Modeling the properties of concrete prepared with recycled aggregates derived from different sources

Pao Yue-kong Library Electronic Theses Database

Modeling the properties of concrete prepared with recycled aggregates derived from different sources

 

Author: Duan, Zhenhua
Title: Modeling the properties of concrete prepared with recycled aggregates derived from different sources
Degree: Ph.D.
Year: 2015
Subject: Aggregates (Building materials) -- Recycling.
Concrete -- Recycling.
Hong Kong Polytechnic University -- Dissertations
Department: Dept. of Civil and Environmental Engineering
Pages: xxxii, 256 pages : color illustrations ; 30 cm
Language: English
InnoPac Record: http://library.polyu.edu.hk/record=b2806845
URI: http://theses.lib.polyu.edu.hk/handle/200/7923
Abstract: Along with the rapid urban development and economic activities, the generation of construction and demolition (C&D) waste has increased substantially in many parts of the world, in particular in China. In Hong Kong, it has been estimated that the annual generation of C&D waste could be as much as 21 million tons. If not managed properly, such a huge amount of waste will bring significant environmental problem. Since the landfill sites in Hong Kong will be saturated in several years’ time, it is important to find a viable way to reuse these waste materials. At the same time, there is critical shortage of natural aggregate (NA) in Hong Kong for the production of new concrete. The concrete industry globally consumes 8-12 billion tons annually of NA, among them about 4 billion tons are consumed in China. The extraction of NA, such as crushed rock and river sand, has significant impact on the environment. Therefore, the concrete industry is exploring ways to utilize C&D waste for concrete production in order to achieve sustainable development. Research studies conducted on the recycled aggregate (RA) and the reuse of them in new concrete have found that the properties of RA are generally weaker than those of NA due to the old mortar attached to the RA, reflected in more angular shape, lower bulk and saturated surface dried (SSD) densities, higher water absorption (Wa), inferior strength, presence of contaminants (e.g. ceramic) and lower resistance to mechanical and chemical actions when compared with NA. Besides, the properties of concrete made with RA are generally found inferior to those concrete made with only NA due principally to their high absorption and low density, which has hindered its use in the production of concrete in practical application, especially for the concrete with the requirement of durability. It is encouraging that a large number of literatures have proved that through adopting alternative design and production measures, the properties of concrete made with RA can be comparable with those made with NA. The findings are helpful for the reuse of RA in new concrete. However, these finding are generally obtained under a relatively ideal situation in the laboratory without considering the case that RAs are generally derived from different sources with vastly different properties. Therefore methods should be developed to provide estimations of the properties of recycled aggregate concrete (RAC) made with different sources and types of RAs. The aim of this study is to develop a scientific approach for the better prediction of the properties of RAC made with RAs derived from different sources, and a large experimental programme is conducted to verify the validity of the approach.
Firstly, data from different published literatures worldwide were collected as the sample data to construct respective artificial neural networks (ANN) models for predicting the compressive strength and elastic modulus. For each model, factors that may influence the concrete properties were firstly selected and the collected sample data were divided randomly into 3 groups as the training, testing and validation sets, respectively. The data number for the latter two sets is no less than 25% of the total data. This helped to provide the established models with generalization abilities. After training, the optimal models for simulating compressive strength and elastic modulus were constructed, respectively, and the network architecture and parameters were also determined. Sensitivity analyses were then made to examine the importance of the selected factors, as well as determine which combination of factors could be used to construct the best model. Then three groups with a total of 46 RAC concrete mixes were prepared to examine the effect of different RAs on the properties of RAC. The RAs used were categorized into 3 groups: (1) RAs derived from laboratory prepared concrete cubes with different compressive strength (35-85 MPa); (2) RAs derived from 3 different sources and crushed by different methods; (3) RAs contained different amounts of added masonry (clay bricks or tiles). As many sources of NAs and RAs are used in these mixes, the aggregate characteristics, such as the fineness modulus (FM) of the fine aggregate, mortar content (MC), 10% fines value (TFV), Aggregate crushing value (ACV), water absorption, SSD specific gravity (SG), impurity (δ) and masonry (m) contents of the coarse aggregate (CA), were quantified. Besides, the mechanical properties of the hardened concrete like compressive strength (fcu) and splitting tensile strength (ftc), elastic modulus (Ec) and durability properties like drying shrinkage and resistance to chloride ion (Cl-) penetration were also investigated to examine the influence of different qualities or sources, or different masonry contents of RAs on the properties of RAC. The experimental results of the above mixes were also used as Cases to test the applicability of the constructed ANN models. Based on the experimental test results, great variations were noticed in both the properties of RAs derived from different sources and those of the produced RAC. Also, the traditional relationships established for NAC were found no longer suitable for use in RAC. The results also indicated that, by constructing ANN models using data collected from many international literatures as sample data, the compressive strength and elastic modulus of RAC made with RAs from different sources could be modeled accurately, with the mean absolute percentage error (MAPE) values all in the range of 5.8%-6.6%. Besides, it was demonstrated that ANN could be also used to determine the relative importance of the factors in affecting the performance of RAC. It was shown that for compressive strength prediction, cement type and specimen size were the most important parameters, and aggregate moisture condition was the most influential parameter amongst all the aggregate characteristics. For elastic modulus prediction, although cement type still played an important part, the characteristics of the aggregates like types of natural and recycled aggregates used were also critical.

Files in this item

Files Size Format
b28068452.pdf 12.21Mb PDF
Copyright Undertaking
As a bona fide Library user, I declare that:
  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

     

Quick Search

Browse

More Information