The increasing population and the consequential demand for dwelling and recreation areas lead to more frequent conflicts between humans and natural hazards. In mountainous regions snow avalanches are, therefore, still a major threat to humans and infrastructure, with a significant impact on the economy and tourism. For the development of design criteria for infrastructure it is crucial to obtain a thorough understanding on the pressure exerted by avalanches, so that they can withstand avalanche impact. Although the differences between avalanche flow regimes reportedly play a crucial role for the avalanche-obstacle interaction, to date the impact pressure is often calculated similarly to the dynamic pressure in inviscid fluids proportional to velocity square and using empirical drag coefficients. Indeed, in the inertial flow regime, which is typical of powder avalanches, the impact pressure is proportional to velocity square. However, in the gravitational regime, which is typical of wet avalanches, the pressure is proportional to the flow depth. The empirical proportionality factor in the gravitational regime is referred to as the amplification factor. Field measurements indicate that the amplification factor and the drag coefficient may range within considerable boundaries. Thus, in the absence of a physics-based framework to make the crucial choice of the drag coefficient and the amplification factor for the impact pressure calculation, engineers need vast knowledge and experience in constructing in avalanche terrain and snow avalanche dynamics. Even for experienced experts it is often unclear how to calculate the impact pressure adequately according to the expected avalanche flow regime or how to consider the obstacle geometry in the calculation. The aim of this project is to develop a physics-based framework for the calculation of avalanche pressure on obstacles. In particular, we want to evaluate drag coefficients and amplification factors as a function of snow properties and avalanche flow regimes. To reach this goal we develop a numerical Discrete Element Method model to investigate the interaction of avalanche flows and obstacles, using a cohesive bond contact law. We test the relevance of the model by comparing simulated impact pressures with field measurements from the Vallée de la Sionne experimental site. By varying avalanche flow velocity and cohesion in the simulations, we show that the impact pressure can be interpreted as the superposition of an inertial, a frictional and a cohesive contribution. Further, we find a novel scaling law, reducing the problem of calculating the pressure induced by cohesive flows, to the calculation of cohesionless flows. We provide evidence that in the cohesionless case the compression inside the influenced flow domain around the obstacle, the mobilized domain, governs the impact pressure of granular flows in the gravitational regime. If the cohesion is high, we find that the cohesive bonds further enhance the stress transmission in the compressed mobilized domain, leading to an increase in impact pressure. Considering an inertial and a gravitational contribution, we quantitatively link the properties of the mobilized domain to the pressure. Finally, the knowledge from previous research and the findings of this thesis allow us to propose a physics-based framework to estimate the impact pressure by applying simple geometrical considerations and fundamental avalanche flow characteristics.