박현준 Hyun-jun Park , 김명수 Myung-soo Kim , 고준혁 Hyeok-jun Ko , 송상훈 Sang-hoon Song , 박민선 Min-seon Park , 윤희영 Hui-young Yun , 이남훈 Nam-hoon Lee , 박진규 Jin-kyu Park

DOI:10.9786/kswm.2026.43.2.33

Abstract

Solid waste landfills require systematic monitoring to manage risks such as landfill gas emissions, leachate generation, surface subsidence, and slope instability. Conventional landfill management methods rely on labor-intensive, point-based field surveys and post-event inspections that provide limited spatial coverage, poor efficiency, and questionable worker safety. Unmanned aerial vehicles (UAVs) have accordingly emerged as an effective tool for landfill monitoring as they enable the rapid, on-demand acquisition of centimeter-level, high-resolution spatial and environmental data at relatively low cost. Current UAV-based landfill management technologies can be applied in either spatial information-based management approaches, which include three-dimensional photogrammetry and LiDAR surveys for accurate volume estimation, remaining airspace assessment, and the time-series detection of settlement and slope changes, or environmental and safety management approaches, such as UAV-based methane monitoring, thermal infrared imaging to detect hotspots and fire risks, and multispectral analyses (e.g., using the normalized difference vegetation index (NDVI) and normalized difference water index (NDWI)) to identify vegetation stress and water ponding related to leachate or gas leakage. Future development of this technology should focus on sensor fusion, artificial intelligence-based automated anomaly detection, and digital twin integration to facilitate predictive, data-driven, and proactive landfill management.

Key Words

UAV, Landfill management, Smart landfill, Digital twin, Remote sensing

이동규 Donggyu Lee , 최수진 Sujin Choi , 장우혁 Woohyeok Jang , 황석환 Seokhwan Hwang

DOI:10.9786/kswm.2026.43.2.50

Abstract

This study conducted seasonal monitoring comprising the collection of 4 samples over 1 year from each of 22 full-scale anaerobic digestion facilities in Korea to compile data describing the physicochemical characteristics of the facility substrate, influent, and effluent as well as the process performance. These data were subsequently applied to establish a data-driven framework for classifying and improving digester efficiency. The methane yield and volatile solids (VS) removal were obtained to quantify process performance, then classified using various machine learning models; the random forest (RF) model exhibited the most robust performance for both metrics, with Macro-averaged F1-score (F1-macro) of 0.746 ± 0.093 for methane yield level and 0.688 ± 0.111 for VS removal rate level, and balanced accuracies of 0.753 ± 0.088 and 0.695 ± 0.113, respectively. Notably, a high VS removal did not always coincide with high methane yield, implying potential bottlenecks in the methanogenesis stage under certain conditions. Interpretation based on SHapley Additive exPlanations (SHAP) indicated that food waste (FW) most strongly increased the probability of achieving high performance in both targets, whereas excessive sewage sludge (SS) and animal manure (AM) generally reduced performance; food-waste leachate (FWL) tended to support VS removal but may involve instability risks when over-applied. Finally, an RF-based surrogate optimization generated 200,000 virtual mixtures under simplex constraints and retained approximately 40,000 realistic candidates via k-nearest neighbors (kNN) filtering. The feasible region satisfying both targets simultaneously was limited, and the best compromise mixture was FW 74.3%, HM 10.1%, PS 7.0%, FWL 6.1%, AM 1.9%, and SS 0.6%, with predicted probabilities for the high methane-yield and high VS-removal classes of 0.63 and 0.58, respectively.

