Energy consumption surveys and end uses estimation in the U.S. EIA

Size: px

Start display at page:

Download "Energy consumption surveys and end uses estimation in the U.S. EIA"

Walter Clarke
6 years ago
Views:

1 Energy consumption surveys and end uses estimation in the U.S. EIA For InterEnerStat Workshop: Energy efficiency and end-use data Paris, France By Eileen M. O Brien, Team Lead, Buildings Surveys Statistics Team U.S. Energy Information Administration Independent Statistics & Analysis

2 Overview Governance of energy consumption data in the U.S. The Energy Consumption and Efficiency Statistics Program: RECS as an example New directions going forward

3 Mission of the U.S. Energy Information Administration The U.S. Energy Information Administration (EIA) collects, analyzes, and disseminates independent and impartial energy information to promote sound policymaking, efficient markets, and public understanding of energy and its interaction with the economy and the environment. EIA s independence is defined in its originating legislation EIA is required by law to conduct the consumption surveys EIA data are widely used; visits to exceed 2 million per month 3

4 Basic design features of RECS & CBECS drive the cost and quality Both use multi-stage area probability sample designs from sample frames constructed by EIA CBECS frame is made up of multiple frames a deduplicated, hybrid list of buildings listed in sampled areas (the area frame) complemented with special lists of large, complex buildings (e.g., airports, hospitals, etc.). Both begin with in-person, voluntary computer assisted personal interviews (CAPI) RECS is experimenting with different modes of collection, principally web and mail CBECS has used telephone collection entirely (1999) and as an opt-in cost mitigation (about 20% in 2012) Field efforts to reduce sampling error are important CBECS administers a special screener to every sampled building; RECS has also surveyed Rental Agents of apartment units for key survey items that household respondents answer poorly (e.g., type, fuel for heating) Both include a follow-on of energy suppliers; by law, suppliers must participate 4

Residential Energy Consumption Survey (RECS) RECS is the only representative, nationwide source of energy related characteristics, consumption, expenditures and end uses for U.S. homes RECS began as an annual survey, but rising costs and budget constraints changed the periodicity.

5 Residential Energy Consumption Survey (RECS) RECS is the only representative, nationwide source of energy related characteristics, consumption, expenditures and end uses for U.S. homes RECS began as an annual survey, but rising costs and budget constraints changed the periodicity. U.S. law now mandates a quadrennial cycle. The 2013 cycle was delayed two years for budget and planning purposes 5

6 RECS begins with drawing a multistage area-probability sample from a complex frame of eligible housing units Primary sampling units (PSUs) vary in size but cover about the same amount of people Secondary sampling units (segments) are groups of Census blocks within PSUs 1. Select PSUs (counties or groups of counties) Most of the residential sample frame is made up of addresses in the U.S. Postal Service Delivery Sequence File (DSF) Unit of analysis: housing units that are occupied as a primary residence 2. Select segments Main St Diagonal Ave 3. Select housing units 6

7 RECS is more than a survey. Its data are derived from a series of surveys, processes, and models. Household Survey Rental Agent Survey Energy Supplier Survey Monthly consumption & expenditures Weather data Census & climate geography Non-survey Data Statistical results energy characteristics (housing & equipment) usage behavior demographics consumption & expenditures Modeled results energy end uses by fuel 7

8 Three types of data are available from the RECS program 1. Household characteristics, including: Structural characteristics of the housing unit (levels, rooms, windows, square footage*): type, number, size Energy systems (heating, cooling and ventilation): types, fuel, age, and usage Water heating: number, type, fuel, age, usage Major appliances (refrigerators, cooking, laundry and other): number, type, age, usage Electronics (TVs, TV peripherals, computers, other): number, type, characteristics Household demographics: number, sex, race/ethnicity*, education, income, ages Big data : census geographies, weather and climate data from NOAA RECS collects both interviewer-measured and self-reported square footage. ** Race/ethnicity is only collected for the householder. Eileen O Brien, Paris, France 8

9 RECS data types (cont.) 2. Annualized energy consumption & expenditures and modeled end uses for these fuels Electricity Fuel oil Natural gas Kerosene Propane Wood 3. Public Use Microdata After ensuring that no individual households can be identified from the data, EIA releases raw data for researchers to perform their own analyses 9

10 How are RECS program data produced? Household surveys Energy supplier surveys Annualize billing data Model end uses 1. Housing characteristics data are collected from respondents on a voluntary basis 2. Consumption and expenditures are reported by energy suppliers, via web forms and standardized spreadsheets submitted through a secure server 3. EIA annualizes billing data to create totals by fuel for the survey reference year, e.g., total electricity consumed 4. End-use consumption and expenditures are the result of complex statistical models now supplemented with engineering parameters Eileen O Brien, Paris, France 10

11 To estimate the end uses, characteristics and energy supplier data feed into disaggregated end-use consumption and expenditure models Total fuel consumption Space Heating A/C Water Heating Refrigerators Appl/ Elec/ Light FUEL MODEL Space Heating A/C Water Heating Refrig-erators Appliances/ Electronics/ Lighting Electricity X X X X X Natural gas X X X Propane X X X Fuel oil X X Kerosene X More info: Eileen O Brien, Paris, France 11

