C97-G31 1

Fundamentals of Moisture Transport in Textiles

Haskell W. Beckham, Johannes Leisen, H. Stephen Lee, Wallace W. Carr

School of Textile and Fiber Engineering
Georgia Institute of Technology, Atlanta, GA 30332-0295

Steven B. Warner
University of Massachusetts, Dartmouth, MA

PROJECT GOALS: To understand and quantify moisture transport in textile structures so that
novel/improved materials for fluid management and novel/improved drying
processes may be developed.

ABSTRACT
Magnetic resonance imaging (MRI) and other techniques are used to investigate moisture
transport in textiles. Industrially relevant through-flow drying of carpet has been simulated on a
laboratory through-flow dryer (LTFD) as well as inside a MRI scanner. A transient two-dimensional
mathematical model is developed to describe the through-flow drying process for tufted carpet. The
heat and mass transfer in a single carpet yarn and the air flowing around it are analyzed separately.
First, thermal and mass circuits are used to analyze the simultaneous heat and mass transfer within the
yarn. Then, the equations of the conservation of mass and energy are written for the airflow. The
resulting system of three non-linear differential equations is numerically solved by an implicit finite
difference method. The numerical solutions are compared with experimental drying results obtained
using MRI and LTFD.

INTRODUCTION
Innovative and improved manufacturing processes can only be developed if the interactions
between process chemicals and textile products are understood on a fundamental level. For example,
current drying processes can be made much more efficient, migration problems can be controlled, and
completely new designs for drying equipment can be proposed. In addition, new materials for fluid
management can only be developed by understanding fundamental interactions between fluids and
fibrous assemblies.
While optical imaging methods are well-suited for the analysis of surface and transparent
substrates, magnetic resonance imaging (MRI) readily provides 3D water distributions within opaque
textile substrates. In this study, standard MRI techniques are being employed to measure one-, two-

National Textile Center Annual Report: November 2000

C97-G31
1
 C97-G31 2

and three-dimensional images of water within wet textile substrates. Because a single one-dimensional
image can be measured in a matter of milliseconds, real-time monitoring of actual drying processes is
possible. Mechanisms of drying are established by following water distributions as a function of drying
time.
Based on MRI results, a transient two-dimensional mathematical model, which considered
primary heat, moisture and momentum transfer mechanisms in the moist porous material (yarn) and
drying medium (airflow), was developed for through-flow drying of tufted carpet. The drying
model consisted of two sets of coupled differential equations with relevant initial and boundary
conditions. The model contained a number of unknown physical and transport properties. With
the number of unknown physical and transport properties, the model is not very useful for
industrial applications. Thus, a simpler model requiring fewer physical and transport properties,
but capable of accurately predicting temporal one-dimensional moisture profiles, and thus overall
moisture regain and drying time, was developed. The simplified drying model is presented here.

THEORETICAL DRYING MODEL

The porous medium studied was unbacked cut-pile nylon carpet. Unbacked tufted carpet
consists primarily of short lengths of yarn, typically 5 to 20 mm long. These yarns, referred to as face
yarn, are held in position by a very thin woven fabric, called primary backing. The face yarns are
packed nearly parallel to each other and run perpendicular to the primary backing. Since the primary
backing is typically made of polypropylene and holds little moisture, it is neglected in our model. Since
the length of the yarn on the backside is small compared to the length of the face yarn, we simplify the
geometry by adding the length of the yarn on the backside to the length of face yarn. The unit cell (see
Figure 1) is a yarn and the surrounding space through which air flows parallel to the yarn. The yarn is
considered to be a porous circular cylinder.
In Figure 2, two-dimensional magnetic resonance (MR) images of moisture in a slice of a single
carpet yarn are shown as a function of drying time. The imaging plane is perpendicular to the yarn axis.
The images clearly reveal the moisture front between the wet and sorption regions recedes from the
outer region of the reference circle to the inner region during drying. Thus, moisture transport varies in
the radial direction. Coupled with the varying airflow temperature and humidity ratio in the airflow
direction (i.e., parallel to the yarn axis), this indicates a transient two-dimensional drying model is
needed to describe mass and heat transfer during the through-flow drying process. The model has two
parts: the first is a model for heat and mass transfer in the yarn, and the second is a model for heat and
mass transfer in the airflow. It is assumed that the airflow is along the surface of the yarn, but does not
penetrate into the yarn.

