ORIGINAL PAPER/PEER-REVIEWED

Cutting Costs Prior to

‘Cutting Metal’ in the
Nonwovens Industry
By A. Mukhopadhyay, and F. J. Kelecy, Fluent Incorporated, Lebanon, NH

Abstract certain point, consumer expectation calls for a design change.

As the North American nonwovens market has matured A successful business model recognizes this trend and makes
over the last ten years, computer-aided design, engineering the necessary decisions. One other important issue to be con-
and flow analysis have established themselves as effective sidered is competition: it puts pressure on every single step in
engineering tools to complement and optimize traditional the business cycle. Thus, in today’s business environment, the
design, analysis and testing. This has lead to significant cost market life of an NW product can turn out even smaller than
and time savings in the product life cycle. More specifically, originally conceived. Competition is one pivotal metric that
these tools mainly impact pre-market activities, thus they calls for changes in process and product design more often
reduce time-to-market and overall development expense. than performance issues.
This paper demonstrates the successful use of computa- Therefore, designers and analysts are being asked to make
tional fluid dynamics (CFD) tools to achieve cost-effective design changes more often than before while reducing the
design analysis and optimization. Examples of airflow deliv- manpower and calendar time required. There is also less time
ery systems (fan/blower design and jets/diffusers in spun available for decision-making. Pilot plant trials and other
bond process) and analysis of filter performance are shown physical tests are the traditional means of design validation,
but the time and resources these entail now make them less
Introduction attractive. In the 1990s, Computer Aided Design (CAD) tools
The nonwovens (NW) industry in North America is a were introduced to solve part of the problem – effective
mature market with respect to traditional products and man- design data transfer to manufacturing. In some application
ufacturing techniques. Now, continuous product innovation areas, CAD was used for further integration between struc-
and reducing costs have assumed greater importance. tural analysis and manufacturing. But, it took some time for
Shortening time-to-market has a big impact on recovering businesses to harness the true potential of such an integrated
development costs and brings a competitive advantage. approach to product and process development: concept, pro-
Two distinct factors are driving innovations in the North totype, pilot to full-scale production. Early on, adverse com-
American NW market today: frequent product design ments by traditional design practitioners indeed helped
changes and new consumer products. Technical innovations streamline and lay the road map for the deployment and uti-
in both of these areas are significantly tied to the time-to-mar- lization of Computer Aided Engineering (CAE) software as
ket as well as optimal pricing. Since innovation is driven by well as model building methodologies [1,2]. CAD and FEA
healthy competition, trade secrecy is getting more attention, tools have since become an indispensable part of all engineer-
too. The question is: how can we match several competing ing design practices. The concept of using high tech tools for
metrics like cost-effectiveness, profitability, time-to-market, routine engineering process and product design has proven
exclusivity of ideas and market capture? The answer lies with- invaluable [3].
in the process of implementing a business idea. Figure 1 shows Over the years, the NW industry has followed suit by
the typical business cycle. Note that a product design is only adopting CAD and CAE tools. Since polymer rheology and
a subset of the business plan. Product release is a key phase in airflow characteristics are so vitally important, the NW indus-
the life cycle because revenue earning begins at this point. try has also begun to realize the value of CFD analysis. A
Earnings pick up as market capitalization occurs, but after a wide variety of other fluid flow phenomena also have an

