We had similar situation when web application was using instance owner id to connect and continuously trying to reconnect. As Paul mentioned above, if your application is using userid which is part of instance owner group , it will manage to re-connect even after QUIESCE was issued. Removing web application user id from instance owner group will fix the problem.
As a temporary fix, we kept all the db2 commands in a file (backup.sql), and invoked db2 command with file as input ( db2 -tvf backup.sql ). There is a possibility that web application might still manage to connect, but using file the chances are low compared to issuing separate commands . Sleep command will not help, if your application is continuously trying to reconnect. I think same thing is happening when backup was taken thru control center (no sleep command) and you are able to take backup thru control center. Permanent fix is to use user id that was not part of instance owner group for web application to connect to database. Another alternative is to take online backups.