interpretation of porosity and fluid constituents from ... · pdf fileinterpretation of...

Interpretation of porosity and fluid constituents from well logsusing an interactive neutron-density matrix scale

Olabode Ijasan1, Carlos Torres-Verdín1, and William E. Preeg2

Abstract

Neutron and density logs are important borehole measurements for estimating reservoir capacity and infer-ring saturating fluids. The neutron log, measuring the hydrogen index, is commonly expressed in apparentwater-filled porosity units assuming a constant matrix lithology whereby it is not always representative of actualpore fluid. By contrast, a lithology-independent porosity calculation from nuclear magnetic resonance (NMR)and/or core measurements provides reliable evaluations of reservoir capacity. In practice, not all wells includecore or NMR measurements. We discovered an interpretation workflow wherein formation porosity and hydro-carbon constituents can be estimated from density and neutron logs using an interactive, variable matrix scalespecifically suited for the precalculated matrix density. First, we estimated matrix components from combina-tions of nuclear logs (photoelectric factor, spontaneous gamma ray, neutron, and density) using Schlumberger’snuclear parameter calculator (SNUPAR) as a matrix compositional solver while assuming freshwater-filled for-mations. The combined effects of grain density, volumetric concentration of shale, matrix hydrogen, and neu-tron lithology units define an interactive matrix scale for correction of neutron porosity. Under updated matrixconditions, the resulting neutron-density crossover can only be attributed to pore volume and saturating fluideffects. Second, porosity, connate-water saturation, and hydrocarbon density are calculated from the discrep-ancy between corrected neutron and density logs using SNUPAR and Archie’s water saturation equation,thereby eliminating the assumption of freshwater saturation. With matrix effects eliminated from the neu-tron-density overlay, gas- or light-oil-saturated formations exhibiting the characteristic gas neutron-densitycrossover become representative of saturating hydrocarbons. This behavior gives a clear qualitative distinctionbetween hydrocarbon-saturated and nonviable depth zones.

IntroductionPorosity calculated from neutron and density mea-

surements is still the most commonly used estimateof pore volume in rock formations penetrated by wells.In complex lithologies, inadequate characterization ofthe matrix could yield inaccurate porosity and satura-tion estimates. The petrophysical effects of lithology,saturating fluid, and borehole conditions on nuclear logsare exhaustively discussed by Ellis and Singer (2007).Using departure curves from log interpretation charts(Schlumberger, 2009), corrections are applied such thatinterpreted properties are representative of the forma-tion only. Extensive studies and publications on neutronand density logs, being ubiquitous for porosity and hy-drocarbon estimation, can be found in the literature.

Historically, total porosity φt in gas-bearing for-mations is approximated with the following formula(Gaymard and Poupon, 1968):

φ2t ≈

φ2N þ φ2

D

2; (1)

where φD and φN are the density- and neutron-apparentporosities, respectively. Mao (2001) studies the correla-tion characteristics of φD and φN for identification ofoil- and gas-saturated zones. Spears (2006) applies lith-ofacies-based porosity corrections derived from neu-tron-density crossplots for φt calculations in geologicand reservoir models. Fertl and Timko (1971) extendGaymard and Poupon’s (1968) formulations for calcula-tion of hydrocarbon density ρhc and detection of oil- andgas-bearing intervals in shaly sands.

The neutron-density overlay technique relies on thedifference between apparent porosities, on a prede-fined matrix scale, for inferring hydrocarbon saturation(Shc), φt, and ρhc. Several petrophysical factors ad-versely affect the reliability of the overlay technique.

1University of Texas at Austin, Austin, Texas, USA. E-mail: [email protected]; [email protected] consultant, Austin, Texas, USA. E-mail: [email protected] received by the Editor 5 June 2013; revised manuscript received 12 July 2013; published online 10 October 2013. This paper appears

in Interpretation, Vol. 1, No. 2 (November 2013); p. T143–T155, 8 FIGS., 6 TABLES.http://dx.doi.org/10.1190/INT-2013-0072.1. © 2013 Society of Exploration Geophysicists and American Association of Petroleum Geologists. All rights reserved.

t

Technical paper

Interpretation / November 2013 T143Interpretation / November 2013 T143

Dow

nloa

ded

05/2

9/14

to 1

28.8

3.16

7.15

5. R

edis

trib

utio

n su

bjec

t to

SEG

lice

nse

or c

opyr

ight

; see

Ter

ms

of U

se a

t http

://lib

rary

.seg

.org

/

For example, gas detection is challenging in shalysands or shale gas reservoirs due to opposite effectsof shale-hydroxyls and gas density in overlay character-istics. Similarly, in oil-saturated or invaded gas zones,the decreased difference between neutron and density-apparent porosities masks the characteristics of light-hydrocarbon crossovers (Mao, 2001). Consequently,application of the overlay technique requires implemen-tation of a suitable matrix correction.

In this paper, we estimate Shc, φt, and ρhc usingan interactive interpretation workflow based on theneutron-density overlay technique, with explicit con-sideration of neutron matrix scale and shale content.The interpretation workflow improves reliability of theoverlay technique in the presence of arbitrary lithologyand fluid effects. These effects and their influence onneutron and density-apparent porosities, along withconventional well-log interpretation methods, are de-scribed with a synthetic example of known lithologyand fluid constituents. Additionally, the interactivematrix scale method is applied to field examples ofvarying geology and lithology, namely, carbonate, sil-iciclastic, and shale reservoirs, where porosity andfluid-saturation estimates are compared to laboratorycore measurements. The calculated ρhc enables differ-entiation between gas- and oil-saturated intervals whenρhc < 0.25 g∕cm3 and ρhc ≥ 0.25 g∕cm3, respectively,for simplified interpretation. Hence, the interactiveanalysis method is implemented for qualitative identi-fication of fluid zones, fluid contacts, and reservoircompartments.

