485

JUNE 2005

485

Stress Relaxation of Tufted Carpets and Carpet Components: Analysis of the Tufted Carpet Structure KRISTIE J. PHILLIPS

AND

TUSHAR K. GHOSH,1

College of Textiles, North Carolina State University, Raleigh, North Carolina 27695, U.S.A.

DAVID A. DICKEY Department of Statistics, North Carolina State University, Raleigh, North Carolina 27695, U.S.A. ABSTRACT Dimensional stability of tufted carpets has been a continuing problem in the carpet industry for years. When a tufted carpet is installed by the stretch-in method, it experiences stress relaxation over time which can cause the carpet to buckle, wrinkle and become loose with the only option being a costly re-stretching of the carpet. Analysis of the various components of the tufted carpet composite structure was performed to identify the role each component plays in the phenomenon of stress relaxation. A biaxial loading system was used to test various samples of the primary backing alone, primary backing after tufting (with tufts), secondary backing alone, and the finished carpet after attaching the backings with various binder weights per area. The four variables under consideration included primary and secondary backing constructions, tufting density, and latex weight. A rheological model that includes representations of each component in the carpet structure was developed and will be presented in a following paper.

Introduction Since the late 1960s, when tufted carpets grew in popularity, the carpet industry has experienced problems with increases of carpet dimensions after installation, otherwise referred to as carpet buckling, wrinkling, rucking, or carpet growth. Carpets that are installed by the stretch-in method, the most popular method for installing residential carpets, experience stress relaxation over time where the tension in the carpet slowly dissipates. Then, when the carpet becomes looser, it is more susceptible to forces imposed by walking, which may result in buckling or wrinkling, making the carpet unsightly, creating trip hazards, accelerating carpet wear, and possibly causing the carpet to delaminate [12]. The only option for correcting a buckled carpet is restretching/re-installing, which is costly for retailers, installers, and consumers. Other factors that may aggravate carpet “growth” include changes in temperature and moisture conditions and incorrect or poor installation especially over large areas [3]. Since woven and knitted carpets seldom require restretching [9], the buckling problem seems to be linked primarily to the tufted construction. Carpet construc-

1 To whom correspondence [email protected]

should

be

addressed:

e-mail:

Textile Res. J. 75(6), 485– 491 (2005) DOI: 10.1177/0040517505053844

tional parameters that contribute to carpet buckling or growth were studied in the late 1960s and 1970s [1, 2, 11] with occasional studies performed since then [5, 7]. However, the majority of the published work has focused on various types of primary and secondary backings on a macroscopic scale, for example, comparing jute versus polypropylene primary and secondary backings, woven versus tufted carpets, and woven, nonwoven, foam or other types of secondary backings. However, no published work has been found that examines differences in backings on a smaller constructional scale such as increases in stitch densities. This research was designed to analyze the various components of the tufted carpet composite structure, identify the role each component plays in the phenomenon of stress relaxation, and thereby to identify the cause of carpet “growth” and buckling as far as it depends on carpet component properties.

Experimental After installation, a carpet is held stretched along both of its principal directions (lengthwise and widthwise) simultaneously over its lifetime. Therefore in this research, a specially designed biaxial testing apparatus, shown in Figure 1, was used to stretch a (45 cm ⫻ 45 © 2005 Sage Publications

www.sagepublications.com

486

TEXTILE RESEARCH JOURNAL

FIGURE 1. Experimental set-up.

cm) square sample in both directions and hold the sample stretched for a period of time. Two load cells measured the load values in the warp (A) and filling (B) directions and two precision dial displacement gauges measured the elongation of the sample in each direction. Deformation of the sample was performed manually. A computerbased data acquisition system, with LabVIEW® software [13], was used to enable continuous monitoring of stress levels and generate stress relaxation data. Test samples were stretched in both warp and fill directions to exactly 1% strain and the load values were recorded continuously for 20 hours. A value of 1% was taken to be the typical strain value that a properly installed carpet experiences, in accordance with the Carpet and Rug Institute’s [4] recommendation of 1-1.5% stretch for carpet installation and considering that, in practice, most installers do not exceed 1% [6].

