National Academies Press: OpenBook

Energy Efficiency in Buildings: Behavioral Issues (1985)

Chapter: PREDICTED AND ACTUAL ENERGY SAVINGS FROM HOME RETROFITS

« Previous: HOME ENERGY RATINGS
Suggested Citation:"PREDICTED AND ACTUAL ENERGY SAVINGS FROM HOME RETROFITS." National Research Council. 1985. Energy Efficiency in Buildings: Behavioral Issues. Washington, DC: The National Academies Press. doi: 10.17226/10463.
×
Page 79
Suggested Citation:"PREDICTED AND ACTUAL ENERGY SAVINGS FROM HOME RETROFITS." National Research Council. 1985. Energy Efficiency in Buildings: Behavioral Issues. Washington, DC: The National Academies Press. doi: 10.17226/10463.
×
Page 80
Suggested Citation:"PREDICTED AND ACTUAL ENERGY SAVINGS FROM HOME RETROFITS." National Research Council. 1985. Energy Efficiency in Buildings: Behavioral Issues. Washington, DC: The National Academies Press. doi: 10.17226/10463.
×
Page 81
Suggested Citation:"PREDICTED AND ACTUAL ENERGY SAVINGS FROM HOME RETROFITS." National Research Council. 1985. Energy Efficiency in Buildings: Behavioral Issues. Washington, DC: The National Academies Press. doi: 10.17226/10463.
×
Page 82
Suggested Citation:"PREDICTED AND ACTUAL ENERGY SAVINGS FROM HOME RETROFITS." National Research Council. 1985. Energy Efficiency in Buildings: Behavioral Issues. Washington, DC: The National Academies Press. doi: 10.17226/10463.
×
Page 83
Suggested Citation:"PREDICTED AND ACTUAL ENERGY SAVINGS FROM HOME RETROFITS." National Research Council. 1985. Energy Efficiency in Buildings: Behavioral Issues. Washington, DC: The National Academies Press. doi: 10.17226/10463.
×
Page 84
Suggested Citation:"PREDICTED AND ACTUAL ENERGY SAVINGS FROM HOME RETROFITS." National Research Council. 1985. Energy Efficiency in Buildings: Behavioral Issues. Washington, DC: The National Academies Press. doi: 10.17226/10463.
×
Page 85
Suggested Citation:"PREDICTED AND ACTUAL ENERGY SAVINGS FROM HOME RETROFITS." National Research Council. 1985. Energy Efficiency in Buildings: Behavioral Issues. Washington, DC: The National Academies Press. doi: 10.17226/10463.
×
Page 86
Suggested Citation:"PREDICTED AND ACTUAL ENERGY SAVINGS FROM HOME RETROFITS." National Research Council. 1985. Energy Efficiency in Buildings: Behavioral Issues. Washington, DC: The National Academies Press. doi: 10.17226/10463.
×
Page 87
Suggested Citation:"PREDICTED AND ACTUAL ENERGY SAVINGS FROM HOME RETROFITS." National Research Council. 1985. Energy Efficiency in Buildings: Behavioral Issues. Washington, DC: The National Academies Press. doi: 10.17226/10463.
×
Page 88
Suggested Citation:"PREDICTED AND ACTUAL ENERGY SAVINGS FROM HOME RETROFITS." National Research Council. 1985. Energy Efficiency in Buildings: Behavioral Issues. Washington, DC: The National Academies Press. doi: 10.17226/10463.
×
Page 89
Suggested Citation:"PREDICTED AND ACTUAL ENERGY SAVINGS FROM HOME RETROFITS." National Research Council. 1985. Energy Efficiency in Buildings: Behavioral Issues. Washington, DC: The National Academies Press. doi: 10.17226/10463.
×
Page 90
Suggested Citation:"PREDICTED AND ACTUAL ENERGY SAVINGS FROM HOME RETROFITS." National Research Council. 1985. Energy Efficiency in Buildings: Behavioral Issues. Washington, DC: The National Academies Press. doi: 10.17226/10463.
×
Page 91

