Computational Fluid Dynamics Simulations and Experiments of

Jan 27, 2016 - Computational Fluid Dynamics Simulations and Experiments of. Meltblown Fibrous Media: New Die Designs to Enhance Fiber. Attenuation and...
0 downloads 0 Views 5MB Size
Article pubs.acs.org/IECR

Computational Fluid Dynamics Simulations and Experiments of Meltblown Fibrous Media: New Die Designs to Enhance Fiber Attenuation and Filtration Quality Mohammad Abouelreesh Hassan,† Nagendra Anantharamaiah,‡ Saad A. Khan,*,† and B. Pourdeyhimi*,‡ †

Department of Chemical & Biomolecular Engineering and ‡The Nonwovens Institute, North Carolina State University, Raleigh, North Carolina 27606, United States S Supporting Information *

ABSTRACT: The meltblowing process employs high-speed hot air jets to attenuate polymer streams injected from a die head. In this study, we examine design strategies to control the air flow field below the polymer injection point to achieve higher fiber attenuation and meltblown webs with smaller fiber diameters. Computational fluid dynamics (CFD) simulations for new die configurations show that vertical or inclined air constrictors around the primary air jets keep the centerline air velocity and temperature at their maximum values for 10−15 mm longer below the die face than the reference die. Polymer streams are kept near their melting temperatures at higher air velocities for a longer period, resulting in higher fiber attenuation. The underlying mechanisms leading to such behavior are discussed. Experimental results show reduction in fiber diameter and pore size, validating the simulation. Improved filtration properties are also obtained from the nonwovens webs.

1. INTRODUCTION Meltblowing (MB) technology is a melt-spinning process used to produce microfibers through injecting molten polymer streams into high-velocity air jets. In the MB process, highvelocity air jets impinge upon the polymer streams as they emerge from the spinneret (see Figure S1). The drag force caused by the air jets rapidly attenuates the fibers, reducing their diameter as much as a hundred times the nozzle diameter. Meltblown fiber diameters can range from 0.5 to 10 μm within the same web with an average fiber diameter of 1−2 μm. These meltblown fibrous webs are known for their large surface area per unit weight, high insulation value, and excellent barrier properties while retaining breathability.1−5 Fiber formation during the MB process is critically dependent on the aerodynamics of the process because the drag force due to high-speed air jets is the main cause of fiber attenuation.6,7 The maximum air velocity plays a significant role because polymer streams spend most of the time in the centerline below the die face. It is also useful to have a high air velocity over a long range of z (distance from die face) values, i.e., the integral of the air velocity versus z curve should be as large as possible. A higher air velocity in the z-direction is desirable because it leads to an increased rate of fiber attenuation (i.e., finer fiber) for a given air flow rate.8,9 The advantages of small fibers are well-documented,4 particularly in filtration applications where meltblown fibers are typically used. Several investigators have thus examined different die geometries to enhance the air flow field below the die face with a view to achieving smaller fiber diameters and better web uniformity. Tate and Shambaugh6 investigated the effect of the spinneret nosepiece geometry on the air flow. They found that blunt MB dies produce a lower maximum centerline air velocity, causing the centerline velocity profile to decay at a rate higher than that of sharp MB dies. Kurtka et al.7−9 examined the effect of the spinneret nosepiece position to the die face. © 2016 American Chemical Society

They found that the inset MB die (die tip above die face) is superior to the outset (die tip below die face) and the flush dies (die tip and face are at same level) because it gives a higher centerline air velocity. Using computational fluid dynamics (CFD), Kurtka and co-workers showed that increasing the nosepiece recession leads to an increase in the maximum centerline velocity and that the air velocity at the highest recession is three times the air velocity for a flush die. This is significant because a higher maximum air velocity in the zdirection leads to an increased rate of fiber attenuation for a given air flow rate. Furthermore, Kurtka depicted the turbulence fluctuations for the inset and outset dies which are quite important in the evaluation of die design because it affects fiber laydown and the web uniformity. He found that as the recession above the die face increases for the inset dies, the velocity fluctuation increases, which may cause fiber to bend and/or stick to the die walls. For outset dies, the velocity fluctuation is found to decrease as the nose piece extends beyond the die face; therefore, the probability of having web defects such as ropes is lowered. Tan et al.10 examined the effect of using Laval nozzles on the airflow field in meltblowing apparatus. Their simulation model showed that a Laval nozzle influences the airflow field by increasing the maximum value of air centerline velocity and eliminating the compression waves at a certain inlet pressure values. Such nozzle configurations would reduce the meltblown fiber diameter but would require working and operating equipment at air velocities above the sound velocity, which is difficult and can cause hearing health problems. Xie et al.11 studied the effect of the swirl die and the slot MB die on the air flow field. Their results showed that fiber Received: Revised: Accepted: Published: 2049

October 26, 2015 December 23, 2015 January 27, 2016 January 27, 2016 DOI: 10.1021/acs.iecr.5b04020 Ind. Eng. Chem. Res. 2016, 55, 2049−2058

