ASHRAE Journal - February 2023 - 58

TECHNICAL FEATURE
variable or whether the use is constant or variable. This
classification makes a distinction between constant or
varying loads (i.e., different rates at which the system
uses energy) versus constant or varying uses (i.e., different
rates at which the system is used) primarily for
purposes of measurement (see Table 1 at https://tinyurl.
com/JournalExtras. This results in the following four
classifications:
1. Constant load, constant use;
2. Constant load, variable use;
3. Variable load, constant use; and
4. Variable load, variable use.
To learn how these four classifications are then used to
classify the type of measurements (i.e., sufficient: onetime
measurements, runtime measurements or continuous
measurements) that need to be made, see Table 2
at https://tinyurl.com/JournalExtras.
The Guideline's appendices provide additional advice
to guide the user in applying the retrofit isolation
approach for pumps (six methods), fans (five methods),
chillers (five methods), boilers and furnaces (12 methods),
lighting (six methods), HVAC systems (four airside
methods), and unitary and condensing equipment
(three methods).
Whole-Building Energy Approach
The Whole-Building Energy Approach is appropriate
when the total building performance is being calculated
versus the performance of a specific retrofit (i.e., retrofit
isolation). Two compliance paths were created for the
Whole-Building Approach, which include a prescriptive
path and a performance path. The Whole-Building
Prescriptive Path is appropriate for projects where the
savings are expected to be greater than 10% of the energy
or demand use, and requires that the pre-retrofit and
post-retrofit data be continuous and complete. The prescriptive
path does not allow for any data to be excluded
from either the baseline model or the post-retrofit
model and has specific requirements on the statistical
goodness-of-fit indicators.
If one is using the Whole-Building Approach but cannot
comply with the requirements of the prescriptive path,
then the Whole-Building Performance Approach can be
followed. This path allows for data gaps and other sorts of
data irregularities by requiring the user to show that the
calculated uncertainty in the cumulative savings be less
than 50% of the total savings reported for the post retrofit
58
ASHRAE JOURNAL ashrae.o rg
F E B R U A RY 2023
reporting period (with a confidence level of 68%).
The Whole-Building Path requires that the user collect
periodic utility data for the facility and includes
stored (i.e., dated inventory or delivery readings of
coal, liquid natural gas, or oil, etc., use) and nonstored
(i.e., dated readings of electricity, steam, or
pipeline-supplied gas, etc., use) energy sources, and
demand data (i.e., amount and date of peak electric,
steam, or pipeline-supplied gas, etc.).
The Whole-Building Method also allows for the use
of whole-building interval data (i.e., 15-minute or
hourly data). Such data are necessary for cost savings
calculations that include time-of-use charges, time-ofday
or real-time electricity pricing. In most instances,
regression models based upon daily data provide the
best statistical goodness of fit.3 Hourly interval data
can provide more accurate insight into the building's
energy use characteristics, which can be useful in
determining why a building's post-retrofit operation
may be performing below expectations, or for use in
fine-tuning a building's energy consuming systems.4,5
Unfortunately, the use of interval data also requires
the collection and storage of coincident weather data
(i.e., hourly dry-bulb temperature, humidity, solar and
wind data) from a reliable, long-term, nearby source
such as the National Weather Service.6
In most cases the models used for the Whole-Building
Energy method will take the form of a linear, changepoint
linear, or multiple variable equations:
E = C + B1V1 + B2V2 + B3V3 + A1Vn + . . .
where
E = Energy use or demand estimated by the equation
C = Constant term in [energy units/day] or [demand
units/billing period]
Bn = Coefficient of independent variable Vn in [energy
units/driving variable units/time interval] or [demand
units/driving variable units/time interval]
A1 = Coefficient of the independent variable for any
adjustment(s)
Vn = Independent driving variable
Models that have been recognized as the most appropriate
for modeling monthly and daily commercial building
energy use include constant or mean models, dayadjusted
models, two-parameter linear models, three,
https://tinyurl.com/journalextras https://tinyurl.com/journalextras https://tinyurl.com/JournalExtras http://ashrae.org
ASHRAE Journal - February 2023

Table of Contents for the Digital Edition of ASHRAE Journal - February 2023
