Abstract
An earlier study confirmed the influence of cotton fiber length characteristics on the High Volume Instrument™ (HVI) strength measurement and devised a quantitative correction factor to compensate for the effect. The current paper investigated the validity of two important assumptions utilized in the previous study. Firstly, single fiber testing confirmed that the particular sample preparation method used to generate samples of different fiber length characteristics from a common cotton sliver did not introduce any inherent damage to the fibers (and so this could not be the explanation for the observed trend in measured fiber strength as a function of fiber length). Secondly, the positioning of the jaws relative to the beard in the HVI strength measurement was explored. This positioning was found to be quite variable for replicate measurements on the same cotton being a function of the size of each individual beard. The average positioning between the different samples was found to be similar and this validated the assumption and approach used previously for deriving the correction factor for that particular sample set. Characterizing the position of the jaws was extended using a wider range of cotton samples. The HVI positioning algorithm appears to not simply be a function of the size of the beard (i.e. the ‘amount’ parameter), but is also dependent on fiber length characteristics. It was also observed that the reported HVI elongation values displayed both a significant bias due to fiber length and also a dependence on the size of individual beards tested.
Both length and strength are important cotton fiber quality parameters, having impacts on textile processing and product quality. As testing of individual fibers is time consuming and relatively expensive, in the commercial arena objective measurement of cotton fiber length and strength characteristics are commonly assessed using a fiber bundle test, for example ASTM Test Method D 4605-86. 1 The Uster High Volume Instrument™ (HVI) is commonly used for this purpose in many Western countries. In recent years this technology and instrument has also been adopted in China, an increasingly important cotton trading and textile manufacturing country.
Sample preparation for the HVI length and strength measurement involves the creation of a beard of aligned fibers by a comb with fibers being gripped at random positions along their length. A non-destructive optical technique is used to estimate the thickness of the beard as a function of position along the beard. Fiber length distribution information is then calculated from this data. Commonly reported length parameters are the Upper Half Mean Length (UHML) and the Uniformity Index (the ratio of the mean length to the UHML). 2 It is believed that the mechanical actions associated with both harvesting and ginning contribute to the wide range of individual fiber lengths in ginned cotton. A number of studies have found that the length distribution can be modeled as a mixture of two Weibull distributions.3–6
The fiber beard is also used for the strength measurement, following the length measurement. Using a gauge length of 3.175 mm (⅛ inch), two sets of jaws clamp the beard at a position towards its base. The breaking force is measured directly and normalized using an estimation of the mass of fiber from the optical sensor (in combination with the Micronaire value) to give the strength in cN/tex. An early version of the HVI software used a fixed distance from the comb for the position and break of the fibers; however, some time ago this was replaced by a new proprietary algorithm where this position is no longer fixed, but rather it changes to ensure that different fiber beards break at a fixed amount of cotton. 7
Fryer et al.8,9 studied the effect of varying the jaw placement on the HVI strength measurement and demonstrated that placing the jaws at different span lengths may introduce errors into strength measurements by changing the subset of fibers in the test.
Gourlot and colleagues10,11 undertook a small study into the effect of fiber length on the HVI strength measurement. Using a cotton sliver, four samples of varying lengths were formed by repetitive cutting of the sliver at different lengths. It was observed that the HVI fiber strength measurement changed significantly as a function of the different cutting lengths. Gourlot’s proposed explanation of this observation is that it is ‘due to a change in the mass of the broken fiber according to the length of the tested fiber’. 11
Recently, Naylor 7 repeated and extended Gourlot’s preliminary study to explore more fully the influence of fiber length characteristics on the HVI strength measurement. Again using a set of samples formed by cutting a parent DII sliver at different lengths, the HVI strength values exhibited a consistent trend as a function of the fiber length properties, with the HVI strength values increasing as the fiber length increased. Under the assumption that the sample preparation technique (i.e. cutting the parent sliver) would not have damaged the cotton, the true fiber strength values should be independent from cut length. In an endeavor to explain these unexpected trends, Naylor 7 noted that not all fibers will span the distance between the jaws, and explored a working hypothesis that the HVI estimates the total mass of fiber at a position between the jaws, which may be an overestimate of the true mass, which contributes to the breaking force. A quantitative model was developed to correct for this overestimation based on the shape of the Fibrogram (a particular form of the length distribution 2 ). It was found that the required correction factor is a function of the mean fiber length and various geometrical parameters of the HVI instrument (i.e. (a) the distance from the base of the beard to the inside edge of the clamp nearest the base of the jaws, and (b) the actual position of the mass determination relative to the clamping jaws). Importantly, this correction factor is independent of the shape of the fiber length distribution. With a further simplifying assumption that the position of the jaws along the beard is fixed, application of this correction factor approach was able to remove the effect of fiber length on the corrected strength values.
