Energy and Buildings 103 (2015) 384395

Contents lists available at ScienceDirect

Energy and Buildings

journal homepage: www.elsevier.com/locate/enbuild

Simplied thermodynamic modeling of chilled water coils based on

bypass factors
Mridul Sarkar
Dar Al-Handasah Consultants Pvt. Ltd, Pune, India

a r t i c l e

i n f o

Article history:
Received 30 March 2015
Received in revised form 15 June 2015
Accepted 26 June 2015
Available online 2 July 2015
Keywords:
Cooling coil
Bypass factor
Sensible load
Dehumidication
Chilled water

a b s t r a c t
This paper aims towards the development of a simple mathematical model for chilled water cooling coil
based on the bypass factor as a principal input parameter. The methodology involves a temperature-based
discretization of the coil that precisely shows when the coil will be dry, wet or partially wet. This procedure
not only depicts the actual coolingdehumidication prole of the coil, but also allows calculation of
sensible and latent coil loads by the summations of individual loads of discrete coil elements. Maximum
deviations in the range of 5% for sensible and total coil loads are obtained between majorities of the
ndings from the theoretical model and experimental data available in literature. A close compliance with
actual results and ease of application make this mathematical model a good choice for energy simulation
tools.
2015 Elsevier B.V. All rights reserved.

1. Introduction
With an increased awareness for the harmful environmental
implications of over-usage of conventional energy resources and
ever-increasing energy demand, it becomes imperative that the
present energy resources be used in an efcient and sustainable
manner. In building sector, air conditioning is a major energyconsuming exponent. Hence, economic and efcient consumption
of energy for air conditioning applications is of foremost importance, if global energy conservation is to be realized.
A nned tube heat exchanger (or simply a coil) is one of the
main mechanical components of any air conditioning equipment.
It forms a thermodynamic link between the refrigerant and conditioned air that is necessary to create comfortable conditions inside
the built environments. Therefore, accurate estimates of the heat
transfer rates through these coils are necessary to optimize the
operation of whole equipment.
Threlkeld has developed the earliest detailed model for
coolingdehumidication coils [1]. His model is based on the Log
Mean Enthalpy Difference (LMED) method by assuming a linear
relationship between the saturation enthalpy and temperatures of
coolant and coil tube. Many researchers like Elmahdy [24] and
Braun [5] took Threlkelds model as a reference to propose new
coil models. Elmahdy et al. have conducted detailed and extensive
experimentation on two kinds of chilled water coils for validating

E-mail addresses: mridul.rns@gmail.com, mridul.sarkar@dargroup.com

http://dx.doi.org/10.1016/j.enbuild.2015.06.066
0378-7788/ 2015 Elsevier B.V. All rights reserved.

Threlkelds LMED model [3]. Braun, on the other hand, has developed a steady state -NTU model for a coolingdehumidifying coil
in counter ow conguration that is applicable for dry, wet and
partially wet coil conditions [5].
Researchers and investigators have used the heat transfer rates
obtained from earlier models to develop new methodologies. The
standard method as per ASHRAE Equipment Handbook [6] and
AHRI Standard 410 [7] consider overall dry surface heat transfer by the Log Mean Temperature Differences (LMTD) between
coolant and air stream, whereas for wet surface heat transfer, the
Log Mean Enthalpy Differences (LMED) between the surface condition and air stream, are considered. Wang and Hihara [8] have
developed a coolingdehumidication model for chilled water coils
based on Equivalent Dry-bulb Temperatures (EDT) to calculate the
heat and mass transfer rates and predict cooling modes of the
coil. Pirompugd et al. [9] proposed new correlations to calculate
heat and mass transfer rates based on extensive experimentation
on n-and-tube heat exchangers under coolingdehumidication
conditions. Xia et al. [10] developed a generalized LMED model,
which is based on non-unity Lewis number for calculating the heat
and mass transfer rates through an air-cooling coil under wet condition.
To simplify the analytical complexity of coolingdehumidifying
coils for air conditioning applications and to predict and evaluate their performance, many researchers have proposed various
numerical methods. Mirth and Ramadayani [11] simplied a cooling coil as a counter ow heat exchanger to develop a numerical
model that involves discretization of coil along air path and application of energy balance principles for each step till the end of

M. Sarkar / Energy and Buildings 103 (2015) 384395

Nomenclature
Symbols
Ain
area of inner side (water side) of chilled water coil
(m2 )
Aout
area of outer side (air side) of chilled water coil (m2 )
overall bypass factor of coil
BF
Cpa
specic heat capacity of dry air = 1.006 kJ kg1 K1
Cpm
specic heat capacity of moist air 1.02 kJ kg1 K1
Cpl
specic
heat
capacity
of
liquid
water =
4.186 kJ kg1 K1
Cpv
specic
heat
capacity
of
water
vapor =
1.86 kJ kg1 K1
Cplv
difference of specic heats of liquid and vapor:
2.326 kJ kg1 K1
ha
specic enthalpy of air (kJ kg1 )
hs
saturation enthalpy of air at effective surface temperature (kJ kg1 )
specic
latent
heat
of
vaporization
at
l0
273 K = 2501 kJ kg1
a
total mass ow rate of air (kg s1 )
m
v
m
mass ow rate of water (kg s1 )
ambient pressure (kPa)
Pt
Q t
coil load capacity (kW)
specic enthalpy change across the coil (kJ kg1 )
q
Ta
temperature of air ( C)
Ts
effective surface temperature (ADP) of coil ( C)
Tw
temperature of chilled water ( C)
Uin
overall heat transfer coefcient on water side
kW m2 K1
Uout
overall heat transfer coefcient on air side
kW m2 K1
Greek Symbols
Humidity ratio (kg moisture/kg dry air)

