+- jeplus.org forums (http://jeplus.org/mybb)
+-- Forum: Building simulation tools (http://jeplus.org/mybb/forumdisplay.php?fid=1)
+--- Forum: jEPlus (http://jeplus.org/mybb/forumdisplay.php?fid=2)
+--- Thread: Weather files (/showthread.php?tid=408)
Weather files - gmavrom - 01-08-2016
Dear jEplus community,
I am trying to run a parametric study of a building for which I have 20 Actual Meteorological Years (AMYs) in weather file form and I will also vary some building parameters in my idf.
What I would like to achieve is have every weather file say 50 times and for each of these times randomly sample all other parameters.
What I want to ensure is that each weather file is selected an equal number of times while the other parameters are treated randomly.
Can I achieve something like that?
Thank you in advance,
George
RE: Weather files - Yi - 01-08-2016
Hi George,
If you do a LHS random sample of 20x50=1,000 jobs, in the results you will probably find 50 cases for each weather file. In theory it should work this way, although I haven't tested it thoroughly. The problem with this approach is that the sets of cases for different weathers are different, which may give you trouble in making comparison.
If you want the sample set for each weather to be identical, you can apparently run the project with each weather file and 50 samples of the other parameters. Change the weather file and repeat it for 20 times, and beware not to change the random seed value. This process does sound tedious, though. The best way may be to use the job list file (see: http://www.jeplus.org/wiki/doku.php?id=docs:manual_1_6_advanced#job_string_and_job_list_file). Do a sample run of the single weather project, first. This will give you a SimJobIndex.csv, which is very similar to a job list file. All you need to do is removing the first row and the first column of the file, I think. You can then copy the list 19 times and change the weather file name each time. In jEPlus, select the run "job list in file" option...
Regards,
Yi
RE: Weather files - gmavrom - 01-11-2016
Dear Yi,
Thank you for your response.
I will test it out with the LHS sampling and I will report back with results so that others can know that too.
I do not mind the fact that the sets of cases for different weather files will be different; that's what I am aiming for actually.
The last solution you suggest, I was thinking as well as a last resort due to its cumbersome nature. However, as I said before, I want a different, random set of sample values for each weather, thus, using the resulting job list file from a first run would not really work.
I think I will resort to just externally writing a job list file, according to the specifications and then feeding it to jEPlus.
Thanks,
George