Key Words

Substrate mixing ratio, Volatile solids removal rate, Methane yield, Classification, Decision-tree-based model

박민선 Min-seon Park , 박현준 Hyun-jun Park , 송상훈 Sang-hoon Song , 이남훈 Nam-hoon Lee , 박진규 Jin-kyu Park

DOI:10.9786/kswm.2026.43.2.62

Abstract

Korea's ban on the direct landfilling of municipal solid waste (MSW), which became effective in the Seoul Metropolitan Area in 2026 and will be implemented nationwide in 2030, is expected to increase the disposal of incineration residues (particularly bottom ash and chelated fly ash) in existing MSW landfills. This study evaluated the potential impacts of this shift and identified essential landfill management strategies for addressing them by drawing on landfill classification and waste acceptance criteria from Germany, Denmark, Austria, and Japan. Major technical challenges associated with the increased disposal of incineration residue in landfills include internal temperature rise and hydrogen generation driven by metal reactions, accelerated deterioration of geomembranes and clay liners, and biological/chemical clogging of geotextiles and leachate collection systems (notably at higher bottom-ash mixing ratios). In addition, elevated Ca and chloride levels can lead to CaCO₃ scaling and chloride-related corrosion, high salinity and residual chelating agents may inhibit nitrification, and hardpan formation may restrict water and gas movement. Recommended countermeasures include metal recovery prior to landfilling, Ca²⁺ removal via Na₂CO₃ pretreatment, strict control of chelating agents used for fly ash stabilization, and reinforcement of leachate drainage systems (e.g., vertical drains). Notably, impermeable final covers may be unsuitable for ash-dominated landfills; therefore, differentiated final cover standards that allow limited, controlled infiltration-as practiced in German Class I landfills-should be considered along with alternative daily cover materials.

Key Words

Municipal solid waste, Landfill, Incineration, Bottom ash, Fly ash

김인복 In-bok Kim , 정명섭 Myeong-seob Jeong , 김춘산 Chun-san Kim , 이재영 Jai-young Lee , 황용우 Yong-woo Hwang

DOI:10.9786/kswm.2026.43.2.74

Abstract

This study applied Life Cycle Assessment (LCA) to evaluate greenhouse gas (GHG) emissions of an automated waste collection system (AWCS) installed in a residential complex in Korea. The functional unit was defined as the treatment of 1 ton of municipal solid waste, and the analysis covered material production, construction, operation, maintenance, and end-of-life stages over a 40-year service life. Total life-cycle emissions were estimated at 1,237.38 kg CO₂-eq per functional unit, with the operational stage accounting for the largest share. Electricity consumption, particularly from high- capacity blowers, was identified as the dominant emission source. Scenario analysis indicated that increasing the share of photovoltaic energy substantially reduces long-term emissions compared to grid electricity alone. These findings highlight that the carbon footprint of AWCS is primarily driven by operational electricity use and suggest that electricity supply transition plays a critical role in achieving emission reduction.

Key Words

Life cycle assessment, Automated waste collection system, Greenhouse gas emissions, Urban waste management, Carbon neutrality

김성준 Seongjun Kim , 연익준 Ikjun Yeon

DOI:10.9786/kswm.2026.43.2.88

Abstract

This study developed models for forecasting daily municipal solid waste (MSW) inflow at the C City incineration facility using operational data collected from 2015 to 2025. Because the influences of seasonal and compositional changes cause MSW inflow to exhibit nonlinear and highly variable patterns, deep learning approaches were applied to predict inflow accurately. Models were constructed based on two recurrent neural network architectures, the long short-term memory (LSTM) and gated recurrent unit (GRU), then evaluated under identical conditions. The input variables were selected through a correlation analysis, revealing relatively high positive correlations between MSW inflow and physical composition variables such as plastics, paper, food waste, and fiber. The dataset was normalized and split into training and test sets, then Bayesian optimization was used to identify the optimal hyperparameter combinations for each model. Early stopping and dropout were applied to both models to prevent overfitting. Model performance was subsequently evaluated using the mean average error, root mean square error, and coefficient of determination for 1, 3, 5, and 7 d prediction periods. The GRU model consistently exhibited lower errors and demonstrated stable convergence, outperforming the LSTM model across all prediction periods. These findings highlight the potential ability of GRU-based deep learning to enhance MSW inflow prediction and support data-driven waste management planning.

