Modelling of combined dissolution-precipitation at the pore scale requires the conceptualization of mineralprecipitation and crystal growth, the formation of a protective diffusive layer by precipitants and slow reaction kinetics that are all coupled with alterations of the microstructure. In this work, we propose an improved approach for handling these challenges in a pore-scale coupled reactive transport model and apply it to portlandite carbonation. The model combines mineral geometry update as a consequence of dissolutionprecipitation reactions during diffusive transport through a saturated porous medium, thermodynamic equilibrium chemistry and dissolution kinetics. Transport of ions is calculated by the lattice Boltzmann transport solver YANTRA. Transport and reaction processes are incorporated at different spatial length scales with the multilevel approach, i.e. mixed liquid-solid nodes in a pore-scale model, which accounts for processes at scales below the model spatial resolution. Instead of defining arbitrary values such as threshold or residual porosities to initiate or halt precipitation, information on crystal shapes, packing and solubility in nano-pores based on interfacial surface energy is used to control precipitation. Additionally, the sensitivity study has been performed on model parameters such as portlandite dissolution kinetics, interfacial surface energy, calcite crystal size, CO2 partial pressure on the rate of carbonation and on the calcite layer properties such as residual porosity and thickness. From the comparison between the modelled calcite growth and the experimental data it has been found that low diffusivity of calcite layer decreases the effect of portlandite dissolution kinetics rate in case of carbonation and diminishes the effect of CO2 partial pressure. Also differences in the structure of the calcite layer were observed for carbonation of portlandite at low and high CO2 partial pressures.