On the use of a high-performance framework for efficient model coupling in hydroinformatics

Many situations in hydraulic engineering require the use of numerical modelling. Developments in this field began about 50 years ago and the resulting computational codes have now reached maturity. A current trend (not restricted to hydraulics) consists in global modelling and building Decision Support Systems. This necessitates taking into account a large number of phenomena and often presupposes coupling appropriate computational codes. Moreover, many models have been built in recent years but this requires much time and effort since whole rivers (and sometimes their tributaries) are often under consideration. So, when undertaking a new study, the reuse of existing models considerably reduces engineering time. The possible lack of information in an existing model can be compensated for by coupling it to another model, thereby giving access to the missing information. This study investigates the available techniques for coupling hydraulic models with different spatial dimensions (1D and 2D). An easy-to-use framework, called PALM, is compared to the other existing tools, and adopted for further study. The way to perform coupling of existing and relatively complex code in such a framework is detailed. We focuss on the reduced invasivity of PALM, which makes it very convenient for algorithm testing and prototyping. Results on real-world hydraulic cases are shown, in order to demonstrate the performance of this approach, compared to less accurate coupling, based on file exchange for example.