See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/254514013

Will Wireline Formation Tests Replace Well Tests?

Article · October 2003

DOI: 10.2118/84086-MS

CITATIONS READS

28 866

3 authors, including:

Alain Gringarten
Imperial College London
159 PUBLICATIONS   3,335 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

New Developments in Well Test Analysis View project

All content following this page was uploaded by Alain Gringarten on 28 October 2015.

The user has requested enhancement of the downloaded file.

WTOver.qxd 1/13/04 1:37 PM Page 56

OVERVIEW
Well Testing

Even in the current U.S. $30/bbl environment, cost reduction and maxi- Additional
mizing efficiencies continue to be a subject of focus within our industry.
Further, on the revenue side of the equation, activities that reduce or
Well Testing
upset the income stream receive their proportionate share of scrutiny. Technical Papers
Well tests continue to be standard procedure for petroleum engineers to
increase their understanding of reservoir characteristics in a given field. Available at the SPE eLibrary:
Stratton Consequently, techniques and tools that lower intervention costs, www.spe.org
enhance the value of existing data, and reduce the downtime associated
with well-testing operations are of great interest. • SPE 84087
Formation Testing While
Technology has a positive effect on our ability to gather the critical reservoir information, Drilling: a New Era in
sometimes even before drilling is complete. Improvements to existing formation-testing Formation Testing
tools widen their use as a replacement of standard well tests. Adapting this same technol-
ogy to logging-while-drilling tools offers some of the same information previously avail- • SPE 84475
able only from wireline-deployed tools. Further enhancements to these tools in the years Integrating Short-Term
to come likely will mitigate their existing shortcomings and widen their appeal, particu- Pressure-Buildup Testing and
larly in extended-reach horizontal drilling. Long-Term Production-Data
Analysis to Evaluate
Well-test-analysis techniques continue to advance with headway being made on the Hydraulically Fractured Gas-
analysis challenges of gas/condensate wells and interference testing in complicated reser- Well Performance
voirs. More marginal properties will benefit from new approaches to well-test analysis
with the use of short buildup tests and decline-curve analysis. Advances in applying pres- • SPE 84290
sure/rate deconvolution to real data may someday result in commercial software packages Practical Application of
offering these analysis capabilities. Combining this analysis with permanent-gauge data Pressure/Rate Deconvolution
can reduce or eliminate effects on production. That is something positive we can all agree to Analysis of Real Well Tests
on. The following pages describe and reference many of these advances in well-test tech-
nology and analysis. JPT

Jay Stratton, SPE, is the Petroleum Engineering Team Lead for Anadarko's Groupement
Berkine partnership with Sonatrach Inc. in Hassi Messaoud, Algeria. His responsibility is
in the area of new well completions, stimulations, testing, and production surveillance.
Previously, he was with Occidental Petroleum of Qatar as Operations Engineering Advisor
in the areas of horizontal multilateral technology, innovative ESP completions, polymer
water-control treatments, and optimization of offshore well interventions. While at Arco
Alaska as a Senior Production Engineer, he was involved in drillsite development,
hydraulic-fracturing technology, and well services. In his 16-year career, Stratton has
worked in the Arctic, North Africa, Middle East, FSU, Pakistan, and West Texas. Stratton
has authored technical papers on application of multilateral completions and polymer
water-control technology. He serves on the JPT Editorial Committee and holds a BS degree
in petroleum engineering from Texas A&M U.

