Hmm..firstly I don't understand what the title of your post has to do with your question.

However, if you have a single mdf file, then that file is your primary datafile and you can't remove it. In order to decrease the size of this MDF file you will first need to:

Create a new datafile on a different physical disk

Create a new filegroup

Add the datafile to the newly created filegroup

Next you need to physically move some of your data from the the primary MDF file into your newly created datafile. There are a number of ways to choose which objects to move into this new datafile but, once you have your plan in place, essentially the easiest way to migrate a table to the new datafile, is to rebuild its clustered index on the newly created filegroup.

To answer your question about time, this step will take as long as it usually takes to rebuild the clustered index on the tables you plan to move. If you have a 1TB database I expect you will have some substantial tables.

Once you have migrated sufficient data to allow you to reduce your MDF to its required size, you can then use the DBCC SHRINKFILE command to shrink the MDF to its 'new' size. You will need to rebuild the indexes on this primary filegroup, once the shrink has completed as the shrink operation will fragment the data. Shrinking a datafile this size probably take several hours and is a task best scheduled out of hours.

At this stage you have two smaller datafiles and subsequent object creation using the ON <FILEGROUP> syntax (or by changing the default filegroup for the database) will give you control over the placement of objects.