Hi!
You can keep on splitting the date (first separating date from time; then separating day/month/year and finally rebuilding the date in the desired format), but let's hope Microsoft Flow Support team can lead us into a more reasonable solution.
If not, this is my ugly proposal, taking the example 27/11/2017 17:00:00
split(outputs('Compose_End_Date_and_Time_(Split)')[1]),' ')[0] will provide 27/11/2017
split(split(outputs('Compose_End_Date_and_Time_(Split)')[1]),' ')[0],'/') will provide the following array ["27","11","2017"]
concat(split(split(outputs('Compose_End_Date_and_Time_(Split)')[1]),' ')[0],'/')[1],'/',split(split(outputs('Compose_End_Date_and_Time_(Split)')[1]),' ')[0],'/')[0],'/',split(split(outputs('Compose_End_Date_and_Time_(Split)')[1]),' ')[0],'/')[2]) will provide 11/27/2017
I haven't tested the expressions myself, let's hope I haven't forgot any parenthesis comma, single quote, etc.