Reverse key index on timestamp column, partitioned table, 3 Node RAC

Here is the scenario;

We have a table with 60 partitions (by range using timestamp), each partition contains at least 5M rows.  The table is being inserted from a 3 node RAC cluster.  To optimize the insert we decided to use REVERSED KEY INDEX on the primary key (doc_id,doc_date). 

Documents says that range scan does not work on RKI.  If that is the case how is partition elimination/pruning happening when the query is executed in the table using doc_date in the predicate?  Execution plan suggests that partition elimination is happening, but it seems slower compared to using another date column that does Full Index Scan across all partitions.  Am I doing it wrong?  I'll appreciate any suggestions.