested at different pressure and temperature. The experiments showed that all the three kinds of seal have good stability under different pressure conditions. The experimental results showed that the leakage rate is influenced by the pressure, waviness amplitude, and temperature, and the leakage rate increases as the pressure increases. The comparison between experimental and numerical results showed that both the temperature and pressure affect the seal performance, and tends to a smaller leakage rate, which is quite different from the numerical values. Therefore, the multi-physics coupling effect should be considered in the numerical analysis of seal performance, and this will be studied in the further works.