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

CHAPTER 6 PREDICTED AND ACTUAL ENERGY SAVINGS FROM HOME RETROFITS Actual energy savings achieved from home retrofits in the United States have varied considerably from predicted values. Engineering models have usually overestimated energy savings on the average, and predictions for individual homes have varied widely on both sides of observed energy use (Goldman, 1984; Goldman and Wagner, 1984; Harris, 1985; Hirst and Goeltz, 1984; Hirst, White, and Goeltz, 1983~. Figure 1 shows a typical scatterplot of predicted energy savings compared with actual savings in a single retrofit PrOoram. ~ The scatterplot represents a correlation of only 0.33; in only 45 percent of the homes was the actual energy savings within 50 percent of the predicted value (Hirst and Goeltz, 1984~. These discrepancies are of more than theoretical concern. They are important both to homeowners who wish to make wise decisions about retrofit and to both public and private organizations that have an interest in retrofit activity. For example, an electric utility com- pany that is counting on home retrofits to obviate the need for a new power plant must be able to tell 10 years in advance whether its con- servation programs will save enough energy to meet that goal and whether programs that meet that goal will cost less than building generating capacity. Regulatory agencies that want utilities to invest in conser- vation rather than constructing new plants have the same needs. For these purposes, savings estimates that are accurate in the aggregate are sufficient. But for other purposes, accurate estimates for individual homes are important. Companies in the home retrofit industry, for example, will only be credible if they can give their customers accurate predictions of energy savings--and the companies could guarantee energy savings if they could make good enough predic- tions. The credibility of savings estimates can also determine whether a utility conservation program will achieve its goals because a program whose predictions are not achieved by its initial customers is unlikely to reach all the customers its sponsors expect. There are many possible reasons that predicted and actual energy savings might not match. The estimates may be unreliable because real homes do not fit the categories used in the estimation models and because operators of the models do not all make the needed approxima- tions in the same way. The estimation procedures may unrealistically assume perfect retrofits when there are only imperfect ones. Or, it 79

80 _ 100 CO a) >. m - - C~ a, - o CD me In a: a: 50- 25 O . . ·~ '< . - / .-. w ·--4 ·w · - . ~~'.-~...-. ~ - ~ o ~ - . . . 346 AUDITED HOMES -50 - 0 25 50 75 100 125 150 AUDIT PREDICTION OF GAS SAVING (MBtu/year) FIGURE 1 Actual and predicted natural gas savings for 346 homes audited by Northern States Power, Minnesota. Six of the homes had values outside the bounds of the graph. SOURCE: Hirst and Goeltz {1984:17~. may be possible to install retrofits of the assumed quality, but installers may be less than thorough or highly variable in their work. Contractors who use estimating techniques may make overly optimistic assumptions to support their sales efforts. And even if contractors, installers, and materials perform up to expectations, home occupants may become heavier energy users as a result: they may reset thermostats for more comfort or use money saved through energy efficiency to buy new appliances. Or, occupants who are induced to change habits when the retrofit is made (for example, by operating shades in a sunspace) may revert to old habits over time. Finally, the accuracy of an estimate may deteriorate over time as wall insulation settles, other materials break down, or occupant behavior changes. It makes a difference which of these hypotheses accounts for most of the discrepancy between actual and predicted energy savings, because they have different implications about how to improve prediction. One hypothesis implies that better estimation can be achieved by refining models, another calls for better training of model operators, another for better quality control in installation, another for better mater- ials, another for regulating deceptive advertising, and yet others for

