Académique Documents
Professionnel Documents
Culture Documents
A R T I C LE I N FO A B S T R A C T
Keywords: Petrophysical cutoff of net reservoir plays a very important role in reservoir characterization for the evaluation
Net reservoir cutoffs of hydrocarbons in place and the estimation of ultimate hydrocarbon recovery. There have been many ap-
Experiment design proaches to quantify cutoffs, yet each of these approaches yields a different reservoir model with some amount of
Proxy model uncertainties due to lack of enough data, insufficiency of knowledge and the heterogeneous nature of petroleum
Dynamic performance
reservoirs. Conventionally, net reservoirs cut-offs are evaluated by applying static petrophysical well-logs
Probability density function
without consideration of dynamic performance, which results in a crisp classification of reservoir or non-re-
servoir zones. This paper takes Q oilfield in Albert Basin of Uganda as an example, provides a new structured
quantitative approach which integrates all core, petrophysical interpretation, modular dynamics test data, re-
servoir fluid data and development wells pattern together to determine a reasonable net reservoir cutoffs. Three
dependent variables including oil initially in place (OIIP), total oil production (FOPT) and total water production
(FWPT) were selected to test the sensitivity with two independent variables, i.e. volume of clay (Vclay) and
porosity. Then experiment design and response surface method were used in constructing the proxy models that
are related to the dependent variable. After running 5000 realizations, probability density function (PDF) was
utilized to locate P50 value on the cumulative probability curve. Finally, the cutoffs of Vclay and porosity were
determined by the arithmetic mean of corresponding P50 value. The case study clearly illustrates how all
available data from a reservoir should be integrated for appropriate determination of the net reservoir cutoffs.
1. Introduction (Masoudi et al., 2012c), shale-gas (Worthington and Majid, 2014) and
fractured reservoir (Kolganov and Kovaleva, 2012).
Net reservoir, a key parameter in reservoir evaluation, has sufficient If the collected data is enough to cover a broad range of parameter
reservoir quality and useful capability to store fluids and allow them to variation, even thin beds reservoir may obtain an accurate cutoff
flow (Worthington, 2010; Hadi, 2017). The principal part of identifi- (Carpenter, 2014). Nevertheless, Q oilfield in Albert Basin in Uganda is in
cation of net reservoir is to determine its cutoffs. However, although the the early stage of the oilfield development (Xu et al., 2017). Cutoff de-
concept of cutoff has been continuously used since the 1950s, there is termination is difficult because of lack of drilled wells and other supported
no universal agreement on its definition and quantification methods yet data. In addition, although the reservoir cutoffs were generated and ap-
(Worthington, 2008). In the most commonly used procedure, it is plied during the petrophysical evaluation, their impacts will be more ob-
usually expressed in term of a log-derived shale volume fraction, Vclay, vious in reservoir engineering. The conventional cut-off-based methods for
being less than a Vclay cutoff and the fractional porosity being greater determining net pays are often based on static data, however Drill Stem
than or equal to a porosity cutoff through a Vclay-porosity cross plot Test and permeability data start to be used in recent studies (Worthington,
(Worthington and Cosentino, 2005). Recently, some new methods in- 2010; Masoudi et al., 2011, 2012a; b; c; 2014a; b). That indicates not using
cluding diffusivity equation (Masoudi et al., 2011), Bayesian (Masoudi dynamic data might increase the uncertainty of predicting reservoir pro-
et al., 2012a), fuzzy classifier fusion (Masoudi et al., 2012b),Demp- duction (Galley, 2016; Testamanti and Rezaee, 2017). This paper takes Q
ster–Shafer theory (Masoudi et al., 2014a), artificial neural network oilfield in Albert Basin of Uganda as an example, providing a new struc-
(Masoudi et al., 2014b) and NMR T2 (Testamanti and Rezaee, 2017) tured quantitative approach that integrates all core, petrophysical inter-
have been adopted to determine net zone. Meanwhile, the study context pretation, MDT test data, reservoir fluid data and development wells
has been expended from terrestrial clastic reservoirs to carbonate pattern for determination of net reservoir cutoffs.
∗
Corresponding author.
E-mail address: weixucug@qq.com (W. Xu).
https://doi.org/10.1016/j.jafrearsci.2018.05.007
Received 18 September 2017; Accepted 21 May 2018
Available online 22 May 2018
1464-343X/ © 2018 Elsevier Ltd. All rights reserved.
