Physical Modeling: Time Lapse, 3D, and VSP
Robert W. WileyAllied Geophysical Laboratories
University of Houston
Outline
• Properties of Physical Modeling
• Fracture Model
• Time-Lapse Model
• 3D VSP Model
• Hardware
• Future Directions
• Conclusions
Properties that scale well
• Distance (typically we use 1km = 10 cm)• Time (typically we use 4 ms = 0.4 s)• Frequency (typically we use 30 Hz = 300 kHz)• P and S velocity
– For acoustic propagation we scale vp by 0.5– For elastic propagation we do not rescale
• Viscosity– Gas by air– Water by water– Oil by glycerol
Properties that do not scale well• Source and receiver size• Pore size (other physics takes place for extremely small pores)• Frequency dependent attenuation mechanisms (which we don’t
yet understand anyway!)• Soft sediments and weathering zones
Other limitations• A model 1 m by 1 m by 0.5 m (corresponding to 10 km x 10
km by 5 km) weighs more than one metric ton!• Models take months to build • We needed bigger, stronger, and more patient graduate
students!
Fracture Model
Fracture ConstructionFracture Construction
1.1 cm
Glass slides
35.5 cm
Glass blocks
Resin
x
Fracture model under construction showing glass slides in situ
y
HTI model
y
x
63.5 cm63.5 cm
50.8 cm50.8 cm
Fracture zoneFracture zone
2.5
3
2
1071 CDP 1360
Line 2151
+ + -
2.2
2.4
2.6
1071 1360
Tim
e (s
)
CDPLine 2140 500 m Offset
Trace Number
200
1
1 200100 300
Line
Damp for 750 m offset – time slice 2430 ms – Perp.
Positive Amplitude
Most Positive Curvature Most Negative Curvature
Principle Comp Filter
Time-Lapse Model
The model as constructed
(Wardana, 2001)
Time lapse modeling
(Wardana, 2001)
wet
half wet
dry
near mid far
Amplitude
Time lapse modeling
(Wardana, 2001)
wet
half wet
dry
near mid far
Coherence
3D VSP Model
Physical model
6 layer alternate blue and black & a salt body ρblue= 2.4 g/cc ρblack = 1.34 g/cc Vblack= 2586.9 m/sVblue =3264.3 m/s
Preprocessing (First break pick)
Hardware
3 C Transducers Shear Transducers
Pin TransducersSpherical Transducers
TRANSDUCERS
Old System (Backup System)
NI 2 Source 4 Receiver System
Dry Tank
NI 8 Source 16 Receiver System
Wet Tank
Initial Configuration
Data Configuration
NI System Control- LabView Based
Main Window
Position Control Window
Future Directions
• Bring the two new systems on line
• Develop a 16 receiver system
• Develop a multiple source system
• Collect porous model data
• Test additional transducers
• Investigate other materials
Conclusions
• We are able to build reasonably complex physical models
• Results are similar to seismic data
• This is an excellent approach to acquiring controlled data for testing algorithms
• Physical modeling is cost effective compared to field experiments
Thank You
CurvatureCurvature2.4
2.6
CDP
Line 2140 500 m Offset
1150 1202
Pinducer
Physical Limitations
• Inaccurate construction of actual model
• Limited selection of velocity and density parameters
Some Limitations
• Limited bandwidth with strong resonance at one frequency
• Limited to materials with specific elastic parameters
• Difficult to build lateral velocity variations
• Labor intensive
• Cannot model all real earth materials/layers
Source and Receiver Issues
• Inaccurate location of source and receiver with respect to the model
• Source and receiver resonate at one frequency
• Source and receiver dimensions are large
• Source and receiver have pronounced radiation patterns
+ + -
2.2
2.4
2.6
1071 1360
Tim
e (s
)
CDPLine 2140 500 m Offset
Properties that do not scale well• Source and receiver size• Pore size (other physics takes place for extremely small pores)• Frequency dependent attenuation mechanisms (which we don’t
yet understand anyway!)• Soft sediments and weathering zones
Other limitations• A model 1 m by 1 m by 0.5 m (corresponding to 10 km x 10
km by 5 km) weighs more than one metric ton!• Models take months to build • We needed bigger, stronger, and more patient graduate
students!
Why Run Models
• To test interpretation – Salt dome, coal seam
• To test algorithm – Migration, Multiple removal
• To test interpretation tools – fracture identification, fluid content
• To test theory – Lamb’s problem, Kirchhoff diffraction, anisotropy
Why Physical Modeling
• No simplifying mathematical assumptions
• No approximations to mathematical functions
• No round-off errors
• No a priori mathematical understanding required
Limitations
• Limited selection of velocity and density parameters
• Difficult to build lateral velocity variations
• Labor intensive
• Inaccurate construction of actual model
• Limited to frequency response of transducers
• Source and receiver dimensions
Benefits of Physical Modeling
• Experimental repeatability and controlled conditions
• Very cost-effective compared to field work
• Physics of elastic energy propagation through physical models is same as real world
• Arbitrary earthlike conditions closer to real earth data than numerical modeling
Spherical Transducer
3C Transducer
www.valpeyfisher-ud.com
Shear Transducer Design