Interpretation of apparent porosityThe porosity value associated with neutron logs is

inherently apparent for given matrix and fluid units.On the other hand, compensated bulk density measure-ments bear no apparentness until density porosity iscalculated with constant values of matrix and fluidproperties. This is a significant difference between den-sity and neutron logs.

Density-apparent porosityThe bulk density measurement ρb principally re-

sponds to formation electron density such that

ρb ¼XMi¼1

ð1.0704ρe;i − 0.188ÞVi; (2)

where ρe;i and Vi are the electron density and volumefraction, respectively, of the ith fluid/matrix componentup to M components. In hydrocarbon-bearing forma-tions, φt can be directly calculated from density logsif and only if matrix density ρm and fluid density ρf areknown precisely. Otherwise, density-apparent porosityφD is obtained using

φD ¼ ρb − ρmρf − ρm

¼ φt þ ΔφD; (3)

where ρm and ρf are assumed matrix and fluid densities,respectively, e.g., limestone matrix of 2.71 g∕cm3 andfreshwater of 1 g∕cm3. The porosity departure ΔφDdue to ρm and ρf assumptions (Ellis et al., 2007) is quali-tatively and quantitatively intuitive such that

ΔφD ≈ρb − ρm

ðρf − ρmÞ2ðΔρm − Δρf Þ −

1ρf − ρm

Δρm; (4)

where Δρm and Δρf are differences in matrix and fluiddensities, respectively, between assumed and trueproperties.

Neutron-apparent porosityThe neutron log is an apparent porosity measure-

ment, given that it refers to an equivalent hydrogen in-dex (HI) response in water-filled lithology units, usuallylimestone. Limestone unit implies an equivalent re-sponse of water-filled limestone formation where thepore volume equals that of the neutron log. As shownby Gaymard and Poupon (1968), the environmentallycorrected neutron-porosity log φN across invaded for-mations can be expressed as

φN ¼ 1HImf

XMi¼1

HIiV i ¼ φt þ ΔφN; (5)

where HI is the hydrogen index, subscript mf identifiesmud filtrate, and ΔφN is porosity departure due to neu-tron-apparent porosity measurement. Equation 5 im-plies that the neutron response is a superposition ofthe volumetric contributions of component hydrogenconcentrations. The apparentness in ΔφN is determinedby the neutron-porosity unit (pu), usually water-filledlimestone. For example, quartz, calcite, and dolomiteblocks yield ΔφN of −2, 0, and 0.5 limestone pu, respec-tively. This matrix effect is qualitatively intuitive be-cause quartz and dolomite have lower and highermatrix densities, respectively, than limestone. On theother hand, unlike φD, the matrix effect is quantitativelyobscure and cannot be calculated directly from equa-tions 3 or 4. This effect is exacerbated in complex mix-tures of various lithologies. Similarly, a gas-saturatedlimestone formation yields negative ΔφN because theHI of gas is typically lower than that of water.

Hence, a physically intuitive parameter representa-tion of neutron-porosity responses is necessary. Usingneutron characteristic lengths, specifically migrationlength Lm a calibration of Lm-to-neutron porosity isused to quantify matrix and lithology effects (Ellis et al.,2007).

Neutron parameter modelThe Schlumberger nuclear parameter calculator

(SNUPAR, McKeon et al., 1989) calculates nuclearproperties such as Lm, HI, photoelectric factor (PEF),capture cross section, Σ, etc., for any given mixture ofrocks and fluids. In this paper, we implement Lm for

T144 Interpretation / November 2013

Dow

nloa

ded

05/2

9/14

to 1

28.8

3.16

7.15

5. R

edis

trib

utio

n su

bjec

t to

SEG

lice

nse

or c

opyr

ight

; see

Ter

ms

of U

se a

t http

://lib

rary

.seg

.org

/

property characterization of wireline neutron poro-sity responses, typically with an americium-beryllium(AmBe) neutron source. It follows that equation 5can be rewritten as

φN ¼ gl

�1Lm

�; (6)

where gl represents the Lm-to-porosity calibration func-tion in limestone water-filled units and gl is obtained byfitting a polynomial function to the inverse of SNUPAR-calculated Lm and limestone pore volume. Figure 1shows gs, gl, and gd water-filled calibration functionsfor sandstone, limestone, and dolomite units, respec-tively. Additionally, the SNUPAR-calculated compen-sated neutron tool (CNT) thermal porosity response,shown in dashed blue, agrees well with gl. Unless oth-erwise stated, neutron-porosity logs in this paper areexpressed in limestone matrix units, where g−1l and glare used to convert neutron porosity to Lm logs, andvice versa, respectively.

Furthermore, we implement a SNUPAR-based com-positional solver (Heidari et al., 2012) for estimation ofmineral and fluid concentrations from nuclear logs. Thesolver uses nonlinear minimization of a constrained-error, quadratic cost function between SNUPAR-predicted properties and nuclear logs (ρb, φN , PEF)for estimation of mineral and fluid volumetric fractions.Additionally, volumetric concentration of shale V sh andwater saturation Sw are calculated using linear scalingof the gamma ray (GR) log and Archie’s equation,respectively.

Lithology effectsMatrix effect

Equation 4 describes the sensitivity of φD such thatthe matrix effect in water zones, i.e., when Δρf ¼ 0 andρf ¼ ρcw (connate-water density), is given by

ΔφD ≈�

ρb − ρmðρf − ρmÞ2

−1

ρf − ρm

�Δρm: (7)

In the neutron-density overlay technique, water-satu-rated zones are expected to overlap only if the matrixscale for density and neutron corresponds to the pre-cise formation lithology. Otherwise, the neutron-densitymatrix effect, Δφmatrix ¼ ΔφN − ΔφD (equations 3 and5), depends on Δρm and the neutron response, φN , ofthe matrix. Unlike φD in equation 3, φN of the matrixis not quantitatively intuitive, and it is only obtainedfrom equation 6 by converting SNUPAR-calculatedLm to neutron porosity. Qualitatively, with a limestonematrix scale in water zones, Δφmatrix < 0 across sand-stone and Δφmatrix > 0 across dolomite.