W. Xu et al. Journal of African Earth Sciences 145 (2018) 261–266
3.1. Dataset
3.2. Approach
262
W. Xu et al. Journal of African Earth Sciences 145 (2018) 261–266
0.60
discrete parameter was generally defined by probability mass function
Test oil
Sample oil
(PMF) (Moeinikia and Alizadeh, 2012; Arinkoola and Ogbe, 2015). In
Vclay: 0.48 Test water
this study, OIIP, FOPT and FWPT were distributed continuously, so the
0.50
Tight PDF curve was used to obtain their probability distribution respectively.
Through Monte Carlo (Mun, 2006; Bailian et al., 2017) simulation, the
No data collected realizations were used to generate PDF. In general, the value at P50 is
0.40
believed to be reasonable, so the corresponding Vclay and porosity at
Vclay (Frac)
Vclay: 0.35 P50 were the realistic cutoff we want to determine (Arinkoola and
0.30 Ogbe, 2015).
No data collected
Por: 0.125
0.10
MDT data in Fig. 2 show that the cutoff can be located anywhere
within the no-data area (highlighted in light blue). In the scope of this
0.00
area, Vclay value ranges from 0.35 to 0.48 and Porosity changes from
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.085 to 0.125. Each four values of Vclay and porosity were chose
Porosity (Frac) among this interval to make a four-lever experiment design (Table 1),
thus 24 = 16 NTG cases which cover all the possible situations were
Fig. 2. Porosity and Vclay cross plot for cutoff determination, dataset of MDT in
finally simulated to test the variation of dependent variable caused by
Q oilfield.
different cutoffs (Table 2).
263
W. Xu et al. Journal of African Earth Sciences 145 (2018) 261–266
Database
Data Analysis Vclay-porosity plotting of MDT points
Table 1 Table 3
The selected Vclay and porosity value for experiment design. The OIIP, FOPT and FWPT experimental results with Vclay and porosity.
Parameter Value 1 Value 2 Value 3 Value 4 Running Case Vclay Porosity OIIP MMbbls FOPT MMbbls FWPT MMbbls
Vclay 0.480 0.450 0.400 0.350 Case 1 0.480 0.085 527 151.1 661.7
Porosity 0.085 0.100 0.110 0.125 Case 2 0.480 0.100 527 151.1 661.7
Case 3 0.480 0.110 527 151.1 661.6
Case 4 0.480 0.125 525 150.0 654.0
Case 5 0.450 0.085 508 144.2 625.7
Case 6 0.450 0.100 508 144.2 625.5
Table 2 Case 7 0.450 0.110 508 144.2 625.2
The four-lever experiment design result by Vclay and porosity. Case 8 0.450 0.125 506 140.5 598.9
Case 9 0.400 0.085 478 132.3 578.2
NTG Case VClay Porosity
Case 10 0.400 0.100 478 132.3 578.2
Case 1 0.480 0.085 Case 11 0.400 0.110 478 132.3 577.8
Case 2 0.480 0.100 Case 12 0.400 0.125 476 131.3 572.4
Case 3 0.480 0.110 Case 13 0.350 0.085 443 121.5 531.4
Case 4 0.480 0.125 Case 14 0.350 0.100 443 121.5 531.4
Case 5 0.450 0.085 Case 15 0.350 0.110 443 121.5 531.4
Case 6 0.450 0.100 Case 16 0.350 0.125 442 121.0 529.9
Case 7 0.450 0.110
Case 8 0.450 0.125 Note: The above data have been processed technically.
