Engineering design

A computational model predicts that a cooling channel keeps a component below its temperature limit. What should engineers do before treating that prediction as dependable evidence?