Open Collections

Applied computational analyses for concrete compressive strength performance assessment

University of British Columbia

The compressive strength of concrete represents a significant essential mechanical property, measured for any given concrete mixture. Traditional methods in form of destructive test methods are used in assessing the strength of concrete. However, concrete strength is influenced by numerous conditions such as its material constituents, mixture designs and environmental exposure which may be complex for the traditional methods. The use of the statistical approach for concrete mixture designs, advanced computational and machine learning approaches for estimating the concrete compressive strength, has received significant attention. However, previous studies have been limited to a single source and small laboratory-produced data sets used for the analysis of concrete. If the adequate nonlinear function is found for several categories and designs of concrete compressive strength condition assessments, then the prediction of concrete strength may tend to become automated. Thereby, reducing cost and the number of destructive testing done in the concrete. In this thesis, some soft computational techniques were employed to address these challenges. The first study investigates the possibility to explore concrete mixture design to produce favorable and optimal compressive strength results for situations where some experimental tests may be difficult to run due to challenges in obtaining certain constituents which may be expensive or not readily available. With intentions to reduce the need for preparing a large number of trial mixes to avoid material wastage, a simple statistical approach for concrete mixture design via a response surface method was proposed to overcome this drawback. In using experimental mixture design data specified for concrete of high performance, the optimization study demonstrated that the proposed method serves as a promising design approach in the concrete domain. The second study investigates the effects of concrete constituents and the constituents’ mixture proportioning in estimating the compressive strength of concrete. A section of this study further tests the sensitivity analysis on concrete relative to the variables used in evaluating its compressive strength. Firstly, measurements of the linear relationships between a series of variables with the compressive strength of concrete were obtained using their correlation coefficients. Thereafter, an ensemble meta-algorithm was employed to investigate the performance of concrete compressive strength with considered concrete features. Based on the prediction performance, the scalability and performance of the ensemble meta-algorithm were validated.

Text

2019-09-09

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/71654

Civil Engineering

University of British Columbia

UBCO

http://creativecommons.org/licenses/by-nc-nd/4.0/