U.S. Energy Information Administration - EIA - Independent Statistics and Analysis
Commercial Buildings Energy Consumption Survey (CBECS)
How was water usage information collected for commercial buildings?
CBECS 2012 - Release date: February 9, 2017
For the second time in its history, EIA has collected water usage data through its Commercial Buildings Energy Consumption Survey (CBECS). Most commercial buildings use water for purposes such as restrooms, kitchens, laundries, showers, building heating and cooling, and landscape irrigation. Information on how much water commercial buildings used was first collected experimentally in the 2007 CBECS. EIA evaluated the 2007 water data and considered the collection a success, prompting a second collection attempt in the 2012 CBECS.
This report discusses the methodology behind the 2012 CBECS water consumption data collection: the collection process, the rates at which respondents provided water consumption data, data quality checks, EIA’s evaluation of the data and publication decisions, and imputation.
Water and energy are connected in commercial buildings
The CBECS is a national sample survey that collects information on the stock of U.S. commercial buildings, including their energy-related building characteristics and energy usage data (consumption and expenditures). EIA, in cooperation with the Environmental Protection Agency (EPA), added water consumption questions to the CBECS because water and energy consumption are connected in many ways. Extracting, purifying, pumping and transporting drinking water, heating and processing water for domestic and other uses, and treating and disposing of wastewater are all activities that use energy.
In commercial buildings, energy is used to pump and heat water, and water is often critical to HVAC equipment. In certain building types, the usage of water can have a major impact on a building’s energy demand and performance. Historically there were no nationwide estimates of water consumption in the commercial buildings sector, and the CBECS was believed to be an appropriate way to collect this data.
Water consumption was collected from CBECS building respondents
The CBECS sample of buildings was selected in a multi-stage, multi-frame area probability frame designed to be representative of the entire commercial building stock in the United States. For more details on the sample design, see How Were Buildings Selected for the 2012 CBECS?. After the sample has been selected, CBECS interviewers collect building characteristics, energy usage data (consumption and costs) , and total water volume from a respondent at the building.
In the water usage section of the 2012 CBECS Buildings Survey questionnaire, respondents were asked to report the total volume of water used (consumption) in 2012, the units of measure (gallons, thousand gallons, ccf, etc.), whether the volume was metered or estimated by the utility, if the sewer flow was metered, if water was used outside the building, and how much of the water was used outside. If the building had a cooling tower, respondents were asked the volume of water used for cooling towers.
EIA asked about the total cost of water (expenditures) in 2007 but not in 2012, after determining that many of the water expenditures included other charges such as sewer costs. These costs can be a significant percentage of the total expenditures, and there was no established methodology for disaggregating the water commodity costs.
Prior to the CBECS interview, respondents were provided with materials to prepare for the interview. A worksheet with the requested water data items was included so the respondent could gather the information, if available.
The full set of water questions begins on page 233 of the CBECS survey form.
The water consumption reporting rate was 31%
The reporting rate for water consumption in the 2012 CBECS was 31%; that is, 31% of the responding buildings that use water were able to provide their 2012 consumption. This rate is substantially lower than the 2007 reporting rate, when 44% provided consumption.
The reporting rates vary by building size and activity. Reporting rates increase with building size; the smallest buildings—5,000 square feet or less—have a rate of 19%, while buildings larger than 1 million square feet have a reporting rate of 56%, as shown in Table 1. Laboratory, inpatient healthcare, and refrigerated warehouse buildings have the highest reporting rates.
After performing data quality checks, EIA decided to publish water data only for large buildings
EIA conducted data quality checks on the water data, including determining if:
- Consumption data were only for a portion of the building
- Consumption data did not cover the whole year
- Gallons per square foot were higher or lower than an expected range for that building type
Buildings flagged by the data quality checks were reviewed by EIA analysts. The analysts reviewed all available information and decided that either a) the data were acceptable as is, b) the data needed a change (for example, the units of measure were incorrect), or c) the data were incorrect and could not be used.