Shale-hydroxyl or matrix-hydrogen effectTypically, shales consist of clay minerals with high

hydroxyl (OH−) content such that φN > φD. The

shale-hydroxyl effect ΔφNsh can be approximated fromequation 6 using the expression

ΔφNsh ¼ V shgl

�1

Lmsh

�; (8)

where Lmsh is the SNUPAR-calculated migration lengthin shale. For example, Lmsh is approximately 15.35 cmfor illite of density 2.78 g∕cm3, whereby ΔφNsh for pureillite (see Figure 1), i.e., V sh ¼ 1, corresponds to 0.156.In unconventional reservoirs with an organic-rich ker-ogen matrix (Passey et al., 1990), the neutron porosityresponse increases due to high hydrogen content of or-ganic matter. The SNUPAR-calculated HI of kerogencould be as high as 0.8, depending on the hydrogen-carbon ratio and kerogen density. Accordingly, equa-tion 8 quantifies the matrix-hydrogen effect, whereV sh ¼ 1 and Lmsh become Vker (volume fraction of ker-ogen) and Lmker (SNUPAR-calculated migration lengthof kerogen matrix), respectively.

It then follows that the total matrix effect on neutron-porosity logs is an addition of Δφmatrix and ΔφNsh,i.e., interactive porosity departures due to apparentlimestone matrix scale (calculated from SNUPAR infreshwater-filled assumptions, equations 6 and 7) andshale-hydroxyl or matrix-hydrogen effect (equation 8).The corrected or rescaled neutron-apparent porosity isgiven by

φNcorr ¼ φN − ðΔφmatrix þ ΔφNshÞ: (9)

Fluid and hydrocarbon saturation effectsGiven equations 3, 5, 6, and 9, fluid and saturation

effects on rescaled neutron-apparent porosity can bewritten as

Figure 1. SNUPAR-calculated water-filled neutron porositycalibration functions gs, gl, and gd for sandstone, limestone,and dolomite units, respectively. The figure also shows neu-tron porosity responses across relevant formations.

Interpretation / November 2013 T145

Dow

nloa

ded

05/2

9/14

to 1

28.8

3.16

7.15

5. R

edis

trib

utio

n su

bjec

t to

SEG

lice

nse

or c

opyr

ight

; see

Ter

ms

of U

se a

t http

://lib

rary

.seg

.org

/

φNcorr ¼ φtð1þ ShcδΣÞ; (10)

where δΣ is the difference in neutron response betweenhydrocarbon- and water-saturated formations. Severalforms of equation 10 are given in Gaymard and Poupon(1968), Mao (2001), and Quintero and Bassiouni (1998).Gaymard and Poupon (1968) characterize δΣ across in-vaded formations as the relative difference in HI be-tween residual hydrocarbon and mud-filtrate; i.e.,

δΣ ¼ ðHIhc − HImfÞ∕HImf ; (11)

where the subscript hc identifies hydrocarbon. For gas-saturated formations at reservoir conditions, one has

HIg ≈ 9ρg½0.15þ 0.2ð0.9 − ρgÞ2�; (12)

where the subscript g describes gas. Equation 12 is re-plicated in SNUPAR for ρhc ¼ ρg < 0.25 g∕cm3, while aSNUPAR-derived functional relationship is obtained foroil (CnH2nþ2) when ρhc ¼ ρg > 0.25 g∕cm3. Estimationof Shc, φt, and ρhc thus requires solving equations 2,10, 11, 12, and inclusion of a water saturation model,e.g., Archie’s equation,

Rt ¼aRw

φmt ð1 − ShcÞn

; (13)

where Rt is the resistivity log, Rw is connate-water ormud-filtrate resistivity, a is Archie’s factor, m is theporosity exponent, and n is the saturation exponent.It follows that δΣ ¼ 0 corresponds to a water or deeplyinvaded zone. Consequently, the magnitudes of δΣ andρhc dictate the hydrocarbon type, i.e., oil or gas.

Interactive analysis of matrix and fluid effectsWell-log interpretation involves conceptual rock mod-

els when evaluating formation rock composition and

saturating fluids. This paper introduces a new inter-pretation method, or interactive analysis workflow,that combines petrophysical effects due to apparentmatrix scale and hydrocarbon saturation. Using a syn-thetic example of a layered earth model, where welllogs are simulated with the University of Texas at Aus-tin petrophysical and well-log simulator (UTAPWeLS,Voss et al., 2009), we describe the estimation ofφt, ρhc, and Shc using the interactive interpretationworkflow.

Interpretation workflowThe first part of the interpretation involves rock/

matrix compositional interpretation from ρb, PEF,and GR logs using the SNUPAR-based solver underthe assumption of freshwater-filled saturation. We as-sume freshwater-filled formations for two reasons:(1) The environmentally corrected φN is typically refer-enced on freshwater-filled units and (2) to independ-ently characterize matrix effects for estimation of ρmgiven that formation fluids have negligible or no effecton PEF and GR logs.

Using the estimated ρm from the matrix solver andequation 3, we calculate density-apparent porosityunder the freshwater-filled assumption, φDwf . Accord-ingly, neutron-apparent porosity under the fresh-water-filled assumption, φNwf , is obtained by convertingthe predicted Lm from the matrix solver to neutronporosity. It follows from equations 8 and 9 thatΔφmatrix þ ΔφNsh ¼ φNwf − φDwf , i.e., the interactiveneutron-density lithology effect in limestone porosityscale, where V sh is calculated assuming linear scalingof the GR log. We then calculate the corrected neu-tron-apparent porosity φNcorr from equation 9 forrescaling with φD. At this point, the overlay character-istics of φNwf and φDwf are solely due to porosity effects,and the overlay of φNcorr and φD is due to hydrocarbonpore volume.

