The objective of this study is to predict the amount of plastic waste generated in South Korea by 2030 using artificial neural networks (ANN) and to analyze
carbon emission scenarios based on different recycling methods. The surge in plastic use in m...
The objective of this study is to predict the amount of plastic waste generated in South Korea by 2030 using artificial neural networks (ANN) and to analyze
carbon emission scenarios based on different recycling methods. The surge in plastic use in modern society has led to a significant increase in plastic waste,
with a high proportion of this waste leaking into the natural environment without proper treatment, causing environmental issues. Managing plastic waste in such a scenario has expanded beyond a mere environmental issue to a social and economic problem, necessitating effective management and treatment strategies. In this study, 23 types of data were collected, considering various socioeconomic and environmental factors, and 11 key factors were selected for
ANN analysis through Pearson correlation analysis. The key factors are the number of single-person households, scale of accommodation and food service
businesses, total plastic production, number of registered households, average population per household, GDP, construction market size, transportation and
storage market size, energy consumption, actual recycling rate of plastics, and export weight of waste plastics. Using Keras Tuner, the optimal hyperparameters were derived, and an ANN model was built to predict plastic waste generation by 2030. The ANN model predicted about 15% higher plastic waste generation with approximately 53% lower error rate compared to the linear regression model. The ANN methodology yielded estimation results that reflected various socioeconomic changes more accurately than traditional statistical methodologies such as regression analysis. Traditional statistical methodologies might have difficulty in rationally estimating waste generation in special situations like the recent COVID-19 pandemic. In contrast, the ANN methodology, by comprehensively considering various factors, provided more reliable results. Furthermore, the contribution of each factor to the model's prediction results was evaluated using SHAP (SHapley Additive exPlanations) analysis. The SHAP analysis revealed that factors with relatively low linear correlation according to the Pearson correlation analysis had a greater impact on the ANN model's prediction results. For instance, energy consumption, GDP, and total plastic production had high SHAP values, positively influencing the model output, whereas the average population per household negatively influenced the model output when its value was high. This indicates that the ANN model can effectively model the complex interactions among various factors. Finally, carbon emission scenarios based on different recycling methods were established, and the carbon emissions for each scenario were calculated. For
scenario setting, several policies and goals of the South Korean government aimed at achieving carbon neutrality by 2030 were investigated. Changes in the
proportion of plastic waste treatment methods were assumed depending on the application of each policy and goal. Based on this, scenarios such as existing
treatment ratios, 10% expansion of chemical recycling, 30% application of physical recycling, zero landfill scenario, and scenarios with high proportions of
physical and thermal recycling were set. The analysis showed that reducing landfill and incineration ratios while increasing the physical recycling ratio was
the most effective strategy for reducing carbon emissions. Specifically, a scenario expanding zero landfill, reducing incineration, and expanding thermal and physical recycling industries showed a reduction of approximately 10.29 million tCO2-eq. The results of this study provide crucial information to policymakers and industry stakeholders in developing recycling industries and waste management strategies. Quantitative comparisons of greenhouse gas emissions according to the development direction of each plastic waste treatment method will contribute to establishing balanced plastic management strategies that pursue both environmental protection and economic sustainability. Additionally, this analysis will play a significant role in determining investment directions for technology development and infrastructure construction in the future plastic recycling industry. This research is expected to make a significant contribution to the establishment of a sustainable resource-circulating economy and the reduction of carbon emissions in South Korea.