Eko Setiawan, Ruzika Azhar, Septin Puji Astuti, Suranto Suranto, Hafidh Munawir, Unung Istopo Hartanto
This study explores the energy potential of spent tea waste (STW) for briquette production by integrating pyrolysis, Taguchi optimisation, and machine learning, a novel approach in biomass fuel research. Unlike conventional biomass briquettes, limited research has explored the optimisation of STW briquettes through thermal decomposition. Pyrolysis is introduced as a pre-treatment to enhance fuel properties, while the Taguchi method systematically optimises combustion performance. Additionally, machine learning models provide a predictive framework, identifying key factors that influence briquette quality. An L9(33) orthogonal array was used to evaluate nine combinations of tea ash composition (100 g, 150 g, and 200 g), starch (7.5 g, 10 g, and 12.5 g), and water (80 mL, 120 mL, and 160 mL). Analysis of Variance (ANOVA) determined the effects of these variables on proximate composition and calorific value, while machine learning identified the predictive model and the most influential parameters affecting briquette performance. The results reveal that water and its interaction with tea ash significantly affect moisture content (R2 = 95.2%, RMSE = 0.849), while starch and tea ash interaction govern volatile matter (R2 = 62.5%, RMSE = 1.010). Fixed carbon, a key factor in combustion stability, is influenced by all briquette materials (R2 = 86.7%, RMSE = 2.210). This is similar to the calorific value, which improves combustion efficiency (R2 = 95.0%, RMSE = 0.313). Meanwhile, this study was unable to determine a predictive model for ash content. The optimal composition, 200 g tea ash, 7.5 starch, 80 mL water, achieves a calorific value of 24.18 MJ/kg, with 2.62 wt% moisture content, 23.05 wt% volatile matter, 61.81 wt% fixed carbon, and 10.81 wt% ash content. These results highlight the potential of STW as a sustainable biofuel feedstock. Although the calorific and moisture content meet European standards, reducing ash content remains a key challenge due to the material’s high mineral content and trade-offs in the pyrolysis process. This study demonstrates the feasibility of combining thermal pre-treatment, statistical optimisation, and machine learning to improve biomass briquette performance and support circular economy initiatives. © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2026.
Department of Industrial Engineering, Universitas Muhammadiyah Surakarta, Central Java, 57162, Indonesia; Department of Environmental Sciences, Universitas Islam Negeri Raden Mas Said Surakarta, Central Java, 57168, Indonesia; Department of Informatics Engineering, Universitas Negeri Surabaya, East Java, 60231, Indonesia
