PETROLEUM SOCIETY PAPER 2004-097

CANADIAN INSTITUTE OF MINING, METALLURGY & PETROLEUM

Downloaded from http://onepetro.org/PETSOCCIPC/proceedings-pdf/04CIPC/All-04CIPC/PETSOC-2004-097/1855532/petsoc-2004-097.pdf/1 by Chulalongkorn University user on 19 July 2022

Quantification of Uncertainty in Production
Forecast Using Experimental Design, Case
Study: Tiguentourine and Taouratine Fields
A.M. ATBI, A. AISSAOUI
SONATRACH, Algeria.

This paper is to be presented at the Petroleum Society’s 5th Canadian International Petroleum Conference (55th Annual Technical
Meeting), Calgary, Alberta, Canada, June 8 – 10, 2004. Discussion of this paper is invited and may be presented at the meeting if
filed in writing with the technical program chairman prior to the conclusion of the meeting. This paper and any discussion filed will
be considered for publication in Petroleum Society journals. Publication rights are reserved. This is a pre-print and subject to
correction.

Abstract Including the development parameters in the design made

possible the estimate of uncertainty in reserves and production
Almost all the data used in reservoir simulation are subject profiles for different development scenarios. One major
to uncertainty, which may be quite large especially for an advantage of using experimental design technique is that during
immature reservoir, consequently, reserves cannot be estimated the progress of this study as a new well has been drilled, the
with accuracy for such reservoirs. Current stochastic probability distributions of some uncertain parameters have
procedures as the multiple realization approach require large been reviewed, using the fit of the reservoir simulation model,
number of simulation runs, such intensive computational effort an update of the production forecasts was possible almost
is sometimes prohibitive. The experimental design method has instantaneously and without any extra simulation run.
been used in this study to quantify uncertainty in production
forecast by running only a reasonable number of simulation
runs. Introduction
This study has been carried out as follow, first the uncertain
parameters have been identified, a screening design was Reserves estimates are generally highly uncertain at early
performed to restrain the analysis to the key parameters, then a stage of field development given the sparse geological data.
more elaborated set of sensitivity runs was designed and run, Considering the large initial capital investments associated
the design included the key uncertain parameters and the with most fields development, decisions should be made in full
development parameters like well spacing and abandonment awareness of geological risk. Many stochastic methods have
pressure, the results of these sensitivity runs have been then been used to quantify uncertainty, those methods generally
used to substitute the reservoir simulation model by a nonlinear require large number of expensive fluid flow simulation runs,
function. After verifying the quality of the fit, this nonlinear most of them may lead to significant time and manpower
function was used to carry out Monte Carlo simulation. This commitment. The problem can be exacerbated if different
procedure was illustrated through Tiguentourine gas development options have to be considered since reserves
condensate field case. should be estimated for each development option separately.

1
 Experimental design was proposed by Damsleth et al1 to get But as a result of the fractionalization the design do not give
around of this problem by running only a limited a set of well full resolution; that is, there are certain interaction effects that
designed sensitivity runs and fit the results of the simulation are confounded with (identical to) other effects. The resolution
model by a mathematical function that is then used as a depends on the number of runs in a fractional factorial design,
substitute of the real reservoir numerical model to carry out The possible resolutions are:
Monte Carlo simulation. Resolution III designs: Where main effects are confounded
The uncertain development design parameters can be with 2 factor interactions. Resolution IV designs: Where two
incorporated into the design, so the fit of the reservoir model factor interactions are confounded with each other. With both
can be used for the analysis of the different possible resolution III and IV designs, the only model that can be