Key Words

MSW, LSTM, GRU, Deep learning, Time-series

김성준 Seongjun Kim , 정지선 Jisun Jung , 연익준 Ikjun Yeon

DOI:10.9786/kswm.2026.43.2.97

Abstract

This study estimated the greenhouse gas (GHG) emissions from a municipal waste incineration facility by training different machine learning models on operational, meteorological, and demographic data to compare their predictive performance. Monthly data from 2015 to 2025 were applied to calculate the total GHG emissions based on Intergovernmental Panel on Climate Change (IPCC) guidelines using random forest (RF), support vector regression (SVR), extreme gradient boosting (XGBoost), and artificial neural network (ANN) models. Although all models successfully reproduced the overall emission trends, the ANN model exhibited the highest predictive accuracy (R² = 0.935, RMSE = 372.317, MAE = 309.345), followed by the XGBoost (R² = 0.855, RMSE = 556.219, MAE = 415.096), SVR (R² = 0.734, RMSE = 753.599, MAE = 529.916), and RF (R² = 0.664, RMSE = 846.712, MAE = 632.756) models. A Shapley Additive Explanations analysis indicated that the waste composition, particularly the plastic content, and number of households exerted the greatest influence on GHG emissions, whereas meteorological variables exerted relatively little influence. The results of this study demonstrate that machine learning models can effectively predict the GHG emissions produced by waste incineration facilities to provide a basis for data-driven emissions management and operational planning.

Key Words

Greenhouse gas emissions, MSW incineration facility, Machine learning, ANN, SHAP

박선균 Sunkyun Park , 곽상훈 Sanghoon Kwak

DOI:10.9786/kswm.2026.43.2.109

Abstract

This study examined how plant-level emissions mitigation translates into product-level embedded emissions calculated using the carbon border adjustment Mechanism (CBAM) based on verified third-party steel data. A 30% reduction in in-plant process emissions yielded only 1-5% declines in product indicators owing to the dominance of precursor emissions, and applying the CBAM production denominator (excluding waste and scrap) shifted the reported emissions by 2-6%. Notably, in a pathway-switching scenario, product-level embedded emissions decreased by approximately 18- 20% when blast furnace-basic oxygen furnace (BF-BOF) routes used for precursor production were partially replaced by scrap-electric arc furnace (EAF) routes, indicating that supply-chain pathway choices can dominate the CBAM-facing mitigation signal.

Key Words

Carbon border adjustment mechanism(CBAM), Circular economy, Embedded carbon emissions, Process- level direct emissions, Supply chain emissions

조해인 Haein Cho , 장용철 Yong-chul Jang , 이호원 Howon Lee , 신동규 Donggyu Shin , 류태안 Taean Ryu , 김서영 Seoyoung Kim

DOI:10.9786/kswm.2026.43.2.120

Abstract

The recent surge in the consumption of single-use plastic products has accelerated global efforts to address their environmental impacts. For example, the European Union enacted the Single-Use Plastics Directive to regulate certain plastic items, including single-use plastic straws. South Korea is also working to reduce reliance on single-use plastic straws by promoting alternative materials. This study accordingly examined various international regulatory measures on single-use plastic straws and quantitatively assessed the consumption and carbon footprints of single-use plastic straws in South Korea. Legislative documents, policy reports, and relevant scientific literature were reviewed to examine national and international strategies for regulating straws. A field survey of 101 coffee shops from 8 major franchises in 2 metropolitan cities of South Korea was undertaken to quantify the consumption footprint associated with single-use plastic straws, and the corresponding carbon footprint was calculated using the Korea Environmental Product Declaration assessment factor provided by the Korea Environmental Industry and Technology Institute. The results indicated that each coffee shop uses an average of approximately 4,000 single-use plastic straws per month. Based on this survey result, the annual consumption footprint of single-use plastic straws in South Korea is estimated at approximately 5.1 billion straws, which is equivalent to about 5,304 tons assuming an average weight of 1.04 g per straw. The carbon footprint calculated based on the single-use plastic straw consumption was determined to be approximately 25,300 tons CO₂-eq. Therefore, the results of this study provide valuable data informing the development of regulatory measures to reduce single-use plastic straw consumption and improve waste management in South Korea.

Key Words

Single-use plastic straw, Consumption footprint, Carbon footprint, Life cycle management