Report of the Panel on Benefits of Lighting R&D
The report presented in this appendix is the product of an expert panel convened by the National Research Council’s (NRC’s) Committee on Prospective Benefits of DOE’s Energy Efficiency and Fossil Energy R&D Programs. The charge to the panel was to complete the preliminary prospective benefits matrix using the guidance from the committee described in Chapter 2 and Appendix E. The panel had considerable flexibility in determining how to fulfill this charge. This appendix summarizes the panel’s findings on the methodology and process.
Although a member of the committee chaired the Panel on Benefits of Lighting R&D (see the section below entitled “Panel Member Biographical Information”), the full committee did not participate in the work of the panel. Rather, the committee reviewed the findings of all three of the panels formed for the purposes of this study1 as the empirical basis for developing the methodology and process that the committee recommends in Chapters 3 and 4 of its full report. As a result, the committee’s recommendations may not reflect the specific suggestions or findings of an individual panel.
The committee wishes to emphasize the following points:
The panel reports were developed for the sole purpose of developing the methodology. As a result, the panel reports are not complete or systematic evaluations of program benefits and should not be interpreted as such.
OBJECTIVES OF STUDY
The NRC’s Committee on Prospective Benefits of DOE’s Energy Efficiency and Fossil Energy R&D Programs is developing a methodology to estimate benefits that may be provided by current R&D programs. The lighting program, part of DOE’s Office of Energy Efficiency and Renewable Energy (EE), was one of three programs selected to test the methodology.
The lighting R&D program has been operational since the mid-1970s and has expended between $55 million and $60 million to improve lighting technology. Electronic ballasts, one of the DOE lighting programs, was a case study for the NRC’s 2001 retrospective benefits study and was shown to have realized billions of dollars in benefits (NRC, 2001). The Solid State Lighting (SSL) Program, the main focus of this panel, was initiated by DOE in FY 2000, with funding at about $2 million. In FY 2004, the program will spend $10 million on solid state lighting, about 80 percent of the total budget for lighting R&D at DOE.
The NRC-appointed panel of experts to review the lighting program and apply the committee’s methodology was drawn from industry (conventional and solid state lighting), academia, state government, and the energy efficiency community. The panel heard presentations from DOE and its contractors on the Conventional Lighting Program and the SSL Program and on the use of the National Energy Modeling System (NEMS) model to determine benefits for the SSL Program. It also heard from Terry McGowan, a lighting designer, and M. George Craford, a panel member and chief technology officer at Lumileds Lighting. The panel estimated the probability of technical, economic, and market success; reviewed the committee’s matrix and modified it; then used the matrix to complete its analysis.
The following sections summarize the DOE lighting program and budget, describe the benefits analysis, and summarize the panel’s findings, including the strengths and weaknesses of the methodology and recommendations to improve it.
SUMMARY OF DOE SOLID STATE LIGHTING PROGRAM AND BUDGET
Lighting has long been a prime target for R&D at DOE because of its high energy demand (see, e.g., Ross, 1978). Despite a dramatic improvement in lighting energy efficiency since the late 1970s, about 30 percent of the electricity used in buildings (an average of commercial and residential use) is for electric lighting.2
The overarching criterion for evaluating all projects funded by the DOE’s Office of Building Technologies (OBT) is energy efficiency, which results in decreased energy consumption (measured in quadrillion British thermal units [quads] or kilowatt-hours [kWh]). For the lighting program, energy efficiency means improving the technology of lighting so that lamps produce more lumens per watt (lpw) of electricity. OBT’s role is limited to supporting R&D. The lighting program is a success if the technology it supports is used in commercial products, but industry is responsible for designing and building the products. A secondary goal of OBT is to help American companies stay competitive with foreign companies, which benefit from the large investments provided by their governments, estimated to be $50 million per year and growing.
The lighting program’s main focus is on solid state lighting (SSL), including both light-emitting diodes (LEDs) and organic light-emitting diodes (OLEDs) (Wessner, 2002).3 Colored LEDs are already in the commercial market in high volume (e.g., brake lights on automobiles and other directional colored lights), and white devices are being produced in lower volumes for niche applications. LEDs are not yet competitive for purposes of general illumination. OLEDs are less advanced but show promise for use in highly diffuse lighting involving large, thin-film panels on walls or ceilings.
The goal of the DOE SSL Program is to develop highly efficacious (high lpw), reliable, low-cost, high-color-quality white light sources that will displace conventional lighting in both commercial and residential buildings. Higher efficacy is key to both energy savings and cost reduction and is therefore the most critical objective. Specifically, the SSL goal is 150 lpw by 2015 for commercially available products. The SSL core technology funding plan is focused primarily on improving efficacy, but the additional goals of cost, reliability, and light quality will be almost as important in ensuring commercial success.
The OBT has sponsored workshops with industry, government laboratories, and universities to establish a technology roadmap for SSL technology. The roadmap includes program objectives, barriers to achieving those objectives, and research areas that should be pursued in order to overcome those barriers (OIDA, 2002). Key tasks are to develop the following: improved materials for LEDs and OLEDs, novel device structures for high power and improved light extraction efficiency, and packaging and wavelength conversion technologies for improved efficacy and device lifetime. The OBT is also providing funding for product development that includes luminaire design and materials, as well as intelligent electronics for LEDs and OLEDS.
