You've got it right that partition switching doesn't physically move the partitioned data. It simply changes pointers that define which table the partitioned data is owned by. But for this to happen, you have to use the same partition scheme and partition function for the table(s) X_n and for table BigDaddy. The partition scheme points to the filegroups that will be used.

What you are proposing only makes sense if each X_n table being loaded has a logical separation from all the other X_n tables that is identical to that used for partitioning. For example an individual year, or an individual state or province, or an individual client. The partition scheme and function will have to use the same column that logically separates each X_N table.

Also, partition switching is all or nothing. If BigDaddy already has a partition with data for 2012, for example, you would have to swap the existing 2012 partition out before swapping X_2012 in. If you want to preserve the existing data, you're back to traditional merging.