56 FEBRUARY 2004
84086.qxd 1/13/04 1:52 PM Page 57

W e l l Te s t i n g

Will Wireline Formation Tests

Replace Well Tests?
Well testing in exploration and appraisal The strongest reason not to perform the
wells has become increasingly unpopu- well test is financial. Environmental cost is This article, written by Technology Editor
lar. Reasons include costs, safety, and increasingly important. The decision to test Dennis Denney, contains highlights of
environmental effects. Well testing also must be made taking into account the cost paper SPE 84086, “Will Wireline
has become rare in production wells of acquiring the information. This decision Formation Tests Replace Well Tests?,” by
because of the potential revenue loss requires an understanding of the informa- T.M. Whittle, SPE, and J. Lee, SPE,
during the buildup tests. Alternatives for tion and whether it can be acquired by Baker Atlas, and A.C. Gringarten,
sampling and reservoir-parameter esti- other means. SPE, Imperial College, U.K., prepared
mation include wireline formation tests The full-length paper focuses on informa- for the 2003 SPE Annual Technical
in exploration and appraisal wells and tion that can be obtained from the pressure Conference and Exhibition, Denver,
continuous recording with permanent transients recorded during a wireline forma- 5–8 October.
pressure gauges in production wells. tion test and how the information compares
Essentially, the interpretation methods with data recorded during a well test.
are the same as those used in well-test With fast PCs and properly designed by the skin factor. Solutions to the spheri-
analysis with the addition of the forma- analysis software, the interpretation of pres- cal-source problem are well documented.
tion-rate analysis, which is particularly sure transients from well tests has become The selection of the correct independent
useful in high-permeability formations somewhat standard. The interpretation of dimensionless parameters and the addition
in which other methods are limited by pressure transients from wireline formation of storage and skin are discussed in
pressure-gauge resolution. tests is a more recent development. Appendix A of the full-length paper.
The established methods used in well Formation-Rate Analysis (FRA). This
Introduction testing are not always applied to wireline technique for analyzing formation-tester
Often, well testing is used in the exploration formation testing. In high-permeability pressure data is based on the material-bal-
and development of hydrocarbon reservoirs zones, gauge resolution may be insufficient ance equation for the formation tester’s
to obtain representative formation-fluid to record an interpretable pressure tran- flow-line volume considering pressure and
samples, measure initial reservoir pressure, sient. Well-test analysis software has not compressibility of the enclosed volume.
demonstrate and/or establish well produc- been adapted for use with wireline-forma- FRA allows for the variation in flow rate and
tivity, determine permeability-thickness tion-test data. Therefore, a service company includes both drawdown- and buildup-
product and skin, identify the drainage area often performs the analysis using its propri- pressure data. It also assumes Darcy flow
of the well along with boundary effects that etary software. from the reservoir to the probe.
may exist, and identify and quantify deple- Transient Analysis. The constant-rate
tion. These objectives can be compared with Wireline-Formation-Test Analysis spherical-source solution with storage and
those of a wireline formation test, which Fig. 1 shows the fundamental types of skin is adapted to a variable rate by use of
include determining formation pressures in wireline formation tests considered in superposition. Comparing its response with
zones of interest, establishing pressure gra- this study. that of the FRA method indicates that a
dients for fluid-type identification, identify- dimensionless geometric skin of approxi-
ing zones in hydraulic communication or Single Snorkel. Several authors have inves- mately 1.8 is required to obtain the same
isolation, collecting representative forma- tigated the theoretical
tion-fluid samples, and estimating forma- pressure-transient re-
tion-fluid mobility. sponse of the single-
An overlap exists between the two tech- snorkel (probe) -type con-
niques, and whether one technique can figuration. The approach
replace the other depends on specific-well in this paper is to consider
objectives. For example, some exploration the withdrawal of a slight-
wells are drilled solely to confirm the exis- ly compressible fluid at
tence of a hydrocarbon column, in which constant rate from a spher-
case a wireline formation test is probably ical source having storage
sufficient. Other wells may be drilled to and skin in an infinite
prove a minimum volume of hydrocarbons homogeneous reservoir.
in place, for which a reservoir-limit well test The difference between
is the only option. Between these two the actual probe-well
extremes, many cases may be unclear geometry and the spheri-
Fig. 1—Types of wireline formation tests.
whether a well test is required. cal source is approximated

