LEDs headline at SALC, speakers predict significant efficiency gains (MAGAZINE)

Oct. 20, 2011
SSL technology dominated both the sessions and exhibits at the annual Street & Area Lighting Conference, reports MAURY WRIGHT, and LED-based lighting that is already succeeding based on efficacy will soon deliver significant additional power savings.
Exhibit hall at SALCThe annual IES (Illuminating Engineering Society) Street & Area Lighting Conference (SALC) took place Sept 19-21 in New Orleans, LA and the first speaker, Entergy Services business systems manager Bob Olsonoski, said “We’re not against LEDs. We just don’t know what to do with them. We don’t know how to price them yet.” Olsonoski likely felt like the Lone Ranger through the remainder of the event because LEDs were central to virtually every other presentation, and dominated the exhibit hall. Despite higher upfront costs, LED-based solid-state lighting (SSL) is winning in outdoor applications based on energy efficiency and the inherent controllability of the technology. The prevailing theme of the conference was that energy savings will escalate significantly through more efficient LEDs, better lighting that can be operated at lower levels, and standards and technologies that minimize over-lighting.

The IES limits the exhibit area to 50 booths and around ten companies took dual booths so there were even fewer companies displaying products. One booth included induction street lights and one had high-pressure sodium (HPS) street lights. More than 20 featured LED street and area lights. Even exhibitors such as Philips Lighting, Osram Sylvania, Acuity, and Cooper that sell legacy lighting products only exhibited SSL.

Why the focus on SSL? Edward Smalley, the director of the US Department of Energy (DOE) Municipal Solid-State Street Lighting Consortium (MSSLC) and the manager of street light engineering at Seattle City Light, pinpointed the reason. Smalley showed a graph from the DOE’s latest SSL Multi-Year Program Plan (MYPP) that charts luminous efficacy against time for various light sources (Fig. 1). While efficacy is slowly increasing for HID and fluorescent sources and has been for 70 years, SSL (both LED and OLED) is on a steep ramp.

FIG. 1. Today HPS and low-pressure sodium (LPS) sources are still more efficient than SSL. But that advantage won’t last long. Moreover, adaptive controls and dimming can deliver energy savings for SSL relative to HPS and LPS sources. And as we’ll discuss shortly, broad-spectrum LED light is simply a better match for the physiology of the eye than are HPS and LPS sources. These were all prevailing themes at SALC.

Of course SSL still has to overcome steep upfront cost, although that premium is certainly dropping rapidly. John Curran, president of LED Transformations, presented a graphic that precisely describes the LED value proposition (Fig. 2). It’s the combination of long life and energy efficiency that provides the LED value proposition.

Improving LED sources

Mark McClear, global director of applications engineering at Cree, discussed the state-of-the-art in LED components and the near-term outlook for improvements. Cool-white LEDs at 6000K CCT (correlated color temperature) are readily available with an efficacy of 160 lm/W. McClear said that a luminaire design typically suffers a 10% loss due to thermal issues, a 10% loss due to optics such as lenses, and a 15% loss due to driver efficiency. So cool-white luminaire system efficacy is a bit over 100 lm/W.

At the other end of the LED CCT spectrum, 2700K warm-white LED efficacy is 115 lm/W, resulting in a system efficacy of around 75 lm/W. McClear said, “I really like the 4300K and 4100K LEDs.” At 4100K system efficacy is 93 lm/W and that CCT is preferable for many people relative to the 6000K LEDs that have more blue energy in the spectral distribution. McClear said the cooler temperatures work better from an economic perspective because you can use fewer LEDs in a luminaire design.

McClear and others including the DOE expect a continued increase in efficacy. McClear pointed out that the first DOE MYPP projected an efficacy plateau at around 150 lm/W. That plateau has been moved to 250 lm/W in the latest MYPP update issued earlier this year.

FIG. 2. McClear said that LED luminaire efficacy has improved from 50 to 90 lm/W, at maximum drive current, over the last six years (Fig. 3). He projected system efficacy for sources in the 4100K CCT range to hit 120 lm/W within the next two to three years. He added, “LEDs will be the most efficient light source available.”

