(291g) Resolved Simulation of Calcium Carbonate Crystallization Using the One-Dimensional Turbulence Model | AIChE

(291g) Resolved Simulation of Calcium Carbonate Crystallization Using the One-Dimensional Turbulence Model

Resolved simulation of calcium carbonate crystallization using the one-dimensional turbulence model

Derek D. Harris, David O. Lignell

Abstract –AIChE 2011 Annual Meeting

Mineralization of carbonate species is an important process with wide ranging use and application. There has been recent interest in technologies that produce carbonate precipitates from CO2 in combustion flue gas, either for cement production, or as a stable means for CO2 sequestration. More generally, carbonate precipitation is a specific application of aqueous crystallization processes.

The modeling and simulation of these processes are a significant challenge. These systems often involve turbulent mixing among two or more input streams, and require the accurate representation of a crystal population balance. Often, only rough details of this balance are of interest, such as product conversion or mean particle size, but the accurate determination of such quantities depends on the detailed physical processes occurring. The final total conversion of reactants may include dependence on particle size, shape factors, crystal structure, as well as the liquid composition. In turbulent environments, turbulent mixing may strongly impact the precipitation and crystal growth processes due to dependence on local liquid state. Disparate time and length scales are present, with scales ranging from fast liquid ionic chemistry, to particle nucleation, to slower particle growth processes. Additional complexities arise due to strong differential diffusion between particles (with essentially no diffusivity) and the mixing liquid streams (so that the solids will transport through the liquid composition space as bulk convection occurs).

Full resolution of these processes is commonly achieved using direct numerical simulation (DNS), where computational costs are extremely high, and configurations are limited to relatively low Reynolds numbers in canonical flows. Reynolds averaged Navier-Stokes (RANS) and large eddy simulation (LES) are effectively employed, but rely on subgrid models for unresolved fine structure, and cannot capture the detailed chemical and mixing processes that occur. The one-dimensional turbulence (ODT) model is an approach that fully resolves all turbulent flow structures in a single dimension, which may be regarded as a notional line-of-sight through a flow. This model has been successfully applied to a variety of turbulent flows, e.g., jets, wakes, mixing layers, buoyant stratified flows, and turbulent combustion, to name a few applications. ODT is computationally affordable since it solves only unsteady diffusion reaction equations in one dimension. Turbulent advection is modeled through stochastic eddy events implemented through domain remapping processes in a manner that is consistent with turbulent scaling laws.

We have applied the ODT model to the problem of aqueous calcium carbonate precipitation via the mixing of two reactant streams. The population balance is treated using direct quadrature method of moments (DQMOM), currently in terms of particle size, and polymorphic composition. Equilibrium is assumed for the nonideal aqueous streams, and carbonate is nucleated and grown kinetically. The yields of calcium carbonate and particle distribution properties will be presented, which depend on mixing rates, stream compositions, and temperature. These results are being applied for model development and validation for LES simulation of practical carbonate precipitation processes.