9 IINJ Spring 2005

 Figure 1
PRODUCT LIFE CYCLE

impact on the manufacturing process and product perfor- than the numerical techniques
mance. Table 1 shows a summary of flow related aspects in the • Extension of CFD capabilities to go far beyond hydrody-
NW industry. Using traditional experiments and testing namics:
In this paper, specific examples
are given to illustrate the cost- TABLE 1
effective deployment of CFD tools Nonwovens fabric making processes and sub-processes where CFD has been success-
to reduce the overall design time fully used for cost-effective design analysis and process optimization methods to quan-
and production costs. In the next tify and improve these processes is either infeasible or extraordinarily expensive for a
section, a general introduction to number of reasons: the measurements themselves are so intrusive that they distort the
CFD will be given. In the third sec- flows being measured; the opacity of the flowing fluid may hinder visualization; the
tion, two specific examples are pre- process transients (or flow instabilities) may be too fast to practically measure. In
sented to demonstrate the utility of response to these difficulties, polymer-oriented industries in general (and the NW
CFD. industry in particular) have adopted high tech tools like Computational Fluid
Dynamics (CFD) along with CAD and structural analysis to reduce the time to market
for new products, and to aid in troubleshooting and scale up of process lines [4-6].
Introduction to CFD
Notably, CFD analysis precedes any pilot testing or experiments involving so called
Computational Fluid Dynamics
"metal cutting" tasks.
(CFD) is an engineering tool used
for conducting virtual experiments
with significant time and cost sav-
ings. It complements and reduces NW Process Sub-processes
physical testing. CFD models fluid Upstream polymer Mixing, pelletizing, polymerization, extruders, gear pumps,
flow and heat transfer problems by processing melt flow distribution
solving the equations governing Spunbond Spin packs, bi-co spinnerets, quench box, diffusers,
conservation of mass, momentum, attenuators, aspirators, fiber laydown, thermal bonding
and energy, given a set of material Meltblown Dies, air delivery systems, Coanda devices, blower/
properties and operating condi- compressor designs, fiber laydown, suction box,
tions. Recent trends have con- thermal bonding
tributed to the rapid growth and Finishing Thermal bonding, air knives, carding, fluff making,
widespread adoption of CFD: fiber shred transport, hydroentanglement
• Easy-to-use Graphical user Product forming Diaper making, web handling, folding, vacuum assisted
interfaces (GUIs) transport, composite lines (SMS…), coating, ink jet printing,
• Dramatic increase in comput- gluing, drying
ing power and substantially lower End Use Moisture and liquid transport, capillarity, wicking, diffusion,
cost! sorption, saturation analysis, breathability,
• Shift of focus to understanding filtration, hydrophilicity
the engineering processes rather

10 INJ Spring 2005

 * Free surface flows, non-Newtonian fluids, viscoelas-
ticity, solidification, crystallization, polymerization,
microstructure, and orientation.
CFD has the additional advantage of being non-intrusive: it
can therefore offer quantitative insight into flows that cannot
be experimentally measured. It allows quick and easy para-
metric analyses leading to faster return on investment. Typical
applications of CFD include product design, process opti-
mization, scale-up and troubleshooting. A large number of
CFD application examples were presented during the last
year’s INTC conference [7].