The second part of the interpretation involves imple-menting the SNUPAR-based solver for hydrocarboncharacterization. In this step, equations 2, 10, 11, 12,and 13 are solved such that a SNUPAR-defined inherentrelationship between δΣ and ρhc is implemented inthe analysis for estimation of ρhc, Shc, and φt. The func-tional relationship between HI and ρhc is derivedfrom SNUPAR for oil (ρhc > 0.25 g∕cm3) and gas(ρhc < 0.25 g∕cm3).

Figure 2 summarizes the interpretation workflowwhere the “Matrix solver” loop is interactive as rockcomponents (e.g., quartz, dolomite, pyrite, etc.) arechosen to quantify their effects on the calculated neu-tron-density matrix scale. Additionally, we compare es-timated ρm to core measurements wherever availableand appraise the solver’s numerical reproduction ofPEF and GR measurements. Based on these compari-sons, an interpretation decision is made concerningthe most representative formation rock components.Consequent with the “Fluid solver” of Figure 2, final in-terpreted results include total porosity, hydrocarbon

Figure 2. Interactive interpretation workflow for interpreta-tion of neutron and density-apparent porosities.


Dow

nloa

ded

05/2

9/14

to 1

28.8

3.16

7.15

5. R

edis

trib

utio

n su

bjec

t to

SEG

lice

nse

or c

opyr

ight

; see

Ter

ms

of U

se a

t http

://lib

rary

.seg

.org

/

density, water saturation, matrix/grain density, and for-mation rock components.

Synthetic exampleThe interactive interpretation workflow is described

for matrix and fluid effects on density and neutron-apparent porosities, using numerically simulated mea-surements across a synthetic and simplistic earthmodel. This model is designed to describe practical sit-uations that present challenges to the interpretation ofneutron and density logs.

Tables 1 and 2 describe the properties assumed forthe synthetic earth model, while Figure 3 shows thesimulated nuclear and resistivity logs. In Figure 4, wedescribe the interpretation results obtained with the in-teractive analysis workflow. Figure 4a and 4f–4h showsthat estimated ρm, φt, Sw, and ρhc, respectively, usingthe interactive interpretation, agree well with modelproperties in Table 1. It is particularly significant thatthe calculated ρhc in Figure 4h distinguishes betweengas- and oil-saturated layers.

Layers I and IV consist of water-saturated shale ofmixed orthoclase and illite clay, where Δφsh ¼ 0.155and shale density ρsh ¼ 2.738 g∕cm3. After correctionfor shale-hydroxyl effects, the actual matrix crossovereffect, due to the shale density greater than limestonedensity, is shaded in brown in Figure 4b. On the otherhand, layer V consists of gas-saturated shale with 20%water saturation (refer to Table 1), such that the gascrossover effect becomes accentuated after correctionfor shale-hydroxyl effect. In this layer because gassaturation and V sh impose opposite overlay character-istics, φN − φD experiences a competition between gasand shale-hydroxyl effects. This behavior in neutron-density interpretation is especially common in logs ac-quired across shale gas formations.

Layers II and III consist of gas- and oil-saturated lime-stone formations, respectively. The matrix effect isirrelevant in these layers because limestone is the refer-ence scale for neutron-density overlay. This behavior iscorroborated by the overlap of φDwf and φNwf in panel c

of Figure 4. Hydrocarbons, in comparison to freshwater, reduce the neutron porosity response becauseof lower HI (equation 5). From equations 11 and 12,the hydrocarbon effect is dependent on ρhc and is ac-centuated in gas-saturated layers when compared tooil-saturated layers. In Figure 4d, layer III shows lowerhydrocarbon effects and could be inadvertently inter-preted as a water-filled layer. Consequently, the fluidsolver incorporates the resistivity measurement, Ar-chie’s model (equation 13), equations 2, 10, 11, and12 for an inclusive calculation of ρhc, Shc, and φt.Figure 4h shows that the estimated ρhc reliably predictsgas and oil densities in gas- and oil-saturated layers IIand III, respectively. In Figure 4f, the φt approximationusing Gaymard-Poupon’s formula (equation 1) is validin layer II but inaccurate in shaly layers.

In layers VI and VII, for oil- and water-saturateddolomite, respectively, the overlay characteristics inFigure 4b indicate a matrix crossover. The matrix effectin panel d shows that Δφmatrix ¼ 0.0072 (i.e., 0.72 pu) forΔφsh ¼ 0. By comparison, SNUPAR-calculated CNT re-sponse yields apparent thermal neutron porosity of0.5 pu in dolomite of 0% pore volume.

Table 1. Layer properties assumed in the synthetic example.

Layer Matrix Saturation fluid properties Interpretation comments

I Shale: 80% illite, 20% orthoclase,ρsh ¼ 2.738 g∕cm3

φt ¼ 0.10 Sw ¼ 1, Shc ¼ 0 Shale and matrix effects

II Limestone φt ¼ 0.28, Sw ¼ 0.05, Shc ¼ 0.95 (methane,CH40.182 g∕cm3)

Gas effect

III Limestone φt ¼ 0.28, Sw ¼ 0.05, Shc ¼ 0.95 (Liquid hydrocarbon,C16H340.757 g∕cm3)

Hydrocarbon effects

IV Shale: 80% illite, 20% orthoclase φt ¼ 0.05, Sw ¼ 1, Shc ¼ 0 Shale and matrix effects

V Shale, 80% Illite, 20% Orthoclase φt ¼ 0.10, Sw ¼ 0.20, Shc ¼ 0.80 (Methane,CH40.182 g∕cm3)

Shale and gas effects

VI Dolomite φt ¼ 0.28, Sw ¼ 0.05, Shc ¼ 0.95 (liquid hydrocarbon,C16H340.757 g∕cm3)

Matrix and hydrocarboneffects

VII Dolomite φt ¼ 0.10 Sw ¼ 1, Shc ¼ 0 Matrix effects

VIII Limestone φt ¼ 0 Limestone reference

Table 2. Summary of assumed Archie’s parameters andfluid properties for the synthetic example.

