A numerical simulation model was designed to evaluate the technical viability of in-situ upgrading using dispersed nanocatalysts in heavy oil reservoirs. Aquathermolysis reactions are represented by a practical kinetic model based on SARA analysis, being consistent with the thermodynamic characterization. With this simplified model, the API gravity enhancement in core-flooding tests was reproduced. The mathematical formulation couples mass and energy transport equations along with a rigorous three-phase equilibrium equation of state. Also, a nanoparticle transport equation was coupled to account for reversible and irreversible non-equilibrium retention, and water-oil partitioning. PVT data were matched successfully, including API gravities and oil viscosities. Reaction rates were adjusted by means of batch-reactor information, while nanoparticle retention was validated using reported single-phase core-flooding tests. Different core-flooding experiments from the literature were reproduced to calibrate the phases transport parameters, and further up-scaled to reservoir conditions. Validation of the model with experimental data suggests that the lumping scheme based on SARA analysis and the modeling of nanoparticle transport and retention, capture the most important phenomena occurring during in-situ upgrading processes. Field-scale simulations, of a sector model from an oil reservoir in the Magdalena Medio Valley basin in Colombia, showed that the in-situ upgrading with nanoparticles can increase the recovery factor up to 5% compared with typical steam injection. However, the oil upgrading achieved in the continuous injection was lower than the one obtained in the core-flooding tests. The numerical model presented in this work, which includes a dynamic nanoparticle retention model, changes on relative permeability alteration due to nanoparticle surface deposition, and a suited kinetic-thermodynamic representation, is able to describe correctly the most relevant phenomena observed during nanocatalysts in-situ upgrading process.