It is well established that the centrifuge can be an important tool to determine permeabilities of porous materials. Nevertheless, up to this date, for geotechnical applications, they are not used outside research laboratories. We attribute this to the cost of a centrifuge, the apparent requirement of measurements in the sample and the more complicated handling of samples for use in the centrifuge. As a first step towards an affordable centrifuge, we present an adaptation to a common bench-scale centrifuge, which allows to obtain measurements in a fraction of the time needed by standard infiltration methods, especially for clayey soils. The permeabilities are determined by measuring global transient data only: water level in an inflow or outflow chamber and gravitational centre of the sample or the rotational momentum of the sample. These are all measured outside of the centrifuge in a start-stop regime. A method of lines approach to solve the Richards' equation is used, combined with interface tracking, which was previously shown to be an accurate method. Permeabilities are assumed to behave according to the van Genuchten-Mualem formula. We present the experimental set-up, accuracy of the measurements and the numerical solution method. For saturated flow, we compare with standard tests, while for unsaturated flow we compare with a pressure plate result (ASTM D2325). The results indicate the strengths and limitations of this approach, but overall we can conclude that measuring the global characteristics during rotation should provide accurate permeabilities.