You could do it if you expose fme_basename attribute in reader and writer, then check Fanout by Dataset (fme_basename) and then batch deploy all dgn's you want. This will write new GDB's named as DGN's.

Hope that will help.

The trick is to use a single merged feature type for the reader in the workspace that you will use in the workspacerunner. You also can create a published parameter for the writer. Then you can set this parameter in the workspacerunner. A Directory and filepath reader is also handy.

thanks for your answers , I'll try !!! thanks again ...

Thanks janin ... it works OK !! , Stalch I haven't tested your solution , sure is good idea also , I've tested the first that I've seen. Thanks to two !!