Four factors were chosen to represent the tufted carpet structure: primary backing construction, secondary backing construction, tufting density, and coating weight per unit area. Details of sample constructions are presented in Table I (components) and Table II (finished carpets). Each component of the carpet as well as the finished carpet samples were tested with three repetitions including the primary backing (alone, before tufting), primary backing after tufting (with tufts), secondary backing alone, and the finished carpet after attaching the secondary with various weights of latex. Carpet samples in this study were produced by a carpet manufacturer with commercial tufting equipment. Choice of test samples was therefore limited by what could be produced by the carpet manufacturer without causing unacceptable disruptions in production. For this reason, warp density in primary and secondary backings as well as latex composition were held constant. The primary backings were made with polypropylene tape yarns in both warp and filling directions while the secondary backings consisted of a leno weave using tape yarns in the warp, similar to the primaries, but with open-end spun polypropylene yarns in the filling. Finished carpet samples, shown in Table II, consisted of samples A-, B-, and C-Lo and -Hi, which focused on the effects of secondary backing construction and coating weight while holding tufting pattern and primary backing construction constant, and samples D, E, and F, which varied the primary backing and tufting density while coating weight and secondary backing construction were held constant. Due to difficulties in controlling the latex weight applied to the samples during production, the actual latex weights were not as precise as desired.

TABLE I. Carpet components evaluated. Warp Sample Name

Ends (cm)

Yarn mass linear densitya

Tape width

Picks (cm)

Yarn mass linear densitya

9.4 9.4

492 den. 492 den.

1.25 mm 1.25 mm

4.3 7.1

1077 den. 1062 den.

2.5 mm 2.5 mm

6.3 7.1

492 den. 471 den.

1.25 mm 1.25 mm

2.0 4.3

1749 den. 1668 den.

spun yarn spun yarn

Primary backings: 11 picks/inch (ppi) 18 picks/inch (ppi) Secondary backings: 16 ends/inch (epi) ⫻ 5 ppi 18 ends/inch (epi) ⫻ 11 ppi Tufted primaries: (Pr ⫽ primary backing ppi; spi ⫽ tuft density stitches/inch) 11 Pr, 8 spi 11 Pr, 12 spi 18 Pr, 8 spi 18 Pr, 12 spi a

Fill

Primary backing used (ends/ cm ⫻ picks/cm) 9.4 9.4 9.4 9.4

As measured by unraveling the fabric.

⫻ ⫻ ⫻ ⫻

4.3 4.3 7.1 7.1

(24 (24 (24 (24

epi epi epi epi

⫻ ⫻ ⫻ ⫻

11 11 18 18

ppi) ppi) ppi) ppi)

Tuft density (stitches/cm) 3.1 4.7 3.1 4.7

(8 spi) (12 spi) (8 spi) (12 spi)

Tape width

JUNE 2005

487 TABLE II. Finished carpets evaluated. “Lo”and “Hi”designations refer to latex content. Primary backing constructiona

Secondary backing constructiona

Name

ends/cm

picks/cm

ends/cm

picks/cm

Tufting density @ 5/32 gauge (stitches/cm)

A-Lo B-Lo C-Lo A-Hi B-Hi C-Hi D E F

9.4 9.4 9.4 9.4 9.4 9.4 9.4 9.4 9.4

4.3 4.3 4.3 4.3 4.3 4.3 7.1 7.1 4.3

6.3 6.3 7.1 6.3 6.3 7.1 6.3 6.3 6.3

2.0 3.1 4.3 2.0 3.1 4.3 2.0 2.0 2.0

3.1 3.1 3.1 3.1 3.1 3.1 3.1 4.7 4.7

Actual latex contentb (g/m2)

Carpet weightc (g/m2)

746 610 712 882 848 848 814 814 882

1702 1611 1733 1838 1831 1865 1824 2119 2140

a b For the primary and secondary backings, yarn mass linear densities can be found in Table 1. Actual latex amount (in g/m2) was calculated c by subtracting the weights of the tufted primary and secondary backing from the average total carpet weight. Average weight for the three carpet samples tested (three repetitions).