The current paper extends the previous work of Naylor. 7 In particular, (a) it examines the assumption that the sample preparation used in the previous study does not introduce any inherent damage to the fibers, which may explain the observed trend in measured fiber strength as a function of fiber length and (b) explores the positioning of the jaws relative to the beard in the HVI strength measurement.
Experimental details
Summary of previously published High Volume Instrument™ data 7 for the primary set of cotton samples presented as mean (standard error of the mean) with n = 16
SFI: short fiber index; UHML: Upper Half Mean Length.
Summary of the assigned values for the Universal Calibration cottons
UHML: Upper Half Mean Length.
Single fiber breaking force measurements were undertaken using the FAVIMAT instrument (Textechno Herbert Stein GmbH, Mönchengladbach, Germany) at a gauge length of 3.175 mm (⅛ inch) with an extension rate of 100%/min. Approximately 300 individual fibers per sample were tested.
In the original study, 7 fiber length and strength measurements were measured using an HVI 1000 instrument (Uster Technologies Inc, Knoxville, TN, USA) at a commercial classing facility in Australia. In the present study these measurements were repeated using the HVI 1000 instrument at the United States Department of Agriculture-Agricultural Research Service (USDA ARS) research laboratory (n = 20 for each sample). The normal routine procedure was used with the additional step that the instrument was interrupted immediately following the strength measurement and prior to the automatic ejecting of the broken portion of the beard from the comb. At this point three different distances along the remaining portion of the beard from the base of the comb (described and defined in the Results and discussion section below) were manually measured using digital calipers. The HVI parameter labeled ‘amount’, as well as the usual tensile data, was recorded for each test. The HVI ‘amount’ parameter is not precisely defined by the instrument manufacturer, but presumably is an estimate of the mass of the beard.
All fiber measurements (FAVIMAT and HVI) were done in a controlled environment (20 ± 1℃ , 65 ± 2% relative humidity (RH)) after conditioning samples for at least 24 hours.
Results and discussion
Single fiber measurements
Figure 1 contains the average breaking force results obtained using the FAVIMAT instrument. No trend is apparent in the average breaking force between the different samples, and an analysis of variance (ANOVA) confirms that there is no evidence of any statistically significant differences between sample differences (Table 3). Also no statistically significant differences between samples were observed in the mean fiber extension to break values, as illustrated in Table 4. Thus it is reasonable to conclude from these single fiber measurements that the sample preparation cutting mechanism does not introduce any significant damage to the fiber’s inherent strength or elongation characteristics.
Average single fiber breaking force values of cut samples with different fiber length characteristics. The error bars represent ± the standard error of the mean (n = 300 or greater). Analysis of variance summary of between-sample mean single fiber breaking force values (cN) Analysis of variance summary of between-sample mean single fiber extension to break values (%)
The relatively short gauge length of 3.175 mm (⅛ inch) was chosen to be compatible with the HVI gauge length, and also to avoid the introduction of any sampling bias that may occur with a longer gauge length, given the relatively short average length of some of the samples. In the case of longer fibers and gauge lengths, the FAVIMAT has the capability of undertaking a fiber linear density determination using the vibration/resonant frequency technique and using this to normalize the breaking force from each fiber to return a breaking strength value. In the current set of measurements the determination of fiber linear density was not possible with the relatively short gauge lengths and so only breaking force is reported. (Further, it is unclear if the algorithms used internally by the instrument for calculating fiber linear density are applicable to a cotton fiber, given its unusual cross-sectional shape and the non-uniformity of shape and dimensions along its length.) Given that each of the five different samples was prepared (i.e. cut) from the same parent cotton sliver, it is reasonable to assume that the average of fiber linear density values will be constant between samples. This, combined with the relatively large number of individual fibers measured per sample, enables a direct comparison of the average breaking force values reported in Figure 1.
HVI measurements
Figures 2 and 3 illustrate the very good agreement between the HVI major length characteristics measured at the USDA with previously published values
7
(the original measurements were undertaken in a commercial cotton classing house). The two sets of short fiber index (SFI) values also correlate well, as shown in Figure 4. The HVI SFI values are known to exhibit considerable machine-to-machine variability. Hence the non-unity slope value observed for the best fit line in Figure 4 is not unexpected.
