By Neil Manewell
For years, groundwater modellers have been plagued with the task of making a groundwater model run efficiently, whilst maintaining sufficient resolution to predict groundwater impacts as accurately as possible. With classic MODFLOW, our only saving grace was grid telescoping, whereby we reduce the cell size until we gradually narrowed down to our intended grid resolution. This was generally restricted to a factor of 10% of the largest cell in the model, and couldn’t be reduced greater than 1.5 x smaller than the neighbouring cell. Model layers had to be continuous, meaning the modeller would have to ‘wrap’ the layers so that a minimum layer thickness was achieved across the model domain.
Finite-element mesh designs were therefore more favourable, particularly when groundwater interactions between the surface water and groundwater systems were high and required a high level of detail. Only the problem was the finite-element method often resulted in slow-runtimes due to high element counts, large water balance errors, and the requirement for continuous layers across the model domain.
MODFLOW-USG is the latest derivative of a long line of MODFLOW releases and represents a breakthrough for modelling efficiency. The code is essentially unrestricted in terms of layer presence, cell resolution changes, cell shape, and cell connectivity. A finite-element style mesh can be used to easily increase resolution where required, e.g. along rivers/creeks/mining areas. Voronoi meshes further reduce cell count by effectively halving the number of cells compared to a triangular mesh. Layers can be ‘pinched’ out, which is particularly useful in coal basins, where a number of units terminate at the surface.
We are currently converting many of our older models to MODFLOW-USG to make them faster, better represent connected and disconnected geological units and more suited to uncertainty analysis.
Enquiries can be directed to [email protected]