Subscripts
0
reference state 273.15 K
a
of air
avg
averaged
design (or desired)
des
dp
dew point
in (or 1) coil inlet condition (on coil)
out
coil outlet condition (off coil)
latent part of coil load
latent
sen
sensible part of coil load
sen-dry sensible part of load capacity for a dry coil
sen-wet sensible part of load capacity for a wet coil
x
of a coil element

coil. Vardhan et al. [12] extended Mirths work by discretizing

the coil into nodes along coolant path and by applying iterative
calculation for convergence of air properties at each node from
coolant inlet to outlet. Mansour et al. [13] employed a row-by-row
method for designing nned tube cooling coils and ascertaining its
coolingdehumidication prole.
In order to determine annual energy consumption by cooling
coils, Morisot et al. [14], Lemort et al. [15] and Chillar et al. [16] have
developed simple coil models for energy modeling applications.
These models derive the required heat transfer resistances applicable at non-nominal operating conditions from available catalogue
data at standard conditions. In recent years, dynamic modeling
has become more popular in order to propose various controlling
strategies for efcient functioning of coolingdehumidifying coils.

385

Among many of the works done on dynamic performance modeling, the work by Yiu et al. [17], Yao et al. [18] and Sekhar et al.
[19] are mentioned here. Yiu and his colleagues have developed a
simulation model for the dynamic performance of dry and wet cooling coils. Yao et al. studied the dynamic relationship between the
coils heat exchanges and various operating parameters by applying
classical control theory. Sekhars work is mainly concentrated on
the development of optimized controlling strategies for enhanced
dehumidication performance of oversized cooling coils for hot and
humid climates. All the cooling coil models or methodologies mentioned here have their own pros and cons, and their application is
entirely based on the objectives targeted.
Most of the thermodynamic modeling techniques for air conditioning coils thrusts more upon the accuracy of the of heat
exchanger framework rather than on their simplication and application for load estimation and energy simulations. These methods
often require detailed inputs of cooling coils and implement iterative procedures for convergence of solutions for coil loads and
supply temperatures. However, most of control sequences for cooling equipment link supply temperatures or mass ow rates or both
with the feedback from zone temperatures. Hence, in energy simulations the intended supply and controllable operating conditions
are afxed based on zone conditions. These parameters are also
determined and included in designs specications based on the
type of applications, climate and equipment.
In a similar way, the bypass factors (BF) of cooling coils at design
load conditions are available via AHU manufacturers catalogue
or customized selection software depending on the design zone
load ratios, application and type of equipment selected. The chilled
water supply temperatures and the net water temperature difference across the coil are obtained from chiller selection and water
distribution network details, respectively. Considering all these
factors, it can be surmised that modeling cooling coils with computational complexities in conjugation with whole building energy
simulation tools are not required.
On the other hand, the coil modeling methodology applied in
many load estimation and energy simulation tools neither substantiate the air supply temperature properly based on the inputs
provided for chiller plant, nor determine the chilled water ow rate
at the correct delta T and realized coil apparatus dew point temperature (ADP). For example, these tools do not model the chiller and
pumping energy consumptions accurately, when the coil is incorrectly selected at a different delta T than the plant. Literature study
reafrms that this is one of the major reasons for inefcient functioning of chillers [20]. In addition to this, many of these tools do
not consider changes in BF with airside parameters (primarily the
air mass ow rate) at part load conditions, which results in incorrect coil loads. Considering the fact that an air-handling unit runs
on part loads more than 90% of operational time, this will lead to
faulty estimation of annual cooling energies.
Hence, a relatively accurate but simplied methodology is
required to model the cooling and dehumidication processes
through a chilled water coil. A method is shown here that addresses
the solution to this problem. The overall objectives of the present
work are summarized as follows:
Development of a simple chilled water cooling coil with bypass
factor and either of chilled water delta T or required supply temperature as an additional input, instead of any detailed physical
data for the coil.
Representation of coolingdehumidication mode (wet, dry or
partially wet) of the coil at various operating conditions.
Computation of sensible and latent loads of the coil by a discrete
load summation or an element-by-element method.
Computation of coil loads at variable conditions by determining
coil bypass characteristics at a reference performance point.

386

M. Sarkar / Energy and Buildings 103 (2015) 384395

2. Mathematical modeling

temperature and the bulk uid temperature across the coil is as

per Eq. (5):

The aim of the mathematical model presented here is to

establish the psychrometric properties of moist air during the
coolingdehumidication process through a chilled water coil,
under the inuence of specic operational parameters and coil
properties. This is accomplished by applying standard mass and
energy conservation principles, augmented with simple assumptions that keep the results under acceptable error limits. Various
researchers have already stated many of these assumptions for
studying and modeling heat exchangers and HVAC systems [112].
Some of them are based on process conditions and coil parameters
as given below:
Steady state analysis.
Airwater mixture assumed to be an ideal gas.
Thermal to mass diffusivity ratio (Lewis number) is assumed as
unity.
Heat transfer coefcients of air and water are constant throughout the process.
Thermal resistance due to coil material is neglected. Thus, radial
non-variance of coil tube temperature is the major implication.
The thermal resistance due to condensate lm is far lesser than
the moist air. Hence, the difference between the thermal resistances of a coil in wet and dry conditions can be neglected.
Uniform coil-prole and n effectiveness throughout.
Chilled water and air stream are in counter-ow conguration.

