Professional Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Article history: The evaluation of building energy consumption usually requires building energy profiles on an hourly
Received 13 November 2009 basis. Computer simulations can be used to obtain this information but generating simulations requires
Received in revised form 3 June 2010 a significant amount of experience, time, and effort to enter detailed building parameters. This paper
Accepted 23 July 2010
presents a simple methodology to estimate hourly electrical and fuel energy consumption of a building
by applying a series of predetermined coefficients to the monthly energy consumption data from electrical
Keywords:
and fuel utility bills. The advantage of having predetermined coefficients is that it relieves the user from
Energy consumption estimation
the burden of performing a detailed dynamic simulation of the building. The coefficients provided to the
Building energy consumption
Benchmark Models
user are obtained by running EnergyPlus Benchmark Models simulations; thus, the simulation process
EnergyPlus is transparent to the user. The methodology has been applied to a hypothetical building placed both in
Atlanta, GA, and in Meridian, MS, and in both cases, errors obtained for the estimated hourly energy
consumption are mainly within 10%.
© 2010 Elsevier B.V. All rights reserved.
1. Introduction input data and time from even experienced users [1,2]. Further-
more, simulation tools may not be cost-effective at the first stage
The Energy Information Administration of the U.S. Department of analysis, which makes others tools, such as screening tools, a
of Energy (DOE), in the 2008 annual energy review presents the better option. Several methodologies to estimate energy consump-
following energy distribution for End-User Sectors: transportation tion have been developed. Some of them are based on statistics and
28%, industry 31%, commercial 19%, and residential 22%. Buildings, other on simulations [1,3–9]. On-line building energy predictions
commercial and residential, account for 41% of the total energy based on neural networks and genetic algorithms [10–12] can also
consumption in the U.S., which justify a variety of initiatives for be used in some applications. In general, it is accepted that weather
building energy consumption reduction. As examples, the Building data can play an important role on forecasting energy consumption
Technologies Program of the DOE Energy Efficiency and Renew- in buildings [3]. Papa et al. [13] proposed a normalized energy use
able Energy (EERE) Office is working to achieve the goal of net-zero index (NEUI) based on a temperature function. In their work, they
energy buildings, and the U.S. Green Building Council promotes discussed the influence of weather variables such as solar radiation
the design, construction, and operation of high performance green and air velocity, and conclude that temperature is the most impor-
buildings through the Leadership in Energy and Environmental tant factor on energy consumption. Their reasoning is that, since
Design (LEED) program. the equipment daily energy consumption is always the same, and
Energy consumption analysis of buildings is a difficult task because there is not significant variation of daily routine, changes
because it requires considering detailed interactions among the in HVAC energy consumption is predominantly a function of tem-
building, HVAC system, and surroundings (weather) as well as perature. To obtain the temperature function to compute the NEUI,
obtaining mathematical/physical models that are effective in char- they used EnergyPlus as a simulation tool. Since even more detailed
acterizing each of those items. The dynamic behavior of the weather building simulations will not reproduce exactly the energy con-
conditions and building operation, and the presence of multiple sumption profile, there is an accepted degree of uncertainty in the
variables, requires the use of computer aid in the design and oper- estimated energy demands as consequence of the accuracy of the
ation of high energy performance buildings. Drawbacks in using tool and inputs that the user needs to be aware of in order to make
computer simulations include the considerable amount of detailed the final conclusions [2,14,15].
Knowledge of energy demand profile and energy consump-
tion of buildings facilitate the implementation of actions to reduce
∗ Corresponding author. Tel.: +1 662 325 6711; fax: +1 662 325 7223. building operational costs [3,14]. For instance, hourly energy
E-mail address: fumo@me.msstate.edu (N. Fumo). consumption data is required for the application of different screen-
0378-7788/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.enbuild.2010.07.027
2332 N. Fumo et al. / Energy and Buildings 42 (2010) 2331–2337
ters defining the heat and mass transfer flow in the actual building • Baseline electricity (Be ):
under accepted statistical weather conditions. Building:Electricity + ExteriorLights:Electricity
EnergyPlus Benchmark Models are proposed as references of • Variable electricity (Ve ):
what is a typical building. The statistical analysis behind the bench- Electricity:Facility − (Building:Electricity +
marks makes them good samples of the entire stock. Therefore, ExteriorLights:Electricity)
beyond the limitations for a particular building with specific char- • Baseline fuel (Bf ):
acteristics, any building that matches the type of building and Gas:Facility − Heating:Gas
climate zone of a Benchmark Model should have an energy con- • Variable fuel (Vf ):
sumption profile similar to the benchmark within a reasonable Heating:Gas
uncertainty for most of the energy analysis.
