Energy management in industrial manufacturing is increasingly driven by decarbonization requirements and cost optimization. This study proposes a digital twin framework that integrates photovoltaic (PV) energy forecasting, carbon footprint accounting, and AI-based dynamic pricing for a vertically integrated textile manufacturer, ATLAS Denim Co. in Adana, Türkiye. Using historical meteorological data and on-site PV generation records, three machine learning algorithms-Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Long Short-Term Memory (LSTM)-are trained to forecast solar energy output. The RF model achieves the highest accuracy, with a coefficient of determination (R²) of 0.993, and is used to estimate PV generation for the January-May 2025 period. The forecasts are combined with production data to compute energy use, carbon emissions, and energy cost per meter of textile output. A carbon-aware pricing model then links product prices to energy mix and emission intensity, enabling the formulation of sustainability-adjusted price offers. Over the five-month horizon, the framework yields an estimated 55.18 MWh of PV generation and a carbon saving of 23.9 tCO₂ compared to grid-only electricity. The results demonstrate that integrating machine learning (ML)-based forecasting and digital twin concepts can support climate-conscious and economically efficient decision-making in textile manufacturing.
Manolya Güldürek (Thu,) studied this question.