Sie sind auf Seite 1von 36

Stratigraphic well correlations for 3-D static modeling of carbonate reservoirs

Jean R. F. Borgomano, Francois Fournier, Sophie Viseur, and Lex Rijkels

AUTHORS Jean R. F. Borgomano  Laboratoire de Geologie des Systemes et des Reservoirs Car ` bonates, case 67, Universite de Provence, 3, Place Victor Hugo, F-13331 Marseille Cedex 03, France; jean.borgomano@univ-provence.fr Jean Borgomano obtained a Ph.D. in carbonate geology in 1987 from the University of Provence in Marseilles, France. In 1988 2003, he worked at Shell as a senior carbonate geologist in various exploration and production Shell companies. He is currently a professor at the University of Provence and the director of the Geology of Carbonate Systems and Reservoirs Laboratory. His research focuses on the geological characterization and numerical modeling of carbonate reservoir architecture and properties. Francois Fournier  Laboratoire de Geo logie des Systemes et des Reservoirs Carbo` nates, case 67, Universite de Provence, 3, Place Victor Hugo, F-13331 Marseille Cedex 03, France; Francois.Fournier@univ-provence.fr Francois Fournier received his M.Sc. degrees from the Nancy School of Mines, France, and from the Institut Francais du Petrole and a Ph.D. in carbonate sedimentology from the University of Provence in Marseilles, France. After a short experience in oil companies as an exploration geologist in France and Angola, he joined the Geology of Carbonate Systems Laboratory, Marseilles, France, as a lecturer in 2005. His research focuses on the relationship between sedimentology, diagenesis, and seismic reflections in carbonate reservoirs. Sophie Viseur  Laboratoire de Geologie des Systemes et des Reservoirs Carbonates, ` case 67, Universite de Provence, 3, Place Victor Hugo, F-13331 Marseille Cedex 03, France; Sophie.Viseur@univ-provence.fr Sophie Viseur is a numerical geologist working as a researcher at the University of Provence. She received her Ph.D. from the Nancy School of Geology in 2001. Her primary interests are in geostatistics for channel simulations and their application to hydrocarbon exploration and production. In recent years, she has worked in developing methods for the integration of outcrop data and geological concepts into 3-D carbonate architecture models.

ABSTRACT The principles and purposes of stratigraphic well correlation in carbonate sedimentary systems are defined and discussed within the context of static reservoir modeling. The challenge of well correlations is to relate the heterogeneities measured at core and well scales to the spatial heterogeneities at reservoir and flow unit scales. The introduction of a priori knowledge in the process of stratigraphic well correlation is critical to support the stratigraphic rules and to establish a coherent geological and petrophysical concept. The links between well correlation and geostatistics are discussed with regard to the stationarity hypothesis and property trend analysis. We stress that wells are incomplete and biased samples of the geological reality, which is not dependent, unlike the dynamic reservoir behavior, on the well numbers, location, and spacing. Stratigraphic rules are applied as a function of the well spacing relative to the geological reality. A simple trigonometric method, combining angle of base profile, paleobathymetry, and well spacing, is introduced to check the validity of the well correlation in carbonate ramp-like systems. Two models, based respectively on outcrop and subsurface with seismic data, are discussed in detail to show the combined influence of the data set, sedimentary systems, and diagenetic transformations on stratigraphic well correlations.

INTRODUCTION This article discusses the principles of stratigraphic well correlations that form the foundation of most carbonate reservoir models used in hydrocarbon flow simulations. The poor

Copyright #2008. The American Association of Petroleum Geologists. All rights reserved. Manuscript received July 13, 2007; provisional acceptance November 7, 2007; revised manuscript received January 14, 2008; final acceptance February 21, 2008. DOI:10.1306/02210807078

AAPG Bulletin, v. 92, no. 6 (June 2008), pp. 789 824

789

Lex Rijkels  Mrsk Olie og Gas AS, Esplanaden 50, DK-1263 Kbenhavn K, Denmark; rij@maerskoil.com Lex Rijkels is a reservoir engineer with Maersk Oil in Copenhagen. He is interested in linking field development decisions to the resolution of geostatistical and production data. He has worked on fractured and carbonate reservoirs, faulted fluvial sandstones, tight gas-condensate chalks, and a range of primary to tertiary recovery developments. He started his career at Shell on the Carbonate Team and on a business unit in Damascus.

ACKNOWLEDGEMENTS The idea of this article was initiated during a reservoir characterization project at Shell Research in 2000. It significantly benefited from stimulating discussions with many staff members on the Shell Carbonate Team and operational business units during the 20002001 period. We specially thank Cathy Hollis from the Shell Carbonate Team who gave constructive suggestions for the improvement of the manuscript. We thank Jean-Pierre and Mugget Masse from the University of Provence who contributed to some outcrop mapping and Ludovic Laugero who helped in finalizing the drawings.

reputation of carbonate reservoirs in term of flow predictability is directly proportional to the expectation that these reservoir units behave like flow units, as most siliciclastic reservoirs. However, in carbonate reservoirs, the spatial correlations between reservoir units and flow units (Amaefule et al., 1993) are commonly weaker than in siliciclastic reservoirs. Complex diagenetic transformations and capillary forces are possible causes of these weaker spatial correlations (Figure 1). In addition, the high degree of heterogeneity of carbonate rock properties at all scales, as measured on well logs and core samples, for example, leads generally to the realization of subsurface models characterized by a high-frequency variability that cannot be predicted nor correlated in the space between wells. The challenge in carbonate reservoir modeling is to relate the heterogeneities measured at core and well scales to the spatial heterogeneities at reservoir and flow unit scales. Well correlations and seismic data are the only possible links between these two scales. The relevance of chronostratigraphical well correlations for siliciclastic reservoir modeling purposes has already been discussed by Ainsworth et al. (1999). The deterministic method of correlating stratigraphic markers between production wells implies spatial correlations between well and seismic data and has an equal influence on the threedimensional (3-D) reservoir rock property model and the seismic and geostatistic modeling methods. The intrinsic complexity of the carbonate reservoir at all scales (from pore network to stratigraphic architectures) makes it necessary to concentrate stratigraphic well correlation efforts on the level of reservoir heterogeneity, which matters in terms of reservoir and flow units. The main objectives of this article are to (1) review the processes of stratigraphic well correlation in carbonate reservoirs, (2) discuss its impact on reservoir and flow unit modeling, and (3) present some recommendations adapted to specific stratigraphic systems and reservoirs.

WHY CORRELATE WELL DATA? Short-range variability has very limited influence on the overall flow pattern (Jennings, 2000). Fluid displacement fronts tend to follow the larger scale structures, such as strata and sedimentary bodies, whereas small-scale heterogeneity, such as pore networks, rock textures, and sedimentary structures, only smears the front of the flow. Such principles also hold when the short-range permeability variation is much larger than the changes of the average permeability of larger scale
790 Well Correlations in Carbonates

Figure 1. Comparison between flow units and reservoir units in carbonate and siliclastic marine sequences. Siliciclastic sedimentary sequences are characterized by simple correlations between grain size, mineralogy, porosity, and permeability. Rock properties can be detected and correlated to stratigraphic sequences allowing the characterization of reservoir and flow units. These relationships are more complex or nonexistent in carbonate sequences characterized by a high level of rock property heterogeneity. OWC = oil-water contact.

structures. Jennings (2000) considerations were in line with Araktingi and Orr (1993), who also studied the effect of heterogeneity on flow, for different mobility ratios. They found that the difference in fluid mobility during water flooding may cause fingering at low heterogeneity, but that the viscous bypassing of oil becomes much less important for more heterogeneous samples. As a consequence, to understand the flow behavior at the interwell scale,

one needs to identify the structures that are correlatable between wells. Anything smaller than those structures is averaged out into the permeability of the correlatable units. This does not imply that one can ignore all the detail below this scale. One needs a reliable measure of the permeability statistics within each unit to identify the representative average properties. However, constructing a geological model at the smallest scale is not always necessary.
Borgomano et al. 791

In principle, one should aim to construct the architecture of the model at the scale of the correlated structures between wells and to use the measurement statistics made on smaller scales (e.g., core plugs) for each structures. Very thin units (<1 m, 3 ft) associated to extreme permeability values (high or low) can be retained in the model, provided they are correlated between wells and can form flow conduits or flow barriers. In an ideal case, let us assume that the rock properties are measured in all available wells with maximum resolution and precision. These are the only direct measurements of the subsurface reality and are the foundations of the static or dynamic reservoir modeling and subsequent field development planning. With the exception of a dynamic well test and production data that can give information on larger volumes of reservoir rock, away from the borehole, the well data only represent the rock volume sampled within the bore hole. This volume, which varies according to the measurement methods and the down-hole tools, is limited to the borehole (cores) and near borehole space (logs), representing a minute part of the reservoir rock volume. To achieve a complete 3-D numerical model of the reservoir parameters, all the available well data are necessarily interpolated and extrapolated (Figure 2). This is justified because the dynamic behavior of the reservoir is mainly controlled by the nonsampled rock volumes between the wells. In the cases of numerical modeling of sedimentary reservoir rocks, the ultimate objective of stratigraphic well correlations is to support the mathematical infilling (or modeling) of the undrilled space between the wells with realistic rock properties (Bashore and Araktingi, 1994). The common mathematical methods used to infill the interwell space can be grouped under the general term of geostatistics, which aims to characterize the intrinsic structure of the properties, allowing the modeling of reality and quantification of the model uncertainty (Matheron, 1989; Deutsch, 2002). Most of the geostatistic methods applied in oil industry are based on the stationarity hypothesis, which assumes that the processes controlling the distribution of any rock properties are independent of space and time (Carle and Fogg, 1996, 1997). This convenient mathematical as792 Well Correlations in Carbonates

sumption is generally too simple to explain realworld processes (Coulibaly and Baldwin, 2005) and especially in carbonate sedimentary systems, which are driven by dynamic and interdependent genetic processes (biological, physical, and chemical) that vary in time and space at all scales (Wilson, 1975; Schlager, 1999). Nonstationary mathematical methods (Prokophi and Barthelmes, 1996) such as wavelet analysis or Fourier transforms are valid for the analysis of the geological time series but are not adapted to spatial modeling of carbonate rock properties from well data. The common and rather intuitive practice in the carbonate reservoir modeling workflow (Figures 2, 3) is to increase the stationary nature of the systems (Bashore and Araktingi, 1994; Aitken and Howell, 1996; Weissmann and Fogg, 1999; Ravenne, 2002; Tinker et al., 2004) by the realization of a stratigraphic framework based on the well correlation. The main goal of the well correlation process is therefore to distinguish, at relevant scales, the correlatable from the noncorrelatable heterogeneities in space and to create subsets of data in which stationary geostatistic methods can be applied. The key question is, therefore, What are the valid well correlation methods to support the modeling of the undrilled space at the relevant scales? This question needs to be answered within the context of the reservoir modeling workflow (Figure 3) and relative to the objectives of the reservoir model; for example, the estimation of the volume of the hydrocarbon in place will favor the modeling of the reservoir units, whereas the planning of water-well injectors will require the modeling of the flow units. Dynamic simulation of hydrocarbon production will in general require the integration of reservoir units (initial conditions) and flow units (production behavior). A general carbonate static modeling workflow, used for the Malampaya field in the Philippines (Neuhaus et al., 2004; Fournier and Borgomano, 2007), is summarized in Figure 3. This workflow is a sequential and iterative workflow, where critical data are used for two main purposes: (1) establishment of a priori knowledge and geological concepts and (2) input to the numerical reservoir modeling. The geological and petrophysical concepts are used

to model the geobodies on the basis of deterministic well correlation and seismic horizon interpretation. Specific facies are attributed to the geobodies to condition the property simulation. The porosity

simulation results from a Bayesian mathematical method (Omre and Halvorsen, 1989) that combined kriging (cokriging) of well and seismic porosity within the stratigraphic framework. Other