National Textile Center Annual Report: November 2000

C97-G31
2
 C97-G31 3

Figure 1. Schematic of a yarn and its immediate surrounding space inside the unit cell.

0 – 15 sec 120 – 135 sec 240 – 255 sec 360 – 375 sec

Figure 2. Two-dimensional MR images of moisture in a single carpet yarn as a function of

drying time. The imaging plane is perpendicular to the yarn axis. Lighter areas represent higher
moisture content. The moisture front recedes radially inward during drying.

Model for Heat and Mass Transfer in the Yarn

The yarn model is developed by considering mass and heat transfer within the wet and
sorption regions of the differential control volume in a given slice of yarn. Heat and moisture
transport are assumed to vary only in the radial direction. First, consider moisture and momentum
transport within the wet region. Capillary flow dominates the unbound moisture flow in this
region. It is assumed that capillary flow maintains a uniform moisture distribution in the wet region
for moisture regain greater than the critical moisture regain. When critical moisture regain is
reached in a given slice of yarn, the evaporation front begins to recede from the yarn surface, and
the moisture regain in the wet region is maintained at the critical value. In the sorption region,

National Textile Center Annual Report: November 2000

C97-G31
3
 C97-G31 4

moisture regain is assumed to be zero. Thus, the region is referred to as the dry region, and bound
moisture transport within this region is neglected. The rate of change in mass of stored water vapor
is ignored in the dry region. Evaporating water at the front diffuses through the drying region to
the yarn surface. Then, the water vapor at the surface is convectively transferred into the airflow.

Dry Region

Ta ∞

ρ vs

ρ va∞

Figure 3. Thermal and mass circuits for the yarn and airflow in section A-A (see Figure 1).

Consider the heat transfer from the air stream to the yarn. The heat transfer rate from the air
stream to the yarn surface, qtot, is equal to rate of the energy stored in the dry region, qsorp, plus rate
of the energy transferred to the wet region, qwet. In the dry region, transient, one-dimensional (r
direction) heat conduction is assumed. The energy stored in the dry region is treated as though it is
all located at the mean radius, rM, which divides the dry area in half; i.e., half of the dry area is from
rf to rM, and half of the dry area is from rM to R (see Figure 3). The energy transferred to the wet

National Textile Center Annual Report: November 2000

C97-G31
4
 C97-G31 5

region by heat conduction evaporates water and/or increases stored energy. The temperature
distribution is assumed to be uniform in the wet region.
In this study, the thermal and mass transfer circuits, shown in Figure 3, are used to write the
energy and mass balance equations in the model. The resistance to mass transfer from the air
stream to the yarn surface is Rm1. The resistance to mass transfer from the yarn surface to the
evaporation front is Rm2. The resistance to heat transfer from the air stream to the yarn surface is
Rt1. The thermal resistance through the dry region is divided into two parts based on a mean
radius, rM, as mentioned above. The mean radius, rM , may be written as

R 2 + r f2
rM =
2

The resistance to heat transfer from the yarn surface to the mean position of the dry region is Rt2.
The resistance to heat transfer from the mean position of the dry region to the surface of the wet
region is Rt3.
Mass transfer from the evaporation front to the air steam can be modeled as a series circuit
as shown in Figure 3. Thus, the evaporation rate, m& , can be written as

[ ρ (r = r , z, t ) − ρ ] va∞
m& =
vs f

Rm tot

where ρvs and ρva∞ are the densities of the saturated water vapor at the evaporation front and the
water vapor in the airflow, respectively, and the total mass resistance, Rm,tot, is

