I know that LED is more energy efficient than HID but how efficient is HID?

Depends on what HID source you are asking about. Generally, mercury vapor is the least efficient, followed by probe start, then pulse start MH, the HPS, and LPS is the most efficient.

Typical mercury vapor lamps have around 50 lumen/watt efficiency.
HPS lamps have around 120 lumens/watt efficiency.
Metal halide lamps (probe start) have around 80-90 lumens/watt efficiency.
LPS lamps usually have about 150 lumens/watt efficiency.

An easy way (but not completely accurate) to find how efficient different HID lamps are is to find the lumen output of equivalent wattages between lamp types, however this doesn’t work for LPS.


Higher wattage lamps are more efficient, so you have to use the same wattage for each lamp type to determine which has better lumen/watt efficiency. (Not including ballast loss)

Example with 1000w lamps:

HPS 1000w is about 140,000 lumens
MH 1000w is about 105-110,000 lumens
MV 1000w is about 55-58,000 lumens.


Obviously 1000w LPS doesn’t exist, so 180w SOX LPS is 32,000 - 40,000 lumens.

Interestingly, the first light source that had an efficacy greater than 100 lumens per watt was the SLI/H linear low pressure sodium lamps used in England. This efficacy was achieved in the early 1960s.

Interestingly, the first light source that had an efficacy greater than 100 lumens per watt was the SLI/H linear low pressure sodium lamps used in England. This efficacy was achieved in the early 1960s.

The question was about HID lighting, LPS is not an HID light source.!

LEDs are around 80-90 lumens per watt, but the existing HPS and MH are as same as efficient or even more, so LED streetlight changeouts are pointless. What's even more pointless are the SOX or SLI streetlights that are changed to LED, those streetlights are more efficient than LED so that's more like a downgrade than an upgrade.

The reason I think why the LED changeouts is just because money. Even if the new LEDs have the same or lower efficiency they save a lot of money in maintenance, Instead of going once a year removing pigeon nests, replacing EOL lamps or clean it up they just leave the fixture there and gets replaced every two years, or never get replaced at all even if they are EOL and it might end up cheaper than the maintenance if they choose cheap enough fixtures, also some governments will pay off part of the cost for support the "green energies".

The biggest problem we have here is that the 'powers that be' have decreed that the light level doesn't need to be as high as it used to be. 
Therefore the L*D fixtures that they have installed do not give nearly the same level of illumination as the HPS lanterns that they replaced. 
Now there are patches of darkness between the columns and we do not feel safe going out when it starts to gets dark, and in our part of world and at this time of the year that's about four in the afternoon. 
Additionally, the colour temperature of the L*Ds means that I can't see clearly. 

LEDs are around 80-90 lumens per watt, but the existing HPS and MH are as same as efficient or even more, so LED streetlight changeouts are pointless. What's even more pointless are the SOX or SLI streetlights that are changed to LED, those streetlights are more efficient than LED so that's more like a downgrade than an upgrade.

The reason I think why the LED changeouts is just because money. Even if the new LEDs have the same or lower efficiency they save a lot of money in maintenance, Instead of going once a year removing pigeon nests, replacing EOL lamps or clean it up they just leave the fixture there and gets replaced every two years, or never get replaced at all even if they are EOL and it might end up cheaper than the maintenance if they choose cheap enough fixtures, also some governments will pay off part of the cost for support the "green energies".

Where are those figures from?, average LED street lighting now are about 125/130 lm/W ?

The question was about HID lighting, LPS is not an HID light source.!

What kind of light source is LPS considered?

Technically speaking LPS is more of a gas discharge source like fluorescent. However it is often grouped under HID in North America especially for marketing purposes by the manufacturers.

What kind of light source is LPS considered?

It's still a discharge lamp, but a low pressure one similar to fluorescent. Technically it is not HID.

Depends on what HID source you are asking about. Generally, mercury vapor is the least efficient, followed by probe start, then pulse start MH, the HPS, and LPS is the most efficient.