Article

Industrial & Engineering Chemistry Research

design to the sharp MB die geometry enhanced air flow and produced finer fibers.6 Modifying the die design by changing the position of the nose piece that has the polymer holes from the flush position to the inset position enhanced the air flow field as well.7−9 Fiber size is also highly dependent on the polymer throughput through the holes, and many investigators4,5,13,14,17 showed that reducing polymer throughput would result in decreased fiber size. To achieve submicron fibers using the meltblowing process, investigators4,11−14,17 had to use a different die design that had much smaller capillaries with larger L/D ratios to achieve uniform polymer flow distribution when working at very low polymer throughputs. Therefore, most instances of achieving smaller MB fiber size would require either a major die redesign, which is cost intensive, or operating at low polymer throughput, which hurts the process economics. The present work offers an easy way to achieve smaller fiber diameters without sacrificing production rate or investing in a new meltblowing die. We combine computer simulations with experiments to show that simple add-on devices to existing die designs can enhance their performance to achieve smaller fiber size at higher production rates. In this regard, we investigate new die configurations that have vertical or inclined air constrictors to produce finer meltblown fibers. Computational fluid dynamics was used to predict the effects of these new die configurations on the air flow field below the die face. The new die configurations are able to not only maintain a high centerline air velocity below the polymer injection point but also keep the polymer temperature around the melting point (Tm) for a longer period near the die face. In addition, we experimentally test the design concept of these new configurations and show their effects on fiber diameter and web properties and concomitant improvement in filtration properties. This complete package, from simulations to experiments to testing of actual applications of meltblown fibers, together with using a facile die design concept provides uniqueness to this work.

whipping in the slot-die melt-blowing process follows a twodimensional motion but takes a three-dimensional spiral path in the case of a swirl-die melt-blowing process. Fabricating nanofibers using meltblowing has also been a target of multiple studies to overcome the significantly low throughput of electrospinning technique typically used to make such fibers. In a prior study in our laboratory,4 we fabricated nano meltblown media with average fiber size of 330 nm using 150 μm polymer capillaries and found high filtration quality factors compared to traditional meltblown media. However, a primary issue of operating such nano meltblowing dies with fine capillaries on a commercial scale would be the lower up-time due to the high probability of polymer hole plugging in addtion to the initial high capital investment of acquiring it. Kolbasov et. al12 showed the possibility of producing nanofibers on a larger scale than electrospinning by using the solution blowing process. They fabricated webs with ∼500 nm fibers using a multirow spinneret with concentric annular nozzles. Although the technology is promising for mass production of nanofibers from nonmeltable polymers, the process was difficult to control and requires solvent reclamation handling procedures. Zuo et al.13 fabricated nanofibers using the meltblowing technology using immiscible blends of polymer in a fashion similar to the islands-in-the-sea spunbond fiber technology. The issues with such technology are the difficulty dissolving the sea polymer, the added cost of washing out and sacrificing one of the polymers, possible hazards from the solvents used, and the high initial capital cost of the bicomponent meltblowing die technology. Uppal et al.14 achieved high filtration quality factors using meltblown nanofibers with average fiber size of 550 nm using a nanofiber modular meltblowing die. To achieve such average fiber size they had to work at 0.022 g/hole/min, which is quite low compared to traditional meltblowing process polymer throughputs (0.3−3 g/hole/min). The rheology of polymer melts plays an important role in influencing fiber size in the meltblowing process although it is often challenging to measure experimentally the rheological properties at the high deformation rates that are typical in meltblowing (∼106 s−1).15 Tan et al.15 and Zhou et al.16 have investigated the effect of polymer viscoelasticity on the average fiber size and the fiber size distribution as a function of shear stress using a one-dimensional slender-jet model. They evaluated different viscoelastic models such as the Newtonian, upper-convected Maxwell, Phan-Thien and Tanner (PTT), and Giesekus constitutive equations using this approach. The results showed a strong dependence of fiber diameter on the air shear stress and variations in fiber diameter with viscoelasticity. The most common MB die technology currently used is the Exxon style slot die that has a single row of nozzles sandwiched between two inclined air jets. The second MB technology that is currently utilized commercially is the biaxial Schwarz MB die that has multirow polymer nozzles with annular air jets. Hassan et al.17 reported a meltblowing die based on parallel plate die design that could be used to fabricate meltblown fibers of 3−5 μm. Although the manufacturing cost of such a design would be 1/10 that of the slot die, the traditional Exxon slot meltblowing die out-performed it at similar operating conditions. Given the high industrial relevance of the slot meltblowing die technology, we have chosen it to be the focal point of this investigation. In many of the referenced studies, investigators had to change the die geometry to enhance the air flow field to achieve higher drag force. For instance, moving from the blunt MB die

2. NUMERICAL MODELING AND SIMULATION PARAMETERS Computational fluid dynamics (CFD) is a useful tool for examining new die geometries without the cost associated with experimental testing. It also provides us with other important parameters that are not easily measured experimentally, such as turbulent intensity, dissipation rate, and air flow field profiles for multiple die configurations in order to identify favorable design parameters. 2.1. Turbulence Modeling. The strong dependence of the fiber formation process on jet aerodynamics inspired us to study the effect of changing the die geometry on the air flow, which is the main cause for the fiber attenuation.18−25 The CFD package used in this investigation included two programs: GAMBIT 2.2.6 and FLUENT 12.1.4. GAMBIT provides the tools necessary for creating the mesh geometrical domain of the simulation model, while FLUENT performs all the computational and postprocessing tasks by using the built-in models. FLUENT has different turbulence models in order to simulate the air flow field. Turbulence models may differ in computation time, accuracy, and suitability of the model to the problem needed to be solved. The most popular models in modeling turbulent flow are the standard K-ε, standard K-ω, realizable Kε, and Reynolds stress models (RSM).8,23,24 Krutka et al.7−9 showed that the standard K-ε and the realizable K-ε models are 2050