81 taking occupant behavior into account. Some of the hypotheses imply that, overall, predicted energy savings can be attained while others imply that they are unattainable. Some hypotheses have implications for the variability in estimation: of these, some imply that predic- tions for single homes can be improved by simple measures such as better training of the people doing the estimating; others imply that variability will be very hard to reduce because it depends on occupant behavior. The hypotheses also have different implications for how interested parties should act. For a utility company investing in conservation, one hypothesis suggests that an investment in inspecting contractors' work could bring energy savings into line with expectations; another hypothesis implies that savings predictions should be lowered because they have not taken into account people's behavioral responses to improved energy efficiency. That implication would lead utilities to decrease support for residential conservation if that support was based on expected relief from demand growth. For a homeowner, support for one hypothesis would underline the importance of choosing a contractor carefully; support for another would mean that all contractors'- estimates are probably too optimistic. For a home retrofit contractor, support for one hypothesis would argue for training energy auditors very carefully before making promises about energy savings; support for another would give a warning to tell customers that expected savings can be achieved only if they do not change their behavior. At present, the relative accuracy of the various hypotheses cannot be determined empirically. This chapter outlines an approach to making such a determination. In the language of Chapter 1, the chapter addresses an issue of technological research and development, focusing on the problem of estimating energy savings while taking behavior into account. RESEARCH STRATEGY The research problem of determining why predicted and actual energy savings are different is difficult because many factors may prove important, because data do not yet exist on some of them, and because some of them are difficult to measure accurately. For these reasons, we believe the problem calls for a two-stage approach that relies heavily on data collection under field conditions. The first stage would use exploratory studies that are very detailed but small in scale to clarify the dimensions of the problem. The aim would be to gauge the approximate range of variation due to each plausible hypothesis, narrow the list of potentially important variables, evaluate possible measures for these, and thus make larger-scale research feasible. The second stage would address a narrower range of questions using low-cost measurement techniques for those variables when those techniques had been demonstrated in the first stage to be sufficiently accurate. A large, representative sample of home retrofits would be used in the second stage to make it possible to generalize quantitative findings.

82 In the f irst stage of research, the primary operating principle should be to assess the full range of possibilities, using a sample of home retrofits chosen for variety rather than representativeness. To assess hypotheses about installation quality, the sample must include retrofits done by contractors with varying levels of experience and expertise, as well as retrofits believed to be of the highest achiev- able quality. First-stage research should be conducted in housing markets where such a wide range is available for sampling. To assess hypotheses about occupant behavior, the sample should include homes whose occupants vary widely in age, income, education, and household composition. To test hypotheses about the adequacy of predictive models, the sample should include houses that vary in age, type of construction, and conformity or nonconformity to typical design and construction methods. In addition, as part of the research, particular retrofits should be modeled by different experts to clarify the size and sources of discrepancies between model outputs. Another principle for the first stage of research is to use multiple measures when it may be possible to validate low-cost indices for use in further research. For example, a regression model of a home's use of heating fuel may prove as accurate for some purposes as more expen- sive methods involving direct monitoring of furnaces (Fels, 1983; Fels, Rachlin, and Socolow, 1984~. If both direct measurement and regression models are used in careful first-stage research, it will be possible to tell whether some version of the less expensive method is adequate for use in a def initive, large-scale study . The f irst stage of research should involve small studies in one to three housing markets in climates with a signif icant heating load . In the cold climates of the northeastern and north central states there is significant potential for cutting energy use in samples of homes that include many of older construction. The Pacific Northwest is another good site for research because, although heating loads are not as great, its history of low energy prices has left a large potential for weather- ization, and considerable research skills have been built through the efforts of the Bonneville Power Administration. There is also need to study energy savings in homes in hot climates, but less is known at present about modeling cooling loads. Initial research in colder climates should advance knowledge, making future studies feasible in warmer climates. If studies proceed one at a time, research should probably begin in an area with heavy heating loads because it may be possible to get reliable determinations of effects with a small sample and even with rather rough measurement techniques . Within each climate zone, a housing market should be chosen for its variety of housing types, con- tractor types, socioeconomic status among homeowners, and the availabi 1- ity of a competent research team. We emphasize that studies in a few markets will not be generalizable nationally: the purpose of the first stage is to determine the major causes of gaps between actual and predicted energy savings so that minor causes can safely be ignored in more comprehensive research. Measurements for each home in the sample should be made for at least a year before and a year or more after retrofitting, with somewhat

