A dynamic model of a chemical CO2 absorption process with aqueous monoethanolamine (MEA) is presented, validated against experimental data. Based on the validated model, a reduced-order model is developed, suitable for an online optimization control strategy. The objective of the optimization is to enable fast adaptations to changes in operating conditions of the power plant, while minimizing the energy consumption in the operation of the CO2 separation plant. The results indicate that model-based online optimization is a feasible technology for control of CO2 separation systems.