I know lightning strikes kills LED lighting in seconds, but that all form of discharge lighting, are immune to lightning strikes.
Are incandescent and halogen lamps also immune to lightning strikes or they may be killed by them?

There aren't much things that are immune to a direct lightning strike. A lightning striking directly a lighting column outdoors will likely blow the luminaire to pieces regardless of the lighting technology, and blow open circuit the 1.5mm2 conductors in the cable running from it to the cutout at the pole bottom

I assume the question is about the residual pulse, conducted through the electrical grid to the luminaire from a remote lightning strike



First there is the effect of the grid before the pulse arrives :

The pulse shape changes (pulse length widens, peak amplitude lowers) just due to going along wires due to induction

The pulse energy decreases as it goes to branches in the grid, where it splits to different directions

The pulse energy decreases as it is absorbed by different components of the grid, those may be surge protection devices, but all devices that are affected by a voltage transient will dissipate some energy

In general, the more far away in the grid the strike happens, the more inductance, branches, and otherwise dissipation will take place before the pulse arrives to the lamp

In addition, a pulse may cause an insulation breakdown in the electrical grid, which will lead a circuit breaker at a substation to switch off and back on. The switching may lead to additional transients on the line, which are not from the lightning itself



Depending on how the strike happened (to the grid, to the ground) and the grounding system used for the installation (TT, TN-CS), the pulse may arrive between L-N or between (LN)-E, with different effects :

A pulse between L-N is applied directly to the input of components facing the line : The lamp itself in case of incandescent lamp, ballast line front end in case of electronic ballast, PFC capacitor, or series combination of few components in case of magnetic ballast

A pulse between (LN)-E is applied to the insulation between the components and ground. It may also be coupled capacitively to individual components "deep" in the circuit



The further damage will depend on what failure mechanism the pulse may trigger, how sturdy is the technology against such failure (how much pulse energy is required to cause the damage), and whether there may be continued additional damage happening after the original pulse is over

Some example cases :

 - A pulse across L-N strikes a discharge in an incandescent lamp parallel to the filament. It may take a significant overvoltage (in the kV range) to strike, but if started, the discharge will continue from line voltage, destroying the lamp

 - A pulse across L-N is less likely to do anything in a magnetic HID or Fluorescent circuit - except the capacitor, as most of it will be very attenuated by the ballast. As long as the ballast insulation withstands (and it is fairly reliable), there won't be long lasting effects in the lamp

 - The PF capacitor of a magnetic circuit is facing the line directly, and insulation breakdown may happen. In many cases the capacitor will have some damage in the form of decreased capacity, but still keep working for the time being

 - A "stupid" HID ignitor is very unlikely to be damaged at all, as its internal circuit starts from a resistor and capacitor of high values, forming an RC circuit

 - A "smart" HID ignitor typically contains additional components powered from a capacitive dropper. The capacitor is likely to let through a pulse, but there is high chance that the Zener diode will still be able to clamp the voltage and prevent damage. The fact that the pulse is greatly attenuated by the ballast helps here

 - A fairly low energy pulse across L-N may cause immediate breakdown of semiconductors in any LED driver that uses some topology of switching a LED array directly from the sinewave line. Those drivers, commonly used in "driverless" modules generally won't survive the minor transients present in everyday line voltage for more than a few months anyway, even without a lightning strike. Once there is a breakdown of switching components, subsequent destruction of the LED array from normal line voltage is likely to happen

 - A higher energy pulse across L-N may cause breakdown of line front ends of most electronic ballasts such as a bridge rectifier. If that happens, short circuit current from normal line voltage will follow and blow the bridge rectifier and fuse

 - A very attenuated pulse on a relatively high impedance line, may be absorbed by the input electrolytic capacitor in an electronic ballast without damage

 - A surge protection device parallel to the inputs of an electronic ballast may attenuate the pulse, preventing damage if the pulse can be attenuated low enough