Q t = (Uin
Ain )

Ts

Twout +Twin



(5)

It should be noted that the surface temperature of the coil varies

along with chilled water temperature across the coil. However,
the effective surface temperature considered above may be lesser
than the outlet water temperature, since the coil conguration is
assumed counter-ow type. Hence, a ctitious water-side thermal
resistance, 1/(Uin *), is introduced in Eq. (5) that would result in the
same overall heat transfer with the effective surface temperature
and average water temperature inside the length of tube as with
varying coil surface and chilled water temperatures.
Similarly, the overall heat transfer between air stream and the
coil surface is:
Uout Aout
Q t =
Cpm

hain haout

(6)

ln((hain hs )/(haout hs ))

Eq. (6) shows the Log Mean Enthalpy Difference (LMED)

between air and the coil that is based on the effective coil temperature and determined by surface effectiveness (BF) of the coil,
with Lewis Factor (Le) approximated to unity for moist air [5].
Equating Eqs. (5) and (6) gives:

(Uin
Ain )

(Uout Aout )

=
Cpm

hain haout



ln((hain hs )/(haout hs ))

 = x1

Ts ((Twout + Twin )/(2))

(7)
2.1. Energy balance for entire coil

The specic enthalpy of the air at any temperature and humidity

ratio is given by:

The ratio x1 shown in Eq. (7) is similar to the coil characteristic

dened by AHRI [7]. Considering turbulent ows into the coil tubes
and noting heat transfer coefcients of water and air, it can be concluded that for any case, x1 > 1. For humidity ratios encountered in
air conditioning applications, the moist air specic heat (Cpm ) used
in Eqs. (6) and (7) can be approximated by a constant value 1.02
kJ kg1 K1 without introducing signicant errors.

ha = Cpa Ta + a (2501 + Cpv Ta )

2.2. Methodology

By enthalpy balance of the whole coil:

a (hain haout ) = m
w Cpw (Twout Twin )
m

(1)

(2)

Furthermore, the effective surface temperature of the coil is

determined from the overall bypass factor by:
Ts =

(Taout BF Tain )
(1 BF)

(3)

The coil surface will be wet and results in condensation of vapor

from the incoming air stream, if its surface temperature determined
by Eq. (3) is below the inlet dew point temperature. In this case, the
saturated humidity ratio of the coil at its effective surface temperature can be easily determined as per the methodology shown in
an earlier work [21].
The outlet humidity ratio is obtained by:
out = (1 BF)s + BF in

In the present methodology, a coil is divided into discrete segments with an equal chilled water temperature drop across it. Heat
transfer and energy balance equations are applied to determine
the thermodynamic properties of air. The determined properties at
the exit of a particular coil section are taken as the inlet condition
for the proceeding segment. This procedure is continued, until the
actual end of the coil is reached. Since, the ows of air stream and
chilled water along the coil are in countercurrent conguration, the
incoming chilled water is at the air supply end of coil.
For a typical section of coil as shown in Fig. 1:
Twi+1 = Twi Twn
where,

(4)

An approximate formulation for the overall heat transfer

between the coil and refrigerant at an effective coil surface

Twn =

Twout Twin
n

Tw
n

n in Eq. (8) is the number of segments of the coil considered.

Fig. 1. Schematic of coil discretization.

(8)

M. Sarkar / Energy and Buildings 103 (2015) 384395

By energy balance across the discretized section:

w Cpw (Twn )
a (hai hai+1 ) = m
m

(9)

The surface heat transfer to the air stream and chilled water
along the coil section gives:

(Uin Ain )i

Txi

Twi+1 + Twi



(Uout Aout )i
(hai hxi )
Cpm

(10)

On rearranging the terms:

(hai hxi )
(Uin Ain )i


=
(Uout Aout )i
Cpm Txi ((Twi+1 + Twi )/(2))

(11)

LHS of Eq. (11) gives the ratios of specic thermal resistances

of the considered coil section. As per literature and the considered assumption, the thermal resistance due to condensate formed
on a coil surface is insignicant in comparison with the remaining
components, and can be neglected [1]. Hence, one can write:

(Uin Ain )i
=
(Uout Aout )i

(Uin Ain )i

=
(Uout Aout )i

Uin Ain
=r
Uout Aout

(12)

From Eqs. (11) and (12), it can be concluded that:

(hai hxi )

 =r

Cpm Txi ((Twi+1 + Twi )/(2))

(13)

The saturation enthalpy of a wet surface is approximated by the

quadratic equation below as stated in the literature [22]:
hxi = 10.90748 + 1.22048Txi + 0.05652Tx2i

hxi = aTxi 2 + bTxi + c

where a, b and c are the regression coefcients. Table 1 shows typical values of these coefcients in variations of saturation enthalpy
and temperature at different ambient pressures. A saturation temperature range of 425 C is considered here.
On substituting Eq. (14) into Eq. (13) and rearranging terms:
0.05652Txi 2 + (1.22048 + rCpm )Txi + 10.90748 hai rCpm Twi

2 4

(16)

where = 0.05652, = 1.22048 + rCpm ,  = 10.90748 hai

rCpm Twi + rCpm (Twn /2). Again, the humidity ratio corresponding
to the surface temperature given by Eq. (16) is determined by the
same methodology mentioned in preceding section.
The local bypass factor for this particular section is given by:
BFxi =

hai+1 hxi
hai hxi

Tai+1 Txi
Tai Txi

Tai+1 = (1 BFxi )Txi + BFxi Tai

(17)

and the humidity ratio by Eq. (18) as:

i+1 =

hai+1 Cpa Tai+1

(18)

2501 + Cpv Tai+1

If the solution of Eq. (16) is greater than the inlet DPT, the particular
coil section will be dry and Eqs. (14) and (15) are invalid. In this case,
the surface enthalpy is substituted in Eq. (13) as:
hxi = Cpa Txi + in (2501 + Cpv Txi )
which gives:

