SPE 93992

The geochoke well test response in a catalogue of systematic geotype curves

Corbett, P.W.M*, Ellabad, Y. 1, Egert, J.I.K. 2, and Zheng, S*
Institute of Petroleum Engineering, Heriot-Watt University, Edinburgh, EH14 4AS
1
now Wintershall, Tripoli, Libya and 2Wintershall, Kassel, Germany
*SPE Members
2004 Society of Petroleum Engineers Inc.
This paper was prepared for presentation at the 14th Europec Biennial Conference held in
Madrid, Spain, 13-16 June 2005.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the SPE, their officers, or members. Electronic reproduction, distribution, or storage
of any part of this paper for commercial purposes without the written consent of the Society of
Petroleum Engineers or the International Association of Drilling Contractors is prohibited.
Permission to reproduce in print is restricted to an abstract of not more than 300 words;
illustrations may not be copied. The abstract must contain conspicuous acknowledgment of
where and by whom the paper was presented. Write Librarian, SPE, P.O. Box 833836,
Richardson, TX 75083-3836, U.S.A.,
fax 01-972-952-9435.

Abstract
A new well test response has been identified. A distinctive
restriction shown as a hump on the derivative curve is seen
in double matrix porosity (dual permeability) reservoirs. High
permeability lenses intersecting the well bore give rise to a
negative geoskin response at early time. After the middle time
period of flow restriction caused by the limits of the lenses
connecting to the wellbore and the presence of additional
limited extent, discontinuous, high permeability lenses within
the reservoir. This phenomenon has been shown in a series of
numerical models and is also observed in two field well tests.
The two field examples come from testing in a braided fluvial
oil reservoir. Other interpretational possibilities (e.g., subseismic faulting, fluid changes), which might also be
considered given the non-uniqueness of pressure data, are
eliminated by careful consideration of the geophysical
interpretation, the depositional environment and the fluid
properies. Sub-seismic faults are often interpreted in well tests
from braided fluvial reservoirs. However, we consider that the
response in these cases, and possibly in a wider range of cases,
is due to geological effects of sedimentary heterogeneity due
to secondary channels within the braided system. Braided
fluvial sedimentary processes result in heterogeneous
reservoirs with double matrix porosity. The heterogeneity and
the double matrix porosity nature of these reservoirs are
evident from the Lorenz Plot. Such sandstones can produce a
variety of well test responses as the two matrices interact at
various length scales.
The geochoke model in double matrix porosity reservoirs is an
end-member response for reservoirs with short lateral and
vertical correlation length. A family of geotype curves can be
generated to consider a range of correlation lengths to
systematically explore the limits of early, middle and late time
responses for specific reservoir cases.

Introduction
The occurrence of negative skin in a braided fluvial reservoir
as a result of reservoir heterogeneity1, rather than fractures, led
to a numerical modelling study2 of the phenomena and finally
to the development of an analytical solution3. The phenomena
exists where there are lenses at least an order of magnitude
higher than the background, the lenses do not extend further
than 10% of the depth of investigation of the well test and
there is cross-flow from the background matrix.
Another characteristic of these variable high net-gross fluvial
reservoirs is that the flow often comes from single, or a small
number of point sources4 and these can be detected by
production logs and are correlated with either the coarser
grained and/or better sorted geological elements. Recent
flume tank experiments have shown those elements to be
deposited in secondary channels within the fluvial system5.
The distribution of patches of relatively high reservoir quality
reservoir sandstones in a matrix of lower reservoir quality
sandstone creates a double matrix porosity (also known as a
Dual Permeability) reservoir. The Lorenz plot can be used to
diagnose this sort of reservoir type. Commonly, the flow in
the well bore will be coming from a small proportion of the
net sandstone interval, as defined by a permeability, rather
than a porosity, cut-off. The stratigraphically-ordered Lorenz
plot has been used to identify the flowing elements in cored
intervals5 and these will correlate well with production logs4 in
wells that are not fractured.
The development of numerical well testing tools and the
combination with 3-D geological models allows the realistic
distribution of these patches in a volume sufficient to carry out
reasonably long well tests. From these simulations we can see
that the geoskin response2 is an early time phenomenum due to
a restricted flow in middle time as the pressure response
extends beyond the high permeability patches connected to the
well. In late time, the presence of additional high
permeability patches beyond the restriction leads to an
expansion of flow following from the restricted flow regime in
middle time, defined here as a geochoke.
Characteristics of Braided Fluvial Reservoirs
Braided fluvial reservoir deposits are formed as the coarser
grained, higher energy part of fluvial systems. These
reservoirs have characteristic patchy distribution of varying
grain size and sorting in outcrop sections (Fig. 1). Recent