1 ln R r f ( )
Rmtot = Rm1 + Rm 2 = +
hm 2π Rdz 2π Deff dz

where Deff is the effective diffusivity and hm is the convective mass transfer coefficient.
The total energy transferred to the yarn, qtot, is

 r f2   ∂T  r f2  ∂T
q tot = m&h fg +  2 ms c ps + mw c pw  wet + 1 − 2 ms c ps M
 R   ∂t  R  ∂t

where hfg is the latent heat of vaporization of water, cps is the specific heat of the fiber, ms is the
mass of fibers, mw is the mass of liquid water, and rf is the position of the evaporation front. The
mean temperature of the dry region, TM, is
ln rM r f ( )
TM = Twet + q
2π k eff dz wet

National Textile Center Annual Report: November 2000

C97-G31
5
 C97-G31 6

where k eff is the effective thermal conductivity.

Additionally, the total energy transferred to the yarn may be expressed as

Ta ∞ − TM
q tot =
Rt 1 + Rt 2
where Rt1 and Rt2 are defined as
1 ln (R rM )
Rt1 = ; Rt 2 =
2π RhT dz 2π k eff dz
After mathematical substitution, the equation of energy conservation may be expressed as

 Pvs YPatm 
 −  h fg
Ta ∞ − TM  R Twet ( 0.622 + Y ) R Ta ∞ 
=
Rt 1 + Rt 2 Rm tot (1)
 r f2   ∂T  r f2  ∂T
  
+  2  ms c ps + m w c pw  wet
+ 1 − 2 ms c ps M

 R   ∂t  R  ∂t

where Y is the humidity ratio and Ta ∞ is the air temperature.

Model for Heat and Mass Transfer in the Airflow

Consider conservation of heat and mass transfer in the differential control volume of the airflow.
Heat and mass balance equations may be written as

 ∂Y ∂Y 
 ρa ∞ + ja∞  dz = m& (2)
 ∂t ∂z 

(c pa
)
+ Yc pv  ρa ∞

∂ Ta ∞
∂t
+ j a∞
∂ Ta ∞ 
∂z 
 dz = q tot (3)

where j a∞ is mass flux in the airflow, and cpa and cpv are the specific heat capacity of air and water
vapor, respectively.
Initial and boundary conditions for each dependent variable (Y and Ta ∞ ) are required to
solve the set of the governing equations. Initial and boundary conditions for humidity ratio and air
temperature are assumed to be constant and are

Y ( z, t = 0) = Y0 ; Ta ∞ ( z , t = 0) = Ta ∞ ,room

Y ( z = 0, t ) = Y0 ; Ta ∞ ( z = 0, t ) = Ta ∞ 0

National Textile Center Annual Report: November 2000

C97-G31
6
 C97-G31 7

RESULTS AND DISCUSSIONS

The implicit finite difference method, which is known to yield stable solutions for any
selection of the space and time nodal points, is used to solve the yarn and airflow models. The
detailed numerical algorithm in solving the drying model is described in our publication. Numerical
solutions for several cases were obtained using the numerical method described above. Local and
overall moisture regains were predicted as a function of drying time for two types of carpet, two airflow
rates, two air temperatures, two airflow directions and three initial moisture profiles. The initial moisture
profiles measured by MRI were used as initial conditions for the numerical simulations.

Comparison of Numerical Results with MRI Results

The predicted local moisture regains versus time agreed very well with the experimental data
for all cases simulated. Representative examples are shown in Figure 4 where the predicted and
experimental results are plotted for carpet samples A and B. For these tests, the airflow, which
enters the carpet face, is at an air temperature of 80°C and an airflow rate of 20 m/min. The initial
overall moisture regain for the two carpets is 50%, but the initial moisture distributions are
different. As illustrated in Figure 4, the predicted moisture profiles agree well with experimental
data.