FEBRUARY 2004 57
84086.qxd 1/13/04 1:53 PM Page 58

mobility from each method, indicating that Only one of either the bed thickness or assumed that the reservoir interval could be
FRA takes into account the nonspherical the permeability anisotropy can be evaluat- described as a sequence of homogeneous
flow geometry resulting from the presence ed from a single-probe-type data analysis. (but possibly anisotropic) layers. Then, the
of the wellbore. Numerical studies that con- Assuming that one of these parameters is average permeability-thickness product of
sider both probe geometry and the wellbore available from other sources of data (usual- such a multilayered reservoir could be calcu-
confirmed that the spherical-source model ly the thickness from logs, cores, or images), lated from the individual-layer properties.
matches the numerically modeled transients then the other (permeability anisotropy) Assuming that the result of a wireline-for-
with errors no greater than 13% when the can be estimated from the analysis. mation-test analysis is a description of
appropriate skin factor is used. spherical permeability for each layer, the
Field Example 1. Fig. 2 shows the meas- Dual Snorkel. In this study, a simplified additional information required to calculate
ured pressure and rate response recorded approach appears to yield adequate results. the average permeability thickness is the
during a wireline formation test. The figure The source probe is considered a spherical permeability anisotropy and the thickness
also includes a match of the spherical- source, and the observation probe measures of each layer. Permeability anisotropy may
source model to the data using the input the pressure at some distance above or be obtained from a wireline-formation-test
data and match parameters. The FRA yields below the producing-source probe. To vali- analysis, but the thickness of each layer is
a higher mobility than that derived from date the model, the response of the obser- obtained from other sources (usually logs,
transient analysis, which also predicts a vation probe, under different anisotropy but at the smaller scale, images may be use-
lower-than-expected skin because of perme- conditions, was compared with that of a ful). It also must be assumed that the spher-
ability anisotropy. Normally, FRA mobility previously published prediction using the ical permeability derived from the forma-
approximates spherical mobility, but it is same set of data. The comparison graph tion-test analysis relates to a single layer and
affected by anisotropy. The transient analy- showed good agreement between the two. is not influenced by adjacent layers.
sis should always predict spherical mobility Otherwise (for example, when the probe is
regardless of anisotropy, although the skin Straddle Packer. The model used to close to a layer boundary), FRA may pro-
will change accordingly. describe the pressure-transient response of a vide a better esimate of permeability.
In this example, the calculated radius of wireline formation test using inflatable Permeability anisotropy may be derived
investigation (120 cm), based on the entire straddle packers is identical to that for a par- from resistivity logs as well as from cores.
shut-in period (204 seconds), is probably tially completed well. In ideal conditions, Having estimated permeability thickness,
realistic. However, in higher permeabilities, such a test can quantify horizontal and ver- the prediction of well productivity can be cal-
the pressure-gauge resolution limit will be tical permeability. culated. For example, the simplest expres-
reached much earlier, after which no mean- Field Example 3: Straddle-Packer Pump- sion for the productivity index assumes pseu-
ingful transients are recorded and the ing Test. The derivative response showed a dosteady-state flow to a fully completed ver-
radius of investigation is reduced. In the negative half slope indicative of spherical tical well in a circular drainage area.
absence of measurable pressure transients, flow. However, radial flow had not devel-
the FRA method becomes the only means oped, and, therefore, the estimate of hori- Conclusions
to estimate mobility. zontal permeability thickness represents a In lower-permeability reservoirs (mobilities
Field Example 2: Application to Thin minimum value because any higher value less than 100 md/cp), the quality of data
Beds. This example includes a repeat test to could result in an equally good match. recorded by wireline-formation-test tools is
validate pressures recorded during the ini- Because the rates are much higher during suitable for pressure-transient interpreta-
tial test and demonstrates the high quality pumping with straddle packers, the transient tion. In higher-permeability reservoirs, the
and repeatability of data that can be response is much better than that obtained resolution of the pressure gauge limits the
acquired during such tests. The inclusion of with a probe-type pressure test. Compared quality of the data, often precluding tran-
upper and lower boundaries to simulate with a well test, these rates are still quite low sient analysis. In this case, the FRA method
thin beds, which is suggested by the stabi- and, in high-permeability zones, gauge reso- provides the best estimate of mobility.
lization in the derivative response, necessi- lution will limit the quality of pressure tran- In general, pressure-transient analysis of
tates introducing permeability anisotropy sients. In open hole, there also is a time limit wireline formation tests provides estimates
into the model. The FRA and spherical on how long the straddle packers can remain of spherical permeability. In thin beds of
mobilities are very similar, which suggests in place safely. This limit may reduce the known thickness or in cases in which an
that the formation permeability is isotropic. available pressure-transient data and could observation gauge is used to measure verti-
compromise the analysis. cal interference, the possibility to evaluate
permeability anisotropy also exists. Scaling
Scaling Up up the permeabilities derived from wireline
After several formation tests were formation tests to a prediction of the perfor-
conducted on a well in a potential mance of a fully completed well is possible
producing interval, the challenge was if several assumptions are made. In particu-
to scale up the interpreted permeabil- lar, the permeability anisotropy must be
ity values of each test to a single per- known or estimated. JPT
meability thickness for the entire
interval in an attempt to predict the For a limited time, the full-length paper
well’s production performance. This is available free to SPE members at
process could be done only with www.spe.org/jpt. The paper has not
Fig. 2—Single-snorkel formation test.
information from other sources (e.g., been peer reviewed.
logs, cores, and images). First, it is

