A physics-informed and data-enhanced tensile stress-strain model for UHPFRC

Wei He Liu, Lu Wen Zhang, Jian Guo Dai

Research output: Journal article publicationJournal articleAcademic researchpeer-review

1 Citation (Scopus)

Abstract

Despite the rapid developments in fundamental investigations and engineering applications of ultra-high-performance fiber reinforced concrete (UHPFRC), there is still lacking of a reliable tensile stress-strain model for UHPFRC in design guidelines. A generalized tensile stress-strain model for UHPFRC was developed for the first time. Through properly identifying unified model parameters, widely acknowledged experimental results were successfully reproduced by using a one-dimensional finite element model (FEM). A rich database was generated and granted with physics by the FEM model. Physical-consistent strength, ultimate strain and stress-strain models of UHPFRC were proposed, trained by model-generated data, and enhanced by experimental data. The proposed strength model and ultimate strain model predicted extensive experimental results with reasonable accuracy, giving mean absolute percentage errors (MAPE) of 12% and 25.3%, respectively. The established stress-strain model also predicted satisfactorily the full-range stress-strain curves tested by different research groups. It was evidenced that higher mean matrix cracking strength leads to higher ultimate strengths, less cracks, higher crack widths of UHPFRC at the ultimate state. This was elaborated for the first time, as caused by the dual action of snubbing effects and multi-crack interactions.

Original languageEnglish
Article number115989
JournalEngineering Structures
Volume285
DOIs
Publication statusPublished - 15 Jun 2023

Keywords

  • Generalized stress-strain model
  • Multiple crack interactions
  • One-dimensional FEM
  • Physics-informed data-enhanced
  • Snubbing effect
  • Ultra-high-performance fiber reinforced concrete

ASJC Scopus subject areas

  • Civil and Structural Engineering

Fingerprint

Dive into the research topics of 'A physics-informed and data-enhanced tensile stress-strain model for UHPFRC'. Together they form a unique fingerprint.

Cite this