Thirty projects have been identified for funding in the Solid State Lighting Program. Approximately two-thirds are devoted to support of LED development. LEDs and OLEDs are quite different technologies, and there is little overlap between the two research programs. The justification for pursuing LEDs and OLEDs in parallel and the appropriate distribution of funding will be evaluated on an ongoing basis through the program.
The entire lighting program within the OBT had been at a level of approximately $4 million per year for the past 10 years. Congress provided significantly greater funding ($7.75 million in FY 2004) for solid state lighting and is expected to increase it again in FY 2005 to at least $10 million per year. The justification for this increased funding stems in large part from the projected energy savings that would be realized by accelerating the development of solid state lighting.
However, the SSL goal of 150 lpw by 2015 is predicated on the program’s receiving a budget of $50 million per year for 10 years. Such a big increase is by no means certain, so the panel requested an analysis of technological progress and market penetration if only $25 million per year were available. As might be expected, technological and economic progress would be slower in the latter case, with SSL achieving only 102 lpw by 2015. The impact of the two sets of assumptions is discussed in the next section.
BENEFITS ANALYSIS OF THE SOLID STATE LIGHTING PROGRAM
Prospective benefits of DOE’s SSL Program are estimated using the generalized matrix framework developed by the full NRC committee. Potential benefits are categorized as economic, environmental, and national security. In the case of lighting, all benefits derive from reduced demand for electricity as a result of the use of more efficient lighting technologies.
The matrix assumes that SSL benefits would be calculated under three scenarios: a baseline scenario, a scenario involving high oil and natural gas prices, and a scenario including carbon constraints. The lighting panel, however, did not find the second and third scenarios useful for its analysis.
High oil prices make virtually no difference to the price of electricity, the only energy form used for lighting, because almost no electricity is generated from oil. Higher natural gas prices and carbon constraints would raise the price of electricity, which would increase the benefits. However, as calculated below, the benefits even under baseline conditions are already high enough to justify the SSL program. Hence, there seemed little reason to extend the analysis.
Gross benefits—the economic, environmental, and national security benefits that accrue from the use of improved lighting technology, as calculated prior to the application of any weighting function—are estimated by DOE using two models: (1) the U.S. Energy Information Administration’s National Energy Modeling System (NEMS), which projects out to the year 2025, and (2) MARKAL,4 which can extend the analysis further. Modeling analysis assumed that the SSL program meets its goals and that SSL technology competes in the marketplace for new and replacement applications. This type of analysis is routinely conducted by DOE to support its programs.
The panel considered the benefits at the following three levels of technological success:
At the first level, DOE’s goal of attaining a lamp efficacy of 150 lpw is met by 2015, along with other necessary characteristics, as discussed below. This very ambitious goal would require several technological innovations and inventions. It is very unlikely to be met without a sustained, high level of R&D investment.
At the second level, DOE falls short of its goal, but SSL R&D nonetheless results in improved lamp efficacy of 125 lpw by 2015. Benefits at this level should still be significant.
At the third level, research progress is slow and lamp efficacy rises to only 100 lpw by 2015, which is close to the efficacy of today’s linear fluorescent lamps but much higher than that of incandescent bulbs.
Two sets of analyses were provided by DOE: one for the full-budget program ($50 million per year for 10 years), which leads to 142 lpw in 2015, and one for a reduced-funding program ($25 million per year for 20 years), which results in lamps of 102 lpw in 2015 and 146 lpw in 2025. The technical results of these analyses are shown in Figure F-1.
Neither analysis corresponds exactly to the technical goals that the panel estimated. The modeling analyses were not available in detail until after the panel had completed its meetings. In fact, the reduced-funding program was not analyzed at all until later. Nevertheless, the full-funding results are reasonably close to the 150 lpw estimate, and the reduced-funding results are even closer to the slow-progress estimate. Therefore, these probability estimates are applied to the benefits analyses below.
NEMS calculates the penetration of new technology by competing it against other technologies, whose characteristics might also improve over time with R&D. DOE first ran a base case without SSL as an option, then a case with it. The benefits, in reduced electricity consumed by the U.S. economy, were then said to be the difference between the two cases. However, DOE’s assumptions for competing technology were essentially currently available technologies, with improvements primarily resulting in lower costs rather than improved efficacy. The panel did not accept this baseline, as it ignored much of the research on SSL in other parts of the world and continued improvements in traditional (non-solid-state) lighting. Successful development anywhere is likely to mean improved lamps available in the United States.
Therefore, at the panel’s request, DOE ran full- and reduced-funding cases that were modified with 5-year delays in the rate of improvement of SSL technology; this 5-year delay is assumed to result from the absence of the DOE program. The full-funding/5-year-delay case resulted in generally faster development than the reduced-funding/no-delay case, suggesting that the former is not a good representation
of the zero-budget baseline. Therefore, the panel used the reduced-funding/5-year-delay as the baseline for both budget cases (i.e., full- and reduced-funding). As shown in Figure F-1, efficacy increases most rapidly under the full-funding case, but all trajectories have nearly converged by about 2040. Even though the programs themselves are concluded by 2015 or 2025, they provide acceleration to the technology that continues to provide benefits. The benefits analysis is continued to 2050 to ensure that the full life-cycle benefits of the technology are included.