Typical mercury vapor lamps have around 50 lumen/watt efficiency.
HPS lamps have around 120 lumens/watt efficiency.
Metal halide lamps (probe start) have around 80-90 lumens/watt efficiency.
LPS lamps usually have about 150 lumens/watt efficiency.

An easy way to find how efficient different HID lamps are is to find the lumen output of their 1000w version, however this doesn’t work for LPS.
For example:

HPS 1000w is about 140,000 lumens
MH 1000w is about 105-110,000 lumens
MV 1000w is about 55-58,000 lumens.


Obviously 1000w LPS doesn’t exist, so 180w SOX LPS is 32,000 - 40,000 lumens.

Going by the 1,000w lamp lumen ratings is actually not very accurate. HID lamp efficiency increases as wattage does, so a 1,000w high pressure sodium lamp is more efficient than a 100w one. (if we're not factoring in ballast loss, 100w HPS is about 95 LPW while 1,000w HPS can range from 130 to 150 LPW. The same applies to metal halide and mercury vapor lamps.

Going by the 1,000w lamp lumen ratings is actually not very accurate. HID lamp efficiency increases as wattage does, so a 1,000w high pressure sodium lamp is more efficient than a 100w one. (if we're not factoring in ballast loss, 100w HPS is about 95 LPW while 1,000w HPS can range from 130 to 150 LPW. The same applies to metal halide and mercury vapor lamps.

Thanks for posting that, I didn’t know about higher wattage lamps being more efficient. I can fix my previous post.

Going by the 1,000w lamp lumen ratings is actually not very accurate. HID lamp efficiency increases as wattage does, so a 1,000w high pressure sodium lamp is more efficient than a 100w one. (if we're not factoring in ballast loss, 100w HPS is about 95 LPW while 1,000w HPS can range from 130 to 150 LPW. The same applies to metal halide and mercury vapor lamps.

The exact opposite thing happens with LEDs, as you increase the power the efficiency dicreases, due to the internal resistance of the chips and the ballast losses, so when a decent 7w LED bulb performs around 80 lumens per watt including ballast losses a 100w or 1kw LED would be around 50 or 40 lumens per watt including ballast losses.

Not really .  My 9w , 60w equivalent LEDs are 800 lumens (88.88 LPW) and my 15w , 100w equivalent LEDS are 1,500 lumens (100LPW)  The GE 450w 1000w equivalent produces 144.44 LPW : https://products.gecurrent.com/sites/products.currentbyge.com/files/documents/document_file/LEDL065-GE-LED-Replacement-HID-TypeB-1000W-SpecSheet.pdf  Plus , the GE's have a 5 year warranty !

The LED efficacy decreasing is valid only if speaking about efficacy of a one single chip (I mean really the one pie e of semiconductor).
So it is the reason why (excep some special applications) higher wattages are assembled using 10's or even 100's chips.

Then there is a problem of heat management: LED is efficient only if prevented from becoming too hot. Quite easy for a 100mW pocket lantern, but becomes quite a challenge when you want to operate on the same temperature a dense 150W "HPS retrofit" assembly (dense, because it is supposed to match the HPS optics).
But the efficacy is limited by how much the lantern will cost and how big it will be. Using too low power chips means there have to be too many of them, making the assembly too expensive. Plus each individual would need its own lens, so it would be rather big assembly.
So practically the power converged to about 1..3W per chip, that seems to be the sweet spot between cost and performance (130lm/W for a normal CRI80, 150lm/W when CRI could be as low as 60 so the phosphor optimized for efficacy alone).
The 50..100mW per chip allow at least 150lm/W for CRI80, but because they are operated quite hot, it drops to about 140lm/W. The extra efficacy is mainly because their use allows them to radiate from all surfaces, but it is limited to just few W total power (in the "filament" style bulbs mainly for home use; so many designs favor a bit lower efficacy in the 120lm/W in exchange for CRI90+).
Speaking only about decent quality products, not the cheepeese garbage...