Artificial neural network modeling of stress in concrete under step-loading using non-linear ultrasonic test results
Date of Publication
11-2016
Document Type
Master's Thesis
Degree Name
Master of Science in Civil Engineering
Subject Categories
Civil Engineering
College
Gokongwei College of Engineering
Department/Unit
Civil Engineering
Thesis Adviser
Jason Maximino C. Ongpeng
Defense Panel Chair
Cheryl Lyne Capiz Roxas
Defense Panel Member
Ronaldo S. Gallardo
Ma. Klarissa G. Martinez
Abstract/Summary
This study deals with the prediction of stress in concrete using the non-linear ultrasonic test and artificial neural network (ANN). Data were obtained from 36 cube specimens and 27 beam specimens subjected to step loading. For the cubes, the ordinary concrete (ORC), fiber reinforced concrete (FRC) and concrete with varying size of aggregate were studied. For the reinforced concrete beams, 7 types of concrete were investigated with varying amount of reinforcements and water to cement ratio (w/c). The fundamental harmonic amplitude (A1), second harmonic amplitude (A2), third harmonic amplitude (A3), strain/neutral axis (NA) and peak to peak amplitude (PPA) were found to be the significant input parameters for ANN based on the results of the Spearman’s rank order correlation. The optimum models were determined based on the Pearson correlation coefficient (R), mean square error (MSE) and soundness of the behavior of the input parameter with the stress of the concrete. The Daponte’s amplitude sensitivity (DA) was used in analyzing the result of the parametric study. Results of the sensitivity analysis show that for the ORC and FRC, the A2 and strain were long range sensitive for all w/c. The A3 decreased its sensitivity as the water to cement ratio was increased. For the size of aggregate study, PPA and A3 decreased their sensitivity as the size of aggregate increases. A1 was only sensitive for small aggregate concrete. The A2 and strain were long range sensitive and varying the size of aggregate did not affect their sensitivity. In the study of reinforced concrete beams, the PPA was sensitive for all types of concrete except WC40B. The A1 was very sensitive to the load having long range sensitivity for all types of concrete except WC40A. The A2 and NA were sensitive for all types of concrete. Lastly, A3 decreased its sensitivity as the water to cement ratio increased. The specific damping capacity (S) of concrete was also investigated and ANN models were produced using the load, neutral axis and loading branch as the input parameters to predict the damping in concrete using the energy method. The highest magnitude for S occurs at the first load of the concrete. S would decrease in magnitude when repetitive load was applied. When a higher load was introduced, a new peak was observed. The results of this study highlight and improve the non-destructive evaluation capabilities of non-linear ultrasonic test.
Abstract Format
html
Language
English
Format
Electronic
Accession Number
CDTG007057
Shelf Location
Archives, The Learning Common's, 12F Henry Sy Sr. Hall
Physical Description
1 computer optical disc; 4 3/4 in.
Keywords
Concrete—Testing; Neural networks (Computer science); Nonlinear mechanics
Upload Full Text
wf_no
Recommended Citation
Soberano, M. C. (2016). Artificial neural network modeling of stress in concrete under step-loading using non-linear ultrasonic test results. Retrieved from https://animorepository.dlsu.edu.ph/etd_masteral/7269
Embargo Period
11-8-2024