It should be noted that DOE did not run MARKAL for the full-funding case. Therefore, the values listed for the 2030-2050 period for the full-funding case were extrapolated, using the reduced-funding case as a model and taking into account the earlier saturation that would be likely to occur. These values should not be taken as authentic projections! In addition, DOE did not directly provide the panel with the data on the modified baseline. Rather, it provided one curve of benefits (measured in billions of dollars) calculated using the original baseline, for both the full- and reduced-funding case, and a second benefit curve calculated using the 5-year-delay baseline. Therefore, to use the reduced-funding/5-year-delay baseline for the full-funding analysis, it was necessary to calculate it by subtracting the reduced-funding/5-year-delay benefits from the reduced-funding/no-delay benefits (see Figure F-1).
Given the lack of information available on R&D programs supported by private companies and other governments, the 5-year-delay baseline appears to be the best way to handle the different potential development rates. These results are shown in Tables F-1 and F-2. In addition, the (undiscounted) economic benefits of the reduced-funding analysis are shown in Figure F-2 to illustrate the general pattern. The data for Figure F-2 are taken from the first three rows of Table F-2.
The methodology relies on expert opinion to estimate the probabilities of the program’s achieving its technological goals and market penetration projections. The benefits that should be credited to DOE are the gross benefits multiplied by both of these probabilities.
As discussed above, success may be achieved at different levels. The panel estimated the probability of achieving three levels of technical success—100, 125, and 150 lpw—by 2015. Each panel member was asked to estimate the probability of success for each of the goals, given a particular level of annual funding. The results are plotted as bars in Figure F-3. The bar closest to the origin shows the results for a scenario in which the goal is to achieve 150 lpw with a research budget of $10 million per year. The length of the bar says that the committee members’ estimates ranged from 0 percent to 40 percent, with an average value of 10 percent. As funding levels increase, moving to the right on this graph, the probability of success increases. As the goal is set lower, moving upward in the graph, the probability of success increases.
Table F-3 shows the same data as those presented in Figure F-3. The top row shows the estimated success if the goal is set to 150 lpw by 2015. If the yearly funding level is $10 million per year, the estimated average probability of success is 10 percent, with a range of 40 percent to 0 percent. If funding is increased to $25 million year, the chance of success rises to 25 percent, with a range of 50 percent to 10 percent.
Each probability implicitly assumes that other lighting characteristics necessary for market adoption are also met—for example, costs will be reasonable and competitive, and longevity and color will be equal to or better than the next-
TABLE F-1 Benefits Corresponding to $50 Million per Year Budget (Full Funding)
TABLE F-2 Benefits Corresponding to $25 Million per Year Budget (Reduced Funding)
best alternative technology. These assumptions are critical to the market acceptance and penetration levels modeled by DOE in the derivation of program benefits. The importance of these attributes was confirmed in panel meetings with DOE staff. Specifically, SSL technology must be cost-competitive at approximately $33 per thousand lumens (DOE
estimates that costs may drop below $4 per thousand lumens); it must have an operating life of 50,000 hours (DOE aspires to 100,000 hours); it needs a reasonable light degradation (70 percent of lumen output at the end of life); and a color rendering index (CRI) of between 80 and 100, comparable to or better than that of fluorescent lighting today, which has a CRI of 80. Other SSL attributes, deemed necessary for SSL to compete effectively with alternative technologies, include the following: having the necessary supporting building and lighting infrastructures available for installation, having known and standardized equipment specifications, and having information available to the lighting industry and consumers to support interior design needs, to name a few.
Each panel member estimated the probability of achieving the technical goals at each level of success and each budget level. The panel’s average for each of the nine probabilities is listed in bold in Table F-3. The column to the right of the average lists the high and low estimates of probability assigned by individual panel members. The same information is shown in Figure F-3, with the high and low probabilities surrounding each average.
The last column in Table F-3 shows the panel’s estimate of the probability that the NEMS market penetration levels can be met or exceeded if the SSL technical goals are met. The actual penetration for the technology could be higher than the NEMS model run, so the expected value analysis conducted using the 70 percent shown in the table should not be viewed as an upper bound on the benefits. It is, rather, a point at which the value could be higher or lower with a 70 percent overall likelihood of being reasonable.
The probabilities in Table F-3 are all based on wide ranges of estimates that are averaged. The averaged estimates are not intended to be taken as accurate. Another panel might produce significantly different numbers. A range of estimates would better represent the collective opinion of the
panel but would be more complex to apply to the benefits and to depict in this report.
Most panel members think that even with only a minimal DOE program the 150 lpw goal might be achieved eventually. DOE sponsors only a small fraction of worldwide SSL R&D but it sponsors a much larger fraction of the research focused on 150 lpw light sources. The panel believes that DOE funding is an important complement and a catalyst to spur non-DOE funding, and that the 150 lpw goal is unlikely to be met without it.
Completing the Matrix
The lighting panel completed the prospective benefits matrixes, shown in Figures F-4 and F-5 for the two completed analyses. In keeping with the committee-designed format, this table shows only the results when the technical goals and all other necessary SSL technology attributes are met as a result of the DOE SSL Program at each of the two levels.
