Energy Savings Potential Estimates Using CBECS and CEUS

1
Energy Savings Potential Estimates Using CBECS
and CEUS

• Michael MacDonald
• Oak Ridge National Laboratory
• macdonaldjm_at_ornl.gov
• ASHRAE SLC Annual Meeting, 6-25-08

2
What will be presented

• Brief info about CBECS and CEUS
• Brief info on building energy performance scoring
  using multivariate normalization
• Brief coverage of sectoral modeling
• Brief info on preliminary sector-wide
  multivariate normalization models for US and CA
  using CBECS and CEUS
• First-ever preliminary results on use of such
  models for estimating nationwide and CA energy
  savings potentials based on performance levels

3
CEUS, Commercial Energy Use Survey (CA)

• In 1996, new law led to first CEUS being
  conducted, with latest survey in 2003, about 60
  building types, about 80 of sector covered
• Very extensive data, used for complicated
  analyses, including calibrated simulations of
  entire commercial sector or subsectors
• Used to develop estimates of statewide floor
  stock, energy intensities, and energy usage by
  building type
• Building / site weights used to scale up to
  entire subsectors, and then results can be
  extrapolated to state levels
• 2003 data currently being studied to examine
  building energy performance system options for CA
• CA est 700,000 buildings, 6 billion sq ft in
  2003

4
CBECS, Commercial Buildings Energy Consumption
Survey

• National survey conducted periodically since
  1979, latest is 2003
• 2003 CBECS identifies about 50 commercial
  building types
• Ignores buildings less than 1,000 sq ft after the
  original 1979 NBECS survey
• Masks buildings gt 1,000,000 sq ft
• Has complicated survey weights that allow
  extrapolation to entire country
• 71 billion sq ft, almost 5 million buildings in
  2003

5
CBECS and CEUS, some important differences
6
Basic EUI Statistics kBtu/sq-ft per yr
7
(No Transcript)
8
CBECS and CEUS Data are already used for savings
potential estimates

• CBECS data provide some of the basis for the
  National Energy Modeling System (NEMS)
• CEUS data used for modeling of savings potentials
  
• Results available based primarily on
  economic-engineering models
• Results presented here are based on performance
  rating models

9
Energy Performance Methods

• Meaningful standard of comparison?
• Compare to what?
• Data sources?
• Comparison method (STD 105-2007)
• Normalization options ... past internal
• Slice-and-dice by specific characteristics
• Additional normalization, e.g., weather
• Simultaneous multivariate normalization

10
ASHRAE Handbook, Fundamentals

• Chapter 32 2005, Energy Estimating and Modeling
  Methods
• Table 10, Capabilities of Modeling Methods (p
  32.31)
• 10 modeling methods mentioned
• Multivariate linear regression is the one that
  allows simultaneous, multivariate normalization
  tools to be developed simple (sometimes), fast,
  medium accuracy (again, compared to what?)

11
Economic-Engineering Models

• Economic-engineering (E-E) models such as in NEMS
  use engineering data and analysis results to feed
  into and partially interact with an economic
  model of energy and investment
• Because change is often slow, this approach often
  works fine for certain types of forecasting
• But many types of energy improvements cannot be
  modeled reasonably, let alone well, with these
  models, and watch out if changes are fast
• To forecast total energy use, normalization of
  energy is not required, as normalized energy is
  not the desired output, but normalized energy can
  account for total energy performance, including
  operational efficiency
• New energy technologies, and impacts of those
  technologies on new buildings, are ably modeled
  in E-E tools at times, but improvements in
  operations are typically not
• Operational improvements are thus typically
  ignored

12
Page 34
13
Simultaneous Multivariate Normalization Compares
Performance

• Tools like the Energy Star buildings rating
  system have been found capable of normalizing
  about 90 of the variation in energy use between
  buildings, leaving the last 10 as the basis for
  performance rating differences
• This approach accounts for total energy
  performance, including operations (other factors
  such as IAQ typically handled separately)
• The resulting performance score or rank gives a
  specific number on building energy performance,
  but not why
• Engineering calculation tools like Energy Plus,
  DOE-2, etc, typically cannot say anything about
  how well a building performs compared to others,
  but can indicate why
• Quantification of total energy performance is
  important, and this presentation will show the
  types of information possible using sectoral-wide
  models as opposed to building type models

14
Building-Type Models

• Tools like Energy Star multivariate normalization
  tools are important for providing performance
  ratings that can be compared for specific
  building types
• But coverage is limited
• Model basis is national-average-driven
• Keep in mind that these tools allow savings
  potential for a building (type) to be calculated
  based on score
• Analysis for CA has indicated that state-level
  tools may be critical in some cases for rating
  building energy performance
• Energy Star multivariate tools may cover 60 of
  the floor area but a much smaller percentage of
  all buildings in CA
• Ratings of CA buildings using the national models
  appear to lead to fairly high rankings for some
  building types, indicating tougher normalization
  may be desirable in CA

15
Sector-Wide Models

• Sector-wide models can cover almost all buildings
  and types
• Performance rating will not be as robust as for
  building-type models, but sectoral coverage is
  essentially achieved
• Savings potential is no longer limited to a
  building (type) but can now be calculated for the
  entire sector and possibly subsectors

16
Or Other Types of Models . . .

• Entire sectors can be modeled, e.g., Buildings,
  Industry, Transportation
• Scoring can be put on a curve to grade the
  entities analyzed
• Normalization at one point in time can serve as a
  baseline to measure future improvements against

17
Example of possible grading
18
CBECS National Model Form

• Energy use index (EUI) as a function of other
  parameters
• EUI itself accounts for 65 of variation in
  energy use
• CBECS 2003 weights used
• Some data screening needed to remove problem
  facility types and include desirable parameters
• Effective R-square 0.85, F 141

19
Basic CBECS Model Parameters

• Heating and cooling degree-days
• Seating density for eating meals
• Hours of operation per week
• Personal computer density
• Worker density

20
Building / Space Types Adjusted
21
(No Transcript)
22
How is California Isolated?
23
Savings Potential based on CBECS Performance
Model Scores
24
California CEUS Model Form

• Ln(energy) as a function of other parameters,
  with Ln(SqFt) as a parameter (not EUI-based,
  heteroskedasticity would not let go)
• CEUS weights used in calculations
• Some data screening needed to only use real fuel
  data and include desirable parameters
• R-square 0.77, F 235

25
California Potential Savings
26
Where to Now?

• Comparisons of CBECS and CEUS energy
  normalization methods indicate CA likely needs
  tougher adjustments than national-average-based
  methods provide
• Several performance rating options will likely be
  available, including a sector-wide normalization
  tool, hopefully within a year
• National sector-wide normalization tools also
  appear potentially important