8.1 Recommended BC Arrangements

This section discusses the recommended boundary condition arrangements for downstream, inflow and internal boundaries, summarised in Table 8.1.

The recommended arrangements of downstream boundaries are:

• Where the downstream boundary is a well-defined water level (e.g. the ocean or a large lake), water level boundaries are typically specified using 1D or 2D HT (water level vs time) boundaries. Note that both outflow and inflow may occur with water level boundaries.
  
• Where the downstream boundary is intended to represent a “reasonable continuation” of a river, a stage-discharge relationship may be specified using a 1D or 2D HQ (water level vs flow, also commonly referred to as a stage-discharge) boundary. Note that real rivers may exhibit a hysteresis in the stage-discharge relationship (sometimes termed a “looped rating curve”). When this occurs the discharge at a particular water level may be different at different stages in the flood, for example a higher flow may occur on the rising limb compared to the same gauge level on the falling limb. A single height-flow (HQ) curve is an approximation of flood behaviour. Sensitivity testing of any downstream HQ boundaries is recommended, if results in the area of interest are sensitive to changes in HQ curve, consider moving the downstream boundary further from the area of interest.
  
• In some situations, a hydraulic structure that is inlet controlled acts as the downstream control, in which case, the boundary specified downstream of the structure will have no influence on the results upstream of the structure. For such cases, a downstream water level (HT) boundary may be used with a water level defined below the ground level. In TUFLOW Classic, with a downstream boundary where a water level has been specified below the ground level, for a positive bed slope (based on slope from upstream cell centre to cell face), upstream friction based flow (which may be supercritical) is assumed based on the bed slope, whilst for negative (adverse) slope, weir flow is assumed. For TUFLOW HPC, with a downstream boundary with a defined water level below ground level, it becomes a ‘weir’ calculation - similar to a broad-crested weir. It is possible that subcritical flow just upstream of the boundary may become critical or supercritical at the face where the weir is. Flow that is already supercritical will remain supercritical.

The recommended arrangements of inflow/internal boundaries are:

• For fluvial flood (river/streams breaking bank) models it is common to use an inflow boundary (flow vs time or QT). However, in some situations upstream levels may be known more accurately than flows (level gauges are common whereas cross-section integrated flow measurements are rare) and a water level boundary (HT) may be used instead - particularly during model calibration.
  
• For 2D pluvial (rainfall exceeding drainage capacity) flood models, inflows may also be defined using 2D flow sources and sinks such as a Rainfall (RF) or SA boundaries.
  
• For models where the rate of inflow changes dramatically, such as in a dam-break scenario, a QT boundary may have instability issues, in which case the flow vs time inflow may be defined using a area based Source-Area (SA) boundary. SA boundaries are typically more stable than QT boundaries. Alternatively the dam break can be modelled via representation of the storage upstream of the dam within the model with the failure being represented via either a dynamic 1D dambreak channel (DF or PF type channel - refer Section 5.9.8) or 2D via variable geometry (variable 2D z shape) - refer Section 7.2.5.3.

For both 1D and 2D models, it is critically important that the inflow and downstream boundary placement location does not affect the model results within the primary area of interest. It is an essential step during the early stages of model creation to sensitivity check the location of the boundary to confirm its location does not cause a material difference in water levels within the region of interest for the modelling. If a difference in result does occur, greater distance between the area of interest and the boundary location should be considered.

With regard to 2D boundaries, particular attention should be given to the placement of water level (HQ and HT) and flow (QT) boundaries. They should be:

• Digitised as line objects approximately perpendicular to the flow direction. The boundary can be digitised at any orientation to the 2D grid (i.e. does not need to be parallel to the grid) and can be defined using a line with multiple vertices and segments (i.e. it does not have to be a straight line). The reason these boundaries should be digitised perpendicular to flow is that the boundary, depending on type and options (refer Section 8.4.1), may force a uniform water level along the boundary cells. Spurious circulations can occur along the boundary if it is digitised along (as opposed to across) a waterway. This is often noticed by viewing the flow velocity vectors in the vicinity of the boundary or by an oscillating delta volume (dV) value in the output window (refer to Section 14.1).
  
• Placed at the edge of the model active area boundary. To achieve this, snap the line vertices to those of the code polygon that defines the model active area. If the boundary is placed inside of the active area, bi-directional and/or recirculating flows are likely to occur. If the boundary line is significantly outside the active code polygon, then flow may not be able to reach the boundary and leave the model.
  
• Have end points at sufficient elevation so that water levels at the boundary never reach the ends of the line.
  
• Specifically for the stage-discharge (HQ) boundary, the boundary should only span one primary flow path. If the stage-discharge relationship is to be auto-calculated, the boundary end points should have elevations that are only just a bit above the highest expected water level, alternative the ‘d’ attribute can be used to set the maximum depth, see Table 8.6. This is to ensure that the digitised stage-discharge table has sufficient resolution over the actual working range of the boundary. Further, HQ curves generated for very shallow surface elevation slopes may cause model stability challenges.

A good check for the boundary locations is to compare peak flood extent (e.g. maximum water level grid) after a simulation with the active model area and associated boundary locations; typically water should not touch the edge of the model unless a boundary is defined. If water does occur at the edge of the domain, a process called ‘Glass-Walling’, boundaries of the active model area may need to be reviewed and added as appropriate.

Table 8.1 provides an overview of the boundary types typically used for different locations.

Table 8.1: Recommended BC Arrangements

Type of Boundary	1D	2D
Ocean or Estuary	HT	HT
Lake	HT	HT
River/Creek Outflow	HQ or HT	HQ or HT
River/Creek Inflow	QT as point on inflow node.	QT line (preferred) or SA region
Local Catchment Inflows around the edge of model or within model.	QT as regions (flow is distributed between nodes within a region)	SA. Use SA with PITS option for directing inflows directly to 1d pits (also referred to as gully traps).
Direct Rainfall (No Hydrology Inflows)	No option at present	RF or possibly SA (with RF option).
Dambreak Hydrograph	QT	SA or QT. SA may offer greater stability and better mass error if mass errors occur using QT.
Pumps	P (Pump) channel	1D linked via SX preferred. SH possible
Infiltration	No option at present other than specifying a negative QT.	Specify rainfall losses on a materail basis (Section 7.2.6.4) or use the soil infiltration feature (Section 7.2.7). Alternatively, RF can be used by specifying negative values.
Groundwater Level	HT	2D HT if on the surface or specify a GT level when using Interflow functionality for sub-surface groundwater level.