7/30/2019 Gpgn303 Dc Methods
1/54
12/11/11
1
Course Notes (Fall 2011)
GPGN 303: Section 3
Introduction to Electrical Methods
Yaoguo Li
Department of Geophysics
Colorado School of Mines
(For class use only, do not distribute!)
Outline
Introduction Electrical conductivity Electrical current in conductive media Charge accumulation
Middle gradient mapping Apparent resistivity
Vertical electrical sounding Survey configuration interpretation
2D and 3D imaging Array and survey geometry Inversion
Induced polarization
7/30/2019 Gpgn303 Dc Methods
2/54
12/11/11
2
Conventional geophysical methods: Gravity Magnetics DC resistivity/Induced polarization Electromagnetic induction (EM) Ground penetrating radar (GPR) Seismic
Energy
Source
Measured
Data
Images
(surface / subsurface)
Interaction
between fields andmaterial properties
Altered fields to be
measured
7/30/2019 Gpgn303 Dc Methods
3/54
12/11/11
3
1. Introduction: Basic concept of electrical method
Inject direct current into conductive ground. Measure the voltage produced on the surface or in boreholes. Always involves four electrodes.
Different components of the model
Energy source: batteries or generator Field: electric field / electrical current Material property: electrical conductivity (resistivity) Interaction: conductivity changes the flow of
electrical current (*)
Altered field: different electrical field distribution Data: voltage measured on the surface or in the
boreholes.
7/30/2019 Gpgn303 Dc Methods
4/54
12/11/11
4
Conceptual model
Given the above configuration: One would measure higher voltages over a resistive
body and lower voltages over a more conductive
body (to be discussed soon).
The voltage senses the change in resistivity orconductivity.
Historical Development The first S.P. survey conducted in 1820 by Fox in
England.
S.P. over a massive sulfide body Schlumberger brothers in 1920s
1D sounding and borehole logging 1940s and 50s: multiple position & multiple offset
used in mineral exploration
1950: Induced Polarization (IP) for disseminatedsulfides
7/30/2019 Gpgn303 Dc Methods
5/54
12/11/11
5
Ohms Law: R = V/I Resistance = Voltage/Current Voltage is directly related to resistivity: Voltage
increases when resistance increases.
Electrical Conductivity
Cross-section area: A
Length: L
V
A
L
Electrical Conductivity and Resistivity
Rock conductivity: Ohms Law
Express in current density (j) in terms of electric field (E) andmaterial property (resistivity !).
Conductivity is the more fundamental quantity
7/30/2019 Gpgn303 Dc Methods
6/54
12/11/11
6
Factors affecting rock conductivities
Porosity Permeability (pore connectivity) Fluid saturation Electrical conductivity of the fluid
7/30/2019 Gpgn303 Dc Methods
7/54
12/11/11
7
(Ward, 1990)
7/30/2019 Gpgn303 Dc Methods
8/54
12/11/11
8
Anisotropy!
in general, electrical conductivity is really a tensor!! The conductivity values are different in different directions
Simple Example:
The conductivities in
longitudinal and transversedirections are different
2. Theoretical background: Current flow
across a boundary between two media
Tangential component
Normal component of 1E
! = conductivity (S/m), S="-1
" = resistivity ("*m)
j = current density (A/m2)
E = Electric field (V/m)
# = charge density (C/m2)
Q = charge (C)
7/30/2019 Gpgn303 Dc Methods
9/54
12/11/11
9
On a planar boundary: (all quantities in thetwo media are labelled with a subscript of 1and 2, respectively)
Integration along a rectangular loop,
Current density is continuous across the boundary(i.e., normal components are continuous)
Electrical field: tangential component is continuous,but normal component is discontinuous!
Summary
Current density does not change across the interface. The electric field is discontinuous across the interface. The only field generator of static electric field is electrical
charges
There must be accumulation of the electrical charges at theinterface.
This leads to the only conclusion:There must be charges accumulating on the interface!
