IPTC 13546

Multi-Layer Testing: Theory and Practice

Asfandiar Ansari and Yahia Mahmoud, RasGas Company Limited (RasGas)

Copyright 2009, International Petroleum Technology Conference

This paper was prepared for presentation at the International Petroleum Technology Conference held in Doha, Qatar, 79 December 2009.
This paper was selected for presentation by an IPTC Programme 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 International Petroleum Technology Conference and are subject to correction by the author(s). The material, as presented, does not necessarily
reflect any position of the International Petroleum Technology Conference, its officers, or members. Papers presented at IPTC are subject to publication review by Sponsor Society Committees
of IPTC. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the International Petroleum Technology Conference 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, IPTC, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A., fax +1-972-952-9435.

Abstract
Pressure transient analysis (PTA) for a multi-layer well, producing commingled from more than two non-communicating
reservoirs, has always been a challenge. To address this challenge, innovative data acquisition techniques and novel multilayer analysis procedures were developed in the 1980s. However, very limited practical application of these theoretical
procedures has been reported in the literature.
This paper attempts to fill in this important gap. The paper summarizes the practical application of different theoretical
multilayer testing (MLT) data acquisition and procedures to commingled wells in the RasGas Company Limited (RasGas)
operated area of Qatars North Field, the largest offshore non-associated gas reservoir.
Comparison of the different MLT analysis methods as applied in RasGas suggests that the rate-normalized (RN) PTA method,
despite its strong theoretical basis, is the least reliable in this field application. On the other hand, in spite of their theoretical
limitations, results from the Selective Inflow Performance (SIP) method obtained during this work are more reliable and
consistent with other observed data (cross-flow, well productivity, etc).
Based on the above work, a way forward in MLT analysis has been developed, relying on the SIP testing approach to provide
not only information on changes in individual layer pressures, but also reliable time-lapse changes in individual layer
productivity, which is very important for monitoring individual reservoirs in commingled wells.
Background
The North Field, located offshore in Qatari waters, represents the largest accumulation of non-associated gas in the world.
RasGas Company Limited is one of the major operators of the North Field in Qatar, producing since 1999.

Figure1: North Field location in Qatari waters

IPTC 13546

Currently, the majority of its producing wells are commingled producers from four different horizons. The four different
accumulations vary in size and in other reservoir properties that include porosity, permeability-thickness (kh), fluid properties,
pressure, skin factor, etc. It is, therefore, very important to know these individual reservoir parameters, and any changes in
them over time to closely monitor changes in the well, the reservoir and the field productivity.
Estimating individual layer parameters from a commingled well test was attempted in the industry in the past. However, none
of the reported MLT analysis methods has been applied to a gas-condensate field. Moreover, all these MLT methods are
limited in ability to estimate a unique set of layer parameters from a commingled well test.
A similar conclusion was noted when different methods were applied to RasGas commingled producers. However, it was
noted that in comparison to more complicated analysis techniques, a simple SIP method yields answers within the accuracy
and consistency required to monitor individual layers over time. This was based on application of different MLT methods to a
number of RasGas commingled producers, some of which were tested more than once over time. The different MLT methods
used, their results, and the successes with the SIP method, are discussed in following sections.
Conventional MLT Analysis
A conventional PTA approach is commonly used to analyze well tests for wells producing from a single reservoir in RasGas.
The same approach was extended to multi-layer commingled producers. However, it was soon noted that the results obtained
from such an approach were not unique. Two analyses, using the same well test data, but different initial parameter values, can
result in substantially different layer parameter results after regression. One such example is shown in Figure 2, where the
pressure and pressure derivative data are plotted (as points) with a pair of model data sets (lines) on a log-log scale. The two
plots show excellent match using two different models; one model is based on very different layer parameters for the two
layers while the other is based on similar layer parameters for the two layers.

