Research development in intelligent generative design methods for steel modular structures

Researchers have developed a systematic review that charts the evolution of artificial intelligence in generative design for steel modular structures, particularly steel box modular buildings, revealing a shift from rule-based, single-objective automation toward integrated, data-driven, and sustainability-aware engineering solutions. Published in Smart Construction, this work analyzes relevant studies through two technical lenses—classic AI (e.g., genetic algorithms, expert systems) and modern AI (e.g., deep learning, large language models)—and examines their applications in intelligent structural scheme generation, single- and multi-objective optimization, and automated layout and drawing production. The review also proposes future directions centered on hybrid knowledge-data integration models, practical engineering deployment, and lifecycle-oriented sustainable optimization, offering guidance to advance intelligent generative design in modular construction.

Modular construction, where buildings are assembled from prefabricated units like standardized steel boxes, promises faster, more efficient, and higher-quality building processes. However, its design phase faces significant hurdles: low efficiency due to repetitive tasks, heavy reliance on manual expertise leading to inconsistent quality, and difficulty in simultaneously optimizing multiple objectives like cost, safety, and sustainability.

To address these challenges, the integration of Artificial Intelligence (AI) into a "generative design" paradigm—where algorithms can automatically and intelligently produce and evaluate numerous design options—has become a critical industry demand. However, a comprehensive overview of this rapidly evolving field was lacking.

Now, a team of researchers from Tianjin University has conducted a systematic review of relevant studies, charting the research landscape and identifying clear future directions. Their work, "Research development in intelligent generative design methods for steel modular structures," is published in the inaugural issue of Smart Construction.

The review categorizes AI technologies into two main categories: "classic AI" and "modern AI". Classic AI methods, including genetic algorithms (GA), expert systems, which rely on pre-defined rules and optimization, have been the dominant approach so far. They have shown success in automating specific tasks and performing single-objective optimizations, but they struggle with complex, multi-faceted design problems due to their dependence on manual rule-setting and high computational costs."

The research shows a clear inflection point around 2021. Driven by the maturation of modern AI and national strategies promoting intelligent construction, publication rates have surged. Modern AI techniques, particularly deep learning models like Generative Adversarial Networks (GANs) and Graph Neural Networks (GNNs), offer a more powerful paradigm. They can learn implicit design rules and performance relationships directly from historical data, enabling the generation of diverse, high-performance structural schemes that meet complex, multi-objective constraints.

The conclusion of this review is explained: The field is undergoing a fundamental shift that the research focus is moving away from local, isolated efficiency gains toward a holistic, system-level integration that considers the entire building lifecycle, including economic cost, structural safety, physical performance, constructability, and carbon emissions.

Despite the promise of modern AI, the review identifies critical gaps. Most current research remains in the simulation or experimental stage, lacking validation in real-world construction environments with their inherent complexities and economic constraints. Furthermore, modern AI models often require vast amounts of data, which is scarce in the specialized domain of modular construction.

To bridge these gaps, four key future research directions are proposed:

(1) Develop hybrid intelligent paradigms that combine structured engineering knowledge with data-driven methods;

(2) Closed-loop verification and process integration for engineering practice by embedding AI tools into real-world engineering workflows and computer-aided platforms, enabling a feedback loop where models self-optimize based on actual project performance data;

(3) Develop more advanced intelligent decision-support methods, to simultaneously considers economic cost, structural safety, physical performance, constructability, and carbon emissions across the full project lifecycle;

(4) Introduce more AI-enhanced form-finding methods combined with modular constructions through synergistic integration of modern AI with classic physics-based approaches such as the Force Density Method (FDM) and Combinatorial Equilibrium Modeling (CEM).

This review is expected to provide researchers and practitioners with a useful reference for understanding the current state and future pathways of intelligent generative design in steel modular construction.

This paper was published in Smart Construction (ISSN: 2960-2033), a peer-reviewed open access journal dedicated to original research articles, communications, reviews, perspectives, reports, and commentaries across all areas of intelligent construction, operation, and maintenance, covering both fundamental research and engineering applications. The journal is now indexed in Scopus, and article submission is completely free of charge until 2026.

Liu Y, Zhou K, Chen Z, Liu J, Liu H. Research development in intelligent generative design methods for steel modular structures. Smart Constr. 2026(1):0005, https://doi.org/10.55092/sc20260005.