Development of an Openfoam Solver for Numerical Simulation of Carbonization of Biomasses in Rotary Kilns

Carbonization is a key process to increase the energy density of high moisture-containing biomasses and biogenic wastes and to provide multipurpose raw chemicals for further applications. Steam-assisted carbonization is a kind of slow pyrolysis, in which wet biomass is treated continuously in superheated steam at elevated temperature and atmospheric pressure. Rotary kiln reactors due to their flexibility and easy control of operating conditions are well suited for this process. In this work, a numerical simulation tool based on an Eulerian-Langrangian approach has been developed to simulate the carbonization of biomasses in rotary kiln reactors resolved in time and space by combining existing OpenFOAM features and developing new physical models. This study demonstrates the features of this extended and validated Eulerian-Lagrangian approach for simulating dense particulate multiphase flows in large-scale rotary kiln reactors. The focus is to use the new tool to aid the design of large-scale rotary kiln reactors by performing parameter studies. The simulations of this kind of large-scale reactors require large computational resources on supercomputers. ... mehrTherefore, a further focus lies in different approaches to reduce the computational effort while keeping the accuracy at an acceptable level. By using the MP-PIC model, computing time increases linearly with the number of biomass particles instead of exponentially with the DPM model. The optimal cell size has been found to be about twice the largest particle diameter. By choosing the optimal domain decomposition method, simulation time can be reduced by a factor of 1/10. Introducing a solver frequency parameter to the DOM radiation model can help to reduce simulation times further by a factor of 1/8 while decreasing the accuracy by only 2%. Parallel scaling tests show good performance with over 1000 CPU cores. These results show that simulations with a total of 40,000 CPU-hours per studied case become feasible proving the developed solver to be an efficient tool for the design of rotary kiln reactors.