Date of Award

Spring 2011

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computational Analysis and Modeling

First Advisor

Jinko Kanno

Abstract

The focus of this research is to model and provide a simulation framework for the packing of differently sized spheres within a hard boundary. The novel contributions of this dissertation are the cylinders of influence (COI) method and sectoring method implementations. The impetus for this research stems from modeling electrokinetic nanoparticle (EN) treatment, which packs concrete pores with differently sized nanoparticles. We show an improved speed of the simulation compared to previously published results of EN treatment simulation while obtaining similar porosity reduction results. We mainly focused on readily, commercially available particle sizes of 2 nm and 20 nm particles, but have the capability to model other sizes. Our simulation has graphical capabilities and can provide additional data unobtainable from physical experimentation.

The data collected has a median of 0.5750 and a mean of 0.5504. The standard error is 0.0054 at α = 0.05 for a 95% confidence interval of 0.5504 ± 0.0054. The simulation has produced maximum packing densities of 65% and minimum packing densities of 34%. Simulation data are analyzed using linear regression via the R statistical language to obtain two equations: one that describes porosity reduction based on all cylinder and particle characteristics, and another that focuses on describing porosity reduction based on cylinder diameter for 2 and 20 nm particles into pores of 100 nm height.

Simulation results are similar to most physical results obtained from MIP and WLR. Some MIP results do not fall within the simulation limits; however, this is expected as MIP has been documented to be an inaccurate measure of pore distribution and porosity of concrete. Despite the disagreement between WLR and MIP, there is a trend that porosity reduction is higher two inches from the rebar as compared to the rebar-concrete interface. The simulation also detects a higher porosity reduction further from the rebar. This may be due to particles aggregating before reaching the rebar that can easily be seen in the graphical representation of the simulation cylinders.

The dissertation author has created a web based framework to allow an interdisciplinary team to work in concert with access to the simulation and the results generated. The results are stored into a MySQL database. The database currently holds 271 simulation runs. Simulation requests can be entered into a web interface and will automatically be processed in the order entered and the results stored into the database. Results can also be retrieved from the database and filtered based on any simulation parameter. Statistical analysis can be completed on the data points stored in the database by using version of Rweb modified by the dissertation author. The result is a collaborative framework that can be extended to address future investigations into pore packing and chloride blocking.

Share

COinS