(a) (b)
140 80
Initial Moisture Profile at t = 0 t = 29.5 sec
t = 55.5 sec
70 t = 63.2 sec
Local Moisture Regain (%)

Local Moisture Regain (%)

120 t = 113.9 sec

t = 94.0 sec
t = 158.4 sec 60
100 t = 116.6 sec
t = 188.2 sec
50
80
Initial Moisture Profile at t = 0 40
60
30
40
20

20 10

0 0
0 5 10 15 20 0 2 4 6 8 10 12

Distance in the Pile Yarn Direction (mm) Distance in the Pile Yarn Direction (mm)

Figure 4. Comparison of MRI experimental and predicted moisture distribution versus time for
(a) carpet sample A with an areal density of 2204 kg/m2, a tufting density of 10.7 stitch/cm2, and a
carpet thickness of 19.0 mm, and (b) carpet sample B with an areal density of 1492 kg/m2, a
tufting density of 11.1 stitch/cm2, and a carpet thickness of 16.6 mm, at an airflow rate of 20
m/min and an air temperature of 80°C. Numerical fits to the data are shown as lines.

National Textile Center Annual Report: November 2000

C97-G31
7
 C97-G31 8

Comparison of Numerical Results with LTFD Results

The simplified mathematical model was verified by the MRI drying results, however, the air
temperature used in the MRI system (80°C) is less than that used in the first few zones of industrial
dryers (typically > 110°C). Thus, the LTFD drying results were used to verify the model at industrial
drying conditions.
Since the LTFD cannot measure the local moisture distribution, the verification focused on
predicting the variations of overall moisture regain (average for entire sample) with drying time. The
initial moisture distribution was assumed to be uniform because initial moisture distribution does not
greatly affect the overall drying results. For the numerical simulations, the same carpet construction
parameters used for drying at 80°C in the MRI scanner were used for simulating the LTFD tests
performed at an air temperature of 116°C. For drying carpet samples A and B at airflow rates of 20
and 30 m/min, the predicted and experimental variations of overall moisture regain are plotted in Figure
5. As shown in these figures, the predicted drying curves fit the experimental data quite well.

50 o
LTAD exp. (A; 20 m/min; 116 C)
Overall Moisture Regain (%)

o
LTAD exp. (A; 30 m/min; 116 C)
o
40 LTAD exp. (B; 20 m/min; 116 C)
o
LTAD exp. (B; 30 m/min; 116 C)
Predicted Data
30

20

10

0
0 20 40 60 80 100 120 140
Time (sec)

Figure 5. Comparison of LTFD experimental and predicted variations of overall moisture regain with
time for carpet samples A and B at airflow rates of 20 and 30 m/min, and an air temperature of 116°C.

Several drying tests were conducted at different airflow rates, air temperatures, airflow
directions and initial moisture profiles and compared with the numerical predictions. Close agreements
with the experimental MRI and LTFD results were obtained. Thus, the simplified mathematical model
can accurately predict drying rate and drying time for through-flow drying of tufted carpet.
CONCLUSIONS
A transient two-dimensional mathematical model has been developed to simulate the
through-air drying process for tufted textile materials. Even after simplifying the model to facilitate
its usability, it provides predictions that closely agree with experimental results. Predicted
variations of one-dimensional moisture distributions within carpet agree well with profiles obtained

National Textile Center Annual Report: November 2000

C97-G31
8
 C97-G31 9

using MRI. Predicted temporal variations of overall moisture regain closely fit experimental results
obtained with both MRI and the LTFD.

