Shared Earth Modeling: Methodologies for Integrated Reservoir Simulations

interesting.

PREFACE

The primary goal of Shared Earth Modeling is to introduce the reader to the processes and concepts needed to develop shared earth models. A Shared Earth Model (SEM) is a multidisciplinary representation of a subsurface resource. Shared Earth Modeling describes the process of integrating static and dynamic data from two or more disciplines to construct a model of a subsurface resource. Reservoir structure and fluid production are examples of static and dynamic data, respectively. The SEM evolves as static and dynamic data are acquired and integrated into the model.

This book is suitable for geoscientists and engineers with an interest in shared earth modeling, and professionals working in multidisciplinary asset management teams. It can be used in multidisciplinary upper division undergraduate and graduate classes, continuing education classes, distance learning (remote or Internet-based) courses, short courses, or self-study. The reader is exposed to topics from several disciplines that have a significant impact on the development of a SEM.

My colleagues in industry and academia, and students in multidisciplinary classes at the Colorado School of Mines, helped me identify important and relevant topics that cross disciplinary lines. I am, of course, responsible for the final selection of topics. I would like to thank Drs. Matt Cole, Craig Van Kirk and Mark Miller for their support and for Mark’s contributions to Chapter 2. I would especially like to thank Kathy Fanchi for her efforts in the preparation of this manuscript.

John R. Fanchi, Ph.D.,     Golden, CO

June 2002

Chapter 1

Introduction to Shared Earth Modeling

Shared earth modeling is an essential aspect of the reservoir management process. Properly constructed shared earth models integrate all available information to yield the most accurate representation of a reservoir possible. The shared earth model is a quantitative representation of a subsurface and can be used to determine the size of the resource and to predict fluid flow performance. Our goal is to introduce the process and concepts needed to develop shared earth models using language that is understandable from a multidisciplinary perspective. This chapter introduces the notions of reservoir scales and reservoir heterogeneity, and provides a working definition of shared earth modeling and reservoir characterization. The value of shared earth modeling is then discussed.

1.1 Porous Media

Reservoirs are examples of porous media. They consist of volumes of rock grains and pore space. Figure 1-1 illustrates a cube of rock with grains of sand filling the cube. Bulk volume is the volume of the cube and includes both grain volume and the volume of space, or pore volume. Bulk volume VB of the porous medium is the product of area A times gross thickness H:

Figure 1-1 Pores and Pore Space

(1.1.1)

The volume that is not occupied by grains of sand is the pore space available for occupation by fluids such as oil, gas, and water.

Porosity is defined as the ratio of pore volume to bulk volume. Pore volume VP is the volume remaining when the volume of grains is subtracted from the bulk volume, thus

(1.1.2)

where ϕ is total porosity, VB is bulk volume and VG is the volume of the grains. While there are several kinds of porosity, porosity can simply be thought of as the void space in a rock. Upon rearrangement, we see from Equation (1.1.2) that pore volume VP is the product of bulk volume and porosity ϕ:

(1.1.3)

When sedimentary rocks are being deposited, the pore space is filled with water. The pores of a rock can be filled at a later time with commercially important fluids ranging from potable water to oil and gas. As a rule, we are interested in void spaces that are connected with other void spaces. Connected pore spaces form a conduit for fluid flow. Permeability is a measure of the connectivity of pore spaces.

The above definitions of bulk volume and pore volume are measures of the gross volume in a system. To determine the volume of the system that is commercially significant, the gross volume must be adjusted by introducing the concept of net thickness.

Net thickness h is the thickness of the commercially significant formation. For example, if the gross thickness includes 5 ft of impermeable shale and 15 ft of permeable sandstone, the gross thickness is 20 ft and the net thickness is 15 ft. If all of the permeable sandstone is not in communication with a production well, then the net thickness can be reduced further. It is often necessary to justify the well log interval that is identified as net thickness.

The net to gross ratio η is the ratio of net thickness h to gross thickness H:

(1.1.4)

The inequality highlights the fact that net thickness is always less than or equal to gross thickness. Net and gross thicknesses are ordinarily determined using well logs. The volume of net pay is the product of pore volume and net to gross ratio:

(1.1.5)

The saturation S . The volume V in the pay interval is the product of net pay volume and phase saturation:

(1.1.6)

The sum of the saturations in the pay interval must equal one. If the system has N phases, the saturation constraint is

(1.1.7)

For an oil-water-gas system, the saturation constraint is So + Sw + Sg = 1 where the subscripts {o, w, g} refer to oil, water and gas respectively.

When fluids are produced from a reservoir, they travel through the reservoir rock from a place of higher energy to a place of lower energy. The wellbore of a producing well is a point of lower energy. The route that the fluid takes can be a straight path, or very circuitous. The property of the rock that measures the length of the path from one point A to another point B relative to a straight line is called tortuosity. If the path is a straight line, which can occur in a fracture, the tortuosity is 1. In most cases, the flow path between points A and B will be longer than a straight line so that tortuosity is greater than 1. Figure 1-2 illustrates the concept of tortuosity.