58 FEBRUARY 2004
84291.qxd 1/13/04 1:53 PM Page 59

W e l l Te s t i n g

Effect of Scaleup and Aggregation on

the Analysis of Inter ference Tests
This paper demonstrates the effect of require a scale-down algorithm. A formal
detailed small-scale heterogeneities on and systematic process to interpret coarse- This article, written by Technology Editor
interference tests. Specific issues encoun- scale adjustments at the fine-scale level is Dennis Denney, contains highlights of
tered when interference tests are ana- not available at this time. Simply making paper SPE 84291, “Effect of Scaleup and
lyzed in reservoirs with complex geologi- adjustments to variables such as permeabil- Aggregation on the Analysis of
cal properties are discussed, which relate ity and porosity at the computational level Interference Tests,” by Rajagopal
to questions concerning the use of low- amounts to ignoring or disregarding the Raghavan, SPE, and Ralph
resolution models, the degree of aggrega- considerations that form the basis of the R. Roesler, SPE, ConocoPhillips, and
tion, method of scaleup, and reliability of geological model. If the purpose of testing is O. Inanc Tureyen, SPE, Stanford U.,
conventional methods of analysis. to calibrate models to verify spatial and prepared for the 2003 SPE Annual
petrophysical relationships, then it is neces- Technical Conference and Exhibition,
Introduction sary to go beyond conventional interpreta- Denver, 5–8 October.
Guidelines concerning the evaluation of tions. Therefore, suggestions are made to
interference or pulse tests in reservoirs with make adjustments at the fine scale.
complex geology are presented. The paper lows bimodal log-normal distributions,
discusses how small-scale heterogeneities Method whereas the vertical permeability distribu-
affect the responses and analysis of interfer- Although a specific reservoir was consid- tion is log-normal and somewhat unimodal.
ence tests. Interference and pulse tests pro- ered in this study, many of the conclusions Fig. 2 is a plan view of the well locations.
vide a measure of reservoir heterogeneity. are general in nature. All wells are assumed to be vertical and are
The consequences of working with a low ver- perforated throughout the vertical extent of
tical resolution were examined, as is normal- Fine-Scale Model. This model was original- the reservoir. Wellbore-storage effects are
ly done in most analyses. The consequences ly generated for use in the Production assumed to be nonexistent at all wells. Well
of this kind of model have a significant bear- Forecast Uncertainty Quantification project. 1 is the pulsing well.
ing on the desired spatial and petrophysical The specific portion of data used here con-
relationships. The vertical resolution is sists of 1,122,000 cells with porosity and Results
important. The scale chosen to analyze data permeability values given on a 60×220×85 For the time period considered here, the
influences the characteristics of the model Cartesian grid. Each cell has dimensions of influence of the boundaries was negligibly
used to describe the reservoir, which in turn 20×10×2 ft (totaling 1,200×2,200×170 small. Because distance was accounted for,
will have a significant bearing on the evalua- ft). The model represents two formations the spread in the curves indicates a nonho-
tion of fluid movement in injection/produc- that form a part of the Brent (Middle and mogeneous system (also nonisotropic). For
tion schemes. In doing so, the consequences Upper Jurassic) sequence in the North Sea: all times considered here, the pressure drop
of aggregation and scaleup are illustrated. the Tarbert and the Upper Ness, 70 ft (35 was greatest at Well 2, the closest location
layers) and 100 ft (50 layers) thick, respec- (130 ft north) to the pulsing well. Similarly,
Literature Review tively. The Tarbert formation is a near-shore, the pressure drop at Well 3, the farthest loca-
The full-length paper details a literature prograding environment, while the Upper tion (255 ft southwest) from the pulsing well,
review of deterministic and statistical meth- Ness is a fluvial environment. The top was the smallest. Consideration of the other
ods of relating reservoir heterogeneity to image in Fig. 1 is a 3D
well performance and pressure tests. Most view of the entire model,
of the studies pertain to groundwater sys- and the lower image is
tems because interference tests are the norm only the Upper Ness for-
in that discipline. Use of geological models mation. The horizontal
to determine reservoir properties with inter- permeability of the
ference tests has not been considered in the Upper Ness formation
petroleum literature. One study considered follows a unimodal log-
a heterogeneous model. The authors used a normal distribution and
numerical model to understand the internal is isotropic; however,
architecture of a reservoir. It should be the vertical permeability
noted that adjustments to model properties distribution is log-nor-
usually are made at the coarse scale and not mal and bimodal. The
at the fine scale. Interpretation of the conse- Tarbert formation is dif-
quences of these adjustments at the geolog- ferent in that the hori- Fig. 1—3D image of fine-
scale model. Fig. 2—Well locations.
ical scale is not a simple matter and would zontal permeability fol-

