Prominence of concrete is characterized by its high mechanical properties and durability, combined with multifunctionality and aesthetic appeal. Development of alternative eco-friendly or multipurpose materials has conditioned improvements in concrete mix design to optimize concrete production speed and price, as well as carbon footprint. Artificial neural networks represent a new and efficient tool in achieving optimal concrete mixtures according to its intended function. This paper addresses concrete mix design and the application of artificial neural networks (ANNs) for self-sensing concrete. The authors review concrete mix design methods and the development of ANNs for prediction of properties for various types of concrete. Furthermore, the authors present developments and applications of ANNs for prediction of compressive strength and flexural strength of carbon nanotubes/carbon nanofibers (CNT/CNF) reinforced concrete using experimental results for the learning process. The goal is to bring the ANN approach closer to a variety of concrete researchers and possibly propose the implementation of ANNs in the civil engineering practice.

Carbon nanotube/concrete composite possesses piezoresistivity i.e. self-sensing capability of concrete structures even in large scale.

External bonding of fiber reinforced composites is currently the most popular method of strengthening building structures. Debonding performance is critical to the effectiveness of such strengthening. Many models of bond prediction can be found in the literature. Most of them were developed based on laboratory research, therefore, their accuracy with less popular strengthening systems is limited. This manuscript presents the possibility of using a model based on neural networks to analyze and predict the debonding strength of steel-reinforced polymer (SRP) and steel-reinforced grout (SRG) composites to concrete. The model is built on the basis of laboratory testing of 328 samples obtained from the literature. The results are compared with a dozen of the most popular analytical methods for predicting the load capacity. The prediction accuracy in the neural network model is by far the best. The total correlation coefficient reaches a value of 0.913 while, for the best analytical method (Swiss standard SIA 166 model), it is 0.756. The sensitivity analysis confirmed the importance of the modulus of elasticity and the concrete strength for debonding. It is also interesting that the width of the element proved to be very important, which is probably related to the low variability of this parameter in the laboratory tests.

Nanomaterials such as carbon nanotubes and carbon nanofibers are used as reinforcement for concrete, enhancing its compressive and flexural strength, and durability, and providing additional properties such as electrical conductivity. Hence, CNT/CNF reinforced concrete composite material is multifunctional material which may be used in structural capacity as well as structural health monitoring purposes. Although this material can be superior to traditional concrete, extensive and costly procedures of fabrication are hindering its practical potential. Concrete mix design methods are commonly used during the design of such composites, however, since these methods cannot give direct connection between the recipe and the end-product, every composite must be put through testing and iteratively adjusted until the appearance of wanted results. This paper proposes application of artificial neural networks for predicting properties of CNT/CNF concrete composite materials. Artificial neural networks in mix design have been developed for various types of concrete, commonly to predict only compressive strength as the primary property of concrete. However, self-sensing concrete is used primarily for its piezoresistivity and enhanced strength is only the consequence of the existence of nanofillers. Hence, the paper investigates prediction of compressive and flexural strength as well as electrical resistivity of 468 concrete mixtures by developing 3 different datasets comprehended by 6 ANN models. The models show some interesting results and point toward the necessity of further investigations on this topic and possible improvements.