Effect of Bulk Density on the Acoustic Performance of Thermally Bonded Nonwovens

Transcription

1 Effect of Bulk Density on the Acoustic Performance of Thermally Bonded Nonwovens Wenbin Zhu 1, Vidya Nandikolla 2, Brian George 1 1 Philadelphia University, Philadelphia, PA UNITED STATES 2 California State University, Northridge, CA UNITED STATES ABSTRACT The effect of different fiber blend ratios and bulk densities at similar thicknesses on air permeability and sound absorption coefficient was investigated. The raw materials used were cotton, polyester, and bicomponent fibers to make acoustic nonwovens through the air-laid and thermal bonding processes. A uniform thermal-conductivity box was designed in order to make thermally bonded nonwovens with fixed thicknesses. The experimental results depict that the air flow resistance of three-layer nonwoven was kpa s/m, which was about four times greater than one-layer of 0.12 kpa s/m. Sound absorption coefficient of 20% polyester-60% cotton-20% bicomponent nonwoven with lower bulk density was greater than the 60% polyester-20% cotton-20% bicomponent nonwoven. The sound absorption coefficient varied based on which fabric side faced the testing apparatus. INTRODUCTION Nonwoven materials are increasingly used for many applications. The automotive industry has incorporated nonwoven textile materials for sound proofing applications such as hood linings, acoustic absorption pads, and trunk trim. Most research on nonwoven materials has focused on factors such as fiber size, air flow resistance, porosity, thickness, and density for the application of acoustic absorption and insulation due to its tremendous potential [1]. The sound absorption and insulation performance of fibrous materials depends on its composition and structure. In general, the thicker the material and the higher the porosity in structure, the better the absorption. Correspondence to: Vidya Nandikolla vidya.nandikolla@csun.edu Research has shown that nonwoven composites with cotton as a surface layer has higher sound absorption coefficients than glass fiber in the frequency range of 100 to 6400 Hz. Also, carbonization and activation of the cotton nonwoven improves the sound absorption performance significantly [3]. The internal structures of other natural fibers such as cashmere, goose down, and kapok show an influence on acoustic performance. The acoustic absorption of the goose down and kapok are much greater than that of cashmere and acrylic fibers. For same type of fiber at certain sound frequencies, the sound absorption coefficient reduces along with the declining mass density [4,5]. The sound absorption properties for nonwovens composed of different fibers have been investigated by Kucuk and Korkmaz. Nonwovens with 70 percent cotton and 30 percent polyester showed the best sound absorption coefficient for higher frequency ranges. Adding acrylic and polypropylene materials into the cotton and polyester blend increased the sound absorption coefficient of the nonwoven in the lower frequency ranges [6]. The acoustical insulation and absorption properties of nonwoven fabrics depend on the fiber geometry and fiber arrangement within the fabric structure. Tascan et al. studied how varying fiber cross-sectional shape, denier, and fabric density affected the acoustic performance of nonwovens [7]. Another important parameter affecting acoustic performance is the bulk density of nonwoven fabrics that makes an influence on the volume of the fabric [5]. In general, textile materials are designed and integrated to increase the sound absorption coefficient. The porosity of materials increases the sound absorption coefficient as the sound wave contacts the fiber surface and dissipates energy [2]. For this research, the effect of different fiber blend ratios and bulk densities at similar thickness on air permeability and sound absorption coefficient was investigated. A thermal-conductivity box was designed to make uniform thermally bonded Journal of Engineered Fibers and Fabrics 39