Downloaded from http://onepetro.org/PETSOCCIPC/proceedings-pdf/04CIPC/All-04CIPC/PETSOC-2004-097/1855532/petsoc-2004-097.pdf/1 by Chulalongkorn University user on 19 July 2022

development strategies. selected is a linear model. There is also Resolution V designs,
This work reports the application of experimental design where main effects and all two-factor interactions are clear of
technique to a real field case where different development each other’s (unconfounded). With resolution V designs,
scenarios have been under study. interaction model can be generated ( quadratic model). This part
of the study using experimental design was carried out in two
stages the first stage consisted in carrying out a screening
Procedure design. The screening design is the first stage of an
investigation where the goal is just to identify the important
Experimental design is how to conduct and plan experiments factors that can cause a substantial changes in the response
in order to extract the maximum amount of information from when it varies. Generally a 2 levels fractional factorial design is
the collected data in the presence of noise, in the fewest number used, after identifying the factors that have a substantial effect
of experimental runs. The basic idea is to vary all relevant on the response(s). A new design is generated including only
factors simultaneously over a set of planned experiments and the important variables in terms of their impacts. The objective
then connect the results by means of a mathematical model. of this new design is to carry out Response Surface Modeling.
This model is then used for the interpretation, prediction, and D-Optimal design was chosen for both screening and
optimization. In our case the mathematical model is used response surface modeling. A Windows-based software was
instead of the real simulation model to carry out Monte Carlo used for this purpose, providing design of experiments,
Simulation. analysis, and optimization in one package. It allows a
One of the most important parts of the experimental design is polynomial fitting with Multi Linear or Partial Least Square
the factorial design, it is used in experiments involving several Regression, it allows refining and reviewing of the models as
variables where it is necessary to investigate the joint effects of well.
the factors on a response variables. The commonly used The variables considered in the design have been the
procedure in determining the main effects of more than one reservoir static, dynamic and development uncertain data2.
variable on a certain response is to design a comparative Uncertainty analysis should be carried out for one
experiment to determine the effects of the first variable while development scenario at the time, including the development
the others are at their most likely value, then a second data in the design can allow the quantification of uncertainty for
comparative experiment to determine the effect of the second different possible development options. The responses of the
variable, and so on. This is one way sensitivity analysis. This experiments are the output results of the simulation runs.
design is very inefficient from an experimental design point of Uncertainty analysis was mainly applied to reserve
view, because the experiment number is large and the results estimates, it can be extended to production profiles3 providing
provide no information about the interaction of the factors. some assumptions.
Factorial design is one in which all levels of a given variable By breaking up the production profile into specific
are combined with all levels of every other variables in the characteristics (periods), the set of simulation runs used to
experiment, depending on the number of levels there can be generate reserve distribution can also be used to generate
different types of factorial designs. distribution for each one of the production profile
2K Experiments: If the complete set of the variables has two characteristics.
levels coded –1 and +1 minimum and maximum respectively, Any production profile can be generally characterized by
the experiment is called 2K factorial design, where K is the three periods and eight key coefficients Figure1. The buildup,
number of variables. This kind of experiment is very useful to plateau and decline periods, each of these production period
illustrate the main and the interactions effects. This design is can be represented by the following simple equations:
generally used in the screening process, or to fit the response by
a first order model. Build up: Q = ((Qp-Qi)/tb).t + Qi 0 ≤ t ≤ tb …(1)
3K Experiments: In this kind of design there is K variables at
three levels coded –1, 0 and +1, respectively minimum, most Plateau period: Q = Qp tb ≤ t ≤ tp…(2)
likely and maximum values. This design plays an important role
in more complicated design problems, because in addition to the Decline period: Q = Qp.(1+ Di b t)-1/b t ≥ tp …..(3)
main effect and interaction this design can be used to determine
the curvature effects. A disadvantage of the factorial design is Abandonment: Q = Qa t = ta …..(4)
that the number of experiment combinations increase rapidly as
the number of variables increase. Assuming standard hyperbolic decline with parameters:
If these data have to be fitted by a polynomial model, many b: Hyperbolic decline exponent.
of these experiments are unnecessary. One way to overcome Di: Initial decline rate.
this problem is to consider only a subset of all possible Because in our case the available number of appraisal wells
combinations, or a so called fractional factorial design. drilled in the prospect insure to reach the plateau at the
Many kind of fractional factorial designs are available. The beginning of the production, it was possible to fully characterize
design used in this study is the D-optimal design. D-Optimal the production profile by two periods and two parameters,
designs are computer generated designs that maximize (1) the plateau period characterized by the plateau length and
information from a limited number of experiments.