Figure 2. Typical reservoir modeling workflow applied in the petroleum industry (modified from Ravenne, 2002; reprinted by permission of Institut Francais du Petrole). Well and seismic data are integrated in the first step of the workflow allowing the sequential realization of facies and rock property models in 3-D. Stratigraphic well correlations are inferred but not explicit in this workflow. Borgomano et al. 793

Figure 3. Reservoir modeling workflow integrating stratigraphic well correlation, seismic interpretation, and a priori geological knowledge in the third step of the workflow on the basis of a geological-petrophysical concept established by combining data analysis and a priori knowledge.

methods integrating the seismic data such as collocated cokriging have also been tested (Xu et al., 1992). Background permeability is derived from the porosity cube, using poroperm transforms, whereas extreme permeability values (very high and very
794 Well Correlations in Carbonates

low) are attributed to discrete geobodies. Property cubes are exported to the dynamic simulator and are used as an input to the volumetric calculations. Note that this workflow is not integrating production data because the gas production had not yet

started at the time of the studies. In this carbonate reservoir modeling workflow, like in many other cases (see, among others, Kerans and Tinker, 1997), stratigraphic well correlation is an early process in the development of a field, on which the modeling workflow is founded. Homewood et al. (1992) correctly stated that it is a critical working process that deserves to be formalized within the context of modeling carbonate sedimentary systems.

in reservoir modeling is not to infill the interwell space with sequences, lithologies, or times but to constrain in 3-D the petrophysical and dynamic modeling, the question is, therefore, What is the most reliable method to establish 3-D stratigraphic envelopes and corresponding bodies from wells? It can be based on various correlation methods, such as sequence stratigraphy, lithostratigraphy, chemiostratigraphy, and biostratigraphy supported by the interpretation of seismic reflectors.

DEFINITION OF STRATIGRAPHIC WELL CORRELATION In practice, correlating well data in the subsurface (Figures 4, 5) consists of a combination of subjective and objective processes based on stratigraphic analyses, interpretations, and assumptions that can lead to a reservoir model with hidden uncertainties. Avoiding a common misunderstanding in subsurface geosciences concerning the stratigraphic methods used with well data is important. To date, the primary focus of geologists has been the choice between sequence stratigraphy and lithostratigraphy (chronostratigraphy, biostratigraphy, cyclostratigraphy, and chemiostratigraphy can also be considered), whereas the basic objective of well correlation is mostly ignored and not well understood. When describing strata (drawing strata) on continuous outcrops and seismic data, all these stratigraphic methods are valid and can be used in combination or alone depending on the objective of the study. Sequence stratigraphy is the stratigraphy of the depositional sequences, whereas lithostratigraphy is the stratigraphy of the lithological units, chronostratigraphy is the stratigraphy of the time units, etc. In this respect, the process of well correlation in reservoir modeling is not a stratigraphic method sensu stricto (as opposed to seismostratigraphy) but a deterministic interpolation method between specific data points that are separated by variable distance in 3-D (along the borehole and between the wells). These points, stratigraphic well markers, deterministically selected along the borehole represent the intersection between the well trajectory and possible 3-D stratigraphic envelopes. Because the ultimate objective of well correlation

SPATIAL RELATIONS BETWEEN ROCK PROPERTIES AND STRATIGRAPHIC CORRELATIONS Before correlating wells, defining the degree of spatial relation between rock properties and possible stratigraphic architectures (Figures 6, 7) as illustrated is important, for example, in the cases of the Malampaya field (Fournier and Borgomano, 2007), the Al Ghubar field (Smith et al., 2003), and the Fahud field (Morettini et al., 2005). This is not an easy task because the wells, in addition to their lateral spacing, sample the rock in one direction, introducing a significant bias. A stepwise petrophysical analysis of the well data (Asquith, 1985) is recommended to establish the spatial relationships between rock properties and stratigraphic architectures. This process includes geostatistical analyses with several iterations based on various well correlation scenarios (stratigraphic, diagenetic, and structural) and the introduction of a priori knowledge from analogs (Figure 3). At a higher hierarchical level, it can be considered that the distribution of rock properties in most carbonate reservoirs is either conformable or not conformable to the sedimentary facies and the stratigraphic units (genetic stratigraphic units, sensu Homewood and Eberli, 2000). In the conformable cases, the diagenetic transformations (from early to late and from shallow to burial) overprint the stratigraphic architecture and the sedimentary facies, whereas the nonconformable cases are characterized by diagenetic transformations cutting through the stratigraphic architecture and the sedimentary facies, in relation to fractures (Stearns and Friedman, 1972),
Borgomano et al. 795

Figure 4. Example of layer-cake reservoir units (A, B, and C) in a giant carbonate field in the Middle East. (A) Logbased stratigraphic well correlation of the main lithostratigraphic units formed by alternating porous limestones with low gamma ray and denser argillaceous limestones with higher gamma ray (from Alsharhan, 1993; reprinted with permission from AAPG). (B) Cross-section showing the porosity model based on well and seismic data (from Melville et al., 2004; reprinted by permission of the AAPG whose permission is required for further use). The tops of the reservoir units A, B, and C correspond to seismic reflectors that allow a coherent structural model of the reservoir throughout the entire structure. GR = gamma ray, N = neutron, OWC = oil-water contact.

unconformities (Budd et al., 1995), depth trends (Ehrenberg and Nadeau, 2005), or karsts (James and Choquette, 1988). Distinguishing the two cases is critical, especially if they coexist in a single carbonate reservoir. In this process, the early identification of genetic and spatial hierarchies is very important where the two systems are mixed and not clearly stratified. One of the biggest sources of error
796 Well Correlations in Carbonates

in 3-D modeling of rock properties is the early introduction in the workflow of unrealistic stratigraphic well correlations not conforming to the diagenetic transformations and subsequent property distribution. Another common error is assuming that diagenetic transformations (and properties) conform to the stratigraphic architectures and sedimentary facies at all scales and hierarchical levels.

Figure 5. Example of layer-cake flow units and mechanical units in a giant carbonate field (Fahud) in the Middle East (modified from Morettini et al., 2005). This figure shows how high-resolution stratigraphical well correlations (A) allow a mechanical stratigraphic unit model to be established and implemented in the static reservoir model (B) prior to flow simulation. (A) Log-based well correlation of 3rd- and 4th-order stratigraphic sequences that are interpreted to control the flow unit architecture of this field on the basis of core and well test analyses. Core and well analyses allow a concept of fracture distribution to be established in relation to the stratigraphical sequences. (B) The bedbounded fracture model of this reservoir is built on the basis of the stratigraphic well correlation of the mechanical units. In this case, the regressive sequences are used to populate the fractures in the property model. Fractures are not associated with transgressive sequences (reproduced by permission of GeoArabia).

Borgomano et al.

797

Figure 6. (A) Depth (in meters subsea) of the top Nido Limestone and location of wells Malampaya-1 (MA-1), Malampaya-2 (MA-2), Malampaya-5 (MA-5), and Malampaya-7 (MA-7). (B) Histograms of porosity (from well logs) within the SM2 sequence showing the eastward lateral increase in porosity. (C) Correlation panel of depositional sequences and seismodiagenetic units between wells Malampaya-1, Malampaya-5, and Malampaya-7, showing the eastward pinch-out of the SM2-IIIb tight unit (modified from Fournier and Borgomano, 2007; reprinted by permission of the AAPG whose permission is required for further use).

GEOSTATISTICS AND WELL CORRELATIONS As introduced above, the process of stratigraphic well correlations must be supported by spatial statistic analysis of rock properties (Sonnenfeld and
798 Well Correlations in Carbonates

Cross, 1993; Pranter et al., 2004). The purpose is to identify and measure the property anisotropy and the spatial correlation between the data and possible stratigraphic correlations (Goovaerts, 1997; Jennings, 2000; Gringarten and Deutsch, 2001). In

Figure 7. Spatial correlation between stratigraphic units and permeability in a Middle East Cretaceous reservoir formed by carbonate platform deposits. Permeability data from 65 cores taken from the same reservoir unit throughout the entire field are depth-plotted relative to a stratigraphic datum (the uppermost sequence boundary). The averaged thickness of the stratigraphic sequences (4th orders) and the lithostratigraphic units interpreted from the cores are plotted on the right of the graph (A). The vertical permeability trend is shown by the variogram (B). Despite the difficulty to interpret it relative to the stratigraphic sequence, such a permeability trend has to be taken into account in the stratigraphic well correlation process especially if the objective is to model the flow units.

Borgomano et al.

