Flow-wave execute problem - D-Flow Flexible Mesh - Delft3D
intro story D-Flow FM
D-Flow Flexible Mesh
D-Flow Flexible Mesh (D-Flow FM) is the new software engine for hydrodynamical simulations on unstructured grids in 1D-2D-3D. Together with the familiar curvilinear meshes from Delft3D 4, the unstructured grid can consist of triangles, pentagons (etc.) and 1D channel networks, all in one single mesh. It combines proven technology from the hydrodynamic engines of Delft3D 4 and SOBEK 2 and adds flexible administration, resulting in:
An overview of the current developments can be found here.
The D-Flow FM - team would be delighted if you would participate in discussions on the generation of meshes, the specification of boundary conditions, the running of computations, and all kinds of other relevant topics. Feel free to share your smart questions and/or brilliant solutions!
Flow-wave execute problem
From a search I found a thread saying about switching off the "online visulation" in the output storage area, but even after doing this it still hangs at the same point.
According to the top of the tri-diag.flow file I am using version "Delft3D-FLOW Version 4.00.07.000000" (struggled to compile myself so had to copy compiled files from someone else)
From going into the report for the tri-diag.flow file, the last line reads
"*** ERROR Calling wave every timestep is not supported. "
I found if I also de-selected the "Online Delft3D-Wave" in processes then it got to "Part VI - Initialisation & checking second part... " before it would once again just hang.
From going into the report for the tri-diag.flow file, the last line reads
"*** ERROR Waves activated but no COM-file present !!"
I'm unsure how I can make the timestep work or how to produce a COM-file, could someone advise me on this please.
You probably set the wave and flow coupling every timestep, which is not supported by the Delft3D system yet as it said in the tri-diag. This function is on the todo list but not yet. The way to walk around is to couple the wave with flow in larger time step.
Thanks for that, appears I was missing the obvious. Having changed that I've managed to get the file to run. It doesn't actually inform me that it has finished running, instead it gives the following:-
Part VII - Initialisation output...
forrtl: severe (64): input conversion error, unit 800, file C:\Users\owner\Documents\Uni work\Dissertation\DELFT3D\working directory\Simulations\flow check 2\FLOW2WAVE_DATA.txt
Image PC Routine Line Source
wave.exe 005ABEEA Unknown Unknown Unknown
wave.exe 0055A96A Unknown Unknown Unknown
wave.exe 00559B6B Unknown Unknown Unknown
wave.exe 0053546B Unknown Unknown Unknown
wave.exe 004EC37E Unknown Unknown Unknown
wave.exe 005001F1 Unknown Unknown Unknown
wave.exe 00460925 Unknown Unknown Unknown
wave.exe 004616FC Unknown Unknown Unknown
wave.exe 00402874 Unknown Unknown Unknown
wave.exe 005BA283 Unknown Unknown Unknown
wave.exe 00593653 Unknown Unknown Unknown
kernel32.dll 74CE339A Unknown Unknown Unknown
ntdll.dll 77279EF2 Unknown Unknown Unknown
ntdll.dll 77279EC5 Unknown Unknown Unknown
Even with this list, a successful tri-diag file was produced with no errors, but when I use quickplot to look at the com, trih or trim file I only get one time value. I'm unsure whether the programs problem with the FLOW2WAVE_DATA.txt file could be a cause, or like in the stand alone wave simulations a wavecon file is also required to supplement the time series.
Is this a problem you have encountered before?
You have to clean up all the old results files before you start a new run.
I saved the flow and wave files to a new directory with the required supporting files, and started the simulation again. The input error and the long list of unknowns have now gone, but the program is once again reaching a point, and then it hangs once again. I have disabled the remote online visualisation, and the software now seems to be happy enough to run.
Running the simulation now ends with
Delft3D-WAVE finished normally.
Flow exited abnormally
Check diagnosis file
child process exited abnormally
Going into the tri-diag file there is a loft of references to the bed change exceeding 5% waterdepth
*** WARNING Bed change exceeds 5.0 % of waterdepth after 31 timesteps, location (m,n) = (9,14)
Bed change messages skipped (more than 50)
Total number of Bed change messages for timestep 31 : 1590
Sediment erosion shortage at NM 3484 Fraction: 1 Mass available : -0.3005E+04 Mass to be eroded: -0.1553E+05
*** ERROR Water level change too high > 25.00 m (per 0.5 DT) after 32 timesteps in the following points:
(m,n) = (68,23), s0, s1, abs(s0-s1): 0.403E-01 , 0.798E+02 , 0.798E+02
*** ERROR Flow exited abnormally
I have closed down the time steps to 60min to try and reduce the amount of change per step, but this doesn't appear to have helped. Is there another way to try and overcome this?
