We use cookies to make your experience of our website better. You can delete or block cookies, but some parts of this website won't work without them. By using this website you accept our use of cookies.

Niamh O'Malley

05/02/2019 12:12:37

I have created a linked 1D-2D model. The 1D model runs without any problems (that I am aware of anyway). When I try to run the linked model the simulation screen opens but then the model doesn't initialise (i.e. nothing further happens even after leaving it open for an hour). The .xml file is the only file created. I have looked at the xml file and there is nothing obvious to me that is causing the problem. I have checked my link lines and the DTM as well and can't see where the problem is. Any suggestions would be appreciated.

Thanks

Niamh

Konrad

Flood Modeller Support

05/02/2019 12:24:08

Hi Niamh - is there a .log file written (usually same folder/name as the xml, just different extension) with any useful further info? Or indeed in the 1D .zzd file? 

I think there may be issues if you have a relative path specified for the 1D .dat file (what it becomes relative to may change, depending on whether it's 1D only or 1D-2D linked)

Regards,

Konrad

Bob

Flood Modeller Support

05/02/2019 12:24:45

Niamh

What version of the software are you using? If it's an older version then the old 2D simulation screen I don't think is compatible with Windows 10. If it's the Free version it could be the 2D model exceeds the limits for Free.

Also have you tried running the 2D in isolation to check that works. You could turn off the linking and just use your link line file as a boundary and assign a constant inflow to this. At least this would confirm the 2D solver can initialise for your model and the issue then lies in the link definition. Finally, in the link definition shapefile are the 1D nodes referenced correctly, i.e. check spellings and case of entries.

Hopefully some of these might help shed some light on the problem.

Regards

Bob

 

Niamh O'Malley

05/02/2019 12:26:48

Hi Konrad,

No the only file it creates it the xml file which hasn't happened to me before. If it would start and create some other files I might have been able to figure out where the problem is.

Should I have the 1D dat file in the same folder as the xml file then?

Thanks

Niamh

Niamh O'Malley

05/02/2019 12:40:25

Hi Bob

I am using Flood Modeller Free version 4.4. The number of 2D cells is 806896 so I thought that was below the limit? I have created 1D-2D models of that size in the past.

I'll try the 2D by itself.

I checked the link line attributes, they are in the correct order and the reference names are correct. There is one line missing in the attributes table(when I used the link line generator) but that is between a cross section and a replicate section immediately upstream of a bridge so I thought that would be ok? 

Thanks Niamh

 

PS I have attached the xml file text below if that helps?

 

<?xml version="1.0" standalone="true"?>

-<ISIS2Dproject description="" name="1% 2D SIM" xsi:schemaLocation="https://www.floodmodeller.com http://schema.floodmodeller.com/4.0/2d.xsd" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns="https://www.floodmodeller.com">


-<link1d weirml="0.9" weircd="1.2" link_to_model="ISIS1D">

<link>C:\Users\Orla1\Desktop\IE1757\GIS\LEFTV LINK LINES.shp</link>

<link>C:\Users\Orla1\Desktop\IE1757\GIS\RIGHT LINK LINES.shp</link>

<ief>C:\Users\Orla1\Desktop\IE1757\Q100 Runs\SIMS\1% UNSTEADY.ief</ief>

<mb frequency="60"/>

</link1d>

<!--content for domain 'Domain 1'-->

 

-<domain domain_id="Domain 1">


-<computational_area>

<xll>556782.0000000000</xll>

<yll>653238.0000000000</yll>

<dx>1.5</dx>

<nrows>1073</nrows>

<ncols>752</ncols>

<active_area>C:\Users\Orla1\Desktop\IE1757\GIS\ACTIVE AREA_ALL.shp</active_area>

<rotation>0</rotation>

</computational_area>

<topography>C:\Users\Orla1\Desktop\IE1757\GIS\DTM_1.asc</topography>


-<time>

<timezero_datetime use_real_dates="false">00:00:00 01/01/2000</timezero_datetime>

<start_offset unit="hour">0.000000</start_offset>

<total unit="hour">192</total>

</time>


-<run_data>

<time_step>2</time_step>

<scheme>ADI</scheme>

<pulsedepth>0.0001</pulsedepth>

<negdepththreshold>-0.001</negdepththreshold>

<velocityheadthreshold>0.5</velocityheadthreshold>

<frictionincreasedepth>0.1</frictionincreasedepth>

</run_data>

<roughness type="global" law="manning">0.035</roughness>


-<output_results>


-<hazard_debris_factor_method>

<static factor="1.0"/>

</hazard_debris_factor_method>


-<output format="SMS9" output_id="output-Domain 1" expand="0">

<variables>Depth Elevation Flow Velocity</variables>

<frequency>3600</frequency>

</output>

<massfile frequency="2">C:\Users\Orla1\Desktop\IE1757\Q100 Runs\1% 2D SIMDomain 1.csv</massfile>

</output_results>

</domain>


-<advanced_options>

<spatial_diagnostics>on</spatial_diagnostics>

</advanced_options>

</ISIS2Dproject>

Bob

Flood Modeller Support

05/02/2019 14:06:52

Hi Niamh

The limit for 2D cells in Flood Modeller Free is 100,000, so I can imagine your 800,000 cell model is going to be a problem for a start. Also if the link line generator has produced a shapefile with a blank field entry in it then I would recommend manually editing that before using it. The link line generator is a tool to help create link lines, but it is not infallible. The output should always be checked and may require further manual editing to produce exactly what is needed for a particular model.

