MEAM Seminar Series Summer 2013

Seminars are held on Tuesday mornings, with coffee at 10:30 am outside Towne Room 337 and the seminar beginning at 10:45 am in Towne Room 337(unless otherwise noted).

To be added to the MEAM Events mailing list (which sends notifications regarding all departmental seminars and events) please email us at

Friday, May 24, 11:00 am, Towne room 337

Steven Gray, Ph.D. Candidate, University of Pennsylvania

Advisor: Dr. Vijay Kumar
"Motion primitives and planning for robots with closed chain systems and changing topologies"

Read the Abstract
When operating in human environments, a robot should use predictable motions that allow humans to trust and anticipate its behavior. Heuristic search-based planning offers predictable motions and guarantees on completeness and sub-optimality of solutions. While motion primitive-based (lattice-based) graphs have been used extensively in navigation, application to high-dimensional state-spaces has, until recently, been thought impractical. In this talk, we will show how these graphs can be applied to mobile manipulation with an illustrative example: planning for non-spring and spring-loaded doors. We also discuss our efforts on the DARPA Robotics Challenge Track B team.

June 18

Drew Cheney, Ph.D. Candidate, University of Pennsylvania

Advisor: Dr. Jennifer Lukes
"Computational modeling of geometry dependent phonon transport in nanostructures"

Read the Abstract
Recent experiments have demonstrated that thermal properties of semiconductor nanostructures depend on both nanostructure size and the geometry of the nanostructure boundary. Phonons are quantized mechanical vibrations that are the dominant carrier of heat in most semiconductor materials and their aggregate behavior determine a nanostructure’s thermal performance. Phonon-geometry scattering processes as well as waveguiding effects which result from coherent phonon interference are responsible for the size and shape dependence of thermal transport in these systems. Nanoscale phonon-geometry interactions provide a mechanism by which nanostructure geometry may be used to create materials with targeted thermal properties. However, the ability to manipulate material thermal properties via controlling nanostructure geometry is contingent upon first obtaining increased theoretical understanding of fundamental geometry induced phonon scattering processes and having robust analytical and computational models capable of exploring the nanostructure design space, simulating the phonon scattering events, and linking the behavior of individual phonon modes to overall thermal behavior.

In this seminar, I will discuss previous computational and analytical approaches to modeling geometry dependent phonon transport in nanostructures and will describe an alternative lattice dynamics based atomistic computational model used to calculate mode-dependent phonon transmission through geometrically irregular nanoscale systems. Through the investigation of shear-horizontal phonon propagation through a simple stepped plate geometry, the model is used to assess the accuracy of previously developed approaches which are based on continuum elastic wave theory. In addition, I will discuss the application of the model to cross-section modulated nanowire systems and compare the results of the atomistic model with results obtained through the derivation of a simple model for phonon transmission based upon long wavelength beam theory.

Friday, June 21, 1:00 pm, Towne room 321 - Master's Defense

Kush Prasad, MSE Candidate, University of Pennsylvania

Advisors: Vijay Kumar and Daniel Lee
"Optimal Control of Traffic Signals at a Single Intersection"

Read the Abstract

Traffic congestion and hence fuel wastage by vehicles is a significant problem in urban areas. With the advent of developments in the field of information technology, communications and computational power of computers, it is possible to adaptive control of traffic signals in real-time. One of the ways to reduce traffic
congestion is through optimal control of traffic signals. The control of traffic signal at a single intersection is formulated as a Mixed-Integer-Programming problem in this work. The formulation takes into account the variable arrival rate of vehicles in a cycle at the intersection. It is desirable to obtain optimal green times along with the optimal sequence of phases. Due to the integer constraints, the formulation is computationally hard to solve.

In this work, Algorithms with polynomial time computational complexity have been developed to solve the problem. They provide an optimal solution if the sequence of phases is fixed and a sub-optimal solution if sequence of phases is variable. The polynomial time computational complexity ensures that the
algorithms can be implemented for adaptive traffic signal control.

