Visualizing Visualizing Tensor Fields Tensor Fields in in Geomechanics Geomechanics Alisa Neeman Alisa Neeman + + Boris Jeremi Boris Jeremi Ć Ć * * Alex Pang Alex Pang + + UC Santa Cruz, Computer UC Santa Cruz, Computer Science Dept. Science Dept. * UC Davis, Dept. of Civil and * UC Davis, Dept. of Civil and Environmental Engineering Environmental Engineering
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Visualizing Tensor Visualizing Tensor Fields in Fields in GeomechanicsGeomechanics
Alisa Neeman Alisa Neeman ++
Boris JeremiBoris JeremiĆĆ**
Alex PangAlex Pang ++
++ UC Santa Cruz, Computer Science Dept. UC Santa Cruz, Computer Science Dept. * UC Davis, Dept. of Civil and * UC Davis, Dept. of Civil and Environmental EngineeringEnvironmental Engineering
Motivation
Geomechanics uses tensors (stress, strain…) – to understand the behavior of soil and their
relation to foundations, structures – analysis of failure of bridges, dams, buildings, etc.
Understand accumulated stress and strain in geological subduction zones (which trigger earthquakes and tsunamis)
Features of Interest Positive stress in piles: cracking concrete large shear stresses: shear deformation or shear failure zones of sign changes: tensile failure These usually occur at soil-pile boundary but can happen anywhere
Bonus Features Capture global stress field Verification and Validation: assess
accuracy of simulation
Limitations of current techniques Hedgehog glyphs inadequate to easily
Marching Cubes designed for isosurfaces in regular grids– Above-below index– Interpolation points
Loop around interpolation points to draw triangles
Some ambiguous cases
Drawing with Marching Cubes!
Edge index: sum of box edges the plane intersects (labels 1,2,4,8,..)
Map from edge intersection index to Marching Cubes index
Intersection points act as interpolation points No ambiguous marching cubes cases
– We build a continuous surface so no holes occur Ambiguous edge index cases, though
Why?
Ambiguous Edge Index Cases
1. Edge lies in the plane (infinite intersections) 2. Plane coincides with a box corner
(three edges claim intersection) Workaround: shift box along an axis slightly - proper
marching cubes case forms Shift box back BEFORE forming triangles - get correct
plane
Filtering With Physical Parameters
Scalar Features – Color, opacity show feature magnitude
Threshold, Inverted Threshold Filtering
Isosurface and isovolume-like selections (without smooth surface)
Opacity Filtering
Goal: find zones of positive stress, sign changes, and high shear
Seismic Moment TensorsIdea: apply moment tensor
decomposition to stress tensors, use scalars as filters to find stress features
GeomechanicsStress Tensor
Seismic/Acoustic Moment Tensor
Symmetric 3 x 3 tensor Symmetric 3 x 3 tensor
Elastic or elastic-plastic material
Elastic material
Describes force on external surface
Force across internal surface (causing movement along fault)
Values throughout volume from simulation
A few point sources from measured acoustic emission
Seismic Moment Tensors
Moment Tensor Decomposition (after diagonalizing):
Mij = isotropy + anisotropy
Isotropy = (λ1 + λ2 + λ3)/3
Describes forces causing earthquake withvector dipoles: two equal and opposite vectors
along an axis orthogonal to both
Mxy:x
Y
Moment Tensor AnisotropyAnisotropy = Double Couple +
Compensated Linear Vector Dipolemi
* = λi – isotropy
sort: |m3*| ≥ |m2
*| ≥ |m1*|
F = - m1* / m3
*
Double Couple: m3* (1 - 2F)
CLVD: m3* F
Seismic Failure Measures1. Pure isotropy: explosion or implosion
2. CLVD: change in volume compensated by particle movement along plane of largest stress. Eigenvalues 2, -1, -1
3. Double couple: two linear vector dipoles of equal magnitude, opposite sign, resolving shear motionEigenvalues 1, 0, -1
Eigen Difference
Measure for double degenerate tensors K = 2λ2 − (λ1 + λ3) K > 0
planar (identical major and medium eigenvectors) K < 0
– linear (identical medium and minor eigenvectors).
Applied universally across volume
Boussinesq Dual Point LoadEasily verify results through symmetry
Linear Scale Isotropy Log Scale color and opacity
0157,861 -157,861 011.97 -11.97
Boussinesq Dual Point Load•Double Couple problem: find high shear
•Selects different regions than isotropy
All values High values0.975-1.0
Mid-range values(0.5-0.715)
1.0
0.0
Bridges and EarthquakesSeries of bents support bridge
Frequency and amplitude vary with soil/rock foundation. Worst case, high amplitude (soft soils)
Two Pile Bridge Bent
Piles penetrate halfway down into soil
Circular appearance in planes’ orientation– boundary effects in
simulation– model needs to be
expanded to more realistically simulate half-space
No pure double couple – Discrete nature of field
1.0
0.0
Dou
ble
Cou
ple
DeckPile
Bridge Bent Pushover Force applied at bridge
deck (top of columns) Simulation: pushover
followed by shaking Eigen difference shows
sudden flip between linear and planar in piles
+15.03
-15.03
Log
Sca
le E
igen
Diff
eren
ce
Isotropy: Inverted threshold Log scale isotropy:
lowest 25% and top 25%
Zones switching sign highlighted
Shadowing effect:right hand pile ‘in shadow’
Border effects(tradeoff withcomputation cost)
14.5
-14.5
0.0
Log
Sca
le Is
otro
py
Shadow
Conclusions
Isotropy most useful scalar feature Thresholding/inverted thresholding
highlights behavior under stress Plane-in-a-box provides global
perspective of stress orientation Algorithm cheap and interactive Assists with simulation verification and
validation
Acknowledgements
Sponsors: NSF and GAANN Thanks to the reviewers for feedback Thanks to Dr. Xiaoqiang Zheng for
discussions
Visualizing Tensor Visualizing Tensor Fields in Fields in GeomechanicsGeomechanics
Alisa Neeman Alisa Neeman ++
Boris JeremiBoris JeremiĆĆ**
Alex PangAlex Pang ++
++ UC Santa Cruz, Computer Science Dept. UC Santa Cruz, Computer Science Dept. * UC Davis, Dept. of Civil and * UC Davis, Dept. of Civil and Environmental EngineeringEnvironmental Engineering