Txi =

rCpm Twi + hai rCpm (Twn /2) 2501in


(19)

(rCpm + Cpa + Cpv in )

Eqs. (17) and (18) are applied as before to obtain the exit condition
of the discretized section. It is to be noted that Eq. (19) should be
applied only when two conditions are met:
Overall effective temperature of the coil is lesser than the inlet
DPT
Surface temperature of a discretized coil element determined by
Eq. (16) is higher than inlet DPT.
In short, a combination of Eqs. (16) and (19) are applicable for
partially wet coils. The whole coil will be dry when the inlet DPT
is higher than effective surface temperature. In that case, discretization of coil is not required to obtain the cooling prole of the
coil.
2.3. Coolingdehumidication prole for non-dry coils
As per ASHRAE handbook [6], the total coil heat transfer for a
fully wet coil is given below by Eq. (20):
Q t = Uin Ain

(Tsin Twout ) (Tsout Twin )

(20)

ln((Tsin Twout )/(Tsout Twin ))

Braun [5] has stated that a partially wet coil condition can
be approximated by assuming fully wet or fully dry conditions,
whichever predicts the highest heat transfer rates. This assumption
does not result in an error of more than 5% for total heat transfer
rates. To simplify the model, Eq. (20) is considered for both fully
wet and partially wet coil conditions here.
From Eqs. (6) and (20):

(hain haout ) ln (Tsin Twout )/(Tsout Twin )

Uin Ain
=



Uout Aout
Cpm ln((hain hs )/(haout hs ))
(Tsin Twout ) (Tsout Twin )

(21)

(15)

The logical root of Eq. (15) is given by:

Txi =

The air stream temperature at the exit of this section is given

by:

(14)

It is to be noted that greater deviations in actual enthalpy from

the above equation are seen with higher variations in ambient
pressures from the reference condition considered above. In that
case, a new regression trendline may be applied based on dew
point temperatures and corresponding saturation enthalpy. In general, regardless of the ambient pressure, the variation of saturation
enthalpy with temperature can be simplied in a quadratic form
as:

+ rCpm (Twn /2) = 0

387

In Eq. (21), the surface temperature of coil at air inlet and exit
points are unknown parameters. However, the method from the
previous section entails a procedure to obtain the surface temperature of each discretized coil element. To apply this, one requires the
factor r. Hence, for initializing the discretization, a simple assumption is made as:
Uin Ain
= r = x1
Uout Aout
By considering x1 = r and substituting it in Eqs. (12)(19), rst set
of coil surface temperatures at inlet and exit points are obtained.
These are substituted back in Eq. (21) and the procedure is iterated until convergence is attained. The factor r is revised after jth
iteration step as:

388

M. Sarkar / Energy and Buildings 103 (2015) 384395

Table 1
Regression coefcients of quadratic function dened for saturation enthalpy.
Coefcients of trendline function: hxi = aTx2 + bTxi + c

Pt (kPa)

108.386
101.325
99
80
75

xj+1 =

Cpm

Coefcient of determination (R2 )

(hain haout ) ln

0.0493
0.0529
0.0541
0.0687
0.0737

1.2557
1.2735
1.2815
1.3141
1.331

10.218
10.925
11.166
14.056
15.007

0.9999
0.9999
0.9999
0.9999
0.9999

((Tsin )j (Twout )avg )/((Tsout )j (Twin )avg )

ln((hain hs )/(haout hs ))

[{(Tsin )j (Twout )avg } {(Tsout )j (Twin )avg }]

(22)

j1

Eq. (22) is expressed in terms of surface temperatures and average water temperatures at air inlet and exit sections. The total heat
transfer rate for the coil after jth iteration march is:
[(Q t )calc ]j+1 = (Uin Ain )j+1

[{(Tsin )j+1 (Twout )avg } {(Tsout )j+1 (Twin )avg }]

ln

 (23)

((Tsin )j+1 (Twout )avg )/((Tsout )j+1 (Twin )avg )

where,
(Uin Ain )j+1 = xj+1 (Uout Aout )
1
a Cpm ln
(Uout Aout ) = m
BF
Fig. 2 below summarizes the steps involved to determine the
coolingdehumidication prole of wet and partially wet coils.
2.4. Bypass factor determination
Applying the more popular -NTU model for heat exchangers,
the bypass factor for a cooling coil can be expressed in terms of
Number of Heat Transfer Units (NTU) as [5]:

BF = exp

Uout Aout
a Cpm
m

(24)

More often, the performance of a cooling coil or an evaporator

is inferred from the manufacturers catalogue, which is given in
terms of main operating conditions (at design load) and particular geometrical parameters pertaining to the coil. It is possible to
ascertain the performance of a particular coil with xed geometry
and physical attributes at varied operating conditions by considering a reference performance point. As per literature, the bypass
factor of a coil with xed geometry is dependent entirely on the air
mass ow rate [23]. The bypass factor is simply expressed as:

 X
0

BF = exp

a
m

(25)

The factor X0 shown in Eq. (24) above is derived from catalogue

at a reference air mass ow rate and bypass factor. With X0 , the
bypass factors at other air mass ow rates, thus, follow Eq. (25).

Fig. 2. Flow chart to determine coolingdehumidication prole of wet and partially wet coils.