Prospective Economic Benefits Results. Figures F-4 and F-5 show that the U.S. economy would benefit by between $37 billion and $50 billion from funding of the SSL Program. These amounts are the current value of the cumulative, probability-adjusted economic benefits through 2050, relative to depending on private-sector and foreign government R&D funding. DOE funding of $500 million would be required over 10 years or 20 years for the two budget levels—full funding and reduced funding, respectively. The DOE funding has not been adjusted to current value.
TABLE F-3 Probabilities of Technical Success at Selected R&D Funding Levels
While the analysis is based on the difference in impacts between the two cases, it is not a net analysis in that the incremental costs associated with SSL market penetration as a replacement for existing lighting or new lighting applications are not considered. The panel suggests that DOE refine its future modeling of benefits for all of its programs to make these adjustments. At least in the case of lighting, they are unlikely to make a large difference in the calculated benefits, but they should be included for the sake of completeness. The panel also notes that there might be other positive benefits that are not incorporated in the NEMS economic analysis, such as the possibility that air-conditioning loads might be lowered in commercial buildings with more efficient lighting. Additional benefits such as lower energy prices at the time of peak electricity demand might also be realized. Quantification of such benefits was not available for the panel’s analysis.
As noted above, the matrix does not quantify the High Oil and Gas Prices scenario or the Carbon Constrained scenario, because benefits were not estimated by DOE for either of these cases. High fossil fuel prices and carbon constraints, if imposed, would raise the value of economic, environmental,
and security benefits of the SSL Program to some degree. Of the three scenarios, the Reference Case represents the most conservative estimate of benefits.
Prospective Environmental Benefits. Environmental benefits were derived from the assumed penetration of the 150-lpw SSL technology and the resulting savings in carbon emissions from power plants due to lower electricity use. The environmental benefits shown in and Figures F-4 and F-5 are estimated in million metric tons of carbon equivalent (MMTCe) from the reductions in carbon dioxide (CO2) emissions from power plants.
Other environmental benefits likely to be achieved from lower electricity use—such as lower emissions of nitrogen oxide (NOx), sulfur dioxide (SO2), particulate matter (PM), and mercury—are not listed here. Since NOx and SO2 emissions are capped and emission reduction credits are tradable, any such reductions probably would result in economic benefits in terms of lower costs of compliance with environmental regulations rather than reductions in emissions.
CO2 emission reductions have value in a carbon-constrained economy. That value will vary considerably: The estimated cost of a ton of CO2 reduction ranges from $3 per ton to as high as $80 per ton for capture and sequestration
and energy efficiency technologies. The panel agreed that setting an economic value on the environmental benefits would be controversial and highly uncertain, and it decided instead to report benefits in terms of MMTCe reduced.
Prospective Security Benefits. Security benefits were estimated as the reduction in natural gas and petroleum used in electricity generation that results from SSL electricity savings. The panel estimates that 5 to 7 quads of natural gas could be saved by DOE’s SSL Program, as reported in Figures F-4 and F-5. By reducing natural gas use and, as a result, likely energy imports in the future, the United States would be less vulnerable to the negative effects of higher energy prices and disruptions in supplies, thereby improving national security.
Note that security benefits do not include any impact on the electric transmission network. Most commercial lighting use is during the day, which is when peak transmission loads occur. Thus, improved efficiency of lighting is likely to reduce peak loads and therefore to some extent the vulnerability of the network. However, the panel was not prepared to estimate either the degree of reduction or its value. The study committee might consider this issue in the future.
Technical and Market Risks
There are technical and market risks that can impede both the SSL technical achievement and market penetration, as discussed below.
Technical Risk. The efficacy of 150 lpw represents about 50 percent of the maximum efficiency with which any device could create white light. Experts disagree on the upper limit to the lumen per watt level that might theoretically be achieved, but it is probably in the 200 to 300 lpw range. If 100 percent conversion of electricity to light results in 300 lpw, then a 150 lpw light source would be 50 percent efficient—half of the energy entering the device as electrical current would end up as usable photons in the right spectrum to create white light. This is a difficult target to meet, as not many known technologies are 50 percent efficient. Typically, technologies that are that efficient are at very large scales, such as gas turbines. Most LEDs are still below 40 lpw despite having benefited from decades of research, nor do they come close to the associated goals.
To achieve the 150 lpw level, there would have to be some significant inventions that are difficult even to envision today. Every aspect of SSL needs improvement and invention, but at least a couple of physical breakthroughs would also need to occur. These inventions could happen sooner, later, or not at all. A higher level of funding increases the probability that the inventions will occur, but even then they are not certain.
The panel believes that if DOE achieves the 150 lpw goals, then other aspects of the technology might also fall into place. A lighting source with this efficiency would be small by its very nature. A single chip or OLED at 150 lpw could emit light equivalent to a 75 watt incandescent lamp while operating on only about 6 watts. There would be little heat to deal with, so costly heat sinks would be unnecessary and the SSL source would degrade very slowly, consistent with the panel’s assumption on longevity. Color would be a manageable issue, because a small trade-off in efficacy would allow designers to attain any color mix desired.