 - A pulse between (LN)-E is capacitively coupled from the metal ballast case (or from the metal luminaire body on which a plastic case ballast is mounted), directly to ELV control circuitry on the board in an electronic ballast, destroying it immediately

 - A high pulse between (LN)-E from a nearby strike causes an insulation breakdown in a HID ballast, causing a flash over. The flash over may or may not continue conducting from line voltage, destroying the ballast and lamp

 - An old mercury lantern on a wooden pole is not grounded, which inherently limits the (LN)-E pulse that may arrive to it

but that all form of discharge lighting, are immune to lightning strikes.
Are incandescent and halogen lamps also immune to lightning strikes or they may be killed by them?

Not true/not true.

Lightning related overvoltage readily kills insulation in ballast chokes leading to short circuit across the winding or to the ground. I believe this happened once in our summer house.

Overvoltage spikes high enough (say 400-600V in 230V network) kill incandescents instantly because of arc flashover across the filament.

Though, electronic ballasts intended for streetlight application are somewhat hardened with input protection circuitry like MOV and typically have some overvoltage rating explicitly stated in datasheet.

There aren't much things that are immune to a direct lightning strike. A lightning striking directly a lighting column outdoors will likely blow the luminaire to pieces regardless of the lighting technology, and blow open circuit the 1.5mm2 conductors in the cable running from it to the cutout at the pole bottom

Not so dramatic actually. @Michael from Switzerland showed us what happens. If I remember right, typically something like a burned/melted spot on the case about 2cm across may be with a small hole.

Airplanes also suffer lightning strikes often. No dramatic damage too, small melted spots, some small holes may be, burnt-off static dissipation rods in the wings.
 

I reached to this conclusion because of this picture: https://www.lighting-gallery.net/gallery/displayimage.php?pos=-254073

Insulation of ballast chokes does withstand repeated pulses of 4.5kV (from ballast-dependent HID ignitors) for long periods, both between the winding ends and to ground. It will likely withstand somewhat more in a single pulse

The line front end of an electronic ballast typically starts from an 1kV or so rectifier bridge straight across the line. In the rest of the circuit there are more semiconductor parts of not so high voltage rating, which may or may not see the same pulse (depending on passive components on the way)

The 5kV vs 1kV difference alone explains the picture linked



I think the arc striking at 400V/600V is not the same fault mechanism as discussed here

If applied for few tens of milliseconds (order of an AC cycle), 400V will overheat and burn the filament. The breaking ends of the filament will ignite the arc

In this discussion we are talking about a pulse of the duration of few 10's microseconds. It is not sufficient time to have effect on the filament. The voltage must directly ignite the gas parallel to the filament, and i dont expect this to happen at 400/600V



Whether an MOV will help there depends to a large extent on the line impedance, in addition to the pulse voltage/energy. In a low impedance line (from the pulse source perspective), both the MOV and the circuit will get enough pulse energy to be destroyed, if sufficient energy arrives. The higher the line impedance, the more effective the MOV is at protecting the circuit

For an electronic converter rated for outdoor use, there is almost certain at least a large MOV across L/N terminals. Also MOVs ( or MOVs + gas discharge tubes or SIDACs in different combinations) may be fit between L/N and ground. This significantly hardens the converter against lightning spikes, both in differential and parallel modes.

MOVs are are actually quite good of arresting lighting induced voltages, short of a direct strike of course, because of their large physical size and so high thermal inertia for the energy dissipated in a bulk of oxide material.

Gas filled incandescents are in fact operated *just a little* below gas ionization potential. This is why they are arcing so easily at EOL. A brief overvoltage, not enough by energy for thermal effect on a filament is certainly able to ignite the arc and kill the lamp immediately. We are not talking kilovolts here.

There in Finland, lightning struck a MV streetlight (Idman 8522). As a result, the lantern and its cable were completely broken. The cable even broke. In addition, a modem in a nearby house broke. Apparently, there was also a fault with the telecommunications mast.
https://youtu.be/hDcZeZB3yJQ