DOI: 10.1021/acs.iecr.5b04020 Ind. Eng. Chem. Res. 2016, 55, 2049−2058

Article

Industrial & Engineering Chemistry Research

Figure 1. Domain size and its boundary conditions for the reference die configuration.

not sufficient to predict the air flow field of a blunt or sharp flush meltblowing die. They reported that the RSM was able to reproduce the blunt-die system with the most success. In a recent study by Begenir,24 she showed experimentally that the standard K-ε model with its default parameters predicted sufficiently the air flow field of the sharp inset meltblowing die. The numerical procedures detailed by Begenir24 have been followed in the present work, and the standard K-ε model with its default parameters from Fluent has been used. The simulations did not require the RSM model, and experimentally updated values for constants were not needed, i.e., the default parameters from Fluent worked well. The equations for the K-ε model takes the form shown below. Equation parameters are defined as follow: k, turbulent kinetic energy; C1ε, first empirical constant for ε equation; C2ε, second empirical constant for ε equation; Cμ, empirical constant for μt; Dh, hydraulic diameter; Gk, generation of turbulence kinetic energy; Sk, user-defined source term of turbulence kinetic energy; Sε, user-defined source term of turbulence dissipation rate; ui, u i-component of velocity vector; ε, turbulence dissipation rate; σk, empirical constant for k equation; and σε, empirical constant for ε equation. The suggested values of the empirical constants embedded in these equations are as follows, C1ε = 1.44, C2ε = 1.92, Cμ = 0.09, σk = 1, and σε = 1.3.8,24 Convergence of the model equations were required to reach 10−6 residuals, and approximately 10,000− 20,000 iterations were necessary to meet this convergence criterion for the different simulations. The K-ε equations model:

G k = −ρu i′u j′

μt = ρCμ

(1)

μ ⎞ ∂k ⎤ ∂ ∂ ∂ ⎡⎢⎛ ε (ρε) + (ρεu i) = ⎜μ + t ⎟ ⎥ + C1εG k ∂t ∂xi ∂xj ⎢⎣⎝ σε ⎠ ∂xj ⎥⎦ k ε2 + Sε k

(3)

(4)

Equation 1 represents the kinetic energy equation, while eq 2 represents the turbulence dissipation rate. The first two terms on the right-hand side of eq 1 represent the production of kinetic energy due to interactions between the mean flow and the products of the turbulent fluctuations. The following term on the right-hand side represents the transformation of kinetic energy at small scales to internal energy. Finally, the last term in eq 1 represents the turbulence kinetic energy due to viscous diffusion, turbulent fluctuations, and pressure (velocity fluctuations). Similar definition can be drawn about eq 2 for the turbulence dissipation rate. Equation 3 is to calculate the generation of turbulence kinetic energy, while eq 4 is used to calculate the eddy viscosity. 2.2. Domain Size and Grid Generation. The computational domains of the new configurations were generated using GAMBIT 2.2.6, and Figure 1 represents the computational domain used in the simulations for the reference die configuration. The origin of the MB domain is at the center of the die face, with the x-direction traversing the major slot axis; the y-axis (not shown) is perpendicular to the plane of the drawing, and the z-direction is in the downward direction below the die. Similar to Krutka et al.7−9 and Begenir,24 the domain dimensions were 30 mm and 100 mm in the x- and zdirections, respectively. These dimensions are large enough to represent the air velocity and temperature profiles below the die. The presence of the polymer and its spinneret was neglected, which is reasonable because of the small dimensions of the polymer spinneret relative to the die dimension. A symmetry boundary condition was used along the z-axis (x = 0) to reduce the size of the computational domain.7−9 The grid is structured with quadrilateral cells because it is convenient for the rectangular nature of the computational domain. The grid resolution of the area that is close to the symmetry line and the die face is fine structured and it becomes coarser toward the

μ ⎞ ∂k ⎤ ∂ ⎡⎢⎛ ∂ ∂ (ρk) + (ρku i) = ⎜μ + t ⎟ ⎥ + G k − ρε + S k ∂x j ⎢⎣⎝ ∂xi ∂t σk ⎠ ∂xj ⎥⎦

− ρ C 2ε

k2 ε

∂u i ∂xi

(2) 2051

DOI: 10.1021/acs.iecr.5b04020 Ind. Eng. Chem. Res. 2016, 55, 2049−2058

Article

Industrial & Engineering Chemistry Research Table 1. Simulated New Die Configurations

effects.19,24 Table 1 shows the different die geometries under investigation in this study. We studied the effect of the air constrictor length (L), width (W), and angle (Φ) on the air flow field below the MB die.

