Current and Graduated ERSL Students

Praveen Yadav

PhD, 2016

CTO, SciArt

Shiguang Deng

PhD, 2016

Research Scientist, MSC Software

Chaman S. Verma

PhD, 2017

Research Scientist, Palo Alto Research Center

Amir M. Mirzendehdel

PhD, 2017

Research Scientist, Palo Alto Research Center

Alireza Taheri

PhD Candidate

Tej Kumar

PhD Candidate

Buzz Rankouhi

PhD Candidate

Subodh Subedi

PhD Candidate

Aaditya Chandrasekhar

MS/PhD Candidate

Bhaghyashree Prabhune

MS/PhD Candidate

Bian Xiang

Visiting Scholar

Anirudh Krishnakumar

MS, 2016

Product Manager GrabCAD

Anirban Niyogi

MS, 2015

Cameron Gilanshah

MS, 2016

Victor Cavalcanti

MS, 2015

Gabriel Elkind

MS Candidate

Nikhil Agarwal

Undergrad Research Assistant

Alex Buehler

Undergrad Research Assistant

Jimmy Herman

Undergrad Research Assistant

Victor Markus

Undergrad Research Assistant

The research group current consists of about 15 highly talented students. Students join ERSL from all over the world including the United States, China, India, Iran, Israel and Europe.

Current Research Projects

  • image
    Topology optimized design for the classic MBB problem.

    Topology optimization benchmark studies

    This is an effort to provide the community with topology optimized models that can be used for 3D-printing, recovery of CAD models, etc. Click here for TO benchMark models

  • image
    A "Thomas Train" structural problem with 50 million degree of freedom solved on a GPU in 24 minutes.

    Limited Memory Deflated Finite Element Analysis

    Large-scale FEA problems with millions of degrees of freedom are becoming commonplace in solid mechanics. The bottleneck in such problems is memory access. The objective of this project is to exploit assembly-free deflation techniques to accelerate FEA.

  • image
    Transient response of a million degree of freedom "Arduino Board" subject to an impulse loading.

    Large-Scale Implicit Structural Dynamics

    The primary computational bottle-neck in implicit structural dynamics is the repeated inversion of the underlying stiffness matrix. A fast inversion technique is proposed by combining the well-known Newmark-beta method, with assembly-free deflated conjugate gradient (AF-DCG) for large-scale problems.

  • image
    A tangled quad mesh over which FEA was solved with high accuracy.

    Finite Element Analysis over a Tangled Mesh

    Classic FEA breaks down if one or more elements gets inverted, i.e., if the mesh gets tangled. But, mesh tangling is unavoidable during mesh generation, mesh morhping and large-scale deformation. The objective of this research is to extend classic FEA to handle tangled meshes.

  • image
    The number of singularities in a quad mesh is reduced without affecting element quality.

    Singularity Removal in Quad Meshes

    In quad meshes, nodes connected to exactly 4 quad elements are called regular; otherwise they are referred to as irregular or singular. Singular nodes are detrimental to FEA accuracy. A new singularity removal method has been developed to dramatically reduce the number of node singularities.

  • image
    The solution to the classic "Michelle Bridge" problem via the topological level-set.

    Topology Optimization via the Topological Level-Set

    The topological level-set method developed by our group directly uses the topological sensitivity field as a level-set for an efficient solution of topology optimization problems.

  • image
    Optimal topology of a flange subject to compliance, stress and eigen-value constraints

    Multi-constrained Topology Optimization

    Topology optimization problems subject to several constraints are both theoretically and computationally challenging. The topological level-set method is combined here with augmented Lagrangian methods to solve such multi-constrained topology optimization problems.

  • image
    The solution to classic cantilever compliance minimization problem, but using three different materials.

    Multi-material Topology Optimization

    As additive manufacturing expands into multi-material, there is a demand for efficient multi-material topology optimization. The classic approach is to impose constraints on the volume-fraction of each of the material constituents. This can artificially restrict the design space. Instead, the total mass and compliance are treated as conflicting objectives, and the corresponding Pareto curve is traced; no additional constraint is imposed on the material composition.

  • image
    A hinge-free solution to the classic cruncher mechanism.

    Hinge-Free Compliant Mechanism Design

    Hinges can lead to high stress concentration in compliant mechanisms. The topological sensitivity concept is exploited here to design hinge-free compliant mechanisms. These mechanisms exhibit high mechanical advantage and low stresses.

  • image
    Client-server architecture underlying cloudtopopt.

    Topology Optimization on the Cloud

    The wide-spread use of topology optimization has been deterred due to high computational cost and significant software/hardware investment. Our group has developed a cloud based implementation of topology optimization, hosted at

  • image
    A microstructure with optimal bulk modulus.

    Microstructural Optimization

    The objective in microstructural optimization is to find the distribution of one or more 'materials' that would result in a desired microscopic behaviour (ex: negative Poisson ratio). Our group has developed a highly efficient topological sensitivity based method for designing such microstructures and tracing their corresponding Hashin-Shtrikman curves.

  • image
    A NURBS based iso-geometric solution to the classic MBB problem.

    Iso-Geometric Multi-material Topology Optimization

    The objective here is to exploit the inherent advantages of isogeometric analysis for multi-material topology optimization. Due to the unified parametrization of geometry, analysis and design space, the sensitivities are computed analytically.

  • image
    A snap-shot of thermo-elastic simulation of the LENS process

    Assembly Free Additive Manufacturing (AM) Simulation

    In AM simulation, repeated meshing and insertion of new elements during material deposition can pose significant implementation challenges. Our group is developing an assembly-free framework for AM simulation that offers several advantages: (1) The workspace is meshed only once at the start of the simulation, (2) addition and deletion of elements is easy since the stiffness matrix is never assembled, and (3) the underlying linear systems of equations can be solved efficiently through assembly-free deflation methods.

Graduated Students

  • 2012 Josh Danczyk (Ph.D.)
  • 2011 Vikalp Mishra (Ph.D.), Inna Turevsky (Ph.D.), Kavous Jorabchi (Ph.D.)
  • 2010 Vaibhav Deshpande (M.S.)
  • 2009 Wa’el Abdel Samad (M.S.)
  • 2008 Sahil Kulakarni (M.S.)
  • 2007 Sankara Hari Gopalakrishnan (M.S)
  • 2006 Himanshu Tiwari (M.S.), Rakesh Vemulapally (M.S.), Murari Sinha (M.S.)
  • 2005 Ameya Sirpotdar (M.S.)

ERSL has graduated 4 PhD and 8 MS students; they are currently employed in academia, as well as semiconductor, machinery and software industries.