SPE 124879 Rapid, Interactive Assessment of Petrophysical and Geometrical Effects on Density and Neutron Logs Acquired in Vertical

and Deviated Wells
Zoya Heidari, SPE, Carlos Torres-Verdín, SPE, Alberto Mendoza, SPE, Olabode Ijasan, SPE, and Ben Voss, The University of Texas at Austin; Bill Preeg, SPE, Consultant

Copyright 2009, Society of Petroleum Engineers This paper was prepared for presentation at the 2009 SPE Annual Technical Conference and Exhibition held in New Orleans, Louisiana, USA, 4–7 October 2009. This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract Borehole, geometrical, and petrophysical effects can significantly affect density and neutron logs. Tool location around the perimeter of the borehole, tool standoff, and wellbore deviation have a measurable effect on density and neutron logs, hence on the estimation of porosity and fluid density. It is difficult to diagnose and quantify these effects a priori without numerical modeling. The multiple-particle, radiation-transport Monte-Carlo code (MCNP) has traditionally been used by the logging industry to simulate borehole nuclear measurements acquired in complex rock formations. Despite its versatility and accuracy, MCNP is not numerically efficient for rapid simulation of nuclear logs and does not lend itself to interactive testing of multiple petrophysical/fluid hypotheses. We describe the successful application of a new method for nuclear-log simulation based on linear iterative refinement of nuclear sensitivity functions pre-calculated with MCNP. The procedure is fast, accurate, and efficient in most practical logging applications, including the simulation of nuclear logs acquired in invaded formations and in highly-deviated wells. Density and neutron logs acquired over a depth segment of 1000 ft can be accurately simulated within minutes of CPU time, compared to days with MCNP. Simulations are successfully verified against MCNP in a number of extreme cases of borehole, petrophysical, and fluid conditions wherein the error of the simulations does not exceed 2 porosity units. We implement the new simulation method to reproduce several field examples where logs are affected by presence of clay and invasion with water- and oil-base muds. Our rapid simulation procedure enables the interactivel quantification of the relative effect of clay, invasion, fluid density, and rock petrophysical properties on field logs. It also permits efficient integration with induction resistivity measurements for assessment of free and clay-bound water saturation, as well as for the assessment of residual hydrocarbon saturation. .Additional simulations of density and neutron logs are performed to quantify the influence of shale laminations on hydrocarbon-bearing and invaded thinly-bedded formations. Because of their longer radial length of investigation compared to density logs, neutron logs may exhibit false cross-over effects across thin beds.

Introduction Monte Carlo algorithms are commonly used to numerically simulate borehole nuclear measurements because of their flexibility to include complex rock formation properties and arbitrary radioactive sources and sensors. However, while Monte Carlo methods are ideal for source/sensor design and for appraisal of borehole environmental effects, they are not sufficiently fast to be used for routine petrophysical interpretation and/or for integration with other borehole measurements. The widespread use of Logging-While-Drilling (LWD) measurements, for instance, emphasizes the need of real-time petrophysical interpretations of borehole measurements acquired in complex rock and geometrical conditions. Petrophysical interpretations also require that nuclear measurements be readily integrated with resistivity measurements to yield more precise and less uncertain calculations of hydrocarbon reserves. Additionally, the petrophysical interpretation of thinlybedded rock formations requires that inversion techniques be used to reduce shoulder-bed effects on the measurements. In turn, inversion requires repeated numerical simulations of the measurements and it is therefore prohibitive with Monte Carlo methods.

the updated FSF is used to simulate a new value of porosity.. In addition.) and borehole properties (borehole size. etc. (2007). Interactive nuclear modeling is also necessary to quantify multiple hypotheses concerning rock quality and for prediction of measurements in complex multi-layer sequences. 2003) via the adjoint formulation of Boltzmann’s equation and enhanced with importance statistical sampling. energy bands. (Mendoza et al. we introduced a new simulation algorithm that approximates nuclear measurements acquired in vertical and deviated wells (Mendoza et al. the examples quantify the influence of invasion on borehole resistivity measurements for comparison purposes. Figure 1 shows the . industry-standard MCNP code (X-5 Monte Carlo Team MCNP. or FSFs) precomputed with the nuclear-radiation. 2009b). permeability. we benchmark the reliability. The procedure repeats itself until the simulated porosity matches the sensitivity function used for the simulation.. The principle behind the linear iterative approximation method is to simulate nuclear logs in non-homogeneous formations using multiple FSFs pre-computed in homogeneous formations (which also include the borehole). borehole fluid. porosity. 2009a). The second step consists of simulating the tool response (i.g. Borehole nuclear measurements acquired across non-homogeneous rock formations are approximated with local iterative (fixed-point) refinements of the FSFs until the final FSF is consistent with the simulated result (Mendoza et al. (Mendoza et al. The algorithm described by Mendoza et al. we briefly explain this method. We consider several field examples as well as synthetic examples based on field measurements. The fast iterative refinement method incorporates the 3D spatial capabilities of the neutron and density FSFs. the iterative FSF method is at least two orders of magnitude faster while delivering relative errors below 25% (Mendoza et al. A pre-computed library of FSFs typically includes a large set of rock formation properties and a fixed set of tool and borehole environmental conditions. 2009a and 2009b). including the challenging case of deviated wells (Mendoza et al. 2009) The first step in the linear iterative refinement is to assume a matrix type and an average porosity value to select an initial FSF for nuclear log simulation. 2009a) For the purposes of this paper. source-sensor separations. and relative permeability) and fluid properties (e.g.e. The variety of geometric factors contained in the library enables the selection of the most appropriate base-case FSFB for the simulation of nuclear borehole measurements across non-homogeneous formations. Compared to MCNP simulation results. (2009a) describe the basis of the method and numerical implementation. and accuracy conditions to be used routinely in well-log interpretation and inversion.. We also consider the important case of nuclear measurements acquired across thinly-bedded rock formations to quantify the relative effect of shoulder beds on both density and neutron logs.. accuracy.) while assuming (albeit not restrictively) a homogeneous and infinite rock formation. The examples consider cases of rock formations invaded with both water-base (WBM) and oil-base (OBM) mud to quantify the effect of invasion on both density and neutron logs.. Below.2 SPE 124879 Watson (1984) for the first time introduced the idea of using linear sensitivity functions to accelerate nuclear log simulation. density and viscosity) on the numerically simulated density and neutron logs. In a third step. (2009) is based on linear iterative refinements performed with spatial sensitivity functions (referred to as Flux Sensitivity Functions. 2009). mudcake thickness. Mendoza et al. and can be used in both vertical and deviated well geometries. Finally. and efficiency of the method with a synthetic model that includes invasion. speed. The linear iterative refinement method updates the FSF at each sample point based on the results of a previous pass... etc. Special emphasis is placed on the effect of both rock petrophysical properties (e. Simulations and comparisons against MCNP for extreme cases of rock formation properties and shoulder beds conclusively indicate that the approximate method to simulate density and neutron measurements meets the necessary stability. Our simulations confirm the need for fast and accurate simulation of nuclear logs for static and dynamic quantification of rock formation properties. mudcake weight. tool standoff. Linear iterative refinement method. detector count rate) which correlates to porosity under specific assumptions of fluid and matrix type. at each sample point a new FSF is calculated by linear interpolation between available functions corresponding to the closest neighbors of the simulated porosity. The objective of this paper is to document examples of application of the FSF-based simulation method for the interactive and quantitative interpretation of borehole nuclear measurements. the numerical simulation method for borehole nuclear measurements adopted in this paper is based on pre-computed FSFs and local iterative refinement. Convergence of the fixed-point iteration is usually achieved in no more than 4 iterations. capillary pressure. The specific tool and source/sensor configuration of density and neutron tools assumed in the numerical simulations corresponds to the “Longhorn” design described by Mendoza et al. the library of FSFs may incorporate several non-homogeneous base-case formations that include tool standoff and mudcake. Moreover. Numerical Simulation Method As indicated above. Quite recently. We have a pre-computed library of FSFs for a range of porosity-matrix-fluid mixtures in homogeneous formations. The FSFs are pre-calculated with MCNP by including all tool (geometrical and material properties.

