CIPC 2003
Louis Mattar
Fekete Associates Inc
Analytical Solutions
in Well Testing
Well Test Equation
P P 1 P
+
=
2
2
y
t
x
2
Solutions
Analytical
Semi-Analytical
Numerical - Finite Difference
Numerical - Finite Element
Boundary Element?
Analytical Element?
Issues
Complexity of Model / Solution
Inverse Problem
Diffusity
Issue
Complexity
Definitions
Analytical Models solve the
problem directly (transforms,
substitutions, calculus)
Solution is continuous across
domain.
Definitions
Numerical Simulation is the process
of dividing the reservoir into
discrete blocks, having different
reservoir properties, in order to
deal with complex reservoir
problems.
Well Test Equations
Analytical Vs Numerical
P P 1 P
=
+
2
2
t
y
x
2
n+1
i1,j
n+1
i,j
2
n+1
i+1,j
2P +P
x
n+1
i1,j
P
+
n+1
i,j
2
n+1
i+1,j
2P +P
y
n+1
i,j
1 P P
=
t
n
i,j
Numerical Vs Analytical
Analytical :
-Whole reservoir
-Single phase
-Exact
-Homogeneous *
-Regular Boundaries**
Irregular Shapes
Irregular Boundaries
Modified Analytical ElementsA
k1
k3
k2
Rubiks Cube (Integral Transforms)
Advantages of Analytical
Solutions
Better Understanding
Cause and effect
Physics of the process
Groups that control response
kh/q or k/c
Numerical
Numerical
- Discretize Reservoir
(1+ Million Cells)
- Complex Reservoir
- Heterogeneous
- Computationally intensive
- Multiphase**
Complexity?
or
Simplicity?
Occam's Razor
If you have two theories
which both explain the
observed facts then you
should use the simplest until
more evidence comes along
Numerical Simulation may be
justified when reservoir complexities
are known.
But how often, or how well, do we
know these reservoir complexities, in
advance?
Often these reservoir complexities
are only discovered through testing
Issue
Inverse Problem
Inverse Problem
Well test interpretation is
essentially an Inverse
problem, and in general,
is better suited to
Analytical Solutions
Direct problem
versus
Inverse problem
Direct Problem: 2 + 3 = ?
Answer : 5
Inverse Problem:
The Answer is 5.
What are the two numbers ?
Characteristics of
Inverse Problems
non-unique solutions
K2
K1
2-Boundaries or Composite?
A good looking history match
is not a good enough answer.
x
x
x
x
x
The selected MODEL must
be appropriate
Numerical models are too
complex-too many Degrees of
Freedom
Reservoir Complexities are
often unknown a priory
Cannot see forest for trees
Analytical models allow us
to focus on the main issues
Create a conceptual analysis.
Pattern recognition
Judgement
Consistency checks
Much better than numerical
simulator
Issue
Diffusivity
Diffusive Nature of PTA
7
9
5
3
6
9
2 4 7
1 2
1
8
9
8 5 4
4
7
6
3
1
6
5
3
Diffusive Nature of PTA
Homogeneous
5.2
The question is, how much
information is contained about
the spatial distribution of
permeability in the well-testing
response in a heterogeneous
formation?
Average permeability in a region
Not
Permeability at a fixed radius
Sageev and Horne (SPEFE 1988, 428 437)
It is possible to have a hole in the
reservoir
as large as half the distance
between a production well and an
observation well, without
any discernible difference
being evident in interference test
Obs
Hole
Well
Linking Analytical & Numerical
Obtain Analytical Solution
Use geostatistical model to
Generate Permeability Field
Populate Numerical Simulator
Example
Naturally Fractured Reservoir
Determine Fractal Parameters
Generate Permeability Field
Modeling can mean 2 things:
a) Model recognition from a set of
data - Well Testing
b) using a model to forecast future
performance - Numerical
Conclusions:
K eep
I t
S imple
S tupid
Einstein
Everything should be made
as simple as possible,
but not simpler
Mathematics:
Great Servant
(Analytical)
Terrible Master
(Numerical)
I would rather be
vaguely right
(Analytical)
than precisely wrong
(Numerical)
Justification for Numerical
Detailed Geological Description
of complex reservoir geometry
available in advance
Multi-phase fluid flow, where gas
is not fluid of interest
Water cut matching