2
(2) the decline period, fitted by a harmonic decline, - Help constrain the structural form of the good reservoir
characterized by an initial decline rate constant Di. intervals and enable to map the base and top of the syn-glacial
package.

Tiguentourine gas condensate field case Uncertain variables:

study Tiguentourine/La Reculé structure is relatively well
appraised compared to the Taouratine area where only few
Tiguentourine field is situated in the South East of the wells have been drilled.

Downloaded from http://onepetro.org/PETSOCCIPC/proceedings-pdf/04CIPC/All-04CIPC/PETSOC-2004-097/1855532/petsoc-2004-097.pdf/1 by Chulalongkorn University user on 19 July 2022

Algerian Sahara desert near the Libyan border in the Illizi basin, The extension and potential of the two Glacial Paleo Valleys
Figure2. in this zone have not been proven with certainty, the acquired
This field was discovered in the 50’s and comprises three 3D seismic helped constrain their shapes only to certain degree
independent reservoirs, the Carboniferous, the Devonian and the given the limits of seismic resolution.
Cambro Ordovician. The extensions and original volumes in place of the two
The Carboniferous and the Devonian are two oil reservoirs. Paleo valleys in the western area of the prospect have been
The Cambro Ordovician, subject of this study, is a gas considered as uncertain parameters and have been attributed
condensate reservoir. three different values, corresponding to three different
The two first Horizons have been the only one produced. The confidence levels.
Cambro Ordovician reservoir was not exploited since the 50’s Using the results of the core, log and well test data gathered,
because of no available gas sailing contract at that time. probability distributions have been also attributed to each one of
In the late 90’s a production sharing agreement was signed the static and dynamic parameters: Porosity, Net to Gross
between two operators for a joint development of the Block Thickness, Water Saturation, Horizontal and Vertical
234, a prospect area including three structures: Permeability and the Critical Condensate Saturation, Wells
Tiguentourine/La Reculée and Taouratine, Figure3. Mechanical Skin, Non-Darcy flow factor and finally the
In order to better constrain the potential of the field and transmissibility of the main western fault separating
optimize development strategy, an appraisal work has been Tiguentourine/La Reculé from Taouratine area.
performed in this prospect, consisting of a 3D seismic survey,
drilling of new wells and a series of well tests. The production
is expected to begin in 2005.
Development Design Parameters:
While this work was under progress no final development
Tiguentourine field Structure : strategy have been adopted yet for Tiguentourine field, different
scenarios were still under consideration.
The Tiguentourine Ordovician structure is a relatively high To be able to quantify uncertainty for the different probable
relief faulted anticline trending in NNE-SSW direction and development scenarios and since uncertainty estimates in
covers 233 km2. reserves should be performed for one development strategy at
The structure is comprised of two peaks: the Tiguentourine the time, the uncertain development parameters have been
peak in the South and La Reculée peak in the North. The incorporated in the Experimental Design.
structural dip is toward the North. The uncertain development parameters considered are:
The structure is bounded on the East and West by two major Well Spacing: three well spacing have been considered 2000,
faults. The dominant fault orientation is Northeast Southwest 1000 and 500 acres spacing.
limiting the structure at the East, the Western fault has the same Minimum Tubing Head Pressure (Abandonment Pressure):
orientation and separates Tiguentourine/La Reculée structure three abandonment pressures have been studied 650, 250 and
from the Taouratine structure Figure4. 120 psi.
And finally Plateau Rate: three plateau rates have been
Reservoir Geology : studied 700, 1000 and 1500 MMscf/D.
The Tiguentourine Cambro Ordovician reservoir was
interpreted as a succession of pre-, syn- and post- Glacial Response variables :
sedimentary packages. A full field 3D simulation model was used to carry out the
They have been gathered in two main reservoir layers or sensitivity study, the responses used from each simulation run
Megasequences. have been the cumulative gas produced after 20 years and gas
1st Megasequence : The lowermost division, it contains all production plateau length, other responses have been used but
the pre glacial sediments. It is interpreted as fluviatile with for brevity considerations not reported here.
some marine influence deposits.
2nd Megasequence : It comprises all the syn and post glacial
sediments. It is interpreted as succession of ice proximal
Screening Design:
tractional and high density turbidites deposits. Mainly found the first stage was to carry out a screening design in order to
along the axis of the two Glacial Paleo valleys, Found one in discriminate between the variables in terms of their importance
the North and the other in the South, called respectively North in affecting the simulation output results (Response variables).
and South Paleo Valley Figure5. This stage is important, its purpose is to restrain the study to
The 2nd Megasequence represent the best play in term of the significant variables in order to limit the number of response
original volumes in place and petrophysical properties . surface modeling sensitivity runs.
the 3D seismic was acquired in this prospect to : The screening design chosen was a D-optimal design with
- Precise structural maps of the top and the base of the resolution IV at 2 levels, in this type of designs the response can
reservoir. only be fitted by a first order polynomial, the screening design
- Locate precisely the faults network. matrix is given in Table1.
The Plot 6 & 7 represent the coefficients of the polynomial
scaled and centered. The scaling makes the coefficients

