School of Mechanical Engineering Seminar
Wednesday, March 18, 2015 at 15:00
Wolfson Building of Mechanical Engineering, Room 206
Offspring Creation Mechanisms for Multi-objective Evolutionary Algorithms
Mor Elias
MSc student of Dr. Amiram Moshaiov
Offspring Creation Mechanisms for Multi-objective Evolutionary Algorithms
Mor Elias
MSc student of Dr. Amiram Moshaiov
Evolutionary algorithms are designed to solve optimization problems using an iterative process. These algorithms are inspired by the biological evolution process. One of the main mechanisms used in evolutionary algorithms is the solution ('offspring') creation method. This study explores the influence of several different offspring creation methods on evolutionary multi-objective optimization. An innovative method of variable-based offspring creation is suggested and studied. It involves independent adaptation of the solution's parameters. The suggested approach is implemented in an algorithm which is termed Variable-based 
-PAES. Another mutation-based algorithm, -MO-CMA-ES (multi objective covariance matrix adaptation evolutionary strategy), is also proposed based on the well-known CMA-ES algorithm. The CMA-ES method takes under consideration the mutual influence of the solution's variables. These mutation-based algorithms are compared with each other and with a state-of-the-art algorithm, known as -MOEA, which uses an offspring creation method based on recombination.
A statistical comparison is performed on a variety of problems, using repeated runs. The comparison refers to the exploration and exploitation capabilities of the algorithms. First, the comparison is carried out on several test functions of two and three objectives. Next, it is performed on several academic path planning problems. The comparison results reveal several characteristics of the algorithms. It is shown that using a combination of multiple runs the ES-based algorithms are better than -MOEA.
School of Mechanical Engineering Seminar
Wednesday, March 18, 2015 at 15:00
Wolfson Building of Mechanical Engineering, Room 206
Mechanical Properties of Poly(Methyl Methacrylate)
Enhanced by Functionalized Carbon nanotubes
Guy Shiber
M.Sc. Student of Prof. Leslie Banks-Silis
Carbon nanotubes (CNTs) were discovered approximately 25 years ago and they are amongst the strongest materials known to man. This material is a carbon allotrope like graphite, diamond and fullerene. Apart from their excellent mechanical properties, they have unique electrical, thermal and optical properties, as well. Upon discovery of this unique material, investigators were led to believe that a huge breakthrough would be made in many fields including composite materials; but soon after their discovery, challenges arose. The two main challenges in using CNTs as a reinforcement phase in a composite are homogenous dispersion and distribution of the CNTs in the matrix, as well as improving the interfacial bond between them. Functionalization of CNTs has proved an effective method for overcoming these challenges. Since this material was discovered, numerous studies have been dedicated to finding effective functionalization methods.
The goal of this study, is to determine the mechanical properties of a composite material using experimental and micro-mechanical methods. The composite consists of a poly(methyl methacrylate), or PMMA, matrix enhanced by different weight fractions (wt%) of nanotubes functionalized by two methods. Carboxilated CNTs were further functionalized by the grafting from (GF) or the grafting to (GT) methods. The effect of these functionalization methods on the CNTs is evaluated through measurement of the effective mechanical properties of the composite; as well as, through a comparison to a previous investigation, in which non-functionalized (NF) CNTs were employed. A micro-mechanical model (HFGMC) was used to evaluate the effective mechanical properties of the material.
Uniaxial tensile tests were conducted according to the ASTM D 638-10 standard (2013) with six batches of dog-bone specimens. Specimens were produced by means of a melt-mixing process, followed by injection molding. These batches were manufactured from neat PMMA, PMMA with 0.8 and 1.5 wt% of GF CNTs, and PMMA with 1.5, 3.0 and 6.0 wt% of GT CNTs. Displacements were recorded using the digital image correlation method (DIC). A comprehensive study regarding the DIC parameters and computational error of this approach was conducted. A MATLAB code was employed for two purposes, to calculate the strain measurements from the displacement field, and to combine these measurements with the applied load reading. Four mechanical properties were determined, elastic modulus, E, Poisson’s ratio, n, tensile strength, suts, and strain to failure, ef.
Both functionalization methods showed improved results for the elastic modulus of the high weight fraction batches, compared to the NF CNTs. The highest rate of change in the elastic modulus was reported for the GF 1.5 and GT 3.0 wt% batches, while the latter is a more robust process. This property decreased or was unaffected, for the low CNT content batches. The tensile strength showed a similar trend for the GT batches and a monotonic increase for the GF batches. In general, the strain to failure decreased with increasing CNT content. A lower CNT content was needed to induce the same change in the mechanical properties, when functionalized nanotubes were used. Good agreement is reported with the micro-mechanical model, as well.