Air Delivery Systems

Shuler and Johnston reported an example of cost-effective
CFD analysis leading to a strategically modified design of
patented attenuator equipment [8]. The improved design not
only required less pressurization but also resulted in more
homogeneous product formation.
Airflow is the primary forcing mechanism in fiber forming
and attenuation in the NW industry. Various processes use Figure 2
different air delivery systems, most of which are designed
THE FAN GEOMETRY
using patented technologies from the past. In recent years,
many of these designs have been revisited to improve process
and product consistency, and to reduce the air handling sys- mine the accuracy that can be expected for this type of flow
tem size and cost. A typical example is the pressure differen- An additional objective of this study is to examine the valid-
tial used in a meltblown die system. Not long ago, heavy com- ity of modeling the entire fan (that is, the inlet, rotor, and cas-
pressors were used to deliver pressures on the order of 50 to ing) using a steady state, multiple reference frame (MRF)
100 psig to the die. More recently, using pressures as little as approach. In reality, the flow exiting the blower wheel is
20 to 40 psig is found to be very effective – thus, less expensive unsteady due to interaction with the casing (especially near
blowers have largely replaced the compressors. Concurrently the cutoff). However, if the unsteady interactions are relative-
the air delivery ducts, diffusers and nozzles were analyzed ly minor compared to the mean flow, a steady state solution
with CFD to look for potential room for improvements. can be a useful approximation and saves calculation time.
Analyses on the original designs revealed problems with The fan was tested in a laboratory (in accordance with
choking the delivery nozzles or stalling the diffusers. As a ANSI/AMCA 210-85, ANSI/SHRAE 51-1985 procedures) by
result of these studies appropriate design parameters and mounting the outlet of the fan to the inlet of a wind tunnel. Air
their limits were identified leading to more efficient and eco- from the ambient entered the fan through the inlet orifice from
nomic hydrodynamic designs [9, 10]. In the next section, the all directions. The pressure rise and flow rate were measured
usefulness of CFD for fan design and performance analysis is in the wind tunnel using conventional techniques (static pres-
shown. sure taps and flow nozzles). Performance data for the fan,
which included pressure rise, shaft power, and sound pres-
CFD for Fan Design Analysis sure level, were collected for a nominal operating speed and a
Centrifugal blowers are often used in the NW industry to range of flow rates. Due to minor variations in speed and air
deliver a relatively large pressure rise for a given flow rate. temperature, all data were corrected to the nominal rotational
Advanced CFD tools let fan engineers employ computer sim- speed and standard atmospheric density (0.075 lbm/ft3). The
ulations to address important performance issues, such as centrifugal fan under consideration is illustrated in Figure 2. It
separated flow, tip clearance effects, and noise. By using CFD is a backward-inclined centrifugal fan of conventional rotor
as a “virtual wind tunnel,” design changes can be tested design consisting of a top rim (shroud), 15 blades, and back
before any hardware is built. This reduces cost while provid- plate. The rotor is installed in a scroll-type casing, which col-
ing the opportunity to develop designs that are more efficient lects the flow from the rotor and discharges it through a rec-
and robust. tangular outlet. An inlet orifice is also employed to help direct
In order for fan engineers to have confidence in the CFD the f low into the rotor with minimal loss. All flow field cal-
predictions, it is essential to validate results using well-docu- culations were performed using the FLUENT CFD software.
mented test cases. The objectives of the present study are as As mentioned previously, one of the goals of this study was to
follows: determine the efficacy of the steady-state MRF approach in
• Calculate the performance characteristics of the fan over a simulating centrifugal fans. Boundary conditions for the
range of flow rates model were prescribed as follows:
• Compare the CFD results with experimental data to deter- • Inlet: Uniform total pressure (0 gauge), turbulence inten-

11 INJ Spring 2005

 Figure 3 Figure 5
PRESSURE-RISE VS. FLOW COEFFICIENT EFFICIENCY OF FAN PERFORMANCE

Figure 4 Figure 6
POWER VS. FLOW COEFFICIENT CONTOURS OF STATIC PRESSURE
(INCHES OF WATER)
sity of 1%, viscosity ratio (turbulent/laminar) of 1.
• Outlet: Uniform velocity corresponding to a desired vol- • Efficiency: η = ΨΦ/Λ
ume flowrate (velocity equals volume flowrate/outlet area), In the above, Q is the volume flow rate, N is the rotational
with a direction pointing out of the domain, normal to the speed, D is the fan diameter, ρ is the density, ∆pt is the total
outlet surface. pressure rise, and P is the shaft power. For the CFD solutions,
• Inlet wall, casing: no slip in absolute frame the shaft power is computed as the product of the rotor torque
• Rotor walls: no slip in relative frame and the rotational speed. Also, static pressures were area aver-
Solutions were obtained for a range of flow rates so that fan aged at the inlet and outlet surfaces, and the total pressure
performance data could be generated and compared with the derived by adding the dynamic pressure based on the given
available test data. The following non-dimensional parame- flow rate to the static pressure.
ters were used to characterize the fan performance: A comparison of the predicted pressure rise coefficient,
• Flow Coefficient: Φ = Q/ND3 power coefficient, and efficiency versus flow rate with the
• Pressure Rise Coefficient: Ψ = ∆pt/ρN2D2 data is presented in Figures. 3 - 5. These results show excellent
• Power Coefficient: Λ= P/ρN3D5 agreement over the entire flowrate range. Figure 6 illustrates