2 nonwovens with fixed thickness. Cotton, polyester, and co-polyester bi-component fibers were used as the raw materials to make acoustic nonwovens through the air-laid and thermal bonding processes. The acoustic performance was measured according to the relevant standards of ASTM. EXPERIMENTAL WORK Fabric Production Cotton, polyester, and bicomponent fibers were blended together at two ratios to make acoustic nonwovens with different bulk densities. Air-laid and thermal bonding methods were essential steps for forming and bonding the web in the manufacturing process. Since high bulk density led to thicker nonwoven when one, two, or three layers were made, the compression process played an important role in order to acquire similar thickness of nonwovens at varying bulk densities. Three different loads were applied to compress the layers in a thermal-conductivity box so as to maintain the same thickness. After hand blending, the fibers were processed with a Rando Webber air laid unit. Both sets of ratios were run through the machine three times in order to better blend the fibers together. After the third run the webs were wound up in paper. The webs were bonded using a Precision Gravity Convection Oven laboratory sized oven at a temperature of 190 for thirty minutes. Fabrics consisting of one, two, and three layers of web were made using this method. In this research the main goal was to compare the effect of bulk density on acoustic properties for samples of similar thickness values. Therefore, a thermal conductivity box (20cm x 20cm x 2cm) with nine holes (5cm in diameter) was designed to compress the nonwoven and make sure that the heat was uniformly conducted. The thermal conductivity box consisted of an aluminum foil body and steel cover. Figure 1 shows the thermal-conductivity box and its basic working principle. Three types of fibers were hand blended at two different ratios. One blend sample was 60% polyester- 20% cotton-20% bicomponent fibers while the other one was 20% polyester-60% cotton-20% bicomponent fibers. Cotton fibers with an average size of 3 denier, 1.5 denier polyester fibers from Wellman and 3.3 denier bicomponent fibers from Hoechst Celanese were utilized. Table I details the process parameters for six samples at different blend ratios and varying basis weight. TABLE I. Physical Parameters for Each Sample. Sample Blend Ratio (%) Basis Weight (gsm) Thickness (mm) Bulk Density (kg/m3) NP1 6P/2C/2B NP2 6P/2C/2B NP3 6P/2C/2B NC1 2P/6C/2B NC2 2P/6C/2B NC3 2P/6C/2B From Table I, NP1 was one layer nonwoven with polyester fibers predominating; NC3 was the three layers nonwoven in which cotton fibers were the majority. Meanwhile, 6P/2C/2B means 60% polyester, 20% cotton, 20% bicomponent fibers. The descriptions of NP2, NP3, NC1, NC2 and 2P/6C/2B samples can be understood similarly. FIGURE 1. Thermal-conductivity box used for compression. The samples were cut into the given size and laid in the box. At the same time, a specified load was applied to the top surface of the nonwoven, and the box was placed in the oven to melt the bicomponent fibers and bond the fabric. The heat was conducted to the sample through the bottom holes and the walls of the thermal conductivity box. Before the process of thermal bonding started, 430g and 610g weights were applied to the metal cover to compress two and three layers respectively in order to acquire similar thicknesses to the one layer fabrics. No load was applied to the single layer when it was placed in the oven. Journal of Engineered Fibers and Fabrics 40

3 Measurement Apparatus The thicknesses of nonwovens were measured according to the test standard of ASTM D5729. A Randall & Stickney thickness gauge was used to measure the thickness of the samples. The average measured thickness of six samples was approximately ten millimeters, because of the applied specific load and high loft surface. The initial apparent thickness of most samples was close to twenty millimeters. An automatic air-permeability tester (Kawabata Evaluation Systems KES-F8-AP1) was used to measure the flow resistivity according to ASTM D737. The rate of airflow passing perpendicularly through a known area of fabric was adjusted to obtain a prescribed air pressure differential between the two fabric surfaces. The described apparatus was used to determine the rate of airflow and the air permeability of the fabric samples. A Bruel and Kjaer sound tester was used for testing the frequency range between 0 Hz to 6.4 khz, which was based on the standard of ASTM E This standard referred to a tube, noise generator, two microphones and a digital frequency analysis system, as shown in Figure 2. A small tube (29 mm diameter) was set up for testing the material sound absorption in the high frequency range. The normal incidence sound absorption coefficient was calculated. FIGURE 3. The average of thickness among six samples. Tukey's method (also called as Turkey-Kramer in the unbalanced case) used an ANOVA to create confidence intervals for all pairwise differences between factor level means controlling the family error rate to a specified level. A family error rate of 0.05 (equivalent to a 95% joint confidence level) was used. In this study, the NP group was made of 60% polyester, 20% cotton, and 20% bicomponent fibers, which included three samples with different bulk densities. The thickness was measured for five specimens from each sample. The ANOVA resulted in a P-value of 0.081, which was greater than 0.05, leading the conclusion that none of the sample thicknesses were different from the others. In the same way, the NC group which consisted of 20% polyester, 60% cotton, and 20% bi-component fibers had a P-value of 0.115, which means there was no difference among the samples data as seen in Table II. TABLE II. One-way ANOVA of NP and NC groups. NP versus Basis Weight FIGURE 2. Configuration for measuring sound absorption [8]. RESULTS AND DISCUSSIONS Thickness The difference of thickness among six samples can be seen in Figure 3. The thicknesses of samples showed little difference and the variation bar refers to a 95% confidence interval of error. The significant differences of these nonwoven samples were determined by using one-way ANOVA and Tukey s statistical methods. Source DF SS MS F P Weight Error Total NP versus Basis Weight Source DF SS MS F P Weight Error Total Minitab calculated the three individual confidence levels as percent in order to obtain the 95 percent joint confidence level. Using Tukey s method, the confidence intervals were examined to see if zero was included, suggesting a significant difference. The reference line at 0 illustrates how the wider Tukey confidence intervals could change the conclusions. Journal of Engineered Fibers and Fabrics 41