3
comparable. The size of the coefficient represents the change in
the response when a factor varies from 0 to 1, in coded units, Where:
while the other factors are kept at their averages. The coefficient Gp: cumulative gas produced after 20 years in this case
is significant (different from the noise), when the confidence tp: is the plateau length.
interval does not cross zero. qp: the plateau rate (MMscf/d).
We noticed that the water saturation Sw, the fault Di: the initial decline constant rate.
conductivity, the Non Darcie flow coefficient Dfactor, the t: in this case is 20 years.
absolute permeability K, the critical condensate saturation Scc, The P90 gas production profiles for the three scenarios are
the skin, the vertical anisotropy Kv/Kh had an insignificant given in the Figure11.

Downloaded from http://onepetro.org/PETSOCCIPC/proceedings-pdf/04CIPC/All-04CIPC/PETSOC-2004-097/1855532/petsoc-2004-097.pdf/1 by Chulalongkorn University user on 19 July 2022

impact on the cumulative gas produced and the plateau length.
all of them were ignored for the Response Surface Modeling Update of the production forecasts:
Design stage.
Before finishing this study, a new well have been drilled in
the western part of the prospect to confirm the extension of the
Response Surface Modeling Design: North Paleo Valley, the analysis of its data suggested that the
The second part consisted in designing a set of sensitivity North Paleo Valley deteriorates in this area.
runs in order to carry out response surface modeling, the design In the light of these new information, the parameter
matrix used for this purpose is given in Table2. “Extension of the North Paleo Valley “ (North-PV) has been re-
A D-optimal design was chosen with resolution V at 5 levels, adjusted.
three center points were included, 0, 0,333 and –0,333. the The last response surface models were used combined with
responses were fitted with quadratic polynomials including Monte Carlo Simulation to update the production forecasts. The
curvature and interaction terms. same input distributions were used to carry out Monte Carlo
The software used allowed the refining and refitting of the simulation, only the distribution of the variable “North Paleo
model, only the significant parameters were kept. Valley” (North PV) was changed (updated).
The refitted coefficient scaled and centered obtained for the The updated unrisked estimates of the Cumulative gas
cumulative gas produced and plateau length are given in the produced and production profile, for the third development
Table3 & 4. scenario, are given in the Figures 12 & 13.
An excellent quality of the fit was obtained for both the Conclusions
cumulative gas produced and plateau length, as indicated by the
values of R2adj, 0.984 and 0.987, for both responses very close The experimental design technique has been applied to
to 1. The very good quality of the fit can also be noticed on the assess uncertainty in production forecasts of Tiguentourine field
plot of the predicted versus observed values, Figure 8 & 9. . From the analysis of the last results the following conclusions
have been drawn :
Monte Carlo Simulation Results: · Uncertainty estimates can be carried out using less
simulation runs by applying Experimental Design.
Monte Carlo Simulation was carried out using a risk analysis · the Screening design gives access to the ranking of
software. The input variables have been sampled 10000 times. parameters in terms of their impacts in affecting the output
The probability distributions of cumulative gas produced, for simulation results, attention can be then focused to appropriate
the three development scenarios are given in the Figure 10. contingency plans to reduce the impact of the key uncertain
From economic standpoint the rate at which hydrocarbons parameters.
can be produced is as important as the ultimate recovery. The · By including the development design parameters in
production profile can be divided into different periods having the Response Surface Model, it was possible to estimate
each one a specific characteristic. The set of sensitivity runs reserves for different development scenarios.
used to generate response surface for reserves can be used to · The response surface model offered the advantage of
generate response surface for each one of the production profile being available to update reserve estimates promptly and
characteristics. without excessive computational effort as new information
In our case the production profile have been divided into two became available.
periods: · At the origin of all uncertainty evaluation there is a
1) a plateau period characterized by the plateau length, and need to assess probabilities in a subjective way. By using the
2) a decline period fitted by a harmonic decline characterized last procedure, it is possible to test different input probability
by an initial decline constant Di. distributions in Monte Carlo Simulation with almost no
Having the cumulative gas produced Gp (t) and the plateau computational effort.
length tp , for a given confidence level, we can get the initial
decline constant rate Di of the same confidence level.
For example: the constant Di P90 is the one that when
combined with a plateau length P90, restitutes a cumulative
Acknowledgement
production P90. The authors would like to thank Sonatrach for its support in
The initial decline constants Di can be calculated solving the the preparation and for permission to publish this paper.
following equation for Di