12 INJ Spring 2005

 the pressure distribution on the rotor and fan casing.
These comparisons, along with the ability to graphically
visualize the flow around the fan, clearly demonstrate the
ability of CFD analysis to serve as a surrogate for experimen-
tal measurements. The added advantage with CFD is the abil-
ity to probe any detail at any location and at any time – CFD
results can be stored for detailed review at any later time. The
Figure 7 CFD model can be developed much quicker than building
MACH NUMBER CONTOURS IN PRIMARY and testing a physical prototype. Furthermore, most of the
effort required to obtain the CFD result is typically expended
AIR SYSTEM
by the computer, not the engineer. In effect, one gets the
analysis of a design accomplished much quicker than even to
begin to build the prototype for testing in the lab.

Primary Air Flow System Design for Spun Bond Units

The primary airflow system in Spun Bond (SB) units
involves use of very high velocity air jets (often supersonic).
In this study, a primary airflow system with an inlet pressure
of 45 psig is analyzed using CFD. Figures 7 and 8 show Mach
Figure 8 number contours and velocity vectors in the domain predict-
VELOCITY VECTORS IN THE PRIMARY AIR- ed by CFD. These results demonstrate the capability of CFD
FLOW SYSTEM to capture flow separation in presence of blunt edges and
expansion channels due to the com-
pressibility effects of the pressurized
air.

Free Jet Flows

The distance between the air sys-
tem die exit plane and the web as
well as the flow pattern therein, have
important influences on the fiber lay
down process and hence the quality
(e.g., fiber orientation, and mechani-
cal properties) of the nonwovens
material. By design, this jet flow is
Figure 9 unconfined to allow significant
TURBULENT SPREADING OF AN UNCONFINED JET entrainment of surrounding air that

Figure 10 Figure 11
TIME-AVERAGED VELOCITY VECTORS IN INSTANTANEOUS VELOCITY VEC-
A PLANAR TURBULENT JET TORS IN A PLANAR TURBULENT JET

13 INJ Spring 2005

 results in large scale eddies. From the fiber lay-down
point of view, the important aspects of the jet expan-
sion include:
• The acceleration in the potential core flow and sub-
sequent expansion
• The formation of large scale eddies at the boundary
of the jet and the surrounding air; and most impor-
tantly,
• Momentum exchange between air and the polymer
leading to the deceleration of the jet and attenuation
and acceleration of the polymer fibers.
The equipment parameters and material properties
are crucial in achieving the right operating conditions.
These parameters are the Die-to-Collector-Distance
(DCD), the jet-width at the die opening, and the fluid
and thermodynamic properties of the polymer.
The characteristics of jet spreading and deceleration
are a function of the length-to-width ratio (L/W) of the
jet, where the length L is the DCD, and W is the slot
width as shown in Figure 4. For example, two different
jets with different exit widths, W, will appear similar in
jet spreading and velocity reduction when they are
plotted on the common basis of the non-dimensional
Figure 12 parameter L/W. This is depicted in Figure 9. Figures 10
and 11 show time-averaged and instantaneous velocity
DIFFUSER ANALYSIS: SPECIFIC CHARTS LIKE THIS
vectors in a planar jet as simulated using the RANS
CAN BE DEVELOPED FOR A SERIES OF DIFFUSER
and LES models.
GEOMETRIES

Figure 13
FLOW FIELD IN DIFFERENT DIFFUSERS: (A) AN UNSTALLED DIFFUSER; (B) A DIFFUSER WITH
TRANSIENT STALL CONDITIONS; (C) A DIFFUSER WITH FULLY STALLED CONDITIONS

