x
size, and flow direction. This research developed a method of test (MOT) on how to obtain the
additional information with minimal effort under the ASHRAE Standard 70-1991.

Keywords: Air distribution, environmental control, indoor air quality, outlet, measurement,
ventilation






Jelena Srebric was a Research Assistant and Qingyan (Yan) Chen is an Associate Professor at
the Building Technology Program in the Department of Architecture at the Massachusetts
Institute of Technology, MA. Srebric is now an Assistant Professor in the Department of
Architectural Engineering at the Pennsylvania State University, University Park, PA.

INTRODUCTION
When using computational fluid dynamics (CFD) for room airflow design, there are
several factors that constrain the CFD applications. The reliability and accuracy of CFD
simulations depends on several factors, such as the turbulence model, the numerical scheme, and
the boundary condition modeling. Correct modeling of the supply air boundary condition is most
crucial, because the momentum from the supply air diffuser mainly dominates the room airflow.
Since a diffuser is much smaller than a room, to model the same detail for the diffuser and the
room would require a very different length scale in the CFD modeling. Since the diffuser
determines the initial jet flow characteristic, several researchers (Emvin and Davidson 1996,
Chen and Jiang 1996) have chosen to simulate detailed diffuser geometry. To model detailed
diffuser geometry would require millions of grid cells, which in turn require a large computer
capacity, but this does not guarantee a successful simulation. On the other hand, to use the room
length scale would ignore the details in the diffuser, which could introduce errors in the
numerical simulation. Therefore, an alternative solution is to develop a simplified but reliable
method to simulate a diffuser while allowing the use of a room-length scale in CFD simulations.
With numerous studies conducted elsewhere (IEA 1993, Emvin and Davison 1996, Chen
and Jiang 1996, Nielsen 1997, and Huo et al. 1996), Chen and Srebric (2000) have investigated
systematically several simplified modeling methods for complex air diffusers. They have
identified two simplified methods, the box and momentum methods, to be most appropriate for
use in CFD simulations of indoor airflow. When the box method is used in a CFD simulation, it
needs the distributions of air velocity, air temperature, and contaminant concentrations around
the diffuser. The box size should also be correctly determined. Similarly, the momentum method
requires the airflow rate, discharge jet velocity or effective diffuser area, supply air turbulence
Srebric, J. and Chen, Q. 2001. A method of test to obtain diffuser data for CFD modeling of
room airflow, ASHRAE Transactions, 107(2), 108-116.
2
properties, supply air temperature, and contaminant concentrations. Unfortunately, not all the
parameters are available from the product catalogues of the diffuser manufacturers, nor can they
be easily obtained from calculations. It is necessary to obtain some of these data through
experimental measurements. The objective of this paper is to present a method of test (MOT) for
obtaining the data.

SIMULATION OF A DIFFUSER WITH SIMPLIFIED METHODS

This section will demonstrate how to use the box and momentum methods to simulate a
complex nozzle diffuser for room airflow modeling, as an example to show the needs for
obtaining the data. The study uses a commercial CFD program (CHAM 1998). The
demonstration will show how to apply the box and momentum methods and what data are
needed in order to develop a suitable MOT.
The nozzle diffuser is the one used for validation exercises in an international project
(IEA 1993). The diffuser consists of 84 small round nozzles arranged in four rows in an area of
28 in.
× 6.7 in. (0.71 m × 0.17 m), as shown in Figure 1(a). Each nozzle has a diameter of ½ in.
(11.8 mm), and the total airflow area for all nozzles is 0.1 ft
2
(0.0092 m
2
). The flow direction of
each nozzle is adjustable. The IEA study adjusted all the nozzles 40
o
upwards. The diffuser was
installed close to the ceiling in the center of the shorter wall in an empty test room, which is 15.8
ft (4.8 m) long, 9.8 ft (3.0 m) wide, and 8.2 ft (2.5 m) high, as shown in Figure 1(b). An exhaust
was installed underneath the nozzle diffuser.
Ewert et al. (1991) measured the air velocity using a laser Doppler anemometer (LDA).
Since the LDA measured the velocity components in three directions, the flow directions are
known. The supply airflow rate was 67 cfm (0.0315 m
3
/s), corresponding to an air change rate of
3 ACH and a maximum jet velocity of 760 fpm (3.8 m/s) at the supply plane. The decay of this
supply velocity was very fast due to the merge of the 84 small jets. The size of the recirculation
zone was approximately 8 in. x 8 in. (0.2 x 0.2 m), as shown in Figure 2. The maximum jet
velocity dropped to 300 fpm (1.5 m/s) at approximately 4 in. (0.1 m) from the diffuser, where the
small jets were already combined into one large jet. The momentum loss measured in front of the
diffuser was approximately 14% of the total momentum (Heikkinen 1991). Hence, the
momentum was not conserved. All the experimental data has been used in the present study to
validate the numerical results.
Box Method

