The study aims to establish a direct relationship between backed yarn ratios, fabric thickness and porosity/permeability, demonstrating the superiority of optical porosity over geometric methods for predicting air permeability in two-fleece and three-fleece fabrics. It highlights the potential of machine learning (ML) models in predicting porosity/permeability, offering practical implications for optimizing the backed yarn ratio to balance cost and performance.
In the study, five different samples were produced for both two-thread fleece and three-thread fleece fabrics with different backed yarn ratios. To obtain different backed yarn ratios, while the pulley degrees of the ground and binding yarns were kept constant in order to keep the feed ratios of these yarns constant, the pulley degrees feeding the backed yarn were changed. Thus, the effects of the changes in the backed yarn feeding ratios on the fabric physical properties, the fabric porosity and in parallel, the air permeability, which is the property most affected by porosity, were investigated. Also, to leverage the advancements in ML, a deep convolutional neural network approach for textile classification based on air permeability was employed. So, porosity of the fabrics was investigated by both image analysis and ML algorithms. To gain deeper insights into optical porosity and air permeability, pore sizes and distributions of fabrics were measured and assessed. In addition, by obtaining geometric porosity values and optical porosity values, important issues such as the effects of different backed yarn ratios on both porosity and air permeability, the relationship between porosity and air permeability and which porosity method gives better results in estimating air permeability were examined. All test results were statistically assessed on the basis of pulley degree, which is the main factor.
The fabric tightness expressed in terms of courses and wales numbers was not affected by the changes in the backed yarn ratio, but as the backed yarn ratio increased, the fabric weight and thickness values increased in parallel, as did both the geometric and optical porosity values, and so air permeability decreased in two-thread fleece and three-thread fleece fabrics. In terms of the porosity determination method, optical porosity, by simply taking and processing images without the need for any fabric physical property tests, was more effective in estimating air permeability. The number and area of pores decreased with the increase of the pulley degree and thus the backed yarn length; the ratio of the backed yarn laid into the pores gradually increased. The high success rate of the ML algorithm demonstrated that fabric properties and quality can be predicted in a much shorter time, effortlessly and with high accuracy rates with artificial intelligence techniques. The differences in physical properties porosity and air permeability are not high when the backed yarn ratio is increased, and there are no major changes between the fabrics that will affect their use. Therefore, knitting of two-fleece or three-fleece fabrics with smaller pulley degrees, which means feeding the backed yarn ratio at a lower level, did not create any problems in terms of the physical properties and air permeability; moreover, lower pulley degrees also reduce yarn waste and energy consumption, aligning with sustainable manufacturing goals.
The fleece fabrics have recently become one of the most preferred fabrics for sports and outdoor clothing due to their good thermal comfort, low air permeability and protection against cold. Changing the backed yarn ratios, which are factors that affect the properties of these fabrics and have not been examined or have been examined inadequately in most previous studies, and examining the effects of these changes on both two-thread fleece and three-thread fleece fabrics constitute the original value of the study. In addition, the use of two different porosity determination methods in estimating air permeability, which is the most prominent property of two-fleece and three-fleece fabrics, and in determining which method is more successful in estimation added originality to the study. Since lower pulley degrees reduce yarn waste and energy consumption, manufacturers could adopt this approach to balance cost, performance and eco-friendliness. Providing manufacturers with an approach that they can apply in this way has greatly increased the value of the study.
