Quantification and correlation of viscose quality characteristics

Abstract

The viscose solutions contain the dissolved cellulose and the undissolved parts in form of undissolved fibres, micro gels, macro gels and the inorganic impurities. These particles influence processing properties of viscose such as filterability and viscosity which lead to industrial problem of clogging of spinneret and breakage of elementary fibres during spinning. The differences in the quality of the pulp and alkali cellulose or an electron beam treatment have a preferential influence on the macro gel particle content, the micro gel particle content is influenced more strongly by the viscose dissolution temperature and composition of viscose dope. When preparing a viscose it is important to add, at a suitable point in the process, an agent which improves the filterability of the viscose solution. Improved filterability is obtained by the ability of the agent to reduce the formation of gel particles in the viscose solution. As an additional effect, certain, especially suitable agents provide fibres or films with less milkiness and higher brightness. In order to improve filterability of viscose and minimize the impact of gel content on spinnability and fibre quality experimental studies were carried out. The work done for gel quantification and particle analysis in viscose dope has been directed towards finding the better of the two techniques available namely, first being microscopy using slides containing a drop of viscose and other being microscopy using flow cell (a cuboidal shaped glass tube for viscose to flow in, designed by ABSTC). The formulation of MATLAB code to detect the particles in images has been done with the help of ImageJ, so as to understand the image processing techniques available and getting the intuition of which technique will work out the best for images take at Aditya Birla. The task to separate out fibres using a code appears challenging. The out of focus images and the non- uniform illumination reduce the devised code’s accuracy. Inch by inch progress has been made to reach the proposed objective of quantifying impurities (gels, undissolved fibres and black particles) in viscose dope and to correlate the distribution with the filterability values. The code is expected to provide particle size distribution of viscose thus segmenting out the gels, undissolved fibres, black particles separately. Aspect ratio, area and geometrical information about particles is the basis for the same using K-means color clustering for RGB images, smoothening followed by gradient on grayscale and HSV images. Along with that an interplay of various other methodologies the quantification protocol has been achieved. The correlation between particle size distribution and filterability values has been investigated. The classical as well as modern filtration theory have been applied to estimate a filtration value using particles size statistics from code and actual filterability, also called the clogging value or the KW values. The correlation between the two is the ultimate approach towards completion of the set target. The correlations obtained from image analysis to KW values is weak. The power law fitting to KW values with cumulative total volume distributions per mL from image analysis gives R-square of 0.255. Cumulative volume per mL from Beckman Coulter shows a better correlation to KW values with R-square of 0.61-0.65. However, the polynomial fit to volume distributions per mL from imaging gives R-square of 0.45-0.61. The logarithmic fit to cumulative fibre volume distribution per mL from imaging gives a fit of 0.36 R-square. The image analysis procedure needs to be made robust to count even the translucent gels and swollen fibres with the viscose flow regulated at a constant velocity for all the batches analyzed. The improvements in this direction can be valuable for studying effect of process changes on each segment of impurities separately.