LEDs have exceeded the lumens-per-watt (luminous-efficacy) of typical incandescent bulbs by about double.  Latest announcements (January 2008) show LED efficacy ratings exceeds fluorescent technology. 

: ef·fi·ca·cy Pronunciation: 'e-fi-k&-sE Function:  noun  Inflected Form(s):  plural -cies  : the power to produce an effect

Some readers are puzzled over the use of this word efficacy verses efficiency.  "Efficiency" cannot relate a measure of input energy to a unit of light output.  Some readers may not be aware that "light" refers only to the visible portion of the spectrum, and not to other emissions from an emitter.  Efficacy is the appropriate term for comparing a light effect from an electrical input, and is commonly used by lighting specialists and LED Engineers worldwide.  Efficiency would be used for example, to compare power in to power out of a power supply, or, compare the light produced by a bulb inside a fixture to the light emitted from that fixture, etc.


The LED manufacturer rates lumen output at a specific junction temperature, typically 25°C, and the component integrator often uses that lumen figure to promote the new widget.  Few LEDs perform in the real world as they did in laboratory tests.  LED marketing has become clever and caused the lighting industry to think in new terms like "wall plug efficiency."

Unlike previous bulb technology, we cannot easily separate the LED from the end product to measure only that element; we can measure output only of the packaged product.  Here we discuss a few hopeful ideas or rumors, and some of the reality to support or dispel them.

Entrepreneurs may need to re-write a few laws of thermodynamics when showing screw-in LED light bulb replacements.  These inventions need more thought to live up to claims.

Will the consumer settle for less light because it's a novel gadget?  Reducing their lighting requirements will itself save energy, but is that what they intended?  This leaves the market open for a rude awakening - buyer beware?

LEDs may produce a little more light per watt than an incandescent bulb.  But most of the watts they consume is still converted to heat, and if not gotten rid of will drastically shorten their life.  Heat from a LED is not easy to manage, and the LED must operate below about 85°C.  Here's an example of proper LED thermal design for a bulb replacement.

A light bulb is fundamentally a heater that just happens to send out a little visible light.  Yes, a heater, and the socket it's screwed into is designed to protect the base and wiring from that heat.  In other words, the socket is a very poor thermal path for cooling that screw-in LED light bulb.  Something must cool the LED.

Estimate heat exchange to get those lumens and keep T_J below 85°C:
Each watt of LED power needs about 9 square inches of surface area in free air for cooling.  (See Lumileds apps. guide Luxeon Thermal Design Guide  pdf file.)
That 3" post-it-note on your desk is 9 square inches.  Roll it up, fold it & shape it to get an idea of the size. 

If the LED has twice the efficacy of incandescent, then lumen for lumen, a 60-watt light bulb replacement must dissipate 30-watts of heat.  That's 270 square inches of surface metal attached to that screw-in LED device.

To get the lumens and keep the life, you gotta cool that thing.  Of course an alternative is forced air with a fan.

That seems like a practical statement since the traffic signal industry showed over 80 percent energy savings when they switched to LED signals.  But beware.

LED traffic signals saved energy because the signals are monochromatic (single color), and LEDs naturally emit only a single color so there's no extra light generated to throw away.  The extra color spectrum from incandescent bulbs was absorbed in a filter and simply unnecessary. 

The issue of saving energy with LEDs comes down to producing only what the task requires, AND using all the light that is produced.  Competing technologies may produce higher lumens, but it's not easy getting all that light where it's needed, and usually any extra light is just masked.

A typical bulb type luminaire may be only 60% effective, or in other words, a lot of the light it produces doesn't go to the intended target (it becomes light pollution or is merely masked off).  And, the higher efficacy HID bulbs require a ballast to start the arc and regulate the gas ionization.  The ballast may consume 20% of the electrical energy going into the luminaire.

   That's  0.6×0.8 = 0.48
       or Metal Halide = 80 lumens/watt ×0.48 = 38 lumens per watt to the target.

In contrast, the LED luminaire may be 90% effective with a power supply that's 90% efficient. 
   That's  0.9×0.9 = 0.81
          or LEDs = 65 lumens/watt ×0.81 = 53 lumens per watt to the target.

Metal Halide shown because of comparative color temperature.

The die-hard Gas in Glass enthusiasts say LEDs put out a fourth of the light, but there's a lot more to the evaluation than mere efficacy.  Obviously energy savings will not sell LEDs in the illumination business, so maybe those few percent of inefficiency may be converted to savings in maintenance.

True, LEDs are now able to compete with halogen incandescents, maybe a little better.  However, review the math shown above. 

The HID bulb industry has created an illusion that high lumens are required for certain tasks, because they get little better than half those lumens out of the fixture.  Where have all those lumens been going, and at what cost to the consumer?  Is this deceptive?  You bet it is.    See this breakdown on Sodium Lighting effectiveness.

LED Luminaire manufacturers are boasting of about 70 lumens per watt (at 25°C T_J  of course). 

