Automating the seismic-resilient design of fiber-reinforced concrete using a physics-informed multi-agent system

Abstract

The design of dynamically resilient concrete materials remains a complex, fragmented process that depends on iterative modelling, expert judgment, and poorly integrated workflows spanning structural analysis, material formulation, and seismic performance evaluation. To address this challenge, we develop an autonomous multi-agent system (MAS) that optimizes the life cycle design of fiber-reinforced concrete (FRC) for earthquake resilience. The system focuses on shear walls, beams, and columns that constitute the seismic load bearing envelope of multi-storey buildings. Rather than functioning as a general-purpose structural tool, the MAS is designed to intelligently generate FRC specifications, tailoring fiber type, geometry, and reinforcement ratios according to seismic inputs and project constraints that are automatically parsed by the system. The framework integrates agentic artificial intelligence (AgenAI) via an Agent2Agent (A2A) protocol with physical artificial intelligence (PhysAI) implemented through physics informed stochastic models for micromechanical prediction and energy dissipation damage analysis. A modified maximum entropy principle models fiber matrix interactions under heterogeneous uncertainty, enabling unbiased optimization of energy dissipative properties. In parallel, a CNN based computer vision pipeline employing a 3D U-Net architecture performs accurate segmentation and interpretation of micro-CT data. The architecture comprises five specialized AI agents built on Qwen-30B-A3B and Phi-4-14B large language models (LLMs), with context grounding provided by agentic retrieval augmented generation (ARAG) for domain specific decision making. Experimental validation shows that the system reduces human intensive design cycles by 75% relative to conventional finite element analysis workflows. It achieves 89.2% reasoning adherence, a mean predictive bias below 4% compared with stochastic simulations, and a failure state classification F1 score of 0.98. These results bridge micromechanical realism and automated, code compliant seismic design.

Keywords

Agentic artificial intelligence (AgenAI) / Physical artificial intelligence (PhysAI) / Agent2Agent protocol (A2A) / Physics-informed multi-agent system / Agentic retrieval-augmented generation (ARAG) / Stochastic-driven seismic resilient design

References

[1] ACI Committee 318. (2019). Building code requirements for structural concrete (ACI 318–19) and commentary. American Concrete Institute. https://www.concrete.org/store/productdetail.aspx?ItemID=31819

[2] ACI Committee 544. (2018). Guide to design with fiber-reinforced concrete (ACI 544.4R-18). American Concrete Institute. https://www.concrete.org/store/productdetail.aspx?ItemID=544418

[3] ASCE. (2022). Minimum design loads and associated criteria for buildings and other structures (ASCE/SEI 7-22). American Society of Civil Engineers. https://doi.org/10.1061/9780784415788

[4] Ascona PP, et al. . Intelligent automated monitoring and curing system for cracks in concrete elements using integrated sensors and embedded controllers. Technologies, 2025, 13(7 284

[5] Atahan AO, Yücel AÖ. Crumb rubber in concrete: Static and dynamic evaluation. Construction and Building Materials, 2012, 36: 617-622

[6] CEN. (2004). Eurocode 8: design of structures for earthquake resistance—part 1: general rules, seismic actions and rules for buildings (EN 1998-1). European Committee for Standardization.

[7] Chen Q, et al. . A stochastic micromechanical model for fiber-reinforced concrete using maximum entropy principle. Acta Mechanica, 2018, 229: 2719-2735

[8] Chen Z, Lai S-K, Yang Z. AT-PINN: Advanced time-marching physics-informed neural network for structural vibration analysis. Thin-Walled Structures, 2024, 196 111423

[9] Cornejo A, Jiménez S, Barbu LG, Oller S, Oñate E. A unified non-linear energy dissipation-based plastic-damage model for cyclic loading. Computer Methods in Applied Mechanics and Engineering, 2022, 400 115543

[10] Ehtesham, A., Singh, A., Gupta, G. K., & Kumar, S. (2025). A survey of agent interoperability protocols: Model context protocol (MCP), Agent Communication Protocol (ACP), Agent-to-Agent Protocol (A2A). arXiv. https://doi.org/10.48550/arXiv.2505.02279

[11] Farea A, Yli-Harja O, Emmert-Streib F. Understanding physics-informed neural networks: Techniques, applications, trends, and challenges. AI, 2024, 5: 1534-1557

[12] fib (International Federation for Structural Concrete). (2013). fib Model Code for Concrete Structures 2010. Lausanne: Ernst & Sohn. https://doi.org/10.1002/9783433604090

