ATE 1046 No. of Pages 14, DTD = 4.3.

1
1 August 2003 Disk used ARTICLE IN PRESS

Applied Thermal Engineering xxx (2003) xxx–xxx

www.elsevier.com/locate/apthermeng

Ceiling radiant cooling panel capacity enhanced by

mixed convection in mechanically ventilated spaces
Jae-Weon Jeong *, Stanley A. Mumma
Department of Architectural Engineering, The Pennsylvania State University, University Park, PA 16802, USA
Received 31 March 2003; accepted 10 June 2003

Abstract
The main thrust of this research is to estimate the impact of the mixed convection effect on the cooling
capacity of a ceiling radiant panel in mechanically ventilated spaces. To estimate panel cooling capacity
enhancement caused by mixed convection, a verified analytical panel model was used. The simplified
correlation for mixed convection heat transfer coefficient which can be easily adopted in panel cooling
capacity estimation was derived from established mixed convection and natural convection correlations. It
was found that the total cooling capacity of radiant panels can be enhanced in mixed convection situations
by 5–35% under normal operating panel surface temperatures.
Ó 2003 Elsevier Ltd. All rights reserved.

Keywords: Ceiling radiant panel; Radiant cooling; Mixed convection; Natural convection; Mechanical ventilation

1. Introduction

Total heat transfer capacity of a radiant panel is determined by two heat transfer mechanisms:
convection and radiation. ASHRAE [1] proposed to use the natural convection (NC) heat transfer
coefficient developed by Min et al. [2] for estimating the convective heat flux on a ceiling panel,
and to use the mean radiant temperature method proposed by Walton [3] for calculating the
radiation heat flux.
When calculating the convective heat transfer rate of a ceiling radiant cooling panel (CRCP) in
a naturally ventilated space, the NC heat transfer coefficient can be used, but in a mechanically
ventilated room the air movement is obviously greater, and the convective heat transfer on a

*
Corresponding author. Tel.: +1-814-683-8313; fax: +1-814-863-4789.
E-mail address: jqj102@psu.edu (J.-W. Jeong).

1359-4311/$ - see front matter Ó 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S1359-4311(03)00211-4
ATE 1046 No. of Pages 14, DTD = 4.3.1
1 August 2003 Disk used ARTICLE IN PRESS

2 J.-W. Jeong, S.A. Mumma / Applied Thermal Engineering xxx (2003) xxx–xxx

Nomenclature

A area (m2 )
AUST area-weighted average temperature of un-cooled surfaces (°C)
bw bond width, (m)
Cp specific heat of the fluid (kJ/kg K)
D tube diameter (m)
De characteristic diameter of room surface (¼ 4Ac =P ) (m)
F fin effectiveness
F0 panel efficiency factor
Fc correction function (W/m2 K)
FR panel heat removal factor
hi fluid to tube heat transfer coefficient (W/m2 K)
h heat transfer coefficient (W/m2 K)
k heat conductivity of the panel (W/m K)
M mass flow rate to the panel (kg/s)
n number of tubes
P parameter of the room (m)
q heat flux to the panel (W/m2 )
q0fin transferred energy to the fin per unit length (W/m)
q0tube heat gain from above the tube region per unit length (W/m)
q0 total sensible heat gain of the panel per unit length (W/m)
T temperature (°C)
DT temperature difference between the space and the panel mean surface temperature (°C)
Uo overall heat transfer coefficient (W/m2 K)
Ue equivalent overall heat transfer coefficient (W/m2 K)
V diffuser discharge air velocity (m/s)
w distance between the tubes (m)
W width of nozzle diffuser (m)
Greeks
d panel thickness (m)
c bond thickness (m)
Subscripts
a air, space
b bond material, fin base
c ceiling, convection
f forced, fluid
i inside
m mean
n natural
o outside, total, overall
p panel
r radiation
 ATE 1046 No. of Pages 14, DTD = 4.3.1
1 August 2003 Disk used ARTICLE IN PRESS