simulated neutron and density porosity logs in a vertical well across horizontal layers of different petrophysical properties. absolute permeability. and electrical resistivity. three-phase fluid-flow equations of porous media. Table 5 summarizes additional properties of this formation. Iterative Algorithm to Simulate Well Logs and to Estimate Petrophysical Properties. and formation fluid and mud properties. These properties can be defined using core measurements or basic well-log interpretation. Fluid flow simulation is performed with CMG (Computer Modeling Group LTD). Available core data also decrease non-uniqueness of results by adjusting some of the unknown input properties. sensitivity of nuclear logs to petrophysical properties decreases in extreme cases when drilling mud is OBM in oil-bearing zones. It is based on initial log interpretation and petrophysical analysis.5 and 1. otherwise they are . resistivity logs improve the estimation. we use an iterative algorithm incorporating well-log interpretation and forward simulation of nuclear and resistivity logs. Table 5 summarizes the petrophysical properties assumed for this geometry. we update the petrophysical properties of the multi-layer model to decrease the difference between measured and numerically simulated well logs. The flowchart in Figure 2 describes the simulation method and the iterative algorithm used to estimate multi-layer petrophysical properties. The CPU time for the MCNP simulations is about 20 hours for each well log. Figure 1: Comparison of fast linear iterative refinement method (solid lines) and Monte Carlo (MCNP) simulations (dashed lines) of 3 neutron and density measurements.SPE 124879 3 corresponding model. panels show: formation bulk density model in [g/cm ]. Mud-Filtrate Invasion. Moreover. The second step includes simulating the process of mud-filtrate invasion to generate spatial distributions of fluid saturation from which we obtain spatial distributions of migration length. rock-fluid properties (relative permeability and capillary pressure curves). For simplicity. initial water saturation. An important application of such a fast and accurate modeling process is to improve the assessment of petrophysical properties. for OBM simulation. Subsequent sections briefly explain the process of mud-filtrate invasion and well-log simulation.2 water-filled sandstone porosity units in neutron and density measurements. The assumed model is based on the field example of North Louisiana’s tightgas sand. it takes 2 minutes to complete the simulation with the fast linear iterative refinement method. In both of the iput models for MCNP and the fast iterative refinement method. Considering nuclear and resistivity logs at the same time in the iterative process reduces non-uniqueness of the results. density. In these extreme cases. and the migration length (Lm) of the model used for the simulation of the neutron log. In the third step. The first step for interpretation consists of generating an initial guess of multi-layer petrophysical properties. The presented multi-layer example contains gas and the drilling mud is WBM. including environmental effects and mud-filtrate invasion. by comparing simulated to measured well logs. constructed from a field example discussed in a subsequent section. By comparison. respectively. formation fluid properties. Two-dimensional (2D) simulations are implemented in cylindrical coordinates to model invasion in vertical wells under the assumptions of cylindrical flow and permeability isotropy. The high accuracy and very low CPU time make the fast linear iterative refinement method an adequate option for forward and inverse simulations of nuclear logs in complex geometries. we use STARS® for WBM simulations and GEM®. Among the various CMG simulators available. The petrophysical model is the input for fluid-flow simulation. a commercial reservoir simulator capable of solving three-dimensional (3D). it consists of non-shale porosity. Finally. invasion parameters. From left to right. To estimate petrophysical properties. numerical simulations described below assume that measured logs have been corrected for borehole environmental effects. we numerically simulate the corresponding resistivity and nuclear logs. Figure 1 shows a good agreement between the simulation results of MCNP and the fast linear iterative refinement method. The maximum error is 0. we assume piston-like invasion. a compositional simulator.