outlet pressure boundary conditions. The number of quadrilateral cells for each die geometry was around 120 000. The air flow entering the computational domain was set as a pressure inlet boundary condition at T = 537 K and P = 4.31 psi. The right boundary and the bottom boundary (see Figure 1) were set as pressure outlets with ambient air conditions (T = 300 K and P = 0 psig [gauge pressure in lb/in2]). The left boundary was set as a line of symmetry. All other boundaries were assigned the default setting of being a wall at a temperature equal to 537 K. The turbulence specification was set with an intensity of 5% and a turbulence scale length of 0.1 mm. These values have been calculated according to eqs 5 and 6. The turbulence intensity and turbulence hydraulic diameter were set to be 4% and 100 mm, respectively. The nonisothermal air flow was modeled as a compressible flow, and its density (ρ) and viscosity (μ) were calculated using ideal gas law and kinetic theory, respectively. The turbulence intensity and the turbulence scale length were calculated according to the following equations: I = 0.16(ReDH)−1/8

3. RESULTS AND DISCUSSION 3.1. Effect of Air Constrictor Length. We examined the effect of the air constrictor length on the air flow field below the MB die (see Table 1, configurations 1−3). We set the width between the two air constrictors at 20 mm, i.e., 10 mm from the spinneret centerline in the x-direction, and varied the length of the air constrictor (10, 20, and 30 mm). The effect of varying the air constrictor length on the centerline air velocity is shown in Figure 2A, while that on the temperature profile is depicted in Figure 2B. We find the maximum air centerline velocity increases slightly from 271 to 277 m/s, and the maximum air centerline velocity was maintained for a distance of 10−15 mm longer below the die face. Maintaining the maximum centerline air velocity for a longer distance than the reference configuration increases the overall fiber attenuation. Centerline air velocity increased because the hot air jets have been constricted by limiting the cross-sectional area underneath the die face and by confining them between the constrictors’ boundaries for a longer distance below the die face. This improvement in velocity profile is expected to result in higher fiber attenuation and hence smaller fiber diameters.7−9 No significant effect on the maximum centerline temperature was noticed; however, the presence of the constrictors maintained the maximum centerline temperature for an increased distance below the die face, from 10 to 18 mm, because air constrictors delayed the interaction between the hot air jets and the ambient air near the die face. Higher attenuation would likely occur because the polymer is kept at a higher temperature and experiences higher air velocities for a longer time. We also notice that the longer the constrictor, the higher the temperature profile along the measured distance from the die face, but a longer constrictor may increase the turbulence and may cause fiber roping or fiber sticking to the constrictors’ walls. 3.2. Effect of Air Constrictor Width. The configurations corresponding to Figure 2 had a width of 20 mm between the

(5)

where I is the turbulence intensity and Re is the Reynolds number based on the hydraulic diameter. l = 0.07L

(6)

where S is the turbulence scale length and L is the characteristic length of the duct or the hydraulic diameter 2.3. Die Configurations. We investigated 11 new die configurations and compared the air flow field in them to the air flow field of a reference MB die provided by Reicofil and operated on a pilot-scale MB line at the Nonwovens Institute at North Carolina State University. The reference die, see Figure S2, is a sharp-edged, inset die with a recession distance of a = 1.524 mm and air plates angle of 45°. The air gap width d = 1.524 mm, the air slot width b = 0.63 mm, and the distance between the outer edges of the air slots h = 1.263 mm (h = 2b), while the overall length of the air slot S = 737 mm. Because the aspect ratio of the air slot S /b = 1168 is significantly larger than 50, the air flow field converging from the air jets in the meltblowing line used in this study could be modeled as twodimensional at positions below the die center and free of end 2052

DOI: 10.1021/acs.iecr.5b04020 Ind. Eng. Chem. Res. 2016, 55, 2049−2058

Article

Industrial & Engineering Chemistry Research

Figure 2. Effect of air constrictor length on centerline air velocity (A) and centerline air temperature (B).

two constrictors and a varying length. Here, we examine the effect of varying the width (6, 10, 30, and 40 mm; see Table 1, configurations 1 and 4−7) keeping the length of the constrictor constant at 10 mm. Figure 3 depicts the effect of air constrictors width on (A) centerline air velocity, (B) centerline air temperature, and (C) centerline air turbulent intensity profiles below the die face. A pronounced effect on velocity and temperature is seen at smaller widths (e.g., W = 6 mm) because of the substantial decrease in the cross-sectional area for the air jets to flow through it. The temperature profile of configuration 4 (W = 6 mm), shows a sudden drop at the beginning due to sudden pressure drop of the air within the narrow area between the air constrictors. This pressure drop is due to the sudden expansion of the air jets inside the limited area between the air constrictors walls, and the corresponding temperature decrease likely occurs because of the Joule−Thomson effect that may take place during this sudden expansion.26 When the distance between air constrictors is reduced, the maximum centerline air velocity and temperature can be increased and/or maintained for a longer distance below the die face, but the potential effect on turbulence should also be taken into consideration. Turbulent intensity and the turbulent kinetic energy are important factors to be considered because they greatly influence fiber formation, entanglement, and roping. Thus, we investigated how the air constrictor width affects the turbulent intensity. Figure 3C shows that as the air constrictor width decreases, the maximum centerline turbulent intensity increases and its peak shifts toward the die face. The turbulent intensity is defined as the root-mean-square of the turbulent velocity fluctuations divided by the mean velocity of the flow. The maximum centerline turbulent intensity shifts toward the die face with decreasing air constrictor width because of the higher velocity fluctuation and more eddies that are generated by these air constrictors. The closer the constrictor walls are to the