Following cases are considered for a chilled water coil, which

is divided into n discretized sections based on the methodology
discussed above.
Case 1 For a completely dry coil:
qsen-dry = (Cpa + in Cpv )[Tain Taout ]

(26.1)

2.5. Net sensible and latent loads

qsen-wet = 0

(26.2)

A simple methodology for determining the net sensible and

latent load of an ideal coolingdehumidication process in terms
of discrete summations has been established in a previous work
[24]. The mathematical model presented above discusses an actual
coolingdehumidication process in terms of key input variables.
A similar methodology can be applied here and net sensible and
latent loads for the coil can be determined by the summations of
corresponding loads of discretized coil sections.

qlatent = 0

(26.3)

Case 2 For a completely wet coil:

i [1, n], Txi < Tdpin

qsen-dry = 0

(27.1)

Table 2
Comparison of heat transfer rates between the model and experiments by Elmahdy and Mitalas on 8-row and 4-row coils.
#

Velocity (m/s)

Entering air
DBT ( C)

Chilled water

(kg/kg)

Velocity (m/s)

Entering air

DBT ( C)

(kg/kg)

Model vs. Experimental data for 4 row coil

1a
4.61
23.3
0.0070
2
4.41
28.5
0.0151
3
1.58
35.6
0.0074
0.90
27.3
0.0067
4
2.40
25.3
0.0072
5
3.10
21.9
0.0069
6
7
1.86
35.5
0.0165
2.76
29.9
0.0150
8
3.39
24.8
0.0135
9
10
4.16
31.2
0.0157
4.35
36.1
0.0180
11
3.74
23.3
0.0129
12
3.06
30.4
0.0155
13
2.97
24.1
0.0136
14
3.22
37.3
0.0171
15
3.94
26.6
0.0143
16
a

Entering ( C)

Leaving ( C)

8.3
8.5
8.3
8.3
8.2
8.2
8.5
8.2
8.2
8.4
8.4
8.7
8.4
8.2
8.7
8.4

10.4
12.4
9.5
10.1
11.1
10.8
11.1
11.9
11.8
10.8
11.9
11.9
10.9
9.9
12.2
11.7

Chilled water

X0

Model BF

0.437

3.684

(*A)

X0

0.100
0.132
0.005
0.030
0.068
0.013
0.001
0.003
0.029
0.003
0.024
0.060
0.086
0.042
0.089
0.115

Model BF

Model ma
(kg s1 )

1.60
1.82
0.69
1.05
1.37
0.85
0.56
0.64
1.04
0.63
0.99
1.31
1.50
1.17
1.53
1.70
Model ma (kg s1 )

Model Taout
( C)

10.9
14.1
9.5
9.8
10.9
9.9
9.9
10.2
11.4
9.7
12.1
12.2
11.4
9.7
13.0
12.6
Model Taout ( C)

Entering ( C)

Leaving ( C)

8.4
8.5
8.4
8.4
8.4
8.4
8.5
8.4
8.4
8.4
8.4
8.4
8.5
8.4
8.4
8.5

11.1
13.1
9.9
9.7
9.8
9.7
12.8
12.6
12.1
13.4
14.5
12
12.9
11.9
13.9
12.7

0.438

3.002

0.226
0.211
0.013
0.0005
0.057
0.110
0.025
0.083
0.132
0.192
0.207
0.160
0.106
0.099
0.119
0.175

2.02
1.93
0.69
0.39
1.05
1.36
0.81
1.21
1.48
1.82
1.90
1.64
1.34
1.30
1.41
1.72

16.05
20.1
13.5
10.1
12.7
12.8
17.8
18.1
16.8
21.1
24
16.6
19.1
16.3
22.1
18.6

Total air side HT (sensible + latent)

Total air side sensible HT

Exp. (kW)

Model (kW)

% ER

Exp. (kW)

Model (kW)

% ER

20.7
37.4
11.4
16.6
24.3
21.8
24.7
30.0
29.1
22.0
33.8
31.2
23.4
17.2
34.1
31.9

20.7
37.3
11.1
16.4
23.9
21.0
24.5
29.7
28.4
21.5
33.6
30.8
23.3
17.3
34.0
31.5

0.0%
0.3%
2.6%
1.2%
1.9%
3.4%
0.8%
1.1%
2.6%
2.1%
0.5%
1.2%
0.4%
0.3%
0.4%
1.3%

20.7
28.2
11.4
16.6
24.3
21.8
14.7
14.9
15.3
12.9
24.3
22.0
14.7
17.2
24.8
21.9

20.7
28.4
11.1
16.4
23.9
21.0
15.3
15.7
15.5
13.3
25.6
22.7
15.3
17.3
25.8
22.7

0.0%
0.7%
2.6%
1.2%
1.9%
3.4%
3.8%
5.4%
1.3%
3.1%
5.3%
3.3%
4.1%
0.3%
4.0%
3.7%

Total air side HT (sensible + latent)

Total air side sensible HT

Exp. (kW)

Model (kW)

% ER

Exp. (kW)

Model (kW)

% ER

14.9
25.7
15.5
6.8
13.5
12.6
23.1
22.6
20.0
27.4
33.5
18.4
24.1
18.5
29.8
22.3

14.9
25.4
15.5
6.8
13.5
12.6
22.8
22.6
20.0
26.9
33.2
18.2
23.8
18.1
29.5
22.1

0.0%
1.2%
0.2%
0.4%
0.2%
0.1%
1.1%
0.2%
0.3%
1.7%
0.9%
1.1%
1.1%
2.2%
0.9%
0.9%

14.9
16.8
15.5
6.8
13.5
12.6
14.1
14.0
11.1
18.8
24.7
10.3
15.0
9.8
21.0
13.6

14.9
16.6
15.5
6.8
13.5
12.6
14.9
14.7
12.2
18.9
23.6
11.3
15.6
10.4
22.2
14.1

0.0%
1.0%
0.2%
0.4%
0.2%
0.1%
5.5%
5.0%
9.9%
0.5%
4.5%
9.7%
3.8%
7.1%
5.5%
4.1%

M. Sarkar / Energy and Buildings 103 (2015) 384395

Model vs. Experimental data for 8 row coil

3.66
23.6
0.00728
1a
4.16
29.3
0.01090
2
1.57
25.3
0.00730
3
4
2.41
25.1
0.00728
3.14
28.0
0.00700
5
6
1.94
34.2
0.00740
7
1.27
36.5
0.01420
8
1.45
34.0
0.01660
2.39
25.9
0.01330
9
10
1.45
30.3
0.01270
2.26
37.2
0.01180
11
3.00
29.1
0.01090
12
3.43
21.2
0.01020
13
2.67
24.2
0.00690
14
3.49
29.4
0.01080
15
3.89
25.6
0.01050
16

(*A)

Reference performance data point.