t
NOMENCLATURE
t tp

∫ ∫ ∫ Di = Initial decline rate.

q

Gp(t) = q dt = p dt + qp/(1+Di (t-tp)) dt………..(5) Gp = Cumulative gas produced.

0 0 tp
tp = Plateau length.
Qp = Plateau rate (MMscf/d).
( )) ..................(6)
qp
G p ( t ) = q p .t p +
Di
.2.303 .log 1+ D i t − t p ( t
Q
=
=
Time.
Rate.
North PV = North Paleovalley.

4
South PV = South Paleovalley.
Phi = Porosity.
NTG = Net to Gross Thickness.
Sw = Water Saturation.
Kh = Horizontal Permeability.
Kv = Vertical Permeability.
Scc = Critical Condensate Saturation.
S = Mechanical Skin.
Dfactor = Non-Darcy flow factor.

Downloaded from http://onepetro.org/PETSOCCIPC/proceedings-pdf/04CIPC/All-04CIPC/PETSOC-2004-097/1855532/petsoc-2004-097.pdf/1 by Chulalongkorn University user on 19 July 2022

FTrans = Fault transmissibility.
WSpacing = Well spacing.
APressure = Abandonment pressure.

REFERENCES
1. Damsleth, E., A. Hage, and R.Volden: “ Maximum Information
at Minimum Cost: A North Sea Field Development Study With
an Experimental Design,” Paper SPE 23139 Presented at the
Offshore Europe Conference held in Aberdeen, 3-8 September
1991.
2. Ovreberg, O., Damsleth, E. and Haldorsen, H.H. : ”
Putting Error-Bars on Reservoir Engineering Forecasts,”
Paper SPE 20512 Presented at the 1990 SPE Annual Technical
Conference and Exhibition, New Orleans.
3. P.J. Smith, D.J. Hendry, and A.R.Crowther: “The
Quantification and Management of Uncertainty in Reserves,”
paper SPE 26056 presented at the Western Regional Meeting
Anchorage, 26-28 May 1993.