83 different schedules for different measures. Comparison of energy use for the year before and the year after retrofit is essential to verify estimation models. Occupant behavior should be assessed before the retrofit and for a year or more afterward so that behavior can be held constant statistically to check the models. Particular behaviors, including thermostat settings, should be monitored over the first month after a retrofit to identify short-term behavior changes that may mitigate energy efficiency. All the above information can help tell whether gaps between actual and predicted energy savings are due mainly to estimation models or to actual changes in buildings and their occupants. Reassessment after two and three years is necessary in at least a subsample of homes to examine hypotheses involving deterioration of retrofit materials, slow reversion to old habits of energy use, or behavioral changes that have the effect of using the money the retrofit saved to increase energy consumption in other ways. There is need to improve understanding of both aggregate predictions and predictions for individual homes. To make estimates credible to building operators, there is a need to improve the accuracy of predictions for individual buildings or to separate the behavioral and structural components of estimates. If much of the inaccuracy in prediction is due to variation in behavior, it will be critical to communicate that information to building owners and occupants. THE FIRST STAGE OF RESEARCH This section discusses possible measurement techniques for the first stage of research. In the second stage, not discussed in this report, the list will be shorter because some measurements will prove to be unnecessary and because some hypotheses will be eliminated from consideration. Measuring Retrofits Air infiltration accounts for approximately one-third of heat loss in conventional housing (Malik, 1978~. Thus, it is a major object of retrofits, especially low-cost ones that emphasize caulking and weather- stripping. Air infiltration rates may be estimated from data on air leakage collected with an instrumented blower door (Dust, Jacobson, and Socolow, 1985~. This measurement takes about half an hour and costs about $100 per home (G. Dutt, Princeton University, personal communica- tion). Infiltration can also be measured with a tracer gas (Malik, 1978), but these measurements are sensitive to momentary weather con- ditions and so are of questionable reliability. A new and promising technique involving continuous sampling of a tracer gas has been developed at Brookhaven National Laboratory (Dietz et al., 19851. Measurement should be made before and after retrofitting and, to test for breakdown of air barriers, after a period of two or three years. Direct measurement of air infiltration is important in the

84 first-stage research because some retrofit activities focus almost entirely on limiting infiltration, and data on the direct effects of such efforts are lacking. An added benefit would be the collection of f ield data relevant for analyzing issues of indoor air quality. Gaps and "bypasses ~ in insulation in attics can account for 3 5 percent of a home 's heat loss in winter (Beyea, Dutt, and Woteki, 1978 ~ . They are a likely explanation for disappointing results in retrofits because small, hard-to-reach attic spaces may easily be passed over by an insulation contractor who is in a hurry or who is not careful. It is possible to locate bypasses by examining the insulation with an infrared scanner. The research team can make such an examina- tion before and after installation, recording the location and size of all gaps in wall and attic insulation. A return visit in a year can detect setting of insulation installed in walls. Such infrared scanning and recording of results for an average home takes roughly an hour and costs about $250, depending on the amount of detail required in the report (Dust, personal communication). At present, precise relation- ships between the size and location of gaps and insulation and heat loss cannot be specified. Therefore, field observations in the first stage can provide only rough indications, not precise estimates, of the size of the problem. This may be enough to tell whether the issue is worthy of more detailed study. Steady-state furnace combustion efficiency is a target of retrofits, especially in older gas- and ail-heated homes that do not have efficient furnaces. Combustion efficiency can also be affected by replacement of furnace filters, replacement of oil burner nozzles, and other main- tenance activities. Maintenance--or the lack of it--is a possible source of errors in estimating energy savings from furnace improvements. A technician with the proper equipment can measure combustion efficiency in 10 or 15 minutes with equipment costing about $150 (T. Vineyard, Oak Ridge National Laboratory, personal communication). This measurement should be made before and after furnace retrofits and again after a full winter to check on maintenance. Note that measuring a furnace's efficiency tells nothing about how much of the heat reaches the living area: methods to directly measure heat losses in a distribution system are not yet well validated. Energy use by furnaces readily converts into a measure of total heat loss, including distr ibution losses and losses through the building shell. It can be measured directly by submetering a fuel line or indirectly by meter fig a furnace's "on time n at the thermostat and multiplying by the rate of fuel feed to the furnace. Metering at the thermostat involves very little incremental cost if the thermostat is also being monitored for other purposes (see below). In at least some homes, the instruments should record the specific times the furnace runs, to compare furnace operation against occupants' reports of their behavior (see below). The timed measure of energy use has several advantages over other indices. Unlike meter readings or the observa- as a direct observation against which to assess the adequacy of regression models of furnace operation based on meter readings. It can reflect and be checked against a range of behaviors, including thermostat talons Rescreen above. It can cove continuous Data. It can be used