J.-W. Jeong, S.A. Mumma / Applied Thermal Engineering xxx (2003) xxx–xxx 3

radiant panel would be expected to be enhanced; therefore, the use of the NC heat transfer co-
efficient is inappropriate for a mechanically ventilated room.
Kochend€ orfer [4] indicated that in real buildings cooling outputs of CRCPs are significantly
higher (over 25%) than measured panel capacities in the laboratory under standard testing con-
dition expressed in DIN 4715 [5]. The reasons for this higher capacity are non-standard sur-
rounding conditions, such as warm windows and outside walls, and mechanical ventilation
systems. However, those effects have not been considered in panel capacity estimation, and NC is
still a general assumption in CRCP system design. If the higher performance of CRCPs is ignored
in the design phase, unnecessary panel area is specified, and the cost of the panels is excessive.
Therefore, in this research the enhanced cooling capacity of a CRCP by mixed convection
(MC) will be estimated with verified analytical steady-state panel model. The simplified yet reli-
able correlation for MC heat transfer coefficient which can be easily adopted in cooled ceiling
capacity estimation will also be derived from the established MC and NC coefficients.

2. Literature review

2.1. Natural convection coefficient

For the last couple of decades, many researchers focused on buoyancy driven heat transfer
coefficients for small free-edge heated plates. Widely used NC heat transfer coefficients are those
in the ASHRAE Handbook Fundamentals [7] and Alamdari and Hammond [8]. However, the use
of above coefficients to room surfaces is somewhat debatable, because the air movement over a
surface in a room is different from that over a small free-edge heated plate.
Some researchers carried out their experiments using full-size enclosures [9–11]. The work of
Min et al. [2] is the first to investigate the NC heat transfer in full-size enclosures. They used three
differently sized testing chambers for their experiments to consider the room size effect. Their
correlation for a cooled ceiling or heated floor is presented in Eq. (1).

hc ¼ 2:13
ðTa  Tpm Þ0:31 ð1Þ

2.2. Mixed convection coefficient

The need for the MC heat transfer coefficient has been growing because the NC heat transfer
coefficient cannot accurately estimate the convective heat flux from a heated or cooled surface in a
mechanically ventilated room. However, there has been a few works into convective heat transfer
from the heated or cooled surfaces in ventilated rooms. Chen et al. [12] performed experiments to
determine MC heat transfer coefficients for the floor and ceiling of the room. The experiments
were carried out in a 5.6 m  3.0 m  3.2 m chamber with two mechanically driven air inlets
located at the floor level. They proposed the MC heat transfer coefficient for cooled ceiling (4.0 W/
m2 K) for mechanically driven ventilation rates ranging from 3 to 7 air changes per hour (ACH).
Spitler et al. [13] also conducted measurements for the convection heat transfer coefficient in a
rectangular office-sized enclosure (5.48 m  3.65 m  3.35 m). Although their experiments were
ATE 1046 No. of Pages 14, DTD = 4.3.1
1 August 2003 Disk used ARTICLE IN PRESS