Saying that though I think the missing attribute would more likely just cause a warning or error message in a 2D log file, whereas exceeding the free limit for 2D cells is more likely to cause the issue you're seeing. Maybe increase your model grid size to reduce the number of cells below 100,000 to check this.

Regards

Bob

 

Niamh O'Malley

05/02/2019 15:52:37

Hi Bob,

Sorry my cell size was below 100000, it was an irregular shape so I have just mutiplied the rows by columns to get the size which was wrong. In any case I increased it grid size to 5 and then 10 and it still made no difference.

I also tried using the link lines as an inflow boundary for the simple 2D sim and it runs ok so the 2D element seems fine - I was able to produce a flood map. Any other suggestions are welcome.

I have uploaded the model files to Dropbox if that is any use: https://www.dropbox.com/sh/kwvfzr80evm4mlj/AABvZDTIqZbEbDTJikW5A2oGa?dl=0

Thanks

Niamh

Bob

Flood Modeller Support

05/02/2019 17:07:01

Hi Niamh

Thanks for sending through your files. Your 2D area seems to be modelling the same area as covered by your 1D sections. I would expect to see separate domains for the left and right bank (each with their own active area and link lines and both independent from the area covered by the 1D sections). When I display the data on the map view the georeferenced 1D cross sections seem to extend a long way into the 2D domain. But then the 1D link lines are located near the middle of each section rather than at the extremities.

I think all (or at least some) of these things will cause issues with your model. However, the model still started for me and was producing log information. So I'm still uncertain as to what is causing the effect you are seeing. I will try to investigate further, but I would advise also improving your model setup as well.

Regards

Bob

Niamh O'Malley

06/02/2019 09:19:05

Hi Bob

I found in the past the 1D model became unstable when I didn't extend the section due too much glass wall effect when the water spilled out of bank. I can try reducing the cross section length and putting the link lines at the section ends (I know this is the way your tutorials show how is should be done). I generated the left and right link lines with seperate active areas for each side but I then used an overall active area to try run the linked model. Thanks for your help, if you come up with any other suggestions that would be great. I will let you know how I get on.

Niamh

Bob

Flood Modeller Support

06/02/2019 09:31:45

Hi Niamh

For a 1D-2D model there should not be a glass wall effect as the water going out of bank in the 1D should be passing into the 2D domain - assuming all link lines are set up correctly. Obviously when checking the 1D by running it in isolation you may need a high glass wall to get it to run, but this shouldn't be left in for the linked model runs (in the areas where you want the 1D to spill into the 2D domain).

The overall linked model should be setup so that the 1D and 2D elements are modelling their own discrete areas without overlap. The link lines you create will then define the areas where the two model components interact, exchanging water between the 1D channel and 2D floodplain.

Hope this helps.

Regards

Bob

 

Niamh O'Malley

06/02/2019 11:32:18

 Hi Bob

So I reduced the sections significantly and have the link lines along either end of the section extents. 1D runs fine but still the same story for the 2D. I used a grid size of 10m to make sure I wasn't over the limit for the free version but still no luck. The new files are in my dropbox if you could have a look? I really appreciate your help.

https://www.dropbox.com/sh/96wtb2uwifpk8eq/AAAeFUMFELK_XGYZ32Y5K1cfa?dl=0

Thanks

Niamh

Bob

Flood Modeller Support

07/02/2019 14:40:53

Hi Niamh

Thanks for making your data available. I've looked at it and found a few issues for you to investigate further:

1.  Some of the link lines are "cutting corners" between points and consequently traversing outside their associated domain active area. In shapefile edit mode you need to use the move and/or delete vertex controls to manually edit your link lines, possibly also using snapping to help with this as well (e.g. snapping to the end point of a cross section). As I said before, the current link line generator can help but is often not a "one-click" solution and some manual editing is often required.

2. Some of the link lines overlap each other, meaning the 2D solver is trying to feed data from2 different 1D nodes into the same cells. Again you need to carefully check your link lines and use manual editng to remove any overlaps (again use move/delete vertex).

3. When the 2 domains for left aand right are setup in the 2D solver with a larger grid size than the 2m resolution of your DEM there is a section about halfway along the reach where cross sections are only about 6m across. Consequently it is possible for the computaional grids set up by the 2D solver for each domain to "touch" which will lead to a model crash. You might be able to see this happening in the iwet check grids written out by the 2D solver. I'm not sure what you can do here other than widen the 1D section in this area to create sufficent space between the domains.

4. Looking at a 1D only steady run of your model, plotted as a long section of the entire reach with calculated levels overlaaid, the water level at the downstream end (from section 6 to the d/ss boundary) seem very high (immediately out of bank). This won't help model stability. You could adjust the d/s level in the HTbdy or change the node type to a normaal depth boundary or QHbdy. Hopefully one of these options will improve things.

So plenty to review in your model. I hope this information is useful.

Regards

Bob

Niamh O'Malley

07/02/2019 16:22:49

Hi Bob

Thanks, I will definitely look at those suggestions. I discovered why the linked 1D-2D wasn't starting at all, I changed to a different PC and that seems to have done the trick - the model at least starts running and then crashes so I have something to work with from the log files. Not sure why my own PC was not behaving as I have used this a number of times to do the same type of modelling.

Thanks for all your help

Niamh

Sign in to create or reply to forum posts.

Privacy & Terms

Copyright