Hello,
Not sure how you are getting on with this, however I am not convinced that resizing a partion without SnapDrive is a good idea. There is no connection between the partion and LUN sitting on the storage without SnapDrive and you cannot therefore guarantee the blocks will align when you perform the resize, i.e. you may lose data! In addition, the storage will not release those blocks since again there is no connection between what Windows sees and the blocks on the NetApp storage. As mentioned already, NetApp does not know which are in-use blocks or blocks that Windows has released/deleted.
As I understand it, the only reason a thick LUN will run out of space is if the volume fills - unless of course the actual filesystem ontop of the thick LUN is full. You need to ensure there is free space in the hosting volume/aggregate. Yo may already be aware but if your volume has guarantee set to "volume", i.e. also thick provisioned, you have fractional_reserve set to anything other than 0% and have at least one snapshot in the volume, then you could reclaim back some space by setting fractional_reserve to 0% - however with the normal caveats (https://library.netapp.com/ecm/ecm_download_file/ECMM1278361 - page 33)
The auto reclaimation referred to is the space_alloc option that becomes available in Data ONTAP 8.2 and is valid for Windows 2012, see: https://library.netapp.com/ecmdocs/ECMP1368845/html/GUID-93D78975-6911-4EF5-BA4E-80E64B922D09.html
However, without upgrading ONTAP and no SnapDrive, the only other option I can think of to release those filesystem deleted blocks, is for you to create a fresh new LUN and copy the files (at the file system level, e.g. xcopy or ndmp) into that new LUN. If you do anything at the block level, then those released/deleted blocks will also follow. The new LUN used size will then match the filesystem size. Although, you will again eventually end up in the same situation as files are deleted/overwritten in the filesystem, without space_alloc or SnapDrive to clean this up.
Hope this help,
Thanks,
Grant.