CORBETT, ELLABAD, EGERT AND ZHENG

flume tank experiments have shown those elements to be

deposited in secondary channels within the fluvial system
(Fig. 26). The geometry of the secondary channels results in
randomly distributed 3-D patches.
Geological model of a high net:gross braided fluvial
system
A numerical petrophysical model, to represent this geological
distribution, is built using a correlated random stochastic
model with varying correlation lengths (Fig. 3) to distribute
the hydraulic units which are then populated by appropriate
petrophysical properties. In the numerical model, the porosity
has been kept constant and the permeability has been allowed
to vary according to the distribution of rock types4. Fig. 3
shows one realisation of a stochastic permeability model.
There are possible geological arguments for making these
patches longer in the downstream direction but as the direction
can radiate over a wide range in braided fluvial systems, the
approximation of an isotropic areal variogram model is
considered sufficient.
Production log data (PLT) from these braided fluvial systems
often shows point source entry points for fluid (Fig. 4). These
high permeability intervals correlate with the best reservoir
property core plugs, classed as hydraulic unit 1 (HU1) in this
example. The Lorenz plot for the interval shows that HU1
contains 70% of the transmissivity in the well in only 15% of
the porosity this PLT response and degree of heterogeneity
is often observed in braided fluvial reservoirs.
In the reservoirs in North Africa that were the basis of this
study the porosity is moderate (10-15%). The background
average permeability was around 3mD with the best
permeability zones approaching and sometimes exceeding
100mD.
Numerical well test model: The geostatistical models (an
example shown in Fig. 3) were incorporated into a black oil
simulator. A single well was placed in the centre of the model
with local grid refinement. Numerical artefacts can often
occur in simulation models when a radial local grid refinement
around the well is nested within a cartesian grid. For this
reason the model described here uses a cartesian grid
throughout. Careful review of simulation models has
eliminated any significant numerical artefact in the middle
time region. However, numerical artefacts may still be present
in the early time period. Because the phenomea we are
describing occurs in some realizations but not in others (Fig.
6), whilst the grid remains constant, we are confident that the
phenomenum that we observe is free of numerical artifact.
Geotype curves for the braided fluvial model
A limited number of cases were simulated in this study. Five
stochastic realizations for each of three varying correlation
length models were run7. The case illustrated in Fig. 3 is for a
medium correlation length (MCL) model of approximately
200 foot. Pressure drawdown and buildup were simulated
and the buildups interpreted as in a real well test. A welldefined hump in the middle time region is seen on two of the
five realizations for this correlation length. Shorter and longer

SPE 93992

correlations were also simulated. For shorter correlation

lengths the hump response was more common. As the
correlation length is extended then the phenomenum becomes
rare and eventually disappears. These observations were made
on a relatively low permeability model. These responses
would have to be scaled for higher permeability systems.
Geochoke phenomenum: The restriction of flow for a short
period of time represents depletion of the high permeability
zones connected to the well, and the delay in recharging from
other patches away from the well. This model requires there
to be a fairly high density of patches and for some to be
intersected by the well. This is why only some numerical
realisations show the hump and some dont to the same
degree.
Field validation: We have identified a humped middle time
response in a number of field examples. Figures 7 and 8 both
show humped derivatives and these data have been interpreted
in by various combinations of radial composite and fault
models. We suggest that these data show the geochoke
phenomena, for which there is not currently an easy method to
apply either an analytical solution or a type curve match.
Fig 7 shows a derivative from a well test in a North African
reservoir in a braided fluvial environment. The mapped faults
in this field were at a distance from the well that was outside
the radius of invesitigation. Clearly sub-seismic faulting could
be invoked. The geochoke response could be reasonably
modelled by an analytical radial composite model with a lower
permeability ring around the well8. Simple faulting would not
give this response. In the full field simulation model more
extensive channels had been modeled and low permeability
regions around the well were needed to improve the match8. It
was beyond the scope of this work to rebuild the reservoir
model. This well test interpretation suggests that a model
should be updated to include short correlation length patchy
high permeability sands, within the more extensive channels.
In a second well from a different field in North Africa (Fig. 8)
a subtle hump can be seen on the derivative in the middle time
region. This is interpreted as a example of a geochoke
phenomena.
Geotype catalogue for double matrix porosity
reservoirs
The geochoke response is seen in double matrix (also call dual
permeability by the well testing community) reservoirs when
the correlation length is short (say up to 10% of the depth of
investigation or 50-500ft) and the proportion is high and
areally reasonably consistent (certainly more than 10%). As
the correlation length in vertical and horizontal orientations
becomes larger this response disappears systematically as
shown in the geotype catalogue (Fig. 9) which is constructed
around an architectural matrix9 for double matrix porosity
reservoirs. A catalogue encourages the development of type
curves for geological phenomena in a systematic framework.
In most geological situations a particular situation around a
well and its resulting well test buildup response are a member
of a family of curves best captured by using a systematic range