Model 1

Model 2

Contrasting
layer skins,kh
& porosity

Similar
Different layer
skins, kh &
porosity

Log-Log plot: (m(p)-m(p@dt=0)).Q/[qn-qn-1] and derivative [psi2/cp] vs dt [hr]

Figure 2: Non-uniqueness in 2-layer model results apparent despite excellent match

This is due to the fact that the conventional method uses pressure transient data recorded above the top layer, representing a
commingled response of all contributing layers. An attempt to match this commingled response by regressing on multi-layer
parameters may give a match, but the match is not unique. More than one set of layer parameters yield a similar match to the
commingled pressure response. Even by simplifying the 4-layer RasGas commingled wells to a 2-layer system, the
conventional method continues to give a non-unique answer; six unknowns (at least three per layer: skin, kh, pressure) against
a single set of commingled pressures and its derivative pose a mathematically indeterminate problem.
Some improvement in this non-uniqueness was achieved when layer contributions from Production Logging Tool (PLT) data
were used as a matching data set in addition to the commingled pressure response. However, the results are not sufficiently
consistent to identify, with confidence, any time-lapse changes in layer parameters.
Non-conventional MLT Analysis
This MLT technique was developed in the 1980s 1, 2, 3. It is based on the premise that if pressure transients of individual layers
are separated, layer parameters can be estimated. The MLT method proposes to facilitate this separation by recording pressure
and rate above each commingled layer, one by one, as shown in Figure 3.

IPTC 13546

Q0Q1

Q1Q2

Tool on top of L3
gives L3 kh and
skin

1E-3

0.01

0.1

Q2Q3
Tool on top of L2
gives L(2+3) kh
and skin.
With L3 known
L2 can be
estimated

10

1E-3

0.01

0.1

Tool on top of L1
gives L(1+2+3)
kh and skin.
With L2 & L3
known L1 can be
estimated

10

1E-3

0.01

0.1

10

Log-Log plot: (m(p)-m(p@dt=0)).Q/[qn-qn-1] and derivative [psi2/cp] vs dt [hr]

Figure 3: Rate-Normalized Pressure Transient Analysis (RN-PTA) testing sequence

Starting with the bottommost layer, where the tool is stationed above the bottommost layer perforations, pressure and rate
transients are recorded simultaneously using the PLT tool at the well rate Q1. Normalizing the pressure transient using the
corresponding rate transient, and applying the conventional PTA to the normalized pressure response, layer parameters for the
bottommost layer are estimated. Placing the tool above the next layer and changing the well rate to Q2, the layer parameters
for the two layers combined are obtained using the conventional PTA on the normalized data. Since the bottommost layer
parameters are already known from the previous analysis, layer parameters of the next higher layer are estimated. All
individual layer parameters are estimated by continuing with this stepwise data acquisition and analysis technique at the next
well rate of Q3.
This MLT technique was applied to RasGas commingled producers, and referred to as RN-PTA. Despite its excellent
theoretical basis, RN-PTA posed some practical problems. Firstly, the rate normalization was very noisy, as shown by two real
examples in Figure 4, where the noise is primarily due to poor normalization.

1 E+7

1 E+6

1 E+5
1 E-5

1 E-4

1 E-3

0 .0 1

0 .1

10

0 .1

1 0

Radial flow or any multi-layer response

very poorly defined to get reliable layer

1 E+7

1 E+6

1 E+5
1 E-5

1 E-4

1 E-3

0 .0 1

Log-Log plot: (m (p)-m (p@dt=0)).Q/[qn-qn-1] and derivative [ps i2/cp] vs dt [hr]

Figure 4: Noisy normalization on pressure and its derivative in above two examples

Therefore, the RN-PTA could not be applied with confidence to get a reliable set of layer parameters and reservoir model.
Secondly, the error in the parameters estimated for the bottom layers have a cumulative effect that results in a significant error
for the topmost layer. It was also noted by applying RN-PTA to wells with contrasting layer contributions that layer
parameters are significantly affected by layer contributions. Well permeabilities seemed to be associated with the biggest
contributing layers, and no layer parameters could be estimated if a relatively low contributing layer was next to a high
contributor. This is shown in the example in Figure 5 where the results for the bottommost layer is the same as the combined
results for the bottom two layers because the bottommost layer has over 80% contribution in the combined analysis.