FEBRUARY 2004 59
84291.qxd 1/13/04 1:54 PM Page 60

wells showed no specific pattern. The key range was used in which the
point is that very few curve characteristics influence of the boundaries was
suggest that the wells are responding to a negligibly small.
nonhomogeneous reservoir. In fact, each To understand the results, it is
response may be readily matched to the line- useful to review the conse-
source solution. quences of using a coarse-scale
model. Those involved in scaleup
Aggregation and Scaleup. An appropriate are primarily concerned with two
model must be selected to analyze pressure issues: ensuring that the pres-
tests. Often, a very-low resolution in the sure/flow relationship is main-
vertical scale and work with single-layer tained at the fine- and coarse-
models is accepted. Because pressure scale levels and ensuring that
responses were considered at a fine scale, it breakthrough characteristics of
is possible to illustrate the consequences of all reservoir fluids are preserved
using a low-resolution model. In practice, at both scales. Because single-
any fine-scale model developed with geo- phase flow is considered, only the
logical and geophysical techniques must be first criterion is of immediate
coarsened to facilitate computations. The interest. The principal conse- Fig. 3—Influence of aggregation on the
authors used flow-based techniques. quence of aggregation is that the response at observation Well 2.
Fig. 3 shows the influence of aggregation interconnectedness within the
in the vertical scale at Well 2. The top curve reservoir is modified. If observations are tical variance is introduced (e.g., fractal
is the fine-scale response, and the other restricted to the consequences of aggregation models), the geological issues underlying
curves are responses for the four coarse in the vertical direction, then the degree of the use of such models are rarely considered.
models. The 5- and 17-layer responses interconnectedness in the horizontal direc- For example, it may be worthwhile to
assume uniform aggregation; in the other tions is increased and that in the vertical explore whether fractal behavior implies a
two models, the aggregation is nonuniform direction is reduced. Because the arrival of hierarchy of scales. The authors’ experience
and formed on the basis of velocity profiles. pressure fronts is dictated by local values of with systems of complex geology suggests
As the system is coarsened, the responses porosity and permeability, the local values of that it is no longer enough to infer estimates
deviate from the fine-scale response. The permeability must be reduced, assuming of permeability or porosity, and it is essential
magnitude of deviation from the fine-scale that the geological description is accurate. to discern factors that influence these esti-
response for a specific system depends on The scaleup process, by definition, pre- mates. The issue of the interconnectedness
the distance between the pulsing and serves pore volume. However, any change to of the reservoir or the interconnectedness of
responding wells. The degree of coarsening porosity affects the value of the resource sig- a network of wells to the reservoir must be a
of the porosity cube to analyze tests will nificantly. One measure of the appropriate- matter of concern in well testing.
affect the results of the analysis. In systems ness of the coarse model to analyze pressure
with complex geology, many layers are tests is that changes in porosity be negligibly Conclusions
needed. In this case, as many as 30 layers are small. Therefore, the degree of coarsening is A significantly large number of layers will be
required to preserve the internal intercon- of concern, and consideration of this aspect required to adequately evaluate spatial and
nectedness of the reservoir. Therefore, the requires considerable care. This observation petrophysical relationships by interference
permitted degree of coarsening appears to also applies to situations in which a distance or pulse tests in reservoirs with complex
be much less than that which is possible in variable must be estimated, such as the dis- geology. Aggregation and scaleup schemes
conventional-model studies. Simply using tance to a fault or the width of a channel. to evaluate pressure tests in complex reser-
scaleup criteria, based on steady-state pro- It is possible to evaluate properties with- voirs need to ensure that the connectivity
cedures, may be inadequate to capture out including porosity as a regression vari- within the reservoir is not compromised. A
responses of interest in transient behavior. able. If this option were pursued, adjust- scale that requires minimal adjustments to
ments to vertical permeability would be sig- porosity provides for an appropriate meas-
Low-Resolution Models. It is possible to nificant—reductions of approximately three ure to test the adequacy of a coarse-scale
use the results in a slightly different way. It orders of magnitude are needed. Fluid flow model. A strategy to evaluate pressure
may be assumed that the fine-scale results in such a model would be very different, responses with 3D geological models in
are the “truth” case and evaluate the conse- and differences in breakthrough times reservoirs with high permeability contrasts
quences of working at low resolution. A five- should be significant. This result points to and highly complex paths was proposed.
layer model was used, and its properties the need to choose the level of aggregation Models that use a very low resolution in the
were adjusted to match those of the porosity carefully because it modifies the geological vertical scale to match interference-test
cube. In this exercise, responses at all wells setting significantly. responses would, in all probability, result in
were analyzed simultaneously by standard a significant underestimate of formation
inversion techniques. It was observed in this Discussion flow capacity and storativity. JPT
case that it is best to adjust the permeability It is claimed that well testing plays a central
(in the three directions) and the porosity in role in reservoir-description methods. For a limited time, the full-length paper
each cell by multiplying them by a constant However, the assumption of homogeneity, is available free to SPE members at
value during the inversion process. The con- which implies a particular scale, is invariably www.spe.org/jpt. The paper has not
stant for each variable to be adjusted, of invoked in most analyses. Even in situations been peer reviewed.
course, is different. For simplicity, a time in which some complexity in terms of statis-

