black>Yahoo! is not affiliated with the authors of this page or responsible for its content.
PII: S0360-1323(02)00054-9
Building and Environment 37 (2002) 857864
www.elsevier.com/locate/buildenv
On approaches to couple energy simulation and computational uid
dynamics programs
Zhiqiang Zhai
a
, Qingyan Chen
a;
, Philip Haves
b
, Joseph H. Klems
b
a
Department of Architecture, Building Technology Program, Massachusetts Institute of Technology, Room 5-418, 77 Massachusetts Ave.,
Cambridge, MA, 02139-4307 USA
b
Lawrence Berkeley National Laboratory, Berkeley, CA, USA
Abstract
Energy simulation (ES) and computational uid dynamics (CFD) can play an important role in building design by providing com-
plementary information of the building performance. However, separate applications of ES and CFD usually cannot give an accurate
prediction of building thermal and ow behavior due to the assumptions used in the applications. An integration of ES and CFD can elim-
inate many of these assumptions, since the information provided by ES and CFD is complementary. This paper describes some e cient
approaches to integrate ES and CFD, such as static and dynamic coupling strategies, in order to bridge the discontinuities of time-scale,
spatial resolution and computing speed between ES and CFD programs. This investigation further demonstrates some of the strategies
through two examples by using the EnergyPlus and MIT-CFD programs. ? 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Energy simulation; Computational uid dynamics (CFD); Integration; Building design
1. Introduction
Energy simulation (ES) and computational uid dy-
namics (CFD) programs provide complementary informa-
tion about building performance. ES programs, such as
EnergyPlus [1], provide energy analysis for a whole building
and the heating, ventilating and air conditioning (HVAC)
systems used. Space-averaged indoor environmental con-
ditions, cooling=heating loads, coil loads, and energy con-
sumption can be obtained on an hourly or sub-hourly basis
for a period of time ranging from a design day to a refer-
ence year. CFD programs, on the other hand, make detailed
predictions of thermal comfort and indoor air quality, such
as the distributions of air velocity, temperature, relative
humidity and contaminant concentrations. The distributions
can be used further to determine thermal comfort and air
quality indices such as the predicted mean vote (PMV), the
percentage of people dissatis ed (PPD) due to discomfort,
the percentage dissatis ed (PD) due to draft, ventilation
e ectiveness, and the mean age of air. With the informa-
tion from both ES and CFD calculations, a designer can Corresponding author. Tel.: +1-617-253-7714; fax: +1-617-
253-6152.
E-mail address: qchen@mit.edu (Q. Chen).
design an energy-e cient, thermally comfortable, and
healthy building.
However, most ES programs assume that the air in an
indoor space is well mixed. Those programs cannot accu-
rately predict building energy consumption for buildings
with non-uniform air temperature distributions in an indoor
space, such as those with displacement ventilation systems.
Moreover, the spatially averaged comfort information gen-
erated by the single node model of ES cannot satisfy ad-
vanced design requirements. The convective heat transfer
coe cients used in ES programs are usually empirical and
may not be accurate. Furthermore, most ES programs can-
not determine accurate air ow entering a building by natural
ventilation, while room air temperature and heating=cooling
load heavily depend on the air ow.
On the other hand, CFD can determine the temperature
distribution and convective heat transfer coe cients. CFD
can also accurately calculate natural ventilation rate driven
by wind e ect, stack e ect, or both. However, CFD needs
information from ES as inputs, such as heating=cooling load
and wall surface temperatures.
Therefore, coupling ES with CFD is very attractive, and is
the objective of the present investigation. After a brief intro-
duction of the principles of ES and CFD, the paper describes
possible approaches to couple ES and CFD. The current
study emphasizes the explicit coupling of individual ES and
0360-1323/02/$ - see front matter ? 2002 Elsevier Science Ltd. All rights reserved.
PII: S0360-1323(02)00054-9 858
Z. Zhai et al. / Building and Environment 37 (2002) 857864
q
ik
Wall
q
i,c
Room
q
i
q
ir
Fig. 1. Energy balance on the interior surface of a wall, ceiling, oor,
roof or slab.
CFD programs by exchanging information linking the two
programs. Due to the di erent physical models and numeri-
cal methods employed by ES and CFD, this study suggests
staged coupling strategies that consist of the static and dy-
namic coupling for di erent problems. The strategies e ec-
tively reduce the computing costs but preserve the accuracy
and details of the computed results, due to the complemen-
tary information from ES and CFD. This paper nally uses
an o ce and an indoor auto-racing space to demonstrate the
strategies.