[13] González DC, et al. . Size effect of steel fiber–reinforced concrete cylinders under compressive fatigue loading: Influence of the mesostructure. International Journal of Fatigue, 2023, 167 107353

[14] Guan X, et al. . A stochastic multiscale model for predicting mechanical properties of fiber reinforced concrete. International Journal of Solids and Structures, 2015, 56–57: 280-289

[15] Henkes A, Wessels H, Mahnken R. Physics-informed neural networks for continuum micromechanics. Computer Methods in Applied Mechanics and Engineering, 2022, 393 114790

[16] Hu Y, Wu L, Li N, Zhao T. Multi-agent decision-making in construction engineering and management: A systematic review. Sustainability, 2024, 1616 7132

[17] Karniadakis GE, Kevrekidis IG, Lu L, Perdikaris P, Wang S, Yang L. Physics-informed machine learning. Nature Reviews Physics, 2021, 3: 422-440

[18] Lv J, Zhou T, Du Q, Wu H. Effects of rubber particles on mechanical properties of lightweight aggregate concrete. Construction and Building Materials, 2015, 91: 145-149

[19] Martinez Y, Rojas L, Peña A, Valenzuela M, Garcia J. Physics-informed neural networks for the structural analysis and monitoring of railway bridges: A systematic review. Mathematics, 2025, 1310): 1571

[20] Najim KB, Hall MR. Mechanical and dynamic properties of self-compacting crumb rubber modified concrete. Construction and Building Materials, 2012, 27: 521-530

[21] Nonato Da Silva CA, et al. . A multiscale model for optimizing the flexural capacity of FRC structural elements. Composites Part b: Engineering, 2020, 200 108325

[22] Nowacka, A., Schladitz, K., Grzesiak, S., & Pahn, M. (2025). Segmentation of cracks in 3D images of fiber reinforced concrete using deep learning. arXiv preprint. https://arxiv.org/abs/2501.18405

[23] Pan L, Hao H, Cui J, Pham TM. Numerical study on dynamic properties of rubberised concrete with different rubber contents. Defence Technology, 2023, 24: 228-240

[24] Parisi F, et al. . On the use of mechanics-informed models to structural engineering systems: Application of graph neural networks for structural analysis. Structures, 2024, 59 105712

[25] Singh, A., Ehtesham, A., Kumar, S., & Talaei Khoei, T. (2025). Agentic retrieval-augmented generation: A survey on agentic RAG. arXiv. https://doi.org/10.48550/arXiv.2501.09136

[26] Wang C, Wu H, Li C. Hysteresis and damping properties of steel and polypropylene fiber reinforced recycled aggregate concrete under uniaxial low-cycle loadings. Construction and Building Materials, 2022, 319 126191

[27] Wang X, Shao J, Wang J, Ma M, Zhang B. Influence of basalt fiber on mechanical properties and microstructure of rubber concrete. Sustainability, 2022, 14: 12517

[28] Wu H, Qin X, Huang X, Kaewunruen S. Engineering, mechanical and dynamic properties of basalt fiber reinforced concrete. Materials, 2023, 16: 623

[29] Xu X, Liu C. Physics-guided deep learning for damage detection in CFRP composite structures. Composite Structures, 2024, 331 117889

[30] Yousefi Maragheh, R., Vadala, P., Gupta, P., Zhao, K., Inan, A., Yao, K., Xu, J., Kanumala, P., Cho, J., & Kumar, S. (2025). ARAG: agentic retrieval augmented generation for personalized recommendation. arXiv. https://doi.org/10.48550/arXiv.2506.21931

[31] Zaitsev A, et al. . Effect of short basalt fibers on energy-dissipating properties of lightweight rubberized concrete shear wall. Sustainable Structures, 2024, 42 000045

[32] Zhang Q, Wang L. Investigation of stress level on fatigue performance of plain concrete based on energy dissipation method. Construction and Building Materials, 2021, 269 121287

[33] Zheng R, Pang J, Sun J, Su Y, Xu G. Damage model of carbon-fiber-reinforced concrete based on energy conversion principle. Journal of Composites Science, 2024, 8 71

[34] Zhu HH, et al. . Maximum entropy-based stochastic micromechanical model for a two-phase composite considering the inter-particle interaction effect. Acta Mechanica, 2015, 226: 3069-3084

RIGHTS & PERMISSIONS

The Author(s)