Running WE-UQ EVT

[Kate]

Hi,

I have encountered 2 problems while using WE-UQ EVT, can anyone fix it or do I need to modify the case?

1. I tried to run a case with CFD-Basic, it failed after a few minutes, the Dakota.err says " at least one variable must be specified".  I'm attaching the .json file.

2. I tried to run CFD-Expert with the case SampleBuilding, when the job is completed, I can not retrieve the result on WE-UQ. It says failed to download. I have the access to the result through DesignSafe portal though.


Thanks,
Kate

[fmk]

1) the error is likely due to fact that you defined no random variables for the building. you need to create at least one random variable for the building when you use the CFD option and then provide information on the variables distribution. you can create a variable named "dummy", specify it to be of type normal and enter a value for mean and stdDev if you have no actual random variables.

2) can you send the .out and .err files from designsafe. also the dakota.in and dakota.out files that are in the tmp.SimCenter directory if present.

[Kate]

Thank you for the quick reply!

1) I'm not sure where I should define the variable for the building, I have only set the GI and EVT, as attached below.

2) also attaching the files.

Thanks!
Kate

[fmk]

1 and 2 could be same issue, i.e. no random variable.

go to the SIM panel. If using the stick model unaltered, put in wFloor for weight. Then go to RV tab and for a normal distribution put in a mean value, say 150 and a stddev say 10 just to see if you get out some numbers .. then try and run again.

[Kate]

Thanks for the reply. I have tried your suggestion, got the error:Local Failure Setting up Dakota.

Then I tried the sample case with exactly the same input provided on the WE-UQ  documentation (I used the settings on the figures instead of the description when they are not exactly the same), got the error:No Dakota.err file - dakotan did not run - problem with Dakota setup or the applications failed with inputs provided.

Any suggestion?

Thanks,
Kate

[fmk]

did you follow the install process .. python, dakota, and OpenSees? .. we should arrange a zoom call to sort out .. if willing, can you send some time slots for today or tomorrow and indicate your time zone .. thanks

[Kate]

Thanks for the reply. That's a good point as I just got a new laptop and didn't check the Python installation. I was able to use the WE-UQ on my old computer.

There are issues with Python installation on the new Mac OS, I' m trying to solve the installation problem then will try the WE-UQ and will keep you updated.

Thanks,
Kate

[fmk]

see what version of python is on your new machine. if python --version retyurns greater that 2.7 you may be able to use it .. if >2.7 try pip install simcenter .. if that works type "where python" in a terminal application and then in WE-UQ Preferences set the path to python equal to result of where python

[Kate]

I'm using Mac Big Sur OS with M1 chip, I had the newest Python version(3.9) as I thought that was for the Mac M1 chip. I don't have problem installing Python, it is the package that I could not install, like hdf5. I have tried a couple of older versions(3.8 and 3.7) and had the same error:Could not find a version that satisfies the requirement hdf5 (from versions: none)

it would be very helpful if it happens you know the solution. Or if you think you can instruct me to install the packages on Zoom with my screen shared.

Thanks,
Kate

[fmk]

you can run w/o that package for what is currently in that code

[Kate]

Thanks a lot! It works now!

The problems I had were:
1.Python installation.
2. Dakota and OpenSees installation. At first I didn't have dakota and OpenSees installed. I went through the Installation session on the documentation, did not see that Dakota and OpenSees needed to be installed. Although everything is clearly demonstrated in the video. It may be helpful to tell users clearly to watch the video or to inform the users of necessity of  OpenSees and dakota installation. Often times the documented instruction is enough and people may skip video instruction unless clearly told, especially for convenience in shared office.

I have another question:
After I ran the job which includes a CFD-basic in EVT, I retrieved the data. Here CFD result is of my interest. I am able to get the CFD result from sim.tar.gz from DesignSafe portal, but not from the retrieved data. I wonder if it is what retrieved data supposed to do or something is messed up?


Thanks,
Kate

[fmk]

1. We are about to release a new version .. Windows users will have nothing to install as we will include python, dakota and OpenSees .. Mac users will need to install python, bit OPenSees and dakota will be included.

2. by CFD result, do you mean the wind field in the model and pressures on the building as opposed to the building response?

[Kate]

Yes, that's what I meant for wind result. when I go to DesignSafe portal, those results are all in sim.tar.gz .

[fmk]

the tool at present does not download those results nor does it show them as the focus is on the building response. i take it that that is something you would like to see in the application for these OpenFOAM runs. Question is it it just for postprocessing to view the flow and if this processing is something that should be done at DesignSafe or on your local machine. In another tool we are working on, the user provides a paraview generated python script which will parse the data at designsafe and create .mov files or whatever for download. The reason is the files are 100's MB, downloading takes time and if the user is going to use paraview to view why not do it were the data is.

[Kate]

Good to know.Thanks.