Porous media, here defined as any permeable structure allowing a fluid to flow through, are relevant to a multitude of engineering applications and natural processes. The observed macroscopic properties of the porous media such as mixing, heat transfer and apparent permeability are properties which are affected by the flow and especially the type of flow, or flow region. The flow regions are characterized by the ratio of the convective to viscous forces, called the Reynolds number (Re). Of these regions the transition from inertial laminar flow to fully turbulent flow is the least understood. In comparison to flows in straight pipes the onset of inertial and unsteady phenomena in porous beds do not coincide, also the transition region stretches over orders of magnitude in Re for most porous beds. In porous media this domain is characterized by temporally long-lived and spatially large scale flow structures which interact in unpredictable ways leading to dramatic shifts of the behavior of the macroscopic properties. To improve the understanding of this transitional domain, ordered materials, that reduce geometrically induced flow complexities, are studied with both numerical and experimental methods.

In Paper A two types of ordered porous media with the same porosity but varying tortuosity are investigated using tomographic Particle Image Velocimetry and pressure measurements. The variation of Re gives an almost complete overview from the onset of inertial effects up to the start of the turbulent region. Two pore-scale phenomena were disclosed from the complex flow patterns that appeared. The first is an inertial steady effect first assumed to be caused by wall effects. In Paper D it was, however, discovered that the phenomenon materializes independently of wall effects. Instead it is a specific case of a more general inertial transition occurring for a wide range of porous media. A second pore-scale effect is a form of inertial core symmetry break-up that occurs in low-tortuosity porous media. This symmetry break-up is correlated to a sharp increase in the average pressure drop. The second flow structure was reproduced using numerical methods in Paper B forming the basis of a more comprehensive discussion on how these structures impact the usage of periodic conditions when modelling porous media.

The possibility of using high performance Graphics Processing Unit (GPU) implementations of the Lattice Boltzmann Method (LBM) for simulating thermal turbulent flows in porous media has also been investigated in Paper C. It is concluded that the GPU LBM implementations provide fast, efficient and accurate simulations of thermal turbulent flows in porous media, as well as for a wide range of other flows. Furthermore, in Paper E, a multiple GPU implementation of a hydrodynamic LBM model is presented.