Knowledge about the density of firn, that is covering the ice sheets, is not only essential for current estimates of mass loss in Greenland and Antarctica, but also to obtain information about the history of those ice sheets. Firn is the intermediate product between snow and ice, originating in areas where the annual accumulation of snow exceeds the amount of melt. Due to the ongoing accumulation of snow the firn densifies until it becomes ice. During this air gets trapped in bubbles within the firn. These air bubbles can be used to learn about past climate conditions on earth. But it is difficult to date the age of these bubbles. Due to circulation within the firn the trapped air can be several hundreds of years younger than the ice surrounding it. With help of firn densification modelling the age of the firn at pore close-off can be constrained. Another topic which requires firn densification modelling is the transfer from volume loss of the ice sheets to mass loss. Modern satellite systems allow us to measure the height change of ice sheets up to a very precise level. Such height changes can only be translated to mass changes if the density of an ice sheet is known. While the density of solid ice is constant, the density of the firn cover varies depending on past and current climate conditions. Only a relatively small amount of firn density measurements were conducted within the last decades compared to the size of the ice sheets. Therefore, it is necessary to simulate the process of firn densification. Attempts in modelling the firn density often follow a so called semi-empirical approach. A great deal of models using this approach but different parameter sets illustrates that there are numerous shortcomings. Therefore a model describing the processes leading to the densification of firn is investigated. The model is able to provide promising results, but using it within an established one dimensional framework uncovers problems. Simplifications regarding the stress acting on the firn have to be overcome to improve current firn density models.