Market Acceptance. The panel believes that if 150 lpw and the OBT’s secondary goal—helping American companies stay competitive with foreign companies—are achieved, market acceptance of solid state lighting will probably follow. It is simply a question of how fast. If, for some reason, fixture losses were to be higher for a solid state device, then some of the efficiency gains could be lost. This eventually would stretch out the payback time. If the lamp had poor-quality white light or if color was not stable over the lamp’s life, market acceptance would be slowed.
If the breakthrough that led to 150 lpw required a hazardous material (LEDs and OLEDs today are fairly benign), market acceptance could be affected. Mercury, when it was first put into lamps, was thought of as a relatively inert, safe substance. Subsequently, however, it has been shown to form compounds that enter the food chain. Environmental risks or unknowns could adversely affect SSL market penetration.
Other forms of lighting will continue to improve. Mercury fluorescents are at 100 lpw currently, with some performing at up to 110 lpw. The traditional lighting industry will not sit still as SSL develops. The industry will spend money to protect the hundreds of millions of dollars already invested in the manufacture of traditional lighting. If the gap between SSL and traditional lighting narrows, then the payback for SSL will be less favorable, slowing or even stopping large-scale SSL adoption.
NOVEL OR UNFORESEEN METHODOLOGICAL ISSUES
Inconsistency of Technical Goals with Current Program Funding Levels
The presentation to the panel from DOE staff revealed that the SSL Program’s 10-year assumed level of funding does not match the ongoing research program at DOE. The budget presented—including the technical research goals and market penetration goals used to calculate benefits—was approximately $450 million over a 10-year period. The FY 2005 funding level for the SSL Program is expected to be $10 million per year. The panel probed what might reasonably be expected to be accomplished in terms of a technical achievement for a program of $100 million over 10 years—with steady-state funding—but no such program plan exists at DOE.
Since the existing funding levels are so out of synchronization with the program technical goals used to estimate benefits, the prospect of achieving success is very suspect at current levels of funding. The panel recommends that DOE present to outside reviewers and to the Congress, as a necessary part of any program plan, technical goals and budgetary levels that are consistent. The panel further suggests that program goals and funding levels be defensible. Prospective benefits should be calculated accordingly. If needed, more than one budgetary level could be presented, and the technical goals, timing of achievement, probability of technical success, or other aspects could be adjusted and the benefits calculations altered as appropriate.
As noted in the previous section, the SSL Program, if successful, will achieve significant benefits under the current Reference Case (see Figures F-4 and F-5). The High Oil and Gas Prices and the Carbon Constrained scenarios would make the benefits even higher, as electricity prices would increase under both of these scenarios. Accordingly, the panel did not consider it worthwhile to ask DOE for more model runs to calculate the benefits under these alternative scenarios, since the program would only produce additional benefits in these cases; the panel was most interested in receiving DOE modeling analysis for the Reference Case to complete the matrix.
Calculation of Benefits
Computational Issues. NEMS is used by DOE to model market transformation times and to estimate prospective benefits. NEMS is powerful, but like all such models, it has weaknesses. While it provides a common basis for the comparison of different government R&D programs, it is difficult and cumbersome to use. It was most useful in calculating SSL market penetration rates, given as input their technical achievements and the status of competing technologies, and the resulting impacts. A simpler model, or even a spreadsheet, could have done much the same and given the panel the opportunity to run parametric analyses. The panel requested that further NEMS analysis be conducted to determine the SSL Program benefits at lower levels of technical success, such as that resulting from lower budget levels. DOE provided the results that might be expected if the budget were $25 million per year instead of $45 million, but these results were not available until this report was almost completed. It would have been more useful to have had results for lower levels of technical success (e.g., 125 lpw and 100 lpw) at the second panel meeting.
This approach of running the model to determine benefits at lower levels of technical success might be useful for all of DOE’s R&D programs. Benefits are likely to be realized even if the research program does not achieve 100 percent of its technical goal. In such cases, the panel believes that it is useful to run NEMS to calculate the differences in market penetration and resulting benefits. While the achievement of a technical goal might be less than 100 percent, the probability of achieving it is likely to be higher. Table F-3 shows the panel’s estimates for the probability of achieving three different levels of efficacy. The net effect on the expected value of benefits is not readily apparent without such a NEMS model run.
It also is necessary to obtain year-by-year calculations from NEMS. DOE frequently prints out only 5-year results, but to calculate total benefits it is necessary to integrate over the period selected.
Panels should be provided with guidance on the period to consider, especially if the 5-year rule (or that for some other time period) is used. This is not as straightforward as in the retrospective study. SSL is expected to start penetrating the market by 2015. If it has not reached 150 lpw by then, it would be expected to continue to improve. In any case, it might continue to improve past 150 lpw. There is an upper limit to what might be achieved with SSL technology. Thus, efficacy improvement curves for different program levels will converge eventually, approaching the same asymptote around 2040 or so. Market penetration (and therefore benefits) curves also will converge. While efficacy gains above this program’s goal would be attributed to future program efforts, those gains are accelerated by the current program, which therefore should be credited with a share of the benefits.
