Soft computing models to evaluate the effect of fly ash and ground granulated blast furnace slag (GGBS) on the compressive strength of concrete in normal and high strength ranges

• Published in: Structures (Oxford) Vol. 58; p. 105459
• Main Authors: Mohammed, Hba Xalid, Mohammed, Ahmed Salih, Hassan, A.M.T.
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.12.2023
• Subjects: Compressive strength range; Fly ash; GGBS; Modeling; Statistical analysis; GGBS; Fly ash; Compressive strength range; Statistical analysis; Modeling
• ISSN: 2352-0124; 2352-0124
• DOI: 10.1016/j.istruc.2023.105459

This study predicts the ultimate compression stress of concrete incorporated with Fly Ash and Ground Granulated Blast Furnace (GGBS) in the normal (NSC) and high (HSC) compressive strength ranges. The utilization of fly ash and GGBS in concrete has been found to enhance its compressive strength at varying stages of curing while also serving as a partial substitute for cement. The present investigation aimed to examine the impact of fly ash and GGBS on the compressive strength of cement-based concrete across diverse mix proportions.A total of 455 concrete mix proportions were examined, evaluated, and quantified for that purpose. The study includes independent parameters such as coarse aggregate (160 to 990 kg/m3), fine aggregate (105 to 1140 kg/m3), cement (56 to 702 kg/m3), fly ash (0 to 305 kg/m3), GGBS (0 to 360 kg/m3), water-to-cement ratio (0.19 to 2), superplasticizer content (0 to 10 kg/m3), and curing time (7 to 365 days). The variable under consideration is the compressive strength, which ranges from 20 to 100 MPa. This parameter is further classified into two categories: normal compressive strength (ranging from 20 to 55 MPa) and high compressive strength (exceeding 55 MPa). The present research investigates several soft computing models, including Linear Regression, Interaction, Artificial Neural Networks (ANN), and Pure Quadratic models, to predict the compressive strength of both normal-strength concrete (NSC) and high-strength concrete (HSC). The accuracy of the models was evaluated based on 249 data points collected from the literature using the Coefficient of Determination, Root Mean Squared Error (RMSE), Scatter Index, Objective (OBJ), Mean Absolute Error (MAE), t-test value, and U95 value. The ANN model demonstrated superior performance compared to all the other models with high accuracy. Finally, sensitivity analysis demonstrated that sand content and curing time are the most influential parameters for predicting NSC and HSC compressive strength.