4 Confidence intervals that contained zero indicated no difference. Multiple comparison results further illustrated the conclusion that no significant difference existed among the thickness values of the samples. Tukey's test provided grouping information and two sets of multiple comparison confidence intervals for each group. The confidence intervals displayed the likely ranges for all the mean differences. NP1 thickness subtracted from the NP2, NP3 means: the first interval in the first set of the Tukey's output ( , , ) gave the confidence interval for the NP1 mean subtracted from the NP2 mean. For this set of comparisons, none of the means were statistically different because all of the confidence intervals included a value of zero. NP2 mean subtracted from the NP3 mean were not statistically different because the confidence interval for this combination of mean ( , , ) also included zero. In conclusion, with Tukey's method, the mean thickness for NP1, NP2 and NP3 did not appear to be statistically different. Similarly the results were obtained for the NC group. Therefore, the thicknesses from different samples were the same and ten millimeters are regarded as the thickness of each sample for comparison s sake. resistance. The samples from NP group have more fine polyester fibers in the nonwoven web. The multilayer nonwovens have a greater air flow resistance due to the more compact structure achieved by compressing multiple layers into the same volume as the one layer fabric, thereby increasing the bulk volume of the fabrics. Sound Absorption Coefficient In Figure 5 the sound absorption coefficient for NP and NC groups for different bulk densities is shown. The average absorption coefficient over the whole frequency range correlated well with the bulk density. Sound absorption coefficients of the nonwoven samples increased with the bulk density and frequency because of the decreasing porosity and increasing tortuosity. Sound absorption coefficients of NC1 and NC2 samples were greater than NP1 and NP2 respectively even though NC1 and NC2 samples had lower bulk densities. This may be due to the natural twisting on the cotton fiber surface along the axial direction increasing the contact area and tortuosity between fibers and sound waves. Thus, much sound energy may have been changed into vibration and frictional energy. Flow Resistance As shown in Figure 4, the airflow resistance for different levels of bulk density presented a significant difference between each other. FIGURE 5. Sound absorption coefficient of different samples. FIGURE 4. Flow resistance of sample at different bulk densities. The value of airflow resistance for NP3 sample was kpa s/m, which was about four times greater than the 0.12 kpa s/m of NP1. There are more fibers in the fabric cross section when the fibers have smaller denier, which creates more opportunities for air molecules to get in touch with fibers in the nonwoven structure, thereby increasing airflow The difference in sound absorption coefficients between NC1 and NP1 increased with the increasing frequency in the low to middle frequency region. It was a similar situation for the difference between NC2 and NP2 samples in sound absorption coefficients. There was no significant difference of sound absorption coefficient between NC1 and NP1 in the high frequency region. The reason was because sound absorption is a function of the wavelength and thickness of the material. Theoretically, 100% absorption was absorbed by a material that is half the Journal of Engineered Fibers and Fabrics 42

5 thickness of the wavelength. Research showed that 100% absorption by a material thickness was closer to 1/10th of the wavelength. Therefore; the thicker the material, the better the absorption. In the high frequency region corresponding to low wavelengths, the thickness of material played a more important role in sound absorption property than the bulk density [9]. From observations of the thick nonwoven samples, the fibers in the top layer were not bonded well compared to the fibers in the bottom layers, which caused the gradual increase in bulk density from the top to bottom surface, as shown in Figure 6. This is because the heat flow through the nonwoven sample was from the bottom to top during the thermal bonding process. The temperature at the top surface was lower than the bottom and thus more bonding occurred at the bottom of the fabric. In this study, the effects of different surfaces on the sound absorption coefficient using front and back surface towards the sound wave were measured. In Figure 7, the different sound absorption coefficients can be seen with the top and bottom of samples facing the sound wave in the impedance tube. NC3-3F is the third specimen of NC3 sample where the front surface facing towards the sound wave. FIGURE 7. Sound absorption coefficients for different surface. The sound absorption coefficient of NC1-3F was greater than the NC1-3B in the low-middle frequency region. The results show that the difference in the sound absorption coefficients between the two samples decreased with the increase in frequency in the high frequency region. For the specimen from NC2 and NC3 samples, a similar change trend was seen, as exhibited in Figure 7. The reason for this phenomenon is depicted in Figure 8. The front surface with a looser fiber structure seemed to be a more porous medium, allowing sound waves to enter the fabric rather than reflecting, whereas the back of the fabric (the bottom during the bonding process) had a more compact surface that created a barrier which prevented the sound waves from propagating into the material as easily. If the sound propagated from the front to the back surfaces, the countless holes and channels created more possibilities for the sound wave to interact with the fibers in the structure and sound energy was reduced by the friction and vibration, as seen in Figure 8 (a). FIGURE 6. Cross section of nonwoven sample. NC3-3B is the third specimen from NC3 sample where the back surface is towards the sound wave. The other samples can be read similarly. (a) (b) FIGURE 8. Sound reflection on front and back surface. Journal of Engineered Fibers and Fabrics 43

