We use cookies to make your experience of our website better. You can delete or block cookies, but some parts of this website won't work without them. By using this website you accept our use of cookies.

31/10/2018 16:51:19

Improvements in remote sensing and computing have helped make it possible to run 2D flood simulations at finer resolutions. Running models at very fine resolutions is also something requested by many clients. However, doing so takes a significant amount of time, with a halving of the resolution typically causing the simulation run time to increase by approximately 8 times. Furthermore, models are being run at increasingly larger scales which also increases the time taken to run a simulation, particularly at finer resolutions.

Modellers are always looking for ways to speed up their simulation times without compromising on the accuracy and precision of their results. Members of the Flood Modeller team at Jacobs have been working with Alireza Shamkhalchian and Dr. Gustavo de Almeida at the University of Southampton to explore a new ‘sub-grid’ modelling approach that would allow modellers to run models at coarser resolution yet at the same time consider local scale topography at finer resolutions. By using two nested meshes, the new model can calculate the 2D shallow water equations on the coarser scale, whilst at the same time considering sub-grid scale topography and friction effects using an upscaling approach.

Initial results during testing using Environment Agency test cases have shown that running the sub-grid model at a 5 times coarser resolution can lead to simulations that run 50 times faster with little deterioration in model accuracy. Furthermore, using the sub-grid approach helps alleviate the drop in accuracy of traditional hydraulic models when run at coarser resolutions.

This behaviour is particularly evident in Figure 1, which shows the results obtained by solving Test 8A (see the Environment Agency’s published benchmarks 1) using both a traditional finite volume solver and the new sub-grid (also based on finite-volumes) solver. Test 8A models a shallow flow in an urban area; hence, sub-grid topographic information such as streets and buildings might affect considerably the resolved dynamics. As can be seen, when run at a relatively coarse resolution of 10m, the traditional finite-volume approach fails to reproduce the time behaviour of the water depth when coarse grids are employed (see green line), while the sub-grid solver performs well (see black line).

Graph.png (1)

Figure 1: time evolution of water level (point 1 in Test 8A) obtained using a traditional solver and the sub-grid solver.

As the research continues, the Flood Modeller team are continuing to investigate whether a sub-grid approach can be incorporated within a new Flood Modeller solver. To find out more about Jacobs’ work with universities or if you are interested in a future collaboration, please contact Dr. James Savage

 

References

1 S. Néelz and G. Pender (2013). Benchmarking the latest generation of 2D hydraulic modelling packages. Technical report, SC120002, Environment Agency.

Privacy & Terms

Copyright