Friday, June 21, 11:00 am, Towne room 337

 Soonkyum Kim, Ph.D. Candidate, University of Pennsylvania

Advisor: Vijay Kumar
"A Topological Approach to Using Cables to Separate and Manipulate Sets of Objects"

Read the Abstract
In this seminar I will discuss the problem of manipulating and transporting multiple objects on the plane using a cable attached at each end to a mobile robot. This problem is motivated by the use of boats with booms in skimming operations for cleaning oil spills or removing debris on the surface of the water. The goal in this research is to automate the task of separating the objects of interest from a collection of objects by manipulating them with cables that are actuated only at the ends, and then transporting them to specified destinations. Because the cable is flexible, the shape of the cable must be explicitly modeled in the problem. Further, the robots must cooperatively plan motions to achieve the required cable shape and gross position/orientation to separate the objects of interest and then transport them as specified. The theoretical foundation for the problem is derived from topological invariants, homology and homotopy. First, I will derive the necessary topological conditions for achieving the desired separation of objects. Then I will propose a distributed search-based planning technique for finding optimal robot trajectories for separation and transportation. The applicability of this method will be demonstrated by using a dynamic simulation platform with explicit models of the cable dynamics, the contact between the cable and one or more objects, and the surface drag on the cable and on the objects. I also describe our preliminary efforts to develop an experimental platform consisting of a system of two cooperating autonomous boats. Finally, I will discuss how we can solve the problem with large number of objects with multiple robots by sequential manipulations.

Friday, June 21, 10:30 am, Fisher Bennett room 140 - Master's Defense

Mengying Li, MSE Candidate, University of Pennsylvania

Advisor: Noam Lior
"Energy and Exergy Analysis and Thermodynamic Optimization of Deep Engineered Geothermal System Energy Extraction and Power Generation"

Read the Abstract

Geothermal energy is a very abundant renewable energy resource, considered to also be environmentally benign in comparison. If depths of 10 km or more can be reached economically, it is estimated that it can supply about three orders of magnitude more than the total current energy used globally. When water is present there, this hot geofluid can be (and is in many existing geothermal plants) brought to the surface for direct use or to generate electricity. However, in the absence of such natural geofluid, the hot underground zone can be fractured and fluid (typically water) can be injected into that zone (reservoir) ad brought to the surface heated for use. This kind of geothermal system is called Engineered Geothermal System (EGS), which is analyzed in this work about its thermodynamic performance when used to generate electricity.

A typical EGS power system contains three parts: injection/ production wells, engineered reservoir and surface power plant. This thesis consists of the analyses of deep EGS well drilling energy, thermodynamic performance of such wells, EGS reservoir hydraulic fracturing, reservoir performance, and flash type and expansion type power plant performance. Exergy analysis accompanies the studies. General conclusions about EGS thermodynamic optimal design are summarized.

Monday, June 24, 10:30 am, DRLB room 2C2 - Master's Defense

David Kim, MSE Candidate, University of Pennsylvania

Advisor: Noam Lior
"Critical Review of Quantitative Sustainability Analysis with sample case for Reverse Osmosis Desalination"

Read the Abstract
The importance of sustainable development and production, as well other human activities that are of significant magnitude either individually or in aggregate, are increasingly recognized. As is the case with any other performance areas or operating parameters of a system, quantitative tools must be utilized in analyzing the sustainability of a system; quantification of sustainability enables an accurate measurement, comparison, as well as possible optimization of its sustainability performance. The use of Composite Indices is the primary tool by which multi-dimensional, i.e. encompassing environmental, social, economic, and other dimensions that are relevant and important to mankind, sustainability is measured. The first part of the thesis is a critical review of the state-of-the-art of the various tools and techniques of quantitative sustainability analysis. This includes the most applied sustainability metrics, metric normalization schemes, weighting methods, and aggregation methods.

