dE/dt and dEps/dt problems leading to Envi-met discontinuation

Discussions about the Version 4 concerning the core model simulations
Post Reply
marmuebn3
Posts: 19
Joined: Fri Oct 23, 2020 5:16 pm

dE/dt and dEps/dt problems leading to Envi-met discontinuation

Post by marmuebn3 »

Hello together,

I'm facing problems with my simulation:

After some dE/dt>200 Instability resp. dEps/dt>200 Instability warning-messages
ENVI-met stops after multiple Max. dE/dt messages.

In an earlier post Tim recommended check of the building structure in the problematic areas (according to the output).
I cannot find unusual structures in the area mainly affected (I mainly looked for single gridpoint surrounded by nuildings, high buildings).

Also Tim recommends to decrease the windspeed to about 1.5 m/s.
I use a full forcing with initial wind speeds of 0.5 m/s during night. They will increase to 3 m/s in daytime. Direction is constant. The crash occurs immediately after the start of the main simulation, that means after initialisation (as calculating sky view factors etc.)
I think, this is a realistic and frequent wind setting that should be covered by ENVI-met to handle common questions.

What is the best way to handle this problem?

To help myself better: how to analyse the PANIC dump in respect to my issues?

Thank you for any hint,

Markus

Tim
Posts: 1989
Joined: Mon May 15, 2017 3:46 pm

Re: dE/dt and dEps/dt problems leading to Envi-met discontinuation

Post by Tim »

Hi Markus,

Could you attach the files here as zip? Maybe, I have another suggestion for you if I see the model area and simulation settings. The forced wind speed should definitely be Ok, especially if wind direction is held constant.

Best regards,
Tim

marmuebn3
Posts: 19
Joined: Fri Oct 23, 2020 5:16 pm

Re: dE/dt and dEps/dt problems leading to Envi-met discontinuation

Post by marmuebn3 »

inx an simx-files and/or something else?

Tim
Posts: 1989
Joined: Mon May 15, 2017 3:46 pm

Re: dE/dt and dEps/dt problems leading to Envi-met discontinuation

Post by Tim »

Hi,

Everything that could lead to the issue (also fox file, maybe the projectdatabase if you use specific materials)...

Best regards,
Tim

marmuebn3
Posts: 19
Joined: Fri Oct 23, 2020 5:16 pm

Re: dE/dt and dEps/dt problems leading to Envi-met discontinuation

Post by marmuebn3 »

here we are, project databases are not in use
Attachments
Lappen_Ist_Sommer.zip
(1.23 KiB) Downloaded 2 times
forcing.zip
(148.85 KiB) Downloaded 2 times
Modell_Ist_Sim_1_3gr.zip
(848.39 KiB) Downloaded 2 times

Tim
Posts: 1989
Joined: Mon May 15, 2017 3:46 pm

Re: dE/dt and dEps/dt problems leading to Envi-met discontinuation

Post by Tim »

Hi Markus,

The issue is most certainly connected to your model area. Did you have a look at it in 3D mode? It looks like the core model area is on a podest and the outer cells are cut away. That is not what is meant by free cells near the model border: Everything should be digitized correctly (plants, surfaces, terrain), only buildings need to be erased to allow a free windflow.
However, I think that you don't need those free cells at all in your model area. Most buildings are a few cells away from the model border and you don't have any densely built street canyons where the wind flow would be blocked. Simply load the status before you added those empty cells and simulate with that one.
However, as I wrote in several posts the last months: Please don't use terrain at the moment, there are some unsolved issues and it won't give you better results. It rather makes simulating and analyzing much more complicated and rather unrealistic. Check other posts where I wrote more in detail about it. So, my guess would be to delete the terrain in total to erase that error source.

Best regards,
Tim

marmuebn3
Posts: 19
Joined: Fri Oct 23, 2020 5:16 pm

Re: dE/dt and dEps/dt problems leading to Envi-met discontinuation

Post by marmuebn3 »

Hi Tim,

thank you for your replay! Its not the best message, that you do not recommend simulations with DEM at present time.

To be true, at least in the 3D-View within SPACES I could not recognise that the model area is like on a base standing on the level of the added grids around the model borders. So I deleted these grids, but the problem still exists.

So I decided to export again from MONDE. But freshly exported from MONE my area had the problem of having undefined walls (I think this is a result from importing from OSM). To avoid fixing this manually finally my question:

Is ist possible in MONDE to assign a standard wall to an existing (!) layer of buildings, so that after exporting to SPACES walls are defined?

Thank you for help

Markus

Tim
Posts: 1989
Joined: Mon May 15, 2017 3:46 pm

Re: dE/dt and dEps/dt problems leading to Envi-met discontinuation

Post by Tim »

Hi Markus,

If you want to change the material for all walls and roofs at the same time, simply go into Settings in Monde (in the tab MyWorld) and change the materials. Otherwise, you need to manually click on each building and choose a different wall material.
The same general setting for changing the wall/roof material for all buildings at once, can also be found in Spaces in the Edit Settings dialog (in Default Settings).

Best regards,
Tim

marmuebn3
Posts: 19
Joined: Fri Oct 23, 2020 5:16 pm

Re: dE/dt and dEps/dt problems leading to Envi-met discontinuation

Post by marmuebn3 »

Hi Tim,

I fixed the wall problem, flattend the domain by switching off the DEM and simplified the building structure a bit, BUT I couldn't get rid of the dE/dt > 200 resp. dEps/dt > 200 problem.
During 3D initialisation of all variables - calculating new turbolence max change dE/dt is growing steadily, issueing about 1.500 warnings until an real overflow happens as dE/dt reaches 1.1E+307. This behavior stayed completely unchanged during the efforts of get the simulation runing.

