Performance assessment of corroding RC beams using response surface methodology

Alexandros Kallias, Muhammad Rafiq

Research output: Contribution to journalArticlepeer-review


Rebar corrosion is the most frequent cause of deterioration in reinforced concrete (RC) structures. Corrosion damage can adversely affect the material and bond properties in RC, which may lead to premature in-service failures. Several studies have investigated the effects of corrosion on the residual performance of RC elements, such as under-reinforced beams; however a significant scatter is observed in the available experimental data. Lack of consistent interpretation of the available data is a major hurdle in reaching reliable conclusions on the behaviour of corroding members, leading to conservative estimation of their residual performance during assessment. In this paper, the performance of corroding under-reinforced beams is investigated using the response surface method (RSM), in which non-linear finite element analysis (NLFEA) serves as “numerical experiments”. Corrosion damage is taken into account using empirical/semi-empirical models derived from the experimental data. Low-order polynomial (RSM) models are obtained for the load at serviceability limit deflection, yield-load capacity, and displacement ductility factor (μδ) of the corroding beams, using suitable experimental designs. Results of an extensive parametric study are presented using the developed RSM models, which quantified the influence of several input variables on the predicted responses. These have assisted in identifying the reasons for the large scatter observed during the laboratory testing of deteriorating beams.
Original languageEnglish
Pages (from-to)671-685
Number of pages15
JournalEngineering Structures
Publication statusPublished - 1 May 2013


  • Rebar corrosion
  • Under-reinforced beams
  • FE analysis
  • Response surface methodology (RSM)
  • Performance assessment


Dive into the research topics of 'Performance assessment of corroding RC beams using response surface methodology'. Together they form a unique fingerprint.

Cite this