The other part of the thesis consists of the application of this methodology to sustainability analysis of the reverse osmosis (RO) desalination process. Reverse osmosis is currently the most used desalination process. Despite its widespread use, its sustainability, especially its environmental impacts, is often neglected. A comprehensive yet sufficiently simple and straightforward sustainability analysis of reverse osmosis desalination process is developed in this work. The metrics and tools used for this purpose include but are not limited to available sustainability analysis methodology. An original methodology by which the sustainability of the reverse osmosis (RO) desalination can be assessed is developed. Metrics are appropriately designed to account for six important sustainability impact areas in RO desalination: 1) contribution to climate change, 2) land footprint, 3) pollution from RO discharge/effluent, 4) impact on local economy, 5) delivered freshwater quality, and 6) social impact of delivered freshwater on local community. Weighting factors are determined using the Analytic Hierarchy Process (AHP) and aggregated using an original concept called ‘Impact Quantification and Monetization’. Finally, conclusions and recommendations on sustainable RO desalination are made, and conclusions on quantitative sustainability analysis are summarized.

July 1

Steven Gray, Ph.D. Candidate, University of Pennsylvania

Advisors: Dr. Vijay Kumar and Dr. Maxim Likhachev
"Motion primitives and planning for robots with closed chain systems and changing topologies"

Read the Abstract
When operating in human environments, a robot should use predictable motions that allow humans to trust and anticipate its behavior. Heuristic search-based planning offers predictable motions and guarantees on completeness and sub-optimality of solutions. While search-based planning on motion primitive-based (lattice-based) graphs has been used extensively in navigation, application to high-dimensional state-spaces has, until recently, been thought impractical. This dissertation presents methods we have developed for applying these graphs to mobile manipulation, specifically for systems which contain closed chains. The formation of closed chains in tasks that involve contacts with the environment may reduce the number of available degrees of freedom but add complexity in terms of constraints in the high-dimensional state space. We exploit the dimensionality reduction inherent in closed kinematic chains to get efficient search-based planning.

Our planner handles changing topologies (switching between open and closed-chains) in a single plan, including what transitions to include and when to include them. Thus, we can leverage existing results for search-based planning for open chains, combining open and closed chain manipulation planning into one framework. Proofs regarding the framework are introduced for the application to graph-search and its theoretical guarantees of optimality. The dimensionality-reduction is done in a manner that enables finding optimal solutions to low-dimensional problems which map to correspondingly optimal full-dimensional solutions. We apply this framework to planning for opening and navigating through non-spring and spring-loaded doors using a Willow Garage PR2. We also apply our approaches to the Atlas humanoid robot from Boston Dynamics for both stationary manipulation and quasi-static walking.

July 9

Nathan Jacobs, Ph.D. Candidate, University of Pennsylvania

Advisor: Dawn Elliott
"Finite Element Predictions of Human Intervertebral Disc Mechanics"

Read the Abstract
Intervertebral disc structure is highly organized, enabling it to bear multi-directional loads, dissipate energy, and permit flexibility of the spine. The heterogeneous and anisotropic composition of the disc structure imparts a nonlinear, anisotropic, and time-dependent behavior to its stress-strain response. Experimental motion segment testing has characterized the global mechanical properties of the disc, however, the internal stress-strain behavior is not well known. Conventional motion segment testing yields little information regarding the stress-strain distribution throughout the disc or the interactions between the different tissues. Additionally, it is difficult to experimentally quantify the effect of targeted structural changes within specific disc tissues without causing secondary structural changes, such as needle puncture, which obfuscate the interpretation of results. Finite element (FE) models have been used to provide quantitative measures of internal disc mechanics, however, the complicated nature of the disc presents a challenge in developing an accurate and predictive computational disc model, especially when material models are selected without proper characterization at a tissue-level. Most disc FE models are therefore limited either in their material models, their validations to experimental studies, or in their abilities to capture the multifaceted nature of disc mechanics. In order to confidently extend the findings from FE simulations to in vivo applications, a new FE model has been created using structural hyperelastic continuum formulations that have previously been validated through tissue-tests. The objective of this study is to quantify the FE predicted equilibrium and time-dependent stress-strain behavior of the intervertebral disc.