Variable Value Units

Connate water resistivity, Rw at 200°F 0.0203 Ω m

Connate water density, ρcw 1.11 g∕cm3

Connate water HI, HIcw 0.936 —

Connate water salt concentration 160,000 ppm NaCl

Archie’s factor, a 1 —

Archie’s porosity exponent, m 1.95 —

Archie’s saturation exponent, n 1.75 —


Dow

nloa

ded

05/2

9/14

to 1

28.8

3.16

7.15

5. R

edis

trib

utio

n su

bjec

t to

SEG

lice

nse

or c

opyr

ight

; see

Ter

ms

of U

se a

t http

://lib

rary

.seg

.org

/

Field examples of applicationIn this section, the interactive interpretation work-

flow is implemented for estimation of ρm, φt, Sw, andρhc in two field examples: (I) gas-bearing carbonatefield of dolomite lithology where ρm > 2.71 g∕cm3,and (II) oil-bearing shale formation where ρm <2.71 g∕cm3.

Field example I, gas-bearing carbonateThis field example consists of conventional wireline

nuclear and dual induction resistivity logs acquiredacross a gas-producing dolomite reservoir. Addition-ally, the well includes routine core measurements.Due to low reservoir pressure, deep mud-filtrate inva-sion affects the nuclear logs and even the deep resistiv-ity log, such that log-derived Shc is considerably lowerthan in situ Shc for water-base mud (Xu et al., 2012).Table 3 summarizes the assumed Archie’s parametersand fluid properties for the gas-bearing carbonate field.

Figure 5 shows the field measurements together withcore measurements, compared to results obtained withthe interactive interpretation. The neutron-density over-lay in Figure 5b emphasizes the matrix crossover be-cause the reservoir is primarily of dolomite lithology.The gas flag in Figure 5j, proportional to hydrocarbonpore volume, is most pronounced across XX45–XX55 mdespite the suppressed gas crossover in Figure 5b.Across the interval in Figure 5, the gas flag providesa qualitative and unequivocal indication of hydrocarbonsaturation despite mud-filtrate invasion and matrixcrossover.

The calculated ρhc in Figure 5h, with an averagevalue of 0.176 g∕cm3, confirms that the reservoir islargely saturated with gas. Conclusively, we implementcombined matrix and fluid volumetric analysis withthe SNUPAR-based solver, where methane gas of0.176 g∕cm3 is assumed as a component of the fluids,thus eliminating the water-filled assumption in the inde-pendent matrix analysis. Figure 5i shows cumulativeplots of the volumetric fractions of shale, quartz, calcite,dolomite, water, and gas, obtained from the SNUPAR-based solver. The estimated ρm and φt (Figure 5e and 5f,respectively), agree well with core measurements. Onthe other hand, log-derived Sw (Figure 5g) within inter-val XX08–XX32 m is considerably lower than core mea-surements. This behavior can be attributed to variationsin Archie’s parameters for differing rock types along thewell. Furthermore, Sw in core samples could increasedue to quick spurt loss in low-porosity, low-pressurereservoirs (Xu et al., 2012).

Field example II, oil-bearing shale exampleIn this example, nuclear and array induction resistiv-

ity logs are acquired in a well drilled with oil-base mudacross an oil-bearing shale formation from the EagleFord shale play. Table 4 describes the assumed fluidproperties and Archie’s parameters for the oil-bearingshale reservoir. Figure 6 shows field measurements,core measurements, and interpreted petrophysicalproperties for the oil-bearing shale example. Here,the SNUPAR-based matrix analysis assumes kerogen(C100H100O8 of density 1.4 g∕cm3), calcite, kaolinite,

Figure 3. Simulated well logs across the syn-thetic multilayer model. (a) GR log, (b) neu-tron and density-apparent porosities on alimestone scale, (c) array induction apparentresistivity logs, and (d) PEF log. Refer toTable 1 for a description of assumed layerproperties.


Dow

nloa

ded

05/2

9/14

to 1

28.8

3.16

7.15

5. R

edis

trib

utio

n su

bjec

t to

SEG

lice

nse

or c

opyr

ight

; see

Ter

ms

of U

se a

t http

://lib

rary

.seg

.org

/

and illite as components of the matrix. Figure 6e com-pares ρm from the interactive analysis to core measure-ments and elemental capture spectroscopy (ECS)lithology analysis. The ECS-derived ρm (dashed blue

curve) is significantly larger than the core ρm (bluecircle points). This result is attributed to the exclusionof the low-density kerogen matrix from the ECS analy-sis. The matrix density ρm from SNUPAR-based matrix

Figure 4. Interpretation results for the synthetic example using the interactive interpretation workflow. (a) Interpreted matrixdensity from SNUPAR-based matrix solver, (b) neutron-density overlay showing shale-corrected neutron log, matrix, and fluidcrossover characteristics, (c) neutron and density-apparent water-filled logs from SNUPAR-based matrix solver, (d) interactiveflag indicators showing matrix effect and gas flag, (e) corrected neutron-density overlay, (f) estimated total porosity, (g) estimatedwater saturation, and (h) estimated hydrocarbon and fluid densities. Refer to Table 1 for a description of layer properties.

Table 3. Summary of assumed fluid properties and Archie’s parameters for field example I, gas-bearing carbonate.






Mud-filtrate water resistivity, Rmf at 96°F 0.84 Ω m

Mud-filtrate water density, ρmf 1 g∕cm3

Mud-filtrate HI, HImf 1 —

Mud-filtrate water salt concentration 5147 ppm NaCl





Dow

nloa

ded

05/2

9/14

to 1

28.8

3.16

7.15

5. R

edis

trib

utio

n su

bjec

t to

SEG

lice

nse

or c

opyr

ight

; see

Ter

ms

of U

se a

t http

://lib

rary

.seg

.org

/

analysis (red curve) agrees well with core measure-ments. The resulting fluid crossover, in Figure 6b, aftermatrix-hydrogen and shale-hydroxyl corrections, is dueto the combined effects of ρm (less than 2.71 g∕cm3 oflimestone), fluid density, and fluid HI. It is found thatthe interactive analysis yields a relatively constantρhc of 0.747 g∕cm3 for the interval in Figure 6h. Further-more, estimated φt and Sw from the interactive analysis

(Figure 6f and 6g respectively), agree well with coremeasurements.

