Infant mortality is a significant problem in mooring design. The objective of this thesis is to design a generic fault detection system for mooring line failures, which can accurately detect a mooring line failure within a certain time frame. The research is done through Catenary Arm Leg Mooring (CALM) buoy systems, as this thesis is a Proof of Concept. Two types of mooring line failures are considered in the XZ-plane, a top segment failure close to the buoy and a bottom segment failure close to the anchor. Both failures characterize themselves by a constant offset in equilibrium position and a changed mooring stiffness. Using theoretic concepts about mooring configurations and waves, it is shown that the shift in equilibrium position has most potential for detecting a mooring line failure. These analysis are verified by hydrodynamic simulations in OrcaFlex. For the bottom segment failure the breaking stage cannot be predicted, as this stage depends on the mooring design, the external loads, and the location of the failure. The focus must lie on what stage the failure becomes detectable, rather than how long it takes. The fault detection algorithm is using a model-based approach for fault detection as a general framework which can be used for any variation in mooring design. Current flows, tankers mooring at the buoy, and non-linear effects of mooring configurations in waves cause a constant offset in position as well. By including measurements of the current flow velocity and the tanker hawser tension as inputs of the CALM buoy model, the fault detection algorithm can distinguish between the sources of the constant offset. Additionally, the non-linear effects of a mooring configuration in waves are for small wave heights not significant and not included in the modeling of the buoy. The mooring tensions and system parameters are introduced as specific parameters per mooring design. Via an Unknown Input Observer (UIO), the two unknown input loads are estimated: the wave load and the residual force. The residual force is chosen to represent the mooring line failure in the model. When there is no mooring line failure this force should be zero. The two unknown input forces are estimated by decoupling them in the frequency domain. Using a shaping filter the wave forces are restricted by a predefined frequency bandwidth. The non-linear effects of a mooring configuration in waves are determining the thresholds for the fault detection algorithm. Therefore, no wave load can be mistaken for a mooring line failure. The load cases show that for small offsets the fault detection algorithm is capable of accurately detecting the mooring line failures. The two unknown forces are decoupled in a way that the residual force can be used as residual for fault detection. The residual signal in RY-direction can be used for fault detection. Additionally, it is shown that no step in position data is required for the algorithm to detect a failure. The answer on the research question, is yes. It is possible to accurately detect to accurately detect a mooring line failure. Since this thesis is only applied for the three Degrees of Freedom (DoF) in the XZ-plane, a six DoF model needs to be investigated in future research. Furthermore, by including the non-linear effects in the modeling of the CALM buoy, higher wave amplitudes could be considered. Another possible way to deal with these non-linear effects is to use adaptive thresholds.