Well Brian, I agree with what you say – as far as you go.
The thresholds for acceptable absorbed power density [in a specified volume of flesh**] and acceptable excess temperature rise [2C} will no doubt have been based on loads of animal tests and extrapolated to human bodies to include the normal heat-removing mechanisms.
If the actual absorbed power density of EMF exceeds the acceptable threshold by more than some safety margin, those heat-removing mechanisms may not be able to remove the generated heat fast enough and we have a non-thermal equilibrium case where the flesh temperature continues to rise and possibly by more than the stated adverse health effect threshold of 5°C.
Conversely [as I previously described] the safe case [continuous over long time] is where the actual absorbed power density remains below that the acceptable threshold so that the heat-removing mechanisms are able to remove the generated heat fast enough and we have the thermal equilibrium case where the flesh temperature rises a bit [up to 2C] but below the adverse health effect threshold of 5°C.
Mammalian muscles are only 25% efficient. For every 4 joules of energy converted from stored chemical energy, only 1J goes to moving the muscle and 3J is wasted in heat. So, generating heat and getting a small internal temperature rise is normal when walking and running over many many hours without harm [we evolved to cope that way!]. A low-level of EMF is doing the same thing without harm.
Also, we should not confuse the heating effect from a RF source with ionizing radiation (e.g. high UV, X-rays, gamma rays) where the duration of exposure is definitely a strong risk factor.
[Specsavers, notwithstanding] the Ofcom calculator seems not to have a duration parameter for the above reason.
**actually, it states 2 sets of absorbed power densities and flesh volumes for different parts of the GHz spectrum. They don’t seem to have spec’d HF and VHF.