Currently, an increasing amount of debris is floating in Low Earth Orbit (LEO) endangering space operations. To limit the contribution of non-operational satellites to the space debris problem a semi-controlled end-of-life de-orbit strategy can be employed. In this strategy the satellite is controlled from the end-of-life orbit down to low altitudes (120 to 200 km). From this altitude the satellite follows a ballistic trajectory towards the surface of the Earth. The target of such maneuver is to phase the impact probability track, with the Earth, such that the resulting impact track covers mostly uninhabited areas.

The objective of this thesis, is to provide insights into the effects of assumptions made in recent research of the semi-controlled de-orbit strategy. Furthermore, the extent to which the casualty risk can be reduced is investigated. Two optimization strategies are carried out. First, the control profile for the complete semi-controlled de-orbit strategy is optimized for minimum casualty risk using a differential evolution algorithm. Second, the initial state of the ballistic trajectory, at the point where the controls are turned off, is also optimized for minimum casualty risk using a differential evolution algorithm. In case similar impact tracks are achieved for both optimizations, future analysis of the semi-controlled de-orbit strategy can be performed which less computational effort.

Employing a semi-controlled low-thrust de-orbit strategy can reduce the casualty risk with 2 to 3 orders of magnitude with respect to an uncontrolled de-orbit strategy. Large ballistic coefficients are beneficial for the reduction of the casualty risk. Furthermore, orbits with high inclinations results in larger reductions of the casualty risk with respect to low inclination orbits. Also the results for low inclination orbits are less sensitive to errors in the re-entry state caused by imperfections in the GNC system of the satellite during the de-orbit maneuver. The lower sensitivity is the result of a better configuration of the impact track with respect to the large land masses on Earth. To find optimal impact tracks on the Earth, it is sufficient to optimize only the last 150 km of the trajectory. In case an accurate estimation for the value for the casualty risk is needed, higher fidelity models must be applied and the uncertainties in initial state at 150 km need to be investigated in more detail. The assumptions made in previous research influence the results with less than 33% and can easily be applied when an optimal impact track is sought for with low fidelity models. For investigating the range of re-entry conditions for which the resulting casualty risk is below the set requirement, some of the assumptions cannot be applied since it limits the solution space for the GNC system to target. Furthermore, in case an accurate value for the risk needs to be found, high-fidelity models are required and in that case, the effect if the assumptions is too large.