16.3 QA Check List and Simulation Logging

A quality assurance check of any model used for a project is recommended, even if initial simulation appears to be trouble free and healthy. Table 16.2 presents a helpful shortlist of general quality control checks. This list is not exhaustive, a more detailed list is available from Modelling Log and Review Template. Model review responsibility should be assigned to an experienced modeller, not someone who is not familiar with TUFLOW modelling.

Table 16.2: Quality Control Check List

Item	Description	Checked
Modelling Log	A modelling log is highly recommended and should be a requirement on all projects. The log may be in Excel, Word or other suitable software. A review of the modelling log is to be made by an experienced modeller. It should contain: \(\cdot\) the TUFLOW executable version used for the modelling; \(\cdot\) Sufficient information to record model versions during development and calibration; \(\cdot\) Observations from simulations; \(\cdot\) Key modelling assumptions; and \(\cdot\) A list of data sources used. A model version naming and numbering system needs to be designed prior to the modelling. The version numbering system should be reflected in input data filenames to allow traceability and the ability to reproduce an old simulation if needed.
File Naming, Structure and Management	A review of the data file management should check: \(\cdot\) Files are named using a logical and appropriate convention that allows easy interpretation of file purpose and content; \(\cdot\) A logical and appropriate system of folders is used to store the files; \(\cdot\) Relative path names are used for input files (e.g. “..\model\geometry.tgc”) so that models are easily moved from one folder to another. \(\cdot\) Documentation of the above in the project Quality Control Document and/or Modelling Log.
2D Cell Size and 1D resolution	Check whether the 2D cell size is appropriate to reproduce the topography needed to satisfactorily meet the objectives of the study (see Section 3.2), and that the 1D spatial resolution is appropriate to reproduce the water longitudinal surface gradient.
Topography	The topography review should focus on: \(\cdot\) Correct interrogation of Digital Terrain Model (DTM); \(\cdot\) Correct datum; \(\cdot\) Modifications to the base data (e.g. breaklines) have been checked. Regarding the latter, can be carried out by reviewing the _DEM_Z check file (see Table 14.3). Note: Reviewing the elevations in the .2dm file is not appropriate as only the ZH Zpt is represented in the .2dm file (the ZH elevation is not used in the hydrodynamic calculations). 1D cross-section locations and conveyance should be reviewed. As a general rule, conveyance should steadily increase downstream. Sudden changes in conveyance need to be cross-checked (these are often easily identified by sudden changes in velocities of successive channels). Are hydraulic controls such as levees, roads and embankments represented in the model?
Bed Resistance Values	Bed resistance values are to be reviewed by an experienced modeller. The review should focus on checking: \(\cdot\) The DEM_M check file; \(\cdot\) The Manning_n attribute values in the uvpt_check file layer created by Write Check Files; \(\cdot\) The Material attribute values in the grd_check file created by Write Check Files; or The reviewer should be looking for: \(\cdot\) Relative consistency between different land-use (material) types; and \(\cdot\) Values are within accepted calibration values. GIS thematic mapping is an excellent way to visually and quickly review the variation in bed resistance and other parameter values.
Calibration / Validation	Check that the model calibration or validation is satisfactory in regard to the study objectives. Identify any limitations or areas of potential uncertainty that should be noted when interpreting the study outcomes.
Mass Conservation	In addition to the mass balance reporting (see Section 14.7), it is good practice to carry out independent mass checks. Standard practice is to place 2d_po flow lines (see Section 11.3.2.1) in several locations throughout the model. They are typically aligned roughly perpendicular to the flow direction. The locations should include lines just inside each of the boundaries. Other suitable locations are upstream and downstream of key structures, through structures and areas of particular interest. The flows are graphed and conservation of mass checked (i.e. the amount of water entering the model equals the amount leaving allowing for any retention of water in the model). Ensure that the flows from any 1D channels crossed by a 2d_po line are also included in the mass check, and that the 2d_po flow lines are digitised so that they cross the 1D channel where there is a change in colour of the linked 2D HX cells as shown in the 1d_to_2d_check files or _TSMB1d2d layers. In dynamic simulations, an exact match between upstream and downstream will not occur due to retention of water, however, examination of the flow lines should reflect this phenomenon. For steady-state simulations, demonstration of reaching steady flow conditions is demonstrated when the flow entering the model equals the flow leaving the model.
Hydraulic Structures	Head losses through a structure need to be validated through: \(\cdot\) Calibration to recorded information (if available). \(\cdot\) Desktop calculations based on theory and/or standard publications (e.g. Hydraulics of Bridge Waterways). \(\cdot\) Cross-checking results using other hydraulic software. Simple checks can be made by calculating the number of dynamic head losses that occur and checking that this is in accordance with what is expected (see Section 7.2.8). It is important to note that contraction and expansion losses associated with structures are modelled very differently in 1D and 2D schemes. 1D schemes rely on applying form loss coefficients, as they cannot simulate the horizontal or vertical changes in velocity direction and speed. 2D schemes model these horizontal changes and do not require the introduction of form losses to the same extent as that required for 1D schemes. However, 2D schemes do not model losses in the vertical or fine-scale horizontal effects (such as around a bridge pier) and may require the introduction of additional form losses. See Syme (2001b) for further details.
Eddy Viscosity	Check that the eddy viscosity formulation and coefficient is appropriate (see Section 7.3.3).