4 J.-W. Jeong, S.A. Mumma / Applied Thermal Engineering xxx (2003) xxx–xxx

focused on the forced convection (FC) heat transfer coefficient with large ventilation rates (15–100
ACH), they found that the space could fall into the MC flow regime at the low ventilation rate.
Fisher and Pedersen [14] proposed MC heat transfer coefficients for the low ventilation rate (3–
12 ACH) using Spitler et al.Õs test chamber. Actually, their correlations for walls, floor, and ceiling
were derived for the isothermal room where buoyancy caused by surface-to-air temperature
differences is negligible. However, they found that the correlations can be applied to non-
isothermal rooms with surface temperature differences of less than 20 °C. Eq. (2) is their MC
correlation for a ceiling.
hc ¼ 0:49
ACH 0:8 ð2Þ
Beausoleil-Morrison [15] built MC heat transfer correlations by blending well known NC and FC
correlations using the Churchill and UsagiÕs approach [16] which has been used to correlate heat
transfer problems which are governed by two or more driving forces. He assumed that, when FC
effect is overwhelmed by buoyancy, surface convection can be adequately characterized with the
Alamdari and Hammond [8] correlations. Whereas when FC effect is dominant, the FisherÕs
correlations [17] are sufficient to calculate surface convection.
The most noted study is by Awbi and Hatton [18]. They proposed MC heat transfer coefficients
for heated room surfaces partially covered by air jet. Experiments were conducted in an office-
sized, well-insulated environmental chamber (2.78 m  2.78 m  2.3 m) with heating plates fixed to
the internal surfaces. A fan box with an adjustable nozzle was placed at one end of the surfaces
within the chamber to create the convective flow. Their correlation for a heated floor or cooled
ceiling is presented in Eqs. (3).
3:2 1=3:2
hc ¼ ðh3:2
cn þ hcf Þ ð3aÞ

2:175 0:308
hcn ¼ ðTa  Tpm Þ ð3bÞ
D0:076
e

hcf ¼ 4:25
W 0:575
V 0:557 ð3cÞ

3. Simplified mixed convection coefficient

3.1. Simplified correlation for mixed convection

In a mechanically ventilated space, the convection heat transfer on a heated (or cooled) surface
is occurred by the combination effect of both NC and FC on the panel surface. Pamelee and
Huebscher [19] tried to include the FC effect on the heat transfer from heated panels as an in-
crement to be added to the NC coefficient. However, their work was not widely accepted by the
industry because one could not be sure the validity of their predicted FC effects.
Recently, Awbi and Hatton [18] proposed noticeable MC correlations for heated room surfaces
by correlating the combined effects of both NC and FC on each surface as a function of four
parameters; characteristic diameter of a space (De ), space-to-panel temperature difference (DT ),
diffuser width (W ), and diffuser discharge air velocity (V ).
 ATE 1046 No. of Pages 14, DTD = 4.3.1
1 August 2003 Disk used ARTICLE IN PRESS

J.-W. Jeong, S.A. Mumma / Applied Thermal Engineering xxx (2003) xxx–xxx 5

In this research, Awbi and HattonÕs MC correlation for a cooled ceiling Eq. (3) was extensively
analyzed, and it was found that the characteristic diameter of a space (De ) or the space size effect
has negligible impact on the MC heat transfer coefficient on a CRCP in most spaces usually faced
in the HVAC design projects. As a consequence of this observation, a simplified MC correlation
which can be easily adopted in panel cooling capacity estimation without concerning the space
size effect was derived as follows.
As shown in Eq. (4), the simplified MC correlation has the form of adding the FC effect to well
known Min et al.Õs NC correlation Eq. (1) widely used to estimate design cooling capacity of
CRCPs. The correction function Fc in Eq. (4) represents additional FC effect on a panel in a
mechanically ventilated room.
hc ¼ Fc þ 2:13
DT 0:31 ð4Þ
In principle, the difference between the MC heat transfer coefficient and the NC heat transfer
coefficient represents the impact of the FC on a panel. The FC effect data required to derive the
correction function (Fc ) were collected by calculating differences between Awbi and HattonÕs MC
coefficients and Min et al.Õs NC coefficients calculated by Eqs. (3) and (1) respectively for the
various values of four parameters: characteristic diameter of a space (De ), space-to-panel tem-
perature difference (DT ), diffuser width (W ), and diffuser discharged air velocity (V ). These FC
effect data were analyzed statistically, and it was found that the space size effect (De ) has insig-
nificant impact on the convection heat transfer on a CRCP surface. Finally, the correction
function (Fc ) for considering additional FC effect on a panel was derived as a function of three
parameters (i.e. DT , W , and V ). More details of Fc were presented in the following sections.