799

fact, it allows the verification of the stationarity or the nonstationarity of the inferred geological processes that controlled the reservoir rock properties. A typical inconsistency observed in carbonate reservoir modeling is introduced where 3-D distribution properties are controlled by diagenetic transformations cutting through the sedimentary stratigraphic architecture (Fournier and Borgomano, 2007). Without accounting for this diagenetic control, the stratigraphic correlations result in unrealistic rock property modeling. The idea is to establish a stratigraphic scenario that considers the spatial correlations between property and strata at specific scales, from the core plugs to the well interval. Common geostatistic libraries such as the Geostatistical Software Library (GSLIB) (Deutsch and Journel, 1998) allow such analysis in most commercial geomodeling softwares (Mallet, 2002). Trend analysis (variation as a function of direction, distance, and position) can help to establish the most valid correlation at the right scale and account for the nonstationarity of the geological process. In carbonates, it is frequent to observe vertical property trends relative to a stratigraphic datum (exposure surface, sequence boundary) (Figures 7, 8). Simple vertical variogram analysis and moving averages can help to identify such trends that can deterministically influence the well correlations. Nonspatial and spatial correlations between a primary continuous variable (permeability, porosity,. . .) and a secondary discrete variable (texture, grain size, facies,. . .), or two continuous variables (porosity and permeability) can also be addressed prior to the well correlation. Transition probabilities between facies (Weissmann and Fogg, 1999) can also be investigated within correlated depositional sequences (Posamentier et al., 1988). This initial statistical analysis is critical for the 3-D permeability modeling and upscaling, which are the most challenging and uncertain processes in carbonate reservoir modeling for the four following reasons: (1) permeability is a tensor (not a scalar number), (2) its detection with seismic and log data is not accurate, (3) averaging permeability (from core to grid cells) is problematic because of significant support effects in relation to the multiscale heterogeneities (core plug to well tests), and (4) the relationship between porosity and permeability is nonlinear in
800 Well Correlations in Carbonates

most carbonate reservoirs and interpolating permeability (i.e., kriging) from well data results commonly in unrealistic and smooth realizations (Matheron, 1989; Araktingi and Bashore, 1992; Bashore and Araktingi, 1994). In this respect, the early identification of spatial correlation between permeability and strata or trend can be taken into account in the correlation scenario. In carbonate reservoir modeling, it must be realized that well correlations can have a greater impact on a 3-D permeability model than on a 3-D porosity model (Jennings, 2000).

OBJECTIVES OF RESERVOIR MODELING The last statement introduces the importance of defining the objectives of the reservoir modeling prior to the well correlation process. A common worst practice in the industry is that the technical goals of reservoir modeling do not direct the (commonly time consuming) well correlation process despite their significant impact on the modeling outcomes. Obtaining high-resolution stratigraphic correlations becomes the primary goal and the causes of many technical arguments among geologists. Two typical modeling strategies are discussed below with the caveat that many other practices and methods are appropriate for varying situations.


Modeling reservoir units and porosity distribution for hydrocarbon volumetric assessment or appraisal well planning is based on stratigraphic architectures established from seismic interpretation matched to the wells, such as illustrated in the examples of giant fields in the Middle East in Figure 4 (Alsharhan, 1993) and Figure 9 (Munn and Jubralla, 1987). The well correlation focuses on stratigraphic envelopes that are meaningful at seismic scale and can condition the seismic inversion (Melville et al., 2004). Well correlation consists of picking the well markers that separate rock units on the basis of porosity, acoustic impedance, and lithological contrasts (Figure 1). When these markers are correlated between wells and are matched to seismic reflectors over the entire structure of the field, they correspond to the primary envelopes of the rock units that

Figure 8. Sequence-stratigraphic correlations between two cored intervals in a carbonate reservoir formed by carbonate platform deposits. The core-plug permeability and the pore types indicate that the three vertical permeability trends are overall correlated to the three sequences that should be correlated between the wells. The stratigraphic correlation of the off-trend high permeability values (related to vugs) is, however, questionable: are they isolated and randomly distributed in space or are they representative of high permeability thin beds? The answer might come from the detail analysis of many cores and will have a very high impact on the flow unit model. This virtual example shows the impact of stratigraphic well correlation on reservoir modeling.

characterize a field: reservoir units, waste zones, and seals (Roehl and Choquette, 1985; Jardine and Wilshart, 1987). Modeling flow units (Amaefule et al., 1993), to support field development, a water flood plan, for example, is based on 3-D high-resolution petrophysical modeling that is generally beyond seismic resolution (Morettini et al., 2005). Establishing the degree of spatial correlation between seismic scale geobodies and potential flow units

(Araktingi and Bashore, 1992; Yao and Journel, 2000; Fournier and Borgomano, 2007) is important; for instance, how does the permeability that controls the well flow correlate to the acoustic property detected in the seismic data? A common error is to assume that the correlation between these two parameters is high without properly checking the relevant scales. Important parameters in modeling flow units are the sweeping radius of each well and the communication
Borgomano et al. 801

Figure 9. High-resolution stratigraphic correlations in the Bul Hanine field, Arab D member, Qatar (modified from Munn and Jubralla, 1987). This graph shows the spatial relationships, in a flat carbonate shelf system such as the Arab D, between time lines and rock units. The two sets of correlations are overlapping or conformable with each other. Such stratigraphic correlations will result in a stratified, layer-cake, reservoir model. The challenge will be to preserve in the dynamic model the extreme property values that are stratigraphicaly correlated and have the potential to form barriers or conduits to flow. GR = gamma ray, FDC = density.

between producing or injector wells. Prior to the stratigraphic well correlations, the interpreter must identify the permeability trends and the low-frequency spatial heterogeneities that are
802 Well Correlations in Carbonates

correlatable between wells (Figure 8). Noncorrelatable, high-frequency spatial heterogeneities are in reality averaged out by the flow and do not affect the flow interference between wells

(Araktingi and Orr, 1993; Jennings, 2000). A common mistake is to force the well stratigraphic correlations on the high-frequency variability that is observed at well and core scales (in response to local diagenetic transformations for example) but is not necessarily correlated between wells (Figure 8). This process will inevitably increase the impact of the noise in the final numerical model (Coulibaly and Baldwin, 2005). The challenge is, however, to link such high-frequency variability to genetic stratigraphic units (sequences, sedimentary, or diagenetic bodies) that can be correlated in space and to apply geostatistics to model them within each unit. These static models are very sensitive to well spacing; their dynamic simulation allows the identification of the most efficient and economic well spacing according to the spatial heterogeneity and the fluid properties.

relate to diagenetic transformations that more or less conform to the stratigraphy (Fournier and Borgomano, 2007). Typical mistakes include correlating noise and forcing stratigraphically valid, but unrealistic, from a well-spacing viewpoint, correlations. Such misunderstandings will have little impact on the initial hydrocarbon productions from immature assets but could have significant negative consequences on long-term field developments and ultimate oil recovery from mature fields (Chilingarian et al., 1992). During the initial phase of the modeling workflow (Figure 2), the interpreters must distinguish the three main categories, geological unit, reservoir unit, and flow unit, according to the available data set. It is, however, expected that, at an early stage of a field appraisal, when data are scarce and production has not yet started, they cannot be separated. Geological Units

WHAT TO CORRELATE? The previous paragraphs summarize the scope and main objectives of the process of stratigraphic well correlation that forms the foundation of many carbonate reservoir models. It is therefore important to distinguish three main types of critical rock units that coexist in each carbonate field and can be correlated between wells: geological units, reservoir units, and flow units. The degree of spatial correlations between these three units can range from very low to very high and from field to field (Roehl and Choquette, 1985; Moore, 2001). A general but critical observation is that reservoir units and flow units always correspond to geological units but not necessarily the same ones and not necessarily the ones we interpret from the data set. The main purpose of this distinction is to focus the attention of the interpreter on the fact that despite the high resolution of the stratigraphic correlation scheme and the time spent to achieve it, the result could well be useless, or even worse, introduce major hidden errors regarding the 3-D property modeling and the flow simulation. A typical error is to assume that the 3-D distribution of porosity and permeability is exclusively related to stratigraphic sequences and sedimentary bodies, whereas, in reality, they corGenerally, a geological unit (geobody) can be considered as a volume of rock material bounded by envelopes genetically related to a set of specific geological processes. In general and in carbonate systems in particular, they are not necessarily equivalent to units with specific reservoir properties such as porosity, permeability, and elastic properties (Figure 1). The body envelopes, which could correspond to genetic stratigraphic surfaces (Homewood and Eberli, 2000), are commonly picked on the basis of well-log responses (Schlumberger, 1972, 1974; Asquith, 1985) and used as correlations between wells (Homewood et al., 1992; Sonnenfeld and Cross, 1993). The relation between stratigraphy, especially high-resolution sequence stratigraphy, and carbonate geobody can be confusing, for example, if the geometry of the sedimentary, diagenetic, or rock property body does not conform to the time surfaces (Figure 10). Four main types of geological objects are commonly correlated in carbonate reservoirs:


Stratigraphic sequences: the most common geological units that are correlated in carbonate fields are sequence-stratigraphic units formed by sets of various sedimentary and diagenetic objects, which are genetically related (Posamentier et al.,
Borgomano et al. 803

Figure 10. Different types of spatial correlations between correlated time lines and simple sedimentary objects. The sedimentary objects are defined by sharp (A) or gradual (B) facies changes that are characterized by different degrees of spatial correlation with the time lines. Stratigraphic well correlations of such sedimentary objects will be influenced by well locations, well spacing, and a priori knowledge on the objects. In both cases, the grainstone objects, which would be critical for the reservoir model, are not entirely conformable with stratigraphic envelopes.

1988; Sarg, 1988; Van Wagoner et al., 1988; Pomar, 1993; Schlager, 1999). They have to be interpreted in a hierarchical system (2nd to 7th orders) and are topped by sequence boundaries, which are considered to be time surfaces ( Vail et al., 1977) and therefore equivalent to paleosedimentary profiles. The sequence boundaries will become an important parameter when analyzing further the relationships between well spacing and stratigraphic well correlations. Maximum flooding surfaces can also be identified and correlated within the sequences. Sequence804 Well Correlations in Carbonates

stratigraphic correlations are not too sensitive to field-scale well spacing in the case of wide and flat sedimentary profiles (e.g., a carbonate shelf ) such as the Natih Formation in the Middle East (Van Buchem et al., 2002) that is illustrated in the well correlation of the Fahud field (Morettini et al., 2005) (Figure 5). Similar considerations can be applied to the Jurassic Arab Formation (Figure 9). In the case of fields located on gentle carbonate slopes, for example, around the Lower Cretaceous Bab Basin in the Middle East (Boote and Mou, 2003; Droste and Steewinkel,

Figure 11. Well correlations of reservoir units in Middle East Cretaceous carbonate platform systems, based on sequence stratigraphy. These correlations display lowangle paleotopographies (clinoforms) that can be very critical for the distribution of reservoir property at basin and reservoir scales. They are not easy to pick in wells without integrating highresolution sequence stratigraphy and seismostratigraphy. (A) The Shuaiba Formation in the Safah field (modified from Boote and Mou, 2003; reproduced by permission of GeoArabia) and (B) the Natih formations (modified from Droste and Steewinkel, 2004) in north Oman. Such clinoforms represent prograding trends expressing the lateral accretion through time of the platform sedimentary profiles.

