Multidimensional optimization of the high-diodicity diaphragm hydrodiode for passive safety systems of nuclear power plants

A novel high-efficiency diaphragm hydrodiode (i.e., fluidic diode) for NPP safety circuits is proposed. To achieve maximum diodicity, a multi-parameter optimization of its geometry is performed using a machine-learning-aided surrogate model. Training the surrogate model is performed using the quasi-random sampling, while the exact diodicity values were provided by the CFD simulations based on the Reynolds-Averaged Navier-Stokes equations closed with the k−ω turbulence model. Iterative complementation of the sampling is employed to further increase the surrogate model accuracy. Genetic and Trust-Region optimization algorithms are executed on top of the surrogate model to arrive at the optimal hydrodiode configuration. The maximum diodicity value reported by both CFD and the surrogate model is D_CFD≈2.74, while the experimentally confirmed diodicity of the optimal diode configuration is found to be D_exp=2.59. Such a high diodicity value for the diaphragm hydrodiode is reported for the first time, thus constituting an achievement in the field. The proposed design and optimization methodology open up possibilities for constructing compact and reliable passive components for safety systems.