Fluid zone identificationConventional methods for fluid contact identification

include interpretation of pressure gradients due to fluiddensity differences in the reservoir hydrostatic column.Additionally, impermeable sealing or geological barriers,

Figure 5. Interpretation results for field example I, gas-bearing carbonate reservoir, using the interactive analysis workflow.(a) GR log, (b) neutron and density porosities on limestone scale, (c) dual-induction resistivity logs, and (d) PEF log. (e) Matrixdensity, (f) total porosity, and (g) water saturation from core measurements and interactive analysis. (h) Calculated fluid densitiesshowing a gas cutoff of 0.25 g∕cm3. (i) Volumetric concentrations of rock and fluid components from the SNUPAR-based solver.(j) Gas flag from interactive analysis workflow.

Table 4. Summary of assumed fluid properties and Archie’s parameters for field example II, oil-bearing shale.








Archie’s saturation exponent, n 2 —


Dow

nloa

ded

05/2

9/14

to 1

28.8

3.16

7.15

5. R

edis

trib

utio

n su

bjec

t to

SEG

lice

nse

or c

opyr

ight

; see

Ter

ms

of U

se a

t http

://lib

rary

.seg

.org

/

often at residual saturations, prevent hydraulic commu-nication between fluid zones such that a higher densityfluid resides above a lower density fluid in the hydro-carbon column.

Occasionally, when pressure measurements are un-available or expensive to acquire, well logs are used toinfer fluid zones. In this section, the estimated ρhc logyielded by the interactive interpretation is used for aquick-look qualitative identification of hydrocarbonzones and fluid contacts along the reservoir columnin two more field examples.

Field example III, identification of hydrocarbonfluid contacts in a North Sea siliciclastic reservoir

This example consists of a siliciclastic reservoirlocated in the central North Sea, where rock formationsconsist of noncalcareous mudstones interbedded withshaly sand deposits (Heidari et al., 2012). Figure 7a–7dshows nuclear and array induction resistivity measure-ments acquired in a vertical well drilled with oil-basemud, while Table 5 summarizes the assumed propertiesand Archie’s parameters for the siliciclastic reservoir. Inaddition, available pressure data in Figure 7f describethree distinct and approximately constant pressure

gradients. Water saturation, Sw, shown in Figure 7e,estimated using the dual-water resistivity model, indi-cates that the hydrocarbon column exhibits a completecapillary transition with an aquifer below X660 m. Pres-sure gradients identify three fluid zones, i.e., gas of0.263 g∕cm2 density, oil of 0.647 g∕cm3 density, andan aquifer at residual hydrocarbon saturation with con-nate water of 1.005 g∕cm3 density, distinguished byred, green, and black intervals, respectively. Figure 7gshows the fluid densities, fluid zones, and fluid con-tacts, where estimated fluid densities, ρf ;p, from pres-sure gradients are juxtaposed with ρf estimated withthe interactive interpretation.

Qualitatively, ρf (Figure 7g) and φt (Figure 7h) fromthe interactive interpretation agree well with pressureand core measurements, respectively, except acrossthe interval between X550 and X600 m. This intervalconsists of highly interbedded sand-shale sequences;evident from the GR log in Figure 7a, whereby log-derived φt and ρf are significantly influenced byshoulder-bed effects, and depth-by-depth analysis isinadequate. Note that the estimated ρf from the interac-tive interpretation agrees well with ρf ;p across the thickbed layers in the gas zone. Nonetheless, assuming no

Figure 6. Interpretation results for field example II, oil-bearing shale reservoir, using the interactive analysis workflow. (a) GRlog, (b) neutron and density porosities on limestone scale, (c) array induction resistivity logs, and (d) PEF log. (e) Matrix density,(f) total porosity, and (g) water saturation from core measurements and interactive analysis. (h) Calculated fluid densities showinga gas cutoff of 0.25 g∕cm3.


Dow

nloa

ded

05/2

9/14

to 1

28.8

3.16

7.15

5. R

edis

trib

utio

n su

bjec

t to

SEG

lice

nse

or c

opyr

ight

; see

Ter

ms

of U

se a

t http

://lib

rary

.seg

.org

/

reservoir compartmentalization and good hydrauliccommunication, the gas-oil contact is located atX600 m, while the oil-water contact is located atX655 m where the water zone is at residual hydrocar-bon saturation.

Field example IV, identification of reservoircompartments in the deepwater Gulf of Mexico

In this example, the reservoir consists of channellevees located in the deepwater Gulf of Mexico,where formations consist of unconsolidated shaly sandintervals and are primarily saturated with oil. Table 6summarizes the assumed properties used in theinteractive interpretation with a dual-water resistivitymodel. In Figure 8, the panels describe well logs andinterpretation results across a hydrocarbon-saturatedinterval in the Gulf of Mexico reservoir. Figure 8eshows that average total porosities in the clean andshaly sand layers are 0.2721 and 0.1724, respectively.In Figure 8c and 8g, we observe a gas-saturated reser-voir compartment between X817 and X819 m, wheregas density is 0.144 g∕cm3 and the neutron-densityoverlay exhibits significant gas crossover. The pri-mary oil-saturated zone, between X778 and X802 m,

with an estimated oil density of 0.43 g∕cm3 is abovethe gas-saturated compartment at X817–X819 m. Thecompartmentalization of the gas layer is possible be-cause hydraulic communication is severed betweenthe oil and gas zones by the interleaving imperme-able nonnet shale barriers. This example verifies the

Figure 7. Fluid zone interpretation results for field example III, North Sea siliciclastic reservoir. (a) GR log, (b) PEF log, (c) neu-tron and density porosities on limestone scale, and (d) array induction resistivity logs. (e) Estimated water saturation, (f) pressuremeasurements, and (g) fluid densities from interactive analysis and pressure gradients. (h) Total porosity from core measurementsand interactive analysis.