5
Exp No Phi Sw NTG Ftrans Dfactor KH Scc S Kv North PV South PV WSpacing Qp Apressure
1 1 -1 1 1 1 -1 -1 -1 -1 -1 -1 1 -1 -1
2 -1 1 -1 -1 -1 1 -1 -1 -1 -1 -1 1 1 1
3 1 1 -1 1 -1 1 -1 -1 -1 -1 -1 -1 -1 -1
4 -1 -1 -1 1 1 1 1 -1 -1 -1 -1 1 1 -1
5 1 -1 1 -1 -1 1 -1 1 -1 -1 1 1 1 -1
6 1 -1 -1 1 -1 -1 1 1 -1 -1 1 -1 1 1
7 -1 1 1 1 -1 -1 1 1 -1 -1 -1 -1 1 -1
8 -1 1 1 -1 1 -1 1 1 -1 -1 1 1 -1 1
9 -1 1 1 1 -1 -1 -1 -1 1 -1 -1 -1 -1 1
10 1 1 -1 -1 1 -1 -1 -1 1 -1 1 -1 1 -1
11 -1 -1 -1 -1 -1 -1 1 -1 1 -1 1 1 -1 -1
12 1 -1 1 -1 1 1 1 -1 1 -1 1 -1 1 1

Downloaded from http://onepetro.org/PETSOCCIPC/proceedings-pdf/04CIPC/All-04CIPC/PETSOC-2004-097/1855532/petsoc-2004-097.pdf/1 by Chulalongkorn University user on 19 July 2022

13 1 -1 -1 -1 1 -1 -1 1 1 -1 -1 1 1 1
14 1 1 1 1 1 1 -1 1 1 -1 1 1 -1 -1
15 -1 -1 -1 1 -1 1 1 1 1 -1 -1 -1 -1 -1
16 1 1 1 1 -1 -1 -1 -1 -1 1 1 1 1 1
17 -1 -1 -1 -1 1 -1 -1 -1 -1 1 1 -1 -1 1
18 1 -1 -1 -1 -1 -1 1 -1 -1 1 -1 -1 1 -1
19 -1 -1 1 1 -1 1 1 -1 -1 1 1 -1 1 -1
20 -1 -1 -1 1 -1 -1 -1 1 -1 1 -1 1 1 -1
21 1 -1 1 1 1 1 -1 1 -1 1 -1 -1 -1 1
22 1 1 -1 -1 1 1 1 1 -1 1 1 1 -1 -1
23 1 -1 -1 1 -1 1 -1 -1 1 1 1 1 -1 1
24 -1 -1 1 -1 1 1 -1 -1 1 1 -1 1 1 -1
25 1 1 -1 1 1 -1 1 -1 1 1 -1 -1 -1 -1
26 -1 1 1 -1 -1 -1 -1 1 1 1 1 -1 -1 -1
27 1 1 1 -1 -1 -1 1 1 1 1 -1 1 -1 1
28 1 -1 1 -1 1 -1 1 1 1 1 1 1 1 -1
29 -1 1 -1 1 1 1 1 1 1 1 1 -1 1 1

Table 1: Screening Design Matrix.