which was drilled with WBM. Migration length is used for the simulation of neutron measurements. Heidari et al. we examine the effect of volumetric shale concentration. high concentration of shale causes a prominent separation of density and neutron logs.70. based on the new petrophysical properties after mud-filtrate invasion. we analyze a North Louisiana’s tight-gas sand. This zone consists of a large portion of residual hydrocarbon saturation causing cross-over between density and neutron logs even after deep mud-filtrate invasion. and resistivity for each simulation grid. Results and Discussions We describe four examples of the interactive assessment of petrophysical and geometrical effects on density and neutron logs. The third field example is another tight-gas sand zone. The fast linear iterative refinement method incorporates the spatial distributions of density and migration length to simulate density and neutron logs (Mendoza et al. The first three examples are challenging field cases that emphasize how modeling makes the estimation of petrophysical properties more accurate. φs is non-shale porosity. Figure 2: Iterative workflow adopted in this paper to estimate unknown petrophysical properties. Rates of mud-filtrate invasion are calculated using the one-dimensional (1D) radial method described by Salazar and Torres-Verdín (2009). where ρsh is shale density. Iterations are intended to progressively improve the agreement between measured and numerically simulated resistivity and nuclear well logs by making adjustments to layer-by-layer petrophysical properties. The reason is the deep mud-filtrate invasion in extremely salty connate water. In these two cases. To calculate migration length for each simulation grid.4 SPE 124879 calculated using the above-described iterative process which simultaneously honors resistivity and nuclear logs. Csh is volumetric concentration of shale. In this field example. From the simulation of mud-filtrate invasion. 1989). In order to simulate resistivity logs. residual hydrocarbon saturation. After obtaining the final multi-layer model based on the iterative approach. Chemical compositions for each grid and their corresponding volumetric concentrations are input to SNUPAR. we calculate the spatial distribution of formation resistivity using the dual-water model.to fine-grained sandstone and some fossiliferous oolitic limestone. The formation was drilled with WBM. In this field example. 2009a). and ρf is fluid density. Simulation of Well Logs. Although oil saturation is above 0. 1 Mark of Schlumberger . (2009) provide further technical details about the iterative approach of nuclear and resistivity simulations as well as of the simulation of resistivity logs. The new petrophysical properties update the values for density. Field Example No. which considers the interplay between mudcake buildup and rock petrophysical properties. Then we numerically simulate array-induction resistivity measurement (AIT1). we obtain new values for petrophysical properties such as water saturation and salt concentration. ρm is matrix density. we use Schlumberger’s commercial software. 1: North Louisiana’s Tight-Gas Sand. The sedimentary rock contains very fine. We calculate bulk density via the equation (1) ρ r = ρ shCsh + ρ m (1 − Csh − φs ) + ρ f φs . to obtain petrophysical properties based on nuclear and resistivity well-log simulation. we investigate the effect of mud-filtrate invasion and petrophysical parameters such as volumetric concentration of shale and residual hydrocarbon saturation on the simulated density and neutron logs. which was drilled with WBM. and second. Nuclear log simulation is expected to improve the assessment of hydrocarbon saturation. The second case is an oil-bearing zone. The fourth example is a multi-layer synthetic case. The objectives are first to investigate the effect of different petrophysical properties on the simulated nuclear logs. The first field example is a tight-gas sand zone. which was invaded with OBM. and mud-filtrate invasion on the simulated nuclear logs. it is not possible to estimate gas saturation based on conventional interpretation methods or using only resistivity logs.. neutron migration length. We simulate density and neutron logs for both a vertical an a high-angle well across the same multi-layer petrophysical model to investigate the effect of deviation angle on nuclear measurements. SNUPAR (McKeon and Scott.

Figure 3 shows the final simulation results. With the initial values listed in Table 4. In the next step. and rock-fluid properties reported for the tight-gas sand.SPE 124879 5 Texturally. The average difference between simulated and measured logs in the cleanest sands is 2% for neutron porosity.1 residual gas saturation in the numerical simulations. In the next step.. we iteratively adjust residual gas saturation and volumetric concentration of shale to match the cross-over between simulated and measured neutron and density logs. mud. Quantifying this separation with the proposed simulation method enables the estimation of residual gas saturation. This process yields 0. Table 5 summarizes the final petrophysical properties obtained at the end of the iterative process. which was impossible with resistivity logs alone. The effect of residual gas saturation is investigated in Figure 4. The time-constant average rate of mud-filtrate invasion is 0.0 to 0. It should be mentioned that combining nuclear logs with resistivity logs makes it possible to obtain accurate estimations of residual gas saturation. Table 3 describes the average values for porosity. In turn. We assume that there is no free water in the formation. These values are calculated by averaging the properties obtained from standard well-log interpretation. more accurate estimations for residual hydrocarbon saturation lead to improved estimations of other petrophysical properties. 3% for density. we iteratively adjust the remaining parameters based on the procedure described in Figure 2. mud-filtrate invasion. and neutron and density logs (right-hand panel) for the multi-layer model of North Louisiana’s tight-gas sand. (2006) based solely on the inversion of resistivity logs. fluid and formation properties. . Tables 1 and 2 summarize the Archie’s parameters.1-0. neutron and density logs exhibit a substantial crossover which is due to residual gas saturation. Neutron and density logs do not match the measured logs anymore. 2006) originated from fluvial deposition during early Cretaceous times. with their cross-over decreased from 8. Despite the fact that invasion is deep. matrix. We assumed 0. This decrease in cross-over is due to lower gas saturation in the invaded formation caused by a decrease of residual gas saturation. and volumetric concentration of shale on the simulated well logs. and 10% for resistivity. initial water saturation. Figure 3: Comparison of numerically simulated (solid lines) and measured (dashed lines) array-induction apparent resistivity logs (center panel). Time of mud-filtrate invasion (tinv) is five days. and permeability. Table 4 describes the petrophysical properties of the initial multi-layer model. There are no core data available to initialize residual gas saturation. the shallow measured and simulated resistivity logs are no longer in good agreement due to lower gas saturation in the radially shallow part of the reservoir. We perform the petrophysical analysis and associated numerical simulations in a 22 ft thick sand interval. The maximum permeability obtained with our interpretation method is 12% higher than the maximum permeability reported by Salazar et al. it is mature quartz arenite and subarkose sand (Salazar et al. The left-hand panel shows the spatial distribution (radial and vertical directions) of water saturation. The iterative simulation process begins with an initial guess obtained from the layer-by-layer averages of porosity. hence initial water saturation is equal to irreducible water saturation.93 [ft3/day/ft] for five days of invasion. Petrophysical properties were obtained from the matching of field logs with numerical simulations (Table 5). we adjust the petrophysical properties of the model based on the the values reported in Table 5 and investigate the effect of residual hydrocarbon saturation. Moreover. permeability. Large separation between resistivity logs with different radial lengths of investigation is an indication of deeply invaded beds (Figure 3). and volumetric concentration of shale.3 residual gas saturation for the tight-gas sand formation.5 sandstone porosity units.