Figure 3. Effect of air constrictor width on centerline air velocity (A), centerline air temperature (B), and centerline turbulent intensity (C).

centerline of the process air, the more fluctuation they will generate at a distance closer to the die face. Fiber roping can take place in this highly turbulent region where polymer streams are in highly turbulent flow in their molten state. Figure 3C also shows that the minimum centerline turbulent intensity decreases with increasing air constrictor width. This is primarily due to the delay of the interaction of the process air with the ambient air. This correlation does not hold in the case of configuration 4 because we have much higher eddy intensity close to the air constrictor walls that are only 6 mm from the centerline. In general, configuration 4 showed the highest turbulent intensity because of the limited expansion for the air jets that leads to higher velocity and higher velocity fluctuation. Configurations with W = 20 mm and above show profiles relatively similar to that of the reference case, especially at the peak point. In general, configurations with widths between 10 and 40 mm achieve better performance over the reference configuration by keeping the maximum centerline air velocity and maximum centerline air temperature constant at their peak 2053

DOI: 10.1021/acs.iecr.5b04020 Ind. Eng. Chem. Res. 2016, 55, 2049−2058

Article

Industrial & Engineering Chemistry Research

Figure 4. Effect of air constrictor width on air velocity contour map (A), air velocity vectors map (B), and air temperature contour map (C).

near to the die face and air constrictors is a velocity recirculation zone, the velocities are relatively low. However, for the case of the temperature contours, this recirculation area is filled with hot air that has limited interaction with the ambient air. Thus, the air temperature in these recirculation zones remains high and would keep the polymer streams at a lower viscosity during the attenuation process (see Figure 4C). 3.3. Effect of Air Constrictor Angle. The effect of air constrictor angle on the air flow field below the MB die was examined. The investigated air constrictor angles were 45°, 60°, 72°, and 90o (see Table 1). The constrictor fixed-end was set 3 mm from the spinneret centerline, while the constrictor freeend was set 27 mm below the die face. Panels A and B of Figure 5 show the maximum centerline air velocity and temperature for different air constrictors, respectively. Configurations with angle between 45° and 60° may perform better than the other two configurations because of maintaining higher centerline air velocities and temperatures for a longer distance below the die face, which would result in higher fiber attenuation. The air constrictor with an angle Φ = 90° exhibits this with a much higher maximum centerline air velocity but with rapid decay. This could result in considerably higher fiber attenuation in the first 10 mm below the die face but much lower fiber attenuation at distances larger than 10 mm. The temperature profile of

values for a longer distance below the die, while also having minimal effect on the turbulent intensity and its dissipation rate. Figure 4 shows the mean air velocity, the velocity vectors, and the static temperature contour maps for air flow field of meltblowing dies with different air constrictor widths. (Static temperature is the temperature that would be measured experimentally if a temperature probe were used in the flow field.) The velocity and temperature contour plots look quite similar, as might be expected from the fundamental relationship between heat- and momentum-transfer processes.7,8 As shown, as we decrease the width of the air constrictors around the air jets, air velocity and air temperature increase (contour color becomes darker red and extended for a longer distance below the die face), but the air turbulence increases and the recirculation zone comes closer toward the centerline where polymer streams exist (see highlighted areas in Figure 4B). The shift of the recirculation zones toward the centerline by decreasing the air constrictor width is not recommended, as it would increase the probability of rope formation, especially if the magnitude of these velocity vectors increased as we see in case 5 (L = 10 mm and W = 10 mm). Near the die face we have a significant difference between the velocity and temperature contour maps. Because the zone between the air inlets and very 2054

DOI: 10.1021/acs.iecr.5b04020 Ind. Eng. Chem. Res. 2016, 55, 2049−2058

Article

Industrial & Engineering Chemistry Research

face. Thus, air constrictors with angles between 45° and 60° would be the best choice because they will keep the maximum centerline air velocity and temperature for a longer distance below the die without a significant increase in the turbulence near the die face. 3.4. Experimental Investigation. Previous studies have validated the use of the K-ε model in CFD simulations for quantitatively predicting the air velocity and temperature profiles for the reference meltblowing die configuration.7,8,24 In this section, we investigate experimentally if constricting the air flow underneath the MB die face will have an effect on the fiber size and web properties as implied by our CFD results. Our objective is to examine if the concept of using constrictors lead to attenuation in fiber diameter; hence, we use constrictor distances that are conducive for conducting experiments. We are by no means attempting any quantitative comparison with simulations. We fabricated three air constrictors and attached them to a small MB die at the Nonowovens Institute located at North Carolina State University. Five different fabrics were produced in each trial, for a total of 20 samples (see Table 2). Table 2. Meltblown Nonwovens Processing Conditions

