Artificial neural network modeling of rheological parameters and compressive strength of self-compacting concrete with zeolite mineral as partial replacement for cement
Date of Publication
Master of Science in Civil Engineering
Gokongwei College of Engineering
Andres Winston C. Oreta
Defense Panel Chair
Bernardo A. Lejano
Defense Panel Member
Lessandro Estelito O. Garciano
Cheryl Lynne Capiz Roxas
The effects of natural zeolite to self-compacting concrete were studied. Natural zeolite was used as partial replacement for cement to increase the apparent viscosity of the concrete. Different rheological tests and compressive strength test were conducted. An artificial neural network models were derived from the tests results and converted to mathematical models. The derived models provide easier steps and neglecting rigorous trial and error method.
Artificial Neural Network (ANN) technique was used because of its predicting capability even in most complicated behavior of results. Different parameters yielded different rheological properties and compressive strength. Non-linear behavior was expected because different dosages produced unpredictable results. In ANN modeling, each parameter is considered significant to the output. A normalization technique in all data and results was also conducted to give each value an equal weight.
The experiments conducted included the influencing factors: SP (0.86%- 2.16% by weight of cement), Zeolite replacement for cement (0%-20% by weight of cement), water/cement ratio (0.30-0.6), sand content (46.04%-60.32% by bulk iv density), cement (408kg-553kg) and coarse aggregates (58.73%-71.20% by bulk density). The output data were rheological properties (viscosity, flowability, passing ability, resistance to segregation) and compressive strength. These rheological properties were based on different test results: T50 test for viscosity, slump flow test for flowability, J-Ring test for passing ability and GTM screen stability test for resistance to segregation. Compressive test was based only on the 28th day strength of concrete.
Although zeolite mineral decreases the compressive strength, it increases the viscosity of the concrete and decreases the possibility of segregation. Replacing up to 20% of cement powder with zeolite mineral decreased the bleeding of concrete caused by excess water and superplasticizer.
All models yielded outstanding results in terms of low mean-square error and Pearson Correlation R. Tests of models based on additional data showed excellent predicting capability of all models. The influence of each factor was also considered using parametric study to help concrete designers determine which factor is influential to each result.
Archives, The Learning Commons, 12F Henry Sy Sr. Hall
computer optical disc.
Clemente, S. (2015). Artificial neural network modeling of rheological parameters and compressive strength of self-compacting concrete with zeolite mineral as partial replacement for cement. Retrieved from https://animorepository.dlsu.edu.ph/etd_masteral/4959