Now the barrier to keep an eye on for the illumination business is not these few percent of inefficiency, it's the LED's high cost per lumen.  HID bulbs can jump in increments of a thousand lumens for mere dollars, whereas LEDs cannot.

On the surface that statement is sort of true.  We don't get a monthly electricity bill for using the Sun's energy, and the process of converting the Sun to electricity produces no harmful side effects.  However the lighting is needed at night when the Sun is not around, so our night lighting is powered by energy saved in batteries during the day.  The batteries contain materials that aren't land-fill friendly, and they must be replaced every few years.  Photo-voltaic cells are the key to harnessing the Sun's rays, but those cells are not cheap.  The solar panels may cost 2 or 3 times that of the light they power, and the amortization may be 10 or more years.

Bottom line may not be clear, but it is clear that accuracy of a catch phrase like that and so many others depends entirely on the application.  Solar powered nighttime lighting can be justified only when there is no electrical service, or, complete independence from the utility company is essential.  It won't likely float when the power source is hydroelectric.

Technology has come into the equation, and with that comes many ways to measure performance and even ROI.  Electronic controls are now involved that can inform the owner of pending failure or lumen depreciation, they can allow remote control of your luminaire.  Can you imagine a communications protocol between streetlights and your central office?  It's here.  Now you need interchangeability, and NEMA can drive manufacturer functional compatibility.

Computer generated Photometric files are here too.  NEMA can draft appropriate performance Standards, and possibly find ways to police deceptive practices.  We used to expect the Photometric file came from an independent lab, but until there are NEMA Type numbers for Luminaires, you should insist on Certified Independent IES Files and Test Lab Report before committing poles and luminaires.

    We can hope that NEMA will raise Manufacturer Standards and Ethics.     Leave your comments here.

Bogus claims like this will make skeptics from the optimists. 

The IESNA (Illuminating Engineering Society of North America) "Standards" provide uniform measurement criteria for comparing lighting fixtures.  They have many lighting categories and types.  They have a category and type number for nearly every type, style and technology of lighting fixture, but the IESNA is not an endorsement of a lighting fixture.

The IESNA has application software to run on most computers that will determine the quantity, and plot the location and spacing for luminaires to provide a lighting criteria, levels and uniformity based on your selection.  For this, a Photometric file (a tabular format of typically 180° x 360° angular luminance measurement) is required for the particular luminaire.  For this layout software, your Manufacturer must have an IES file available for the luminaire, and it should be a file of actual measurements from a certifed independent lab.
  

LEDs have a reputation for long life, but they are also easily abused.  Light Emitting Diodes are more fragile than some designers are aware, but electrical or thermal deficiencies in their fixture design will not often cause a sudden failure.  Too much current or too much heat for example, could result in a 50,000 hour part lasting only 10,000 hours. 

High quality, high output monochromatic LEDs can provide 70% of their original lumen output at 100,000 hours IF they are not abused.  Note that Monochromatic is stipulated because the Phosphor in most white LEDs can degrade faster than the LED junction itself.  This can be caused from thermal overload or from excessive ultra-violet exposure.  Some manufacturers have overcome these problems, now the integrator must adhere to good science.

Time will tell.

Correlated Color Temperature Chart

Color Rendering Index  (CRI) indicates how well a test source renders eight standard colors of intermediate saturation, when compared to a reference lamp of the same color temperature.  Lab measured CRI is a comparison against a spectrally continuous red-weighted reference.  Field conducted CRI tests are subjective with Human observers when luminance levels are below 3 cd/m2. 

The industry is discovering that CRI is not the best metric for comparing LED light sources, especially at Mesopic levels.  Originally developed in 1964, this index is based on outdated color models and assumes illumination sources with broad spectral distributions, whereas LEDs are narrow-band sources.  And, nighttime lighting requirements fall primarily in the Mesopic range where our color sensitivity shifts with luminance, and there is no defined index.  Several standards bodies are addressing this deficiency, and in the interim, Color Temperature may be the most suitable tool for comparison because it is independent of observer subjectivity.
        


Correlated Color temperature (CCT) defines a color as the temperature in degrees Kelvin that a "black body" source must reach in order to produce that same color.  CCT describes the dominant color without regard to Human visual response or the source technology and is more appropriate for comparison of visual effectiveness at lower light levels and among different technologies. 






Which measurement shall I use:  Candela or Lumens?

The user community, contractors, agencies and municipalities, have established lighting criteria so they can maintain a quality of illuminance through time.  Specifying a lighting fixture by lumens is merely requiring a power consumption and not a lighting criteria.  The size of an area illuminated to a specific value does define a minimum amount of lumens, but lightmeters measure footcandles or lux, not lumens. 

Properly managed LEDs in a suitable luminaire have been field-measured as placing 3-times more foot-candles on the ground per watt than other technologies, partly because they can be directed where most suitable, and partly because they produce light only where needed. 

Be careful of the Lumen Legacy.  Select fixtures based on application footcandles or lux rather than lumens.