All fabrics were produced at the same processing conditions, and the only variable was the aspirator pressure that was varied between 15 and 35 psi (1 psi = 6 894.8 Pa). The polymer used was an isotactic polypropylene manufactured commercially by the ExxonMobil Chemical Company under the trade name Achieve 6936G1. This resin has a melt flow rate of 1550 g/10 min, according to ASTM Standard D 1238, at 230 °C and 2.16 kg. This resin is the most common polypropylene used for producing meltblown webs because it is easy to spin because of its higher melt flow rate that requires lower operating pressures and temperatures. Achieve 6936G PP resin has a numberaverage molecular weight of 25 000 g/mol, and a weight average molecular weight of 60 000 g/mol.27 Steady shear measurements conducted using a TA Instruments AR2000 rheometer with a parallel plate geometry (1 mm gap) reveal the sample to have zero shear viscosities of 18 and 10 Pa·s at 180 and 220 °C, respectively (Figure S3). During the meltblowing process, the polymer melt temperature was held constant at 500 K and air temperature was kept constant at 537 K. The polymer throughput was held constant at 0.3 g/hole/min. The fabrics produced had a basis weight of 30 g/m2. Figure 6A shows an SEM image for fibers produced using a MB die with air constrictors positioned 10 mm from the spinneret centerline, while Figure 6B shows another SEM micrograph for a MB sample produced at operating conditions using the same MB die, but with air constrictors positioned 4 mm from the spinneret centerline. We find the fibers in the first image to be slightly larger than those in the second image. Note, however, that the second image shows significant roping. We noticed shot formation and roping defects in the case of the

Figure 5. Effect of air constrictor angle on centerline air velocity (A), centerline air temperature (B), and centerline air turbulent KE (C).

configuration 11, Φ = 90°, shows a sudden drop at the beginning due to a pressure drop of the air within the narrow area between the air constrictors and the concomitant Joule− Thomson effect.26 Centerline temperature starts to increase again, possibly as the pressure increases and equalizes with the outside atmosphere; the centerline velocity stays constant at 530 K for ∼20 mm and then starts to decrease at a rate similar to that of the other die configurations once the air starts to interact with the ambient air near the edge of the constrictors. Figure 5C shows the turbulent intensity for the simulated die configurations with different air constrictor angles. Configurations 8 and 9, Φ = 45° and 60°, respectively, do not show a significant difference in their turbulent intensity from that of the reference die configuration. Configurations 10 and 11, Φ = 72° and 90°, respectively, show the highest centerline turbulent intensity at a closer distance to the polymer injection point, and their maximum centerline turbulent intensity was higher than the reference case by 20−40%. This is not favorable because it will increase the likelihood for fiber entanglement and roping of the polymer streams that are in the molten state near the die 2055

DOI: 10.1021/acs.iecr.5b04020 Ind. Eng. Chem. Res. 2016, 55, 2049−2058

Article

Industrial & Engineering Chemistry Research

Figure 6. SEM images for fibers produced at 35 psi using (A) air constrictors positioned 10 mm from the spinneret centerline line and (B) air constrictors positioned 4 mm from the spinneret centerline line.

second and third air constrictors, but in the case of the first air constrictor that was 10 mm from the spinneret centerline there were almost no shots or roping defects. We believe that the high turbulence near the die face is the main reason behind these defects, and is consistent with our predictions from CFD simulations on the turbulent intensity. Such high turbulence near where the polymer exits would increase filament perturbation and fiber whipping probabilities. Whipping is thought to contribute to roping and shots defects in the meltblowing process and is primarily caused by high inlet oscillation and strong longitudinal stresses along the jet axis.28,29 In addition, any imperfection in fabrication or alignment of the two air constrictors around air jets may perturb the molten polymer streams and direct them toward one of the constrictors’ side and initiate rope formation. Figure 7 shows the average fiber diameter for the fabrics produced using a MB die with different air constrictors. We

higher centerline air velocities and temperature profiles in the case of MB dies with different air constrictors widths. In the case of the air constrictors positioned 10 mm from the spinneret centerline, we reduced the fiber diameter by 20% compared with the fibers produced using the reference MB die at the same process conditions when air aspirator pressure was 15 psi. In the case of the air constrictors at 7 mm from the spinneret centerline, we obtained fibers that were 20−40% smaller than the fibers produced using the reference die. Figure 8 shows the average pore diameter of the produced meltblown nonwovens. Mean flow pore diameter was measured

Figure 8. Mean flow pore diameter versus aspirator pressure.

by using PMI capillary porometry.30,31 The fabric produced by using MB die with air constrictors had relatively smaller pore diameters compared to those produced with the same MB die without using air constrictors. This is attributed to the smaller fiber size as a result of using the air constrictors that accelerated the air flow and increased the drag force and fiber attenuation on the spun polymer streams. Figure 9 shows the filtration efficiency and filtration quality factor of webs obtained using the air constrictors. Filtration properties of the fabrics are measured using a TSI 3160 instrument32 by challenging them against 0.3 μm DOP particles at a face velocity of 77.5 mm/sec. The results show enhancement in filtration efficiency for most of the cases of the fabrics produced by using MB die with air constrictors over the reference fabrics. Figure 9B shows the normalized filtration properties per unit thickness in the form of the filtration quality factor (Qf). Qf is defined as the average fractional capture per unit thickness divided by the pressure drop per unit thickness.33,34 It is evident that filtration quality factor is

