The 100 ohm resistor doesn't limit the current flowing in the circuit - it's limited by the reactance of the 2.2 uF capacitor.
These values you've given, are they measured on the actual unit or are they from a simulation?
I've re-run the simulation. Please check that my schematic matches yours. Don't worry about the zener voltage being slightly different. It makes very little difference to the zener current. I've assumed the relay is NOT energised, since this means more power dissipation in the zener.
The red trace is the zener current. When it's above the 0mA line, the zener is conducting in the forward direction and the voltage across it is only ~0.7V so the power dissipated is negligible. It's only when current flows in the other direction, shown below the 0mA line, that significant power is dissipated.
I've filled in a couple of these excursions to show you where power is dissipated, and how much. I also did a higher resolution capture and calculated the area inside the wave as a fraction of the total cycle time and the peak current, and it works out to about 23%. The wave peaks around 185 mA, and 23% of that is 43 mA. So the mean zener power dissipation is about 1.16 watts in this simulation (assuming a 27V zener).
Can you tell me WHERE your simulation, or your actual circuit, differs from that simulation?
In my simulation, the RMS current in the 100 ohm resistor is about 155 mA which is about 2.4 watts dissipation! You can reduce the resistor to, say, 47 ohms to halve this dissipation (P = I^2 R, and I is mainly determined by the 2.2 uF capacitor). Another way to reduce the dissipation in the 100 ohm resistor would be to use a bridge rectifier, because the existing design wastes half the incoming current, but that's a major change so let's not go there.
Do you have access to an oscilloscope? An isolating transformer?