Predicting fiber‐reinforced polymer–concrete bond strength using artificial neural networks: A comparative analysis study

Abstract

The repair efficiency of fiber‐reinforced polymer (FRP) is crucially linked to bond strength between FRP and concrete. Artificial neural networks (ANNs) technique is employed for the prediction of FRP–concrete bond strength based on more than 440 data points collected from literature work for training and testing of the proposed ANNs model. Such a model facilitates investigating the effect of various key parameters in controlling the bond. These are concrete compressive strength, maximum aggregate size, FRP thickness and modulus of elasticity, FRP‐to‐concrete length and width ratios, and adhesive tensile strength. The proposed ANNs model shows high fitting and prediction capability of training and testing data, respectively, with low mean square errors. Its accuracy of prediction far exceeds that of literature empirical models. Furthermore, the present comparative and sensitivity study of the predicted bond strength promotes the understanding of the impact of the above key parameters.

Experimental Investigation and Artificial Neural Network Based Prediction of Bond Strength in Self-Compacting Geopolymer Concrete Reinforced with Basalt FRP Bars

“…The only guidelines available for FRP bars in concrete-namely, the ACI [22,59] and CSA [27] guidelines-do not include basalt FRP bars or geopolymer concrete. The current practice is to use the same coefficients for glass FRP and BFRP bars, which creates non-conformity owing to differences in the material properties of glass and basalt fibres [18,61,62]. Numerous studies have focused on finding accurate coefficients corresponding to the type of FRP bar, material, and surface properties in order to produce an accurate prediction model [31,52,63].…”

Research on the safety assessment of RC workshop buildings in earthquake site based on the fuzzy comprehensive evaluation

“…The indexes of highperformance concrete are often better than those of ordinary concrete. [36][37][38][39][40][41] Therefore, if the high-performance concrete is used to column, earthquake damage to the column will be greatly reduced. Taking the workshop column and the plaster layer as an example, the detail categories and the corresponding detail seismic damage descriptions are shown in Table 1.…”

“…In general, the column is made of normal concrete rather than high‐performance concrete. The indexes of high‐performance concrete are often better than those of ordinary concrete 36–41 . Therefore, if the high‐performance concrete is used to column, earthquake damage to the column will be greatly reduced.…”

Prediction of compressive strength and ultrasonic pulse velocity of admixtured concrete using tree model M5P

“…In the recent years, diverse methods of soft computing in civil engineering have drawn heed of numerous researchers. These methods were used to predict the slump, compressive strength, crack propagation, and elastic modulus of various types of concrete [16–24]. Milicević et al [20] compared the formulae given in the code provisions for estimating the concrete elastic moduli for lightweight and normal‐weight concretes with values obtained experimentally.…”

“…Wang et al [21] analyzed the main factors influencing early strength of mortar by gray correlation analysis and established a neural network model to predict the early strength value and relative error of steam curing mortar. Haddad and Haddad [23] predicted FRP–concrete bond strength by using artificial neural networks (ANNs) technique with data points collected from literature work. Cvetkovic [24] proposed a novel approach with available data using adaptive neurofuzzy inference system to predict effect of cement, blast furnace slag, fly ash, water, superplasticizer, coarse aggregate, fine aggregate, and age on concrete strength.…”