389

390

M. Sarkar / Energy and Buildings 103 (2015) 384395

Fig. 3. Coolingdehumidication proles at different bypass factors (Twin = 3 C, Tw = 6 C, Tdes = 17 C).

Fig. 4. Coolingdehumidication proles at different inlet DBT (Twin = 5 C, Tw = 8 C, Tdes = 12.8 C, BF = 0.06).

M. Sarkar / Energy and Buildings 103 (2015) 384395

Fig. 5. Coolingdehumidication proles at different inlet humidity ratios (Twin = 5 C, Tw = 8 C, Tdes = 13 C, BF = 0.1).

Fig. 6. Coolingdehumidication proles at different inlet water temperatures (Tw = 8 C, BF = 0.1).

391

392

M. Sarkar / Energy and Buildings 103 (2015) 384395

Fig. 7. Coolingdehumidication proles at different delta Tw (Twin = 3 C, BF = 0.1).

Fig. 8. Coolingdehumidication proles at different supply temperatures (at Twin = 5 C, Tw = 8 C, BF = 0.1).

M. Sarkar / Energy and Buildings 103 (2015) 384395

qsen-dry =

n


[Cpa + i+1 Cpv ][Tai Tai+1 ]

(27.2)

i=1


n

qlatent =

[lo Cplv Txi ][i i+1 ]

(27.3)

i=1

Case 3 For a partially wet coil:

i [1, k], k < n, Txi Tdpin

qsen-dry =

k


(Cpa + in Cpv )[Tai Tai+1 ]

(28.1)

i=1

and,

i [j, n], j k + 1, Txi < Tdpin

qsen-wet =

n


[Cpa + i+1 Cpv ][Tai Tai+1 ]

(28.2)

i=j

qlatent =

n


[lo Cplv Txi ][i i+1 ]

(28.3)

i=j

Hence, the net cooling loads are summarized by:

a qsen-dry
Q sen-dry = m
a qsen-wet
Q sen-dry = m
a qlatent
Q latent = m
For a dry coil:
Q t = Q sen-dry
However, for a partially or fully wet coil:
Q t > Q sen-dry + Q sen-wet + Q latent
The actual refrigeration load is higher than the sum of sensible
and latent loads of a wet coil due to the energy loss by condensate
formation and its subcooling. As per ASHRAE Handbook, the load
due to condensate formation and removal is between 0.5 and 1.5%
of the applied refrigeration [6].

393

heat transfer rate is compared with the corresponding experimental result. In order to get a close match, the bypass factor of reference
performance is revised accordingly and the bypass and mass ow
characteristic constants (X0 and A) for the particular coil type are
determined. Subsequently, with these constants, the air mass ow
rates and bypass factors are determined at varied on-coil velocities
for the coil. For the coil discretization and determination of total and
sensible heat transfer rates, a uniform chilled water temperature
drop of 0.1 C is considered for each coil element.
As per Table 2, greater coherence in results are obtained in case
of dry coil conditions. However, higher level of deviations in some of
the results from the experiment and present model are seen in case
of wet coil conditions. These can be explained by the arguments
summarized below:
The present model is based on an element-by-element method
that theorizes the net coil loads as discrete summations of individual loads of constituent coil elements. However, the sensible
and total heat transfer rates reported for the experimental runs
are based on AHRI procedure that requires inlet and outlet conditions of air [7]. Hence, these differences in results will be higher
for wet coils, since the present model considers the changes in
temperatures and humidity ratios along the coil for calculating
sensible and latent loads.
A combined cooling and dehumidication process creates unpredictable airow patterns due to condensation on the coil and
between ns that are very difcult to model with the simplied
methodology presented here. Due to this, the results obtained
from the theoretical model may deviates from the experimental
predictions in some cases.
The results from the model are based on bypass factor (BF) which
is a consolidated parameter depending on a number of factors.
Hence, the actual experimental conditions and the precision of
measurements may have caused the deviations of test results
from the theoretical predictions.
Nevertheless, the difference between the total and sensible heat
transfer rates from the model and Elmahdys experimental data are
predominantly within a range of 5% (Table 2). Considering the
maximum deviations in results and the uncertainty in measurements [12], the given difference levels in sensible and total heat
transfer rates are acceptable and hence validates the model.