Another factor to consider is the life expectancy of the lamps. Those installed just before the convergence of the efficacy curves will keep operating long after that time. DOE suggested that 100,000 hours of operation is possible. This analysis continued to 2050 to capture that effect, but further examination of the MARKAL output is warranted to ensure that the model handled it correctly.
Next-Best Technology Baseline. DOE needs to consider and include in its NEMS analysis the likelihood of further improvements in existing technology. In examining the technology assumptions modeled in NEMS, it became clear that the input to NEMS did not adequately reflect likely developments in competing technologies over the next 10 years. NEMS did not incorporate any technical efficiency improvement in existing fluorescent technology—although there appears to be a price decline modeled in NEMS for fluorescent technology. There is a good likelihood that fluorescent technology will improve by 5 to 10 lpw over the next 10 years. The NEMS model also did not reflect the possibility that SSL with 100 lpw technology would be developed in the next 10 years by private industry, either by its own research or research with the assistance of significant non-U.S. government funding that is now being conducted. These factors, if incorporated into the NEMS model, would reduce the prospective benefits that were projected by the model. This pos-
sibility was accounted for in this analysis by the 5-year rule, which produced a more plausible baseline. However, the analysis would have been much simpler if DOE had used a reasonable baseline in the first place, as there would have been no need for the 5-year rule.
Federal Role. In considering the effect of the federal funding for SSL, the panel estimated both the amount of private investment currently going into research and the amount of funds being spent by foreign governments in this field, based on historical funding for R&D in the industry. Research sponsored by foreign governments appears to be focused on a technical goal of 100 lpw by approximately 2010 (about where the DOE programs are expected to be by 2010), without necessarily expecting to proceed further toward 150 lpw. The need for two or three breakthroughs in research in order to achieve the U.S. goal of 150 lpw must be recognized. The panel believes that the DOE funding is necessary in order to accelerate the achievement of the technical goal of 150 lpw. Acceleration will vary with the amount of funding provided and the time frame over which it is provided. The panel believes that acceleration in the 5- to 10-year time frame for a program of $250 million over 10 years is reasonable. Crediting DOE with 5-year acceleration may be too little.
Interpretation and Use of the Matrix
The generalized matrix (Figure F-4) is based on the assumption that the DOE’s SSL Program will attain SSL lamp efficiencies of 150 lpw along with other implicit program goals such as cost, reliability, and color quality. The matrix specifically evaluates this particular case, with the attendant research program funded at the full $500 million, 10-year level; it presents the panel’s estimation of the probability that the program’s technical goals will be met when the program is funded at this level.
Following the technical success metric, the matrix then shows the panel’s assessment of the probability of market acceptance, given that these technical objectives are met. Market acceptance is defined as the level of market penetration predicted by the NEMS model. Thus, the panel is assessing the probability that market penetration rates will be as good as or better than those predicted by the model. The panel does not necessarily agree with the model’s particular prediction of market penetration, but it saw no advantage in trying to estimate penetration rates independently.
Finally, the matrix summarizes the economic, environmental, and national security benefits that will accrue to the nation if both the technical objectives and market acceptance rates are achieved.
These results are the distillation of a considerable amount of analysis and discussion by the panel, of both qualitative and quantitative data. The matrix should thus be used with care, recognizing both its capabilities and its shortcomings: It is a good way to present the results of the analysis of a complex program, but it cannot convey all of the considerations and caveats that went into the analysis. It is only one factor that must be used in developing priorities for R&D funding.
The matrix, in conjunction with the supporting materials, is a useful tool for comparing DOE’s SSL research program benefits with those of other research programs that have undergone a similar analysis. The results can be used to show how benefits accrue to the nation given various funding scenarios and what the probability is that these benefits, or partial benefits, will materialize. This combination of risk (probability of success) and return (benefits) can be used as part (but not all) of a comparison of the return on investment of various research programs. These data also can be used to construct a portfolio of projects at the DOE that is balanced among high-risk/high-return and low-risk/low-return projects.
Neither the full-funding nor the reduced-funding matrix can fully represent all facets of the panel’s considerations, especially when attempting to quantitatively describe important qualitative information. For instance, attainment of the full 150 lpw technical performance target would require major breakthroughs in several areas of SSL. The panel’s probability estimate (labeled “Overall Probability of Success” in the matrix) attempts to quantify the likelihood that the research program will influence these breakthroughs. But this number alone, not taken in the context of where and when these breakthroughs might occur, is an insufficient descriptor of the program. The number, presented as a percentage chance of success, also gives the illusion of precision. Attention should be paid to the range of inputs that resulted in the number presented in the matrix. The matrix presents the average of the individual panel members’ estimates, but in reality the range is probably a better way to represent the probability of success.
The panel struggled with this last point for a considerable amount of time. Its concern was that a false sense of precision would be provided by a percentage numerical estimate. Future research programs might be compared solely on a quantitative basis, with a marginally higher percentage beating a marginally lower one. Hence, the panel’s preference was to present the probabilities on a 1 to 5 scale, which at least would prevent this. However, the panel found it very difficult to do the actual estimation work and to show measurable difference between the various funding and technical cases using a 1 to 5 scale, so the scale was changed to 1 to 100. However, the reader is cautioned to be careful when using these numbers for comparison purposes.