2004), the well spacing will influence the well correlation depending on the paleoslope angle that controls the stacking of the stratigraphic sequences (Figure 11). Sedimentary bodies: these bodies can be considered as continuous volumes of carbonate sedi-

ments (biostrome, bioherm, bank, patch reef, channel, lobe, mound, dune,. . .), resulting from a single set of depositional processes (Borgomano et al., 2001) occurring in specific environments of deposition (Scholle et al., 1983). In theory, their envelopes should represent time surfaces
Borgomano et al. 805

that can be correlated between wells within a sequence-stratigraphic framework. Sedimentary bodies can also be formed by stacked diachronous deposits of the same nature, which can be assembled into objects bounded by composite surfaces or gradual changes (Figure 10). Dimensions of carbonate sedimentary bodies, such as the Cretaceous rudistid bodies common in Middle East reservoirs, vary significantly even in a single platform system depending on their environments of deposition (Borgomano, 2000; Borgomano et al., 2002). In this case, well correlations can be very sensitive to well spacing (Figure 12). Diagenetic bodies: these bodies can be considered as volumes of carbonate rock material transformed or formed by diagenetic processes (cementation, dissolution, recrystallization, and dolomitization) and confined to a discrete spatial entity (bed, lens, etc.) (Budd et al., 1995). Such bodies can be defined by envelopes that can be picked in the wells or by gradual changes more difficult to correlate. Diagenetic bodies can conform or not conform to sedimentary bodies depending on the types and timing of diagenetic transformations (see, for example, Fournier and Borgomano, 2007). It is therefore important to establish the relationships between these two genetic types of bodies when they coexist in a single reservoir. Diagenetic bodies can also conform to structural deformation (faults, fractures, and folds). Structural objects: these objects correspond to the volume of rock bounded by the structural discontinuity of tectonic origin such as faults or thrust planes. Carbonate reservoir architectures based on the well correlations of tectonic discontinuities are rare.

Reservoir Units Reservoir units in carbonate rocks correspond to storage units characterized by specific porosities and pore entry pressures that control the distribution of hydrocarbon within the trap (Chilingar et al., 1972; Jardine and Wilshart, 1987; Roehl and Choquette, 1985; Chilingarian et al., 1992). Capillary or static seals formed by specific lithologies
806 Well Correlations in Carbonates

such as shale, salt, anhydrite, and dense carbonates bound the reservoir unit. Pore volume is the primary parameter that controls porosity. Faults can also provide effective seals. From a storage point of view, the most complex carbonate oil fields can be characterized by long transition zones, multiple and tilted oil-water contacts, and vertical and horizontal pressure compartments. The identification and mapping of such reservoir heterogeneity at an early stage of the field development (based on 3-D seismic data associated with appraisal wells) are the foundation of a successful economic development. This first level of heterogeneity controls the hydrocarbon volume in place in the carbonate fields but not necessarily the dynamic behavior of the reservoir. In carbonate systems, hydrocarbons are commonly stored in all the carbonate rock units (limestone and dolomite) independently of the grain size, unlike in siliciclastics reservoirs (Figure 1). For example, chalky reservoirs with microporosity are typical in carbonate petroleum provinces as a result of favorable entry pressures and capillary forces. Interbedded noncarbonate units, such as salt, anhydrite, or shale, form the nonreservoir, the seal, or the waste zones. Correlating reservoir units in carbonate systems consists generally of separating carbonate from noncarbonate units or porous from nonporous carbonates, using core and well logs (Asquith, 1985), within a consistent stratigraphic framework (Figures 4, 5, 9, 11). In addition to noncarbonate units, wellcemented and nonporous carbonates can form nonreservoir units, which have to be discounted from the reservoir units. In giant carbonate fields, reservoir units are generally easy to correlate between wells and to image with seismic reflection as consequences of their wide development at a regional scale, combined with their significant rock property contrast with nonreservoir units (Eberli et al., 2004; Neuhaus et al., 2004; Fournier and Borgomano, 2007). Flow Units A flow unit, or hydraulic flow unit (Ebanks, 1987; Amaefule et al., 1993), is defined as the representative volume of total reservoir rock within which

Figure 12. Spatial relationships between well distribution and sedimentary systems. Hydrocarbon fields are generally smaller than the entire sedimentary system. The well spacing, locations, and numbers determine the sampling pattern of the facies heterogeneity and determine the stratigraphic rules applied for the well correlations. If the well spacing is more than the average dimensions of the sedimentary objects, only sequence-stratigraphic correlations can be applied to condition object-based or pixel-based facies models. If the well spacing is less than the average dimensions of the sedimentary objects, sequence and lithostratigraphic correlations can be combined to condition pixel-based facies models.

geological properties that control fluid flow, from the reservoir into the wells, are internally consistent and predictably different from properties of other rocks. A flow unit is a discrete rock unit that is in flow communication in a field with at least one well. Unlike reservoir units, flow units have no meaning as a subsurface object without physical connection to wells (producers or injectors). Flow units can also vary according to the dynamic

conditions applied to the reservoir. Their geological properties are a function of pore geometry, pore throat, and pore distribution that are, in turn, controlled in carbonate rocks by the depositional texture, the grain mineralogy, and the diagenetic transformations. In this article, we restrict our definition of flow units to units of absolute permeability defined as the permeability of a rock to a fluid when the rock is 100% saturated with that
Borgomano et al. 807

fluid. In this manner, only rock properties are considered to control the flow units. Instead of absolute permeability, it is possible to consider effective permeability, defined as the ability of a rock to flow a particular fluid when another immiscible fluid is present in the pore space. Effective permeability, which is always less than the absolute permeability, is therefore dependent of the pore saturation (oil and water), introducing a fluid component to the definition of the flow units. Effective permeability units might be investigated in case of a complex fluid distribution in the reservoir, especially when gas, oil, and water are mixed following a long period of production and reservoir stimulation (water flooding, gas injection, etc.). The identification of flow units relies on the permeability detection in the subsurface, which is very limited and uncertain in carbonate systems for the following reasons: (1) direct down-hole measurements of permeability (as a physical property) are not routinely acquired by logging tools, unlike porosity or water saturation; only nuclear magnetic resonance (NMR) logs proved to be successful in some sandstone reservoirs (Kenyon, 1997; Coates et al., 1999); (2) the seismic reflection method is not capable of detecting permeability heterogeneities, especially within carbonate rocks with homogeneous porosity and acoustic impedance (Anselmetti and Eberli, 1993); (3) depending on the depositional and diagenetic factors, flow units can be very thin (<0.5 m, 1.6 ft) compared to reservoir units (Lucia, 1995, 1999) and beyond seismic and log resolutions. In practice, it should be concluded that seismic and logging methods do not allow the straightforward detection of permeability and flow units in carbonate reservoirs. The only two methods supporting without equivocation the detection of the permeability are (1) laboratory permeability measurements on core samples and (2) well tests. In practice, flow units will be assessed by an integrated petrophysical and geological numerical approach based on a core-calibrated petrophysical log evaluation, including borehole image, resistivity, NMR, and production tests. The spatial and genetic relationships between flow units and geological units must be established prior to attempting the well correla808 Well Correlations in Carbonates

tions, especially in uncored wells (Morettini et al., 2005). A typical worst practice in the industry is to correlate directly flow units from well logs, as stratigraphic units, without integrating all the previous data. Figure 8 shows an example of flow unit correlations based on core data. In this synthetic example, only the vertical permeability trends, corresponding to the stratigraphic sequences, have been correlated.

WELL SPACINGS AND STRATIGRAPHIC CORRELATION The ultimate objective of stratigraphic well correlation is to support and condition rock property modeling for simulation and prediction of fluid flow in hydrocarbon fields. One of the most fundamentals laws of hydrocarbon reservoir engineering is the famous Darcy law that clearly states that spatial dimensions are critical parameters of the fluid dynamics in porous media (see, among others, Dake, 1978). To simplify, for a given rate of total oil production at field scale, the spacing of producing wells would increase with increasing rock permeability and decrease with increasing oil viscosity. Well spacing is also equally critical to the process of modeling rock properties (Figures 2, 3), with geostatistic methods involving variograms (Armstrong, 1984). Although it might be intuitive to most geologists that flow rates and property modeling at field scale are a function of well spacing, it is not a certainty that this critical parameter is correctly taken into account when correlating stratigraphic markers between wells. The paradox is that the recent and fundamental introduction of high-resolution sequence-stratigraphic concepts in the well correlation process (Kerans and Tinker, 1997; Homewood and Eberli, 2000) is potentially misleading as it consists of the correlation of a time series intrinsically independent of the well spacing, especially at field scale. The purpose of this paragraph is to illustrate the impact of well spacing on stratigraphic well correlations and to indicate some practical approaches to improve the validity of the correlations and avoid unrealistic reservoir models (Figures 1117). In the

Figure 13. Spatial relationship between well distribution and stratigraphic trends in carbonate platform systems. This figure illustrates three typical stratigraphic architectures with a different facies partitioning and sedimentary profile expressing the regression of the carbonate platform system: ramp (1), flat-topped shelf with a low-angle slope (2), and flat-topped shelf with steep slope (3). The stratigraphical correlation of vertical trends in wells is strongly influenced by the position and the spacing of the wells relative to the geometry of the platform system especially to the zones of progradation and aggradation. Distinguishing these three architectures in the subsurface, especially if they are beyond seismic resolution, is commonly a challenge (see well spacing sensitivity analysis in Figure 14).

process of stratigraphic well correlation, it must be clearly considered that production wells and well spacing determine the fluid flow and influence the property modeling but have no impact on the geological reality. Wells are random or deterministic samples of this geological reality, which could even overestimate reservoir and flow units if they target the best areas on the basis of concepts and models. In fact, the process of stratigraphic well correlation is introduced in the reservoir modeling workflow to compensate for the lack of interwell geological information and improve the validity of the geological and numerical models (Figures 2, 3). The conceptual representation and the numerical modeling of the reservoir depend significantly on how accurately the wells are sampling the subsurface reality. Thus, it is not a straightforward exercise

to estimate the validity of the well sampling relative to an uncertain geological reality in 3-D and furthermore to compare the well spacing against the lateral heterogeneity of the reservoir (Figure 12). The knowledge of the geological reality is based on the subsurface data, including the wells themselves and the seismic and the dynamic data. The major problem, generally addressed by geostatistics (Matheron, 1989), is that the optimized integration of those three sources of information gives only an approximation of the geological reality, especially if the information is not representative. A common practice is to validate the stratigraphic model and the subsequent geological reservoir model by production dynamic data (Bashore and Araktingi, 1994; Jennings, 2000). This is a practical and valid approach, especially in mature fields
Borgomano et al. 809

Figure 14. Well spacing sensitivity analysis for stratigraphical correlation. The purpose of this figure is to illustrate the influence of well spacing and well density on the stratigraphic correlation of the three architectures of carbonate platform systems illustrated in Figure 13. Well correlation of aggrading zones is not sensitive to well spacing and to stratigraphic rules: sequence stratigraphy and lithostratigraphy can be indifferently applied and have identical results. Well correlation of prograding zones is, however, very sensitive to well spacing and to stratigraphic rules: sequence stratigraphy is better adapted in this situation. Correlation of the ramp system (A) is less sensitive to well spacing and to stratigraphic rules than the two flat-topped shelves (B, C). In these two cases, the correlation of the prograding zones is strongly dependent of the well spacing. In the three cases (A, B, C), when well spacing exceeds mostly a threshold, equivalent to the average lateral dimensions of the facies zones (A4, B4, and C4), the well correlations will be unrealistic and uncertain for any stratigraphic rules. In these cases, well correlation of different stratigraphic architectures can look identical for any stratigraphic rules (A4 = B4 = C4). Only seismic images could help to differentiate the three cases.