Case 9 0.400 0.085
Case 10 0.400 0.100 4.4. Probability density function
Case 11 0.400 0.110
Case 12 0.400 0.125
Case 13 0.350 0.085 The above three formulae were respectively used as response sur-
Case 14 0.350 0.100 face proxy to calculate OIIP, FOPT and FWPT by running 5000 reali-
Case 15 0.350 0.110 zation through Monte Carlo stochastic algorithm. Then, PDF including
Case 16 0.350 0.125 probability density distribution and cumulative probability curve were
utilized to find the most reasonable results (P50) (Fig. 6). Cumulative
probability curve of OIIP showed that P50 value is 486 MMbbls and the
corresponding Vclay and porosity cutoff is 0.414 and 0.106 respec-
FWPT = −5639.68-239.967*cos(acos(x)) + 3656.692*cos(acos(y))
tively. Meanwhile, the P50 value of FOPT cumulative probability curve
+ 885.076*cos(2*acos(x)) −2510.55*cos(acos(x))
is 135.7 MMbbls and the corresponding Vclay and porosity cutoff is
*cos(acos(y))−6784.77*cos(2*acos(y)) (3)
0.418 and 0.106 respectively. FWPT cumulative probability curve also
264
W. Xu et al. Journal of African Earth Sciences 145 (2018) 261–266
0.60
Test oil
Sample oil
0.50 Vclay: 0.48 Test water
Tight
Vclay: 0.414
0.40
Vclay (Frac)
Vclay: 0.35
0.30
Porosity: 0.107
Por: 0.085
Por: 0.125
0.20
0.10
0.00
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Porosity (Frac)
Fig. 7. The final determined net reservoir cutoffs projected on porosity and
Fig. 5. Three-dimensional response surface plots of OIIP (a), FOPT (b), FWPT Vclay cross plot.
(c) as a function of Vclay and porosity with corresponding residual plot (d) and
(e).
proposed a mobility-based method which incorporates economic-pro-
440 460 480 500 520 120 124 128 132 136 140 144 148 152 550 600 650 duction rate in the quantification of cutoffs. The same point on the
a OIIP 99.5 b FOPT c FWPT
99.5 99.5
application of dynamic data implies that comprehensive utilization of
Cumulative Probability
Cumulative Probability
Cumulative Probability
98 98
98
90
70
90
70
90 static as well as dynamic information will significantly improve the
70
50
30
50
30 50
accuracy of cutoff determination.
30
10
2
10
2 10
Distinct reservoir parameters should be used to determine cutoffs in
0.5
300
0.5
350
5
formation evaluation for different environments. Compared to clastic
400
250 300
250
and carbonate environment, natural fracture, total organic carbon
200 300
Counts
Counts
Counts
150
200
200
maturity and free gas saturation could be utilized to define cutoff for
150
100
100
100
net shales gas reservoir besides volume of shales and porosity
50
50
0 0 0
(Worthington and Majid, 2014). Furthermore, the same assessment of
440 460 480 500 520 120 124 128 132 136 140 144 148 152 550 600 650
the ANN-based method has showed precision of 86% and 91% in
Fig. 6. Probability density function and cumulative probability curve for OIIP classifying pay zones in carbonate and clastic reservoirs respectively
(a), FOPT (b), FWPT (c), and the blue line locates at P50 value. (For inter- (Masoudi et al., 2014a). This is due to the distinct characteristic of pore
pretation of the references to colour in this figure legend, the reader is referred throat structure and heterogeneity of different reservoir formation in
to the web version of this article.)
different environment. In essence, cutoff determination is to find the
most appropriate parameters to effectively identify the net interval that
gives the P50 value at 591 MMbbls and the corresponding Vclay and is in ability to store hydrocarbon and allow them to flow.
porosity cutoff is respectively 0.419 and 0.109. The cutoffs derived In general, the approach proposed in this paper is preferred because
from 3 different dependent variables’ cumulative probability curve are of several advantages. First, it takes into account dynamic performance
very close. at reservoir conditions. Second, experiment design avoids time-con-
fusing simulation in the models. Third, response surface function pro-
4.5. Cutoff determination vides a very fast test for static and dynamic result versus cutoffs. Fourth,
it is simple, straightforward and practically rationalized with a model
OIIP as a static parameter was integrated with the dynamic para- grid.
meters FOPT and FWPT to determine a very close cutoff through
probability density function. Therefore, an arithmetic mean value of the
three Vclay cutoffs was adopted as the final Vclay cutoff for Q oil field, 5. Conclusion
and so did porosity cutoff. Fig. 7 showed the locations of final cutoffs on
cross plot. The final Vclay cutoff is 0.414 with a loss rate at 3% and the This paper provides an integrated quantitative approach to de-
ultimate porosity cutoff is 0.107 with the corresponding loss rate at 5%, termine reservoir cutoff by taking static and dynamic parameter into
both of the loss rates of reservoir are in a reasonable scope. This in- account, which overcomes deviation that may be caused by lack of core
tegrated quantitative approach has formed a positive effective circle of test and MDT test data. The method presented in this study is in-
the static and dynamic data interaction. dependent of the core test and MDT test data, besides it considers re-