Results and Discussion Typical stress-relaxation curves obtained in warp and fill directions for one of the carpets and its components are shown in Figures 2 and 3. Each data point in the plot represents an average load value of three tests at that time interval. As shown in the figures, the load sustained by the finished carpet is significantly higher than for the individual components. This indicates that the layer of latex used to adhere the carpet components together plays a significant role in the amount of load the carpet can support. ANALYSIS

OF THE

STRESS RELAXATION BEHAVIOR

The data from the biaxial tester were initially obtained as load versus time curves, containing more than 800 data points, for the 20-h testing period. Four characteristic parameter values were extracted from the data in order to represent each sample, including:

FIGURE 2. Stress relaxation curve (warp direction) of finished carpet sample D, plotted with its components, tufted primary (18 ppi primary, 8 spi tuft density) and secondary backing (16 epi ⫻ 5 ppi) (see Table I for sample details).

(1) initial load: the highest or peak load experienced by the sample immediately after it was stretched to 1%; (2) residual load: the residual load on the carpet sample after being stretched to 1% in both directions for 20 hours; (3) percentage retained load: the residual load after 20 hours expressed as a percentage of the initial or peak load; and, (4) residual rate of load decay: an estimate of the rate of load decay after 20 hours by taking the slope of the last part of the stress-relaxation curves using a local linear regression. An analysis of variance (ANOVA) was performed for each of six dependent variables, including the residual load, percentage retained load, and residual rate of load decay in both the warp and fill directions, on each of four data sets, including primary backings alone, secondary backings alone, tufted primaries, and finished carpets. For the finished carpets, linear regression analysis was

488

TEXTILE RESEARCH JOURNAL

FIGURE 3. Stress relaxation curve (fill direction) of finished carpet sample D, plotted with its components, tufted primary (18 ppi primary, 8 spi tuft density) and secondary backing (16 epi ⫻ 5 ppi) (see Table I for sample details).

used to isolate the carpet components that had significant effects on the dependent variables. PRIMARY

AND

SECONDARY BACKINGS

A summary of the results for the primary backing fabrics is presented in Table III. As shown in the table, the 18 ppi (7.1 picks/cm) primary retained 9 –10% more stress and maintained nearly twice the actual load as the 11 ppi (4.3 picks/cm) in both warp and fill directions after 20 hours. However, the only dependent variable to show a statistically significant difference between the 11 and 18 ppi primaries was the residual load in the fill direction, although some other p-values were close to 0.05. Apparently, more fill yarns in the 18 ppi fabric resulted in higher residual loads and retained load values in the fill direction. No statistically significant difference was found in the warp direction (at 95% confidence), which is understandable since the warp thread density in the two primary backings was the same (9.4 ends/cm or 24 ends/inch).

TABLE III. Results of the statistical analysis of the biaxial test data for the 11 ppi and 18 ppi primary backings when tested alone. (p- values ⬍ 0.05 indicate statistical significance at 95% confidence. Four (4) error degrees of freedom.) Sample means: primaries Dependent variable

11 ppi

18 ppi

p-value (for difference)

Residual load1, warp Residual load1, fill Retained stress, warp Retained stress, fill Rate of load decay, warp Rate of load decay, fill

1.90 kg 3.13 kg 17.93% 16.60% ⫺0.6986 ⫺0.9203

3.51 kg 6.09 kg 26.33% 26.83% ⫺0.7850 ⫺1.4795

0.0607 0.0195 0.0917 0.0534 0.5050 0.0888

1

Per 45 cm sample.

