Lifting hydrofoils are used on modern sailing boats to lift the hull(s) out of the water, greatly reducing resistance and thus increasing the speed. Since limited budgets or competition rules often exclude the use of model testing and restrict the number of prototypes which can be constructed, accurate numerical simulation is crucial for naval architects to optimise the performance of these hydrofoils. To simplify routine simulations of complex flows in hydrodynamics, we are working towards automatic computations where accuracy and reliability are obtained, with almost no user intervention, thanks to adaptive mesh refinement. These simulations combine adaptive anisotropic refinement and derefinement of hexahedral meshes, local mesh deformation to keep the mesh conformal with the geometry, and physics-based guidelines to choose the simulation parameters in an automatic way. Recently, we have established such a procedure for ship hulls. This paper presents a similar computational approach for lifting hydrofoils. Apart from their importance for naval architects, the interest of studying hydrofoils is: * Carbon-fiber foils are flexible and their deformation should be taken into account for hydrodynamic performance calculations. We combine fluid-structure interaction and radial-basis-function based mesh deformation to model the foil deformations, with adaptive refinement. * Modern foils have intricate curved planforms, which make the generation of good quality hexahedral meshes a challenge. We show that a combination of body-following original grids and local refinement leads to efficient meshes and high-precision simulations. * Simply speaking, a hydrofoil is not a ship. If automatic setup procedures can be produced for two types of geometries which have such different flow physics, then it is likely that similar procedures can be found for all classes of hydrodynamic simulations which are typically performed in industry. We demonstrate the efficiency of our procedure by simulating several different hydrofoils and comparing the results with other simulation methods and experimental data.