3.2. 2k factorial experiment design approach

The 2k factorial experiment design method [20] was used to determine which parameters and
their interactions would show significant effects on the increment of the convection heat transfer,
and to derive the correction function (Fc ) with a small number of experiments (or calculations).
The superscript k means the number of parameters considered in an experiment. In the 2k factorial
experiment design, it is typical to select only two values (i.e. minimum and maximum) for each
parameter. Table 1 shows the ranges of four parameters considered in this research. Once the
individual parameters and their interactions which have significant effects on a dependent pa-
rameter are identified, a linear correlation for the dependent parameter can be developed as a
function of those parameters.

Table 1
Typical ranges of each parameter
Label Parameter Low High
A Temperature difference between the space 1 14
and the panel mean surface temperature
(DT ), °C
B Diffuser discharged air velocity (V ), m/s 2 6
C Characteristic diameter of a space (De ), m 1 30
D Diffuser width (W ), m 0.2 0.8
ATE 1046 No. of Pages 14, DTD = 4.3.1
1 August 2003 Disk used ARTICLE IN PRESS

6 J.-W. Jeong, S.A. Mumma / Applied Thermal Engineering xxx (2003) xxx–xxx

In this particular case, the dependent parameter is the difference between the MC and the NC
coefficient (Dhc ), and the full factorial (or 24 ) experiment can be performed without making as-
sumptions to reduce the number of calculations because only four parameters (i.e. De , DT , W , and
V ) are considered in this problem.
Individual parameters or interactions which have significant effects on the dependent parameter
can be identified by a simple way proposed by Daniel [21]. He suggested examining a normal
probability plot of the effects of each parameter. Effects are defined as the average change in
response that occurs as a result of changing each parameter from its low value to its high value.
Formulas for computing effects can be found in statistics texts [20]. According to this method, the
negligible effects are normally distributed, and will tend to fall along a straight line representing
the normal probability distribution, whereas significant effects will not lie along the straight line.

3.3. Derivation of the correction function Fc

Fig. 1 clearly shows that three individual parameters (DT , V , W ) and one interaction (V
W )
have significant effect on the dependent parameter (Dhc ). Based on this analysis result, a first order
regression equation for the correction function (Fc ) was derived as a function of those parameters
Eq. (5). The coefficients for Fc were presented in Table 2.
Fc ¼ f ðDT ; V ; W Þ ¼ a0 þ a1 ðDT Þ þ a2 ðV Þ þ a3 ðW Þ þ a4 ðV
W Þ ð5Þ

On the other hand, the regression equation can be verified by the normal probability plot of
residuals or the differences between the actual Dhc (collected data) and the predicted Dhc by Eq.
(5). If the points on this plot lie reasonably close to the normal probability distribution line, one

Fig. 1. Normal probability plot of effects on response variable Dhc .

 ATE 1046 No. of Pages 14, DTD = 4.3.1
1 August 2003 Disk used ARTICLE IN PRESS

J.-W. Jeong, S.A. Mumma / Applied Thermal Engineering xxx (2003) xxx–xxx 7

Table 2
Coefficients for correction function
a0 a1 a2 a3 a4
0.28021 )0.13931 0.11416 1.25013 1.22058

Fig. 2. Normal probability plot of residuals of the correction function Fc .

can conclude that the regression equation is satisfactory. Fig. 2 clearly shows that Eq. (5) is
satisfactory.
By substituting Eq. (5) into Eq. (4), the simplified MC correlation is represented. The proposed
correlation will return to the NC correlation if a space is not mechanically ventilated (i.e. V and W
equal 0), although there is a small deviation (i.e. a0 þ a1 ðDT Þ) from original Min et alÕs NC co-
efficients. To estimate the MC impact on a cooling capacity of a CRCP, the proposed correlation
was integrated into the verified analytical CRCP model presented in the following section.