Friday, July 12, 12:00 p.m., Towne 227 - Doctoral Defense

Graham Wabiszewksi, Ph.D. Candidate, University of Pennsylvania

Advisor: Robert Carpick
"Interrogation of Single Asperity Electrical Contacts using Atomic Force Microscopy with Application to NEMS Logic Switches"

Read the Abstract
Energy consumption by computers and electronics is currently 15% of worldwide energy consumption, and growing. The fully-electronic transistor, which is the fundamental computational element of these devices, consumes a significant fraction of this energy. These transistors have been aggressively scaled for increased speed and decreased size for the past half century. However, transistor scaling has lead to undesired energy losses, and device sizes are now approaching the physical limitations of the technology. Ohmic nanoelectromechanical systems (NEMS) logic switches have recently been recognized as a potential transistor replacement technology with energy savings of one to three orders of magnitude over traditional, fully-electronic transistors.

However, NEMS logic switches suffer from tribological failure mechanisms, and this is inhibiting their commercialization. Successful operation of these nanoscale switches is dependent on reliably making and breaking an electrical contact separated by a physical gap. The nanoscale dimensions of NEMS logic switches confer limited contact and restoring forces that lead to permanent device seizure (stiction) or the formation of insulating films of mechanical origin when using conventional, adhesive contact materials (i.e. metals). Of critical need is a method to efficiently identify and interrogate low adhesion, chemically stable electrical contact material pairs using conditions and scales relevant to NEMS logic switch contacts.

This work presents the development of a nanoscale electrical contact testing method based on atomic force microscopy (AFM) that enables 2+ billion contact cycles in laboratory timeframes. Using this method, single asperity Pt-Pt contacts were cycled using contact forces and environments representative of NEMS logic switch operating conditions. Contact resistance before cycling significantly exceeded theoretical predictions for a clean Pt-Pt interface due to adsorbed contaminant films. This resistance increased by up to six orders of magnitude due to cycling-induced insulating tribopolymer growth. Degradation of the contacts was most pronounced when cycled in humid air or under the presence of a voltage (hot switching). Sliding of the contact with nanoscale amplitudes lead to significant recovery of conductivity through displacement of the insulating films. Since typical NEMS logic switch designs result in closure normal to the electrical interface (no shear), this result suggests that lateral actuation (wiping of the contact) should be considered as one possible mechanism for improved NEMS logic switch designs.

Based on this observation, AFM was then used to investigate the role of shear, load, voltage, and environment on the electrical robustness of nanoscale contacts consisting of Pt tips in contact with Pt and nitrogen-incorporated ultrananocrystalline diamond (N-UNCD) flat samples. N-UNCD was selected because similar diamond films have demonstrated low adhesion, chemical inertness, and compatibility with the microfabrication processes required for NEMS logic switches. Pt/Pt interfaces and Pt/N-UNCD interfaces subjected to high loads while sliding resulted in a cleaning of the contact. However, Pt/N-UNCD interfaces subjected to low loads with sliding demonstrated decreased conductivity that correlated with electrical power during contact.

Taken in concert, these finding demonstrate the capability of AFM to investigate nanoscale electrical contact phenomena without the need for time-consuming and expensive integration of unproven materials in NEMS logic switches.

Tuesday, July 30, 10:30 a.m., Towne 227

Jonathon Yoder, Ph.D. Candidate, University of Pennsylvania

Advisor: Dawn Elliott
Title: "Intervertebral Disc Mechanical Function Under Physiological Loading Quantified Non-invasively Utilizing MRI and Image Registration"

Read the Abstract
The intervertebral disc functions to permit motion of the spine while distributing the multidirectional loads experienced during daily activities, including tension, compression, torsion, and bending.  Intervertebral disc degeneration widely affects the aging population, often manifesting itself in low back pain.  This progressive and irreversible disease causes deleterious changes to the discs structural integrity, mechanical function, and nutritional pathways.  Quantification of internal IVD mechanics can improve knowledge on the effects of degeneration on disc mechanical function.  Measuring disc internal mechanics is a complicated challenge; in situ boundary condition replication is difficult with excised tissue testing samples.  Motion segment testing permits the study of overall disc stress and strain behavior, but it does not present detail of the discs internal mechanics and interactions between its constituents.   Magnetic resonance imaging (MRI) has recently been utilized to study IVD internal deformations within our laboratory, providing a non-invasive technique to visualize the discs substructures in two-dimensions (2D).  However, this technique has only been applied to measure 2D strain under single loads.  The IVD deforms in three-dimensions (3D), single 2D images are not able to capture out-of-plane deformations, which are typical of a loaded disc.  This work will develop techniques utilizing 3D MRI and image registration, allowing intervertebral disc structural visualization and quantification of its deformations under load.  The objective of this study is to measure the discs 3D internal deformations when subjected to multiple physiological loading levels to study the effect of degeneration.

