The aim of the current study is to recommend and compare the estimates of finite element model (FEM), analytical model, and artificial neural networks (ANN) model for capturing the LCC of FCSC members. A database comprising 325 FCSC columns was constructed from previous studies to propose FEM and ANN models while the analytical model was proposed based on a database of 712 samples and encasing mechanics of steel tube and FRP wraps. The concrete damage plastic model was used for concrete along with bilinear and linear elastic models for steel tube and FRP wraps, respectively. Analytical and ANN models effectively considered the lateral encasing mechanism of FCSC columns for accurate predictions.
The study aimed to compare the prediction accuracy of finite element (FEM), analytical, and artificial neural network (ANN) models for the load-carrying capacity (LCC) of fiber reinforced polymer (FRP)-encased concrete-filled steel tube (CFST) compression members (FCSC). A database of 325 FCSC columns was developed for FEM and ANN models, while the analytical model was based on 712 samples, utilizing encasing mechanics of steel tube and FRP wraps. FEM used a concrete damage plastic model, bilinear steel tube, and linear elastic FRP models. Statistical accuracy was evaluated using MAE, MAPE, R², RMSE, and a 20-index across all models.
Based on the experimental database, the FEM presented the accuracies in the form of statistical parameters MAE = 223.76, MAPE = 285.32, R2 = 0.94, RMSE = 210.43 and a20-index = 0.83. The analytical model showed the statistics of MAE = 427.229, MAPE = 283.649, R2 = 0.8149, RMSE = 275.428 and a20-index = 0.73 while ANN models portrayed the predictions with MAE = 195, MAPE = 229.67, R2 = 0.981, RMSE = 174 and a20-index = 0.89 for the LCC of FCSC columns.
Although various investigations have already been performed on the prediction of the load-carrying capacity (LCC) of fiber reinforced polymer (FRP)-encased concrete-filled steel tube (CFST) compression members (FCSC) using small and noisy data, none of them compared the accuracy of prediction of different modeling techniques based on a refined large database.