After removing the incorrect and unusable data, only 26% of the buildings that used water had valid water consumption data. Table 1 shows the rates after the data quality checks, again showing that larger buildings have higher rates of reporting consumption data. As a result of the low rate of reported data for smaller buildings, EIA decided to publish estimates only for large buildings greater than 200,000 square feet. Of these large buildings, 40% reported data that passed EIA’s quality checks.
|Before data quality checks||After data quality checks|
|1,001 to 5,000 square feet||19%||17%|
|5,001 to 10,000 square feet||23%||20%|
|10,001 to 25,000 square feet||25%||22%|
|25,001 to 50,000 square feet||33%||28%|
|50,001 to 100,000 square feet||35%||29%|
|100,001 to 200,000 square feet||38%||33%|
|200,001 to 500,000 square feet||44%||37%|
|500,001 to 1 million square feet||53%||45%|
|Over 1 million square feet||56%||50%|
|Large buildings > 200,000 square feet||48%||40%|
|Public order and safety||35%||31%|
|Outpatient health care||26%||24%|
|Inpatient health care||51%||44%|
|Retail other than mall||23%||23%|
Missing water consumption for large buildings was imputed with a non-linear model
The water consumption for large buildings that could not accurately provide their data was imputed using a non-linear regression model, similar to the method used to impute missing energy consumption variables. For each building, an initial engineering estimate of approximately how much water that type of building uses was calculated using a measure of size and the estimated water usage per the measure of size. Frequently the measure of size was square footage, but for specialized building types such as hospitals and schools, the measure of size was more closely related to the building’s activity type. These initial estimates were based on research of water and sewer requirements of various building types  .
The regression model used actual water consumption as the dependent variable and fit an equation using these initial engineering estimates, plus the following predictors: estimates of water used by a cooling tower if present, cooling degree days, percent of the building that is cooled, building activity, weekly operating hours, months of use, water used outside (and if it is metered), whether the building is on a campus, year constructed, number of employees, laundry, cooking, pool, large amounts of hot water, and electricity consumption.
Results are published in tables, a report, and a public use microdata file on EIA’s website
For an overview of the results, detailed tables, and public use microdata, see Water Consumption in Large Buildings Summary.
Frequently Asked Questions (FAQ’s)
Why does EIA only provide water consumption data for large buildings? After extensive data quality checks, EIA determined that the reporting rate and data quality of water consumption data for large buildings greater than 200,000 square feet met EIA’s quality standards for publication. The reporting rate for smaller buildings was too low to build a satisfactory imputation model.
Why are there no food sales or food service buildings in the tables? The 2012 CBECS sample selected zero food sales or food service buildings that are greater than 200,000 square feet, either because none exist in the population, or so few exist that they were not sampled.
Do the figures include water used outside the building such as for irrigation or landscaping? As with energy sources, the intended scope of water consumption data collection is water used inside the building. Some buildings use water outside for purposes such as irrigation and landscaping, and this water may or may not be separately metered. If it is separately metered, the outside amount is subtracted from the total. If it is not metered and therefore cannot be subtracted from the total to get the usage inside the building and the usage was minimal (as determined by water intensity data quality checks), the outside water was included in the figures. If the outside water was separately metered but the respondent could not provide the outside consumption, or if there was some other indication that the amount used outside was not minimal, the total building consumption was imputed. Of the large buildings that provided water data, 58% said that some was used outside. In those cases, 33% separately metered it, and most were able to provide the outside water consumption, which was subtracted from the total water consumption figure.
Do the figures include water from wells? The CBECS questionnaire did not ask the source of the water. If the building used well water that was measured, and if water consumption was provided and passed the data quality checks, it is included in the figures.
Is there information on cooling tower water consumption? Cooling towers, which discard excess heat from HVAC equipment, can use substantial amounts of water in a commercial building. If a building had a cooling tower, the respondent was asked to provide the cooling tower consumption if it was separately metered. Only 8% of the buildings had a cooling tower and could provide this figure, which was not enough buildings for EIA to publish accurate estimates about cooling tower consumption._____________________________