Yikes. That is some bad data consistency.
That helps some but sounds likeyou did not add the formula field for the T/F.
Sorry I can't give you a straight formward answer here but with that many variations in the data it is hard to make a conversion from a string to a date type.
My guess is that it is choking on the likes of the second example where it has no year.
When there is no year how do you know if the data is from the previous day or the previous year and a day?
Try and use an isDate function in your select criteria.
You can either filkter all the records that are not or include them and see if there is a consistent patternt o which fields are not able to be converted.
So if you put this as your select criteria:
if isDate({TEC_T_EVT_REP.DATE_EVENT}) then
date({TEC_T_EVT_REP.DATE_EVENT})=dateadd('d',-1,currentdate)
else
NOT(isDate({TEC_T_EVT_REP.DATE_EVENT}))
It should give you all the records from yesterday and all the records that are not considered real dates.
What do you get back from this?