85 settings, which can be independently monitored at low cost, and the opening of windows, which cannot. A disadvantage of recording specific times compared to cumulative time is the extra cost of data collection and analysis. For this reason, complete data should be collected only in homes where it will be used to compare against other time-specific data, such as observed or reported behavior. The measure of total energy use reflects the combination of dis- tributional losses in a heating system and inefficiencies due to an oversize furnace or oil burner nozzle. Because the measure does not distinguish among these inefficiencies, it has limited value for evaluating retrofits aimed at correcting them. But in combination with other measures, a measure of total furnace operation can clarify the importance of some factors in retrofit that are not easy to measure directly. Assessing Engineering Models in Practice Discrepancies between actual and predicted energy savings may be due to errors or omissions in engineering models or inconsistencies in their application. One issue is how much accuracy can be improved by making models more detailed. In stage-one research, models offering different degrees of detail should be tested on the same data. To the extent that simpler models are adequate for addressing the questions determined to be important, the burden of data collection can be reduced in stage-two research. Another question concerns the variance in estimates as a function of judgments by auditors and model operators. It would be useful to have some homes examined by a second auditor and to have some audit reports entered into models by a second operator to estimate the mag- nitude of these two possible sources of variability. If differences are large, it may be worthwhile to experiment with training to standardize the experts' judgments. A third question concerns the modeling of interactions between retrofits. The effect of insulation on energy use is smaller when a home has an efficient furnace, but engineering models of this sort of interaction have not been tested against data on actual retrofits. If the stage-one research includes a variety of retrofit packages, its data can help improve engineering models of these interactions. Other Direct Measures Indoor temperature regulation is the short-term behavioral adjust- ment with the greatest effect on a home's energy use. Indoor tem- perature must be monitored to test the frequently asserted hypothesis that people, especially those in low-income households, who have been sacrificing comfort to pay energy bills will respond to improved energy efficiency by resetting thermostats. It is also critical to a less often-stated reverse hypothesis: since drafts raise the temperature level needed for comfort in winter (Nishi, 1978), reducing air infiltra-

