Hi Lucie,
I would recommend using the latest version of SAM, which is SAM 2021.12.02 r1. I understand if you are using the legacy SAM 2015.6.30 version to try to replicate the results shown in the 2017 paper.
If you want to use the legacy NSRDB 1961 - 1990 (often referred to as "TMY2") weather data with the latest version of SAM, you can download those files from this link:
nsrdb.nrel.gov/data-sets/archives and add them to your SAM solar resource library. (You can also copy the files from the solar_resource folder your SAM 2015.6.30 installation folder to another folder on your computer and add the files from that folder.)
To add the files to the solar resource library, put them files in a folder, and click
Add/remove weather file folders on the Location and Resource page to add the folder to the list that SAM scans for weather files and add them the solar resource library. That will make it possible to choose the TMY2 files from the library list.
As for the difference between the values shown in the paper and yours, did you model an array with 18 modules in a single 6 x 3 row as described on p. 11 of the paper? I've attached a file created in SAM 2015.6.30 that shows the same results as those in the paper.
Best regards,
Paul.