Solvent extraction is a liquid-liquid multi-phases process. The design and scale of the solvent extraction equipment is very difficult due to the small gravity different, droplets breakage- coalescence and high axial diffusion, et al. This work studied the hydrodynamic and mass transfer of the liquid-liquid system by experiment and simulation methods. The results are as below:
- The drag force coefficient is a key parameter for the simulation of multiphase flow. Numerous drag coefficient models have been presented in the literature but most of them are for steady cases. It has been reported that the unsteady effect accounts for ∼50% of the standard drag coefficient in the range of low Reynolds number. We measured the accelerating and decelerating motion of water droplets in an organic phase using a high-speed camera. An unsteady drag coefficient (UDC) was proposed as a function of acceleration (An), Reynolds (Re), and Archimedes (Ar) as follows:
- For single droplet, droplet clusters are also an important research direction. Considering of the deficiencies of Eulerian–Eulerian framework, a new method based on Monte Carlo approach was proposed to quantify droplet size distribution in PDDC. The new model used random sampling to instead averaging method. Thus, the effect of droplets with different sizes can be included in momentum equation. Simulation results showed that new model predicted the “U-shape” of holdup accurately and reduced the AARD of holdup
- Droplets are often the minimum mass transfer units of liquid−liquid extraction. However, the internal circulation and surface update of small droplets are limited. We developed a concept of “discrete liquid film”, which uses a wire immersed in the continuous phase and the droplets wet and move along the wire. The experimental results showed that the internal circulation was significantly strengthened. This new extraction increased mass transfer efficiency by 40−65%, specific surface area by 53.3%−69.7, and the residence time by 27.6%−51.9%.
- Machine learning algorithms as well as advanced optimization methods can aid in the design, operations management and intensification of complex chemical and manufacturing systems. The ML methods were used for predicting the PDDCs performance based on a well-prepared dataset of more than 1000 data points in our previous work1. The predicting AARDs of holdup, droplet diameter, axial dispersion coefficient and mass transfer coefficient are all below 15%. Feature importance of predicted parameters was analyzed using random forest, which would support the establishment of solvent extraction expert system.
- Scale-up of pulsed columns is an important process for industrial applications. However, the design of pilot-scale columns was mainly depended on experience of engineers because the lack of scale methods. Through orthogonal experiment of column diameter, plate spacing, and physical properties, we obtained several scale-up models to predict droplet diameter and holdup. The AARD were 15.4 % for holdup and 13.7 % for droplet diameter. The new models have been used for the design of industrial PDDC.
- More than 30 solvent extraction columns with different kinds of internals are designed and applied in hydrometallurgy and chemical process, including the separation of Rb/Cs from the mother liquor of lithium precipitation in the smelting process of lepidolite, the separation of Zr/Hf in MIBK-(NH4)2SO4 process, the separation of Ni/Co in MHP process, the extraction Li from the Qinghai Lake and the purification of pesticides process.