When the box method is used to simulate the diffuser for room airflow modeling, it is
necessary to specify the distributions of air velocity, air temperature, and contaminant
concentrations in the box surface through which the flow is discharged. The best approach is to
find a suitable jet formula to calculate the distributions. For isothermal attached jets, Verhoff
(1963) proposed the following formula that is extensively used for diffuser jets (Skovgaard et al.
1990, Jacobsen and Nielsen 1992):

(
)
[
] =
68
.
0
erf
1
48
.
1
u
u
7
1
m







(1)

where u = the jet velocity in direction x at a distance y from the wall

u
m
= the maximum jet velocity in the x direction
3
Um
2
/
1
y
y
=
(
)
b
12
x
073
.
0
y
Um
2
/
1
+
=
(Rodi 1982),
(
)
45
.
0
08
.
0
2
/
1
+
=
x
y
Um
(Skovgaard et al. 1990)
x = the distance from the jet origin
b = diffuser coefficient
Although the jet from the diffuser attached to the ceiling at approximately x = 8 in. (0.2
m), it did not immediately develop into a normal attached jet. At 1.0 m (3.3 ft) from the diffuser,
the jet flowed parallel to the ceiling, but the jet profiles cannot be described by Equation (1) as
an attached jet. The jet formula can only be applied at x = 7.2 ft (2.2 m), where the jet profile can
be well predicted, as shown in Figure 3. However, the distance is too large for the box method
because the box would cover half of the ceiling length, and would influence the numerical
solution of elliptic equations (Emvin and Davidson1996). Under a non-isothermal condition, the
thermal plumes in the room would have a strong interaction with the jet, and the formula also
cannot be applied. Therefore, the jet formula is not very useful for most practical HVAC
applications.
Since the jet formula cannot be used, the box method needs measured data to set the
boundary conditions. The question is at which location the flow data should be measured. In
other words, how big should the box be? In previous studies, such as the one conducted by IEA
(1993), the box size was selected to be sufficiently small to avoid the impact of the room size
and thermal plumes. However, the box size could be larger than the domain with the
recirculation. Then, the calculation does not need to handle the difficult estimation of flow in this
recirculation area.
The commonly used box size, such as the one used by Heikkinen (1991) and Ewert et al.
(1991), is
x × y × z = 40 in. × 20 in. × 14 in. (1.0 m × 0.5 m × 0.4 m), as shown in Figure
4(a). With this box size, the velocity profile in the front surface where flow was discharged
should be described as the boundary conditions for the CFD simulation. The boundary conditions
for all the variables at all other surfaces can be assumed to be zero-gradient. The measured
velocity, as shown in Figure 5, was used to set the boundary conditions for the box method. The
data shows that the jet had a very strong three-dimensional feature. Our simulations used 3
× 3
supply-patches, as illustrated in Figure 4(a). Each of supply-patch used a velocity averaged from
the measured data for the represented area.
The figure on the right in Figure 6 compares the calculated results with the measured data
in the middle section of the room at 7.2 ft (2.2 m) from the diffuser. In the upper part of the
room, the CFD over-predicted the jet velocity by 25%. The reason for the discrepancies might be
that the box was too large, and the 3
× 3 supply patches may be too coarse for such a large box.
With such a large box, it is impossible for us to validate the calculated results at 3.3 ft (1.0 m)
from the diffuser, where the experimental data is available. At this location, the box size is the
same as the distance.
Therefore, the next simulation used a smaller box that neglects the lower part of the
measured velocity profiles, as shown in Figure 4(b). The justification for this approximation is
that the jet in the upper part of the box is responsible for most of the momentum transport.
Although the results are not presented here, the smaller box does not improve the results much.
Finally, this study used a tiny box 12 in. x 16 in. x 10 in. (0.3 m x 0.4