U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Project goal Methodologies and tools for interface between design and manufacturing for Micro-Electro-Mechanical Systems (MEMS) NSF Grant CISE #
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Optimal topology design of compliant mechanisms Automatic image interpretation and creation of solid model from optimized topology image Image capturing of fabricated devices Image processing for edge-extraction and object- finding Writing image data as an IGES file for Pro- Engineer Vision-based metrology for meso-scale ceramic devices Past accomplishments
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Performance Specifications Synthesis Solution Refined Design Solution Meshed Model for Analysis Solid Model from the Optimized Device Digital Format for SFF or CNC Mask Layout for Microfab. Micro Prototype Macro Prototype Refined Prototype CAD model from the macro prototype Fabrication/Prototyping Digital Interface Design Reverse Engineering Solid model from microscopic images
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia From optimized compliant topology image to a solid model IGES model with line and arc segments Optimized compliant topology using material density design parameterization Image processing, edge extraction, and object finding
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Image from the optical microscopeAfter edge extraction The lack of sharp edges corresponding to the actual structure poses a challenge in extracting the correct topology. Edge extraction from 2-D images
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia From fabricated MEMS device to a solid model Optical microscope image of a compliant micro crimper After edge extraction 5 µm
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia From fabricated MEMS device to a solid model IGES model exported into Pro-E with line and arc segments Extruded solid model ready for behavioral simulation
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Macro prototype of a micro wedge motor (Allen, 1998) The micro prototype was made using Sandia’s SUMMiT process.
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Lithography masks for a micro wedge motor
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Premise: Assuming material behavior and properties, forces can be estimated from underformed and deformed geometry of a flexible structure. Vision-based force sensing Goal: Non-contacting, non-interfering force sensor based on vision and computation.
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Image capturing before and after deformation Displacements Strains Stresses Forces Material properties
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Inverse elastic analysis problem. Linear (small deformations) problem is trivial. Large deformation, small/large strain problem is of interest here. Stiffness matrix Forces Displacement Assumption: Correspondence between undeformed and deformed geometries is known.
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia y z At time t = 0 (Undeformed) At time t (Deformed) o x Computing the deformation gradient Deformation gradient
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Computing the large strain Logarithmic strain
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Computing the stress and element forces Material properties for plane-stress condition Stress Element forces
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Force recovery Large strain caseSmall strain case
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Macro-scale experiment
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Corner detection
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Comparison of computed and measured displacements
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Recovered forces (experimental displacements)
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Recovered forces (numerical displacements)
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia The procedure for computing strains from displacements is prone to numerical error. The computed strain is very sensitive to even small perturbations in displacement data. The sensitivity is worse for small strains than large strains.
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Sensitivity analysis
U. of Pennsylvania, Jan University of Pennsylvania, Philadelphia Sensitivity analysis