3. Model validation

4. Application range of model

To show the validity of the theoretical model, the results are

compared with actual data available in literature. Extensive experimental data on chilled water coils are available due to many
researchers like Elmahdy and Mitalas [4], McQuiston [25], Threlkeld
[1] and Braun [5]. Elmahdys experiments were mainly directed
towards the performance of 4 row and 8 row coils for a wide range
of operating conditions. The on-coil velocities and chilled water
ow rates were adjusted to obtain both dry and wet coil conditions.
For the comparative study, the experimental heat transfer rates
of 8 row and 4 row coils reported by Elmahdy and Mitalas are compared with the present model. A reference performance point for
each coil type is selected from the given dataset. It is to be noted
that for a coil to be completely dry, there is a maximum limit on its
bypass factor. Hence, for the ease of applicability, a dry coil condition is selected as the reference performance point. Since, Elmahdy
el al. have not explicitly mentioned the bypass factors for the coils
during experimental runs, the maximum value of the bypass factor for the reference performance point is initially assumed. Using
this, the bypass factor for another performance data point (preferably a wet condition) is determined by Eq. (25) and its sensible

The present model performs well for a varied range of psychrometric parameters applicable for typical comfort and commercial
air conditioning utility. As per literature study, bypass factors in
the range of 00.3 are applicable for typical comfort air conditioning and coolingdehumidifying equipment [26]. The BF restricts
the ADP of equipment based on the inlet conditions and the desired
outlet temperature. The inlet condition and the resulting ADP affect
the accuracy of the model. Typically, ADP of a chilled water coil is
above 4 C [21]. Considering a range of 922 C for design supply
temperatures and an inlet humidity range of 0.0070.022 kg/kg DA,
the present model shows minuscule deviations in order of 00.2 C
for supply temperatures in the aforementioned bypass factor band.
However, on increasing the upper limit on the humidity ratio range
to 0.035 kg/kg DA, the error in supply temperature rises up to 0.6 C.
5. Results and discussion
A number of operational and process parameters affect the
performance of chilled water coils. This section discusses the
implication of these factors and the impact they impose on the

394

M. Sarkar / Energy and Buildings 103 (2015) 384395

coolingdehumidication prole of the coil. For the psychrometric representation of coolingdehumidication process, a uniform
chilled water temperature difference of 0.1 C is considered through
each discretized coil element.
5.1. Bypass factor
A bypass factor is a consolidated parameter that denes the performance of a coil and the variation of psychrometric properties of
cooled and dehumidied air. It is directly dependent on the geometric descriptors of a coil (coil dimension, n density, coil rows,
compactness of coil) and interaction (air mass ow rate and on-coil
air velocity) between process uid and heat exchanger during the
process. A higher bypass factor implies that a greater percentage of
air crosses the coil without coming in contact with it. On the other
hand, the cool-dehumidied air obtained from an evaporator with
lower bypass factor will be closer to its saturation point. For a common inlet condition and coil capacity, ADP of a coil decreases with
an increase in its bypass factors. Hence, the exit humidity ratios of
air after coming in contact with a coil of higher bypass factor will
be lower and vice versa. Fig. 3 shows the psychrometric representation of coolingdehumidifying process through three different
coils with varying bypass factors and supplying air at a common
temperature.

5.5. Net chilled water delta T

For a given coil with a constant chilled water temperature at
inlet and consistent exit air temperature, a reduction in net chilled
water temperature difference across the coil is expected with an
increase in chilled water ow rate through it. However, the bulk
uid temperature decreases with a net drop in water delta T across
the coil. Under these conditions, the lowest possible ADP of the coil
also reduces. Fig. 7 shows the variation of air supply conditions at
varying chilled water difference across a coil. As such, for a xed
chilled water supply temperature, an increase in water ow rate
through a coil shows similar trends.
5.6. Supply air temperature
When a lower supply air temperature is intended, a reduced
coil ADP and an increased coil capacity are required. For a coil with
constant bypass factor, air mass ow rate, water inlet temperature
and chilled water delta T, this is attained by increasing the water
ow rate through the coil. As discussed previously, this leads to
higher dehumidication loads and higher coil wetness fractions.
Fig. 8 shows the variation of air temperature and humidity through
an evaporator at a common inlet condition, but at different supply
temperatures.

5.2. Inlet DBT

6. Conclusions

At a common inlet humidity ratio, supply temperature and

bypass factor, the required coil ADP are lower when inlet air DBT are
higher. The reversed scenario is true, when inlet air temperatures
are lower. In this case, the inlet DPT remains the same. However,
the dry sensible load varies directly with the dew point depression
(DPD). Fig. 4 illustrates the effect of inlet DBT on the psychrometric
variation of air. These representations are based on a common airow rate, inlet chilled water temperature and net water delta T, but
at varying water ow rates through the coil to meet the required
cooling loads. In practical applications, modulating valves control
the chilled water ow rates through coil headers to vary or maintain
the supply air temperatures.

A simple steady state model for chilled water cooling coils is

presented in this paper. This model requires bypass factor and
another additional parameter (net chilled water temperature difference or air supply temperature) as principal inputs to depict a
coolingdehumidication prole of the coil. Although, the discretization scheme adopted here to determine coil loads is different
from the conventional methodology, a comparison of coil loads
from model and experimental data in literature shows close compliance, which proves its validity. The computationally simple and
easily applicable nature of the present model makes it a suitable
choice to be integrated into building energy simulation tools.

5.3. Inlet humidity ratio

Fig. 5 shows the representation of coolingdehumidication
processes through a coil for the same inlet and supply temperatures but at different inlet humidity ratios. It also illustrates that
the percentage wetness of coil increases with an increase in inlet
humidity ratios. Here, for a coil with xed parameters and air supply
temperature, the required ADP and its depression from inlet temperature remains the same, even though the inlet DPT increases.
This is contrary to the effect on DPD and dry sensible loads, which
decrease with an increase in inlet humidity ratios. When the difference between inlet temperature and ADP is lower than the DPD,
the coil will be completely dry. However, when it is greater than
the DPD, the coil will be partially or fully wet.
5.4. Inlet chilled water temperature
As per the present methodology, the lowest possible supply air
temperature is restricted by the bulk chilled water temperature
through the coil. It implies that for a particular coil with constant
chilled water delta T across it, lower water inlet temperature results
in lower possible supply air temperatures and ADP. In addition, a
lower water inlet temperature results in greater percentage of coil
to be wet. Fig. 6 shows the effect of inlet water temperatures on the
lowest possible supply air temperatures for a given coil.