Exp n° Phi NTG WSpacing Apressure Qp North PV South PV

1 -1 1 -1 -1 -1 -1 -1
2 -1 -1 1 -1 -1 -1 -1
3 1 1 1 -1 -1 -1 -1
4 1 1 -1 1 -1 -1 -1
5 -1 1 1 1 -1 -1 -1
6 -1 1 1 -1 1 -1 -1
7 1 -1 -1 -1 -1 1 -1
8 -1 1 1 -1 -1 1 -1
9 1 1 -1 1 -1 1 -1
10 1 -1 1 1 -1 1 -1
11 1 1 -1 -1 1 1 -1
12 1 1 1 1 1 1 -1
13 1 -1 -1 -1 -1 -1 1
14 1 -1 1 1 -1 -1 1
15 1 1 -1 -1 1 -1 1
16 1 -1 1 -1 1 -1 1
17 1 1 1 1 1 -1 1
18 1 1 -1 -1 -1 1 1
19 1 1 1 1 -1 1 1
20 1 -1 -1 1 1 1 1
21 -1 -1 -1 -1 1 -1 0,333
22 -1 -1 -1 1 -1 -1 0,333
23 -1 -1 -1 1 -1 -0,333 -1
24 -1 -1 -1 1 0,333 1 -1
25 -1 -1 -1 -0,333 -1 1 1
26 -1 -1 -1 0,333 1 -1 -1
27 -1 -1 1 -1 1 0,333 -1
28 -1 -1 1 -1 -0,333 1 1
29 -1 -1 1 1 1 -0,31 1
30 -1 -1 -0,333 1 1 -1 1
31 -1 -1 0,333 1 1 1 -1
32 -1 1 -1 1 1 1 -0,333
33 -1 1 -1 1 0,333 -1 1
34 -1 1 -1 0,333 -1 -1 1
35 -1 1 1 -0,333 1 1 1
36 -1 1 0,333 -1 1 1 1
37 -1 1 -0,333 1 -1 1 1
38 -1 -0,333 -1 -1 1 1 1
39 -1 0,333 -1 1 1 -1 -1
40 -1 0,333 1 -1 -1 -1 1
41 -1 -0,333 1 1 -1 1 1
42 1 -1 -1 -1 -0,333 -1 -1
43 1 -1 -1 1 1 -0,333 -1
44 1 -1 1 -1 -1 0,333 1
45 1 -1 1 -1 1 1 -0,333
46 1 -1 1 1 0,333 -1 -1
47 1 0 -0,333 0 0 0 0
48 0 1 -0,333 0 0 0 0
49 0 0 -0,333 0 0 0 -1
50 0 0 -0,333 0 0 0 0

Table 2: Response Surface Modeling Design Matrix

6
 Cum Gp (20years) Coeff, SC Std, Err, Conf int(+/-)
Constant 6,88393 0,0777407 0,16011
Phi 0,544558 0,0333825 0,0687526
NTG 0,36529 0,0331433 0,0682599
WSpacing -0,617144 0,0347597 0,0715889
Apressure -0,372236 0,034586 0,0712312
Qp 0,964206 0,0339501 0,0699216
North PV 0,805326 0,0344999 0,071054
South PV 0,365549 0,0343684 0,0707831

Downloaded from http://onepetro.org/PETSOCCIPC/proceedings-pdf/04CIPC/All-04CIPC/PETSOC-2004-097/1855532/petsoc-2004-097.pdf/1 by Chulalongkorn University user on 19 July 2022

Qp*Qp -0,494485 0,090945 0,187305
North PV ^ 2 -0,651427 0,0905195 0,186428
Phi*WSpacing -0,0417602 0,0349823 0,0720474
Phi*Apressure 0,0698683 0,0346624 0,0713886
Phi*Qp 0,332011 0,0339113 0,0698416
Phi*North PV -0,148256 0,0359863 0,0741152
Phi*South PV -0,0613698 0,0350723 0,0722327
NTG*Apressure -0,10439 0,0354661 0,0730439
NTG*Qp 0,179685 0,0357043 0,0735344
NTG*North PV -0,119487 0,0350317 0,0721492
NTG*South PV -0,0647714 0,0350145 0,0721137
WSpacing*Apressure 0,0635609 0,036604 0,0753875
WSpacing*Qp -0,420409 0,0356203 0,0733615
Apressure*Qp -0,291591 0,0352552 0,0726094
Apressure*North PV 0,121795 0,0354191 0,072947
Apressure*South PV 0,103441 0,0355256 0,0731664
Qp*North PV 0,568667 0,0361532 0,0744591
Qp*South PV 0,196107 0,0358164 0,0737652
North PV*South PV -0,0925707 0,0364533 0,075077
Table 3: cumulative gas produced response surface model coefficients (scaled and centered).