Comparison of the High Volume Instrument™ (HVI) Upper Half Mean Length values from the current study and previously reported values.
7
Comparison of the High Volume Instrument™ (HVI) length uniformity values from the current study and previously reported values.
7
Comparison of the High Volume Instrument™ (HVI) short fiber index values from the current study and previously reported values.
7
Figure 5 shows the new HVI strength values in comparison with the previously measured and reported values and Figure 6 shows the increase in measured HVI strength with the UHML, similar to that previously reported.7,9,10 (In Figure 5 small differences in the strength calibration of the instruments probably account for the slope being significantly greater than unity. This difference between the data sets does not affect the conclusions of this study.) Figure 7 shows the new observation that the HVI reported elongation values also vary significantly as a function of length.
Comparison of the High Volume Instrument™ (HVI) strength values from the current study and previously reported values.
7
The observed change in measured High Volume Instrument™ (HVI) strength as a function of the HVI Upper Half Mean Length as measured in the current study. The observed change in measured High Volume Instrument™ (HVI) elongation as a function of the measured HVI Upper Half Mean Length.
At the end of each HVI length and strength test measurement, the instrument was manually interrupted and stopped prior to the instrument automatically disposing of the broken beard. In this way, the broken portion of the beard still attached to the comb was visible, as shown in Figure 8. The broken edge of the beard is clearly visible as well as two ‘impression lines’ on the remaining portion of the beard, as marked in Figure 8. It is assumed that these ‘impression lines’ are the imprint of the clamping jaw. The distances from the base of the jaw to the two impression lines and the edge of the broken beard were measured using digital calipers for each of the 10 measurements. These values are summarized in Table 5. It is interesting that for each of the three separate distances, the average, standard deviation and range (e.g. min and max values) are very similar for each of the five different samples. This indicates that, on average, the positioning of the jaws along the beard for the strength measurement is similar for all five different samples. For the data set under consideration, this confirms the validity of one of the important assumptions in the previous paper
7
that was used both to derive a quantitative explanation for the observed bias of the HVI strength measurement and to develop a correction factor to compensate for the bias.
Image of the broken portion of the beard still attached to the comb, immediately following the High Volume Instrument™ strength test. Summary of measured distances (mm) from the base of the beard (n = 20)
Summary of measured width of clamp (second impression – first impression) and an estimate of the gauge length (broken edge – second impression) in mm (n = 20)
Fibers should break at random positions across the width of the beard; there is clear evidence in Figure 8 of this variation, and secondly some relaxation of the broken section may occur. Both of these effects will contribute to and explain the small difference between the measured average estimate of the gauge length reported in Table 4 (2.65 mm) and the nominal gauge length (3.175 mm).
Returning to Table 5, there is a significant variability in the observed distances between replicate measurements on the same sample, indicating that the positioning of the jaws varies significantly between replicate measurements. Noting the manufacturer’s statement that the instrument uses an internal algorithm to determine the positioning along the beard of the jaws for the strength measurement, Figure 9 plots the measured distances as a function of the ‘amount’ output from the HVI. From the instrument perspective, control of this ‘amount’ parameter is important to ensure adequate light transmission through the beard for the length measurement. It is highly likely that the ‘amount’ is measured very close to the base of the beard, that is, not at the ‘position of break’. Further, given the general non-linear shape of the Fibrogram, this ‘amount’ parameter may not be linearly related to the actual amount of fiber between the jaws at the position of break. Figure 9 illustrates that as the amount increases, the position of the strength measurement moves further away from the base of the beard. The observed relationships in Figure 9 are approximately parallel. This is consistent with the physical interpretation above of the fixed distances between the markings and edge of the broken portion of the beard. The trends in Figure 9 are consistent with the manufacturer’s statement claim, that is, as the amount of mass of the beard increases, the strength measurement shifts further along the beard to a less dense region, and perhaps gives an insight into the ‘constant mass’ algorithm used by the instrument. It is particularly interesting that the positioning of the beard appears to be a smooth function of the amount and not simply a small number of discrete positions. Further, all five samples fall on the same lines in Figure 9 and indeed the individual points appear to be similarly randomly positioned along the available range. The instrument has specified minimum and maximum ‘amount’ values for a valid measurement.
For the broken portion of the beard, individual measured distances from the base of the beard to the broken edge (diamonds) or the outside edge of the clamping jaw (first impression) (circles) for each of the five different cotton samples are plotted as a function of the High Volume Instrument™ ‘amount’ parameter (n = 20).
