Carbonate reservoirs contain more than half of the world’s conventional hydrocarbon resources. Hydrocarbon recovery in carbonates, however, is typically low, due to multi-scale geological heterogeneities that are a result of complex diagenetic, reactive, depositional and deformational processes. Improved Oil Recovery (IOR) and Enhanced Oil Recovery (EOR) methods are increasingly considered to maximise oil recovery and minimise field development costs. This is particularly important for carbonate reservoirs containing fractures networks, which can act as high permeability fluid flow pathways or impermeable barriers during interaction with the complex host rock matrix. In this thesis, three important contributions relating to EOR simulation and optimisation in fractured carbonate reservoirs are made using a high-resolution analogue reservoir model for the Arab D formation. First, a systematic approach is employed to investigate, analyse and increase understanding of the fundamental controls on fluid flow in heterogeneous carbonate systems using numerical well testing, secondary and tertiary recovery simulations. Secondly, the interplay between wettability, hysteresis and fracture-matrix exchange during combined CO2 EOR and sequestration is examined. Finally, data-driven surrogates, which construct an approximation of time-consuming numerical simulations, are used for rapid simulation and optimisation of EOR processes in fractured carbonate reservoirs while considering multiple geological uncertainty scenarios.