For the secondary backings tested alone, most of the stress relaxation data did not show any statistically significant differences. This is probably due to the very low load values experienced by the secondary backings initially and after 20 hours. The secondary backings were difficult to test because of their thin, net-like structure which allowed the yarns to move more freely, dissipating the stress quickly. The biaxial tests performed at 1% strain apparently were not deforming the yarns themselves but were simply moving the yarns within the fabric structure. There was a significantly higher residual load in the warp direction than in the fill, which is undoubtedly due to the spun yarn structure of the filling yarns. TUFTED PRIMARIES:

THE

EFFECT

OF

TUFT DENSITY

The four tufted primaries tested were combinations of the 11 and 18 ppi primary backings and the 3.1 and 4.7 stitches per cm (8 and 12 spi) tufting densities. Experimental data is summarized in Table IV. As shown in Table IV, in general, the residual load and retained load (%) values for the tufted primaries in the fill direction are two to three times higher than the untufted primary backings. This was verified statistically using two linear contrasts to compare the primaries alone to the corresponding tufted primaries: (1) 11 ppi primary versus 11Pr8spi and 11Pr12spi tufted primaries and (2) 18 ppi primary versus 18Pr8spi and 18Pr12spi. For the residual load in the fill direction, equality was rejected with p-values of 0.0006 and ⬍0.0001 for linear contrasts (1) and (2), respectively (with 12 error degrees of freedom (df)). For the retained load (%) in the fill, equality was rejected with p-values of ⬍0.0001 for both linear contrasts (with 12 error df). At first, this result does not make sense, since previous studies have shown that the

JUNE 2005

489 TABLE IV. Sample means (average of three repetitions) for the primaries and tufted primaries tested. Initial load1 (kg)

Residual load1 (kg)

Retained stress (%)

Residual rate of load decay

Sample

Warp

Fill

Warp

Fill

Warp

Fill

Warp

Fill

11 ppi primary 11Pr 8 spi (tufted) 11Pr 12 spi (tufted) 18 ppi primary 18Pr 8 spi (tufted) 18Pr 12 spi (tufted)

10.6 6.6 8.8 13.2 10.5 9.3

18.7 17.6 16.9 22.9 34.8 29.8

1.9 2.4 3.4 3.5 4.2 3.6

3.1 7.4 7.4 6.1 15.0 12.5

17.9 36.1 39.2 26.3 40.0 38.5

16.6 42.4 43.9 26.8 43.0 41.8

⫺0.6986 ⫺0.7861 ⫺0.9589 ⫺0.7850 ⫺1.0394 ⫺1.0733

⫺0.9203 ⫺1.9696 ⫺2.1407 ⫺1.4795 ⫺3.5921 ⫺3.4468

1

Per 45 cm sample.

primary backing loses 30 – 60% of its strength and other properties suffer after tufting [2], and it was expected that the tufted primaries would show faster rates of stress decay and lower amounts of retained load. However, the process of tufting forces yarns (tufts) through the primary backing, including through the interstices between the warp and fill yarns as well as penetrating the yarns themselves. Consequently, in the tufted primary, yarns in the backing have less freedom of movement within the structure and cannot dissipate stress as quickly. A third linear contrast was used to compare tufted primaries with the 11 ppi primary backing to those with the 18 ppi primary (11Pr8spi and 11Pr12spi versus 18Pr8spi and 18Pr12spi). A significant difference was found for the residual load in both warp and fill directions with p-values of 0.0030 and ⬍0.0001, respectively (with 8 error df). Another observation, which was not verified statistically, is that the residual load in the fill direction is consistently higher than in the warp across all of the backing/tufting combinations. This is most likely due to the difference in crimp levels between the warp and fill yarns. The fill yarns have less crimp, about 0.5%, compared with the warp yarns, at about 2% [6]. Therefore, when stretched 1% during testing, the fill yarns are un-crimped and stretched to a greater degree than the warp, resulting in higher levels of load.

