Experimental evaluation and modeling of drying shrinkage behavior of metakaolin and calcined kaolin blended concretes

Highlights

• Concretes were produced with commercial metakaolin and four types calcined kaolins.

• Free shrinkage and weight loss measurements were carried out over 60 days of drying period.

• Noticeable enhancement in shrinkage behavior of concretes was observed.

• Mathematical models from genetic programming and linear regression were derived.

• The models yielded comparable predictions of shrinkage strain to actual ones.

Abstract

In the first stage of the study presented herein, the findings of an experimental study on drying shrinkage behavior of concretes incorporated with high reactivity commercial metakaolin (MK) and calcined kaolins (CKs) were reported. Free shrinkage strain measurements as well as corresponding weight loss were measured over 60 days of drying. Four different types of kaolins obtained from local sources were calcined and used as mineral admixture for concrete production. Moreover, commercial metakaolin of high purity was also used as reference material for comparison. In the second stage of the study, prediction models through gene expression programming (GEP) and multiple linear regression (MLR) were derived. The data set used for training and testing covers the experimental data presented in this study as well as additional ones collected from the literature. The parameters considered for developing the prediction model are related to the characteristic properties of mineral admixture, concrete composition, and drying period. As a result, CK incorporated concretes revealed comparable performance with MK incorporated ones in terms of drying shrinkage and weight loss. Furthermore, the prediction models yielded strong correlation with the experimental results. Statistical analyses also revealed that the proposed models can be handful tools in predicting the drying shrinkage strain of the concretes modified with MK.

Introduction

Time dependent deformation of the concrete can be considered as one of the most detrimental effect leading to deterioration of concrete [1]. The volumetric reduction of the concrete due to shrinkage is generally obstructed by either internal or external restraints resulting in tensile stresses [2]. Once exceeding a certain limit, the induced tensile stress causes the concrete to crack [3], [4], [5], [6], [7], [8]. These facts have directed the researchers to investigate the ways for mitigating the effect of drying shrinkage of concrete. Utilization of mineral admixtures for this purpose is the most popular one [8], [9], [10], [11], [12], [13], [14]. For example, in the study of Al-Khaja [9], it was reported that the shrinkage and creep of plain concrete were significantly or moderately diminished with the incorporation of silica fume, revealing a 1 month reduction in strain of 34.9% and 18.5% for shrinkage and creep, respectively. This provided a total reduction of 20.8% in deformation. In the study of Jianyong and Yan [10], it was also shown that ultrafine ground granulated blast-furnace slag and silica fume can importantly contribute to the hydration of cement and rise in the quantity of AFt crystal hydrates and C–S–H gel hydrates in cement paste, which offers hardened concrete a stronger structure and higher resistance to deformation caused by applied force. Moreover, these two binders may also fill small pores and voids which are harmful to the structure of concrete. That may be considered as the mechanism of reducing effect of ultrafine mineral admixtures (i.e. GGBS, SF) on creep and drying shrinkage of concrete.

Being one of the most popular mineral admixtures high reactivity metakaolin has been used to improve the mechanical and durability properties of the concrete for last two decades [12], [13], [14], [15], [16], [17], [18], [19], [20], [21]. The improvement of shrinkage behavior of concretes with MK modification was also previously reported [12], [13], [16], [17]. For example, Brooks and Megat-Johari [17] studied the effect of MK on the creep and shrinkage of properties of concrete mixtures. The replacement level of MK was 0%, 5%, 10% and 15% by weight of cement. According to their results the total shrinkage (autogenous plus drying shrinkage) measured from 24 h was reduced by the use of MK, while the drying shrinkage was significantly less for the MK concretes than for the control concrete. In the study of Güneyisi et al. [12], it was revealed that MK incorporated concretes demonstrated similar trends as silica fume (SF) concretes in terms of free shrinkage strain development as well as formation and propagation of shrinkage cracks in restrained samples. They reported that both SF and MK containing concretes had far better performance than control concrete during overall drying period.

Metakaolins used in the aforementioned studies are all obtained from calcining high purity kaolins of which kaolinite contents are approximately 95 ± 5%. On the other hand, it was pointed out that non purified ground kaolins having relatively lower kaolinite content can also be benefited as a mineral admixture after a proper thermal treatment [22], [23], [24]. In the literature, there has not yet been a study on the shrinkage behavior of the concretes including such calcined kaolins.

In this study, the effects of calcined kaolins obtained from different types of kaolins on the drying shrinkage and weight loss of the concretes were investigated. Four different types of kaolins from western region of Turkey were obtained and subjected to a thermal treatment to obtain calcined kaolins for concrete production. In order to evaluate the performance of the produced mineral admixtures, high purity commercial MK from Czech Republic was also utilized. Two different replacement levels, namely 5% and 15%, were considered for concrete production. A reference plain concrete was also designed for comparison. The development of the free shrinkage strains and corresponding weight loss of the concretes were monitored for 60 days of drying period. After completion of the experimental study, the obtained data were combined with the previously reported experimental data from the literature to obtain a data set for derivation of prediction models from gene expression programming (GEP) and multiple linear regression (MLR). The database was divided into two sections as training and testing databases. Training database was used to construct prediction models while testing database was employed for evaluation of the performance of the proposed models. Statistical analyses were also performed to assess the accuracy of the prediction models.