Hallo Peter,

you wrote:

Only the average power determine, if a heat sink was needed or not.

This seams to be obvious. But let me tell you about a strange experience I made many years ago:
I wanted to supply a car radio for testing purpose with the help of 7812. Unfortunately car radio had some booster and contained a switching mode power supply. Fold back characteristic of voltage regulator made it to refuse to work. Inrush current at power on of such car radio was too high. The strange thing was, that at the beginning of a session the regulator worked! But after running some minutes regulator refused to work when again powering up of car radio.
I first thought that 7812 regulator was damaged somehow. So, I used another 12V supply from another application laying around on the table. This worked without trouble.
I tried to find out what was happening. And the only difference was that the proper working 12V regulator was mounted on a heat sink. So, I put the trouble making regulator also on heat sink, and now everything worked fine!
The strange thing was, that the regulator wasn't heating so much, at all, that you would mount it on a heat sink. Much less than 50°C (which is the temperature I can put my finger on it for longer period, without having pain).

The solution is that fold back characteristic of regulator is temperature dependent. And if high currents pulses are to be sourced by regulator in a certain application, good heat sink should be used, even if average power is so small, that normally a heatsink need not to be used.

Regards,
Kai