SPE 93992

GEOCHOKE WELL TEST RESPONSE IN A CATALOGUE OF SYSTEMATIC GEOTYPE RESPONSES

of geostatisical and petrophysical model parameters.

Nomenclature

Sub-seismic faults and their interpretation in Braided

Fluvial Systems
With knowledge of the existence of the geochoke model, the
well test interpreter should be very careful not to misinterpret
the phenomenum as a fault response, particulary in braided
fluvial systems. It is often said that sub-seismic faults
interpreted in braided fluvial systems have to be removed
during later production as they are not seen in later test
interpretations or during production. These faults, we
believe, can be misinterpreted geochoke phenomena. Other
high net:gross environments with significant petrophysical
heterogeneity (e.g., mixed fluvial aeolian, deep-water
channelised systems) may also have property variations and
architectural arrangements that will cause a similar buildup
pressure response. The Lorenz plot is a useful screening tool
to identify whether the contrasts in permeability exist to set up
the right conditions for the choking phenomena. Production
log data are also a useful indicator of point source entry points.

HU
h
k
MCL
PLT

Conclusions
Patches of coarse and/or well sorted, high
permeability sand are often present in braided fluvial
systems and are identified as the preserved secondary
channels of limited lateral extent.
These patches produce restricted point entry points
for flow into the well bore in high net to gross
braided fluvial reservoirs.
The Lorenz plot and production log data can provide
useful diagnostic information in such reservoirs and
should be used in support of the well test
interpretation.
The middle time well test response in a double matrix
porosity reservoir with short vertical and horizontal
correlation lengths might display a geochoke
restriction (depending on the location of the well with
respect to local high permeability patches, the
correlation length, and the contrast between rock type
properties).
Interpreting the geochoke restriction as infinite acting
radial flow, if the test is not long enough to see the
full hump, will result in a negative skin response.
The geochoke phenomena can be confused with a
sub-seismic fault response as they will be very
difficult to diagnose. Well test interpreters should be
particularly careful in braided fluvial systems to keep
this alternative model in mind.
Acknowledgements
The authors acknowledge the support of ConocoPhillips,
Shell, Wintershall, and Yukos as sponsors of the Geotipe
Project at Heriot-Watt University. The authors also would like
to thank, GeoQuest RT, Edinburgh Petroleum Systems for the
providing use of their software. YE and JE also acknowledge
the support of Wintershall during their PhD and MSc studies,
respectively.

Hydraulic Unit
Thickness
Permeability
Medium Correlation Length
Production Log
Porosity

References
1.

Toro-Rivera, M., Corbett, P.W.M., and Stewart, G., 1994,

Well test interpretation in a heterogeneous braided fluvial
reservoir, SPE 28828, Europec, 25-27 October

2.

Corbett, P.W.M., Mesmari, A., and Stewart, G., 1996, A

method for using the naturally-occuring negative geoskin
in the description of fluvial reservoirs, SPE 36882,
presented at Europec, Milan, 22-24 October.

3.

Sagawa, A., Corbett, P.W.M., and Davies, D., 2000,

Pressure transient analysis of reservoirs with a high
permeability lens intersected by the well bore, Jounal of
Petroleum Science and Engineering, Sept., 27(3-4), 165177.

4.

Ellabad, Y., Corbett, P.W.M., and Straub, R., 2001,

Hydraulic Units approach conditioned by well testing for
better permeability modelling in a North Africa oil field,
SCA2001-50, Murrayfield, 17-19 September, 2001.

