Caution: Pitfalls Using Department of Energy energy Use and OpenEI energy and hourly load data for utility customer market-oriented analysis
US Department of Energy residential and commercial databases provide information on a small sample of US utility customers
as well as engineering modeled estimates of 8760 hourly loads developed by NREL, National Renewable Energy Laboratory and provided on
the OpenEI Web site.
Pitfall 1: Applying Department of Energy Residential (RECS) and Commercial (CBECS) energy use/customer data to assess energy-related marketsThe US Department of Energy EIA's RECS (Residential Energy Consumption Survey) and CBECS (Commercial Buildings Energy Consumption Survey) surveys provide annual energy use information on individual utility customers throughout the United States. While RECS and CBECS data are useful for some national and large region analysis, small survey sample sizes on the order of 6,000 - 8000 customer records means that developing information for smaller geographic areas or market segments result in unacceptably large confidence intervals. For example, see the description above where, the CBECS 95% confidence interval on food sales fuels consumption in the West Census region is +/- 95%. Drilling down to more building-specific segments defined by say square feet, electric space heat or other variables generates even more uncertainty. MAISY Utility Customer Databases overcome this inability to evaluate market segments by detailed geographic areas, building and occupant characteristics and other variables by maintaining databases composed of more than 7 million utility customer records. See the section above "MAISY Database Accuracy/Comparison With Other Data Sources" for a more detailed comparison of the Department of Energy's RECS/CBECS and Jackson Associates utility customer databases.
Pitfall 2: Applying The Department of Energy's OpenEI NREL Hourly Loads data to assess energy-related marketsThe US Department of Energy funded development of several large energy use and hourly load data sets currently available on OpenEI NREL.org. Data sets reflect engineering model-based energy use and hourly loads for 16 commercial building types and 3 residential building specifications for 1020 TMY3 weather locations. Model-simulated hourly kW loads data are presented for several individual end uses (heating, air conditioning, etc.) for each building type/location.
Prototype building shell and other model inputs are documented in OpenEI-linked references. Specifications for residential prototype buildings are presented in the graphic below. These datasets appear to have been developed and provided on OpenEI in 2013.
It is not clear who the DOE intended to use these data; however it is clear that the prototpye building engineering energy/hourly load analyses ARE NOT APPROPRITE for market-related evaluations such as market sizing, product design/development, sales and profitability analysis and other market-oriented applications. The OpenEi energy and hourly load databases ARE NOT RELIABLE when used for the following applications:
OpenEi NREL prototype building energy use and hourly loads:
1. Market Segment Mismatch Market-based information nearly always requires data for selected market segments.
Energy/hourly loads of customer segments defined by income, householder age, electric appliance holdings and other variables will be significantly different than energy and hourly loads of OpenEI NREL prototype engineering simulation results.
For example, energy technology sales targeting applications often target certain income segments while smart grid applications typically need to differentiate among customer segments based on contributions to peak electricity demands during system peak periods.
The chart below shows a comparison of OpenEI NREL load data for the first 7 days in the high energy-use residential dwelling unit for Napa, California and data for a sample of MAISY utility customer records for the same location. The 73 records in the MAISY data sample are selected to conform to the OpenEI NREL dwelling unit model parameters including natural gas space and water heating, electric cooking, electric clothes dryer. The average dwelling unit size in the MAISY sample is 3,000 square matching the OpenEI NREL dwelling unit assumption.
To illustrate the importance of segment-based analysis, the MAISY sample of customers was divided into three household income categories with boundaries of less than $75,000, between $75,000 and $150,000 and greater than $150,000. Assuming that a technology a sales strategy targets households with income above $75,000, the chart shows load shapes for these two income categories. What the chart also shows is that there is a significant difference between mid- and high- income load profiles. High income customers have significantly greater peak kW and greater kW variation than middle-income customers.
A comparison of loads for the two categories shown below the chart provides the basis for several conclusions
shows several that s that this difference in
2. Limited Representation of Utility Customer Diversity OpenEI NREL residential engineering model simulations are based on dwelling unit/appliance characteristics documented in the following graphic. Three building designs are specified for each of five climate categories. Engineering model simulations are conducted using the three building designs for each of the 1020 TMY3 weather station locations.
However, a quick review of prototype buildings shows an extremely limited representation of the diversity in utility customer characteristics and energy use. Three dwelling unit sizes are represented with 1,000, 2,000 and 3,000 square feet prototypes. Energy use determining characteristics within each size category are similarly limited. For example in the mixed dry/hot dry climate zone, space heating and water heating fuels are specified as natural gas. If we assume a reasonably typical average annual electricity use of 8,000 kWh for a 3,000 square foot California dwelling unit represented by this high-use prototype, consider what would happen if the prototype used electricity for water heating and space heating. The annual use would increase to about 12,000 kWh with a significantly different hourly load profile. Electric space heating increases overnight loads, often peaking in the morning and declining through the day. Water heating load profiles typically reflect peaks in the morning and later in the day. Adding in a second refrigerator would increase annual kWh to about 13,000 kWh while a swimming pool pump would bring the total to more than 16,000 annual kWh with accompanying changes in the kW load profile. Other diversity to be considered includes household income, number of family members, clothes dryer fuel, use of a free-standing freezer, etc.
The question for market analysis is what dwelling unit/appliance/demographics/operational characteristics are likely to define the market segments of interest, how do the segments differ in energy/hourly loads characteristics and how many potential customers are in each segment. As indicated in the section above, utility customer segment profiles can differ dramatically across many of these dwelling unit, occupancy and appliance combinations. The OpenEI NREL data provides no energy or hourly load information on these important issues.
Conclusion: OpenEI NREL hourly loads reflect extremely limited characterizations of dwelling unit structure, occupancy and appliance choices that are unlikely to reflect actual utility customers associated with energy market related analysis.
3. Questionable Load Profiles
Engineering building models are notoriously poor at reflecting actual building load profiles. While weather-sensitive end use components reflect traditional heat-load model analysis (which may or may not be of acceptable accuracy), non-weather sensitive end use load profiles reflect assumed values. Since nonweather sensitive loads typically reflect the vast majority of electricity use, relying on the OpenEI NREL assumed nonweather sensitive load profiles is risky.
MAISY nonweather sensitive loads are based on metered appliance data reflecting more accurate and more realistic load data. The chart below shows a comparison of OpenEI NREL and MAISY nonweather sensitive loads for an area around Napa California. The OpenEI NREL data reflect the high-energy use model which assumes gas space heating and gas water heating, electric cooking, electric clothes dryer. The MAISY data reflect the average of 48 residential dwelling units averaging 3,000 square feet also reflecting the same appliance characteristics as the OpenEI NREL prototype building
Several observations can be drawn from this comparison based on the fact that MAISY hourly loads are based on actual metered data and OpenEI NREL loads are based on engineering model assumed loads:
Click Here for additional detail on MAISY Database content, variables, and software.
OTHER MAISY LINKS
Products and Services
Smart Grid Analysis
Energy Budgets at Risk