Hi Stefan,
the options to recover from corrupted data structures in SAP HANA are basically the same as in all other DBMS platforms.
Data that is not stored somewhere else too (e.g. in an index, in the source column of a computed column, in a backup, ...) can only be recreated (e.g. by application logic) but not recovered.
To get existing copies back into place, it depends on where the data is located.
If for example the corruption occurs in an internal compound column (that is sometimes created automatically) then one option is of course to re-create this compound column from the original part columns.
Exactly this seemed to have been the case here, as the compound column ($BUKRS$MANDT$) data in the copied table had been re-created from the original part columns MANDT and BUKRS.
With the changed select statement I proposed earlier I _suspect_ that for the join the compound column was not used anymore - but the filter had been applied to MANDT on both tables first and then the join was executed via BUKRS only. That's why it worked in this case.
(once again just a lucky shot into the blue 
).
- Lars