The tufting density was found to have no clear effect on the residual load or retained load (%). This was again verified with a linear contrast comparing tufted primaries with 8 spi versus 12 spi. None of the p-values obtained were lower than 0.1174 (with 12 error df). BEHAVIOR

OF

FINISHED CARPET

Analysis of the finished carpet data is presented in Table V. Interestingly, the results for the finished carpets with 11 and 18 ppi primary backings followed the results reported for primary backings earlier. In the fill direction, the residual loads were significantly higher for finished carpets with the 18 ppi primary, indicating that the effect of additional fill yarns in the primary is not obscured when the carpet is finished. Analysis of the finished carpet data was more complicated than analysis of the components alone since there were four factors that came into play simultaneously: the primary and secondary backing, the tufting density, and the latex content. Linear regressions were used to examine all four factors/components (using actual magnitudes of the variables) as well as two possible interaction effects between the factors (primary ⫻ tuft density and secondary ⫻ latex). Dependent variables included the residual load and retained load (%) in the warp and fill directions. The initial linear regression models were re-

TABLE V. Summary of the residual load and retained stress values for the finished carpet samples tested (average of three repetitions). Residual load1

Percent (%) retained stress

Residual rate of load decay

Sample

Warp (kg)

Fill (kg)

Warp (%)

Fill (%)

Warp

Fill

A-Lo (11 Pr, 8 spi, 16 ⫻ 5) B-Lo (11 Pr, 8 spi, 16 ⫻ 8) C-Lo (11 Pr, 8 spi, 18 ⫻ 11) A-Hi (11 Pr, 8 spi, 16 ⫻ 5) B-Hi (11 Pr, 8 spi, 16 ⫻ 8) C-Hi (11 Pr, 8 spi, 18 ⫻ 11) D (18 Pr, 8 spi, 16 ⫻ 5) E (18 Pr, 12 spi, 16 ⫻ 5) F (11 Pr, 12 spi, 16 ⫻ 5)

19.0 18.3 21.4 13.7 16.5 20.5 19.7 17.7 18.9

25.5 21.8 24.2 22.9 20.5 21.9 29.8 29.9 21.7

29.6 30.0 30.7 25.2 27.4 30.4 27.0 28.1 27.8

33.8 31.9 33.0 30.0 30.0 32.0 31.8 32.5 29.0

⫺4.3938 ⫺4.1819 ⫺5.8640 ⫺3.2337 ⫺3.4572 ⫺4.5234 ⫺5.2639 ⫺3.4660 ⫺4.4182

⫺5.3557 ⫺4.9190 ⫺6.6902 ⫺5.1316 ⫺4.1501 ⫺4.1355 ⫺7.4781 ⫺5.2077 ⫺4.6001

1

Per 45 cm sample.

490

TEXTILE RESEARCH JOURNAL

duced by removing the insignificant factors one by one, starting with insignificant interaction effects, until only the significant factors were left in the model. For the residual load of the finished carpets in the fill direction, the primary backing construction was the single most significant contributing factor in the reduced model, overpowering the effects of the other factors (with p-value ⬍ 0.0001, R2 value 0.5981, and 28 error df in the reduced model). For the residual load in the warp direction, most of the factors tested were found to significantly contribute to the model, indicating that no one factor was the most critical (95% confidence, R2 0.4899, 23 error df). The latex content was found to be significant in three out of four dependent variables, including residual load in the warp and retained load (%) in the warp and fill (p-values 0.0071, 0.0020, and 0.0066, respectively). The effect of latex content on carpet performance is discussed further in the following section. COMMENTS

ON

LATEX CONTENT

Two levels of latex content were examined in this study, “Lo” (712–746 g/m2 or 21–22 oz/yd2) and “Hi” (848 – 882 g/m2 or 25–26 oz/yd2). While the latex content was expected to be a significant factor, after examination of the data in Table V, it appeared that the high latex levels had lower residual and retained load (%) values than the low latex levels in both warp and fill directions. This result seems, at first, counterintuitive. It would generally be expected that higher latex levels would bond the carpet backings more securely together, restricting their movement and their ability to dissipate stress, and therefore having higher residual load values. However, the latex used in carpets generally contains large amounts of filler [8, 10], so increasing the amount of latex and filler may result in unexpected effects. Furthermore, the “Lo” latex level is the typical level used for this type of carpet, so in attempting to obtain the

