Statistically averaging of failures of fluorescent lamps in commerical lighting.
Statistically averaging of failures of fluorescent lamps in commerical lighting.
I am addressing this in context of large scale commercial lighting implementation. Not just a storage room with one or two fixtures. Fluorescent lamps are used in a large quantity and the failure model is very well established.With a large number in use, the failure rate is very very very predictable and well spread out. Of course, ballasts can fail and that will wipe out the entire fixture, but this does not make a difference if you're using fluorescent or LED. the only difference is the name.
Yes, I'm familiar with the N80, but your spin on the technology is completely backwards. They actually use less power the first few years, resulting in even more savings.
The baby thinks you're driving in reverse if he's in a rear facing seat
You are confused, because you were told with a spin of sales or you read something with a spin of sales.
There are legacy street light systems that use a constant current 6.6A circuit. As filaments thin out, the voltage goes up and wattage goes up. On a constant voltage circuit, as filaments thin out, current drops and wattage goes down. You're confusing variation compensation with degradation compensation. A constant wattage driver can hold the lamp at the same wattage, but you will still lose lumens, because the evaporated filament slowly tints the glass. If the filament power was raised to perfectly counter the increasing darkness of glass, you will have a constant output and this would be called a DEGRADATION compensation.
The lamp power regulation is entirely up to the driver. Incandescent, standard magnetic ballasts,some electronic ballasts and some LED lights (especially earlier re-wire retrofits) are affected by normal line voltage variation. The extent of variation depends on the ballast. CWA type MH ballasts have a much better lamp wattage regulation. Most modern electronic HID/LED/fluorescent ballast-driver-power supplies are immune from wattage variance within the range of ordinary voltage variation (+/- 10%). Standard magnetic HID ballasts see the most impact of voltage variation. Some designers have included minus 10% voltage operation in design if the system was seriously overspec'ed and the primary concern was not being caught dead under delivering under any circumstance and waaaay over lit most of the time.
Metal halide, mercury vapor and and LED (as they're applied now) have the same pitfall and that is the extremely high level of lumen depreciation during their designed useful life. The lights need to live up to the required performance range for years. Not just for one week in an LED product booth at the Light Fair. So, if the maintained light level requirement is 1 at half way into useful life, you need to up size the initial output accordingly so you maintain 1. For HPT8, remaining output factor is 0.94 mid way and 0.92-0.93 at end of life (70% of rated life when you lose around 10% of lamps burn out).
Take a look a this:
http://www.grainger.com/ec/pdf/Philips-ICN4P32N-Spec-Sheet.pdf
Most T8 ballasts bump up the lumens on remaining lamps if one of the lamps fail. So, when a 4 lamp fixture loses one lamp, the ballast boosts the power by 12.4% on the remaining 3. you lose some lumens, but you lose some wattage too. So, if you have 10 fixtures with 40 lamps, you can expect to see 4 lamps out. So, with six fixtures running 4 lamps and 4 fixtures running three lamps and each lamp still giving 92%:
If the end of life cut off is 90%, you only need to oversize about 5%.
Initially you had 10 fixtures x 4 fixtures x 100% of starting lumens = 40 units
At end of life, I estimate...
6fixtures x 4 lamps x 0.92 = 22.08
4 x 3 x 0.92 x 1.124 boost(BF bumps up from 0.89 to 1.00 on remaining lamps) =12.41
Those add up to 34.49 units
So, the estimated lumen remaining is 86%.
The system uses 1110W initially (111W x 10)
At end of life...
111W x 6 fixtures = 666W
93W x 4 fixtures = 372W (fixtures that are running at lower wattage because one lamp is out)
1038W
93.5% of original kW demand. 6.5% temporary demand offset, until relamped.
86/93.5 = 0.92... which is the same as lamp's lumen maintenance.
0.90 required lumen maintenance at end of life /0.86 remaining = 1.046. 4.6% overspec needed. I'd say that this is a small enough level that it's not worth the expense of a custom programmable micro controller compensation system.
Light loss due to dust needs to be addressed regardless of lamp technology, so it is excluded from this discussion...
Using LEDs, 90% required divided by 70% remaining at end of life (50,000 hours) vs (56,000 hr for 80,000 hour Philips 2XL lamps used to 70% rated life),
you need to spec out 28.6% higher initial lumens.
This is where nLight comes in. You use a programmed controller that is calibrated to the ASSUMED degradation characteristic of LED decay and start the system dimmed down to 78%. As the actual LEDs develop tolerance to electricity and need more and more dosage to produce the same effect, the computer raises the dosage until the maximum allowed is reached. The maximum dosage is continued and the fixtures are considered end of life when the final output no longer meets the requirements. If the real life degradation pattern and predicted calibration profile deviate significantly, constant lumen maintenance may not work as planned. Unless all the fixtures are linked by a data bus, it's going to be a lot of work to fix them.
iwire, what are you doing today? Reflashing the firmware in every light fixture at the store.
It produces kWh saving compared to LED fixtures without degradation compensation, but
I do not believe it will yield a kW demand saving for the utility infrastructure, because the increase in demand is predictable and it is simply offset a few years.
I think we both have an idea of how it works, but you see my view as backwards, because slick sales played mind games put it in backwards in your mind. Metal halides and LEDs suffer the same shortcoming of high lamp lumen depreciation percentage during the useful lifespan.