Figure 1-2 Tortuosity

1.2 Reservoir Heterogeneity and Reservoir Scales

The description of reservoir heterogeneity depends on how the values of the property of interest vary spatially in the reservoir, and the size of samples used to measure the rock property. Every measurement of any rock property is made on a sample of some specific size. For example, porosity may be measured using core plugs of various sizes. The value of the rock property depends on the number of measurements made, which depends on the number of samples. Thus the accuracy of a characterization of the reservoir depends on the number of samples. Given these concepts, we can think of a reservoir as a collection of elements. Each element has the volume of one of our samples (Figure 1-3).

Figure 1-3 Ensemble of Volume Elements

Rock property values have meaning only for samples of rock of some macroscopic size which contain large numbers of pores and grains. These values will not be unique because of random events in the depositional process and subsequent diagenetic history. We can expect to find a statistical distribution of values associated with measurements of any rock property on many neighboring samples within a given depositional structure. A depositional structure can be defined as homogeneous in a statistical sense if the same frequency distribution of a rock property is found throughout the structure. Given this definition of homogeneity, we define heterogeneity as the absence of homogeneity.

The determination of the heterogeneous quality of a depositional structure requires the comparison of frequency distributions of any rock property in spatially distinct regions of the depositional unit. Collins [1961] showed that every measured value of the rock property should be on a sample of the same size because the form of the frequency distribution is determined in large part by the sample size for each measurement. We can illustrate this concept by a special and relatively simple case. Suppose we measure porosity ϕ on samples as small as a sand grain. The porosity measurements will exhibit one of two values, zero (all grain) or one (all pore space). The observed frequency distribution is therefore bi-modal with peaks at ϕ = 0 and ϕ = 1. For larger samples, such as core plugs, the distribution changes from a discrete bi-modal distribution to a continuous Gaussian distribution. If we wish to be consistent when we compare frequency distributions, we should use measurements from samples of comparable size.

The depositional environment determines the statistical heterogeneity of rock when viewed on any size scale. A depositional unit might be statistically heterogeneous on the scale of core plugs but be statistically homogeneous on a larger scale. In this case, rock properties within the depositional unit could be treated on the larger scale as deterministic functions of spatial position {x, y, z}. As deterministic functions, the rock properties would be adequately represented by smooth functions and it would be unnecessary to introduce statistical fluctuations. A critical sample volume should exist at the larger scale that would be large enough to mask statistical fluctuations in values of the rock property among adjacent samples. We may still see regional trends, or deterministic variations in rock property values as a function of spatial position, but the observed values of the rock property could be contoured with reasonable accuracy at the larger scale. The resulting contour map would provide a deterministic representation of the rock property as a function of spatial position in the depositional unit.

The scale associated with data measurements depends on the type of measurement, the sampling technique, and size of the sample. For example, core and well log information sample a very small part of the reservoir. A seismic section expands the volume of the subsurface environment that is sampled, but the interpretation of seismic data is less precise. Surface seismic data can cover the entire region of interest, but surface seismic data are often viewed as soft data because surface seismic measurements are remote, indirect measurements. The reliability of surface seismic data can be improved when correlated with hard data such as core and well log measurements.

The importance of sample size in characterizing rock properties has been stressed by many authors. As early as 1961, Collins [1961] showed that variance in porosity distribution depended on sample size. Bear [1972] stressed the need to identify an appropriate averaging volume for assigning a macroscopic property to a porous medium. Haldorsen and Lake [1989] defined four conceptual scales to account for variations in the range of data applicability. Their scheme is schematically illustrated in Figure 1-4.

Figure 1-4 Reservoir Scales (after Haldorsen and Lake, 1989; reprinted by permission of the Society of Petroleum Engineers)

The giga scale includes information associated with geophysical measurement techniques. An example of a giga scale measurement is the imaging of reservoir architecture. Theories of regional characterization, such as plate tectonics, provide an intellectual framework within which giga scale measurement techniques, like seismic and satellite data, can be interpreted. The mega scale is comparable in size to the reservoir being characterized. Mega scale measurement techniques include well logging, well testing and 3-D seismic analysis. Data sampling at the level of core analysis and fluid property analysis provides macro scale information. Thin section analysis and measurements of grain size distribution provide pore size information at the micro scale. Fayers and Hewett [1992] have noted that scale definitions are not universally accepted, but can be used to illustrate the relative scale associated with reservoir property measurements. The shared earth model will integrate information from all scales.