The panel cautions that the economic and other benefits presented using the expected-value approach based on the technical and market penetration probabilities are but one element to consider as DOE conducts its portfolio of research. The matrix presents an expected-value approach, but several important points are missed if this number is the sole focus of consideration.
First, this number is the result of a range of probabilities that represented the views of the panel. The probability range is important to consider. The numbers in the matrix should not be viewed as being too precise or subject to incremental analysis. Second, the probability aspect is but one element to consider as an R&D portfolio is constructed. Projects should not necessarily be selected on the basis of highest to lowest “expected value.” There are many elements to consider in selecting the projects that will constitute a well-constructed and balanced R&D portfolio.
This report has focused on evaluating the DOE-funded SSL Program based on the goal that solid state lighting will—at some point in the future—replace conventional lighting. The panel also noted that other potential benefits are likely to arise. As new light sources are developed, new applications will also emerge, yielding economic benefits that have not been discussed here. Indeed, even if the program fails to produce a light source that replaces an appreciable number of conventional lighting fixtures, it is possible that these new applications alone will benefit the overall economy in such a way as to make the program worthwhile. As a caution regarding energy consumption, however, such new applications are likely to increase demand for energy, not reduce it.
SUMMARY OF PANEL FINDINGS
The Panel on Benefits of Lighting R&D found the methods proposed by the full committee to be workable. Results should be helpful to DOE and Congress in considering their budgeting priorities. The panel noted several points that should be considered in future endeavors. The presentations and support by DOE representatives were excellent. Their help and professionalism are appreciated. The panel was made up of members who were all very interested in the subject and had great expertise in the lighting and energy savings field. The chair kept the discussions on track, focused, and on subject. Several members had facilitation skills that also were very helpful in keeping the group focused and working in an orderly and cooperative manner. The wide range of experience and opinions among panel members ensured that the issues were fully discussed and understood by everyone. Disagreement was encouraged, resulting in conclusions that represent diversity and balance, a nearly impossible expectation for committee meetings. Although the NEMS database lacked some flexibility, it did not detract from the validity of the conclusions. The use of both voice voting and secret balloting on the probability estimates permitted a comparison of methods. The voice voting early in the meeting brought out some great debates and thoughts, but the final vote by secret ballot probably produced a truer and more thoughtful result. The combination worked well.
A few items should be considered in order to improve the process. Consideration should be given to viewing a project such as this one on a total system basis. For example, the panel could have used a fixture representative and a general construction expert during its deliberations. Solid state lighting will cause a major change in fixture and building design and, although the panel reached no conclusions on these issues, it is worth noting that it may have underestimated the savings in energy consumption and the speed of market acceptance.5
Consideration should be given to including on the panel laboratory scientists (e.g., a discharge lighting physicist), who could be helpful in determining the exact scientific requirements for a major breakthrough in existing technologies.
The panel’s experience was very positive, and it hopes that this methodology will be a great resource as attempts are made to determine how the nation’s precious national research dollars are spent. Thanks are extended to DOE for the opportunity to participate and for their leadership and professionalism.
PANEL MEMBER BIOGRAPHICAL INFORMATION
Maxine Savitz (NAE) (chair) is retired general manager of Technology Partnerships, Honeywell, Inc. She has managed large R&D programs in the federal government and in the private sector. Some of the positions that she has held include the following: chief, Buildings Conservation Policy Research, Federal Energy Administration; professional manager, Research Applied to National Needs, National Science Foundation; division director, Buildings and Industrial Conservation, Energy Research and Development Administration; deputy assistant secretary for conservation, U.S. Department of Energy; president, Lighting Research Institute; and general manager, Ceramic Components, AlliedSignal Inc. (now Honeywell). Dr. Savitz has extensive technical experience in the areas of materials, fuel cells, batteries and other storage devices, energy efficiency, and R&D management. She is a member of the National Academy of Engineering. She has been, or is serving as, a member of numerous public- and private-sector boards and has served on many energy-related and other NRC committees. She has a Ph.D. in organic chemistry from the Massachusetts Institute of Technology.
M. George Craford (NAE) received a B.A. degree in physics from the University of Iowa and a Ph.D. degree in physics from the University of Illinois. Dr. Craford began his professional career as a research physicist at Monsanto Chemical Company. Initially, his research dealt with the development of optoelectronics materials and devices using a variety of compound semiconductor materials. In 1974 he became the technical director of the Monsanto Electronics Division, with management responsibility for silicon wafer development as well as compound semiconductor materials and device development. In 1979, Dr. Craford joined the Hewlett-Packard Company as manager in the Optoelectronics Division, responsible for the development of technology and processes for manufacturing visible light-emitting diodes. In 1999 he assumed his current position as chief technology officer of Lumileds Lighting, a joint venture of Agilent Technologies and Philips Lighting.