4. Ceiling radiant cooling panel model

The analytical CRCP model developed by Conroy and Mumma [6] was modified to consider
the MC effect on panel cooling capacity. Fig. 3 shows the cross section and the geometry of a
CRCP.
The temperature distribution between the tubes Eq. (6) was derived from the energy balance on
the panel (fin) element with temporarily assuming that the temperature gradient in the flow
ATE 1046 No. of Pages 14, DTD = 4.3.1
1 August 2003 Disk used ARTICLE IN PRESS

8 J.-W. Jeong, S.A. Mumma / Applied Thermal Engineering xxx (2003) xxx–xxx

Fig. 3. CRCP cross section and the geometry.

direction (y-direction) is negligible. The energy transferred to the fin base per unit length in the
flow direction (q0fin ) is calculated with Eq. (7). The fin effectiveness (F ) is a ratio of the actual heat
transfer to the ideal heat transfer when the entire fin is at its base temperature Tb Eq. (8).
Tp ðxÞ  Ta coshðmxÞ
¼ ð6Þ
Tb  Ta cosh½mðw  Do Þ=2

q0fin ¼ FUo ðw  Do ÞðTb  Ta Þ ð7Þ

tanh½mðw  Do Þ=2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

F ¼ where; m ¼ Uo =k
d ð8Þ
mðw  Do Þ=2
The sensible heat gain of the panel also includes the heat gain from above the tube region Eq. (9).
The temperature above the tubes can be considered at uniform temperature (Tb ). By adding Eqs.
(7) and (9), the total sensible heat gain of the panel per unit length (q0 ) is determined Eq. (10).

q0tube ¼ Do Uo ðTb  Ta Þ ð9Þ

q0 ¼ ½ðw  Do ÞF þ Do 
Uo
ðTb  Ta Þ ð10Þ

The resistance to heat flow to the fluid results from the bond and the tube-to-fluid resistance Eq.
(11). By eliminating Tb from Eqs. (10) and (11), q0 can be expressed in terms of known dimensions,
physical parameters, and the local fluid temperature (Tf ) as shown in Eq. (12). The panel efficiency
factor (F 0 ) Eq. (13) is a ratio of overall heat transfer coefficient between fluid and room to overall
heat transfer coefficient between fin and room. For more simplicity, the bond and the tube-to-fluid
resistance are neglected, then F 0 can be reduced to Eq. (14).
Tb  Tf
q0 ¼ ð11Þ
1 c
þ
hi pDi kb bw

q0 ¼ wF 0 Uo ðTf  Ta Þ ð12Þ

1=Uo
F0 ¼   ð13Þ
1 1 c
w þ þ
Uo ½Do þ ðw  Do ÞF  hi pDi kb bw
 ATE 1046 No. of Pages 14, DTD = 4.3.1
1 August 2003 Disk used ARTICLE IN PRESS

J.-W. Jeong, S.A. Mumma / Applied Thermal Engineering xxx (2003) xxx–xxx 9

Do þ ðw  Do Þ
F
F0 ffi ð14Þ
w
The temperature distribution in the flow direction was derived from the mass and energy balance.
If the fluid enters the panel at temperature Tfi and increases in temperature until at the exit it is Tfo ,
the fluid temperature at any position in flow direction can be expressed as Eq. (15).
 
Tf ðyÞ  Ta nUo wF 0
¼ exp  y ð15Þ
Tfi  Ta MCp
If the panel length is L, the mean fluid temperature (Tfm ) is found by integrating Eq. (15) from y ¼
0 to L. Performing this integration, and after some algebraic manipulation, the mean fluid tem-
perature can be expressed as follows;
 
qo FR
Tfm ¼ Tfi þ 1 0 ð16Þ
FR Uo F
The panel heat removal factor (FR ) relates the actual sensible heat gain of a panel to the heat
gain if the whole panel surface were at the fluid inlet temperature Eq. (17), then the total sensible
heat flux of the panel (qo ) can be expressed as Eq. (18). The total sensible heat flux can also be
expressed in terms of the mean panel temperature (Tpm ) Eq. (19). By equating these two equations
and solving for Tpm , the expression for the mean panel temperature is derived as shown in Eq. (20).
MCp ðTfo  Tfi Þ
FR ¼ ð17Þ
Ap Uo ðTa  Tfi Þ