810

Well Correlations in Carbonates

Figure 15. Trigonometric relationships of carbonate sedimentary profiles. (A) This simple prograding carbonate ramp system is formed by stacked shallowing upward sequences that can be detected in wells by the analysis of vertical sedimentary trends (triangles). (B) Well correlations of such stratigraphic sequences can help to establish the geometry of the carbonate ramp as a simple sedimentary prism (triangle) characterized by a thickness (E), a length (L), and an angle (A) relative to a given horizontal stratigraphic datum. (C) Assuming that L is the distance between the correlated wells, a simple trigonometric relationship between this distance, the angle (A), and the thickness (E) is seen. Such a relationship can be used to check the validity and the self-coherence of the well correlation as explained in Figures 16 and 17. In specific cases, it can also be assumed that E is equivalent to the maximum paleobathymetry along the sedimentary profiles. This parameter can be interpreted from the core analysis and introduced in the validation of the well correlation as explained in Figures 16 and 17.

with a long production history. It needs, however, to be funded on realistic geological scenarios and stratigraphic models to avoid a random validation of the dynamic models, which could have a significant negative impact on longer term field development.

The suggested approach for the static (geological) data is to integrate a priori stratigraphic knowledge, compare the well information to stratigraphic and reservoir analogs, and carry out several iterations (Figure 3). This approach formalizes the intuitive
Borgomano et al. 811

812

Well Correlations in Carbonates

Figure 17. Crossplot showing theoretical trigonometric relationships between well spacing (L), base profile angles (A), and paleobathymetric differences (N) calculated along the base profile between two correlated wells. The curves show the relation between sedimentary profile angles and spacing of two correlated wells for various paleobathymetric differences along the sedimentary profile between the two wells. Based on the simple trigonometric relationship (tanA = N/L), similar to the equation in Figure 15, such plots could help to check the geometrical self-coherency of the stratigraphic correlations given the paleobathymetric interpretations in the wells.

approach typically applied by geologists in the well correlation process. Our purpose here is to discuss the impact of well spacing on stratigraphic well correlations in the simple cases of carbonate platforms and ramps (Read, 1985), which are hosting the most prolific hydrocarbon reserves (Carmalt and St. John, 1986; Jardine and Wilshart, 1987).

Sedimentary Bodies In fields where it is established that reservoir and flow units are controlled by discrete sedimentary bodies (sand shoals, rudist banks, etc.), it is critical to compare their dimensions relative to the well spacings and to the field extension (Figure 12).

Figure 16. Trigonometric relationships for the well correlation of carbonate ramp systems. (A) The diagram displays two successive sedimentary profiles or base profiles (BP) consequent to a regular increase in accommodation without differential subsidence. The successive sea levels are noted as SL. Assuming the paleobathymetry of the two couples of stratigraphic markers that are correlated between the wells, a simple trigonometric relationship between the well spacing (L), the base profile angles (a, b), and the sediment thickness (E1, E2) is observed. (B) The diagram displays two successive base profiles consequent to a differential increase in accommodation (rotation or differential subsidence). The equation, derived from the previous one, expresses the trigonometric relationships between the well spacing (L), the sediment thickness (E1, E2), the base profile angles (a, b), and the angle of rotation between the two base profiles (g). This angle is referenced to as a horizontal datum (dashed line, SL b) parallel to the sea level. Borgomano et al. 813

Deterministic well correlation of such sedimentary bodies must be based on the valid assumption that well spacing is less than the lateral extent of the bodies (Borgomano et al., 2001). This assumption should be validated at the regional and global scales with the support of an analog database. If well spacing is greater than the lateral extent of the bodies (Figure 12), the deterministic well correlation of sedimentary bodies will force unrealistic sedimentary architectures and subsequent reservoir models. In such cases, the recommended approach is to correlate sequence-stratigraphic units, which comprise the bodies, and model the sedimentary bodies with stochastic methods (Dubrule, 1993; Goovaerts, 1997). The two main stochastic methods that can be applied are pixel-based (Koltermann and Gorelick, 1996) and object-based (Deutsch and Wang, 1996; Viseur, 2001). The pixel-based method simulates a facies property in a 3-D grid while taking spatial statistical parameters and available data into account (Suro-Perez and Journel, 1990). To capture and reproduce facies spatial repartition, most of these approaches rely on the variogram, which is a two-point statistic and cannot detect complex shapes. The object-based method aims to, first, define template objects that mimic sedimentary body geometries and, second, to distribute them stochastically in the modeled volume while accounting for available data and facies proportions (Viseur, 2001). Most of these approaches integrate in the simulation processes geological rules to generate 3-D models that look geologically sound. Sequence-Stratigraphic Systems In hydrocarbon fields hosted by extensive and flat carbonate ramps or platforms, the sequencestratigraphic architecture is commonly the key to the reservoir and flow models (Kerans and Tinker, 1997; Homewood and Eberli, 2000). Where the spatial and genetic relationships between petrophysical trends and stratigraphic sequences are clearly established (Figures 68), sequence-stratigraphic well correlations become one of the most critical steps of the reservoir modeling workflow. Ideally, the well correlation should be founded on a coherent stratophysical concept integrating the data (wells
814 Well Correlations in Carbonates

and seismic) and a priori knowledge (analog database and regional trends) (Figure 3). At this stage of the workflow, it is critical to compare the well spacing and the spatial heterogeneity of the stratigraphic systems (Figures 12, 13). In such flat and extensive systems, subtle changes of sedimentary profiles will affect the rock property and facies partitioning (Figures 11, 13). Therefore capturing this depositional architecture from the seismic data and from the well correlation is critical. Although it is generally admitted that the sequencestratigraphic well correlation is the most reliable method, it is important to consider the well spacing relative to the stratigraphic architectures (Figures 13, 14). In the aggrading parts of the systems, well correlations are not sensitive to well spacing and to correlation methods: lithostratigraphic and sequence-stratigraphic correlations result in the same stratigraphic architecture independently of the well spacing (Figures 13, 14). In the prograding parts of the systems, well correlation is very dependent of well spacing, and lithostratigraphic methods should be avoided. This principle should also be applied to transgressive systems. This sensitivity analysis shows also that the sequence-stratigraphic method does not restore realistic stratigraphic architectures if the wells are not sampling the spatial heterogeneity of the stratigraphic systems (A4, B4, and C4 in Figure 14). The lithostratigraphic method would be equally wrong in this case. Sedimentary Profiles In extensive and flat carbonate ramps or platforms, volume partitioning of sedimentary facies is mainly controlled by the evolution through time of the base profiles, or sedimentary profile (Quirk, 1996), in response to changes of accommodation space and sedimentary fluxes (Robin et al., 2005). Facies partitioning is in turn expressed by the evolution of the sedimentary profiles as illustrated by outcrops and subsurface data in modern and ancient carbonate sedimentary systems (Pomar, 1993; Adams and Schlager, 2000; Boote and Mou, 2003; Droste and Steewinkel, 2004; Pranter et al., 2004). It implies that the reservoir model should be based on facies

distribution consistent with the spatial evolution of the sedimentary profiles. In theory, this consistency is guaranteed by the well correlation of stratigraphic sequences, assuming that the correlated sequences are representative of the sedimentary profiles (Figure 15A, B). The validity of the well correlation is therefore dependent of the well spacing relative to the sedimentary profiles (Figure 15C). Flat and wide carbonate ramps or platforms are characterized by low-angle-average sedimentary profiles that are beyond seismic resolution at field scale (Boote and Mou, 2003; Droste and Steewinkel, 2004). The only method to estimate the average angle of the sedimentary profiles is to integrate the paleobathymetry of the correlated well markers and the well spacing (Figures 15, 16). The introduction of the paleobathymetry parameter (inferred from biota and sedimentary structures) in the correlation process of carbonate platforms and ramps was implicit in the Neptune modeling reservoir modeling workflow (Massonat, 2001). It infers the existence of a horizontal datum sea level, allowing simple trigonometric calculations to assess the validity of the correlation (Figure 15C). Although the quantification of the paleobathymetry in wells is uncertain, this method should be used to estimate the internal coherency of the stratigraphic model. The crossplots in Figure 18 illustrate the relationships between the average angle of the sedimentary profile and the well spacing for a given paleobathymetric gradient between wells. Such graphs could be used in principle to estimate the range of possible correlations depending on the well spacing and the interpreted paleoenvironments (Figure 13). The graphs could also support the validation of lowangle clinoforms interpreted from seismic data (Schwab et al., 2005). The previous trigonometric calculations involved a single sedimentary profile correlated between wells. Actually, the sequence-stratigraphic correlation process infers multiple and stacked sedimentary profiles that are spatially and genetically related to each other. Two cases have been investigated without and with differential subsidence or tectonic unconformity (Figure 16). In this trigonometric calculation, we are trying to relate

critical parameters to assess the validity of the well correlation. In the first case without differential subsidence (Figure 16A), a simple relationship between the thickness (decompacted) between the two markers (E), the interwell length (L), and the two successive sedimentary profile angles (a and b) is observed. In the second case with differential subsidence between the two wells (Figure 16B), the trigonometric calculation introduces the angle of tectonic rotation (c). These oversimplified calculations, which apply only to ramp profiles, indicate clearly the possibility to assess the validity of well correlations on the basis of sedimentary profiles and paleobathymetric estimation.