6 By contrast, when the back surface with higher bulk density and highly compact structure faced towards the sound source, as shown in Figure 8 (b), the results were slightly different. A greater amount of sound energy was reflected instead of propagating into the material, meaning more sound pressure signals was absorbed by the microphone rather than the fabric. This was the reason for the difference of sound absorption coefficients between the front and back surfaces. In order to find the value of sound absorption coefficients corresponding to certain density and frequency, two contour plots for NP and NC samples, were produced using Minitab. Figure 9 shows the contour plot of sound absorption coefficient for NP sample and Figure 10 for the NC sample. FIGURE 9. Contour plot of sound absorption coefficient for NP. Some differences are shown in the two contour plots described in Figures 9 and 10. The results show the maximum density of NP was close to 90 kg/m 3 and 60 kg/m 3 for NC sample. In the contour plot of NP sample, some big curves appeared at the density of 50 kg/m 3 that seems to be an abnormal phenomenon. This was due to the limitation of sample size and uneven data distribution of bulk density. From the original data, it was observed that the sound absorption coefficient at 49.3 kg/m 3 was greater than at 50.5 kg/m 3, which led to the big curve at around 50 kg/m 3. However, a perfect contour plot was seen in Figure 10 for NC sample due to the relatively uniform data distribution. In conclusion, the NC sample with 20% polyester, 60% cotton, 20% bicomponent fibers showed better sound absorption coefficient compared to NP with 60% polyester, 20% cotton, and 20% bicomponent fibers. There are other uncertain structural factors such as tortuosity and fiber shape factors, which may have affected the sound absorption coefficient even though airflow resistance played an important role in the acoustic performance. CONCLUSION A thermal-conductivity box was used to control the thickness of nonwovens at different bulk densities. The bulk density of the nonwoven was directly proportional to the sound frequency and sound absorption coefficient. The experimental results concluded the airflow resistance of NP samples was greater than NC samples. The difference in airflow resistance increased significantly with the increasing bulk density. Airflow resistance of NP3 sample showed kpa s/m, which was about four times greater than the NP1 of 0.12 kpa s/m. The sound absorption coefficient of NC1-3F with a less compact surface towards the sound source was slightly greater than that of NC1-3B with compact surface towards the sound source in the lower to middle frequency region. The difference of sound absorption coefficient between the two samples decreased with the increase of frequency in the high frequency region. FIGURE 10. Contour plot of sound absorption coefficient for NC. ACKNOWLEDGEMENT We appreciate the assistance of Eric Staudt and Janet Brady in evaluating the performance of these materials. REFERENCES [1] Seddeq H.S.; Factors Influencing Acoustic Performance of Sound Absorptive Materials; Australian Journal of Basic and Applied Science 2009, 3, [2] Shoshani Y.; Yakov Y.; Numerical Assessment of Maximal Absorption Coefficients for Nonwoven Fiberwebs; Applied Acoustics 2000, 59, Journal of Engineered Fibers and Fabrics 44

7 [3] Jiang N.; Chen J.Y.; Parikh D.V.; Acoustical Evaluation of Carbonized and Activated Cotton Nonwovens; Bioresource Technology 2009, 100, [4] Dweib M.A.; Hu B.; O Donnell A.; Shenton H.W.; Wool R.P.; All Natural Composite Sandwich Beams for Structural Applications; Composite Structure 2004, 63, [5] Yang S.; Yu W.D.; Pan N.; Investigation of the Sound-absorbing Behavior of Fiber Assemblies; Textile Research Journal 2011, 81, [6] Kucuk M.; Korkmaz Y.; The Effect of Physical Parameters on Sound Absorption Properties of Natural Fiber Mixed Nonwoven Composites; Textile Research Journal 2012, 82, [7] Tascan M.; Vaughn E.A.; Stevens K.A.; Brown P.J.; Effects of Total Surface Area and Fabric Density on the Acoustical Behavior of Traditional Thermal-bonded Highloft Nonwoven Fabrics; The Journal of The Textile Institute 2011, 102, [8] Shen D.H.; Wu C.M.; Du J.C.; Laboratory investigation of basic oxygen furnace slag for substitution of aggregate in porous asphalt mixture; Construction and Building Materials 2009, 23, [9] Staudt, E.; Personal Communication; AUTHORS ADDRESSES Wenbin Zhu Brian George, PhD Philadelphia University Engineering 4201 Henry Ave and School House Ln Philadelphia, PA UNITED STATES Vidya Nandikolla Nordhoff St Mechanical Engineering, CSUN Northridge, CA UNITED STATES Journal of Engineered Fibers and Fabrics 45