CFD simulation and experimental measurement of gas holdup and liquid interstitial velocity in internal loop airlift reactor

This paper documents experiments and CFD simulations of the hydrodynamics of our two-phase (water, air) laboratory internal loop airlift reactor (40 l). The experiments and simulations were aimed at obtaining global flow characteristics (gas holdup and liquid interstitial velocity in the riser and in the downcomer) in our particular airlift configurations. The experiments and simulations were done for three different riser tubes with variable length and diameter. Gas (air) superficial velocities in riser were in range from 1 to 7.5 cm/s. Up to three circulation regimes were experimentally observed (no bubbles in downcomer, bubbles in downcomer but not circulating, and finally the circulating regime). The primary goal was to test our CFD simulation setup using only standard closures for interphase forces and turbulence, and assuming constant bubble size is able to capture global characteristics of the flow for our experimental airlift configurations for the three circulation regimes, and if the simulation setup could be later used for obtaining the global characteristic for modified geometries of our original airlift design or for different fluids. The CFD simulations were done in commercial code Fluent 6.3 using algebraic slip mixture multiphase model. The secondary goal was to test the sensitivity of the simulation results to different closures for the drag coefficient and the resulting bubble slip velocity and also for the turbulence. In addition to the simulations done in Fluent, simulation results using different code (CFX 12.1) and different model (full Euler–Euler) are also presented in this paper. The experimental measurements of liquid interstitial velocity in the riser and in the downcomer were done by evaluating the response to the injection of a sulphuric acid solution measured with pH probes. The gas holdup in the riser and downcomer was measured with the U-tube manometer. The results showed that the simulation setup works quite well when there are no bubbles present in the downcomer, and that the sensitivity to the drag closure is rather low in this case. The agreement was getting worse with the increase of gas holdup in the downcomer. The use of different multiphase model in the different code (CFX) gave almost the same results as the Fluent simulations.