7/30/2019 Gpgn303 Dc Methods
10/54
12/11/11
10
The questions are then:
What is the sign of the charge and how much is there? Lets approximate the charges locally as an infinite sheet
with surface charge density, , which produces an
electrical field normal to the interface.
The normal field is given by
Such that the total normal components are given by
Set:
Surface density of charges on the interface
7/30/2019 Gpgn303 Dc Methods
11/54
12/11/11
11
Summary of charge accumulation
Two cases:1. From conductive to resistive region
Positive charges accumulate
V
2. From resistive to conductive region
Negative charges accumulate
^
7/30/2019 Gpgn303 Dc Methods
12/54
12/11/11
12
How does the current flow?
Combining the two boundary conditions toderive the angle of current flow with respect
to the normal vector
i.e., current bends towards normal when
entering resistive medium, and away from
normal when entering conductve medium!
Case II:
Negative charge
pulls E in towards
interface
Case I:
Positive charge
pushes E away from
interface
V ^
7/30/2019 Gpgn303 Dc Methods
13/54
12/11/11
13
Case Charge Electrical current
Conductive to Bends towards
Resistive the normal
Resistive to Bends away
Conductive from the normal
Key Points:
charges accumulate and current flow
(Burger, 1992)
7/30/2019 Gpgn303 Dc Methods
14/54
12/11/11
14
Current flowing in and around conductivity anomalies
Current channels into
a conductor
Current flows arounda resistor
Arrowed lines are
current flow; dashedlines are supposed
to be equipotentiallines.
Background electrical field:1. Point current on surface of a uniform half-space.
Equivalent charges
at the source point:
7/30/2019 Gpgn303 Dc Methods
15/54
12/11/11
15
2. Buried point current source in a half-space
spreads out on the surface!
at the source location!
is the mirror point of
w.r.t. the sound surface
Mise-a-la-masse method
An illustration of the role of charge accumulation in
DC resistivity method
7/30/2019 Gpgn303 Dc Methods
16/54
12/11/11
16
1. Mise-a-la-masse method
Direct application of charge accumulationCharge an in-place conductive body, creating
charges on the surface of the body as current
flows outwards.
Measure the potential at the surface.Similar to gravity, except in this case you are
measuring potential.
Schematics of a mise-a-la-masse survey
7/30/2019 Gpgn303 Dc Methods
17/54
12/11/11
17
Conceptual summary of DC resistivity method
We inject electrical current into the ground (source andfield)
Electrical charges accumulate on the interfaces betweenregions with different electrical conductivity (physical
property and interaction with the input primary field).
Accumulated charges produce additional electric fieldthat is superimposed on the primary field (altered field).
We measure the resultant field (data), which carryinformation about the conductivity variation.
(Hattula and Rekola, 2000, Geophysics)
Mise--la-masse surveys:
delineated the extent of sulfides method for correlating drill holeore intersections
guide drilling during a deepexploration program.
7/30/2019 Gpgn303 Dc Methods
18/54
12/11/11
18
2. Middle Gradient Survey
Middle Gradient Survey
Inject current into the ground using a pair of widely separatedA &B electrode, so as to create nearly horizontal priamry field
in the middle.
As current passes through regions of varying conductivity,charges accumulate at the interfaces between regions with
different conductivities.
Measure the potential differences produced by theseaccumulated charges together with the primary potential.
Rely on the fact that the horizontal component of the primaryelectric field is nearly constant with the central 1/3 bewteen
the AB electrodes
7/30/2019 Gpgn303 Dc Methods
19/54
12/11/11
19
Set-up of Middle Gradient
Measures the horizontal variation in voltage. Use fixed electrode (A and B) positions at large separation. Move MN around inside AB (central 1/3) Originally: measurements only along the line passing AB Commonly: measurements over a grid MN electrodes are parallel to AB electrodes (important!)
A B
Primary Field in the direction paralle to AB
-1000 m 1000 m
Linear scale log scale
7/30/2019 Gpgn303 Dc Methods
20/54
12/11/11
20
Primary Field in X-direction: in the central area
Linear scale log scale
Middle gradient: over a conductive Body
Primary field
Total field
7/30/2019 Gpgn303 Dc Methods
21/54
12/11/11
21
Middle gradient: over a resistive Body
Primary field
Total field
Charge accumulation & anomalous potential Two parts of potential difference Primary Anomalous
The anomalous potential can be eitherconstructive ordestructive to the primary field, depending on location.