EXAMPLE OF OUTCROP SEQUENCE-STRATIGRAPHIC CORRELATION AND FACIES MODELING The following paragraph illustrates the impact of sequence-stratigraphic well correlation and well spacing on stochastic facies modeling on the basis of an outcrop example (Figures 18, 19). This sensitivity analysis is based on a Lower Cretaceous outcrop example studied in the south of France and represents typical marine carbonate platform systems. At the scale of the studied outcrop (4 km2, 1.5 mi2), the internal architecture of this 25-m (82-ft)-thick sedimentary system corresponds to three correlated shallowing upward sequences. It is dominated by grainstone bodies (submarine dunes) that are generally connected and wackestone bodies, formed below wave base, that are not connected as a result of the erosive nature of the grainstones. This sedimentary system can be represented by the deterministic correlation of 14 pseudowells resulting in a well spacing ranging between 68 and 648 m (223 and 2126 ft) with an average of 243 m (797 ft). The body dimensions exceed the well spacing at any location, but the mudstone bodies (300700 m, 984 2296 ft) are characterized by overall smaller dimensions than the grainstone bodies (0.52 km, 0.31.2 mi) and a lower net-to-gross ratio (030%). No or limited stratigraphic interpolation was involved in the representation of this sequence.
Borgomano et al. 815

Figure 18. Impacts of different stratigraphic correlation rules on facies interpolation between two wells. The gray shades in the two logs represent the three main carbonate facies. The workflow includes four main steps (1 4): the first step represents the higher level of stratigraphical correlation that could be based on seismic reflectors. The second step illustrates the two possible correlation scenarios based, respectively, on lithostratigraphy and sequence stratigraphy. The third step corresponds to the restoration of the structural deformations whereas the fourth step displays the facies interpolation within the correlated units. Lithostratigraphy forces the lateral correlation of facies between the two wells and the realization of layered reservoir architectures with a strong polarization of the spatial correlation of the properties. Sequence stratigraphy allows lateral facies variations (sharp or gradual) within correlated units and the realization of more complex and heterogeneous reservoir architectures. 816 Well Correlations in Carbonates

Facies units and sequence boundaries were walkedout on the outcrop. Facies are distributed according to a sequence-stratigraphic framework that results in the development of pinch-outs and sharp changes between facies. Gradual changes, for example, between wackestones and grainstones, have not been observed. The 14 pseudowells, the facies, and the sequence correlations were imported in a numerical geomodeler. The main objective of the sensitivity analysis was to test the impact of the sequencestratigraphic correlations and of the well spacing on the stochastic facies modeling. A standard pixelbased stochastic method, founded on facies proportion curve and variogram (Ravenne, 2002), was used with or without constraints on object dimensions and shapes. The workflow in Figure 18 illustrates the differential impact of sequence-stratigraphic correlation and lithostratigraphic correlation on stochastic facies modeling. Lithostratigraphic correlation implies the spatial correlation and the lateral continuity of facies between the wells, whereas sequencestratigraphic correlation allows gradual or sharp lateral facies transition between the wells. In this case, lithostratigraphic correlation is intrinsically not compatible with stochastic facies infilling between wells. It can be applied if the overall spatial heterogeneity of the sedimentary system is sampled by the wells (Figure 19A). When this is not the case (Figures 13, 14), applying a high-resolution sequence-stratigraphic correlation is recommended. The correlation of the three sequences between the 14 wells results in a stochastic facies model very close to the reality (Figure 19B). The high precision of the proportion curves and variograms, which are established for each sequence, is controlled by high well numbers and small well spacing. Well spacing and well numbers are nevertheless insufficient on their own to guaranty a realistic model without the correlation of the three sequences (Figure 19C). It just confirms that the deterministic correlation of the three sequences improves, for this particular data set, the quality of the proportion curves and variograms. Despite the correlation of the three sequences, the great spacing between the wells would result in a nonrealistic facies

model unless facies body dimensions are introduced in the modeling workflow (Figure 19D). In a subsurface case, such knowledge would come from reservoir or outcrop analogs. High net-to-gross and thick facies (grainstone) are less sensitive to well spacing and sequence correlations than the low netto-gross and thin facies (Figure 19B, C, D). These thin objects, which could act as a flow barrier, must be deterministically correlated between the wells by a lithostratigraphic method, which shows that sequence-stratigraphic and lithostratigraphic correlations can be complementary in the same reservoir modeling workflow.

EXAMPLE OF SUBSURFACE WELL CORRELATIONS (MALAMPAYA GAS FIELD, TERTIARY, PHILIPPINES) The Malampaya oil and gas accumulation is located in the deep-water Block Service Contract 38, offshore Palawan (Philippines), at a depth of 3000 m (9840 ft) below present sea level, within a northeastsouthwestoriented carbonate buildup. Like several hydrocarbon accumulations in the north Palawan block, the Malampaya field is situated within the upper Eocene to lower Miocene Nido Limestone (Sales et al., 1997; Williams, 1997). Various models of carbonate buildup growth history were proposed (Grotsch and Mercadier, 1999; Fournier et al., 2005) using 3-D seismic data, well logs, cores, and sidewall samples. Two levels of well correlations were defined within the Nido Limestone: (1) sequence-stratigraphic well correlations individualizing time-correlatable units and (2) reservoir unit well correlations defining rock volumes characterized by specific diagenetic patterns and petrophysical properties. Figure 20 presents the overall workflow of the well correlation as well as the spatial relationship between reservoir units and depositional sequences. Sequence-Stratigraphic Well Correlations Decameter-scale depositional sequences have been identified in wells, within the inner shelf, on the basis of (1) the vertical evolution of depositional
Borgomano et al. 817

818

Well Correlations in Carbonates

facies, (2) the vertical evolution of meteoric diagenetic features, and (3) the presence of major unconformities associated with a significant time hiatus (Fournier et al., 2005). The sequences and bounding unconformities have been correlated between wells using (1) the chemiostratigraphic (Sr isotopes) and biostratigraphic constraints (large benthic foraminifers), (2) the well-to-seismic data tie, and (3) the sequence stacking pattern. Inner shelf sequences are mainly composed of mud-rich shallow water deposits (mainly wackestone to packstone textures). The upper part of the sequences is always characterized by a strong meteoric overprint related to repeated phases of subaerial exposures. Reservoir Unit and Seismodiagenetic Unit Well Correlations On the basis of cores and well logs, the Nido Limestone was subdivided into 2070-m (66231-ft)thick units characterized by specific pore-type diagenetic patterns and petrophysical properties (= reservoir units). The alternation between tight and porous units is mainly controlled by meteoric diagenesis (leaching and pedogenesis) and by late burial cementation and leaching (Fournier and Borgomano, 2007). The correlation between wells of these reservoir units rests on a detailed well-to-seismic data tie using synthetic seismograms. Figure 20 shows the spatial relationship between rock properties, pore type, and reservoir units. Seismodiagenetic units (sensu Fournier and Borgomano, 2007) are carbonate rock volumes whose petrophysical properties (porosity, permeability, density, and sonic velocity) are dominantly controlled by diagenesis

and whose boundaries display acoustic impedance contrasts that are high enough to generate interpretable seismic reflectors. In 3-D seismic data, seismodiagenetic units whose thickness exceeds the vertical resolution of seismic data can be accurately related to reservoir units and can be delineated in 3-D by picking selected seismic horizons. In Malampaya, most of the tight burially cemented units can be imaged in 3-D. The resulting low-resolution petrophysical model can be used for guiding the further porosity inversion of the seismic amplitudes (Neuhaus et al., 2004). Relationship between Depositional Sequences and Reservoir Units Figure 6 shows the relationships between depositional sequences and reservoir units. In Malampaya, the tight units are formed by burial-related cemented lenses that progressively pinch out from the western to the eastern margin of the buildup. In contrast, the depositional architecture displays an aggrading pattern and consists of a stack of sequences of constant thickness that are bounded by flat unconformities. As a consequence, the boundaries of the reservoir units are not necessarily parallel to depositional surfaces and to sequence boundaries. Impact of Well Spacing on Well Correlations Previous studies (Fournier et al., 2004, 2005) have shown that despite the significant effects of synsedimentary tectonic deformation, the Malampaya buildup developed mainly as an aggrading flattopped shelf, with very low lateral variations of

Figure 19. Well spacing and correlation sensitivity analyses for facies simulation of an outcropping carbonate shelf. (A) The reality of the stratigraphic architecture (mapped from the outcrop) is obtained by the deterministic correlation of the two main facies within three correlated shallowing-upward sequences between 14 pseudowells. (B) This simulation is realized with 14 wells and the three correlated sequences. In this case, the facies are not correlated. The stochastic facies infilling, based on facies proportion curves and variogram, is closer to the reality because of the well density and spacing relative to the object dimensions (minimum well spacing < minimum body dimensions) combined to the sequence correlations that improve the quality of the proportion curves and variograms. This model could be improved by introducing constrains on the object shapes and dimensions. (C) Same as B without the correlations of the three sequences. The high net-to-gross facies (grainstone) is not sensitive to the sequence correlations as opposed to the low net-to-gross facies (wackestone). The spatial correlation of the less abundant and thin wackestone beds must be deterministically imposed by the interpreter. (D) Same as B with only four pseudowells. In this case, probalistic discrete objects (coral wackestone for example) are introduced on the basis of the proportion curves and a priori knowledge. This compensates only partially for the well decimation. Borgomano et al. 819

820 Well Correlations in Carbonates Figure 20. Workflow of sequence-stratigraphic and seismic-based well correlation in the Malampaya carbonate reservoir (depths are in true vertical depth subsea, TVDSS). Steps 1a to 1d show the definition of a sequence-stratigraphic framework based on larger foram biostratigraphy (letter-scale classification), Sr-isotope stratigraphy (the step-like shape of the Sr curve indicates significant hiatuses), and vertical changes in depositional environment. The water depth trends are dominated by keep-up intervals, meaning that the carbonate sedimentary system is able to grow despite the increase in accommodation. Steps 2a to 2d present the process of seismic-based well correlation and seismodiagenetic unit definition, integrating diagenetic interpretation of cores, well logs, and petrophysical measurements, and well-to-seismic tie. Step 3 establishes the spatial relationship between depositional sequences and seismodiagenetic units, and therefore specifies the chronostratigraphic or diagenetic significance of correlation surfaces. TVDSS = true vertical depth subsea; Tf1 = Upper Burdigalian; Upper Te = Aquitanian-Lower Burdigalian; Lower Te = Chattian. (Fournier et al., 2005)

depositional facies. In such a layer-cake depositional system, the well spacing will not significantly influence the stratigraphic framework of the reservoir, within the shelf deposits. In addition, except for few cemented lenses (e.g., SM2-IIIb unit), most of the large-scale seismodiagenetic units (2070 m [66231 ft] thick) can be picked continuously using 3-D seismic. As a consequence, a higher density of wells is not expected to have a significant impact on the reservoir model at this scale, within the Malampaya shelf, but it can help to better characterize smaller scale heterogeneity. Further modeling of the build-up flanks, where depositional facies (and possibly petrophysical properties) are expected to change laterally within a short distance, could be more sensitive to well spacing. Specific diagenetic environments, such as the meteoric-marine mixing zone, combined with transitional shelf-to-slope environments of deposition could increase the lateral heterogeneity of the reservoir rocks and justify a much denser well spacing for realistic reservoir modeling. The combined use of detailed stratigraphic interpretations of well data and a synthetic seismogram to tie wells to seismic data allows one to answer the three following questions that are commonly encountered in reservoir case studies involving 3-D seismic and well data: (1) the spatial relationship between stratigraphic sequences, depositional bodies, seismodiagenetic units, and reservoir units; (2) the geological significance (chronostratigraphic versus diagenetic) of seismic-based correlations; and (3) the relevance of sequence-stratigraphic and biostratigraphic well correlations for the construction of reservoir models. Such an integrative approach linking sedimentology, diagenesis, petrophysics, and seismic data is therefore recommended for carbonate reservoirs that have undergone intense diagenetic alteration, in a meteoric and/or burial realm (Zampetti et al., 2005; Fournier and Borgomano, 2007).