The cross-over between neutron and density porosity logs in the sand zone increases by 5 sandstone porosity units compared to Figure 5a. The left-hand panel shows the spatial distribution (radial and vertical directions) of water saturation. we consider the effect of volumetric concentration of shale. Time of mud-filtrate invasion (tinv) is five days. Resistivity curves do not exhibit separation before invasion. (a) (b) Figure 5: Comparison of numerically simulated (solid lines) and measured (dashed lines) array-induction apparent resistivity logs (center panel). the effect of volumetric concentration of shale on the cross-over of nuclear logs is 50% of the initial cross-over. thus verifying that the separation of apparent resistivity curves is not due to post-processing artifacts. . We observe no separation between apparent resistivity curves. and neutron and density logs (right-hand panel) for the multi-layer model of North Louisiana’s tight-gas sand before invasion. Remaining layer petrophysical properties are equal to those of estimated values. Figure 4: Comparison of the numerically simulated (solid lines) and measured (dashed lines) array-induction apparent resistivity logs (center panel). Residual gas saturation is assumed equal to 0. the cross-over between neutron and density logs increases by 5 sandstone porosity units when we eliminate the effect of shale. This behavior is expected because of higher gas saturation in the pre-invaded formation.6 SPE 124879 Figure 5a shows the numerically simulated nuclear and resistivity logs before mud-filtrate invasion. (a) In this model. Volumetric concentration of shale is another parameter that serously affects neutron and density logs. In other words. Remaining layer petrophysical properties are equal to those of estimated values in Table 5. Figure 5b illustrates the numerically simulated nuclear and resistivity logs before mud-filtrate invasion where we neglect shale in the formation. Compared to the results shown in Figure 5a. and neutron and density logs (right-hand panel) for the multi-layer model of North Louisiana’s tight-gas sand. Simulated density and neutron porosity logs exhibit a larger cross-over compared to the one measured after invasion. (b) The sand is assumed to be clean and the effect of clay is neglected.1. The left-hand panel shows the spatial distribution (radial and vertical directions) of water saturation.

Table 8 summarizes the corresponding mud-filtrate and fluid properties. we simplify the invasion profile as a piston-like front to define the geometry for input to MCNP simulations. We assume a piston-like invasion profile to construct the MCNP geometry.25. We conclude that the fast linear iterative refinement method is a better choice for simulating nuclear logs.6 sandstone porosity units for neutron porosity.35 [ft3/day/ft] for one day of invasion. Core data indicate values of absolute permeability and irreducible water saturation of 50 [md] and 0. Tables 6 and 7 summarize the Archie’s parameters. In addition. but also the possibility of including invasion and environmental effects in the numerical simulation. Based on well-log interpretation. The time-constant average rate of mud-filtrate invasion is 1. Even though resistivity logs suggest high water saturation in the bottom zone. Separation of resistivity curves indicates null free water saturation in the top zone (Figure 7). Another explanation is the presence of high irreducible water saturation due to variable particle-size distribution and associated mixed wettability conditions. it is known that all the layers in this sand are in hydraulic communication. increasing to 0. it follows that volumetric concentration of shale is very high in this formation due to high gamma-ray values and large separation between density and neutron porosity logs. Table 10 lists the initial values of multi-layer petrophysical properties. A hypothesis to explain this behavior is the presence of high residual oil saturation in the bottom zone. Figure 6: Comparison of Monte Carlo (MCNP) simulations (dashed lines) and rapid approximation results (solid lines) of density and neutron measurements for the multi-layer model of North Louisiana’s tight-gas sand. The lack of neutron-density cross-over renders the analysis even more complicated. Reported hydrocarbon components from PVT analysis are lumped into five pseudo-components for the in-situ oil and three for OBM (Angeles et al. 2: Oil-Bearing Zone. The maximum difference between the two simulation methods is equal to 2. formation properties. and rock-fluid properties for this field example. Based on pressure measurements. The second field example considers an oil-bearing sand saturated with high viscosity oil. Final petrophysical properties were obtained from the matching of field logs with numerical simulations (Table 5).3 ten feet below. matrix. it is likely that overbalance and production pressures are lower than water capillary trapping forces. Field Example No. and 2. This simplifying assumption makes the MCNP-simulated density and neutron porosity higher than those simulated with the fast linear iterative refinement method. Simulation of mud-filtrate invasion seems the only viable interpretation alternative to diagnose residual oil saturation in the bottom depth zone and distinguish the effect of residual hydrocarbon and capillary pressure on the distribution of vertical fluid saturation.SPE 124879 7 In the next step. Even though the well was drilled with a deviation angle of 27° we approximate it with a vertical well. Figure 3 shows the spatial distribution of water saturation and corresponding resistivity logs. A reason explaining this difference is the effect of piston-like invasion assumed in the input geometry defined for MCNP versus the actual invasion distribution defined for the fast iterative refinement method. Table 9 describes the average petrophysical properties assumed for the oil-bearing sand in this field example. The reason is not only the more efficient CPU time. respectively. Time of mud-filtrate invasion (tinv) is five days for the rapid approximation simulations. . production tests conclusively indicate null water production. The formation of interest was drilled with OBM. 2009). Figure 6 compares the rapid approximation of neutron measurements against MCNP simulations for the North Louisiana’s tight-gas sand field example penetrated by a vertical well. Initial values for petrophysical properties were calculated by averaging log interpretation results across the desired depths.3 for density porosity.