Table 5. Summary of assumed properties and Archie’sparameters for field example III, North Seasiliciclastic reservoir.









Shale porosity, φsh 0.10 v∕vShale resistivity, Rsh 1.50 Ω m


Dow

nloa

ded

05/2

9/14

to 1

28.8

3.16

7.15

5. R

edis

trib

utio

n su

bjec

t to

SEG

lice

nse

or c

opyr

ight

; see

Ter

ms

of U

se a

t http

://lib

rary

.seg

.org

/

capability of the interactive interpretation workflow todistinguish between oil- and gas-saturated layers, irre-spective of formation lithology. The workflow alsoprovides an efficient qualitative method for identifica-tion of reservoir compartments separated by sealingbarriers.

ConclusionsThe interactive interpretation workflow rescales the

neutron-density overlay with corrected neutron anddensity-apparent porosities in a variable matrix scale,for independent characterization of fluid effects. Itwas found that the SNUPAR-based matrix analysis, as-suming freshwater-filled formations, renders accurateestimations of matrix density even across hydrocar-bon-saturated intervals. Such a result is due to the factthat formation fluids have negligible or no effect on PEFand GR logs. One limitation of the SNUPAR-based ma-trix analysis is that a priori qualitative knowledge ofmatrix components, i.e., lithology, clay mineral, etc.,is essential for accurate estimation of matrix density.This is achieved by preliminary lithology or matrix iden-tification crossplots, e.g., PEF-ρb, thorium-potassium,and PEF-potassium crossplots. Furthermore, the work-flow assumes minimal shoulder-bed effects such thatdepth-by-depth analysis is adequate for SNUPAR calcu-lations. The uncertainty in estimated ρhc increases inthinly bedded intervals with pronounced shoulder-bed effects.

The merits of the SNUPAR-based interactive inter-pretation workflow and its contributions to the practiceof interpretation include the following:

Figure 8. Fluid zone interpretation results for field example IV, deepwater Gulf of Mexico reservoir. (a) GR log, (b) PEF log,(c) neutron and density porosities on sandstone scale, and (d) array induction resistivity logs. (e) Total porosity, (f) water sat-uration, and (g) fluid density logs estimated using the interactive analysis.

Table 6. Summary of assumed properties and Archie’sparameters for field example IV, deepwater Gulf ofMexico reservoir.


Connate water resistivity, Rw at 150°F 0.030 Ω −m







Shale porosity, φsh 0.15 v∕vShale resistivity, Rsh 1.0 Ω −m


Dow

nloa

ded

05/2

9/14

to 1

28.8

3.16

7.15

5. R

edis

trib

utio

n su

bjec

t to

SEG

lice

nse

or c

opyr

ight

; see

Ter

ms

of U

se a

t http

://lib

rary

.seg

.org

/

• unequivocal identification of hydrocarbon-satu-rated zones,

• estimation of model-consistent/lithology-indepen-dent formation porosity, and

• calculation of hydrocarbon density for gas/oil-zone identification.

It was shown that the workflow incorporates inter-active matrix corrections such that Gaymard-Poupon’sformulation for lithology-independent porosity and hy-drocarbon identification can be implemented for anyneutron-density matrix scale and lithology (clean and/or shaly), especially in wells with limited data.

Synthetic and field examples of application indicatethat lithology-independent porosity and hydrocarbondensity can be efficiently estimated from conventionalnuclear and resistivity logs for reliable and quantita-tive detection and appraisal of hydrocarbon-saturatedsweet spots and nonviable zones. Furthermore, identi-fication of fluid types in the reservoir column provides aqualitative means for determining fluid contacts andreservoir compartments.

Symbols and nomenclature

a = Winsauer factor in Archie’s equation ( )AmBe = Americium-berylliumAO10 = 25.4-cm (10-in) array induction one-foot

resistivity (Ω −m)AO30 = 76.2-cm (30-in) array induction one-foot

resistivity (Ω −m)AO90 = 228.6-cm (90-in) ray induction one-foot

resistivity (Ω −m)CNT = Schlumberger-compensated neutron

toolgd = Neutron-porosity calibration function,

in dolomite unitsgl = Neutron-porosity calibration function,

in limestone unitsGR = Gamma ray American Petroleum Insti-

tutegs = Neutron-porosity calibration function,

in sandstone unitsHI = Hydrogen index ( )ILD = Deep induction resistivity (Ω −m)ILM = Medium induction resistivity (Ω −m)Lm = Neutron migration length (cm)m = Archie’s porosity exponentM = Number of componentsn = Archie’s saturation exponentNMR = Nuclear magnetic resonancePEF = Photoelectric factor (b∕e)Rt = True resistivity (Ω −m)Rw = Water resistivity (Ω −m)SFLU = Spherically focused resistivity (Ω −m)SNUPAR = Schlumberger nuclear parameter calcu-

latorSw = Water saturation (%)UTAPWeLS = The University of Texas at Austin petro-

physical and well-log simulatorVi = Volumetric concentration (v∕v)

V sh = Volumetric concentration of shale (v∕v)Σ = Capture cross section (cu)ρ = Density (g∕cm3)Δ = Departureφ = Apparent porosity (v∕v)ρm = Matrix density (g∕cm3)φt = Total porosity (v∕v)δΣ = Neutron fluid effect parameter ( )

Subscripts

b = Bulkcorr = Correctedcw = Connate watere = Electronf = Fluidg = Gashc = Hydrocarboni = Component indexker = Kerogenmf = Mud-filtrateN = Neutronnsh = Nonshalesh = Shalet = Totalwf = Water-filled

AcknowledgmentsThe work reported in this paper was funded by the

University of Texas at Austin’s Research Consortiumon Formation Evaluation, jointly sponsored by Afren,Anadarko, Apache, Aramco, Baker-Hughes, BG, BHPBilliton, BP, Chevron, ConocoPhillips, COSL, ENI,ExxonMobil, Halliburton, Hess, Maersk, Marathon OilCorporation, Mexican Institute for Petroleum, Nexen,ONGC, OXY, Petrobras, PTT Exploration and Produc-tion, Repsol, RWE, Schlumberger, Shell, Statoil, Total,Weatherford, Wintershall, and Woodside Petroleum Lim-ited. We are indebted to Shell Oil Company for providingthe core and well-log measurements used in this study.