In a last attempt I switched off the forcing and - surprise - it works! It took 9 hours for the first simulation hour, but it works.

I have no idea, why, because we had already checket the forcing file. Also the diagramms of the meteorological variables providded by the forcing manager looked pretty reliable. Do you have any idea?
As I'm running out of license, I will not be able to initiate my simulations again with the DEM and the original building structure but without forcing.

Also, I'm wondering,

- why no log is witten? That would it make much more easy to analyse i.e. the location of the problems (instead of manually checking the output window (see screencopy, no copy&paste possible!; this is - may be - because the core is still running after the final error message, but not consuming CPU-time).
- is there any experience in analysing the panic dump (or is that only a tool for the developers?)
- and - not 100% serious - why I didn't stay at ENVI-met 3 (as I lost the two features, I was focussed on: DEM and external forcing. I don't want to offend the developers, there are for sure advantages in the numerics and I can run models at any size.)

Additional note: I discovered that adding grids at the model top in the tools menu in SPACES deletes all builings. Adding grids at the horizontal borders workes fine.

Best regards and thanks in advance

Markus
Attachments
Modell_Ist_Sim_2Dcleared_area3.zip
(230.29 KiB) Downloaded 2 times
screen.gif
Lappen_Ist_Sommer.zip
(1.24 KiB) Downloaded 2 times
forcing.zip
(148.85 KiB) Downloaded 3 times

Tim
Posts: 1989
Joined: Mon May 15, 2017 3:46 pm

Re: dE/dt and dEps/dt problems leading to Envi-met discontinuation

Post by Tim »

Hi Markus,

Okay, now we know that it is connected to the forcing file. I opened your attached files and found that you want to simulate the 23rd of June but the forcing file only contains invalid data for that day. If you attached the correct forcing file, this is the issue for the simulation crashes.

- A log file is written in the end of the simulation. If the simulation crashes before, the log file does not exist.
- The panic dump is a standard output file that can be found in the output folders. It prints out the current status of all variables at the moment of the crash. If e.g. a high turbulence happens during midday of the simulation, wind channels etc. can be found causing the issue by analyzing the panic dump. Building structure or wind profiles can then be changed to avoid the issue in the next run.
- About your last point: I think I need to clarify the DEM topic a bit. The terrain simulation has always been at the current status. There are no new changes that make the simulation of DEM more problematic. However, people nowadays seem to use DEM for every study that they want to simulate. This is probably caused by the easy DEM import that can be done in Monde. In the following, some reports came up that the results seem to be rather unrealistic and that some more simulation crashes did appear when simulating with DEM. We thus had a closer look at it and are not satisfied with the current implementation of DEM anymore. It simply came into our focus now, especially since more people are using it for simulations where the DEM gives no valuable additional information. We thus have now the wish to improve the DEM implementation to give more reasonable results and be more stable. It will take however some more time to apply those changes since the whole topic is pretty complex. You are still able to get the same results with DEM as before but we simply don't recommend it at the moment. In addition, FullForcing algorithms were improved since ENVI-met 4.4.3 (I think that you meant that version since ENVI-met 3 did not give you the possibility of forcing at all), so I don't see why you should want to rather use that old version.

Adding grids at the top is easiest done by simply using the Edit Area dialog.

If you want to attach multiple files, please make one zip that contains all files instead of attaching every file individually.

Best regards,
Tim

marmuebn3
Posts: 19
Joined: Fri Oct 23, 2020 5:16 pm

Re: dE/dt and dEps/dt problems leading to Envi-met discontinuation

Post by marmuebn3 »

Hi Tim,

thank you for the response!

About the forcing file I can't follow you, may be I misunderstand you.
Looking into the .for-file I can see, that it contains "data" for all the year, most invalid resp. missing values, but exactly from June 23rd to June 24th noon there are data in the forcing file that are exactly the same as the ones in my forcing data (it isn't fully implemented, so I don't know what the models assumes - given the forcing data for near ground - for higher levels). Also, the forcing manager displays diagramms of the forcing data, that are correct.

There will be no time to check what went wrong as I have to finish my simulations, but I don't think its through invalid forcing data.

Of course I will put all attached files in one zip in future.

Best regards

Markus

Tim
Posts: 1989
Joined: Mon May 15, 2017 3:46 pm

Re: dE/dt and dEps/dt problems leading to Envi-met discontinuation

Post by Tim »

Hi Markus,

Ah okay, sorry for the irritation. Another user also named his fox file "forcing.fox" and I loaded that one instead of yours.

I have another idea about the issue: Could you open the DBManager and check all the soil profiles and wall/roof materials that you use (also the default materials). I have seen it twice that the albedo (or another parameter) was changed to an incredible high value without notification of the user.

Best regards,
Tim

marmuebn3
Posts: 19
Joined: Fri Oct 23, 2020 5:16 pm

Re: dE/dt and dEps/dt problems leading to Envi-met discontinuation

Post by marmuebn3 »

Hi Tim,

thank you for the hint.
Checking the profiles I cannot fin extreme values for albedo or any other property. The albedo of Loamy Soil is exact zero (a very bit too low I think), the others vary between 0.2 and 0.8 (0000PL Concrete Pavement Light). I'm not having so much experience, but from my knowledge I woould say, the values are ok.

Best regards

Markus

Tim
Posts: 1989
Joined: Mon May 15, 2017 3:46 pm

Re: dE/dt and dEps/dt problems leading to Envi-met discontinuation

Post by Tim »

Hi Markus,

Okay, that sounds good to me. 0 is correct for Loamy Soil since it is a natural soil profile and the albedo is calculated dynamically during the simulation.
I will try to debug your simulation to see why the crash happens.

Best regards,
Tim

Post Reply