Bond of HPC and reinforcing steel under fatigue loading (data collection)
Table of contents
Project data
Titel | Title Verbundverhalten von HPC und Bewehrungsstahl unter Ermüdungsbeanspruchung (Datengrundlage für Vortrag 7. IPW, Delft) | Bond behaviour of HPC and reinforcing steel under fatigue loading (data base for lecture at 7th IPW, Delft) Förderer | Funding Institut für Massivbau, TU Dresden | DAfStb Zeitraum | Period 2009 Leiter | Project manager Prof. Dr.-Ing. Manfred Curbach Bearbeiter | Contributor Dr.-Ing. Lars Eckfeldt |
Report in the yearbook 2009
Biostatistics and Bond Fatigue
Before high performance concrete (HPC) could formally be approved for use in structural members under fatigue loading, (e.g., in bridge engineering), it was necessary to clarify whether or not premature fatigue failure could be attributed to the brittleness of the material. Several series of pullout-tests (POT) were developed and conducted for short bond lengths to investigate this notion during the initial stage of this research project.
Specimens in these tests were subjected to static loading and constant, highcyclical, tensile fatigue loading until failure or
1 million load-cycles, whichever occurred first. Tests were conducted on bars of different sizes having a variety of concrete covers. This testing regime also required investigation into different amplitudes and mean stress levels selected based on the ultimate bond strength. The upper stress bounds chosen for fatigue loading were 55 %, 70 % or 85 %, respectively. The test results led to the conclusion that the bond fatigue behaviour of HPC was not significantly different than that of normal strength concrete (NSC).
Biostatistics Methods for Analysis
The analysis and model building were generally hindered to a great extent by: limited levels of multiple factors, fatigue run-outs of over 1 million loadcycles; and the isolated censoring that occurs after unexpected failures of steel reinforcing. Methods borrowed from survival analysis in biostatistics allowed for more in-depth research into parametric effects under these circumstances. Application of this model required translation of the original time axis into the number of load-cycle axis which is more suitable, and the addition of censoring or truncation status. Original longitudinal data also needed to be randomized and transformed into data duples that indicated quasi-independent, sampled, load-cycles with the corresponding status of survivability falling within the range of 30 to 1 million load-cycles. Functions, direct referencing to survivability, were produced using logistic regression analysis to develop corresponding Smith- and Wöhler-diagrams. Kaplan-Meier survival estimates helped to evaluate impacted of censoring. The Cox-PH Model in the figure indicates that the improved material properties of HPC (i. e., tensile strength, modulus of elasticity) actually can lead to higher mean survival rates. Higher damage rates occur when the lower stress bound is relatively high and the upper bound is held constant.
Publication
Eckfeldt, L.; Curbach, M.: Application of Biostatistics in the Analysis of Fatigue Data in Bond Research. Proceedings of the 7th International Probabilistic Workshop (IPW), Delft, The Netherlands, 25.– 26.11.2009, 26 pages