August 6

David Argudo, Ph.D. Candidate, University of Pennsylvania

Advisor: Dr. Prashant Purohit
"The mechanical response of DNA: twisting the molecule"

Read the Abstract
Since the structure of double stranded DNA was discovered about six decades ago, there have been great
developments in the fields of biology, genetic engineering, molecular biology, biophysics and numerous others.  It is the objective of our work to carry on with these developments by providing new insights and understanding of DNA behavior/function using mechanics as our primary tool. To achieve this goal we focus on building analytical models for DNA molecules using rod theory and statistical mechanics. We model a key
experiment in which a tensile force is applied on the DNA while simultaneously twisting it using optical or
magnetic tweezers. In these experiments, DNA has been seen to form helical structures (supercoils), collapse into tightly condensed structures (toroids) and undergo structural changes (phase transitions).
Our work focuses on studying all these phenomena by accounting for DNA elasticity, entropic effects due to thermal fluctuations and electrostatics.

Friday, August 16, 9:00 a.m. - Doctoral Defense, Levine 512

Soonykum Kim, Ph.D. Candidate, University of Pennsylvania
Advisor: Dr. Vijay Kumar
"Robot Motion Planning Under Topological Constraints"

Read the Abstract
Path planning or trajectory generation of robotic system is one of the most active areas in robotics research. Various algorithms have been developed to generate path or trajectory for different robotic systems. Our goal is to demonstrate how complicated tasks are solved in motion planning or trajectory generation problem by adapting topology class constraints.We propose the optimal trajectory generation problem under topology class constraints, which can be solved in finite-time and guarantees the global optimal solution. Then, we present the mathematical framework and algorithms for multi-robot topological exploration of unknown environments in which the main goal is to identify the different topological classes of paths. Finally, we demonstrate how to to manipulate and transport multiple objects with only two robots connected with a cable by restricting the cable configuration and paths of robot into specific topology classes.

August 20

Morteza Hakimi Siboni, Ph.D. Candidate, University of Pennsylvania

Advisor: Dr. Pedro Ponte
"Electro-active polymer composites: effective response and stability analysis"

Read the Abstract
Electro-Active Polymers (EAPs) are smart materials that can change shape and size in response to externally applied electric stimuli. This unique property is known as electrostriction, and makes such materials promising candidates for a wide range of practical applications. EAPs are now widely used in the robotics industry, and as sensors and actuators in many other applications. Also, a huge body of research has been dedicated to developing EAPs that can mimic biological muscles and for this reason they are also referred to as artificial muscles. In this work we investigate the possibility of enhancing the electrostriction by making composites consisting of one family of stiff but highly dielectric inclusions firmly embedded in a soft ideal dielectric.

We start this talk with a brief overview of the background material on electro-elasticity and a homogenization framework for Electro-Active Polymer Composites (EAPCs). Then, using a partial decoupling strategy/approximation (developed in earlier works) together with available estimates for the purely mechanical response of such composites, we obtain estimates for the effective electro-mechanical response of particulate EAPCs. Using the homogenization estimates obtained in this work, we are able to study the effect of microstructural parameters (i.e. the concentration of the inclusions, their shape, and their distribution) on the electro-mechanical response of the EAPCs when an electric potential is applied to them. We also investigate the effect of three different failure mechanisms on the performance of dielectric actuators made of particulate EAPCs: loss of strong ellipticity, loss of positive definiteness, and dielectric breakdown, and attempt an optimal design of the microstructure for enhanced electrostriction. In addition, we will briefly discuss the possibility of driving pattern changing instabilities in EAPCs by means of an externally applied potential.