IPTC 13546

At top of L4

Log-Log plot: (m (p)-m (p@dt=0)).Q/[qn-qn-1] and derivative [ps i2/cp] vs dt [hr]

At top of L3

Log-Log plot: (m (p)-m (p@dt=0)).Q/[qn-qn-1] and derivative [ps i2/cp] vs dt [hr]

Figure 5: Layer dominance of L4 over L3 failed to identify reservoir properties for L3 in the above example

This last observation points to an important limitation of RN-PTA, which will remain even if the other two issues (noisy
normalization and error accumulation) are resolved.
Selective Inflow Performance Method
The literature on the SIP method appeared in the 1980s. However, it has gained renewed attention fairly recently 4, 5. Like the
RN-PTA method mentioned above, it also uses the PLT-based data, but from a conventional PLT run as opposed to the
specialized runs required for the RN-PTA method. The SIP method is based on using rate and flowing pressure obtained for
each layer from the PLT passes across the layer during conventional PLT runs, or from stationary PLT data commonly used to
estimate layer contributions without correcting for tool movements. Since PLT passes are usually made at more than two
flowing well rates during a PLT campaign on a well, three to four sets of rate-pressure data are acquired for each layer.
Plotting these data sets for each layer gives a conventional Inflow Performance Relationship (IPR) plot for each layer as shown
in Figure 6. The data sets shown in Figure 6 are from four different PLT surveys: first one following the initial shut-in (Q=0).
The other three surveys were recorded following 12-hr flow periods (Q1, Q2, and Q3). Conventionally, the SIP method is used
to estimate layer pressures by extrapolating the IPR curves for each layer to zero rates.

Figure 6: Selective Inflow Performance (SIP) testing sequence

IPTC 13546

The SIP method was applied to selected RasGas commingled wells as only few wells have conventional PLT surveys recorded
at multiple well rates. Most of the resulting IPR plots fall on the conventional IPR trend, as shown in Figure 7. The layer
pressures estimated from the IPR extrapolation matched well with corresponding layer pressures measured by open hole
wireline pressure tests. The shapes of the resulting layer-by-layer IPR plots give relative productivity of individual layers of
the commingled well.

Figure 7: SIP-based Inflow Performance Relationship (IPR) plots for different layers

Unlike the noisy plots from the RN-PTA method that result in a wide error bar on layer parameters, the SIP method gives
fairly well defined curves and as a result, a reliable set of layer pressures. Based on this observation, the SIP method was
extended beyond its conventional use of estimating layer pressures. By fitting an IPR equation, such as the one from the Darcy
equation, layer skin can also be estimated for a given set of layer kh values. This was successfully done on a number of
RasGas wells, and layer skin factors were estimated using layer kh values estimated by distributing commingled well total kh
according to the open-hole log-based layer kh distribution.
It is important to note that the layer IPR plots are based on short flow durations. Low permeability layers may still be in the
transient state. Hence, the layer pressure estimated from the SIP may not represent the true reservoir pressure of that layer.
Moreover, if the layer IPR is still in the transient state, the shape of the layer IPR may vary with changing flow durations.
Despite this theoretical limitation on the SIP method, its results can still be compared from one test to another as long as the
same rate sequence and rate durations are maintained for each survey. The resulting changes in the layer pressure and the layer
skin will represent the true change in the layer parameters over time. This has been successfully demonstrated in the time-lapse
SIP analysis discussed in the following section.
Time-Lapse SIP
Many RasGas wells have multiple PLTs obtained during their production history. However, only a few of these wells are
commingled wells, and even fewer have multi-rate PLT surveys. Time-lapse SIP was applied to one of these wells (Figure 8).
The analysis showed that the initial layer IPR plots (Figure 8a) suggest pressure differential between layers. This is in line with
the MDT pressures for these layers, and the observed cross-flow between the layers. The SIP analysis for the next PLT run
(Figure 8b) showed that all layer IPR plots have shifted down over time suggesting some pressure depletion for each layer, as
expected. Moreover, the shift was more pronounced in one layer than the other. This resulted in a reduced pressure differential
between the two layers. This is in line with the expected larger pressure depletion in a layer with smaller volume of gas in
place, and the reduced cross-flow observed from the PLT surveys. By overlaying the layer IPRs on top of each other (Figure
8c), it was noted that the shape of only one layer IPR has changed over time. This suggests that only this layer has a skin
change over time.