qo ¼ FR Uo ðTa  Tfi Þ ð18Þ

qo ¼ Uo ðTa  Tpm Þ ð19Þ

MCp ðTfo  Tfi Þ

Tpm ¼ Tfi þ
ð1  FR Þ ð20Þ
A p F R Uo
In principal, the total heat flux (qo ) is the summation of the convection heat flux (qc ) and the
radiation flux (qr ) as shown in Eq. (21a), and each heat flux can be expressed as Eqs. (21b) and
(21c), respectively.
qo ¼ qc þ qr ð21aÞ

qc ¼ hc
ðTa  Tpm Þ ð21bÞ

qr ¼ hr
ðAUST  Tpm Þ ð21cÞ

The overall heat transfer coefficient Uo can be easily determined by summing the convection
heat transfer coefficient (hc ) and the radiation heat transfer coefficient (hr ) if Ta ¼ AUST; however,
it is usually not true. Therefore, the equivalent overall heat transfer coefficient (Ue ) was defined
Eq. (22).
qo
Ue ¼ ð22Þ
ðTa  Tpm Þ
ATE 1046 No. of Pages 14, DTD = 4.3.1
1 August 2003 Disk used ARTICLE IN PRESS

10 J.-W. Jeong, S.A. Mumma / Applied Thermal Engineering xxx (2003) xxx–xxx

By substituting Eq. (21) into Eq. (22),

qc þ qr ðAUST  Tpm Þ
Ue ¼ ¼ hc þ hr
ð23Þ
ðTa  Tpm Þ ðTa  Tpm Þ
The simplified mixed convection heat transfer coefficient Eq. (4) was derived in previous section,
and the radiation heat transfer coefficient Eq. (24) was found in the literature [1].
h i
2 2
hr ¼ 5  108
ðAUST þ 273Þ þ ðTpm þ 273Þ
½ðAUST þ 273Þ þ ðTpm þ 273Þ ð24Þ

An approximated expression for AUST given by Kilkis [22] was used in this study.
AUST  Ta  d
z ð25aÞ

7
zffi where; 26 °C 6 Toa 6 36 °C ð25bÞ
ðToa  45Þ
Where, d is the room position index; d is 0.5 for an interior space, 1.0 for a room with one outdoor
exposed side with fenestration less than 5% of the total room surface area or 2.0 for a room with
fenestration greater than 5%, and 3.0 is for a room with two or more outdoor exposed sides.
The equivalent overall heat transfer coefficient Eq. (23) can replace the overall heat transfer
coefficient (Uo ) of the panel model; however, Ue can not be determined explicitly because the mean
panel surface temperature (Tpm ) is unknown. This unknown is determined by solving panel model
equations Eqs. (6)–(20) and Eq. (23) iteratively for given boundary conditions. Once Ue and Tpm
are converged to certain values, other quantities such as the panel cooling capacities (qo , qc , and
qr ), heat transfer coefficients (hc , and hr ) are determined.

5. Panel cooling capacity estimation

5.1. Model space

It is assumed that five aluminum radiant cooling panels (0.6 m  3 m) are installed on the
ceiling of a 3 m  3 m  3 m model space (Fig. 4). The topside of the panel is perfectly insulated,

Fig. 4. Schematic of model space.

 ATE 1046 No. of Pages 14, DTD = 4.3.1
1 August 2003 Disk used ARTICLE IN PRESS

J.-W. Jeong, S.A. Mumma / Applied Thermal Engineering xxx (2003) xxx–xxx 11

and four-row of parallel copper tubes is attached on the topside of each panel by high conduc-
tivity bond material. The tube outside diameter is 0.01 m, and the mass flow rate is 0.04 kg/s for
each panel. The thermal resistance between the panel and the fluid is negligible. The model space
has one exterior wall with small window area. The outdoor air (OA) temperature is 30 °C, and the
space temperature is maintained at 26 °C. The air is supplied by the 0.5 m  0.05 m nozzle diffuser
located on a wall near the ceiling. The diffuser discharge air velocity varies from 2 to 6 m/s, and
the panel inlet chilled water temperature changes from 12 to 25 °C.