Acknowledgement
The author would like to thank Dr. Mandar Tendolkar, VJTI
Mumbai, for reading the manuscript and offering valuable suggestions.
References
[1] J.L. Threlkeld, Thermal Environmental Engineering, 1st ed., Prentice-Hall,
Englewood Cliffs, NJ, 1962.
[2] A.H. Elmahdy, Analytical and Experimental Multi-Row Finned-Tube Heat
Exchanger Performance During Cooling and Dehumidifying Processes, Ph.D.
thesis, Carleton University, Ottawa, Canada, 1975.
[3] A.H. Elmahdy, G.P. Mitalas, A simple model for cooling and dehumidifying
coils for use in calculating energy requirements for buildings, ASHRAE Trans.
83 (2) (1977) 103117.
[4] A.H. Elmahdy, G.P. Mitalas, Facility for Testing Cooling and Dehumidifying
Coils: Description, Procedures and Test Results, Paper No. 756, Division of
Building Research, National Research Council of Canada, Ottawa, Canada,
1978.
[5] J.E. Braun, Methodologies for the Design and Control of Central Cooling Plants,
Ph.D. thesis, University of WisconsinMadison, USA, 1988.
[6] ASHRAE Handbook HVAC Systems and Equipment, Chapter 23 Air Cooling
and Dehumidifying Coils, American Society of Heating, Refrigeration and Air
Conditioning Engineers, Inc., Georgia, USA, 2012.
[7] AHRI Standard 410-2001, Forced-Circulation Air-Cooling and Air-Heating
Coils, Air-Conditioning, Heating and Refrigeration Institute, Virginia, USA,
2001.
[8] J. Wang, E. Hihara, Prediction of air coil performance under partially wet and
totally wet cooling conditions using equivalent dry-bulb temperature
method, Int. J. Refrig. 26 (2003) 293301.

M. Sarkar / Energy and Buildings 103 (2015) 384395

[9] W. Pirompugd, S. Wongwises, C. Wang, Simultaneous heat and mass transfer
characteristics for wavy n-and-tube heat exchangers under dehumidifying
conditions, Int. J. Heat Mass Transfer 49 (1/2) (2006) 132143.
[10] L. Xia, M.Y. Chan, S.M. Deng, X.G. Xu, A modied logarithmic mean enthalpy
difference (LMED) method for evaluating the total heat transfer rate of a wet
cooling coil under both unit and non-unit Lewis Factors, Int. J. Therm. Sci. 48
(2009) 21592164.
[11] D.R. Mirth, S. Ramadhyani, Prediction of cooling-coil performance under
condensing conditions, Int. J. Heat Fluid Flow 14 (4) (1993) 391400.
[12] A. Vardhan, P.L. Dhar, A new procedure for performance prediction of air
conditioning coils, Int. J. Refrig. 21 (1) (1998) 7783.
[13] M.K. Mansour, M. Hassab, in: J. Mitrovic (Ed.), Thermal Design of Cooling and
Dehumidifying Coils, Heat Exchangers Basics Design Applications, In Tech,
2012.
[14] O. Morisot, D. Marchio, P. Stabat, Simplied model for the operation of chilled
water cooling coils: under non nominal conditions, Int. J. HVAC R Res. 8 (2)
(2002) 135158.
[15] V. Lemort, J. Lebrun, C. Cuevas, Development of simple cooling coil models for
simulation of HVAC systems, ASHRAE Trans. 114 (1) (2008) 319328.
[16] R.J. Chillar, R.J. Liesen, Improvement of the ASHRAE secondary HVAC Toolkit
SIMPLE cooling coil model for simulation, SimBuild 2004, in: IBPSA-USA
National Conference Boulder, Colorado, USA, August 46, 2004.
[17] X. Yiu, J. Wen, T.F. Smith, A model for the dynamic response of a cooling coil,
Energy Build. 37 (2005) 12781289.

395

[18] Y. Yao, Z. Lian, Z. Hou, Thermal analysis of cooling coils based on a dynamic
model, Appl. Therm. Eng. 24 (2004) 10371050.
[19] S.C. Sekhar, L.T. Lan, Optimization of cooling coil performance during
operational stages for improved humidity control, Energy Build 41 (2009)
229233.
[20] S.T. Taylor, Degrading chiller water plant delta-T: causes and mitigation,
ASHRAE Trans. 108 (1) (2002).
[21] M. Sarkar, Theoretical comparison of cooling loads of an air handling unit in
blow-through and draw-through congurations, Energy Build. 64 (2013)
239248.
[22] P. Napon, S. Wongwises, A study of the heat transfer characteristics of a
compact spiral coil heat exchanger under wet-surface conditions, Exp. Therm.
Fluid Sci. 29 (2005) 511521.
[23] R. Raustad, A variable refrigerant ow heat pump computer model in energy
plus, ASHRAE Trans. 119 (1) (2013).
[24] M. Sarkar, A new theoretical formulation of dew point temperatures
applicable for comfort air-cooling systems, Energy Build. 86 (2015)
243256.
[25] F.C. McQuiston, Heat mass and momentum transfer data for ve
plate-n-tube heat transfer surfaces, ASHRAE Trans. 84 (1) (1978)
266293.
[26] Chapter 8 applied psychrometrics, in: Part 1 Load Estimating, Carrier
System Design Handbook, Carrier Corporations, Syracuse, New York, 1960,
pp. 127.