Partial results have also been produced which I can view in QUICKPLOT, but I can't seem to locate the cross sections I set in the monitoring section. Where abouts would these be produced?
It seems the model is not stable. You should check the time step. Maybe it is far too big.
The results of computation are stored in two types of files map and his files. Try his files if you cannot see in the map files. The his files could be viewed by Quickplot as well.
Thanks for your help.
I've reduced the time steps to 1min intervals and it ran to completion, though such small time steps seems excessively small when considering morphology, though will try playing with the scale factor to see how this effects the results.
I still seem to be unable to locate anything relating to the cross-sections I specified in the monitoring section. There doesn't appear to be either a .his or .map file in the directory where I have run the simulation. Using Quickplot I have the option of opening 1 of 4 file types [com, trih, trim or wavm] Am I missing file types, or is it embedded in one of these?
What I have tried is to view a section by specifying a fixed value of M and all values of N which gives me a grid slice. This seems to work okay for a number of different variables within the trim file, giving me the option of going through the 121 different time steps (10days at 120min intervals, going to reduce this to make it quicker during setup).
The trih file appears to only have 2 time steps, the wavm file has 10 time steps, and the com file only has 1 time step.
Unfortunately what I am most interested in is "time-varying bed level" located in the com file, and this output doesn't have the option to view through the time steps. Since a "time-varying bed level" with only one time step would be rather strange, I assume I have missed an option somewhere of applying the time steps, but am unsure as to where I maybe able to locate this. Could you please direct me to where this option would be?
Indeed, morphology scale is usually larger than the hydrodynamic scale. Using morphological factor might bridge the gap a bit.
Using Quickplot I have the option of opening 1 of 4 file types [com, trih, trim or wavm] ... That is right.
If you only find 2 time stpes in trih-runid file, then the interval for his output is not properly set.
Com files is used to communicate with different modules, thus only 1 step of the results was stored in the com files.
Hope this help a bit!
I'm a bit confused by that, as "time-varying bed level" would suggest multiple times steps. Does the "time-varying bed level" only refer to the last step? and the only way to check the bed level in the com file is to compare this to the initial bed level? Is there another way in which I can monitor the bed level?
Are the cross sections stored in any of the 4 files, or is that purely for online visualisation as the simulation is running?
I've found that using timesteps of 5min for a 24hr period works fine with a morphological scale factor of 60, though runs into problems with either bed change being to high or the change in water depth being to great, if I try to reduce the scale. Is there a typical ratio of time steps to morphological scale factor?
On the occasion that I managed to get past this problem, I encountered another error, this time involving the velocity.
*** WARNING Velocity change too high > 5.00 m/s (per 0.5 DT) after 24 timesteps in the following points:
Would you know how I would be able to get past this?
Thanks for you help so far, I'm getting there slowly and apologies for so many questions.
Regarding to the situation, you may forget about the com file for the time being.
Cross section information should be stored in his file.
The model complained the velocity change is too big, since the model is not stable any more. You may check the Cr number of your model.
Afraid I haven't come across a "his file"
I've looked in the flow manual regarding the Courant number (Cr), and according to equation [4.1] (on page 40), I have a value well under 10, so should be stable (t = 1min, H = 10m, x = 15m)
It seems when ever I manage to get the simulation working for the bed change if stops working due to excessive velocity.
Is there a worked example of a morphology model that II could download to look through? That would probably be a good guide for me, and help me identify why I am getting so many errors for different things.
Sorry. The his file I mean trih-runid files.
You can find the example 03 with the source code with mor and sediment transport activated.
If you still cannot manage, would you send me your model to have a quick check? Qinghua.firstname.lastname@example.org.
I would like to see the example 03 for mor and sediment transport model, where can I find it?
I've got one idea of what might be going wrong, as I experienced a similar issue the first couple of times I tried to run a FLOW model --
Try double checking that the bathymetric sample points you used to define your depth file are defined with z-positive downward.
Many public elevation/bathymetry data sets are not formatted this way to begin with.
In other words, rather than the ocean floor being located at a depth of h = -4000m, make sure that its defined as h = 4000m.\