Richards Bay Port, located in the East Coast of South Africa, was built during the 1970s. Two rubble mound breakwaters were constructed to protect the deep-water entrance channel and create a sheltered area for the vessels. Since the completion of these breakwaters in 1976, they have withstood several major storms, including cyclones that have caused significant damage to the dolos armour layers. To restore their functionality, two major reparations were carried out in 1976 and 1996, respectively. In addition, a severe storm that occurred in March 2007 caused relevant damages to the breakwaters of Richards Bay Port. Their damage level was established after the survey conducted in May 2007. This survey concluded that most of the breakwaters sections had an intermediate damage, except from the South Breakwater’s roundhead, which was in failure and it required urgent repairs. Since then provisional measures have been adopted to avoid the spread of damage along the breakwater while new repair works are designed. The main objective of this thesis was to determine the most suitable design for the repair works that should be applied in the roundhead of the South breakwater at Richards Bay Port through a Quasi Three-Dimensional (3D) model testing. This was achieved by reproducing the observed damage at the structure’s roundhead in one of CSIR’s hydraulic laboratory flumes and testing three repair alternatives. These repair alternatives consisted of covering the damaged structure with new armour units. Dolos, Core-Loc and antifer cubes were the armour units used in this research. The wave basin used to conduct this research had a length of 32m, a width of 4m and an available height of 1m. A transitional slope of 1:15 that extends about 4.5m long was built inside the basin to connect the deep water with the shallower water close to Richards Bay Port. Thereafter, the seabed profile corresponding to the South East direction was constructed along the next 20m of the basin. The structure was placed at a distance of 26m from the wavemaker. Graded gravel was used to construct the core, underlayer and toe protection of the roundhead, with a nominal size of 4.2g, 4.8g and 12.2g, respectively. The existing armour layer was built using dolos of 68g and gravel that represented the broken pieces. Above this damaged armour layer, the new armour units were placed with a nominal size of 82g for the dolos, 102g for the Core-Loc and 100g for the antifer cubes. The new armour units were placed trying to replicate the placement conditions at the roundhead. A total of 8 to 9 tests were conducted per armour unit. Five wave conditions were tested with increasing significant wave heights varying from 7cm to 18cm. Two water levels were set up per wave condition (High Water and Low Water). The tested wave conditions were generated with a JONSWAP spectrum and a duration that corresponded to a 1000 waves approaching the structure. Prior to and after each test, pictures were taken from three fixed positions perpendicular to the roundhead. These images were visually compared with the Armour Track software developed by CSIR to identify and quantify the movement of the armour units. This software is based on the superposition technique and it is useful to determine the stability of the structure. For each test, the stability number and the measured damage within the reference area were estimated. Generally the movements of the units occurred along the water line. However for higher wave heights (return periods of 20, 50 and 100 years), the waves overtopped and the damage started to concentrate in an area located between the angles 120 and 150 degrees from the direction of the incident wave, until failing with the overload condition. From these experiments it followed that the Core-Loc repair alternative does not perform as good as the other two options. Although all the repair options have difficulties to achieve the placement requirements at the roundhead, this phenomenon has a larger impact in the Core-Loc armour layer because it consisted of a single layer and any unit displacement resulted in failure of the structure. Therefore repairs should be undertaken more frequently, which leads to larger maintenance costs. The remaining repair options had a similar performance, even though the resistance mechanism of dolos and antifer cubes is different. The first one resists by the interlocking between the units, whereas the antifer cubes resist by their mass. Both are placed as double armour layers and thus some damage is allowed before carrying new repair works. The main difference between them is the actual feasibility to construct the units. The antifer cubes do not have any size restriction. Therefore heavier units can be manufactured without changing the stability of the unit. However, dolos have a size limitation because of its slenderness, and right now dolos heavier than 30-tonne cannot be built. Overall it could be concluded that the repair alternative consisting of antifer cubes is the one that should be applied at this particular location due to its performance and its construction feasibility.