Figure 7. Average fiber diameter versus aspirator pressure for experimental runs using different die configurations.

were able to produce fabrics using the MB die with air constrictor at 10 mm from the spinneret centerline and with air constrictors at 7 mm from the spinneret centerline. In the latter case, we could not produce a uniform web because of the fiber roping and sticking to the constrictor walls, but we were able to collect some fibers and measured their diameter distribution using scanning electron microscopy (SEM). As shown in Figure 7, the MB die with different air constrictors was able to reduce the fiber size in the fabrics produced. This is consistent with the expectation from our CFD simulation results that showed 2056

DOI: 10.1021/acs.iecr.5b04020 Ind. Eng. Chem. Res. 2016, 55, 2049−2058

Industrial & Engineering Chemistry Research



Article

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.iecr.5b04020. Schematic drawing of the meltblowing process; crosssectional views of a 45° sharp, inset meltblowing die in the z- and the y-directions; and viscosity versus shear rate at 180 and 220 °C for Achieve 6936G PP resin (1550 MFR) (PDF)



AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. Tel.: 919-515-6551. *E-mail: [email protected]. Tel.: 919-515-4519. Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS The authors gratefully acknowledge the Nowovens Institute of North Carolina State University for supporting this research. REFERENCES

(1) Shambaugh, R. L. A macroscopic view of the melt-blowing process for producing microfiber. Ind. Eng. Chem. Res. 1988, 27 (12), 2363. (2) Lee, Y.; Wadsworth, L. C. Effects of melt-blowing process conditions on morphological and mechanical properties of polypropylene webs. Polymer 1992, 33 (8), 1200. (3) Ellison, C. J.; Phatak, A.; Giles, D. W.; Macosko, C. W.; Bates, F. S. Melt blown nanofibers: Fiber diameter distributions and onset of fiber breakup. Polymer 2007, 48, 3306. (4) Hassan, M. A.; Yeom, B.; Wilkie, A.; Pourdeyhimi, B.; Khan, S. A. Fabrication of nanofiber meltblown membranes and their filtration properties. J. Membr. Sci. 2013, 427, 336. (5) Hassan, M. A. Structure Property Process Relationships for Meltblown Fibrous Media. Ph.D. Dissertation, North Carolina State University, Raleigh, NC, 2013. (6) Tate, B. D.; Shambaugh, R. L. Modified Dual Rectangular Jets for Fiber Production. Ind. Eng. Chem. Res. 1998, 37 (9), 3772. (7) Krutka, H. M.; Shambaugh, R. L.; Papavassiliou, D. V. Analysis of a Melt-Blowing Die: Comparison of CFD and Experiments. Ind. Eng. Chem. Res. 2002, 41, 5125. (8) Krutka, H. M.; Shambaugh, R. L.; Papavassiliou, D. V. Effects of Die Geometry on the Flow Field of the Melt-Blowing Process. Ind. Eng. Chem. Res. 2003, 42 (22), 5541. (9) Krutka, H. M.; Shambaugh, R. L.; Papavassiliou, D. V. Effects of Temperature and Geometry on the Flow Field of the Melt Blowing Process. Ind. Eng. Chem. Res. 2004, 43 (15), 4199. (10) Tan, D. H.; Herman, P. K.; Janakiraman, A. F.; Bates, S.; Kumar, S.; Macosko, C. W. Influence of Laval Nozzles on the Air Flow Field in Melt Blowing Apparatus. Chem. Eng. Sci. 2012, 80, 342. (11) Xie, S.; Zheng, Y.; Zeng, Y. Influence of Die Geometry on Fiber Motion and Fiber Attenuation in the Melt-Blowing Process. Ind. Eng. Chem. Res. 2014, 53, 12866. (12) Kolbasov, A.; Sinha-Ray, S.; Joijode, A.; Hassan, M. A.; Brown, D.; Maze, B.; Pourdeyhimi, B.; Yarin, A. L. Industrial-Scale Solution Blowing of Soy Protein Nanofibers. Ind. Eng. Chem. Res. 2016, 55, 323. (13) Zuo, F.; Tan, D. H.; Wang, Z.; Jeung, S.; Macosko, C. W.; Bates, F. S. Nanofibers from Melt Blown Fiber-in-Fiber Polymer Blends. ACS Macro Lett. 2013, 2, 301. (14) Uppal, R.; Bhat, G.; Eash, C.; Akato, K. Meltblown Nanofiber Media for Enhanced Quality Factor. Fibers Polym. 2013, 14 (2), 660. (15) Tan, D.; Zhou, C.; Ellison, C. J.; Kumar, S.; Macosko, C. W.; Bates, F. S. Meltblown Fibers: Influence of Viscosity and Elasticity on Diameter Distribution. J. Non-Newtonian Fluid Mech. 2010, 165, 892.

Figure 9. Filtration efficiency (A) and filtration quality factor (B) for the produced nonwovens.

greatly enhanced especially at aspirator pressures from 15 to 25 psi. The enhancement in the filtration properties is mainly due to the smaller fabricated fiber diameters that we obtained by using MB die with air constrictors that enhanced the air flow field as predicted in the CFD simulations results.

