Physical modelling occupies a space between two forms of numerical computer models that are used to simulate small geographical areas, and really large areas. Between them, there is a size and type of simulation – for example for breakwater stability across full port or harbour areas, or erosion and accretion of beach areas – where it is still more cost-effective and efficient to use physical models.
For large geographical areas – of the order of tens to thousands of kilometres – “large area” models, such as the TELEMAC suite, are used. These models have a resolution (the distance between points at which calculations are performed) of metres to kilometres, and are used for simulating, for example, the sediment transport along a large area of coastline, or the wave climate approaching a port.
These models are extremely useful for providing approximate results across a large area, but lack the fine resolution often required for structural design studies. Due to their size, these models must necessarily contain simplifying assumptions or equations for the models to run efficiently. This, coupled with the resolution of the model, often makes them less suitable to look at the interactions between, for example, waves and a breakwater, as the model is not sufficiently refined to allow the motion of individual rocks or armour units to be estimated.
For very detailed studies of small areas, CFD (Computational Fluid Dynamics) models such as OpenFOAM are used. They are ideal for small geographical areas – of the order of centimetres to metres, such as the end of a breakwater or a power station outflow. The resolution of CFD models is much greater than that of the “large area” models.
However, due to their complexity, CFD models are computationally expensive, and therefore not as well suited as physical models to simulating complex areas, such as scour protection around bridge piers or offshore wind turbines, in terms of cost and effectiveness.
Physical modelling also has several advantages over computational techniques:
- Physical modelling requires few simplifying assumptions, with no requirement to, for example, select a turbulence closure scheme or bed roughness formula, as nature takes it course;
- Immediate visual feedback is provided. There is no need to wait for a model run to complete to view results if damage occurs part way through a test;
- Uncertainties are reduced, as models are less reliant on empirical formulae which may need to be verified or extrapolated outside of their experimental range;
- Users have a high degree of control over the experiments. They are able to make quick changes between, or even during, tests to maximise benefits;
- These combined factors provide surety – clients can believe the results they obtain.