Over a conductive body: we measure a central lowwith two positive side lobes
Over a resistive body: we measure a central high withtwo negative side lobes
7/30/2019 Gpgn303 Dc Methods
22/54
12/11/11
22
Variation of Potential Difference
The current is closer to the surface near theelectrodes. Therefore, the potential is also larger near
the electrodes, but this is due simply to the geometry
of the survey (not ideal for interpretation)
V
X
Uniform half-space
Apparent Resistivity:
Lets define a quantity, apparent resistivity. It should:
Be constant over a uniform half-space (preferred) have the units of resistivity
X
Uniform half-space
7/30/2019 Gpgn303 Dc Methods
23/54
12/11/11
23
The apparent resistivity is the measured voltage scaled by thecurrent and by a geometrical factor (K).
The geometrical factor depends on the type of survey, basedon the electrode and voltmeter positions.
Generally:
Apparent Resistivity: Definition
A few comments about the apparent resistivity
It is a convenient quantity to work with for thefollowing two reasons:
It has the units of resistivity (!m) It is equal to the true value if we have a uniform halfspace
However, apparent resistivity is NOT the primarydata. Measured voltages are the primary data, which
are used by most modern inversion algorithms
In true 3D acquisition where the MN and ABelectrodes are not aligned in the same direction, we
may NOT be able to define an apparent resistivity,
because the geometric factor is undefined (infinite).
7/30/2019 Gpgn303 Dc Methods
24/54
12/11/11
24
Middle Gradient Array Over a Vertical Contact
A (-1000 m) A (+1000 m)
MN=20 m
Apparent resistivity mapfrom a middle gradient
Survey at the BallengerRanch, NM
Two surveys merged
AB1: (-20,60 ) (280,60) AB2: (-40,0) (260,0) MN= 10 m
7/30/2019 Gpgn303 Dc Methods
25/54
12/11/11
25
Utility of middle gradient surveys
Areal mapping to characterize the lateral variation ofsubsurface conductivity
Has little information about the vertical variation Often used:
to map elongated linear features such as a vein, an intrusivedyke, or a paleochannel
As a reconnaissance tool to locate targets for furtherinvestigation
Vertical electrical sounding (VES)
7/30/2019 Gpgn303 Dc Methods
26/54
12/11/11
26
VES (Vertical Electrical Sounding)
Used to detect the vertical variation of resistivity as afunction of depth.
Designed to work in 1D environment (layer cakedearth): Assumption!
Length AB increases sequentially For each AB separation, measure the potential using a
small MN separation
Schlumberger sounding
Designed to work over 1D earth MN located in the center between A & B Measure electric field
7/30/2019 Gpgn303 Dc Methods
27/54
12/11/11
27
Schlumberger Sounding
When ,
Geometrical factor:
Consider a half-space with 2 layers of differing resistivity.Resistivity of top layer is and the bottom layer is .
At small AB, the current does not flow deep enough tosignificantly detect the deeper layer. Most current flows in thetop of the first layer.
The potential reflects the resistivity of the top layer. At large AB, most of the current flows in the deeper layer. The potential reflects the resistivity of the bottom layer.
7/30/2019 Gpgn303 Dc Methods
28/54
12/11/11
28
Summary of 2-layer half-space
Depth of investigation: on the same order as AB/2 For small AB (Lh): The apparent resistivity changes monotonically
between layers (it only increases or decreases).
h
h
7/30/2019 Gpgn303 Dc Methods
29/54
12/11/11
29
Plot the apparent resistivity data as a function of AB/2 on a log-log plot.
Depth of current penetration
AB determines the current distribution The current distribution determines the depth of
investigation.
Depth of investigation is proportional to ABseparation (not linearly, though!)