86 Lion may induce people to lower indoor temperatures because they need less heat to compensate for drafts. Indoor temperature can be monitored at thermostats along with thermostat settings, at a cost of at least $150 per home (Dust, personal communication). It would be advisable to keep a running record of temperatures in some homes to assess the extent and timing of any behavioral changes that occur and as a check on the accuracy of occupants' self-reports of temperatures. Thermostat settings are closely related to indoor temperatures during the heating and cooling seasons when the systems function properly. They are not the same, however, and may sometimes be what people report when asked for the temperature in the home. Thus, both thermostat settings and indoor temperatures should be monitored. Hot water use is a major energy variable in homes that is behav- iorally controlled: Kempton (1984) has observed a threefold variation in hot water used per day per person among seven households. Hot water use should be monitored in some homes to see if it changes after a retrofit. It may increase if people feel they can afford to use more, or it may decrease, for example, if baths and hot drinks were being used to warm people in a cold house. Accurate measurement would require a flow meter on the outlet of the hot water tank, an investment of about $100 per home (Dust, personal communication). Additional metering of energy use by the water heater would give useful data on the average operating efficiency of water heaters and on the actual energy savings from retrofits or replacements of water heaters. The operation of appliances can be assessed by submetering appliances whose use is likely to be affected by home retrofits. The most obvious of these are room air conditioners and space heaters; stoves may also be affected when they double as space heaters. These appliances should be monitored in a subsample of homes. Fuel and electricity bills are, of course, essential for studying the effects of retrofits and for building and validating single-home energy models that may greatly decrease the cost of the second stage of research. The importance of billing data is that they are regularly available without adding instruments to a home. They give less perfect accounts of fuel use than complex data recorders, but promising tech- niques are being developed for assessing energy savings with models based only on billing and weather data. One such method has been developed by Margaret Fels and her associates at Princeton University (see Fels, 1983~. In this method, consumption of fuels used for heating is estimated for each home by an equation: Fi = a + bHi ' where Fi is the average daily fuel consumption in time interval i, and Hi is the heating degree-days per day during time interval i below a reference temperature, Tref. Ordinary least-squares regression equations are calculated for several values of the Tref, and the equation giving the best fit is selected as the model of energy use in that home. Tree is interpreted as the average daily temperature below which the home uses heat; a is interpreted as the base load for l ,

87 the fuel in question, used for cooking, lighting, and other functions that do not change with the ambient temperature; and b is interpreted as the dependence of fuel use on temperature. For assessing energy savings from retrofitting, normalized annual consumption (NAC) equations are calculated for a year before and a year after the retrofit, using the equation: NAC = 365a + bHo(Tref) , where Ho(Tref) is the number of heating degree days (base Tref) in a "typical" year. Thus, the NAC formula can be used to partition energy savings into those due to base-load effects, to the weather sensitivity of energy use, and to changes in the reference temperature, the last two of which can be expected to fall after retrofitting the building shell. The above system is very inexpensive compared to instrumenting homes, but two important operational problems must be recognized. First, the system is sensitive to the quality of billing and weather data. Careful checking of billing data to eliminate estimated bills, changes in occupancy, and other sources of error is essential in developing NAC models (Berry and Vineyard, 1985~. Second, the need to calculate Tref for each home puts the procedure beyond the capability of inexperienced analysts using standard statistical programs. A sim- plified system that calculates NAC based on an assumed uniform value of Tref has been developed at Oak Ridge National Laboratory (Berry and Vineyard, 1985~. In analyses of gas-heated homes in Minnesota and electrically heated homes in the Pacific northwest, the simplified system, setting the reference temperature at 60°F, has yielded estimates of NAC very close to those obtained with the Princeton method, although estimates of the heating and base-load components of energy use diverge farther from those calculated when the reference temperature is determined empirically. Further testing of both calculation procedures should be a part of stage-one research because of the value that can be gained in later research from an acceptable estimation technique that does not require instrumentation. Self-Reported Behavior Temperature and thermostat settings should be measured directly, but the first-stage research should also collect self-reports to check their validity. If it proves possible to get reliable data from self-reports, it will greatly decrease the cost of the second-stage research. Many researchers mistrust the accuracy of self-reports because of the possibility that they will be influenced by "social desirability effects: householders reporting something between the truth and what they think the researcher would like to hear or the neighbors expect. It is possible, however, that careful survey techniques can yield reliable estimates. For example, in one study of 779 households in the Pittsburgh area (Beck, Doctors, and Hammond, 1980), 71 percent of households reported their thermostat settings

