No abstract

In classical stitching process, needle holes occur during the penetration of the needle through the fabric. If the waterproofness of the sewn product is important, the water leakage from these holes must be prevented. To prevent this negative situation, different techniques such as sealing of seams with waterproof tapes, joining the textile materials by bonding or welding are used. Among these techniques, there is no needle damage in bonding and welding and all the seam area is covered by thermal or chemical bonding. In sewing technology, the water leakage is prevented by covering all the seam area with seam sealing tape. These three methods have different effects on the physical properties of the seams obtained. Instead of covering the whole seam area, covering just the needle damages is the focus of this research.  With this aim, fusible sewing threads were used to cover the needle damages to increase the waterproof performance of seam line. The fusible sewing threads have not been used for obtaining waterproof seams before.  In this research, the fusible sewing threads were used as lower thread in different combinations. Initial results of waterproofness test show that, melted fusible threads improve the waterproof performance of seams. In other words, the needle damages on sewn fabric can be covered by melted fusible sewing thread. However, unbalanced seam is the negative side of this research because of using different threads as needle and bobbin thread. Additionally, there is no variety of fusible threads to select an appropriate one for this method. The study is hoped to be a sample for the further studies on this method, using different fusible threads, fabrics, seam types and even improving new fusible threads for this waterproofing method.

The aim of this study was to estimate waterproofness values of seams composed of the combination of fusible threads and antiwick sewing threads through artificial neural networks (ANN). Fusible threads were used to obtain waterproof seams for the first time. Therefore, estimating the value of the waterproofness variable with the help of models created from test values can contribute to accelerating the progress of further studies. Hence, ten different samples were prepared for two fabrics, and the waterproofness values of the seams obtained were tested using a Textest FX 3000 Hydrostatic Head Tester III. For the prediction of waterproofness values of the seams, the Levenberg-Marquardt backpropagation algorithm was used for artificial neural network pattern models with sigmoid and positive linear transfer functions. Finally, the ANN model was successful in estimating the waterproofness of the seams. The highest correlation coefficient was R = 0.95081 which indicated that the prediction made by the neural network model proved to be reliable.