In order to have 50% of current flowing beneath agiven depth, the length AB must be twice that depth.
Nominally: the depth of investigation is half ABspacing
7/30/2019 Gpgn303 Dc Methods
30/54
12/11/11
30
Current density at depth h directionbelow th emiddle point between AB
Fraction of current above depth h
Depth of 50% current partition: h=AB/2
Current distribution in a uniform half-space
(Burger, 1992)
However, the current flowdepends strongly on electricalconductivity.
Less current goes to the depthwhen there is a more conductive
surface layer:smallerdepth of
investigation
More current goes to the depthwhen there is a more resistive
surface layer:greaterdepth of
investigation
7/30/2019 Gpgn303 Dc Methods
31/54
12/11/11
31
Practical implementation of
Schlumberger Sounding Select a set of AB separations that increases
logarithmically, so there are several values per decade
Keep the MN separation much smaller, so weeffectively measuring the horizontal electric field atthe mid-point between A and B..
To keep the measured #VMN well above noise, weincrease the MN separation for every half decade ofAB separations.
Two AB separations are repeated using two adjacentMN separations (see next slide for example).
Sample separations for
Schlumberger sounding
7/30/2019 Gpgn303 Dc Methods
32/54
12/11/11
32
7/30/2019 Gpgn303 Dc Methods
33/54
12/11/11
33
Depth of investigation as a function of MN spacing
(in VES over 1D earth only)
For a given AB separation, an increase in MNseparation leads to decreased depth of investigation.
The reason: As M and N electrodes move apart, they each become
closer to the current electrode A or B
Correspondingly, the relative contribution to the measuredvoltage by deeper charges decrease, so the data are more
sensitive to shallower conductivity.
An end-member scenario: As MN separationapproaches AB separation, the depth of investigation
becomes zero.
Measurements with same AB
but different MN
Exapnding MN causes the curve to shift to the right!
-- decreasing depth of investigation
7/30/2019 Gpgn303 Dc Methods
34/54
12/11/11
34
3-Layer Cases
1. A:2. Q:3. K:
4. H:
7/30/2019 Gpgn303 Dc Methods
35/54
12/11/11
35
Wenner array beside a
dipping contact(parallel to contact)
1D Interpretation
Curve matching by manual approach (ancient): comparing measured curve with pre-calculated curves
to find a macthing one, thus the resistivities andthickness
Works for two- and three-layered cases Least-squares solution to find the resistivities and
thicknesses of a small number of layers.
Generalized nonlinear inversion to find a minimumstructure function of resistivity as a function of depth.
7/30/2019 Gpgn303 Dc Methods
36/54
12/11/11
36
7/30/2019 Gpgn303 Dc Methods
37/54
12/11/11
37
Theoretically:
Uniqueness Theorem: Lange 1932 1D: perfect data for all AB offset and fixed MN Practically: Many similar solution exists because we have only
finite number of data with measurement errors.
Least-squares solution (parametric inversion): find theresistivities and thicknesses so as to minimize the data
misfit between observed and calculated apparentresistivities.
Parametric inversion for 1D parameters
We assume known number of layers, andparameterize the model by the resistivity and
thickness of each layer.
For example, a two-layered earth has threeparameters:
We can calculate the predicted apparent resistivityknowing the values of these variables and surveys
geometry:
7/30/2019 Gpgn303 Dc Methods
38/54
12/11/11
38
We solve a non-linear least squares problem to findthe values of that would have
produced the measured apparent resistivity data.
This is done by minimizing the following datamisfit function;
Because the forward modeling is non-linear, this issolved iteratively by starting from an initial guess
for
Equivalence:
one form of ambiguity Pertains to the cases when
A a thin resistive layer is sandwiched between moreconductive layers: K type:
Or: a thin conductive layer is sandwiched between resistivelayers: H type:
As long as the transverse resistance (K type) or thelongitudinal conductance (H type) remains the same,resistivity profiles with different middle layer thicknesswill produce the same apparent resistivity curve within
error tolerance.Transverse resistance
Longitudinal conductance
7/30/2019 Gpgn303 Dc Methods
39/54
12/11/11
39
Leyden Schlumberger Sounding
Four-layer interpretation
Leyden Schlumberger Sounding
Five-layer interpretation
This thin layer is clearly not required.