This study introduces the definition of “EnergyPlus normalized Once the hourly energy consumption is obtained, each value
energy consumption coefficients” (E + NECC) as normalized energy is divided by the corresponding monthly energy consumption to
profiles to estimate hourly building energy consumption from obtain the normalized energy consumption coefficient for that
utility bills information. Hourly energy consumption from the sim- hour of that particular day of the month. Mathematically can be
ulation of the Benchmark Model is used to obtain the E + NECC expressed as
for the building type and climate zone represented by the Bench-
(Be )i
mark Model. The building actual energy consumption obtained (Be,coef )i = (1)
from utility bills is multiplied by the E + NECC to estimate the build- (Be )i m
ing hourly energy consumption. Section 4.2 describes how to obtain
the E + NECC and Section 5 presents an example of how the E + NECC (Ve )i
(Ve,coef )i = (2)
can be used to estimate hourly energy consumption. (Ve )i m
(Bf )i
4.2. E + NECC (Bf,coef )i = (3)
(Bf )i m
In order to increase accuracy based on the time interval between
(Vf )i
one time of the analysis and the following, the simulation of a (Vf,coef )i = (4)
Benchmark Model is performed with a timestep of 15 min. Report- (Vf )i m
ing frequency can be as low as the timestep or as high as annual,
where i is the hour based on a 8760 h a year, and m (month) indi-
but the most common reporting frequency is hourly. Thus, this
cates that the coefficient is computed for the hour (i) corresponding
study focuses on obtaining hourly energy consumption. Outputs
to the month. Since weekdays have similar operational loads and
from EnergyPlus simulations include meters that allow reporting
schedules, but different from weekend days, it is important to
all kind of energy consumption at different points of the energy
mention that the first day for simulations of Benchmark Models
flow path and at the end point where the energy is actually
is Sunday.
used. Generally, analysis of energy consumption requires informa-
While Vf actually represents the fuel thermal energy for space
tion of the baseload and the total consumption. Therefore, in this
heating, for cooling an additional analysis is required. Ve accounts
study five sets of “normalized energy consumption coefficients” are
for all components of the HVAC system to handle the cooling and
obtained. Each set represents the normalized energy profile or how
heating loads. For cooling, the electricity used by auxiliary equip-
the energy consumption is distributed in an hourly basis. In this
ment does not exist or is commonly neglected, leaving the variable
methodology five sets are proposed. One set for the baseline for
energy consumption composed by cooling equipment electricity
electricity, one set for the baseline of fuel, one set for the variable
(chillers) and air-side equipment (fans) electricity. Fans’ energy
electricity, one set for the variable natural gas use, and one set for
consumption is not necessarily proportional to cooling load since
cooling electricity. As suggested by other authors, and what seems
the fans can be operating as constant volume while the cooling load
to be a logical approach, for typical days (weekly or weekend days)
is variable. Therefore, to estimate electricity from cooling, the set
that keep the same operational loads and schedules, the variation
of cooling electricity coefficients is obtained as
in energy consumption is due to weather conditions. Therefore, the
mentioned sets of energy consumption are calculated based on this
• Cooling electricity (Ce ):
approach.
Using the meters output variables from EnergyPlus simulations, Cooling:Electricity
the energy consumption sets are obtained as
(Ce )i
(Ce,coef )i = (5)
(Ce )i m
Due to seasonal factors, the baseline for electricity and fuel con-
sumption is not exactly the same for all the months. In addition,
since no all months have the same number of days, the baseline
energy consumption does not have the same contribution on the
monthly utility bill. As example, Fig. 1 illustrates how the contribu-
tion of the baseline, electricity and fuel, on the total monthly energy
consumption varies along the year for the benchmark representing
Small Office Buildings in Atlanta, GA (Climate Zone 3A). Therefore,
the methodology also incorporates the “baseline fraction contribu-
tion factors” of the baseline, electricity and fuel, for the monthly
energy consumption. These factors are the measure of how much
the baseline represent from the total monthly energy consump-
Fig. 1. Baseline contribution on the month energy consumption for the benchmark tion. Mathematically, these baseline fraction contribution factors
building for Climate Zone 3A. are defined as
2334 N. Fumo et al. / Energy and Buildings 42 (2010) 2331–2337
Table 1
Sample data of E + NECC for Small Office Buildings in Climate Zone 3A.
Be Ve Bf Vf Ce
Fraction factors
Bf • Energy consumption for baseline fuel (ECBf ):
(fBf ) = (7)
m Bf + Vf
m
(ECBf )i = Fm (fBf ) [(Bf,coef )i ]m
m
Since variable electricity embraces cooling and fans electricity,
in order to apply the cooling electricity coefficients, an additional where Fm is the monthly fuel consumption from the utility bills.
