Core+ DynamicFlood – Understanding the hydrodynamic simulation

TUFLOW HPC (Heavily Parallelized Compute) is a 2D fixed grid hydrodynamic solver that uses an explicit finite volume solution that is 2nd order in space and 4th order in time. TUFLOW HPC uses adaptive time stepping with the ability to revert back in time should a numerical inconsistency occur, thereby providing a high level of numerical stability. The solution solves the full 2D free-surface equations including the inertia and sub-grid turbulence (eddy viscosity) terms.

This document explains how the model is set up when running a TUFLOW computation from within SCALGO Live. For a full technical description of TUFLOW we refer to TUFLOW's manual. The SCALGO Live integration always uses the most recent TUFLOW release.

Buildings

The elevation model (DEM) used in the simulation is identical to the DEM in the workspace except for cells covered by building objects. 

Building cells are inactive in the computation and rainfall on buildings is therefore handled as a special case by computing the runoff from the buildings as a precomputation step before running TUFLOW. This runoff is then placed on the regular cells on the outside perimeter of the buildings. Thus from TUFLOW's perspective, there is extra rain on the perimeter of the building, corresponding to their runoff.
The runoff from the building cells is computed as usual from the building rainfall by applying the runoff function for the "building" land cover class. For the purposes of the runoff functions we currently assume that all buildings are connected to the drainage system. Note that we only support constant infiltration rate runoff functions for this class when running the hydrodynamic engine, so if you have selected a different type of runoff function, you will get an error when you start the computation.

To find out which perimeter cells receive water from the building roof, we use a flat routing heuristic to route across the building footprint surface, similar to how water is routed across depressions in our depression-free flow analysis.  There is no guarantee that building rainfall is distributed evenly on the perimeter cells.  Some cells receive a large amount of the rainfall while others receive none or very little for certain building footprint shapes. For example, this can happen for concave building footprints.

The boundary condition applied to the boundaries between the regular DEM and the removed buildings is set to no crossing. This entails that the building perimeters function like infinitely high walls.

Workspace boundary

The boundary condition applied to the outer boundary of the workspace as well as to any coastline within the workspace is of the type HT (=Head versus Time), where H*T is set to zero. This entails that water flowing into a model cell  touching this boundary is immediately removed from the model.

Infiltration and drainage system

Rainfall on any cell in the model is turned into runoff at each time step after infiltration is subtracted. We apply two distinct infiltration schemes based on the type of land cover in the cell.

For land cover types in the natural group (e.g. bare land, shallow/dense vegetation), some rainfall is assumed to infiltrate into the soil, and this is simulated using Hortons infiltration equation. The Horton parameters are set based on the soil type and the expected degree of compaction of the soil, which is determined based on the land cover type.

For land cover types in the artificial group (e.g. paved areas, buildings), rainfall is assumed to run off the surface and into a nearby inlet to a drainage system, if there is any. To determine whether an artificial surface is connected to a drainage system or not, we use different data sources in different countries, see table below. Generally, we assume all artificial surfaces in urban areas to be connected to a drainage system, and all artificial surfaces outside urban areas to not be connected to a drainage system. If the artificial surface is not connected to a drainage system, we assume almost 100% runoff and set the runoff to 1mm/hr. We do not set this value to 0mm/hr to avoid stability issues with the model. If the artificial surface is connected to a drainage system, we represent the drainage capacity of the system using a constant potential infiltration rate, set based on expert advice regarding the typical capacity of drainage systems in the country, see the table below.

Country

Dataset (source)

Potential infiltration rate (mm/hr)

Denmark

Kloakoplande (plandata.dk)

36

Norway

tettsteder (Statistics Norway)

20

Sweden

Tätorter (SCB)

36

Finland

YKRKaupunkiseutu2021: Keskustaajama & Lahitaajama

20

Table: Sources of data used to determine whether artificial surfaces are connected to a drainage system, and which potential infiltration is used to simulate the capacity of the system.

Surface roughness

Roughness, in the form of Mannings n, is set according to the land cover type, as specified in the table below. To learn more about the land cover map, see the specific documentation for your country (e.g. Denmark, Sweden, Norway). Note that we have not yet released our land cover map in all countries, please contact us if you have questions about availability.

Land cover type

Mannings

Notes

Code

Name

M

1

Bare land

5

2

Water

50

3

Other paved

40

4

Snow-ice

10

6

Shallow vegetation

5

 

7

Dense vegetation

2

 

8

Farmland

10

9

Paved road

50


10

Unpaved road

40

 

15

Bare rock

10

 

16

Building

50

Not used 

17

Unclassified

5

 

Table: Roughness parameters (Mannings M) for all land cover types.

Soil parameters

The hydrodynamic setup carefully uses soil type information alongside the land cover map to setup soil infiltration in TUFLOW. We use the Horton infiltration equation to model infiltration and the parameters used can be found in the dedication section about soil parameters.

Culverts

Culverts are simulated using TUFLOWs 1D model, also known as ESTRY. You can create culverts using the culvert object tool inside your workspace. Culverts are always circular (ESTRY type=C). You can set a culvert diameter in the workspace, however if this diameter is not specified, we use a fallback diameter of 3 m.  Note that hydrological corrections imported unto the workspace from the national analyses are treated just like any other culvert. We do not provide diameters for hydrological corrections, but you can specify the diameter yourself in the workspace.

The length of the culvert in ESTRY is set equal to the length of the polyline defining it in the workspace. Any bends along the line have no impact in the simulation except for defining the culvert’s length in ESTRY. The invert levels upstream and downstream are set equal to the level of the terrain in the grid cells that the line starts and ends inside, plus 2 mm. The Mannings n inside the culvert is set to 0.02 (equivalent to Mannings M of 50). The entry loss coefficient is set to 0.5 and the exit loss coefficient is set to 1.

The flow model through culverts is described in the TUFLOW manual, section 5.7.1 (Culverts and Pipes) and the differential equations in section 5.3 (Solution Scheme). Throughout the simulation, the flow through the culvert can switch between different flow regimes, described in Figures 5-1 and 5-2.

The ESTRY culvert properties configured by SCALGO Live are summarized in the following table.

LengthLength of 2d polyline (including bends, if any)
Upstream invertTerrain elevation at higher endpoint plus 2 mm
Downstream invertTerrain elevation at lower endpoint plus 2 mm
Manning's n0.02
Bend loss0
Entrance loss0.5
Exit loss1
Number of barrels1

Table: ESTRY culvert properties used by SCALGO Live.

The 1D model of the culvert is connected to the 2D surface model using an SX flow boundary. This means that water level in the 1D node is determined based on the average water level along the 2D cells that are directly involved in the connection, called 2D SX cells. Conversely, water level in the 2D SX cell(s) is determined based on the flow from the 1D node. Flow is proportioned via depth if multiple SX cells are connected to a single node. The number of 2D cells directly involved in the 1D2D-connections is determined by TUFLOW (using TUFLOW’s “Sag” approach).

Area subsurface structures are converted to a series of adjacent pipes of the above type with the same fallback diameter before the simulation is run.