7/30/2019 Gpgn303 Dc Methods
40/54
12/11/11
40
2D and 3D Imaging (ERT)
2D imaging of subsurface
Multiple measurements at different surface locations withfixed array geometry: detects lateral variation of electricalconductivity
Multiple measurements at the same locations with expandingarray geometry: detects vertical variation of electricalconductivity
To detect the variation of conductivity horizontally andvertically in the surface, we require measurements at multiplelocations using expanding arrays (multiple electrode off-set)
Equivalently: multiple source locations and multiplemeasurements for each source location
7/30/2019 Gpgn303 Dc Methods
41/54
12/11/11
41
Traditional surveys conducted along lines(co-linear arrays)
Common survey geometriespole - polepole - dipole (PDR)pole - dipole (PDL)dipole dipoleWenner
2D acquisition: Co-linear survey geometry
I V
I
V I
V
I V
I
V
Apparent resistivity:
Organize data by TX-RX (current and potential)electrode locations
Plotting each datum (apparent resistivity directly belowthe mid-point of the array) at a pseudo-depth (array
separation)
Pseudo-sections: plotting raw data
7/30/2019 Gpgn303 Dc Methods
42/54
12/11/11
42
Example: Dipole-dipole pseudo-section
Pole-pole
Pole-dipole (R)
Pole-dipole (L)
7/30/2019 Gpgn303 Dc Methods
43/54
12/11/11
43
Pole-pole
Pole-dipole (R)
Pole-dipole (L)
7/30/2019 Gpgn303 Dc Methods
44/54
12/11/11
44
Comments on psudo-sections
Pseduo-section were developed as an easy way to organize andplot the data It has little to no correspondence to real geo-electrical sections
Recall: each datum is affected by the entire charge accumulationassociated with electrical conductivity variation in the subsurface
(volumetric effect)
Thus: CANNOTmake one-to-one correspondence between a givendatum and a point in the subsurface
Earlier interpretation using on pseudo-section was based onmatching anomaly patterns in the pseudo-section with known
conductivity anomalies in the subsurface
Can be effect when the subsurface is simple (such as a singleconductivity anomaly)
Difficult when multiple anomalies or geological noise is present
Example pseudo-sections: Illustration of Geological Noise
Resistivity model
Pseudo-section
Dipole-dipole; n=1,8; a=10mSimple scenario: clear anomaly pattern and interpretableMultiple bodies: overlapping anomalies dominated by effect of shallowconductivity variations
Ohm-m Resistivity model Ohm-m
Pseudo-section
7/30/2019 Gpgn303 Dc Methods
45/54
12/11/11
45
Comments on numerical modeling: 2D
There is no analytic solution for complicated 2D or 3D problem We utilize numerical solution of differential equation governing the
electrical potential in conductive media (refer to course on static field)
Finite difference or finite element methods Discretize a much larger region of earth than that of interest
Region of interest Entire discretized region
Interpretation of 2D DC resistivity data: Inversion
Similar to the 1D case, we resort to inversion to quantitatively interpret the measureddata by finding asimple conductivity distribution that could have produce the data
This is accomplished by requiring conductivity (referred to as model) to satisfy twocriteria:
It must reproduce the observed potential difference data to within the error tolerance(quantified by data misfit function, as in 1D)
It must be simple and geologically interpretable (quantified by a model objective function) Parameterize the conductivity by a piece-wise constant 2D function, such that the
number of cells is much greater than the number of data
Inversion finds the conductivity values in all cells
Mis the number
of unknown
conductivity
values
7/30/2019 Gpgn303 Dc Methods
46/54
12/11/11
46
Data misfit
Model objective function
N: number of data !i is the standard of error in ith datum
m=ln("): model used in inversion (log conductivity) m0is a reference model(we want the inverted model to be
close to it)
Inversion solution
Obtain the solution by Tikhonov regularization:
where is the regularization parameter determines the balance between the two parts
