tempdb can also grow due to maintenance jobs, e.g. if CHECKDB spills. Have you established a timeline when tempdb does most of its growing? If it is happening during a maintenance window you may have your culprit in a scheduled index maintenance or integrity check job.

Please read this article to check some basic settings of tempdb. If tempdb is growing for a valid reason that you cannot prevent then you should consider initializing tempdb at the proper size, 50 GB, i.e. do not rely on autogrow.

Has tempdb grown since the last restart?[/url]

In the References section of the article there is a link showing a way to check the Default Trace to see when the autogrow events occurred. That will help you establish a timeline of when the autogrowth is occurring to know if it is random during the day or during a maintenance window.