88 correctly to within 2°~. Furthermore, the errors fell about equally on each side, suggesting random rather than systematic reporting errors. Comfort is a critical subjective variable in most hypotheses about changes in occupant behavior after retrofits. It is also important in interpreting data from evaluations of retrofit programs. When a retro- fit program saves less energy than anticipated, it may not matter to a utility whether the reason was an inadequacy in an estimation model or an increase in occupants t comfort levels--in either case, the program does less than expected to obviate the need for new power plants. But for a state or local government, a retrofit contractor, or a homeowner, human comfort may be as important a reason for retrofitting as lowered energy costs. It is therefore essential to measure it, which can only be done by self-report. Survey questions should can assess relative comfort on a rough (e.g., five-point) subjective rating scale. They should focus on not only the degree of discomfort but on its frequency (nonly on the coldest or hottest days," "often," or "almost always," during the heating or cooling season). They should assess which members of the household have discomfort, recognizing that the most uncomfortable person may not be one who responds to the survey. And questions about comfort should be asked during the same period of the year for pretests and posttests (preferably when discomfort is expected to be great) because of the unreliability of people's memories. Draftiness is a subjective comfort-related variable that should be assessed directly. Air infiltration and convection near cold windows and exterior walls can produce sensations of draft to which people may respond by adding clothing, by resetting thermostats, by using space heaters, or by feeling discomfort. Direct measurement of air infiltra- tion can be compared with reported feelings of draftiness and discomfort and-reports of behavioral adjustments to those feelings to determine how much the reduction of air infiltration does to change the experience and behavior of occupants of retrofitted homes. Use of supplementary heating and cooling equipment should be assessed by instrumenting space heaters in some homes and also by self- reports because it is impractical to meter all space heaters in all homes in a sample. Direct measures and self-reports can be checked against ongoing measures of furnace operation and against regression models of energy used for heating. According to recent research with regression models, non-furnace energy use in winter increases in northern climates due to some combination of the use of space heaters, increased energy use to heat water, and increased cooking and lighting (Fels, Rachlin, and Socolow, 1984~. Where space heaters are not metered, survey questions should inventory space heating equipment, get specifications for the equipment if feasible, and ask about the frequency of use for each piece of equip- ment. To validate regression models of the impact of this equipment when direct measurements cannot be made, it will be useful to conduct surveys during the heating or cooling season and to ask about the use of supplementary heating and cooling during the previous day. People can report this more accurately than they can estimate average use over a season, and if total household energy use is being continuously .<

89 _ __ monitored, the reports can be checked against those measurements to directly assess the effects of supplementary heating and cooling on total energy use. Closing off rooms to save on heating is a common expedient in cold climates, and it is at sometimes a predictor of participation in retro- fit programs (Tone and Berry, 1984~. It is a fairly simple behavioral adaptation that may have important effects on energy bills. It is also easily reversible if energy costs become a less salient problem for households. Thus, opening closed rooms is a likely behavioral reversion after retrofits. It is a simple matter to ask, in each administration of a household survey, whether rooms are closed off in winter, how many rooms are involved, and for how much of the winter. Since the effects of this adaptation on energy use will depend on the heat distribution system, data on that system should be collected for the homes surveyed. Ventilation is an important behavioral variable in European studies of residential energy use (e.g., van Raaij and Verhallen, 1983) and may become important in the U. S. as homes are retrofitted to higher stan- dards of air tightness. Surveys can assess how often windows and doors are opened to decrease stuffiness, to vent smoke or odors, or to air out rooms. Such information may help make sense of apparent anomalies in data on furnace operation and might also be an indicator of behav- ioral responsiveness to changes in indoor air quality after retrofits. Adding or removing clothing is a simple behavioral adaptation to indoor temperature. However, it is difficult for researchers to get accurate measures of the insulating value of clothing over the course of several winters. An approximation can be made with interviewers' observations, which are useful when cumulated over many homes. Survey questions about the usual behavior of household members would also have some value. But in the early phase of research, the best way to explore the importance of clothing as a behavioral adaptation may be through ethnographic interviews of members of a few households before and after retrofits. A good example of this technique applied to hot water use is presented by Kempton (1984~. Economic adjustments related to energy use include depressed living standards before retrofitting because of energy costs and increases in living standards because of energy savings from the retrofits. Retro- fitting may also affect a household's ability to make timely payments of utility or other bills. Preretrofit adjustments must be assessed and may include less expenditures on food, clothing, entertainment, transportation, or other expenditures the household considers normal. Self-reports cannot give a reliable dollar value for such sacrifices, but they can tell which households perceived sacrifice and what items in the household budget were affected (one study that used such survey items is by Dillman, Rosa, and Dillman, 1983~. After retrofitting, surveys should repeat the questions, along with questions about whether the retrofitting has enabled the household to afford things and activities it could not previously afford. Surveys might also inquire about appliance purchases as an indication of whether the retrofits have affected major household purchases or their energy intensity. l