1.3 Shared Earth Modeling Defined

A shared earth model is a multidisciplinary representation of the subsurface resource. The shared earth model is developed from static geological information such as reservoir structure, and dynamic information such as fluid production performance. A shared earth model may change as static and dynamic data are acquired and integrated into the model.

The correct relationship between data measured at two different scales may be difficult to determine. For example, permeability is often obtained from both pressure transient testing and routine core analysis. The respective permeabilities, however, may appear to be uncorrelated because they represent two different measurement scales. The integration of data obtained at different scales is often referred to as the scale-up problem [for example, see Oreskes, et al., 1994].

An important task of the scale-up problem is to develop a detailed understanding of how measured parameters vary with scale. The focus on detail in one or more aspects of the reservoir modeling process can obscure the fundamental character of the reservoir in a flow model study. One way to integrate available data within the context of a big picture is to apply the concept of flow unit. Ebanks defined the flow unit as a volume of rock subdivided according to geological and petrophysical properties that influence the flow of fluids through it [Ebanks, 1987].

All rock properties measured for use in reservoir characterization are measured on samples of a specific size and scale. The measured values are usually averages over smaller elements of rock making up the sampled region. For example, core plug measurements are macro scale averages over micro scale elements, while pressure transient testing provides a mega scale average over both macro scale elements and, on a finer scale, micro scale elements. The four scales defined above are related to averaging volume in Table 1-1.

Table 1-1

Reservoir Characterization Scales

We can see how scale influences the analysis of a porous medium by considering a statistical description of porosity. Let us begin with a porous medium, such as a core plug, that has the bulk volume V. We subdivide the volume V into a number n so that

(1.3.1)

The volume and associated elements are shown in Figure 1. Each element is sampled to determine if it is either grain or pore space. This means the elements are small enough to be considered micro scale samples.

Figure 1-5 Elements of Volume

Let x be the number of elements that are pore space. Porosity is expressed in terms of x as

(1.3.2)

and pore volume is

(1.3.3)

The distribution of samples in this formulation is described by the binomial distribution since the sampling of each element yields one of two possible outcomes: either grain or pore space.

The probability that in n independent trials an element of pore space is encountered x times is given by the binomial distribution. Let p denote the probability that pore space is encountered in a single trial. Then the probability that grain is encountered is q = 1 − p. The resulting binomial distribution is

(1.3.4)

with the binomial coefficient

(1.3.5)

The DeMoivre-Laplace Central Limit theorem says that the binomial distribution becomes the normal distribution

(1.3.6)

when the number of elements n is large, which is reasonable, even for a core plug. The distribution in Equation (1.3.6) has mean μ = np and σ becomes

(1.3.7)

The effect of sample size on the distribution can be estimated using Equation (1.3.7) and the following procedure.

ab. A relationship between the number of samples in each volume and sample size is obtained by recognizing that bulk volume is constant, thus

(1.3.8)

A standard deviation for each distribution can be obtained from Equation (1.3.7). The ratio of standard deviations is

(1.3.9)

a b has a smaller standard deviation σa < σb, as shown in Figure 1-6. The distribution with larger samples is more sharply peaked about the average porosity than the distribution with smaller samples. Figure 1-6 is based on a micro scale analysis. The samples are assumed to be small enough at the micro scale that each sample is either a pore or a grain. The analysis would have to be modified if it was conducted on a different scale [for example, see Collins, 1961, Chapter 1].

Figure 1-6 Effect of Sample Size on Porosity Distribution at the Micro Scale [Fanchi, 2001a]

The above discussion of porosity illustrates a general observation: the value of a rock property is a function of the averaging volume. This observation is illustrated in Figure 1-7, which shows that the value of a rock property can vary within a scale and from one scale to another. An important task of reservoir characterization and associated reservoir flow modeling is to find a suitable scale for developing a flow model that can be used to prepare reliable performance predictions.

Figure 1-7 Rock Property as a Function of Averaging Volume

1.4 The Shared Earth Modeling Process

Shared earth modeling is the process of integrating static and dynamic data from two or more disciplines into a model of a subsurface resource. Static data are data acquired at a single point in time, such as reservoir structure from a seismic survey or a well log. Dynamic data are data that change with time, such as reservoir pressure and production performance. An essential element of the shared earth modeling process is reservoir characterization.

Reservoir characterization is the process of preparing a quantitative representation of a reservoir using data from a variety of sources and disciplines. All of the information collected at various scales in the reservoir characterization process must be integrated into a single, comprehensive, and consistent representation of the reservoir. The reservoir characterization process should include the acquisition and analysis of static and dynamic data. The model that results from the reservoir characterization process is a shared earth model.

Shared earth modeling may be viewed as a reservoir characterization process that integrates both static and dynamic data. Shared earth models are refined as additional data are acquired and incorporated into the reservoir description. In addition to representing our best understanding of the subsurface resource, shared earth models are used in reservoir flow simulators to predict production