The lack of neutron-density cross-over is due to high density of in-situ oil.25 [g/cm3] between oil and water.6 and 1. Figure 9 compares the rapid approximation of neutron measurements against MCNP simulations for the field example of the oil-bearing zone. Figure 8a shows the numerically simulated well logs before mud-filtrate invasion. relative permeability end points. originating from the corresponding decrease of the differences between density and migration length of mud-filtrate and in-situ hydrocarbon. Fast simulation of nuclear logs makes it possible to seperately quantify the effect of different petrophysical properties such as residual hydrocarbon saturation. respectively. Table 11 describes the final values of porosity. and deep radial invasion. which definitely improves log interpretation. By neglecting the presence of clay and mud-filtrate invasion. simulated density and neutron porosity measurements decrease by 2% and 40% of the initial value. This behavior can be explained by the small density difference of approximately 0.8 SPE 124879 We iteratively match numerically simulated and measured well logs by adjusting residual oil saturation. Figure 7: Comparison of numerically simulated (solid lines) and measured (dashed lines) array-induction apparent resistivity logs (second panel from right). porosity and volumetric concentration of shale. and 5% for resistivity logs. volumetric concentration of shale. non-shale porosity.8 water-filled sandstone porosity units for neutron porosity and density porosity. This fast numerical method enables us to infer a high value of residual oil saturation at the bottom zone. and wetting-phase relative permeability obtained after securing an acceptable match between simulated and measured well logs. which was impossible to quantify with classical well-log interpretation. Comparison of these results against those obtained for the gas-bearing zone of Field Example No. and neutron and density logs (right-hand panel) for the multi-layer. thereby decreasing the difference with MCNP results (which is mostly due to the effect of invasion). and water saturation. The maximum difference between the two simulation methods in the sand zone is equal to 1. We observe that the difference between the fast iterative refinement method and MCNP simulations decreases in the oil-bearing zone compared to the tightgas sand. on nuclear logs. indicates that the density log is less sensitive to mud-filtrate invasion in the oil-bearing than in the gas-bearing sand. respectively. Therefore. 1. We observe that. However. The average matching error is 6% for neutron porosity logs. oil-bearing field example model. a cross-over develops between neutron and density logs. The assumption of piston-like invasion in defining the geometry for MCNP simulations is the reason for this difference. Figure 7 shows the final match between numerically simulated and measured array-induction resistivity and nuclear logs. the effect of mud-filtrate invasion on nuclear logs decreases. The left-hand panel shows the gamma-ray log. . 1% for density logs. and a density-neutron cross-over develops in the sand zone (Figure 8b). neutron logs remain sensitive to oil saturation since the migration length of in-situ hydrocarbon is approximately 5 [cm] longer than that of water. Time of mud-filtrate invasion (tinv) is one day. in the absence of mud-filtrate invasion. residual oil saturation. high volumetric concentration of shale. The effect of invasion is about 1% on density logs and 13% on neutron logs in the oil-bearing sand. water saturation. Final petrophysical properties were obtained from the matching of field logs with numerical simulations (Table 11).

The effect of invasion is a 4-porosity-unit decrease on the neutron log. We assume a piston-like invasion profile to generate the MCNP geometry.6 and 1. We observe that the difference between the fast iterative refinement method and MCNP simulations decreases in the oil-bearing zone compared to the tight-gas sand. We observe no separation between apparent resistivity curves.8 water-filled sandstone porosity units for neutron porosity and density porosity. Figure 9: Comparison of Monte Carlo (MCNP) simulations (dashed lines) and rapid approximation results (solid lines) of density and neutron measurements for the multi-layer. Neutron porosity decreases by 6 sandstone porosity units and the cross-over between neutron and density porosity logs increases by 6 sandstone porosity units. respectively. Figure 7 shows the corresponding resistivity and gamma-ray logs. . Remaining layer petrophysical properties are equal to those of estimated values in Table 11. oil-bearing field example model. The left-hand panel shows the gamma-ray log. and neutron and density logs (right-hand panel) for the multi-layer model of the oil-bearing field example before invasion. compared to results shown in Figure 8a. Final petrophysical properties were obtained from the matching of field logs with numerical simulations (Table 11). Time of mud-filtrate invasion (tinv) is one day for the rapid approximation simulations. Figure 9 compares the rapid approximation of neutron measurements against MCNP simulations for the field example of the oil-bearing zone. The maximum difference between the two simulation methods in the sand zone is equal to 1. The assumption of piston-like invasion in defining the geometry in MCNP is the reason for this difference.SPE 124879 9 (a) (b) Figure 8: Comparison of numerically simulated (solid lines) and measured (dashed lines) array-induction apparent resistivity logs (center panel). Neutron and density porosity logs overlap in the sand zone. (a) In this model we consider the effect of volumetric concentration of shale. (b) The sand is assumed to be clean and the effect of clay is neglected.

estimating gas saturation from resistivity logs in the Hugoton field is difficult to impossible because of deep mud-filtrate invasion of freshwater mud and extremely salty connate water (Torres-Verdín et al. Petrophysical properties were obtained from the matching of field logs with numerical simulations (Table 15).12 in this formation.98 [ft3/day/ft] for one day of invasion. neutron and density logs exhibit a large cross-over due to residual gas saturation. we iteratively adjust petrophysical propertries to match the simulated resistivity and nuclear logs to the measured ones. low-permeability nonmarine siltstones. Time of mud-filtrate invasion (tinv) is one day. In the middle zone. 3: Upper Fort Riley’s Tight-Gas Sand. took less than three hours. mud-filtrate invasion is very deep. and permeability for the sand under consideration. Although invasion is deep. 2% for density. permeability and other petrophysical properties. Figure 11 shows the numerically simulated nuclear and resistivity logs before mud-filtrate invasion. and volumetric concentration of shale. The left-hand panel shows the spatial distribution (radial and vertical directions) of water saturation. and neutron and density logs (right-hand panel) for the multi-layer model of Upper Fort Riley’s tight-gas sand. mud. matrix. After obtaining an initial guess from the layer-by-layer averages of porosity. Neutron porosity is displayed in limestone porosity units. 2006).. Simulated density and neutron porosity logs exhibit a larger cross-over compared to the one measured after invasion. including log simulation and 15 iterations to obtain final petrophysical parameters. Table 14 describes the average values for porosity. Figure 10: Comparison of numerically simulated (solid lines) and measured (dashed lines) array-induction apparent resistivity logs (center panel). This behavior is expected because of higher gas saturation in the pre-invaded formation. Table 15 summarizes the final petrophysical properties achieved at the end of the iterative process. we estimated .10 SPE 124879 Field Example No. We perform the petrophysical analysis and associated numerical simulations across a 31 ft-thick sand interval. The sand interval includes gas saturation above 0. and 16% for resistivity. and rock-fluid properties. In this field example. Interpretation of this depth interval. The formation was drilled with WBM. initial water saturation. These values are calculated by averaging the properties obtained from standard well-log interpretation. fluid and formation properties. we analyze the Upper Fort Riley’s tightgas sand formation in the Hugoton field. Tables 12 and 13 summarize the corresponding Archie’s parameters. These layers are stacked in series of marinenonmarine sedimentary cycles. permeability.8 on average. Accordingly. Quantifying this separation with the proposed simulation method enables the estimation of residual gas saturation along with porosity. Figure 10 shows the final simulated resistivity and nuclear logs on top of measured ones. The main objective in this part is to achieve a fast petrophysical analysis to estimate movable and residual hydrocarbon saturation based on the simulation of nuclear and resistivity logs. Reservoir flow units consist of relatively thin marine carbonates separated by thin. The average difference between simulated and measured logs in the cleanest sands is 12% for neutron porosity. The results of this analysis are estimations of petrophysical properties with accuracy difficult to achieve via conventional log interpretation. Due to low porosity of about 0. Adding nuclear logs to the modeling process improves the estimation of gas saturation in the Hugoton field. The time-constant average rate of mud-filtrate invasion is 0.