servoir connectivity to reflect the rationality of cutoff through geolo-
4.6. Further discussion gical and recoverable reserves and water production meanwhile.
According to the analysis results, the reasonable reservoir cutoffs in this
Cutoff determination has obvious impact on static parameters like oil field are Vclay at 0.416 and porosity at 0.107. It is worth noticing
net pay thickness and hydrocarbon volume, consequently dynamic re- that Vclay may have greater influence on reservoir compared to the
sults alters as cutoff changes. The procedure in this paper carries out an porosity. Moreover, it is more sensitive for cutoff impacting on thin
analysis of FOPT and FWPT to detect the fluid flowing in the porous interbedding reservoir sands. Consequently, integrated static and dy-
media. The wider range of FOPT and FWPT than OIIP indicates the namic data analysis should be fully utilized in the process of reservoir
impacts on dynamic response by reservoir connectivity. Hadi, 2017 cutoff determination.
265
W. Xu et al. Journal of African Earth Sciences 145 (2018) 261–266
Acknowledgments Masoudi, P., Tokhmechi, B., Jafari, A.M., 2012a. Application of Bayesian in determining
productive zones by well log data in oil wells. J. Petrol. Sci. Eng. 47–54 94–95.
Masoudi, P., Tokhmechi, B., Jafari, M.A., Moshiri, B., 2012b. Application of fuzzy clas-
We acknowledge the CNOOC Uganda Limited for their support and sifier fusion in determining productive zones in oil wells. Energy Explor. Exploit. 30
permission to present this work. We also acknowledge other staff from (3), 403–416.
CNOOC Research Institute, specifically Hu Guangyi, Sun Lichun, Zhang Masoudi, P., Tokhmechi, B., Bashari, A., 2012c. Identifying productive zones of the
Sarvak formation by integrating outputs of different classification methods. J.
Yingchun and Gao Boyu, who made contributions to this study. The Geophys. Eng. 9, 282–290. http://dx.doi.org/10.1088/1742-2132/9/3/282.
conscientious reviews of the two reviewers and editorial guidance of Dr. Masoudi, P., Arbab, B., Mohammadrezaei, H., 2014a. Net pay determination by Dempster
Mapeo are gratefully acknowledged. rule of combination: case study on Iranian off shoreoil fields. J. Petrol. Sci. Eng. 123,
78–83.
Masoudi, P., Arbab, B., Mohammadrezaei, H., 2014b. Net pay determination by artificial
References neural network: case study on Iranian off shore oil fields. J. Petrol. Sci. Eng. 123,
72–77.
Moeinikia, F., Alizadeh, N., 2012. Experimental Design in reservoir simulation: an in-
Aanyu, K., Koehn, D., 2011. Influence of pre-existing fabrics on fault kinematics and rift
tegrated solution for uncertainty analysis, a case study. J. Petrol. Exploration Prod.
geometry of interacting segments: analogue models based on the Albertine Rift
Technol. 2 (2), 75–83.
(Uganda), Western Branch-East African Rift System. J. Afr. Earth Sci. 59, 168–184.
Moras, B., Rey, S., Vilarem, G., Pontalier, P.Y., 2017. Pressurized water extraction of
Arinkoola, A.O., Ogbe, D.O., 2015. Examination of experimental designs and response
isoflavones by experimental design from soybean flour and soybean protein isolate.
surface methods for uncertainty analysis of production forecast: a Niger delta case
Food Chem. 214, 9–15.
study. J. Petrol. Eng. 3, 1–16.
Mun, J., 2006. Modeling Risk : Applying Monte Carlo Simulation, Real Options Analysis,
Bailian, Chen, Jincong, He, Xian-Huan, Wen, Wen, Chen, Albert, C., Reynolds, 2017.
Forecasting, and Optimization Techniques. John Wiley & Sons, Inc, Hoboken, New
Uncertainty quantification and value of information assessment using proxies and
Jersey p.74.
Markov chain Monte Carlo method for a pilot project. J. Petrol. Sci. Eng. 157,
Oguz, Y., Cochrane, C., Koncar, V., Mordon, S.R., 2016. Doehlert experimental design
328–339.
applied to optimization of light emitting textile structures. Opt. Fiber Technol. 30,
Carpenter, C., 2014. Thin Beds and Cutoffs: Improved Assessment and Modeling of Net
38–47.
Porous Volume. Society of Petroleum Engineershttp://dx.doi.org/10.2118/0814-
Roberts, E.M., Stevens, N.J., O'Connor, P.M., Dirks, P.H.G.M., Gottfried, M.D., Clyde,
0071-JPT.
