Industrial waste, leftover materials and chemical residues constitute a major environmental challenge in Bangladesh where the textile industry annually produces some 400,000 tons of fabric waste that is a significant source of pollution. Manufacturing 4.0 technologies are making possible advanced manufacturing systems that can optimize production and reduce waste, using technologies such as those supported on data analytics and the Internet of Things (IoT). The objective of this study is to build machine-learning based predictive analytics framework for minimizing textile production waste, evaluate the developed framework using a practical context in Bangladesh and finally observe the environmental and socio-economic impact caused by the approach. The design employed a mixed-methods case study. Data were collected from a medium-sized textile dye-house in Dhaka from January to March 2025, with IoT tracked measurements on fabric consumption, machine productivity, and wastewater output (n = 1,000 production cycles). Python generates a Random Forest regression model to predict waste, while simulation is carried out through a digital twin to optimize production parameters. The model obtained a mean absolute error of 5.4% and was able to accurately predict the pattern of waste. Application of the optimized parameters resulted in 20% less fabric waste (from 500 to 400 kg/day), 15% less use of water in dyeing (from 10,000 to 8,500 liters/day) and 10% lower CO₂ emission (0.5 tons/day). The greatest waste reduction was observed in the urban area, due to better cutting techniques. These findings highlight the opportunities provided by Industry 4.0 analytics for sustainable manufacturing towards UN SDG 12. Additional investigation is also required on low-cost IoT deployment and policy enablers, to achieve widespread adoption and impactful change sustainably in developing economies.
Uddin et al. (Thu,) studied this question.