Plateau Length Ceff. SC Std. Err. Conf. Int(+/-)

Constant 126,428 6,88098 14,2704
Phi 40,837 2,81083 5,82934
NTG 26,3095 2,7265 5,65445
Wspacing -60,2347 2,85514 5,92122
Apressure -23,9758 2,87959 5,97192
Qp -108,483 2,79018 5,7865
North PV 62,4574 2,87891 5,97052
South PV 26,4127 2,84492 5,90004
Wspacing ^ 2 25,4432 8,68455 18,0107
Qp*Qp 31,4799 7,66338 15,893
Phi*NTG 6,75875 2,74963 5,70241
Phi*Wspacing -139597 2,85053 5,91166
Phi*Apressure -8,04552 2,85927 5,92979
Phi*Qp -25,2632 2,80146 5,8099
Phi*North PV 18,1468 2,94211 6,10159
Phi*South PV 9,77953 2,8452 5,90061
NTG*Apressure -6,57921 2,94681 6,11133
NTG*Qp -14,638 2,96916 6,1577
NTG*North PV 10,9491 2,93095 6,07845
NTG*South PV 11,031 2,89191 5,99749
WSpacing*Apressure 6,76581 3,01157 6,24565
WSpacing*Qp 17,551 2,9175 6,05056
WSpacing*North PV -22,8966 2,9229 6,06175
WSpacing*South PV -9,0012 2,91109 6,03726
Apressure*Qp 14,8774 2,90164 6,01766
Apressure*North PV -11,3285 2,90028 6,01484
Apressure*South PV -10,8727 2,91855 6,05273
Qp*North PV -32,4671 2,992 6,20505
Qp*South PV -13,1886 2,94477 6,10711
North PV*South PV 5,24748 2,99689 6,21519
Table 4: plateau length response surface model coefficients (scaled and centered).

7
 Q

Downloaded from http://onepetro.org/PETSOCCIPC/proceedings-pdf/04CIPC/All-04CIPC/PETSOC-2004-097/1855532/petsoc-2004-097.pdf/1 by Chulalongkorn University user on 19 July 2022

Qp
Qi
Qa
t
tb tp
Figure 1 : Characteristics of production profile.

Figure 2 : Tiguentourine field Location. Figure 3 : Contract Area.

Figure 4 : Tiguentourine Field structure. Figure 5 : Situation of the North & South
Paléo Valleys in the field.

8
 Downloaded from http://onepetro.org/PETSOCCIPC/proceedings-pdf/04CIPC/All-04CIPC/PETSOC-2004-097/1855532/petsoc-2004-097.pdf/1 by Chulalongkorn University user on 19 July 2022
APressure
FTrans

Wspacing

North PV

South PV
Phi

NTG

Dfactor
Sw

APressure
Scc
Kh

Qp

Wspacing
Kv

North PV

South PV
S

Dfactor
FTrans
Phi

NTG

Scc
Kh

Kv
Sw

Qp
Figure 6 : Cumulative Gas produced coefficients, Figure 7 : Plateau Length Coefficients, screening design.
screening design.

Figure 8 : Cumulative Gas Produced , observed versus Figure 9 : Plateau Length, Observed versus predicted values plot..
predicted values plot.

100% 1200

90%

1000
80%

70%
800
Gas rate(MMscf/d)

60%
Pobability(%)

50% Scenario1 600

Scenario1
Scenario2
Scenario2
Scenario3
40% Scenario3
400
30%

20% 200

10%

0
0% 0 5 10 15 20 25
2 3 4 5 6 7 8 9 10 ti ( )

Figure 10 : The three development scenarios Cumulative Figure 11 : The three development scenarios P90
Gas Produced, Cumulative probability chart. Production Profiles.

9
 100%

90%

80%

Downloaded from http://onepetro.org/PETSOCCIPC/proceedings-pdf/04CIPC/All-04CIPC/PETSOC-2004-097/1855532/petsoc-2004-097.pdf/1 by Chulalongkorn University user on 19 July 2022

70%

Probability(%) 60%

50%
Old Cum Gas
New Cum Gas
40%

30%

20%

10%

0%
3 4 5 6 7 8 9

Figure 12 : Old and Updated Cumulative probability chart of the

Cumulative Gas produced for the third development scenario.

1200

1000

800
Gas Rate (MMscf)

600
Old Production Profile
New Production Profile

400

200

0
0 50 100 150 200 Time(Month) 250 300

Figure 13 : Old and Updated P90 production profile for the

third development scenario.

10