We look for an optimal balance between the two components so that we fit the signal in the data, but not the noise one simple condition: data misfit equal to it expectation
Solution obtained iteratively by starting with an initial guess
7/30/2019 Gpgn303 Dc Methods
47/54
12/11/11
47
A mystery example: DC resistivity inversion
dipole dipole
Pole dipole right
Pole dipole left
Pole pole
Apparent resistivity pseudo-sections Inverted resistivity sections
Inversions with different :
"-m
best model: m0 = 400 "m
m0 = 40 "m
m0 = 4000 "m
7/30/2019 Gpgn303 Dc Methods
48/54
12/11/11
48
Quantifying the Depth of Investigation
Invert data with two different reference models The regions that dont change very much are considered to be resolved
(investigated)
The regions that change a lot with reference model are not seen by data
Regions of investigation
Regions to which data
are insensitive
3D acquisition and imaging
Most readily (approach-I): co-linear arrays along multiple lines Line spacing should be shorter than the maximum depth of
investigation of the 2D arrays
More effectively (approach-II): True 3D acquisition with cross-line measurements Distributed data acquisition: measure potential data over a
portion of the 2D grid for each current-electrode location
Cross-line acquisition analogous to cross-holeacquisition
7/30/2019 Gpgn303 Dc Methods
49/54
12/11/11
49
Induced Polarization (IP) Methods
Phenomenological observation of induced polarization effect
7/30/2019 Gpgn303 Dc Methods
50/54
12/11/11
50
IP effect is characterized by a phase lag between the observed potential
and input current in the frequency domain
Definition of chargeability
Primary property characterizing the IP effect in a rock unit is thechargeability #: ratio of secondary potential over total potential
The secondary potential is a function of delay time, so is the chargeability
A similar quantity, apparent chargeability can be defined for fieldmeasurements using co-linear arrays
Intrinsic chargeability (measured on rock samples)
Commonly used as IP data,
but not always defined just as apparent resistivity
7/30/2019 Gpgn303 Dc Methods
51/54
12/11/11
51
Relationship between apparent chargeability and
intrinsic chargeabilities
Assume the subsurface is divided intoMsubregions (or cellsas in the DC inversion)
To the first order, apparent chargeabilities are the weightedsum of the intrinsic chargeabilities
The weighting factors in the equations are called IPsensitivities (see next slide)
This relationship also forms the basis for the most practical IPinversion used in interpretation.
Siegels (1959)
dilation equation
7/30/2019 Gpgn303 Dc Methods
52/54
12/11/11
52
Linear forward modeling of IP data: sensitivities
Plots of sensitivities with a chargeable block. Contributionfrom block$for#=0.1, is #a.I V I V
I VI V
At n=1, #a= 0.0608 At n=5, #a= -0.0897
At n=2, #a= 0.0358 At n=6, #a= -0.0103
Inversion of IP data
A two-stage process First, we invert the accompanying DC resistivity data to
recover the conductivity distribution
Second, we use that conductivity to invert IP data use the recovered conductivity to calcuate the sensitivities in Siegels
dilation equation
Invert the IP data (apparent chargeabilities) using a similar approach asin the DC resistivity inversion: finding a simple intrinsic chargeability
model that predicts the observed apparent chargeabilities.
We also impose the condition that the chargeability must be positive
7/30/2019 Gpgn303 Dc Methods
53/54
12/11/11
53
A mystery model - IP inversion results
dipole dipole
Pole dipole right
Pole dipole left
Pole pole
Apparent chargeability pseudo-sections Inverted chargeability sections
Apparent
conductivity data
mS/m
Apparent
chargeability data
mrad
7/30/2019 Gpgn303 Dc Methods
54/54
12/11/11
DC resistivity Example: McDermott deposit
Observed
data
Predicted
data
mS/m
mS/m
Recovered
model
mS/m
Overburden is delineated.
IP - McDermott deposit
Observed
data
mrad
Predicted
data
Recovered
model
mrad
Ore bodys depth and position are located.