PEMANFAATAN GENERATIVE DESIGN DALAM OPTIMASI KINERJA ENERGI BANGUNAN
生成デザインを活用した建物エネルギー性能最適化：系統的文献レビュー (AI 翻訳)
Suryawinata, Bonny, Darmadi, Herru, Mansuan, Melki
🤖 gxceed AI 要約
日本語
本論文は、生成デザイン（GD）手法を用いた建物エネルギー消費、日射、温熱環境最適化に関する系統的文献レビュー（2019-2025年、47編）である。Rhino/Grasshopperと多目的遺伝的アルゴリズムが主流（87.2%）だが、GANや機械学習サロゲートモデル（17%）が急速に普及。熱帯・亜熱帯気候への適用が僅か8.5%と地理的偏在を指摘。
English
This systematic literature review (47 papers, 2019-2025) examines generative design (GD) for optimizing building energy, daylighting, and thermal comfort. Rhino/Grasshopper with multi-objective genetic algorithms dominate (87.2%), while generative AI and ML surrogate models (17%) are rapidly accelerating simulation. A major geographic gap is identified: only 8.5% of studies cover tropical/subtropical climates.
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
日本ではZEB・ZEH基準や2025年省エネ基準全面適合があり、建物セクターの脱炭素が急務。本レビューはGDによるエネルギー最適化の手法整理を提供し、日本の建築設計実務へのAI応用や熱帯地域向け基準（沖縄など）への示唆も含む。
In the global GX context
Globally, buildings account for ~40% of energy-related CO₂. This review provides a state-of-the-art taxonomy of generative design methods for performance optimization, highlighting the rapid adoption of AI surrogates and the critical underrepresentation of tropical climates—relevant for ISSB/TCFD-aligned building asset decarbonization strategies.
👥 読者別の含意
🔬研究者:Valuable taxonomy of GD methods and algorithms, including emerging AI surrogates, plus identification of the tropical-climate research gap.
🏢実務担当者:Guidance on tooling (Rhino/Grasshopper, EnergyPlus) and algorithms (NSGA-II) for early-stage building energy optimization.
🏛政策担当者:Evidence of geographic bias in building energy research; supports rationale for tropical-region green building code development.
📄 Abstract（原文）
TINJAUAN LITERATUR SISTEMATIS: PEMANFAATAN GENERATIVE DESIGN DALAM OPTIMASI KINERJA ENERGI BANGUNAN The building sector accounts for approximately 30–40% of global final energy consumption and nearly 40% of energy-related CO₂ emissions, making it a critical target for decarbonization efforts. Generative Design (GD), an emerging computational paradigm leveraging parametric modeling, multi-objective optimization algorithms, and machine learning, offers transformative potential for optimizing building performance from the earliest conceptual stages. This paper presents a systematic literature review (SLR) of 47 peer-reviewed publications (2019–2025) indexed in Scopus, examining how generative design methods are applied to optimize building energy consumption, daylighting, and thermal comfort. Using the PRISMA framework, we evaluate publication trends, journal quality, methodological taxonomies, optimization algorithms, simulation engines, and climate contexts. The findings reveal a dramatic acceleration in research output, with 57.4% of key studies published in 2024–2025 and 68.1% featured in Q1 Scopus journals. The Rhino/Grasshopper ecosystem combined with Ladybug Tools and EnergyPlus dominates current practice (87.2%), while Multi-Objective Genetic Algorithms (MOGA/NSGA-II) remain the most widely adopted optimization solver (76.6%). Furthermore, Generative AI (GANs, CycleGAN, SolarGAN) and Machine Learning surrogate models (17.0%) are rapidly transforming the field by accelerating physics-based simulation speeds up to several hundred times. Crucially, a major geographic disparity is identified: only 8.5% of studies focus on tropical/subtropical climates. This review establishes an integrative framework for performance-driven generative design and highlights future directions for tropical climate adaptation, explainable AI, and local green building standard integration.
🔗 Provenance — このレコードを発見したソース
- Zenodo https://zenodo.org/records/21667209first seen 2026-07-30 04:33:42
🔔 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。