In the 19th Century, hydraulic engineering was typically addressed in two parallel approaches. On one hand, classical theoretical fluid mechanics with the use of analytical approaches to study various flow occurrences. Due to the difficulty and complexity of the phenomena under investigation and the lack of rational basis, hydraulic experimentation as the second approach played more and more an important role. Therefore, since more than one hundred years, physical experiments were relevant in science and engineering. With the rapid development of computational fluid mechanics and affordable high performance computing power, CFD is getting competitive with physical modelling that allows overcoming scale effect. Based on many years of experience in physical modelling at the LCH, some present and future needs and on-going changes observed are considered. Increasingly complex models are under investigation where regularly entire systems with dynamic behaviour or transient and unsteady flow discharge scenarios have to be considered. There is a trend towards larger all-inclusive models. These studies often exceed the essential hydraulic phenomena, and several parameters are included in one single model. These demands frequently lead to some suboptimal scale requirement. In these cases, and as there is also an increasing demand for hybrid numerical and physical modelling, the previously mentioned aspects can be avoided by treating each hydraulic phenomena at its proper scale and with the appropriate tool, either CFD or physical modelling. This allows reducing the physical model perimeter defining the boundary conditions from numerical modelling. Three examples are presented and discussed to highlight the potential of physical and numerical modelling. First rapid hydraulic transients due to the creation and collapse of a single cavitation bubble will be shown. Here the challenge comes from the observation and measurement techniques and time-space analysis, with acquisition frequencies up to 500 kHz for acoustic signals and 100 kHz for imaging and propagation velocities around the speed of sound in water. The second case of a complex surge tank with varying geometry. The capacity to physically model transient phenomena including a free surface is illustrated. The case of a simulated first rise due to mass oscillation between the upstream reservoir and the surge tank of a high head hydropower plant is presented. To terminate the study of air entrainment on a spillway chute is presented, where the limits of physical modelling due to scale effects are visible. In principle, there is no possibility to scale down the physics of air transport. Here numerical modelling could play a significant role in the future. As closing remark, besides the importance of hydro-engineering experimentation for the design process, and for the safe and efficient operation of present and future hydraulic structures and scheme, the author also emphasize the importance for fundamental research. Currently, the majority of scientific papers in peer-reviewed journals in the field of hydraulic research and engineering are still based on data that was collected in physical models and experimental installations.