低炭素設計のための鉄筋コンクリート建物の体化炭素の統計的特性評価と予測
Statistical Characterization and Prediction of Embodied Carbon in Reinforced Concrete Buildings for Low-Carbon Design (原題)
Fan Zhang, Bo Wen, Ditao Niu, Anbang Li, Qingxi Zhao, Zheng Fang, Yao Lv
🤖 gxceed AI 要約
日本語
本研究は、ライフサイクル炭素会計、統計分析、モンテカルロシミュレーション、機械学習を統合し、鉄筋コンクリート建物の体化炭素排出量を評価する。陝西省の124建物のデータを用い、材料生産段階が総排出量の90.09%を占め、鋼材とコンクリートで87.38%を占めることを示した。床面積は総体化炭素と強い相関（R²=0.95）を持ち、構造タイプが炭素原単位を決定する。決定木モデルが最良の予測性能（R²=0.885）を示し、構造システム別の炭素原単位の基準値を提案した。
English
This study integrates life-cycle carbon accounting, statistical analysis, Monte Carlo simulation, and machine learning to assess embodied carbon in RC buildings. Using data from 124 buildings in Shaanxi Province, it finds material production dominates (90.09% of total emissions), with steel and concrete contributing 87.38%. Floor area strongly correlates with total embodied carbon (R²=0.95), and structural type determines carbon intensity. A decision tree model achieves best prediction (R²=0.885), and benchmarks for different structural systems are proposed.
Unofficial AI-generated summary based on the public title and abstract. Not an official translation.
📝 gxceed 編集解説 — Why this matters
日本のGX文脈において
日本の建設業界では、SSBJ開示やカーボンニュートラル政策に対応するため、建物の体化炭素評価が重要になっている。本研究の統計的ベンチマークと機械学習予測手法は、日本のRC建物の低炭素設計や環境性能評価に応用可能であり、国内の建設会社や不動産デベロッパーにとって有用な知見を提供する。
In the global GX context
Globally, embodied carbon in buildings is gaining attention in climate disclosure frameworks like TCFD and ISSB, as well as in green building certifications. This study provides a data-driven approach to benchmark embodied carbon intensity, which can inform international efforts to standardize carbon accounting in the construction sector. The use of machine learning for rapid prediction offers a scalable tool for practitioners worldwide.
👥 読者別の含意
🔬研究者:Provides a novel integrated framework combining statistical analysis and ML for embodied carbon prediction, with benchmarks that can be compared across regions.
🏢実務担当者:Offers practical embodied carbon intensity benchmarks and a decision tree model for quick estimation, useful for low-carbon design and sustainability reporting.
🏛政策担当者:Suggests potential regulatory benchmarks for embodied carbon limits in buildings, which could inform building codes and carbon reduction policies.
📄 Abstract（原文）
With the rapid expansion of the construction sector, reducing embodied carbon in reinforced concrete (RC) buildings has become an important issue for low-carbon development. However, practical benchmarks and reliable prediction tools at the structural level are still limited. This study presents an integrated approach combining life-cycle carbon accounting, statistical analysis, Monte Carlo simulation, and machine learning to evaluate embodied carbon emissions in RC buildings. A dataset of 124 buildings in Shaanxi Province is used to examine emission distributions and identify key influencing factors. The results show that most life-cycle stages approximately follow normal distributions, while the construction stage is better described by a lognormal distribution due to higher variability. The material production stage is the dominant source of embodied carbon, accounting for 90.09% of total emissions, with steel and concrete contributing 87.38%. Building floor area shows a strong correlation with total embodied carbon (R 2 = 0.95), whereas structural type plays a key role in determining embodied carbon intensity. Based on the statistical results, embodied carbon intensity limits are proposed for different structural systems, including frame structures (471.316 kg CO 2 e/m 2 ), shear wall structures (443.757 kg CO 2 e/m 2 ), and frame–shear wall structures (423.724 kg CO 2 e/m 2 ), providing practical benchmarks for low-carbon design. Among the tested models, the decision tree achieved the best performance (R 2 = 0.885), indicating its suitability for rapid prediction. Overall, the proposed framework offers a quantitative basis for evaluating embodied carbon and supports decision-making in low-carbon RC building design.
🔗 Provenance — このレコードを発見したソース
- openalex https://doi.org/10.1016/j.cscm.2026.e06436first seen 2026-08-24 04:53:41
🔔 こうした論文の新着を逃したくない方は キーワードアラート に登録（無料・3キーワードまで）。
gxceed は公開メタデータに基づく研究支援データセットです。要約・翻訳・解説は AI 支援で生成されています。 最終的な解釈・検証は利用者が原典資料に基づいて行うことを前提とします。