12 Home energy consumption over the past 30 years is a story of competing trends Total energy consumption in homes has stayed relatively constant, but electricity now makes up a larger share Efficiency gains have offset increases in the number of appliances (ex: more than one refrigerator) and electronic plug loads Energy performance standards to improve A/C efficiency have been offset by the growing share and size of homes using it Newer homes are built with more efficient features, but this efficiency is in part offset by trends in larger homes 12

13 Shares of energy end uses have changed significantly Appliances & Electronics 1.94 (19%) Refrigerators 0.46 (5%) Total energy use in homes quadrillion Btu and percent Appliances & Electronics 3.04 (30%) Refrigerators 0.48 (5%) Water Heating 1.83 (18%) Total Space Heating 5.32 (53%) Total Space Heating 4.23 (42%) Air Conditioning 0.46 (5%) Water Heating 1.80 (18%) Air Conditioning 0.64 (6%) 13

14 Number of homes with 2 or more refrigerators is approaching 30% in the Midwest Homes with 2+ refrigerators, selected survey years Percent of homes 40% 35% 30% 25% 20% 15% 10% 5% 0% RECS 1984 RECS 1997 RECS 2009 West South Midwest Northeast April Lee Austin, TX June 4,

15 Newer homes are much more likely to have efficient features, such as double or triple-pane windows 15

16 Homes built in last two decades are much larger than homes built in earlier decades Average total square footage, ,500 3,000 2,500 2,000 1,500 1, s 1980s 1990s 2000s 16

Commercial Buildings Energy Consumption Survey (CBECS) CBECS is the only national-level data source for characteristics and energy use of commercial buildings Mandated by Congress, it has been

17 Commercial Buildings Energy Consumption Survey (CBECS) CBECS is the only national-level data source for characteristics and energy use of commercial buildings Mandated by Congress, it has been conducted every 3 to 5 years since 1979 CBECS data are an essential input to EIA s demand projections, and energy performance standards for buildings, building energy systems and other end uses 17

CBECS is also a multi-year, multi-phase project 2012 2013 2014 2015 2016 Construct frame & select sample Design survey & collect building data Design forms & collect energy supplier data

18 CBECS is also a multi-year, multi-phase project Construct frame & select sample Design survey & collect building data Design forms & collect energy supplier data Process energy data & model end uses Complete data dissemination Releases span several years Characteristics summary tables Characteristics microdata Energy summary tables Energy microdata 18

19 Many aspects of CBECS are similar to but more complex than the RECS The unit of analysis is a building not a business, which complicates many aspects of design and collection CBECS requires more effort to develop its complex frame. This ensures we can make statistical inferences from a relatively small sample to the population of buildings CBECS is the benchmark, so we must be cautious in testing and implementing new methods. 19

20 The 2012 CBECS showed the average size of new commercial buildings has increased relative to vintage buildings average commercial building size thousand square feet built before 1920 (6%) built (20%) built (24%) Percent of total commercial building stock built (32%) built (18% of total) 20

21 Health care and lodging lead the growth in average building size average building size (thousand square feet) 60 constructed before Health care Lodging Retail (other than mall) Principal building activity Religious worship 21

22 Large buildings (over 100,000 square feet) are just 2% of the building stock but about 35% of the total floorspace and 41% of consumption 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1.6% 3.6% 6.0% 15.9% 22.1% 50.0% 0.1% 0.7% 8.1% 11.0% 12.3% 14.3% 16.0% 13.7% 14.7% 14.9% 15.3% 11.8% 16.2% 12.6% 10.2% 9.3% 9.2% 10.4% Buildings Floorspace Energy consumption (Percent of total commercial) Size of buildings in square feet Over 500, ,001 to 500, ,001 to 200,000 50,001 to 100,000 25,001 to 50,000 10,001 to 25,000 5,001 to 10,000 1,001 to 5,000 EIA Secretary of Energy Brown Bag Briefing, 2012 CBECS Results, June 16,

23 With the latest CBECS (2012), we see that total electricity consumption in commercial buildings has almost doubled since 1979 quadrillion Btu Total consumption Fuel oil Electricity Natural gas District heat Year of survey 23

24 While electricity share is up 23 percentage points from 1979; natural gas share is down 12 percentage points. 100% 90% 80% 70% 60% 50% 40% 44% 32% 61% Fuel oil 0.1 Quads District heat 0.3 Quads Natural gas 2.2 Quads Electricity 4.2 Quads 30% 20% 10% 38% Total energy use in Quadrillion Btu 0% Electricity Natural gas District heat Fuel oil 24

25 Energy usage per square foot is down significantly from 2003 (-12%) kbtu per square foot Total consumption Natural gas Electricity Note: In 2012, electricity conditional floorspace totaled 84.9 billion square feet; natural gas conditional square feet totaled 58.7 billion square feet Year of survey EIA Secretary of Energy Brown Bag Briefing, 2012 CBECS Results, June 16,