14 INJ Spring 2005

 maintaining the same slot exit velocity) without increasing the
length L, then L/W ratio is reduced and the average velocity
at the web will be increased. That would have detrimental
effects by increasing the strength and velocity at which large
scale eddies hit the web. This is equivalent to reducing the
distance to the web at which the average jet velocity drops
below the fiber linear velocity. This latter effect would reduce
the amount of time and distance the fibers can randomly
weave and twist prior to laydown.
Diffusers are widely used for air delivery systems along
with controlled reduction of air velocity. Different regimes of
planar diffusers as obtained from experimental data are
shown in Figure 12. Based on the values of diffuser exit to inlet
area ratio and the ratio of diffuser length to diffuser inlet
Figure 14 width, the flow field in the diffuser will fall under different
GEOMETRY OF A DUCT WITH A FILTER-WEB regimes.
Figure 13 shows the type of flow fields that are obtained in
an unstalled diffuser, a diffuser with transient stall as well as
a diffuser with fully stalled conditions, respectively. These
results are obtained using the LES models available in FLU-
ENT commercial CFD software. In particular, the LES
approach helps in simulating the transient flow features,
which are very important in determining the quality of the
fabric produced [9,10].
Airflow plays a significant role in determining the fabric
quality in the nonwovens fabric manufacturing processes. A
defect-free nonwovens web manufactured using wet-laid or
air-laid system requires near perfect distribution of jets as
much as the polymer conditions to attain the designed condi-
tions for optimum entanglement, size distribution, and over-
all fabric quality [11,12]. ‘Logs’ or ‘sticks’ (bundles of fibers
Figure 15 with aligned cut ends that are never dispersed) may be
PATH LINES IN THE DUCT. RED ARE SOME formed due to a gross under-agitation of fibers during their
REPRESENTATIVE AIR PATH LINES; GREENS initial dispersion in the surrounding fluid. ‘Ropes’ may be
ARE FOR DUST PARTICLES formed when fibers encounter a vortex of about the same size
as the fiber length. In Spun Bond (SB) systems, airflow plays
The spreading of the jet shown in Figures 9 and 10 is a time- several important roles:
average characterization. The jet boundary is very unsteady • A ‘cold air stream’ cools continuous fibers leaving the
where large scale eddies are formed by the viscous shear spinneret. The characteristics of cold air streams can have a
between the high velocity jet and the quiescent surrounding significant impact on the fiber properties.
air (as shown in Figure 11). If the motion of the eddies in Figure • High-speed airflow (with Mach number ~1.0) is used in
11 is averaged over time then the gradual spreading profile of the fiber drawing process.
the jet is obtained as depicted in Figures 9 and 10. Since the • Turbulence in the airflow is utilized to obtain desired
large-scale eddy motions are present in reality, they have a Machine-Direction to Cross-Direction (MD/CD) strength
substantial impact on web formation because they entrain ratio.
fibers before they hit the web and eddies can also impact the
web and disturb material already laid down on the web. Filter webs in ducts
These latter effects can contribute to the blotchy appearance of This is a typical air filter application studied in a relatively
the nonwovens material. simple geometry to characterize the performance of the filter
Finally, the average jet velocity needs to drop below the web. Following similar pressure drop correlation as men-
fiber lineal velocity in order for the fibers to be able to weave tioned in the preceding example, here the analysis is per-
and twist in a random manner prior to laydown. The distance formed for transients. The geometry as shown in Figure 14,
above the web, at which this occurs, will have an influence on involves a corrugated nonwoven web covering the entire
the quality of the laydown process. cross-section of a circular duct. This type of filter webs are
All of the above effects need to be taken into account when fairly common in the industry and the simulation framework
adjusting the design of the slot hardware and how it is oper- is tailored to handle geometric configurations in a parametric
ated. For example, if the slot-exit width W is increased (while form: the pitch, height, basis weight/web denier can all be

