Wave impact tests are performed in the LNG industry to design the structure of the cargo containment system (CCS) of the vessels and to study the physics of breaking wave impacts. The state-of-the-art methodology divides the wave impact flow into two parts: the global flow and the local flow. The global flow considered as the solution of the incompressible Euler equations for the liquid and gas in the tank; and the local flow as a perturbation of the global flow. However, researchers have found it to be a challenge to generate repeatable global flows of focused breaking waves and, therefore, to draw definitive conclusion about the contribution of the global flow variability in the impact pressure variability. In this context, a new facility was designed and built at MARIN in the framework of the SLING project to further investigate the physics of sloshing impacts: the Multiphase Wave Lab (MWL), a wave flume of 10 m in length, 0.6 m in width and water depth of 400 mm. This thesis aimed to identify and evaluate the sources of variability of the global flow of focused waves, to define the repeatability criteria and to determine the theoretical conditions that would lead to global flow repeatability in the MWL. To achieve the objectives, both theoretical and experimental work have been required. Three main sources of global flow variability were identified: (1) water depth variation, (2) long bounded waves (seiching) and (3) currents induced by seiching. These drivers of variability were modeled in a wave generation and propagation algorithm from which the sensitivity of the global flow to the sources of variability has been addressed. The results showed that the water depth is the most critical driver of variability and that repeatability would be achieved if its difference between experiments is below 0.5 mm. Based on the sensitivity study and the characteristics of the wave maker, a criterion to experimentally quantify global flow repeatability was derived from the Sobolev norm of the Fourier space of free surface elevations at a distance from the focal point. To validate the theoretical value, impact waves were generated using a wave focusing technique. Linear wave and wave making theories were used to compute the paddle motion generating the wave impact of interest. The breaking wave was designed using a two-parameter Ricker amplitude spectrum formulation, which defines the contribution of each frequency to the total breaking wave energy and therefore changing its characteristics (crest thickness, crest stability, gas pocket size…). Image processing techniques were used to measure wave maker motion and free surface elevations from video recordings. While more repetitions are required to confidently conclude about the validity of the criterion, the experimental results showed that when the ‘dissimilarity’ value was below the theoretical threshold, exceptional repeatability of the global flow was obtained.