5.2. Convection and radiation heat fluxes

Fig. 5(A)–(C) clearly shows that the convection heat flux (qc ) is enhanced by increasing dis-
charge air velocity (V ). The qc values for simplified correlation, Awbi and HattonÕs, and Fisher
and PedersenÕs correlation increase with discharge air velocity except for Chen et al.Õs coefficient.
Chen et al.Õs coefficient is inherently insensitive to the diffuser discharge air velocity because they
proposed a constant value (4W/m2 K) as a MC coefficient for cooled ceiling.

Fig. 5. Convection and radiation heat flux.

ATE 1046 No. of Pages 14, DTD = 4.3.1
1 August 2003 Disk used ARTICLE IN PRESS

12 J.-W. Jeong, S.A. Mumma / Applied Thermal Engineering xxx (2003) xxx–xxx

Note that the simplified correlation and Awbi and HattonÕs correlation return higher con-
vection heat fluxes than those determined by Min et al.Õs NC correlation. Whereas convection heat
fluxes for Fisher and PedersenÕs and Chen et al.Õs MC correlation are not much different from or
even smaller than those for the NC correlation.
Fig. 5(D) shows that radiation heat fluxes (qr ) are not much affected by diffuser discharge air
velocities and kind of convection coefficients since the panels operate with a very small water
temperature rise or a nearly constant surface temperature.

6. Cooling capacity enhancement

6.1. Total cooling capacity

In Fig. 6, as the discharge air velocity increases from 2 to 6 m/s, the total cooling capacity
(convection heat flux plus radiation heat flux) of the panel increases significantly due to the
convection heat transfer enhancement, while the radiation part of the panel capacity is not much
affected by the MC effect. On the other hand, when the diffuser discharge air velocity is less than 2
m/s, the total panel cooling capacity is not enhanced noticeably.
The percent of total cooling capacity enhancement caused by the MC effect is presented in Fig.
7. In practice, the inlet chilled water temperature of the cooling panel is around 15 °C (with 16 °C
design surface temperature). In Fig. 7, the total cooling capacity is enhanced by the MC effect
from 5% (V ¼ 2 m/s) to 35% (V ¼ 6 m/s) at 15 °C panel inlet chilled water temperature. This result

Fig. 6. Total cooling capacity of the panel.

 ATE 1046 No. of Pages 14, DTD = 4.3.1
1 August 2003 Disk used ARTICLE IN PRESS

J.-W. Jeong, S.A. Mumma / Applied Thermal Engineering xxx (2003) xxx–xxx 13

Fig. 7. Panel cooling capacity enhancement.

corresponds very well to the experiment results of Kochend€ orfer [4]. His field measurement for
real buildings showed the cooling output of the panel increases over 25% compared with DIN
4715 test results. This increased capacity will finally reduce required panel area and initial cost.

7. Conclusions

The objective of this piece of work was to show the impact of the mixed convection on the
cooling capacity of a ceiling radiant panel, which occurred in most of the mechanically ventilated
space. The total cooling capacity of the panel was enhanced significantly with increasing diffuser
discharge air velocity from 5% to 35% when the panel was at typical design temperature. It means
that engineers can use 5–35% increased panel cooling capacity usually overlooked at the design
stage. This increased capacity will finally reduce required panel area and initial cost. However,
when the diffuser discharge air velocity is less than 2m/s, the impact of MC on the panel cooling
capacity is small. Therefore the correlation for natural convection heat transfer coefficient can be
used to estimate panel cooling capacity instead of the MC correlation for low velocities.

