It's that "delta T avg" that points to the problem. It is apparently an attempt to get around the problem of thermal gradients within the wort; the simplified linear equations assume the only gradients are between wort and cooling water. For plate chillers this assumption is probably justified, as the distances between plates are small and the wort velocity probably results in complete mixing (within each channel). For counterflow chillers it may be untrue but probably not far enough off to cause major error. But for an immersion chiller, the assumption is not justified unless the wort is completely mixed at all times, such that the temperature at any point is very close to the average temperature. Unless you stir really vigorously and continuously, this is far from true; there is a significant thermal gradient within the wort (from the center of the kettle to the chiller coils) and this greatly increases the time to chill the wort. (The time required for heat to travel through the wort toward the chiller coils is nowhere accounted for in the equation; and the actual gradient at the coil/wort contact is much lower than the equation assumes.)

Failure to recognize, identify, and examine all the assumptions behind a mathematical model is the #1 cause of model failures. That is the best and most lasting lesson my hydrogeology prof (Michael Campana) taught me many years ago.