My first idea:
read the table with the SQL statements with an Execute SQL Task and store the results in a recordset. This recordset is stored in an object variable.
Then use a for each loop to loop over the recordset and map the SQL Statements to a variable.
The only problem is that when the different SQL statements return different columns with different datatypes, the metadata is always screwed over, so that becomes problematic in the DFT.
Maybe read the variable with the SQL stament in a script task and do all the handling there?