• Energy consumption for variable fuel (ECVf ):
set of fraction factors is required to define the contribution of the
cooling electricity on the monthly variable electricity. Mathemati-
cally, the cooling fraction factors are defined as (ECVf )i = Fm [1 − (fBf ) ][(Vf,coef )i ]m
m
C
e • Energy consumption for cooling (ECCe ):
(fCe )m = (8)
Ve m
(ECCe )i = Em [1 − (fBe )m ](fCe )m [(Ce,coef )i ]m
The five sets of hourly energy consumption coefficients (Be,coef ,
Ve,coef , Bf,coef , Vf,coef , Ce,coef ) and the three sets of fraction factors (fBe ,
fBf , fCe ) constitute the E + NECC.
4.4. Uncertainty
Table 1 shows an example of the E + NECC for the Benchmark
Model representing Small Office Buildings in Climate Zone 3A. Since
Uncertainty in building energy simulations is related to the
it is impractical to present in this paper a table with the coefficients
models and methods used to simulate the flow of energy and mass,
for each of the 8760 h, Table 1 presents only selected coefficients
the inputs describing the characteristics of the building and its
and the full set of fraction factors.
operation, and the weather file. The dynamic behavior of the sys-
tem (building) and the amount of details required to reproduce
4.3. Using the E + NECC it makes impossible to reproduce the building energy consump-
tion. However, since detailed inputs do not compromise the overall
To obtain the estimated energy consumption from energy utility result, simulations has been accepted and validated in order to be
bills using the E + NECC, the following equations must be used. continuously used for building energy consumption analysis and
optimization.
• Energy consumption for baseline electricity (ECBe ): For the development of the Benchmark Models, lack of many
details that are not available from standard data sources, assump-
tions where required to complete the models. Assumptions
(ECBe )i = Em (fBe )m [(Be,coef )i ]m included thermal zoning, aspect ratio, orientation, number of floors,
N. Fumo et al. / Energy and Buildings 42 (2010) 2331–2337 2335
Table 2
Monthly energy consumption for the example hypothetical building located in
Atlanta, GA.
window to wall ratios, HVAC types, internal loading, and schedules. (ECBf )3252 = Fmay (fBf ) [(Bf,coef )3252 ]may
may
However, the methodology used intended to minimize the poten-
tial for personal bias on selecting the characteristics of buildings
[15]. (ECBf )3252 = 1751 · 1.00 · 0.001706 = 3W
The uncertainty associated with the E + NECC is related to the
EnergyPlus as building energy simulation tool and the Benchmark • Energy consumption for variable fuel:
Models. Since EnerglyPlus is worldwide accepted and the bench-
marks are accepted by the U.S. government as national standards, (ECVf )3252 = Fmay [1 − (fBf ) ][(Vf,coef )3252 ]may
may
the E + NECC have a reasonable degree of uncertainty related to the
use of EnergyPlus and the Benchmark Models.
(ECVf )3252 = 1751(1 − 1)0 = 0W
5. Example – evaluation
• Energy consumption for cooling:
To illustrate the use of the proposed methodology, monthly
energy consumption from electrical and fuel utility bills are (ECCe )3252 = Emay [1 − (fBe )may ](fCe )may [(Ce,coef )3252 ]may
required. Since monthly data from a real building was not available,
monthly and hourly energy consumption data for a hypotheti-
cal office building located in Atlanta, GA and Meridian, MS was (ECCe )3252 = 7, 944, 763(1 − 0.67)0.50 · 0.003494 = 4, 580W
obtained using the EnergyPlus Example File Generator [26]. The
monthly energy consumption was used to apply the proposed
methodology while the hourly energy consumption was used to Following the same procedure for all hours, the hourly energy
analyze the error associated with the methodology. consumptions are estimated from utility bills information. The esti-
The E + NECC are obtained for the Small Office Buildings Bench- mated hourly energy consumptions were compared with the actual
mark Model for Atlanta–Climate Zone 3A. Appendix B shows the hourly energy consumption from the simulation of the example
main differences between the hypothetical office building and the hypothetical building, and the results summarized in Fig. 2. This
Small Office Building Benchmark Model. Tables 2 and 3 show the figure shows in the abscissa the percentage error range between
monthly energy consumption for the hypothetical building located the estimated results using the coefficients and the “actual” results
in Atlanta and Meridian, respectively, which are considered as the obtained from the Energy Plus simulation of the hypothetical build-
energy consumption obtained from utility bills. ing. The figure is interpreted as follows.
The numerical computation example (for the city of Atlanta) is There are six ranges of errors: (<−10), (−10 to −5), (−5 to −0), (0
done for the hour 3252 (May 16 at noon). The energy consumption to 5), (5 to 10), and (>10) percent. Thus, an error of 8% falls into the
is obtained from Table 2, while the coefficients and fraction factors range from ‘5 to 10%’ error. The ordinate indicates the frequency
are obtained from Table 1. of errors corresponding to an error range. Fig. 2 illustrates that for
2336 N. Fumo et al. / Energy and Buildings 42 (2010) 2331–2337
mark, it can be concluded that the closeness of the results from the
methodology to the actual building energy consumption depend
on how well the evaluated building resembles the benchmark in
terms of type of building, floor area, and climate data.