ReferencesEllis, D. V., and J. M. Singer, 2007, Well logging for earth

scientists: Springer.Fertl, W. H., and D. J. Timko, 1971, A distinction of oil and

gas in clean and shaly sands as derived from well logs:The Log Analyst, 12, 21–32.

Gaymard, R., and A. Poupon, 1968, Response of neutronand formation density logs in hydrocarbon bearing for-mations: The Log Analyst, 9, 3–12.

Heidari, Z., C. Torres-Verdín, and W. Preeg, 2012, Improvedestimation of mineral and fluid volumetric concentra-tions in thinly bedded and invaded formations: Geophys-ics, 77, no. 3, WA79–WA98, doi: 10.1190/geo2011-0454.1.

Mao, Z.-Q., 2001, The physical dependence and the corre-lation characteristics of density and neutron logs: Pet-rophysics, 42, 438–443.


Dow

nloa

ded

05/2

9/14

to 1

28.8

3.16

7.15

5. R

edis

trib

utio

n su

bjec

t to

SEG

lice

nse

or c

opyr

ight

; see

Ter

ms

of U

se a

t http

://lib

rary

.seg

.org

/

http://dx.doi.org/10.1190/geo2011-0454.1



McKeon, D. C., and H. D. Scott, 1989, SNUPAR — Anuclear parameter code for nuclear geophysics applica-tions: IEEE Transactions on Nuclear Science, 36, 1215–1219, doi: 10.1109/23.34634.

Passey, Q. R., S. Creaney, J. B. Kulla, F. J. Moretti, and J. D.Stroud, 1990, A practical model for organic richnessfrom porosity and resistivity logs: AAPG Bulletin, 74,1777–1794.

Quintero, L. F., and Z. Bassiouni, 1998, Porosity determina-tion in gas-bearing formations: Presented at SPE PermianBasin Oil and Gas Recovery Conference, SPE 39774.

Schlumberger Limited, 2009, Log interpretation charts,http://www.slb.com/resources/publications/books/log_charts.aspx.

Spears, R. W., 2006, Lithofacies-based corrections to den-sity-neutron porosity in a high-porosity gas- and oil-bearing turbidite sandstone reservoir, Erha field, OPL209, Deepwater Nigeria: Petrophysics, 47, 294–305.

Voss, B., C. Torres-Verdín, A. Gandhi, G. Alabi, and M. Lem-kecher, 2009, Common stratigraphic framework to sim-ulate well logs and to cross-validate static and dynamicpetrophysical interpretation: Presented at Transactionsof the SPWLA, 50th Annual Logging Symposium.

Xu, C., Z. Heidari, and C. Torres-Verdín, 2012, Rock clas-sification in carbonate reservoirs based on static anddynamic petrophysical properties estimated from con-ventional well logs: Presented at SPE Annual TechnicalConference and Exhibition, SPE 159991.

Olabode Ijasan received a B.S.(2006) in electrical and electronicsengineering from the University ofLagos, Nigeria, and an M.S. (2010) inpetroleum engineering from the Uni-versity of Texas at Austin, where heis currently a Ph.D. candidate devel-oping modeling-based techniques forinterpreting borehole nuclear and

resistivity measurements. His research interests includewell-log modeling, inversion, and petrophysical interpreta-tion. He is a member of SEG, SPE, SPWLA, and IEEE.

Carlos Torres-Verdín received aPh.D. (1991) in engineering geosci-ence from the University of Californiaat Berkeley. During 1991–1997, heheld the position of research scientistwith Schlumberger-Doll Research.From 1997–1999, he was a reservoirspecialist and technology championwith YPF (Buenos Aires, Argentina).

Since 1999, he has been affiliated with the Departmentof Petroleum and Geosystems Engineering of the Univer-sity of Texas at Austin, where he is currently a full profes-sor, holds the Zarrow Centennial Professorship inPetroleum Engineering, and conducts research on bore-hole geophysics, formation evaluation, well logging, andintegrated reservoir characterization. He is the founderand director of the Research Consortium on FormationEvaluation at the University of Texas at Austin, which iscurrently sponsored by 32 companies. He has publishedmore than 115 refereed journal papers and 130 conferencepapers, has served as a guest editor for Radio Science, asan associate editor for the Journal of ElectromagneticWaves and Applications, SPE Journal, and Petrophysics(SPWLA), and is currently associate editor for GEOPHYSICS

and an editorial board member of The Leading Edge

(SEG). He is the recipient of the 2006 Distinguished Tech-nical Achievement Award from the SPWLA, the 2008 For-mation Evaluation Award from the SPE, the 2003, 2004,2006, and 2007 Best Paper Awards in Petrophysics bythe SPWLA, the 2006 Best Presentation Award, and the2007 Best Poster Award by the SPWLA.

Biographies and photographs of the other authors arenot available.


Dow

nloa

ded

05/2

9/14

to 1

28.8

3.16

7.15

5. R

edis

trib

utio

n su

bjec

t to

SEG

lice

nse

or c

opyr

ight

; see

Ter

ms

of U

se a

t http

://lib

rary

.seg

.org

/

http://dx.doi.org/10.1109/23.34634

http://dx.doi.org/10.1109/23.34634

http://dx.doi.org/10.1109/23.34634

http://www.slb.com/resources/publications/books/log_charts.aspx





interpretation of porosity and fluid constituents from ... · pdf fileinterpretation of...

Documents