stratigraphic method must be adapted to the goal of the reservoir model. The main objective of the well correlation process is to distinguish, at the relevant scales, the correlatable from the noncorrelatable heterogeneities in space and to create subsets of data in which stationary geostatistic methods can be applied for the reservoir modeling. The stratigraphic well correlations must therefore be guided by consistent geological and petrophysical concepts that are established on the basis of data analysis and a priori knowledge. The ultimate objective of the correlation is to capture in the model the correlatable petrophysical heterogeneities that matter for the definition of reservoir and flow units. The largest error is introduced when the stratigraphic rules force unrealistic spatial correlations of random noise sampled in the wells. Stratigraphic rules are chosen on the basis of the geological factors that control the distribution of the rock properties, the spatial heterogeneity of the reservoir, the well spacing, and the objectives of the reservoir model. If the average well spacing is less than the lateral dimensions of critical sedimentary objects, then lithostratigraphic rules can be applied. If, on the contrary, average well spacing is far beyond these object dimensions, then sequence-stratigraphic rules are more valid. Well correlation of the time series is, however, not sensitive to well spacing but it does not necessarily represent a valid correlation of rock properties. In the case of sequence-stratigraphic correlations of base profiles, several genetic assumptions can be made that allow the validation of the correlations based on simple trigonometry.

REFERENCES CITED
Adams, E. W., and W. Schlager, 2000, Basic type of submarine slope curvature: Journal of Sedimentary Research, v. 70, p. 814 828. Ainsworth, B. R., S. Montree, and S. T. C. Duivenvoorden, 1999, Correlation techniques, perforation strategies, and recovery factors: An integrated 3-D reservoir modeling study, Sirikit field, Thailand: AAPG Bulletin, v. 83, no. 10, p. 1535 1551. Aitken, J. F., and J. A. Howell, 1996, High resolution sequence stratigraphy: Innovations, applications and future prospects: Geological Society Special Publication 104, p. 1 9. Alsharhan, A. S., 1993, Asab field-United Arab Emirates,

CONCLUSIONS The process of stratigraphic well correlations is very critical for carbonate reservoir modeling and the

Borgomano et al.

821

Rub Al Khali Basin, Abu Dhabi, in N. H. Foster and E. A. Beaumont, eds., Structural traps VIII: AAPG Treatise of Petroleum Geology, Atlas of Oil and Gas Fields, p. 6997. Amaefule, J. O., M. Attunbay, D. Tiab, D. G. Kersey, and D. K. Keelan, 1993, Enhanced reservoir description: Using core and log data to identify hydraulic (flow) units and predict permeability in uncored intervals/wells: Society of Petroleum Engineers, SPE Paper 26436, 16 p. Anselmetti, F. S., and G. P. Eberli, 1993, Controls of sonic velocity in carbonates: Pure and Applied Geophysics, v. 141, p. 287 323. Araktingi, U. G., and W. M. Bashore, 1992, Effects of properties in seismic data on reservoir characterization and consequent fluid flow predictions when integrated with well logs: Society of Petroleum Engineers, SPE Paper 24752, 14 p. Araktingi, V. G., and F. M. Orr Jr., 1993, Viscous fingering in heterogeneous porous media: SPE Paper 18095, Society of Petroleum Geology Advanced Technology Series, v. 1, p. 71 80. Armstrong, M., 1984, Common problems seen in variograms: Mathematical Geology, v. 16, no. 3, p. 305313. Asquith, G. B., 1985, Handbook of log evaluation techniques for carbonate reservoirs: AAPG Methods in Exploration Series 5, 47 p. Bashore, W. M., U. G. Araktingi, M. Levy, and W. J. Schweller 1994, Importance of a geological framework and seismic data integration for reservoir modeling and subsequent fluid-flow predictions, in J. M. Yarus and R. L. Chambers, eds., Stochastic modeling and geostatistics: Principles, Methods, and Case Studies: AAPG Computer Applications in Geology 3, p. 159 175. Boote David, R. D., and D. Mou, 2003, Safah field, Oman: Retrospective of a new concept exploration play, 1980 2000: GeoArabia, v. 8, no. 3, p. 367 430. Borgomano, J. R. F., 2000, The Upper Cretaceous carbonates of the Gargano-Murge region (southern Italy): A model of platform-to-basin transition: AAPG Bulletin, v. 84, no. 10, p. 1561 1588. Borgomano, J. R. F., J. H. van Konijnenburg, and J.-C. Jauffred, 2001, Anatomy of carbonate bodies for hydrocarbon reservoir modeling: Applications and future developments: Ge ologie Me diterrane enne, v. 28, no. 12, p. 2326. Borgomano, J., J.-P. Masse, and S. Al Maskiry, 2002, The lower Aptian Shuaiba carbonate outcrops in Jebel Akhdar, northern Oman: Impact on static modeling of Shuaiba petroleum reservoirs: AAPG Bulletin, v. 86, no. 9, p. 1513 1529. Budd, D. A., A. H. Saller, and P. M. Harris, eds., 1995, Unconformities and porosity in carbonate strata: AAPG Memoir 63, 313 p. Carle, S. F., and G. E. Fogg, 1996, Transition probabilitybased indicator geostatistics: Mathematical Geology, v. 28, no. 4, p. 453476. Carle, S. F., and G. E. Fogg, 1997, Modeling spatial variability with one and multi-dimensional continuous Markov chains: Mathematical Geology, v. 29, no. 7, p. 891 917. Carmalt, S. W., and B. St. John, 1986, Giant oil and gas fields, in M. T. Halbouty, ed., Future petroleum province of the world: AAPG Memoir 40, p. 11 53.

Chilingar, G. V., R. W. Mannon, and H. H. Rieke, eds., 1972, Oil and gas production from carbonate rocks: New York and Amsterdam, Elsevier, 408 p. Chilingarian, G. V., S. J. Mazzullo, and H. H. Rieke, eds., 1992, Carbonate reservoir characterization A geologicengineering analysis: Amsterdam, Elsevier, Developments in Petroleum Science, v. 30, 639 p. Coates, G. R., L. Xiao, and M. G. Prammer, 1999, NMR logging principles and applications: Halliburton Energy Services, Houston, p. 253. Coulibaly, P., and C. K. Baldwin, 2005, Nonstationary hydrological time series forecasting using nonlinear dynamic methods: Journal of Hydrology, v. 307, p. 164 174. Dake, L. P., 1978, Fundamentals of reservoir engineering: Amsterdam, Elsevier, Developments in Petroleum Science, v. 8, p. 443. Deutsch, C. V., 2002, Geostatistical reservoir modelling: New York, Oxford University Press, 376 p. Deutsch, C. V., and A. G. Journel, 1998, GSLIB: Geostatistical software library and users guide, applied geostatistics series, 2d ed.: New York, Oxford University Press, 384 p. Deutsch, C. V., and L. Wang, 1996, Hierarchical object-based geostatistical modeling of fluvial reservoirs: Society of Petroleum Engineers, Annual Technical Conference and Exhibition, Denver, Colorado, October 1996, SPE Paper 36514, v. 1, p. 221 236. Droste, H., and M. Van Steenwinkel, 2004, Stratal geometries and patterns of platform carbonates: The Cretaceous of Oman, in seismic imaging of carbonate reservoirs and systems: AAPG Memoir 81, p. 185 206. Dubrule, O., 1993, Introducing more geology in stochastic reservoir modelling, in A. Soares, ed., Geostatistics Troia 92: Dordrecht, Kluwer Academic, p. 351 369. Ebanks Jr., W. J., 1987, Flow unit concept-integrated approach to reservoir description for engineering projects (abs.): AAPG Bulletin, v. 71, no. 5, p. 551 552. Eberli, G. P., J. L. Masaferro, and J. F. Rick Sarg, 2004, Seismic imaging of carbonate reservoirs and systems, in G. P. Eberli, J. L. Massaferro, and J. F. Sarg, eds., Seismic imaging of carbonate reservoirs and systems: AAPG Memoir 81, p. 1 9. Ehrenberg, S. N., and P. H. Nadeau, 2005, Sandstone vs. carbonate petroleum reservoirs: A global perspective on porosity-depth and porosity-permeability relationships: AAPG Bulletin, v. 89, no. 4, p. 435 445. Fournier, F., and J. Borgomano, 2007, Geological significance of seismic reflections and imaging of the reservoir architecture in the Malampaya gas field (Philippines): AAPG Bulletin, v. 91, no. 2, p. 235 258. Fournier, F., L. F. Montaggioni, and J. Borgomano, 2004, Paleo-environments and high-frequency cyclicity in the Cenozoic south-east Asian shallow water carbonates: A case study from the Oligo-Miocene build ups of Malampaya (offshore Palawan, Philippines): Marine and Petroleum Geology, v. 21, p. 1 22. Fournier, F., J. Borgomano, and L. F. Montaggioni, 2005, Development patterns and controlling factors of Tertiary carbonate buildups: Insights from high-resolution