It’s also noteworthy to mention that prices are dropping at the same time that LED components are improving and volumes are increasing – a recurring trend in the semiconductor industry. McClear said, “The semiconductor industry has always been a massive solution looking for a problem.” The message is that the same juggernaut that delivered cheap PCs and cell phones will drive lighting going forward, and the escalation in LED manufacturing has begun. McClear said that more MOCVD (metal-organic chemical vapor deposition) reactors, used for epitaxial growth in LED manufacturing, have been installed in the past two years than existed previously.

The fact is that LED cost has already diminished significantly in terms of the bill-of-materials (BOM) cost in luminaires. According to McClear, LED cost accounted for around 70% of the BOM in 2008 and has dropped typically to around 25%. The driver is now the biggest part of the BOM, but McClear said, “The driver community is now just as engaged [in SSL] as the LED community.” And the drivers are largely comprised of semiconductors so prices should drop as efficiencies increase.

CCT and broad-spectrum light

While McClear had noted the economic advantages of cooler CCTs, other speakers described the benefits of white light with a broad spectral distribution – typical of today’s LED sources. Ron Gibbons of the Virginia Tech Transportation Institute, said, “White broad-spectrum light may provide equivalent task performance at lower illuminances than a less-broad source.” Gibbons presented a graph that depicted the luminosity function of the human eye both for bright photopic and dark scotopic conditions. And he showed the energy peak of a LPS light that falls almost completely outside the spectrum of scotopic response by the eye at night.

Across CCTs that range from 3500K to 5000K Gibbons showed that LED sources have significant spectral content in both the photopic and scotopic bands. Gibbons said, “The physiology of the eye lends itself to broad-spectrum sources.”

FIG. 3. The ultimate goal of Gibbons' research is to determine whether white light sources can be operated at lower levels than have been conventionally required, and still provide driver and pedestrian safety. Indeed many people believe the world is significantly over-lit and reducing light levels would provide direct energy savings.

Of course there are both scientific and emotional challenges to white-light in general and a move to lower levels. Many people including several questioners in the SALC audience insist that driving under yellow- and orange-tinted HPS and LPS lights is more relaxing than driving under white lights. Gibbons, however, insisted that every study conducted finds that “people like white light better” although they may not realize it until going through a controlled experiment.

Gibbons also said that white lights render colors better. And color contrast is important in enabling the eye to detect objects, especially in the peripheral vision, he said.

Leveraging light research

Gibbons and Rick Kauffman, of Kauffman Consulting and the chairman of the IES committee working on the latest update to the ANSI RP-8 standard for roadway lighting, both discussed how the research on lower light levels will be applied in the near term. The RP-8 update due imminently will allow for lower light levels in some cases in mesopic conditions (relatively low light levels where the eye combines photopic and scotopic response).

The new standard will specify calculations called Effective Luminance Factor (ELF) and Effective Luminance Multiplier (ELM) that rely on photopic and scotopic luminance values: these are presented in a table relative to various light sources. The math is beyond our scope here, but Kauffman presented a relatively simple example in a case where a light source has a scotopic to photopic ratio of 2 and photopic luminance of 0.3 cd/m2. In such a case the required minimum light level could be reduced by 16.6%.

There is a caveat to the change to RP-8. For now the standard prescribes that the lower light levels can only be used on streets where the speed limit is 25 mph or less. At such speeds, drivers don’t even need street lights according to Gibbons because the headlights illuminate a distance greater than the stopping distance of the car. Street lights in such cases are primarily intended for pedestrian safety. Gibbons has begun another research project that will determine if lower light levels are also safe at higher speeds.

Minimizing over-lighting

In addition to discussing the ability to reduce light levels, Kauffman addressed the larger issue of over-lighting and wasting energy. We routinely install lights that operate at higher light levels than necessary to compensate for light loss. Lighting specifiers typically calculate a light-loss factor (LLF) when planning a project. The calculation can be quite complicated, said Kauffman, and includes accounting for thermal issues, driver or ballast issues, and even ambient temperature at an installation. But primary factors are lumen depreciation and luminaire depreciation caused by dirt.

