Based on user feedback, training requests and reviews of published papers these new features have been very well received! We thank you for your support and look forward to working with you in 2021.
From Phillip Ryan, Greg Collecutt and the TUFLOW Classic/HPC Development Team
Quadtree is the division of one cell into four and then those cells into four and so on, thereby allowing a regular grid-based scheme to vary cell size across the 2D domain. The 2020 release offers this powerful feature as part of the 2D TUFLOW HPC solver.
In addition to the extra calculation resolution that quadtree provides, one of key strengths of this feature, unique to TUFLOW, is the ease of its implementation. It is exceptionally quick to create a quadtree mesh in a matter of minutes, as demonstrated in our “How to Implement Quadtree” video. Simply use one or more GIS polygons to define regions where you wish to refine the 2D cell size (or make it coarser). All other model inputs scale automatically to suit.
The benefits of quadtree are vast. For whole of catchment direct rainfall modelling, narrow streams can easily be defined using smaller cells. Key flow paths through urban areas, i.e. gutters, roads, and open channels, can be represented on a finer scale improving the hydraulic representation and conveyance, along with improved pit flow capture into the underlying pipe network. Areas well away from the key areas of interest can use a coarser resolution, greatly reducing run-times and memory footprints. Conversely, a development impact assessment can utilise a high resolution mesh relative to the surrounds when it’s beneficial to do so.
Sub-Grid Sampling (SGS) uses cell depth / volume and cell face depth / width and area curves to incorporate sub-2D cell terrain resolution. This provides much improved storage definition within cells and conveyance between cells, compared to the traditional approach of a single elevation for the volume definition and single cell face elevations.
Benchmarking has shown the benefits of SGS to be substantial for many types of applications.
Over the course of the past two years Dr Greg Collecutt and Dr Shuang Gao from the TUFLOW Team have been researching and testing new sub-grid turbulence models. Their work, published at the IAHR 10th Conference on Fluvial Hydraulics (River Flow 2020) in Delft (Collecutt et al, IAHR River Flow Delft 2020), produced a cell size independent turbulence scheme that benchmarks across all scales; a feature no other 2D solver provides.
Modellers can now confidently model at scales from sub-centimetre cells for a flume to tens of metres for a large river using the same default turbulence parameters - the existing most commonly used turbulence models such as Smagorinsky and/or Constant have a strong dependency on cell size and need to be calibrated for different scales or when the cell size becomes significantly less than the depth.
The new turbulence approach is pivotal for variable cell size meshes such as quadtree and our flexible mesh solver TUFLOW FV in 2D mode (3D requires a different approach).
TUFLOW HPC modellers can now reduce or vary cell sizes without seeing significant changes in results due to prior limitations associated with turbulence scheme assumptions, especially in situations where cell sizes are less than the flow depths. It is not an understatement that this research is a game-changer for 2D modelling.
For the 2020 release TUFLOW HPC supports modelling of non-Newtonian fluids. Utilise the exceptional numerical stability of HPC via computed adaptive time stepping to model non-Newtonian flow. Benchmarking against published tests has been carried out to confirm performance.
The user is required to specify yield stress, viscosity coefficient, shear thickening exponent, upper and lower viscosity, to model both shear thickening or shear thinning fluids. The 2D viscosity is computed as the derivative of the power law (HershelBuckley) viscosity model.
The TUFLOW 2020 release sees the expansion of advection-dispersion modelling from TUFLOW Classic into TUFLOW HPC, leveraging the speed and stability gains of HPC on GPU hardware.
This addition to TUFLOW HPC will pave the way for us to port across TUFLOW FV’s cohesive and non-cohesive sediment modules to the fixed grid product line.