References

[1] ASHRAE, ASHRAE Systems and Equipment Handbook, American Society of Heating, Refrigerating and Air-
Conditioning Engineers Inc., Atlanta, GA, 2000.
[2] T.C. Min, L.F. Schutrum, G.V. Parmelee, J.D. Vouris, Natural convection and radiation in a panel heated room,
Heating Piping and Air Conditioning (HPAC) (1956) 153–160.
[3] G.N. Walton, A new algorithm for radiant interchange in room loads calculations, ASHRAE Transactions 86 (2)
(1980) 190–208.
[4] C. Kochend€ orfer, Standard testing of cooling panels and their use in system planning, ASHRAE Transactions 102
(1) (1996) 651–658.
ATE 1046 No. of Pages 14, DTD = 4.3.1
1 August 2003 Disk used ARTICLE IN PRESS

14 J.-W. Jeong, S.A. Mumma / Applied Thermal Engineering xxx (2003) xxx–xxx

[5] DIN, DIN 4715––Cooling surfaces for rooms; part 1: measuring of the performance with free flow, Deutsches
Institut fur Normung, 1997.
[6] C.L. Conroy, S.A. Mumma, Ceiling radiant cooling panels as a viable distributed parallel sensible cooling
technology integrated with dedicated outdoor air systems, ASHRAE Transactions 107 (1) (2001) 578–585.
[7] ASHRAE, ASHRAE Handbook Fundamentals, American Society of Heating, Refrigerating and Air-Condition-
ing Engineers Inc., Atlanta, GA, 2001.
[8] F. Alamdari, G.P. Hammond, Improved correlations for Buoyancy-driven convection in rooms, Building Service
Engineering Research and Technology 4 (1983) 106–112.
[9] A.J.N. Khalifa, R.H. Marshall, Validation of heat transfer coefficients on interior building surfaces using a real-
sized indoor test cell, International Journal of Heat and Mass Transfer 33 (1990) 2219–2236.
[10] S.R. Delaforce, E.R. Hitchen, D.M.T. Watson, Convective heat transfer at internal surfaces, Building and
Environment 28 (1993) 211–220.
[11] H.B. Awbi, A. Hatton, Natural convection from heated room surfaces, Energy and Buildings 30 (1999) 233–244.
[12] Q. Chen, C. Meyers, J. v.d.r Kooi, Convective heat transfer in rooms with mixed convection, International Seminar
on Indoor Air Flow Patterns in Ventilated Spaces, Liege, Belgium, Feb., 1989, 69–82.
[13] J. Spitler, C.O. Pedersen, D.E. Fisher, Interior convective heat transfer in buildings with large ventilative flow rates,
ASHRAE Transactions 97 (1) (1991) 505–515.
[14] D.E. Fisher, C.O. Pedersen, Convective heat transfer in building energy and thermal load calculations, ASHRAE
Transactions 103 (2) (1997) 137–148.
[15] I. Beausoleil-Morrison, An algorithm for calculating convection coefficients for internal building surfaces for the
case of mixed flow in rooms, Energy and Buildings 33 (2001) 351–361.
[16] S.W. Churchill, R. Usagi, A general expression for the correlation of rates of transfer and other phenomena,
AIChE Journal 18 (6) (1972) 1121–1128.
[17] D.E. Fisher, An experimental investigation of mixed convection heat transfer in a rectangular enclosure, Ph.D.
thesis, University of Illinois at Urbana-Champaign, 1995.
[18] H.B. Awbi, A. Hatton, Mixed convection from heated room surfaces, Energy and Buildings 32 (2000) 153–166.
[19] G.V. Pamelee, R.G. Huebscher, Forced convection, heat transfer from flat surfaces, ASHVE Transactions 53
(1947) 245.
[20] D. Montgomery, Design and Analysis of Experiments, fifth ed., John Wiley and Sons, 2001.
[21] C. Daniel, Use of Half-Normal Plots in Interpreting Factorial Two Level Experiments, Technometrics, vol. 1, 1959,
pp. 311–342.
[22] B.I. Kilkis, S.S. Sager, M. Uludag, A simplified model for radiant heating and cooling panels, Simulation Practice
and Theory 2 (2) (1994) 61–76.