W.C., Armstrong, R.A., Kemp, A.I.S., Hemming, S., 2012. Initiation of the western
Cheong, Y.P., Gupta, R., 2005. Experimental Design and Analysis Methods for Assessing
branch of the East African Rift coeval with the eastern branch. Nat. Geosci. 5,
Volumetric Uncertainties. Society of Petroleum Engineershttp://dx.doi.org/10.2118/
289–294.
80537-PA.
Rohmer, J., Bouc, O., 2010. A response surface methodology to address uncertainties in
Daghbandan, A., Chalik, S.M., 2015. The prediction of the performance of an oil reservoir
cap rock failure assessment for CO2 geological storage in deep aquifers. Int. J.
by proxy model: a case study. Int. J. Chemoinf. Chem. Eng. (IJCCE) 4 (2), 46–58.
Greenhouse Gas Control 4, 198–208.
Galley, S.V., 2016. Pay Cutoff Definition Based on Dynamic Reservoir Parameters. Society
Testamanti, M.N., Rezaee, R., 2017. Determination of NMR T2 cut-off for clay bound
of Petrophysicists and Well-Log Analysts.
water in shales: a case study of Carynginia Formation, Perth Basin, Western Australia.
Hadi, Saboorian-Jooybari, 2017. A Structured Mobility-based Methodology for
J. Petrol. Sci. Eng. 149, 497–503.
Quantification of Net-pay Cutoff in Petroleum Reservoirs. Society of Petroleum
Venkataraman, R., 2000. Application of the Method of Experimental Design to Quantify
Engineershttp://dx.doi.org/10.2118/183643-PA.
Uncertainty in Production Profiles. Society of Petroleum Engineershttp://dx.doi.org/
Karner, G.D., Byamungu, B.R., Ebinger, C.J., Kampunzu, A.B., Mukasa, R.K., Nyakaanad,
10.2118/59422-MS.
J., Rubondod, E.N.T., Upcottc, N.M., 2000. Distribution of crustal extension and re-
Worthington, P.F., 2008. The application of cutoffs in integrated reservoir studies. SPE
gional basin architecture of the Albertine rift system, East Africa. Mar. Petrol. Geol.
Reservoir Eval. Eng. 11 (6), 968–975.
17, 1131–1150.
Worthington, P.F., 2010. Net pay-what is it? What does it do? How do we quantify it?
Karp, Tobias, Scholz, C.A., McGlue, M.M., 2012. Structure and stratigraphy of the lake
How do we use it? SPE Reservoir Eval. Eng. 13 (5), 812–822.
Albert rift, east africa: observations from seismic reflection and gravity data. In:
Worthington, P.F., Cosentino, L., 2005. The role of cut-offs in integrated reservoir studies.
Baganz, O.W., Bartov, Y., Bohacs, K., Nummedal, D. (Eds.), Lacustrine Sandstone
SPE Reservoir Eval. Eng. 8 (4), 276–290.
Reservoirs and Hydrocarbon Systems. AAPG Memoir. vol. 95. pp. 299–318.
Worthington, P.F., Majid, A.A., 2014. The Use of Net-Pay Concepts in the exploitation of
Kolganov, V., Kovaleva, G.A., 2012. Approaches for Estimating Carbonate Fractured
shale-gas deposits. J. Petrol. Sci. Eng. 120, 78–85.
Reservoir Property Cutoffs. Society of Petroleum Engineershttp://dx.doi.org/10.
Xu, Wei, Fang, Lei, Zhang, Xinye, Wang, Pengfei, Liu, Jun, Yang, Xipu, 2017. Object-
2118/162071-MS.
based 3D geomodel with multiple constraints for early Pliocene fan delta in the south
Mason, J.C., Hsndscomb, D.C., 2003. Chebyshev Polynomials. CRC Press Company, pp.
of Lake Albert Basin, Uganda. J. Afr. Earth Sci. 125, 1–10.
155–157 chapter 6.
Zhang, X., Scholz, C.A., 2015. Turbidite systems of lacustrine rift basins: examples from
Masoudi, P., Tokhmechi, B., Zahedi, A., Jafari, M.A., 2011. Developing a method for
the lake kivu and lake Albert rifts, east africa. Sediment. Geol. 325, 177–191.
identification of net zones using log data and diffusivity equation. J. Min. Environ. 2
(1), 53–60.
266