IPTC 13546

Figure 8: SIP-based time-lapse analysis identifying layer performance changes

This apparent skin change was quantified by estimating skin factors for each one of the layers from the two SIP data sets. By
using the open-hole log-based kh for each layer, skin factors are estimated for the two layers by matching with the SIP trends
from the two tests over time. The skin estimates confirmed (Figure 9) that one layer has an increase in skin whereas the other
layer does not show any change in skin.

Figure 9: SIP-based time-lapse analysis estimates skin changes over time

IPTC 13546

Conclusions & Recommendations

1. Multi-layer Testing (MLT) is the best way to monitor layer parameters in a commingled well, as conventional pressure
transient analysis (PTA) cannot uniquely estimate all layer parameters.
2. MLT techniques are more desirable since the analyses use flowing data. Conventional PTA is carried out on the final buildup, which is quite often affected by the cross-flow in the wellbore during shut-in.
3. Out of different MLT techniques available, theoretically most sound method is the rate normalized pressure transient
analysis (RN-PTA), as it accounts for rate changes during the test. However, this approach is quite involved in terms of data
acquisition, and even more complex in terms of performing the analysis. Consequently, the results are affected by the
quality of both, data acquisition and analysis.
4. Another MLT method, called the Selective Inflow Performance (SIP) method, is fairly straightforward with respect to the
data it uses and the analysis methodology. Consequently, the results are of a consistent quality that can be used for timelapse comparison of layer parameters.
5. The time-lapse SIP approach has been proven successful in estimating changes in layer parameters for RasGas wells.
6. Based on this, the standard MLT method adopted to monitor RasGas commingled producers is SIP. All current and future
well testing campaigns are designed to obtain multi-rate conventional PLT surveys which will provide the necessary data to
perform time-lapse SIP analysis.
Acknowledgements
The authors extend their thanks to RasGas Reservoir Engineering Group for providing the opportunity, support, and technical
peer review to the work presented in this paper. The authors are also grateful to Petroleum Engineering Department and
Subsurface Group of RasGas for approving the work. Special gratitude is due RasGas shareholders, Qatar Petroleum and
ExxonMobil, for permission to present this work.
Figure 1 was obtained from: Challenges of Monitoring the largest Gas Condensate Field, presented at the offshore Middle East
conference held in Doha, Qatar 28-30 October 2008.
Figure 2 to Figure 9: All generated by the authors in RasGas.
References
1. Kuchuk, F. J., Karakas, M., and Ayestaran, L.; Well Testing and analysis Techniques for Layered Reservoirs, SPE Formation
Evaluation (August 1986).
2. Ehlig-Economides, C. A. and Joseph, J.A.: A new Test for Determination of Individual Layer Properties in a Multilayered Reservoir,
SPE 14167 presented at 1985 SPE ATCE, LasVegas.
3. Kuchuk, F. J., Shah, P. C., and Ayestaran, L.: Application of Multilayer Testing and Analysis: A Field Case, SPE 15419 presented at
1986 SPE ATCE, New Orleans.
4. Schnorr, D. R.: More Answers from Production Logging Than just Flow Profiles,: Transactions of the 1996 SPWLA 37th ALS, New
Orleans.
5. Ayestaran, L.: Multi-Layer Testing, A Method to Obtain Formation Properties in completed Wells, SPE presentation in Balikpapan,
June 2007.