so GUIDELINES FOR RESEARCH DESIGN The costs of the f irst stage of research are influenced primarily by the number of homes sampled, the extent of instrumentation, the effort needed to simplify data collected by continuous monitoring, and the number and type of occupant surveys conducted. Additional survey questions involve smaller incremental expense. The first stage of research can obviously be done with different degrees of thoroughness, at different cost. Because we cannot predict the level of resources that will be available for the research or the number of sources from which the resources will come, we have not outlined a detailed research program. We can, however, offer some guidelines for setting priorities. ,~, _ 1. Research should ensure that all data collected are cleaned, . documented, and analyzed. Too often, research projects collect valuable data but exhaust their funds before the data become usable. Research sponsors should consider holding back a portion of funds to prevent the t eventual) ty . 2. A single research project should not collect more data than it can clean, document, and analyze. If resources are limited, it is better to conduct a small, focused study than to collect large amounts of data in the hope they will be analyzed later. 3. Resources can be conserved by narrowing the research question or by reducing sample sizes. To address the full range of technical and behavioral questions about the effects of retrofits and their interactions requires full instrumentation of homes and repeated and detailed surveys of occupants. However, a detailed study of even 25 homes would give valuable information if they spanned the range of a housing market and were followed for two to three years after retro- f itting. In addition, less than complete instrumentation can generate useful knowledge about some important questions. For example, a study of behavioral changes after retrofitting requires repeated and detailed surveys, but valuable information can be gained with limited instrumen- tation to assess thermostat settings, indoor temperature, furnace use, and perhaps use of space heaters and hot water. Variability in estimates due to expert judgment can be estimated by using additional energy auditors or operators of engineering models, without any survey research or instrumentation of homes. And the validity of regression models of retrofit effects can be assessed by comparing different models using data from the same homes. Instrumentation of fuel use at the furnace would be valuable for determining if the temperature coefficient in a model corresponds to furnace use. This chapter illustrates the importance of understanding both tech- nology and human behavior for developing policies affecting the energy efficiency of buildings. Physical models are unreliable predictors of the effects of physical changes on a building's energy use in part because those models do not account for systematic var. iation in the behavior of builders, retrofit contractors, and building occupants.

91 understand the causes of the unsatisfying performance of the models and to improve the ability to predict the effect of retrofits requires simultaneous assessment of equipment and human behavior; to make policies that could improve the effectiveness of retrofits will also require attention to both technology and behavior. In this way, the issue of predicted versus actual energy savings underlines the value of the kinds of analysis considered in this report: energy use is a human activity that occurs through technology; to understand it, one must comprehend not only the relevant technologies but also the people who use them.

Next: REFERENCES »
Energy Efficiency in Buildings: Behavioral Issues Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF
  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!