FIG. 4. The RP-8 standard includes a graph for dirt depreciation factors. For very clean environments the factor can be as high as 0.9 over 8 years. In very dirty areas the factor can be in the 0.8 to 0.9 range for one year but as low as 0.3 over 8 years. Kauffman described DOE tests that have measured dirt depreciation of 3.7% to 5.3% per year but said more testing is needed.

Lumen depreciation is a well-known phenomenon and has been documented for a number of light sources. Ironically many have questioned LED performance over time, but LEDs actually provide superior lumen maintenance to most other sources with only HPS matching SSL. The graph in Fig. 4 depicts the typical advantage LEDs offer in terms of uniformity over time.

The specifier will often utilize a lumen depreciation factor of 0.7 for LEDs essentially tied to the widely-accepted definition of the L70 LED lifetime that describes how long a light will maintain 70% of the initial lumen output. Multiply lumen depreciation of 0.7 by dirt depreciation of 0.9 and you get 0.63 as the LLF. This is used to de-rate lumen output. A 100-lm source would be used in an application requiring 63 lm.

LEDs, however, offer our best chance yet of minimizing over-lighting. Some companies are designing luminaires that slowly increase drive current over time in a way that matches the projected lumen depreciation curve. And it turns out that L70 may not be the right lumen-depreciation factor with LEDs getting better and the new IES TM-21 standard available to project LED life.

Constant light output

David Baum, director of sales and marketing at Philips Roadway Lighting, addressed constant light output relative to the company’s Fortimo linear LED module. Baum compared the Fortimo light with a legacy source and a typical LED source (Fig. 5). The legacy light provides significantly too much light each time it is relamped over time. The legacy LED source provides too much light initially and gradually degrades over time.

Rick Kauffman, Kauffman Consulting The Fortimo design gradually increases drive current over time thereby maintaining the target lumen output – although that also means the power consumption gradually increases as well. Baum showed an example where luminaire system power increased from 26W to 31W over 50,000 hours. But he said a conventional source with comparable light output would require 38W.

In the case of TM-21, meanwhile, a solution to reduce over-lighting may be an unintended consequence. TM-21 was in the works for a long time as we covered in a feature earlier this year. The idea behind the standard was development of a mathematical model that allows accurate projection of LED lifetime. LEDs rarely fail over time but rather degrade to the point of being ineffective. That led to concepts such as L70 life. TM-21 provides a way to take the reams of data that are produced in LM-80 LED component testing and produce projections across a range of scenarios.

The details of TM-21 are beyond our scope here, but let’s examine the basics. TM-21 utilizes data from a 5000-hour window of LM-80 testing. If a particular LED model has been tested for the requisite 6000-hour minimum required by LM-80, then TM-21 uses data collected from the 1000-hour mark forward. If an LED has been tested for 10,000 hours, then TM-21 data uses data collected from the 5000-hour mark forward.

TM-21 and over-lighting

TM-21 results are reported in hours alongside a descriptor in the form Lxx(Yk) where Lxx is the level of lumen maintenance and Y is the number of hours tested. A rating described by L70(10k) would infer that the LED would maintain 70% of its initial light output and was based on 10,000 hours of LM-80 testing. The TM-21 methodology delivers two results, one called calculated and one called reported. The former is the calculated output of the TM-21 math. The latter is limited to the lesser of the calculated life or 6 times the number of LM-80 test hours. An LED tested for the 6,000 hour minimum can have no greater reported life spec than 36,000 hours.

FIG. 5. You can calculate TM-21 results for any lumen maintenance value you desire. For example, Cree’s McClear showed an actual example where an L70(10k) test included a calculated life of 290,000 hours. And while TM-21 will be broadly used to project life, here’s how it comes into play in reducing over-lighting. The example McClear showed also included a L80(10k) calculated life of 186,000 hours and a L90(10k) calculated life of 94,200 hours. McClear’s point is that maybe we should use 0.9 as a lumen-depreciation, life-loss factor rather than 0.7. McClear said “LEDs are unshackled from L70” and the result can be additional energy savings.