higher latex level, the thicker, heavier coating may have been under-cured. LONG TERM STRESS RELAXATION TESTS A limited number of samples were tested for longer periods of time than the initial 20-hour period. Figure 4 shows the long-term residual load data as a function of time for one of the samples. After the initial 20 hours were up, these samples were left on the biaxial tester and load measurements were taken manually after additional time had elapsed. After 600 hours, the four samples had retained load (%) values ranging from 18.7 to 20.7% in the warp and 17.5 to 24.8% in the fill. The two samples that were tested for longer than 600 hours continued to dissipate stress slowly over extended time periods.

Conclusions From the results presented here, it is clear that primary backing construction plays a very important role in determining the stress relaxation behavior of carpets. It is also apparent from the data that latex as a binder provides significant synergy to the carpet system; as a result the force required for stretching a finished carpet to 1% extension is much higher than the sum of the forces required to stretch the individual components by the same amount. At the same time, higher latex content showed inferior stress relaxation performance. However, before concluding that lower latex levels perform better, other factors affecting buckling need to be considered, including various kinds of loading such as the forces imposed by walking. In addition, other performance properties were not tested in this study, such as tuftbind. Obviously the role of latex needs to be investigated further. For the primary and secondary backings tested alone at 1% strain, the residual loads have generally leveled off to their limiting values after the 20-hour

FIGURE 4. Long-term stress relaxation curve of finished carpet sample E (kg per 45-cm sample) in both warp and fill directions.

JUNE 2005

491

testing period; however, finished carpet samples did not reach their limiting residual load values during the 20hour testing period, but took significantly longer. ACKNOWLEDGEMENTS The authors would like to express their appreciation to Hugh Gardner, Carroll Yawn, and Tom Baker from BP-Amoco for their help in obtaining test samples and for making the biaxial tester available to us for an extended length of time.

Literature Cited 1. British Standards Institution. “Methods of Test for the Dimensional Stability of Textile Floor Coverings. Part I. Determination of Extension under Mechanical Action.” BS 4682: Part 1: 1971. British Standards Institution, London, 1971. 2. Gentry, D. R., The Dimensional Stability of Carpets in Installations. Part I: Stability to Mechanical Actions, Textile Res. J. 47(7), 459 – 463, (1977). 3. Goodman, P. J., Down to Basics, Carpets & Floorcoverings Review -Supplement 10 –30, (1986).

4. “Guidance for Restretching to Remove Buckles, Wrinkles, and Bubbles.” Technical Bulletin. The Carpet and Rug Institute. Dalton, Georgia, 1998. 5. Monson, J. A., A Study of the Relationship Between Carpet Mechanical Properties and Increases in Carpet Dimensions Following Installation. Ph.D. Thesis, Clemson University, 1982. 6. Private communications with H. Gardner, C. Yawn, and T. Baker, BP-Amoco. 7. Schaff, A. J., and Ogale, A. A., Tensile Viscoelastic Properties of Spunbonded Nonwoven Polypropylene Backing. Textile Res. J. 61(7), 386 –392 (1991). 8. Scott, R. L., Carpet Laminating. J. Coated Fabrics 19, 35–52 (1989). 9. Smith, G. W., Backings are the Foundation, Not Secondary. Carpet & Rug Industry, 24(9), 28 –29 (1996). 10. Stamper, K., An Overview of Carpet Laminates. Carpet Manufacturing Conference. Northwest Georgia Trade & Convention Center, Dalton, Georgia, Aug. 15–16, 2000. 11. Sudnik, Z. M., Dimensional Stability of Carpets: Rucking of Carpets in Use. Textile Institute and Industry. 7, 278 – 281 (1969). 12. www.carpet-rug.com (The Carpet and Rug Institute). 13. www.ni.com (National Instruments).