PUBLICATIONS
1. Lee HS, Carr WW, Beckham HW, and Leisen, J, "A Model of Through-Air Drying of Tufted
Textile Materials", submitted to International Journal of Heat and Mass Transfer, June 2000.
2. Lee HS, Carr WW, Leisen J, Beckham HW, "Through-Air Drying of Unbacked Tufted Carpet",
Textile Research Journal, accepted for publication, September 2000.
3. Lee HS, Carr WW, Beckham HW, Wepfer WJ, "Factors Influencing the Air Flow Through
Unbacked Tufted Carpet", Textile Research Journal, October 2000, in press.
4. Lee HS, Study of Industrial Through-Air Drying Process for Tufted Carpet, Ph.D. Dissertation,
Georgia Institute of Technology, May 2000.
5. Leisen J, Beckham HW, Good J, Warner S, Carr WW, "Magnetic Resonance Imaging of Water
Ingress and Distribution in Fluorochemical-Finished Polyester Cut-Pile Carpet", Textile Chemist
and Colorist 1999, 31(4), 21-26.
6. Leisen J, Hou L, Beckham HW, Carr WW, "Observation of the Water Distribution during Drying
of Textiles", In Spatially Resolved Magnetic Resonance: Methods, Materials, Medicine, Biology,
Rheology, Geology, Ecology, Hardware, P. Bluemler, B. Bluemich, R. Botto, E. Fukushima, Eds.,
Viley-VCH 1998.
7. Beckham HW, Fülber C, Spiess HW, Blümich B, "Nuclear Magnetic Resonance Imaging of Water
Distributions in Loop-Pile Nylon Carpet Tile", J. Text. Inst., Pt 1 1998, 89(2); 436.
8. Zhou W, Warner S, “Pore Sizes in Carpet", Textile Research Journal 1998, in press.
9. Beckham HW, Leisen J, Carr WW, "Magnetic Resonance Imaging of Water Distribution in Carpet"
AATCC Book of Papers, 1997 International Conference & Exhibition.

PRESENTATIONS
10. Leisen, J, Beckham HW, "Magnetic Resonance Imaging of Fluids in Nonwovens", 9th annual
TANDEC NONWOVENS Conference; Knoxville/ TN; 10-12 November 1999.
11. Leisen J, Beckham HW, Hojjatie B, Doffin D, "Quantitative Application of Magnetic Resonance
Microscopy for the Characterization of Moisture Distributions in Fibrous Substrates", 5th
International Conference on Magnetic Resonance Microscopy; Heidelberg, Germany; 5-9
September 1999.
12. Leisen J, Kinser H, Beckham HW, "Fluid Distribution and Diffusion in Engineered Fibrous
Substrates via Diffusion NMR and NMR Imaging", 218th ACS national meeting; New Orleans, LA;
21-26 August 1999.

National Textile Center Annual Report: November 2000

C97-G31
9
 C97-G31 10

13. Lee HS, Carr WW, Leisen, J, Beckham HW, "Observation and Simulation of Heat and Moisture
Transport during Through-Air Drying of Cut-Pile Tufted Carpet", the 12th Annual Polymer
Education & Research Review Meeting; Atlanta, GA; June 1999.
14. Lee HS, Carr WW, Leisen, J, Beckham HW, "Observation and Simulation of the Industrial
Through-Air Drying Process for Cut-Pile Tufted Carpet", the 4th Annual Conference on Recycling
of Fibrous Textile and Carpet Waste; Dalton, GA; May 1999.
15. Beckham HW, Leisen J, Carr WW, "Magnetic Resonance Imaging of Moisture in Textiles",
Nonwovens Conference, Proceedings of TAPPI; Orlando, FL; 20-24 April 1999.
16. Beckham HW, "Magnetic Resonance Imaging of Fluids in Textiles", Fachhochschule Niederrhein;
Krefeld, Germany; 18 June 1998.
17. Leisen J, Beckham HW, Carr WW, "Observation of Drying Processes in Textiles by Magnetic
Resonance Microscopy", 39th Experimental NMR Conference; Pacific Grove, CA; 22–27 March
1998.
18. Beckham HW, "Magnetic Resonance Imaging of Water Distribution in Carpet: Drying Studies",
AATCC International Conference & Exhibition; Atlanta, GA; 29 September 1997.
19. Leisen J, Beckham HW, Carr WW, "Observation of Water Distribution and Diffusion during the
Drying Process in Textiles", 4th International Conference on Magnetic Resonance Microscopy and
Macroscopy; Albuquerque, NM; 21–25 September 1997.
20. Beckham HW, "Magnetic Resonance Imaging of Fluids in Engineered Fibrous Substrates",
DuPont; Chattanooga, TN; 19 September 1997.

National Textile Center Annual Report: November 2000

C97-G31
10