26 Lighting and space heating contributed the most to reductions in energy intensities total consumption quadrillion Btu % of total energy in commercial buildings in % of total energy in commercial buildings in % of total energy 10% of total energy Totals for all end uses 6.5 Quadrillion Btu in Quadrillion Btu in 2012 CBECS End Uses 26

27 Decrease in total energy intensity driven by office and education buildings Food service Inpatient health care Food sales Other Enclosed and strip malls Lodging Outpatient health care Public order and safety Public assembly All All commercial bldgs (*) Office (*) Education (*) Retail (other than mall) Service Religious worship Warehouse and storage Vacant Total energy use per square foot (thousand Btu) (*) Office, education, and commercial buildings overall showed statistically significant decreases from 2003 to 2012 Food service Inpatient health care Food sales Other Enclosed and strip malls Lodging Outpatient health care Public order and safety Public assembly Office Education Retail (other than mall) Service Religious worship Warehouse and storage Vacant % total energy (2012) % total floorspace (2012) Percent 27

28 New directions in the consumption program Presenter name, Presentation location, Presentation date 28

29 EIA runs two complex periodic surveys of energy consumers against this agency budget profile Budget of the US Energy Information Administration mil dols US real budget (2007=base) nominal budget It costs more to achieve the same or less success Survey costs are rising faster than budgets Budgets are declining in real terms fiscal year Source: OMB, Statistical Programs of the United States Government 29

30 sample size These studies are expensive. Over time, tight budgets and rising survey costs have reduced sample size and frequency. EIA energy consumption surveys 20,000 RECS CBECS 15,000 10,000 5, survey reference year 2018? 11 surveys 8 surveys 5-6 surveys Eileen O Brien, Paris, France December 12,

31 Complexity and challenges are motiving changes in how we will conduct the consumption data program in the future. Timeliness Cost Quality Project resilience and oversight 31

32 RECS 2015 example: At week 10 (Nov 3), effort, contacts and output varied widely across RECS sample areas. Not worked No contact RECS ATD Report, Nov 3,

33 Completes Weekly Production Rates RECS 2015 in CAPI Completes by Week * Interviewer attrition training * Field blitz * Closeout week and ending date RECS 2015 ATD Report Jan

34 Multi-mode studies are going to be more common and also more important to design and monitor well Mode differences - differences in response patterns within questionnaires (e.g., conflicting responses), item missing rates, nature and amount of write in responses, and quality of responses Different response distributions across respondents, possibly resulting from varying interpretations, but also instrument differences by mode, e.g., use of screener questions Weighting the sample and imputation is more complex Differences in survey questions and quality may affect how data items may be published (e.g., we did not collect comparable detail on secondary heating equipment in web/mail) 34

35 New directions in the end uses estimation End use estimation has moved away from statistical-only based models to a hybrid approach In 2003, engineering models were incorporated into the CBECS. CBECS 2012, RECS 2015 and future rounds will include such parameters. We expect future rounds will experiment with more objective measurement of end uses. 35

36 Algorithm: Imputation, Engineering, and Calibration Models, 2003 Imputation Model Engineering Model Purpose Estimate end-uses for buildings with known totals and extrapolate to those with unknown consumption (statistical model) Calculate consumption for end uses using deterministic formulas (algebraic model) Model is fit using data from: Buildings with good consumption N/A Model is applied to: Buildings with missing consumption All buildings Calibration Model Make the total consumption from engineering model as close as possible to actual total (statistical model) Buildings with good consumption All buildings 36

37 Algorithm: Engineering, Calibration, and Imputation Models, 2012 Engineering Model Purpose Calculate consumption for individual end uses using deterministic formulas (algebraic model) Model is fit using data from: N/A Model is applied to: All buildings Calibration Model Imputation Model Make total consumption from engineering model as close as possible to actual total (statistical model) Combine end-use estimates from engineering model and parameters from calibration model to impute missing consumption data (calculation) Buildings with good consumption data N/A All buildings Buildings with missing consumption data 37

38 For more information Residential Energy Consumption Survey Commercial Buildings Energy Consumption Survey Manufacturing Energy Consumption Survey U.S. Energy Information Administration home page Energy Projections Annual Short-Term International Monthly Energy Review Today in Energy State Energy Profiles 38

39 Resources Executive Office of the President, Office of Management and Budget. (2013). Statistical programs of the U.S. government: Fiscal year Federal Committee on Statistical Methodology, Methodology Reports. Journal of Official Statistics, Statistics Sweden: Meetings and Events of the American Statistical Association, International Conference on Establishment Surveys, Conference Proceedings: Principals and Practices of a Federal Statistical Agency by the National Academy Press: Statistical Programs and Standards of the United States Summer Institute of Survey Research Techniques at the University of Michigan, Ann Arbor, MI Presenter name, Presentation location, Presentation date 39

EIA Short-Term Energy and

EIA Short-Term Energy and Winter Fuels Outlook Washington, DC U.S. Energy Information Administration Independent Statistics & Analysis www.eia.gov Overview EIA expects higher average fuel bills this winter