thin beds. In the thin beds where the thickness is shorter than the vertical resolution of the neutron tool. On the other hand. and shoulder beds. Effect of Well Deviation Angle on Density and Neutron Logs: Synthetic gas-bearing zone. The left-hand panel shows the spatial distribution (radial and vertical directions) of water saturation. the linear iterative refinement method for the numerical simulation of nuclear measurements makes it possible to assess petrophysical parameters quantitatively across different geometries and bed thicknesses. we consider a multi-layer case with large contrasts of petrophysical properties.05 and 0. Diagnosing these misleading effects improves the reliability of well-log interpretations. which reduced the estimation of producible gas. where the model includes gas-saturated sands.SPE 124879 11 large residual gas saturation. Moreover. 2000). It leads to a difference between estimated neutron porosity values and those estimated from density measurements across the water-saturated thinly-laminated interval. This example emphasizes that simulating nuclear logs in different geometries helps to diagnose misleading effects observed in the well logs due to deviation angle. . radial length of investigation of neutron measurements is larger than that of density measurements. The objective of this synthetic case is to compare the simulated neutron porosity and density porosity logs across layers penetrated by vertical and HA/HZ wells. The low CPU time for nuclear simulation. We also observe a small cross-over across the thin layers near the bottom. Figure 12 compares simulated neutron and density logs for the cases of vertical and highly-deviated wells. Figure 11: Comparison of numerically simulated (solid lines) and measured (dashed lines) array-induction apparent resistivity logs (center panel). A neutron-density gas cross-over occurs near the top. Neutron porosity is displayed in limestone porosity units. across the water-saturated beds. Sand beds are water-saturated in the lower zone and gas-saturated in the upper zone. The false gas effect in the bottom thin beds is due to the difference in spatial resolution between neutron and density measurements (neutron logs exhibit lower vertical resolution than density logs) in both the vertical and deviated well cases (Ellis and Chiaramonte. and neutron and density logs (right-hand panel) for the multi-layer model of Upper Fort Riley’s tight-gas sand before invasion. increasing the angle of well inclination decreases shoulder-bed effects in the low-porosity water sand at the bottom. is the key to fast an accurate petrophysical interpretation.20 porosity bounded by shale shoulder beds. To this end. Simulations shown in Figure 12 would require several days of CPU time with MCNP in the same depth interval as opposed to approximately 15 minutes with iterative refinement approximations for the vertical well and 18 minutes for the deviated well. The model consists of sandstone layers of 0. The difference becomes smaller in the highly-deviated well compared to the vertical well.

type of mud. panels show: 3 formation bulk density model in g/cm . yet accurate algorithm for nuclear-log simulation as well as for the interactive assessment of petrophysical and geometrical properties of rock formations on measurements. and saturation properties on both density and neutron logs for the accurate calculation of porosity and fluid saturation. the migration length (Lm) of the model used for the simulation of the neutron log. and the bulk 3 density model [g/cm ] for the deviated well. From left to right. simulated neutron and density porosity logs in a well deviated 85 degrees from the vertical. the migration length (Lm) of the model.05 porosity water-saturated beds. where the model includes gas-saturated sands. A neutron-density gas cross-over occurs near the top. The numerical simulation method described in this paper could be readily interfaced with inversion procedures for the automatic estimation of static/dynamic multi-layer petrophysical properties in the presence of invasion and in combination with resistivity logs.20 and 0. reliability. quantitative assessment of multi-layer petrophysical properties in combination with resistivity logs. By contrast. petrophysical properties could be misinterpreted without reliable quantitative appraisal of shoulder. we showed that mud-filtrate invasion may have unequal effects on density and neutron logs depending on the radial length of invasion. for instance. This bottom section is a laminated sequence of 0. . Such unequal effects may give rise to biased estimations of porosity and fluid saturation if not accounted for in the interpretation. simulated neutron and density porosity logs in a vertical well across horizontal layers of different petrophysical properties. Simulation of borehole nuclear measurements has been elusive in the past because of extremely time consuming and cumbersome Monte Carlo methods. The great variety of geometrical and petrophysical properties of rock formations make it necessary to quantify a multitude of hypotheses for reliable appraisal of static and dynamic petrophysical properties. bed thickness.12 SPE 124879 Figure 12: Neutron and density logs simulated with the linear iterative refinement approximation. The false gas effect is caused by the difference in spatial resolution between neutron and density measurements in both the vertical and deviated well cases (Ellis and Chiaramonte. and well-inclination effects on nuclear logs. and internal consistency of measurements acquired with borehole nuclear tools. In addition. simulation of borehole resistivity measurements is routinely performed by oil companies to appraise (and correct) shoulder-bed and invasion effects on processed well logs. and dip. they respond in different ways to invasion and shoulder beds. 2000). An interesting observation is a small false apparent gas cross-over effect across the thin layers near the bottom. among other things. The vertical axis in shown on the left panel describes true vertical depth (TVD) of the vertical well and the right panel shows the measured depth (MD) depth across the same interval penetrated by a well deviated 85 degrees. Conclusions Synthetic and field examples of the simulation of density and nuclear measurements confirm the need of fast simulation methods for quantitative and interactive assessment of petrophysical properties. Consequently. The algorithm permits. This paper introduced a computer efficient. Experience with the interpretation of field data sets indicates that numerical simulation is a valuable procedure to verify the accuracy. fluid. Because density and neutron logs do not exhibit the same volume of investigation. We considered examples of the assessment of matrix. invasion.