60 FEBRUARY 2004
81039.qxd 1/13/04 12:56 PM Page 61

W e l l Te s t i n g

Analysis of Well-Test Responses in

Gas/Condensate Reservoirs
Single-phase analysis based on pseudo- gas/condensate reservoir behavior. How-
pressure has been used extensively in the ever, important differences exist between This article, written by Technology Editor
oil industry to analyze well-test data the resultant fluid distributions; oil satura- Dennis Denney, contains highlights of
from gas/condensate reservoirs. However, tion increases from zero in the condensate paper SPE 81039, “A Novel
the presence of a two-fluid system, which case, while oil saturation decreases from Methodology for the Analysis of Well-
forms when the pressure drops below the 100% in conventional core measurements. Test Responses in Gas/Condensate
dewpoint, requires advanced techniques Also, it has been demonstrated that steady- Reservoirs,” by K. Barrios (now with
to derive accurate skin values. This quan- state relative permeability determined with PDVSA), G. Stewart, SPE, (also with
tity contains mechanical, non-Darcy, and gas/condensate fluids is sensitive to flow Edinburgh Petroleum Services), and D.R.
liquid-dropout components. Many reme- rate, and both gas and condensate relative Davies, SPE, Heriot-Watt U., Edinburgh,
dial well treatments designed on such permeabilities increase as the flow velocity prepared for the 2003 SPE Latin
incorrect analysis result in poor treat- increases. An understanding of the con- American and Caribbean Petroleum
ment success. densate-banking process and condensate Engineering Conference, Port-of-Spain,
mobility is essential in predicting well Trinidad, 27–30 April.
Introduction behavior and the performance of gas/con-
Many high-pressure and -temperature densate reservoirs.
gas/condensate reservoirs are being devel- densate, a near-critical fluid, a volatile oil,
oped. These reservoirs display complex Gas/Condensate Reservoirs and a highly undersaturated oil. A clear
behavior because of a two-fluid system Reservoirs can be classified by the location oil/gas contact is not present; instead, a
after the pressure drops below the dew- of their initial reservoir pressure and tem- transition zone exists in which gas and oil
point. Liquid dropout caused by retrograde perature with respect to the two-phase properties cannot be distinguished. High-
condensation leads to a condensate bank gas/liquid region. Gas/condensate reser- pressure produced-gas injection was imple-
building up around the producing wells. voirs are distinguished by two characteris- mented to increase reserves and minimize
The growing condensate bank progressive- tics. First, a liquid phase can condense dur- rapid reservoir-pressure decline.
ly impedes gas flow into the well. The ing isothermal pressure depletion (retro- The 2,500-ft-thick Naricual and Cretaceo
effect of a condensate blockage region grade behavior). Second, this liquid revap- formations are the main producers. These
depends on several reservoir parameters, orizes with further pressure depletion. complex reservoirs are characterized by
including formation relative permeability A three-region model is used to charac- numerous faults with up to 1,000 ft of ver-
and the fluid properties. terize gas/condensate flow. The first region tical displacement. Pressure-analysis data
Fluid-property and relative permeability is the outer part of the reservoir (only the suggest that some of these faults are sealing,
data are essential to diagnose and analyze a gas phase is present because the pressure is causing reservoir compartmentalization.
well test accurately. However, many chal- higher than the dewpoint). The second
lenges are involved in sampling gas/con- region has a pressure less than the dew- Pressure Analysis
densate reservoirs. Accurate simulation of point, but the condensate that forms This study established a systematic method
the liquid-dropout process is based on an remains immobile because of its low satu- for analyzing production-well-test pressure
equation of state (EOS). Without proper ration. The third region is near the well- data in gas/condensate reservoirs. A three-
pressure/volume/temperature (PVT) data, bore with both gas and condensate flow- phase 1D model was used. Different syn-
the EOS can be very uncertain and the ing. The existence of the third zone thetic features were simulated to illustrate
results inaccurate. reduces well productivity because of the the buildup pressure response in gas/con-
Measuring relative permeability for a liquid dropout. This three-region model densate reservoirs. The full-length paper
gas/condensate reservoir is difficult. forms the basis for many gas/condensate details the analysis procedure.
Initially, the retrograde liquid forms in the flow studies. During reservoir depletion, The steady-state method does not accu-
vicinity of producing wells, where the the condensate saturation increases from rately predict the oil-saturation values in
greatest rate of pressure reduction exists, zero and is present in the second and the region in which oil is condensing from
causing a gas-phase relative permeability third regions. the gas and accumulating until it reaches
reduction. It is important that experimen- the critical oil saturation. Applying the line-
tal procedures used to generate the relative Santa Barbara Reservoir fitting method in the two-phase analysis, a
permeability data in the laboratory be rep- The Santa Barbara field is in the North zero-slope line can be fitted to the deriva-
resentative of the condensing process that Monagas Area of the Eastern Venezuela tive data points in the middle-time region.
occurs in such a reservoir. Often, conven- basin. The fluid column is complex with an This procedure allows the gas relative per-
tional gas/oil drainage relative permeability extreme vertical-compositional variation. It meability value at the connate-water satura-
measurements are used to represent shows, from top to bottom, a low-yield con- tion and the mechanical skin factor to be