Of course the LEDs are only one part of the SSL system-life puzzle. For example, McClear mentioned things like gaskets and paint that may not last 100,000 hours and of course the driver is an issue. Philips’ Baum said quality drivers have a maximum life of 100,000 hours. But he said that driver life declines rapidly when case temperatures exceed 65°C.

Adaptive controls and dimming

Not surprisingly, adaptive controls and dimming was a popular topic at SALC since dimming lights during periods of low activity can compound energy savings. And LEDs are dimmable to a fine level of granularity with commensurate energy savings whereas other light sources lack that attribute. The talks included the need for standards, ongoing field trials, and activity on controls within the DOE MSSLC.

Rob Gibbons, Virginia Tech Transportation Institute Let’s start with the MSSLC, which launched a control task force about one year ago. Tod Rosinbum from the city of Portland is a member of the task force and described a wish list that is being molded into a model specification that municipalities and communities can utilize in specifying control systems. The document will be very similar in concept to the Model Luminaires Specification that the MSSLC released in draft form earlier this year and that is due for final release imminently. The controls specification will be released in beta form later this year for review and a final version is due next year.

Rosinbum said on/off, dimming, and scheduled-based control are widely desired by consortium members. The members also want diagnostic capabilities, energy measurement for billing purposes, and automation of the work-order process for repairs. He also said that the membership is universal in wanting to own and control the data in-house. The most complete controls solution on the market today, by contrast, is the Acuity Roam system in which fees for maintaining the system and managing the data are part of the Acuity business model.

Michael Poplawski, another controls task-force member from the Pacific Northwest National Laboratory (PNNL), addressed some obstacles and lamented the lack of network standards that can be used in a wireless network. He noted that Zigbee is a possibility but as we covered last year, Zigbee doesn’t include the definition of a complete protocol stack or a layer specific to the lighting application. Poplawski also noted that work needs to be done to ensure that different luminaires operate similarly when dimmed. He showed a graph of the performance of three luminaires that revealed noticeable differences in light level and power consumption when set to the same level by a 0-10V controller. But he said that operation could be normalized with standardization.

Roadway reclassification

Laura Stuchinsky heads the controls task force and is also managing a pilot controls project in San Jose, CA where she serves as the sustainability officer in the city’s Department of Transportation. San Jose has been a leader in trialing control technology and dimming. Indeed the city’s work has been seminal in pushing the concept of reclassifying roads at night so that light levels can be reduced. Table 1 shows an example where light-level is reduced by as much as half later at night when there is little pedestrian activity.

San Jose had previously worked with Pacific Gas & Electric (PG&E) to negotiate a lower fixed tariff for LED lighting. The city subsequently asked for even lower rates based on operating the lights at reduced levels during portions of the night. PG&E didn’t offer to lower the rate now but agreed to participate in a 3-year pilot program.

TABLE 1. The utility mandated that the luminaires include a power meter with 2% accuracy. Moreover PG&E has insisted that the metering include the power consumption of the control electronics – not just street-light power. San Jose must also monitor the power used by the wireless gateway that connects to a group of lights. San Jose’s work is being used as a guideline in the development of the MSSLC controls specification.

The combination of controls and baseline LED efficiency appear to be the right match for broad deployment of SSL. And there are more savings coming through better LEDs. There is also more potential for savings in system design. Earlier this year, Tom Geist of the Electric Power Research Institute (EPRI) contributed an article to our publication on LED street-light field trials, and at SALC presented some additional data. Geist said EPRI has documented energy savings in the range of 25-70% in different trial installations. But he added that there is other low-hanging fruit in terms of savings. He said driver-efficiency improvements, temperature compensation in fixtures, and better quality control by fixture manufacturers could deliver more than 10% in additional savings.

Also it's important to realize that LEDs are being held to a higher standard. A couple of times during Q&A sessions at SALC audience members asked why there is no standard such as LM-79 with which legacy lighting must comply. Smalley of the MSSLC said, “We are asking more out of LEDs than we ever have of HID.”