Deep geological repositories with a multi-barrier concept are foreseen by various countries for the disposal of high-level radioactive waste. Advanced simulation tools based on a detailed process understanding need to be developed for a close-to-reality description of repository evolution scenarios over time scales of some hundred thousand years. The construction of underground galleries and geotechnical barriers in the host rock formation and the emplacement of nuclear waste packages will create perturbations induced by chemical, thermal and pressure gradients at the interfaces of the different barriers, leading to mineral dissolution and precipitation to achieve re-equilibration. These processes can lead to an alteration of permeability, diffusivity and other physical characteristics of the rock matrix that can have significant effects on subsurface solute and gas transport. The understanding of these phenomena at the pore scale is a prerequisite for the development of predictive conceptual approaches to describe the evolution of the subsurface. Our lab on a chip concept, which combines microfluidic experiments and reactive transport modelling, has proven to be a powerful tool to (i) evaluate the impact of hydraulical heterogeneity on nucleation mechanism, (ii) decode oscillatory zoning exhibited by solid solutions crystallizing in porous media, (iv) assess the effects of confinement of crystallization, and finally (v) parameterize porosity-diffusivity relationships with respect to coupled mineral dissolution-precipitation reactions. In this work, we discuss how this pore-scale knowledge can be integrated into reactive transport models to achieve a realistic description of coupled processes at perturbed interfaces.