FEBRUARY 2004 61
81039.qxd 1/13/04 12:56 PM Page 62

obtained, provided the pseudopressure sure of 9,100 psi was used, 104 psi greater from the Santa Barbara full-field-simulation
function has been estimated accurately. than the dewpoint pressure. The simula- model and were scaled manually using the
The skin factor estimated by the two- tion consisted of reproducing a well test in same endpoints as those established in the
phase analog would have an error of up to which the well was flowed for a consider- Santa Barbara input data file.
two units (i.e., the maximum error in the able period of time (drawdown), then The scaling process was used to be cer-
calculated skin value will be −2). This closed for 4 days for a pressure-restoration tain that the relative permeability curves
overcorrection will depend on the real period (buildup). used in the well model to generate the
mechanical-skin value. Simulations varying pseudopressure function were the same set
the mechanical skin from negative values PVT Modeling. Original compositions of relative permeability curves used in the
(wells with fractures) to positive values from two different condensate reservoirs simulation model. Six additional relative
(wells with formation damage) were per- were used as the condensate mixture. permeability curves were used to investi-
formed. A correlation between the skin val- These compositions made it possible to gate the effect on the extent of the conden-
ues obtained from the two-phase pseudo- incorporate all the features associated with sate bank formed near the wellbore region
pressure analysis and the mechanical skin the phase behavior relevant to the problem in a condensate reservoir after the bottom-
was derived. (well productivity, deliverability, prediction, hole flowing pressure decreased to less than
The skin factor estimated with the single- and well-test analysis). the dewpoint pressure.
phase analog may contain contributions
from the mechanical skin as well as the liq- Relative Permeability Model. A key para- Conclusions
uid-dropout effects. Therefore, taking the meter in determining well-deliverability A 1D radial compositional-simulation
difference between the two estimates and loss is the relative permeability. Therefore, model was used to simulate a gas/conden-
applying the skin correction yields a very it is very important that the experiments sate well-test response under multiphase
close approximation of the skin value used to measure relative permeability data conditions. A range of sensitivity studies
caused by the liquid-dropout effect. This consider the process that occurs in these illustrated the most important features
procedure can help discriminate between reservoirs. Conventional gas/oil drainage occurring in a gas/condensate reservoir
mechanical skin and liquid-dropout skin relative permeability data are derived from under depletion conditions. Thereafter, a
and enable correct decisions regarding steady-state- or unsteady-state-flow labora- method using steady-state two-phase
required remedial well treatments (e.g., tory experiments usually conducted to pseudopressure function was established to
stimulations or hydraulic fractures) neces- measure gas relative permeability in the obtain the most realistic results from a
sary to optimize well production. presence of oil or condensate. The typical gas/condensate well-test analysis. This
drainage relative permeability data do not method assumed that the available relative
Compositional Simulation apply to this situation. permeability and fluid-property data were
Compositional simulation is required to Condensate and gas relative permeability correct. The full-length paper details the
model reservoir production processes, such will increase with increasing velocity when following findings.
as depletion of volatile-oil and gas/conden- measuring relative permeability using con- • If a condensate bank formed in the
sate reservoirs, miscible flooding, and gas densing fluids in long cores. This phenom- near-wellbore area when the bottomhole
cycling. Such simulation models assume enon is called the “positive-coupling pressure was less than the dewpoint pres-
that reservoir-fluid properties depend on effect.” These results point out the need to sure, a radial composite model was
the reservoir temperature and pressure as use an appropriate experimental technique observed in the pressure derivative when
well as on the changing composition of the in which the phase distributions represent using single-phase pseudopressure analysis.
reservoir fluid during production. those in gas/condensate reservoirs. • If the reservoir pressure was greater
In gas/condensate reservoirs, the effect In the reservoir, gravity and capillary than the dewpoint pressure, then formation
of relative permeability changes and non- forces are the main controls of the conden- transmissibility, reservoir pressure, and
Darcy flow vary significantly across rela- sate mobility. However, close to the well, total skin factor can be estimated accurate-
tively short distances. Also, complex the viscous and capillary forces dominate ly by use of a single-phase pseudopressure
phase-behavior effects, such as condensa- flow. Within this regime, relative perme- analog.
tion and revaporization in the near-well- ability is affected most by fluid velocity, • The length of the condensate bank esti-
bore region, require the use of a properly which is contrary to the conventional non- mated with this single-phase analysis was
tuned EOS. Hence, accurate simulation of Darcy flow in which the permeability shown to equal a length corresponding to
near-wellbore phenomena requires the use decreases with increasing velocity in iner- an averaged oil-saturation value and not
of compositional-simulation models with a tial flow. Core tests were modeled in the equal to the total extension of length of the
very fine grid definition near the produc- near-wellbore flow regime by rotating the condensate bank.
tion well. core to minimize gravitational forces. • The two-phase pseudopressure func-
This study assumed the reservoir to be a In this study, a compositional-simulation tion, which incorporated the influence of
homogeneous porous medium of uniform model was used to evaluate the effect of changes in relative permeability and fluid
thickness in which gravity and capillary fluid velocity on the well-productivity loss properties, could be used to estimate both
pressure effects are negligible. Therefore, and condensate-banking extension on the absolute permeability and mechanical-
the porosity and permeability data for each basis of the correlation between relative skin factor. The latter has an error of
gridblock are assumed to be the same, with permeability and the capillary number. The −2 units.
values of 8.5% and 50 md, respectively. imbibition relative permeability curves • A correlation was derived to estimate
The original pressure of the Santa from a Santa Barbara condensate reservoir the mechanical-skin effect from the
Barbara reservoir was 11,900 psi. However, were used as a base case to perform all the pseudopressure-skin value for the cases in
to save computational time, an initial pres- sensitivity cases. These curves were taken which the mechanical skin was not zero.

