Geostatistics for Reservoir Characterization
In this lecture you will The origin of randomness
What is required for randomness to exist The definitions of random, deterministic and stochastic
learn
Randomness is
behavior of a system or process which produces uncertain results systems may show extreme sensitivity to initial conditions
Such systems have 2 properties
Nonlinear physical law Iterative processes
Apparently simple mechanical systems can show unpredictable behaviour
The Outcome of a Flipped CoinFig. 1.1
Height of Toss
Rate of Spin
Examples of Nonlinear Physical Laws
Basic well flow equation
q=
2khP (In(re/rw)+s)
nonlinear in re, rw and s
Two phase fractional flow
fw =
krw(Sw)/w ____ krw(Sw)/w + kro(Sw) /o
nonlinear in krx krx nonlinear in Sw
Examples of Iterative Processes in Geology
Diagenesis
Growth of clays Leaching Cementation
Deposition
Sedimentation Erosion Breakage
Sedimentation
Rock breakage is a non-linear iterative process
1-f0 1-f1
Dpo Dp1 f1
f0
Dp2
...
Dpi = (f0Dp0) x (f1) x (f2) x (f3) x ..
Systems with well-defined boundary conditions can appear to behave unpredictably
May's Equation
xi+1 = xi(1 - xi)
1.0 0.8 0.6 x 0.4 0.2 0.0 3.5
3.6
3.7
3.8
3.9
4.0
Summary so far
Randomness = unpredictable For unpredictability, all we need is a process that is
non-linear iterative
Definitions
Random - Having no predictable aspects Deterministic - Completely predictable Stochastic - Partly predictable, some randomness
Measurements
Are snap-shots of the reservoir, providing needed values for engineering calculations Have a scale which represents a rock/fluid property over a certain volume V
WELL TESTING
core plug
V = 10-6 m3
WELL LOGGING
V=
10-2
m3
V = 106 m3
May be biased, inaccurate, or misleading
Example Core Plugs
Core from barrel Approx 30ft sections Incomplete recovery
Gaps - where? How much?
1.2 ft 1.7 ft
Biased sampling
1ft sample spacing? Core photos useful
What does each observation represent?
Summary so far
Systems can have various amounts of unpredictability
Random Stochastic Deterministic
Need care deciding if a quantity is random
The measurements can fool us The geology might be deterministic (next example!)
Reservoir Properties and measurements example
9950
Non-linearity and randomness cause variability (e.g., perm k) But deterministic part is also there (can you find it in this example?)
10000
Depth, ft
10050
10100
10150
10200 1 10 100 1000 10000 Plug Permeability, mD
Reservoir Properties usually have a deterministic component
Lamina-scale (mm to cm) k
Nested scales of variation can make property look very unpredictable Physical laws give some predictability (scales or stratal elements)
Bed-scale (cm to m)
Formation-scale (m to100's m)
Reservoir Properties and measurements: some details
Stratal element
PARASEQUENCE Thickness (m) Extent (sq. km) Time (yrs)
1
Depending on volume V, measurements
capture some variable & deterministic parts miss out smaller scales
Measurement resolution
10 1 0.1 50 5 0.5 1000
BEDSET
Seismic
BED
Plug
LAMINASET
LAMINA
Probe Thin section
Well log Whole core
Another example
Changing value with measurement volume V
kh/kv
from Cowan and Bradney, 1997
Same example (replotted)
Change value with measurement volume V
kv/kh and volume of measurement
1 kv/kh 0.1 0.01
probe plug tester tidal
0.001 0.001
0.1
10
1000
"scale" of msmt, ft
Summary
Even deterministic systems may exhibit randomness
need iteration need nonlinearity
Measurements
are snap-shots of a reservoir property have a volume (scale), depending on measurement may be wrong/misleading
Reservoir properties
are variable have some systematic elements measurements have to capture both aspects