2. Fundamentals of ES and CFD thermal coupling
2.1. The principles of ES
Energy balance equations for room air and surface heat
transfer are two essential equations solved by many ES pro-
grams. The energy balance equation for room air is
N
i=1
q
i;c
A
i
+ Q
other
Q
heat extraction
= V
room
C
p
T
t
;
(1)
where
N
i=1
q
i;c
A
i
is the convective heat transfer from en-
closure surfaces to room air, q
i;c
is the convective ux from
surface i, N is the number of enclosure surfaces, A
i
is the
area of surface i, Q
other
is the heat gains from lights, peo-
ple, appliances, in ltration, etc., Q
heat extraction
is the heat ex-
traction rate of the room, V
room
C
p
T= t is the energy
change in room air. is the air density, V
room
is the room
volume, C
p
is the speci c heat of air, T is the temperature
change of room air, and t is the sampling time interval,
normally 1 h.
The heat extraction rate is the same as the cooling=heating
load when the room air temperature is maintained as constant
( T =0). The convective heat ux from a wall is determined
from the energy balance equation for the wall surface, as
shown in Fig. 1. A similar energy balance can be obtained
for each window. The energy balance equation for a surface
(wall=window) can be written as
q
i
+ q
ir
=
N
k=1
q
ik
+ q
i;c
;
(2)
where q
i
is the conductive heat ux on surface i, q
ir
is
the radiative heat ux from internal heat sources and so-
lar radiation, and q
ik
is radiative heat ux from surface i to
surface k.
The q
i
can be determined by transfer functions, by weight-
ing factors, or by solutions of the discretized heat conduction
equation for the enclosure surface using the nite-di erence
method. The radiative heat ux is
q
ik
= h
ik;r
(T
i
T
k
);
(3)
where h
ik;r
is the linearized radiative heat transfer coe cient
between surfaces i and k, T
i
is the temperature of interior
surface i, and T
k
is the temperature of interior surface k, and
q
i;c
= h
c
(T
i
T
room
);
(4)
where h
c
is the convective heat transfer coe cient and T
room
is the room air temperature.
The convective heat transfer coe cient, h
c
, is unknown.
Most energy programs estimate h
c
by empirical equations
or as a constant. If the room air temperature, T
room
, is as-
sumed to be uniform and known, the interior surface tem-
peratures, T
i
, can be determined by simultaneously solving
Eq. (2). Space cooling=heating load can then be determined
from Eq. (1). Thereafter, the coil load is determined from the
heat extraction rate and the corresponding air handling pro-
cesses and HVAC system selected. With a plant model and
hour-by-hour calculation of the coil load, the energy con-
sumption of the HVAC system for a building can be deter-
mined. It is obvious that the interior convective heat transfer
from enclosures is the explicit linkage between room air and
surface energy balance equations. Its accuracy will directly
a ect the energy calculated.
2.2. The principles of CFD
CFD applies numerical techniques to solve the Navier
Stokes (NS) equations for uid ow. CFD also solves the
conservation equation of mass for the contaminant species
and the conservative equation of energy for building ther-
mal comfort and indoor air quality analysis. All the govern-
ing conservation equations can be written in the following
general form:
@
@t + (V ) 2
= S ;
(5)
where
is the V
j
for the air velocity component in the j
direction, 1 for mass continuity, T for temperature, C for
di erent gas contaminants, t is the time, V is the velocity
vector,
is the di usion coe cient, and S is the source
term.
could also stand for turbulence parameters.
C can stand for water vapor and various gaseous con-
taminants. For buoyancy-driven ows, the Buossinesq ap-
proximation, which ignores the e ect of pressure changes
on density, is usually employed. The buoyancy-driven force
is treated as a source term in the momentum equations. Z. Zhai et al. / Building and Environment 37 (2002) 857864
859
Because most room air ows are turbulent, a turbulence
model must be applied to make the ow solvable with
present computer capacity and speed.
Since the governing equations are highly non-linear and
self-coupled, it is impossible to obtain analytical solutions
for room air ow. Therefore, CFD solves the equations by
discretizing the equations with the nite-volume method.
The spatial continuum is divided into a nite number of dis-
crete cells, and nite time-steps are used for dynamic prob-
lems. The discrete equations can be solved together with the
corresponding boundary conditions. Iteration is necessary to
achieve a converged solution [2].
The accuracy of CFD prediction is highly sensitive to
the boundary conditions supplied (assumed) by the user.
The boundary conditions for CFD simulation