4. CONCLUSION We used CFD simulations to investigate the effects of different meltblowing die configurations on the air flow field of the MB process. In particular, the effects of air constrictors of different widths, lengths, and angles were examined. The main goal behind generating these configurations was to identify designs that would maintain high centerline air velocity below the polymer injection and keep polymer temperature around the melting point near the die face, so as to achieve higher fiber attenuation. Simulation results showed that vertical air constrictors with width between 20 and 30 mm and length between 10 and 20 mm would result in both higher maximum centerline air velocity and higher maximum centerline air temperature. In addition, both maxima would be maintained for a longer distance (10−15 mm) below the die for some configurations, resulting in higher fiber attenuation, vis a vis smaller fiber diameter, which is highly desirable. Simulation results also revealed higher turbulent kinetic energy. Air constrictors with width less than 10 mm and length greater than than 20 mm may cause fiber roping during the web formation. Experiments conducted with air constrictors attached to an existing MB die produced webs with 20−40% smaller fiber diameter with air constrictor widths of 14 and 20 mm. Meltblown media produced using these dies showed 10− 30% reduction in mean pore size and enhanced filtration quality factor by 10−50% depending on operating conditions. 2057

DOI: 10.1021/acs.iecr.5b04020 Ind. Eng. Chem. Res. 2016, 55, 2049−2058

Article

Industrial & Engineering Chemistry Research (16) Zhou, C.; Tan, D. H.; Janakiraman, A. P.; Kumar, S. Modeling the Melt Blowing of Viscoelastic Materials. Chem. Eng. Sci. 2011, 66, 4172. (17) Hassan, M. A.; Khan, S. A.; Pourdeyhimi, B. Fabrication of micro-meltblown filtration media using parallel plate die design. J. Appl. Polym. Sci. 2016, 133 (7), 42998. (18) Uyttendaele, M. A. J.; Shambaugh, R. L. Melt Blowing: General Equation Development and Experimental Verification. AIChE J. 1990, 36, 175. (19) Rao, R. S.; Shambaugh, R. L. Vibration and Stability in the Melt Blowing Process. Ind. Eng. Chem. Res. 1993, 32, 3100. (20) Marla, V. T.; Shambaugh, R. L. Three-Dimensional Model of the Melt-Blowing Process. Ind. Eng. Chem. Res. 2003, 42, 6993. (21) Marla, V. T.; Shambaugh, R. L. Modeling of the Melt Blowing Performance of Slot Dies. Ind. Eng. Chem. Res. 2004, 43, 2789. (22) Harpham, A. S.; Shambaugh, R. L. Flow Field of Practical Dual Rectangular Jets. Ind. Eng. Chem. Res. 1996, 35, 3776. (23) Harpham, A. S.; Shambaugh, R. L. Velocity and Temperature Fields of Dual Rectangular Jets. Ind. Eng. Chem. Res. 1997, 36, 3937. (24) Begenir, A. Structure-Process-Property Relationships in Elastic Nonwovens Made From Multi-Block Elastomer. Ph.D. Dissertation, NC State University, Raleigh, NC, 2008. (25) Begenir, A.; Michielsen, S.; Pourdeyhimi, B. Meltblowing Thermoplastic Polyurethane and Polyether-Block-Amide Elastomers: Effect of Processing Conditions and Crystallization on Web Properties. Polym. Eng. Sci. 2009, 49 (7), 1340. (26) Roy, B. N. Fundamentals of Classical and Statistical Thermodynamics; John Wiley & Sons: UK, 2002. (27) Datta, S.; Fu Tse M.; Sahnoune A.; Sims C., Coffey J. Blends of low crystallinity, low molecular weight propylene copolymers and styrenic block copolymers. U.S. Patent US8052822B2, 2011. (28) Xie, S.; Zeng, Y. Online Measurement of Fiber Whipping in the Melt-Blowing Process. Ind. Eng. Chem. Res. 2013, 52, 2116. (29) Chung, C.; Kumar, S. Onset of Whipping in the Melt Blowing Process. J. Non-Newtonian Fluid Mech. 2013, 192, 37. (30) Jena, A.; Gupta, K. Characterization of Pore Structure of Filtration Media. Fluid/Part. Sep. J. 2002, 14 (3), 227. (31) Jena, A.; Gupta, K. Liquid Extrusion Techniques for Pore Structure. Evaluation of Nonwovens. Int. Nonwovens J. 2003, 45. (32) Automated Filter Tester Model TSI 3160, Operation and Service Manual, 2002. (33) Wang, J.; Kim, S. C.; Pui, D. Y. H. Figure of Merit of Composite Filter with Micrometer and Nanometer Fibers. Aerosol Sci. Technol. 2008, 42, 722. (34) Wang, J.; Kim, S. C.; Pui, D. Y. H. Investigation of the Figure of Merit for Filters with a Single Nanofiber Layer on a Substrate. J. Aerosol Sci. 2008, 39, 323.

2058

DOI: 10.1021/acs.iecr.5b04020 Ind. Eng. Chem. Res. 2016, 55, 2049−2058