822

Well Correlations in Carbonates

3D seismic and well data in the Malampaya gas field (offshore Palawan, Philippines): Sedimentary Geology, v. 175, p. 189215. Goovaerts, P., 1997, Geostatistics for natural resources evaluation: Applied Geostatistics Series: New York, Oxford University Press, 496 p. Gringarten, E., and C. V. Deutsch, 2001, Teachers aid variogram interpretation and modeling: Mathematical Geology, v. 33, no. 4, p. 507 534. Grotsch, J., and C. Mercadier, 1999, Integrated 3-D reservoir modeling based on 3-D seismic: The Tertiary Malampaya and Camago buildups, offshore Palawan, Philippines: AAPG Bulletin, v. 83, p. 1703 1727. Homewood, P. W., and G. Eberli, 2000, Genetic stratigraphy on the exploration and production scales: Bulletin des Centres de Recherches Exploration-Production, ElfAquitaine. Memoir 24, p. 290. Homewood, P., F. Guillocheau, R. Eschard, and T. A. Cross, 1992, Corre lations haute re solution et stratigraphie ge ne tique: Une de marche inte e: Bulletin des Centres de gre Recherches Exploration-Production Elf-Aquitaine, v. 16, p. 357 381. James, N. P., and P. W. Choquette, eds., 1988, Paleokarst: New York, Springer-Verlag, 416 p. Jardine, D., and J. W. Wilshart, 1987, Carbonate reservoir description, in R. W. Tillman and K. J. Weber, eds., Reservoir sedimentology: SEPM Special Publication 40, p. 129 152. Jennings, J., 2000, Spatial statistics of permeability data from carbonate outcrops of west Texas and New Mexico: Implications for improved reservoir modeling: Austin, Texas, Bureau of Economic Geology, Report of Investigations, no. 258, 50 p. Kenyon, W. E., 1997, Petrophysical principles of applications of NMR logging: The Log Analyst, v. 38, no. 2, p. 21 43. Kerans, C., and S. Tinker, 1997, Sequence stratigraphy and characterization of carbonate reservoirs: SEPM Short Course, v. 40, 130 p. Koltermann, C. E., and S. M. Gorelick, 1996, Heterogeneity in sedimentary deposits: A review of structure-imitating, process-imitating, and descriptive approaches: Water Resources Research, v. 32, p. 2617 2658. Lucia, F. J., 1995, Rock-fabric/petrophysical classification of carbonate pore space for reservoir characterization: AAPG Bulletin, v. 79, no. 9, p. 12751300. Lucia, J. F., 1999, Carbonate reservoir characterization: New York, Springer-Verlag, 225 p. Mallet, J.-L., 2002, Geomodeling: New York, Oxford University Press, 599 p. Massonat, G., 2001, Stochastic modelling of sedimentary facies constrained by paleobathymetry: The Neptune approach: Geologie Mediterraneenne, v. 18, no. 1 2, p. 121 126. Matheron, G., 1989, Estimating and choosing: Berlin, SpringerVerlag, 141 p. Melville, P., O. Al Jeelani, S. Al Menhali, and J. Grotsch, 2004, Three-dimensional seismic analysis in the characterization of a giant carbonate field, onshore Abu Dhabi, United Arab Emirates, in G. P. Eberli, J. L. Massafero,

and J. F. Sarg, eds., Seismic imaging of carbonate reservoirs and systems: AAPG Memoir 81, p. 123 148. Moore, C. H., 2001, Carbonate reservoirs: Porosity evolution and diagenesis in a sequence-stratigraphic framework: Amsterdam, Elsevier, Developments in Sedimentology, v. 55, 444 p. Morettini, E., et al., 2005, Combining high-resolution sequence stratigraphy and mechanical stratigraphy for improved reservoir characterisation in the Fahud field of Oman: GeoArabia, v. 10, no. 3, p. 17 44. Munn, D., and A. F. Jubralla, 1987, Reservoir geological modelling of the Arab D reservoir in the Bul Hanine field, offshore Qatar Approach and results: Proceedings of the 5th SPE Middle East Oil Show, Bahrain, SPE Paper No. 15699, p. 109 120. Neuhaus, D., J. Borgomano, J.-C. Jauffred, C. Mercadier, S. Olotu, and J. Grotsch, 2004, Quantitative seismic reservoir characterization of an Oligocene Miocene carbonate buildup: Malampaya field, Philippines, in G. P. Eberli, J. L. Massafero, and J. F. Sarg, eds., Seismic imaging of carbonate reservoirs and systems: AAPG Memoir 81, p. 169183. Omre, H., and K. B. Halvorsen, 1989, The Bayesian bridge between simple and universal kriging: Mathematical Geology, v. 21, p.767 786. Pomar, L., 1993, High-resolution sequence stratigraphy in prograding Miocene carbonates: Application to seismic interpretation, in R. G. Loucks and J. F. Sarg, eds., Carbonate sequence stratigraphy: AAPG Memoir 57, p. 389 407. Posamentier, H. W., M. T. Jervey, and P. R. Vail, 1988, Eustatic control on clastic deposition I Conceptual framework, in C. K. Wilgus, B. S. Hastings, C. G. St. C. Kendall, H. W. Posamentier, C. A. Ross, and J. C. Van Wagoner, eds., Sea-level changes: An integrated approach: SEPM Special Publication 42, p. 109 124. Pranter, M. J., N. F. Hurley, and T. L. Davis, 2004, Sequence-stratigraphic, petrophysical, and multicomponent seismic analysis of a shelf-margin reservoir: San Andres Formation (Permian), Vacuum field, New Mexico, United States, in G. P. Eberli, J. L. Massafero, and J. F. Sarg, eds., Seismic imaging of carbonate reservoirs and systems: AAPG Memoir 81, p. 59 89. Prokophi, A., and F. Barthelmes, 1996, Detection of nonstationarities in geological time series: Wavelet transform of chaotic and cyclic sequences: Computers and Geosciences, v. 22, no. 10, p. 1097 1108. Quirk, D. G., 1996, Base profile: A unifying concept in alluvial sequence stratigraphy, in J. A. Howell and J. F. Aitken, eds., High resolution sequence stratigraphy: Innovations and applications: Geological Society Special Publication 104, p. 37 49. Ravenne, C., 2002, Stratigraphy and oil: A review: Part 2. Characterization of reservoirs and sequence stratigraphy: Quantification and modeling: Oil & Gas Science and Technology: Revue de lInstitut Francais du Pe trole, v. 57, no. 4, p. 311 340. Read, J. F., 1985, Carbonate platform facies models: AAPG Bulletin, v. 69, p. 1 21. Robin, C., D. Rouby, D. Granjeon, F. Guillocheau, P.

Borgomano et al.

823

Allemand, and S. Raillard, 2005, Expression and modelling of stratigraphic sequence distortion: Sedimentary Geology, v. 178, p. 159186. Roehl, P. O., and P. W. Choquette, eds., 1985, Carbonate petroleum reservoirs: New York, Springer-Verlag, 622 p. Sales, A. O., E. C. Jacobsen, A. A. Morado, J. J. Benavidez, F. A. Navarro, and A. E. Lim, 1997, The petroleum potential of deep-water northwest Palawan Block GSEC 66: Journal of Asian Earth Sciences, v. 15, p. 217 240. Sarg, J. F., 1988, Carbonate sequence stratigraphy, in C. K. Wilgus, B. S. Hastings, C. G. St. C. Kendall, H. W. Posamentier, C. A. Ross, and J. C. Van Wagoner, eds., Sea level changes An integrated approach: SEPM Special Publication 42, p. 155 181. Schlager, W., 1999, Sequence stratigraphy of carbonate rocks: The Leading Edge, August 1999, v. 18, p. 901 907. Schlumberger, 1972, Log interpretation Principles: New York, Schlumberger, v. 2, 116 p. Schlumberger, 1974, Log interpretation Applications: New York, Schlumberger, v. 2, 116 p. Scholle, P. A., D. G. Bebout, and C. H. Moore, eds., 1983, Carbonate depositional environments: AAPG Memoir 33, 708 p. Schwab, A. M., P. Homewood, F. S. P. van Buchem, and P. Razin, 2005, Seismic forward model of a Natih formation outcrop The Adam Foothills transect (northern Oman): GeoArabia, v. 10, no. 1, p. 17 44. Smith, L. B., G. Eberli, J. L. Masaferro, and S. Al-Dhahab, 2003, Discrimination of effective from ineffective porosity in heterogeneous Cretaceous carbonates, Al Ghubar field, Oman: AAPG Bulletin, v. 87, no. 9, p. 1509 1529. Sonnenfeld, M. D., and T. A. Cross, 1993, Volumetric partitioning and facies differentiation within the Permian upper San Andres Formation of Last Chance Canyon, Guadalupe Mountains, New Mexico, in R. G. Loucks and J. F. Sarg, eds., Carbonate sequence stratigraphy Recent developments and applications: AAPG Memoir 57, p. 435 474. Stearns, D. W., and M. Friedman, 1972, Reservoirs in fractured rocks, in R. E. Kings, ed., Stratigraphic oil and gas fields Classification, exploration methods and case histories: AAPG Memoir 16, p. 82106. Suro-Perez, V., and A. G. Journel, 1990, Stochastic simulation of lithofacies: An improved sequential indicator approach, in D. Guerillot and O. Guillon, eds.: Proceedings of the 2nd European Conference on Math of Oil Recovery: Paris, Technip, p. 3 10.

Tinker, S. W., D. H. Caldwell, D. M. Cox, L. C. Zahm, and L. Brinton, 2004, Integrated reservoir characterization of a carbonate ramp reservoir, South Dagger Draw field, New Mexico: Seismic data are only part of the story, in G. P. Eberli, J. L. Massafero, and J. F. Sarg, eds., Seismic imaging of carbonate reservoirs and systems: AAPG Memoir 81, p. 91 105. Vail, P. R., R. G. Todd, and J. B. Sangree, 1977, Seismic stratigraphy and global changes of sea level: Part 5 Chronostratigraphic significance of seismic reflections, in C. E. Payton, ed., Seismic stratigraphy applications to hydrocarbon exploration: AAPG Memoir 26, p. 99 116. Van Buchem, F. S. P., P. Razin, P. Homewood, J. M. Philip, W. H. Oterdoom, and J. Philip, 2002, Stratigraphic organization of carbonate ramps and organic-rich intrashelf basins: Natih Formation (middle Cretaceous) of northern Oman: AAPG Bulletin, v. 86, no. 1, p. 21 53. Van Wagoner, J. C., R. M. Mitchum, K. M. Campion, and V. D. Rahmanian, 1988, Siliciclastic sequence stratigraphy in well logs, cores, and outcrops: AAPG Methods in Exploration Series 7, 55 p. Viseur, S., 2001, Simulation stochastique base e-objet de chenaux: Doctorat de the Institut National Polytech`se, nique de Lorraine, Nancy, 250 p. Weissmann, G. S., and G. E. Fogg, 1999, Multi-scale alluvial fan heterogeneity modeled with transition probability geostatistics in a sequence stratigraphic framework: Journal of Hydrology, v. 226, p. 48 65. Williams, H. H., 1997, Play concepts Northwest Palawan, Philippines: Journal of Asian Earth Sciences, v. 15, no. 2 3, p. 251273. Wilson, J. L., 1975, Carbonate facies in geologic history: New York, Springer-Verlag, 471 p. Xu, W., T. T. Tran, R. M. Srivastava, and A. G. Journel, 1992, Integrating seismic data in reservoir modeling: The collocated cokriging alternative: Society of Petroleum Engineers, SPE Paper 24742, 10 p. Yao, T., and A. Journel, 2000, Integrating seismic attribute maps and well logs for porosity modeling in a west Texas carbonate reservoir: Addressing the scale and precision problem: Journal of Petroleum Science and Engineering, v. 28, p. 65 79. Zampetti, V., U. Sattler, and H. Braaksma, 2005, Well log and seismic character of Liuhua 11-1 field, South China Sea, relationship between diagenesis and seismic reflections: Sedimentary Geology, v. 175, p. 217 236.

824

Well Correlations in Carbonates

Das könnte Ihnen auch gefallen