Figure 10 shows that for each of these particular samples, and the pooled data set over all five samples, on average the recorded HVI strength value reduces linearly as the ‘amount’ parameter increases. It would appear that for these particular samples the instrument internal positioning algorithm has not completely removed the effect of different size beards and indeed has perhaps overcompensated for beard size.
Individual High Volume Instrument™ (HVI) strength values as a function of the HVI ‘amount’ parameter (a) for each of the five different samples including the least squares trend lines for Samples 1 and 5, and (b) as a pooled data set. The linear regression coefficients are presented as (estimate ± standard error).
It is interesting that the observed elongation values also exhibit a similar decreasing trend as a function of ‘amount’, as shown in Figure 11. The mechanism causing the observed changes in the average HVI fiber bundle elongation values as a function of fiber length reported in Figure 7 and the trends observed as a function of bundle size in Figure 11 are not clear. Different degrees of (a) slippage and/or (b) fiber alignment may be the underlying mechanisms.
Individual High Volume Instrument™ (HVI) elongation values as a function of the HVI ‘amount’ parameter as a pooled data set for the five different fiber length samples (i.e. n = 100). The linear regression coefficients are presented as (estimate ± standard error).
The HVI observations were extended by examining the relationship between amount and positioning along the beard using Upland and Pima long/strong and short/weak calibration cottons. The distance measurements are summarized in Tables 7 and 8. Table 8 again illustrates that the distance between the two impressions and the distance from the outer impression to the broken edge show very little variation between replicate measurements (compared to the considerable variability of the primary distance measurements in Table 7) and their average values are very similar to that observed previously (Table 6). This is further evidence consistent with the physical interpretation of these markings on the beard. The relationship between the position of the jaw and the HVI ‘amount’ parameter are shown in Figure 12. The two short/weak calibration cottons follow the same trend as observed previously. However, the data for the two long/strong calibration cottons form separate individual linear relationships with increasing slopes as the length increases. Thus, the position is not simply determined by the overall size of the beard as characterized by the ‘amount’ parameter, but appears to also be a function of length such that for a given amount if the fiber is inherently long, the instrument positions itself further along the beard for the strength measurement. Again this is qualitatively consistent with the concept of constant mass for the strength measurement.
For the broken portion of the beard, individual measured distances from the base of the beard to the outside edge of the clamping jaw (first impression). The circles represent the data for the cut samples, the squares for the Upland calibration cotton and the triangles for the Pima calibration cottons (n = 20). Summary of measured distances in mm from the base of the beard for the two sets of calibration cottons (n = 20) L/S: long/strong; S/W: short/weak. Summary of measured width of clamp (second impression – first impression) and an estimate of the gauge length (broken edge – second impression) in mm for the two sets of calibration cottons (n = 20) L/S: long/strong; S/W: short/weak.
Conclusion
The current work has investigated and confirmed the validity of two important assumptions utilized in the previous study of the observed length bias in the HVI strength measurement. 7 Firstly, single fiber testing confirmed that the particular sample preparation used to generate a sample of different fiber length characteristics from a common cotton sliver did not introduce any inherent damage to the fibers (and so this could not be the explanation for the observed trend in measured fiber strength as a function of fiber length). Secondly, the positioning of the jaws relative to the beard in the HVI strength measurement was explored. This positioning was found to be quite variable for replicate measurements on the same cotton and was demonstrated to be a function of the size of each individual beard, that is, the HVI ‘amount’ parameter. The average positioning between the different samples was found to be similar and this validated the approach used previously for deriving the correction factor for that particular sample set. Characterizing the position of the jaws was extended using a wider range of cotton samples and this proved to be helpful in understanding the operation of the HVI strength measurement module. Then, thirdly, it was demonstrated that the HVI positioning algorithm appears to be not simply a function of the size of the beard (i.e. the ‘amount’ parameter), but is also dependent on fiber length characteristics. These observations are not only consistent with the statement from the manufacturer that the instrument seeks a position of constant mass, but also imply that the approach utilized in the previous paper to develop a correction factor may need to be modified to take into account this additional variable. Fourthly, it was also observed that the reported HVI elongation values displayed both a significant bias due to fiber length and also a dependence on the size of individual beards tested.
Footnotes
Funding
This work was supported by the Australian Cotton Research and Development Corporation, CSIRO and the USDA.
Acknowledgment
This work was undertaken while one of the authors (GRS Naylor) spent time at the USDA ARS Laboratory in New Orleans. The authors also wish to thank Holly King for her expert technical assistance in skillfully and carefully collecting both the single fiber FAVIMAT and the HVI data reported in this paper.