15 INJ Spring 2005

 varied for comparative study. Variation in the pipe diameter
can be considered as well.
Results discussed here are limited to mild flow rates. As a
result, the flow is mostly laminar. Figure 15 shows an instan-
taneous distribution of the path lines in the duct. Only select-
ed tracers are monitored in this graphic. The red lines corre-
spond to the airflow whereas the green path lines are for the
dust particles. Note that in the current simulation, such web
properties are used that all particles are trapped by the web.
Using FLUENT’s discrete phase model, the transient trans-
port of the dust load is simulated (Figure 16) as part of the
overall flow solution. Simultaneously the increase in flow
resistance on the web surface is recalculated every time step
based on the distribution of the accumulated dust particles on
Figure 16 the surface from the beginning of the transient process (Figure
INSTANTANEOUS DISTRIBUTION OF DUST 17). It is interesting to note the distribution of the growth in
PARTICLES UPSTREAM OF THE FILTER IN resistance on the surface. Due to the resistance of the porous
THE DUCT web material itself, the flow impingement on the surface is
causing radial redirection of the flow upstream of the web. As
a result larger number of particles would hit the web away
from axis of the pipe. Figure 17 thus reflects the stronger
growth of resistance (blue is the weakest values, red being the
strongest) in an annulus region. Also note that this annulus is
not perfectly symmetric due to the corrugation of the web
material. However, the pattern is fairly symmetric along the
orthogonal alignments of the corrugation. Figure 18 shows the
resistance values along a diameter plane.

Conclusion
In recent years, CFD has been established as an effective
engineering tool that complements traditional design, analy-
sis and testing activities. Along with other CAE software
tools, CFD can provide significant cost and time savings dur-
Figure 17 ing the product life cycle. More specifically, as these tools
INSTANTANEOUS FLOW RESISTANCE MAP mainly impact pre-market activities, they can reduce both
ON THE FILTER MEDIUM time to market, as well as overall product development
investment.

Figure 18
RESISTANCE ALONG A DIAMETER (Z=0) OF THE PIPE ON THE FILTER SURFACE

16 INJ Spring 2005

References:
1. Thornton G.; The Enterprise-Wide Impact of Computer-
Aided-Engineering on Manufacturers 1996.
2. Management Roundtable; Product Development in the
1990’s, 1997.
3. Smith P., Make Time-to-Market Technologies a Bottom-
Line Issue, Computer-Aided Engineering, vol. 17, n4, p. 78, 1998.
4. Baer, T. A. and Finlayson, B. A., "Comparison of
Numerical Simulations of Polymer Flow with Experimental
Data", Theoretical and Applied Rheology, P. Moldenaers, and R.
Keunings (ed.), vol. 1, pp. 247-249 1993.
5. Staples P., and Giles A., Computer Simulation Helps Triple
Engineers' Efficiency While Improving Quality of Nonwoven
Fabrics, DuPont Nonwovens, Old Hickory, TN, Journal Article
by Fluent Users.
6. Mukhopadhyay A., Sun J., Troshko A., and Prasad R.O.,
Modeling and Analysis of Polymer and Air Flow Configurations in
Nonwovens Industry, Beltwide Cotton Conferences,
Nonwovens Conference, Atlanta, GA January 8-12, 2002.
7. Mukhopadhyay A., Avalosse T., Troshko A., and Prasad
R.O., An Album Of Computational Modeling In Nonwovens
Industry, 3rd INTC Conference, Atlanta, 2002. Shuler B.,
Johnston B., and Mukhopadhyay A., Design Improvement
and Flow Visualization of Patented Attenuator Using CFD
Tools, TANDEC, Knoxville, TN, Nov. 18-21, 2002.
9. Prasad, R.O.S. and Runstadler, P.W., Proceeding of the
INTC Conference, September 26-28, Dallas, TX, USA (2000),
pp.13.1-13-16.
10. Mukhopadhyay, A., , R.O.S. Prasad, E. Grald, J. Sun, and
N. Lifshutz, Performance Analysis of Melt-blown Dies Using
Computational Fluid Dynamics, 2nd INTC Conference,
Baltimore, 2001.
11. White C., Nonwovens Training Course Material, INDA,
March 2000.
12. Goswami, B.C., Spunbounding and melt-blowing processes,
in Manufactured Fiber Technolgy, Edited by Gupta, V.B. and
Kothari, V.K., Chapman & Hall, London, (1997). — INJ

Return to Table of Contents

17 INJ Spring 2005