[frac. Salazar.. in Transaction of The American Nuclear Society. C. RWE. C. Ellis.. [frac. [hours] Porosity. Mendoza. 28 . TX.] Residual wetting-phase saturation. The Woodland. Computer Modeling Group Ltd. [md] Non-wetting-phase relative permeability krnw end point Wetting-phase relative permeability krw end point Migration length [cm] Archie’s cementation exponent Archie’s saturation exponent Capillary pressure.] Density porosity. J. 2003... and Torres-Verdín. and Klein. Habashy. M. Baker Hughes. 2009. and Wang.. and Scott.] Snwr Sor Sw Swr Swt tinv : : : : : : : : : : : : : : : Non-wetting phase residual saturation. version 5. and Torres-Verdín. C. 655. volume III.] Time of mud-filtrate invasion. [psi] Coefficient for Pc equation.. 6.. 2007... C. TOTAL... Alpak. BG. 1215–1219. C. Torres-Verdín. ConocoPhillips. [frac. ENI. J. Geophysics. Torres-Verdín. O. 2006.] Neutron porosity. Mexican Institute for Petroleum. Monte Carlo modeling of borehole nuclear measurements in vertical and horizontal wells in the presence of mud-filtrate invasion and salt mixing. [frac. Part II: high-angle and horizontal wells. Hess. Part I: vertical wells. 41. Preeg. Torres-Verdín.. and Elshahawi. C.. vol. T. [frac. 2009. A general Monte Carlo N-Particle transport code. Schlumberger. [frac. New Mexico. D.] Bulk density. 2009. E. F. SNUPAR – a nuclear parameter code for nuclear geophysics applications: IEEE Transactions on Nuclear Geoscience. . 1.. vol. Canada. W. Estimation of permeability from array induction measurements: applications to the petrophysical assessment of tight-gas sands: Petrophysics. 46. Mendoza. F. Alpak. no. M. C. and Preeg.. [psi. [frac.. jointly sponsored by Anadarko. J. W. and Habashy. Torres-Verdín. O. H. Petrophysical inversion of borehole array-induction log: Part II . R. History matching of multi-phase flow formation-tester measurements acquired with focused sampling probes in deviated wells: 50th Annual Logging Symposium Transactions: Society of Professional Well Log Analysts. [frac. E57-E73. 2008. McKeon. 71. Improving the assessment of residual hydrocarbon saturation with the combined quantitative interpretation of resistivity and nuclear logs. no. The Woodlands. C. Quantitative comparison of processes of oil..] Total water saturation. Alberta. in Transactions of the SPWLA 50th Annual Logging Symposium. [frac. Interpreting neutron logs in horizontal wells: a forward modeling tutorial. D. 2000. TX.darcy1/2] Residual gas saturation. and Chiaramonte. v. 48.Field data examples. June 21 – 24. no. 1989. M. Los Alamos National Laboratory. Los Alamos.. and Weatherford. vol. pp. C.. in press. vol. J.. D. C. 2009. pp..] Non-shale porosity. User’s guide: STARS: Computer Modeling Group Ltd. Mendoza. M. A. L.and water-base mud-filtrate invasion and corresponding effects on borehole resistivity measurements: Geophysics. Chevron. 1984. February.] Residual oil saturation. Salazar. [frac. G. Watson. pp. E. developer’s guide: University of California. [frac. no. 1. ExxonMobil. 47. Torres-Verdín.. Mendoza. Petrobras. pp. BP. no. [g/cm3] Shale density.. Halliburton.. 23 -32. [g/cm3] Fluid density. A. Calgary.] Wetting-phase saturation. A. Nexen.. Petrophysics. 36... Petrophysics. 2006. M. StatoilHydro. Marathon.. pp. C.] Total porosity. A. submitted to Geophysics. Z.] Pore-size distribution exponent Experimental exponent for krnw equation Experimental exponent for krw equation Permeability. 2009. Geophysics. 1.. 74. Monte Carlo computation of differential sensitivity functions. June 21-24.. [frac.. pp. no. Linear iterative refinement method for the rapid simulation of borehole nuclear measurements. 5. Torres-Verdín. D. paper FFF. [g/cm3] Matrix density. Heidari. 1. Linear iterative refinement method for the rapid simulation of borehole nuclear measurements. T. pp. and Preeg. Aramco.44. H. References Angeles. G261-G268. E.SPE 124879 13 Nomenclature a Csh ep enw ew k krnw k0mw krw k0rw Lm m n Pc Pc0 Sgr AIT® CMG FSF FSFB GAPI GEM® GR HA/HZ MCNP MD : : : : : : : : : : : : : : : : Archie’s tortuosity factor Volumetric concentration of shale. BHP Billiton.. V. [g/cm3] φ φD φN φs φt ρb ρf ρm ρsh Acronyms : : : : : : : : : : Array-Induction Tool Computer Modeling Group Ltd Flux Sensitivity Function [1/cm3-eV] Flux Sensitivity Function for an assumed base-case background formation [1/cm3-eV] Gamma-ray American Petroleum Institute Units Compositional Simulator from CMG Gamma Ray High-angle/Horizontal Monte Carlo N-Particle Program Measured Depth OBM ppm SNUPAR STARS® TVD WBM 3D 2D 1D : : : : : : : : Oil-Based Mud Parts Per Million Schlumberger Nuclear Parameter Code Multi-Component Thermal Reservoir Simulator from CMG True Vertical Depth Water-Based Mud Three Dimensional Two Dimensional One Dimensional Acknowledgements The work reported in this paper was funded by the University of Texas at Austin’s Research Consortium on Formation Evaluation. 527–544. X-5 Monte Carlo Team MCNP. vol. W. vol.