5.

Gunter, G.W., Finneran, J.M., Hartmann, D.J., and Miller,

D.J., 1997 Early determination of reservoir flow units
using an integrated petrophysical approach, SPE 38679,
Proceedings of the Society of Petroleum Engineers
Annual Technical Conference and Exhibition, Formation
Evaluation and Reservoir Geology, Part 1, 373-380.

6.

Moreton, D.J., Asworth, P.J., and Best, J.L., 2002, The

physical scale modeling of braided alluvial architecture
and estimation of subsurface permeability, Basin
Research, 14, 265-285.

7.

Ellabad, Y., 2003, A method for reservoir modeling

incorporating geostaistical models of flow-storage
elements calibrated by dynamic tools, Unpubl. PhD
Thesis, Heriot-Watt University, Edinburgh, 290p.

8.

Egert, J.I.K., 1004, Numerical well test forward modeling

and interpretation of extended pressure transient tests
from an oil field in North Africa, Unpubl. MSc Thesis,
Heriot-Watt University

9.

Tyler, N. and Finley, R.J., 1991, Architectural Controls on

the Recovery of Hydrocarbons from Sandstone
Reservoirs, SEPM Concepts in Sedimentology and
Palaeonotology, V3, p3-7.

CORBETT, ELLABAD, EGERT AND ZHENG

SPE 93992

Figures

Fig. 1: A view of a typical braided fluvial system at outcrop

(Eocene Escanillia Formation, N. Spain). The figure on the
left is identifying the coarser grained elements the figure of
the right, the finer. It is likely that the better sorted finer
elements will have a significantly different permeability for
the poorly sorted coarser elements. Note that there are no
obvious flow barriers between the elements to stop crossflow.

Fig. 2: The result of a laboratory analogue study of a braided

fluvial system. The high permeability class units in D relate to
the secondary channel fills in C that have been mapped in
cross sections (A and B) through modern sediments deposited
in a flume tank (from Moreton et al., 2002)

Fig. 3: Isotropic correlated random field to capture the

distribution of high permeability patches (the rock type to the
left of the scale bar which shows four rock types with
reducing properties from left to right).

Fig. 4: Point inflow from thin high permeability (HU1)

interval in a braided fluvial reservoir. The production log
(PLT) inflow (solid line) is predicted by the cumulative core
permeability. Note that the inflow corresponds to plugs from
hydraulic unit 1 (HU 1), the rock type with the best reservoir
properties.

SPE 93992

GEOCHOKE WELL TEST RESPONSE IN A CATALOGUE OF SYSTEMATIC GEOTYPE RESPONSES

Fig. 5: Lorenz Plot for a braided fluvial reservoir. The best

permeability units (HU1) contain 70% of the k*h
(transmissivity) but only 15% of the *h (storage). Braided
fluvial reservoirs contain double matrix porosity part of the
matrix is transmissivity-dominated (in this case with only 15%
of the pore volume) and part is storage-dominated (85% of the
pore volume).

Fig. 6: Numerical simulations from five realizations (MCL1-5)

of the model shown in Fig. 3. The hump in the middle time
regime is only obvious in only two of the realisations. This
result suggests that the geochoke phenomena will occur only
for particular arrangements of the high permeability patches
around the well. It also shows that the effect can be quite
subtle. Upper set of curves show pressure buildup, lower set
show derivative.

Fig. 7: Field example of a pronounced humped middle time

region in a well test buildup from a braided fluvial reservoir in
a North African fluvial reservoir example. Upper curve is
pressure buildup and lower curve is the derivative.

Fig. 8: Field example of build-up data from a test a second

North African reservoir example showing more subtle
humping (arrowed) of the derivative. Upper curve is pressure
buildup and lower curve is the derivative.

CORBETT, ELLABAD, EGERT AND ZHENG

Fig. 9: The geochoke response occurs in double matrix

porosity systems with short lateral and vertical correlation
lengths. The Lorenz plot on the left defines the transmissive
and storage elements. The matrix in the centre shows
decreasing correlations in the vertical and horizontal
correlation lengths from top left to bottom right (based on the
Tyler and Finley architectural matrix9). The various derivative
responses relate to well test characteristics for various
scenarios.

SPE 93992