The following suggestions and recommendations are provided when troubleshooting a model. The list is not complete, but offers solutions to the more commonly found problems.
If the DOS window does not appear at all, check virtual memory congestion (see Section 4.7).
If the DOS window disappears for no apparent reason (this is now very rare!) first check the following:
You have sufficient disk space on the drive you are writing your results to and where the .tcf or .ecf files are located (this is where the .tlf or .elf files are being written to).
Your computer network is/was not down.
Check the water level to be used for detecting instabilities. If you have not allocated every Z point an elevation (the default Z elevation is 99999.) or if you have very high Z points in your geometry (relative to your water levels), this allows any instabilities to oscillate in a very large range. Consequently, the instability can become so extreme that floating point errors (ie. the computation is unresolvable) may occur before TUFLOW stops the simulation and declares it unstable. If this occurs, the DOS window disappears. However, in most cases there should be some water level exceedance warnings at the end of the .tlf file and/or negative depth warnings in the _messages.mif file. To remedy the situation use Instability Water Level to set a realistic maximum water level. This same effect can occur in 1D domains if the maximum height in a node storage table or a channel cross-section is very high or the Depth Limit Factor is set unrealistically high.
If the problem persists, please contact WBM (tuflow@wbmpl.com.au). There is no charge for fixing bugs of this nature!
If TUFLOW or ESTRY indicate that GIS objects are not snapped, not connected, could not be found or are outside the model domain, check that the relevant GIS layers are in the correct projection and that the objects are snapped to each other. A GIS can handle layers of different projections, however, TUFLOW and ESTRY require that all layers be in the same projection. This projection must be a Cartesian projection (not lat/long) and is specified using MI Projection (.ecf file) or MI Projection (.tcf file).
If you are having stability problems, check that the computational timestep is appropriate (see Section 3.6).
Discontinuous initial water levels, particularly at 2D/1D interfaces are a common source of instabilities. If the model is going unstable near a 2D/ID interface shortly after starting, check that the initial water levels in the 1D and 2D domains are similar.
It is not possible to specify a node as a flow boundary as well as being connected to a 2D SX boundary (which automatically applies a HX boundary to the node). This combination of flow boundary and water level (HX) boundary is incompatible. The result is that the HX boundary overrides the flow boundaries. An ERROR check for this occurrence was incorporated in Build 2003 03 AD.
Deep bends with “bumpy” topography may cause instabilities in 2D models. Smoothing the topography, rather than reducing the timestep is recommended.
Under-sized 1D node storage (NA tables) connected to 2D HX boundaries may cause instabilities near the 2D/1D interface. Over-sized storage attenuates the inflow hydrograph. As a rule-of-thumb, the node surface area should be similar to the width of the 2D/1D interface times 3 to 10 cell widths.
Irregular topography just inside a 2D boundary may cause instabilities. If problems occur, smooth the rough topography.
Use a timestep that divides neatly into 3600, ie. 1, 2, 3, 4, 5, 6, 7.5, 10, 12, 15, 20, 30, 45, 60, etc.