62 FEBRUARY 2004
81039.qxd 1/13/04 12:56 PM Page 63

• A two-phase pseudopressure analysis

cannot be applied if there is no indication
Resources
of radial composite effect in the pressure-
derivative response.
• The length of the condensate bank Echometer Well Analyzer
depends on the composition of the reser-
voir fluid. It was found that a richer fluid
generates a longer condensate bank and a
greater skin effect.
• The revaporization effect during the
buildup period was minimal in the cases
studied.
• The velocity-dependent relative per-
meability effect can minimize, or in some
cases prevent, the accumulation of con-
Data acquisition and analysis are per-
densate in the near-wellbore area. This case formed using a powerful notebook com-
showed homogeneous reservoir-model puter housed with an accurate analog to
behavior in the pressure-derivative digital converter in a rugged briefcase.
response.
A cost advantage is realized by using one
• In tight reservoirs, a total gas-block
electronics package with the desired sen-
effect was observed because oil was the sors to obtain a complete well analysis.
only mobile phase present in the near-well-
bore area when the well was shut in. This • Liquid Level Instrument
effect was attributed to the likely high pres- • Dynamometer
• Pressure Transient Tester
sure drops occurring in the near-wellbore • Motor Power Sensor
region.
• If the reservoir pressure was less than Echometer Company
5001 Ditto Lane
the dewpoint pressure, the pressure-deriva- Wichita Falls, TX 76302
tive response showed a radial composite Phone: (940) 767-4334
model attributable to the variations of oil Fax: (940) 723-7507
saturations with pressure. Single-phase e-mail: [email protected]
www.echometer.com
analysis in these reservoirs yielded a
reduced permeability value as a result of the
oil saturation, and the inner-zone skin
value reflected the real mechanical skin.
• It was found that the steady-state two-
phase pseudopressure method was not
applicable to the analysis of a gas/conden-
sate well test when the reservoir pressure
was less than the dewpoint pressure. The
oil-saturation values predicted by this
method were much higher than the simu-
lated values.
• Horizontal wells minimized the accu-
mulation of condensate in the region of the
producing well as a result of the smaller
pressure-drop values observed during the
drawdown period compared with those for
vertical-well performance.
• Compositional simulation is an impor-
tant tool in gas/condensate well-testing
design to estimate how long the flow period
should be to detect a condensate-bank for-
mation in the well-test pressure-derivative
response (radial composite model). JPT

For a limited time, the full-length paper

is available free to SPE members at
www.spe.org/jpt. The paper has not
been peer reviewed.

FEBRUARY 2004 63

View publication stats