n Connate water salt concentration Bound water resistivity Shale density Matrix density Formation temperature Initial formation pressure Mud hydrostatic pressure Mudcake reference permeability Mudcake reference porosity Mud solid fraction Mudcake maximum thickness Mudcake compressibility exponent Mudcake exponent multiplier Wellbore radius Formation maximum invasion time Value 1.30 0.05 0. mud.043 0.00 1.5 Swr 0.40 0.14 5 25.14 SPE 124879 Table 1: Archie’s parameters.038 2.06 1.15 2 7.25 0. a Archie’s cementation exponent.153 0.45 0.154 0.574 0.30 0.29 0.06 0.8 2.4 Table 7: Rock-fluid properties assumed for the oil-bearing field example.00 210 5000 5825 0.07 1 0.65 1.06 Units ft md fraction fraction fraction fraction Table 8: Mud and fluid properties assumed for the oilbearing field example.00 1.60 2. matrix.76 2. m Archie’s saturation exponent.30 0. matrix. 0.03 0.87 5.16 3 0.darcy1/2] 1. and formation properties assumed for North Louisiana’s tightgas sand.12 Units ft md fraction fraction fraction fraction .08 Swt 0.117 0. Pc0 [psi.9 2.631 0. Pc0 [psi.40 0.00 Units kppm Ohm.00 1.176 0.30 0. fluid.000 Table 9: Averaged petrophysical properties calculated for the oil-bearing field example. n Connate water salt concentration Mud-filtrate salt concentration Shale density Matrix density Water density Water viscosity Gas density Gas viscosity Mud-filtrate density Mud-filtrate viscosity Formation temperature Initial formation pressure Mud hydrostatic pressure Mudcake reference permeability Mudcake reference porosity Mud solid fraction Mudcake maximum thickness Mudcake compressibility exponent Mudcake exponent multiplier Wellbore radius Formation maximum invasion time Irreducible water saturation Value 1. Layer thickness [ft] 3 2 5 5 3 3 1 k [md] 58 312 257 334 259 30 0.976 1.10 Table 6: Archie’s parameters.033 0.30 0.02 0. m Archie’s saturation exponent.45 0.75 160 3. a Archie’s cementation exponent.09 0. Table 5: Final multi-layer petrophysical properties obtained after matching the measured logs with numerical simulations for North Louisiana’s tight-gas sand.10 10.4 ep 2 k0rnw 0.515 Swt 0.358 0. Variable Water density Water viscosity In-situ oil density In-situ oil viscosity Mud-filtrate density Mud-filtrate viscosity Value 1.06 1.448 0.170 0.00 0.10 10.145 Csh 0.30 0.09 0.1 105 0.02 0.02 1.452 0.00 0.79 1.0004 φs Csh 0.121 0.48 0.159 3 0.5 k0rnw 0.278 0.26 5.5 k0rw 0.80 0.481 0.60 0.015 0.45 0.03 0.00 1.68 2.10 0.80 Sgr 0.95 1.m g/cm3 g/cm3 °F psi psi md fraction fraction cm fraction fraction cm Days Table 2: Rock-fluid properties assumed for North Louisiana’s tight-gas sand.48 ew 2 Variable Thickness Absolute permeability Non-shale porosity Total water saturation Volumetric concentration of shale Shale porosity Value 22 210 0.0001 Variable Thickness Absolute permeability Non-shale porosity Total water saturation Volumetric concentration of shale Shale porosity Value 15 35 0.21 0. Table 3: Averaged petrophysical properties assumed for North Louisiana’s tight-gas sand.20 0.069 0. Variable Archie’s tortuosity factor.40 Units kppm kppm g/cm3 g/cm3 g/cm3 cp g/cm3 cp g/cm3 cp °F psi psi md fraction fraction cm fraction fraction cm Days fraction Layer thickness [ft] k [md] φs 3 1.00 Units g/cm3 cp g/cm3 cp g/cm3 cp Table 4: Initial values of model properties assumed for North Louisiana’s tight-gas sand in different petrophysical layers.00 1.126 0.90 0.darcy1/2] 20 ep 4. Variable Archie’s tortuosity factor.9 enw 2.30 0.100 0.10 0. and formation properties assumed for the oil-bearing field example.18 5 45.176 0.926 0.00 0.40 0.85 ew 4.92 enw 2.50 0.65 254 18949 19500 0.10 0.50 0.50 0.016 0.04 0.

230 Table 15: Final multi-layer petrophysical properties obtained from the matching of measured logs with numerical simulations for Upper Fort Riley’s tight-gas sand field example. Layer thickness [ft] Swt Sgr k[md] Csh φs Csh 0.19 0.10 0.69 1. Table 14: Averaged petrophysical properties assumed for Upper Fort Riley’s tight-gas sand.00 1.56 0. m Archie’s saturation exponent.45 0.18 0.30 0. fluid and formation properties assumed for Upper Fort Riley’s tightgas sand.5 k0rw 0.30 0.42 0.35 0.12 0.19 0.20 3 0.18 0.33 Swt 0.28 0.22 0.3 2.46 0.75 0.26 0.30 0.30 0.18 0.3 0.4 enw 2.24 0.25 0.12 0. matrix.14 0.11 0.0 96 92 1153 0.26 0.40 0.53 0.19 0.12 0.30 krw 0.22 0.145 0. Layer thickness [ft] 2 3 2 3 2 1 1 1 φs 0. n Connate water salt concentration Mud-filtrate salt concentration Shale density Matrix density Water density Water viscosity Gas density Gas viscosity Mud-filtrate density Mud-filtrate viscosity Formation temperature Initial formation pressure Mud hydrostatic pressure Mudcake reference permeability Mudcake reference porosity Mud solid fraction Mudcake maximum thickness Mudcake compressibility exponent Mudcake exponent multiplier Wellbore radius Formation maximum invasion time Irreducible water saturation Value 1.001 φs 0.6 0.38 Variable Thickness Absolute permeability Non-shale porosity Total water saturation Volumetric concentration of shale Shale porosity Value 31 3 0.00 0. Pc0 [psi.00 0.12 0.33 0.10 0.30 0.5 Swr 0.6 0.186 1.3 3 8 2 4 3 5 4 0.76 2.12 0.150 0.26 0.02 0.08 Units ft md fraction fraction fraction fraction Table 11: Final multi-layer petrophysical properties obtained from the matching of measured logs with numerical simulations for the oil-bearing field example.190 0.10 0. Layer thickness [ft] 2 3 2 3 2 1 1 1 k [md] 31 53 30 71 14 2 1 0.20 3 3 3 3 0.220 0.10 10.30 0.18 0.00 0.12 0. Variable Archie’s tortuosity factor.00 1. a Archie’s cementation exponent.21 0.3 0.220 0.14 0.25 0.230 0.12 0.00 0.SPE 124879 15 Table 10: Initial multi-layer petrophysical properties for the oil-bearing field example.33 Sor 0.25 0.20 Csh 0.13 0.12 0. mud.06 1.08 0.40 0.35 0.51 200 0.30 0.2 0.17 0.24 0.05 0.20 3 3 Table 12: Archie’s parameters.25 0.29 Sw 0.22 0.00098 0.220 0.01087 1.50 0.00 0.35 0.3 0.08 .20 0.12 0.19 0.11 0.0 1.0 ep 10 k0rnw 0.03 0.17 0.89 1.darcy1/2] 1.49 0.10 0.08 Units kppm kppm g/cm3 g/cm3 g/cm3 cp g/cm3 cp g/cm3 cp °F psi psi md fraction fraction cm fraction fraction cm Days fraction Table 13: Rock-fluid properties assumed for Upper Fort Riley’s tight-gas sand.12 0.15 0.6 0.3 ew 4.23 0.

Sign up to vote on this title
UsefulNot useful