Paul A. DeCotis is director of the Energy Analysis Program responsible for statewide energy policy analysis, planning, and evaluation at the New York State Energy Research and Development Authority (NYSERDA). Mr. DeCotis directs statewide energy planning, energy policy and legislative analysis, corporate strategic planning, energy program and R&D program evaluation, and energy emergency planning. He oversees statewide energy demand and price forecasting for all fuels: economic, electricity and natural gas system, and environmental modeling; and energy markets assessments. He is the record access officer to the State Energy Planning Board and chair of the Interagency Energy Coordinating Working Group, comprising the Departments of Public Service, Environmental Conservation, Transportation, and Economic Development. He is also a member of the New York Independent System Operator Management Committee. Prior to joining NYSERDA in 1995, Mr. DeCotis was chief of policy analysis at the New York State Energy Office. He previously served as a staff economist, financial analyst, and policy analyst. Mr. DeCotis also is president of Innovative Management Solutions, a management consulting business specializing in strategic planning, executive management development, and mediation. Mr. DeCotis is an adjunct professor at the Sage Graduate School in the master of business administration program and previously was an adjunct professor at Cornell University in the School of Industrial and Labor Relations. His broad experiences include holding elected office as a member of the Ballston Spa Central School District Board of Education, serving as chair and chief fiscal officer, and also serving as chair of the Saratoga County School Boards Association. He is currently a board member and executive vice president of the Association of Energy Service Professionals International. Mr. DeCotis holds a B.S. in international business management from the State University College at Brockport, an M.A. in economics from the State University of New York at Albany, and an M.B.A. in finance and management studies from Russell Sage College.
Todd Graves is the GE Global Research Business Program Manager for GE Consumer and Industrial (GECI), where he coordinates the advanced research and development for lighting and appliance technologies, including significant programs in both organic and inorganic light-emitting diodes. He is also responsible for strategic planning of the R&D efforts in these areas, developing long-term research plans to support future business needs and coordinating product development efforts at GECI to take advantage of research breakthroughs. Mr. Graves also manages the research portfolio of the GELcore, the GE/Emcore joint venture that is developing high-brightness light-emitting diode (LED) products and LED-based signaling and signage products. Prior to joining GE, Mr. Graves was on active duty in the U.S. Air Force, where he worked on DOD acquisition and development programs in the areas of turbine engine technology and the Global Positioning System. He participated in and led several major government source selections and acquisition management programs and was certified as a DOD acquisition manager. He has a master’s degree in aerospace engineering from the University of Notre Dame.
Pekka Hakkarainen is director of technology and business development at Lutron Electronics. Previously he had been general manager, Commercial Lighting Control Systems, and had held several other positions in his 14 years at Lutron. Dr. Hakkarainen also was research scientist in nuclear fusion at the Massachusetts Institute of Technology (MIT). He is a member of the Illuminating Engineering Society of North America (IESNA), the National Electrical Manufacturers Association (NEMA), and the American Physical Society. He has chaired several IESNA and NEMA committees and served on other industry committees and advisory groups. Dr. Hakkarainen received a B.A./M.A. in mathematics from Cambridge University, England, and a Ph.D. in plasma physics from MIT.
Dean T. Langford retired as president of OSRAM SYLVANIA in 2001, a position he had held since January 1993. Mr. Langford, who had been president of the GTE Electrical Products Group, a global lighting and precision materials company, since 1984, joined GTE in 1983 as vice president for marketing of the former GTE Communications Products Group. The GTE Electrical Products Group was acquired by OSRAM GmbH, a Siemens company, in January 1993. Prior to OSRAM SYLVANIA, Mr. Langford spent 19 years with the IBM Corporation, where he held numerous positions with varied responsibilities. He has held leadership positions on several executive boards and has served as co-chair of the Alliance to Save Energy. He also serves on the board of directors of the National Park Foun-
dation. A graduate of the University of Illinois, Mr. Langford has a bachelor’s degree in mathematics and holds an honorary doctor of humane letters degree from Salem State College in Massachusetts.
Mark S. Rea has been director of the Lighting Research Center at Rensselaer Polytechnic Institute (RPI) since 1988. He is also a professor at the School of Architecture and in the Department of Philosophy, Psychology, and Cognitive Science at RPI. Previously, he had been manager of the indoor environment program, Building Performance Section, at the National Research Council of Canada. He also has been visiting scientist at the Electricity Council Research Centre, Capenhurst, United Kingdom. Dr. Rea is a fellow of the Illuminating Engineering Society of North America and of the Society of Light and Lighting (United Kingdom). He is on the international editorial advisory board of Lighting Research and Technology and editor-in-chief of the Illuminating Engineering Society of North America’s Lighting Handbook (8th and 9th editions). He received the William H. Wiley Distinguished Faculty Award from RPI and the Gold Medal from the Illuminating Engineering Society of North America. Dr. Rea received a B.S. in psychology and an M.S. and Ph.D. in biophysics from Ohio State University.
James Wolf is an independent consultant working with companies to design new products and services for deregulating electric utility markets. He was formerly vice president of energy and environmental markets for Honeywell, Inc., where he focused on business development opportunities to develop new products and services and to market existing services to energy and environmental concerns. Previously, he was executive director at the Alliance to Save Energy, a nonprofit coalition whose board of directors is composed of U.S. senators, chief executive officers of major corporations, and environmental leaders. He also served as acting deputy assistant administrator for policy and planning with the U.S. Department of Commerce’s National Oceanic and Atmospheric Administration. There he helped design and supervise policies and programs addressing marine pollution, global climate change, alternative energy resources, and international scientific research protocols. Mr. Wolf has a J.D. degree from Harvard Law School.