Wednesday, August 21, 1:00 p.m.- Doctoral Defense, Towne 337

Ian Cosden, Ph.D. Candidate, University of Pennsylvania

Advisor: Dr. Jennifer Lukes
"A Hybrid Atomistic-Continuum Model for Liquid-Vapor Phase Change"

Read the Abstract
Boiling, evaporation, and liquid-vapor phase change are inherently
multiscale processes.  Current continuum-based numerical models fail
to capture the atomistic nature of the local density fluctuations that
lead to vapor nucleation.  Atomistic methods, such as molecular
dynamics (MD) simulations, are capable of fully resolving the effects
of individual atomic interactions and nanoscale surface structure on
the incipience of liquid-vapor phase change.  Macroscopic problems,
however, are still well beyond the reach of MD simulations due to the
prohibitively large computational expense of modeling discrete
particles. Hybrid atomistic–continuum (HAC) models offer a solution.
HAC models limit the use of MD simulations to only a small region
where atomistic-level resolution is necessary, such as near a wall or
heater surface, and use continuum methods away from this region.
In this work a fully parallelized hybrid atomistic–continuum model is
developed to resolve nanoscale features of liquid-vapor phase change.
The domain is decomposed into an atomistic domain, where individual
atomic interactions are computed, and a continuum domain, where the
Navier–Stokes equations are solved.  The two domains are coupled
through an overlap region in which the solutions in both domains are
consistent.  The accuracy of the HAC model is demonstrated through the
simulation of sudden start Couette flow, unsteady heat transfer, and
the bulk flow of a liquid-vapor interface.  The new HAC model is used
to model vapor nucleation at a heater surface and compares well to
analytic solutions for evaporation.  Unlike continuum-only methods,
the new HAC model is able to nucleate vapor from liquid naturally,
given the correct thermodynamic conditions, without any assumptions on
the nucleation location or frequency.  The new highly-parallelized HAC
model is shown to reduce computation time by a factor of five for
Couette flow in a 78 nm channel as compared to a fully-atomistic
simulation.  This speedup is expected to become even greater for
larger systems.

August 27

Reza Avazmohammadi, Ph.D. Candidate, University of Pennsylvania

Advisor: Dr. Pedro Ponte
"Homogenization of Fiber-Reinforced Elastomeric Composites"

Read the Abstract
For the last decade, it has been a focus of attention in the mechanics community to find analytical estimates for effective mechanical behavior of elastomeric composites with fiber-reinforced microstructure under large deformation. However, studies on composites containing short fibers are still lacking. Short-fiber-reinforced composites offer great advantages in a variety of engineering applications because of their high strength to weight ratio together with favourable mechanical properties. In addition, short-fiber reinforced composites are very easy to process and offer a greater design flexibility compared to long-fiber reinforced composites. Examples where the properties of short-fiber reinforced composites can be advantageous include flexible underwater vehicles, compliant aircraft structures and car tires.

In this work, we aim to develop an analytical, homogenization-based model to study the constitutive behavior of short-fiber reinforced composites subjected to finite-strain deformations. Assuming hyperelastic behavior for the constituent phases in the elastomeric composites, we propose an approximate homogenization method to obtain estimates for the effective stored-energy function, associated microstructure evolution, and macroscopic instabilities in the composites. The method can be used to provide constitutive relations for the incompressible composites with ellipsoidal microstructure while taking into account the evolution of characteristic features of the microstructure. We make use of this method to obtain analytical estimates for the effective behavior of elastomers containing random distributions of aligned, rigid, spheroidal particles under general loading conditions. We consider both prolate and oblate particles, and for the special case of incompressible neo-Hookean matrix under aligned loading, we provide closed-form estimates. We further conduct a detailed study of the analytical estimates for several representative values of the microstructural and loading parameters, as well as matrix properties.

The method is also applied to two-dimensional composites with random distributions of (long) aligned, elliptical fibers, and the results are compared with corresponding results of earlier homogenization estimates and finite element simulations. In this connection, we have shown that the earlier estimates (obtained via the generalized second-order method) for constitutive behavior of 2-D composites are not stable for a certain range of applied loading, and, in fact, they bifurcate to another set of estimates which are stable. These post-bifurcation estimates are demonstrated to be the associated quasiconvex envelope for the pre-bifurcation ones.