Log analysis for unconventionals

To understand the petrophysical properties of an unconventional reservoir, well log data are used to investigate the rock properties of the shales in the areas of interest. The petrophysical evaluation comprises quality checking and editing of the available well logs. Additionally, the description of their properties, evaluation of the intervals of interest, and calculations to estimate the shale or clay volumes, TOC content, and brittleness index for the available wells are the main steps that should be followed in the initial log analysis.

Data requirements

• The suite of logs should include gamma ray, dipole sonic, deep resistivity, neutron porosity, and density measurement.
• In addition to log data, at least one well in the study area should contain spectral gamma-ray data.
• Core TOC evaluation is recommended to validate the results.

Shale volume estimation

The estimation of the shale volume in the zone of interest can be performed in the available wells using the normalized version of the Gamma-Ray log (GRn), and using both neutron and density porosity logs.

Fig. 1: GR vs. Corrected Spectral GR for Well B, color-coded by formation. The correlation between both GR responses allows an adequate linear fit of 0.981. The linear empirical relationship is used to calculate the corrected GR for uranium in the Well A. (Modified from author's research results ).

Using gamma-ray logs

Following the concept that the increase in radioactivity of the organic-rich shales is related to their organic matter content, the GR and spectral GR responses need to be corrected for uranium before estimating clay content. This element forms compounds that sorbs to clays and organic material in both cases where their depositional environment is anoxic marine or oxidizing lacustrine. In case spectral GR data are not available for a particular well (Well A), its correction for uranium should be conducted by means of a linear empirical relationship constructed using both, the corrected spectral GR of another well in the area of interest (Well B) plotted against its original log (Figure 1). The following equation is used for the GR correction for uranium:

$GRc=GR-8U$

Where, GRc = corrected GR, GR = total GR, and U = uranium in ppm.

After correcting the GR for uranium to remove the effect of the organic matter, the normalization of the Gamma log is conducted as a method of reducing mud weight and hole size effects The normalization process follows the assumption that all pure shales in an area have the same GR values, and that all clean sands have the same GR log reading using the following equation:

$GRn=GRmin+[(GRmax-GRmin)*({\frac {GR-GRlow}{GRhigh-GRlow}})]$

Where, GRn = normalized corrected GR in API units, GRmin = GR clean sand value to normalize to, GRmax = GR shale value to normalize to, GR = total GR, GRlow = GR clean sand value in the well/zone, GRhigh = GR shale value in the well/zone.

To calculate Vshale from GRn, the following methodology could be applied:

$Vsh_{\mathrm {GRn} }={\frac {GRn-GR_{\mathrm {0} }}{GR_{\mathrm {100} }-GR_{\mathrm {0} }}}$

Where, Vsh_GRn = shale volume from normalized GR log, GRn = normalized GR, GR₀ = GR log reading in 100% clean zone, GR₁₀₀ = GR log reading in 100% shale.

Fig. 2: Well A logs. From left to right, total GR, the corrected GR for uranium and its normalized readings (thick black curve), neutron porosity and density, density porosity and the comparison between the estimated shale volumes (Vsh_Grn, Vsh_NPHI, Vsh_PHI). Shale volumes calculated with the three methods follow the separation trend between NPHI and RHOZ except in the dolomitic formation (in grey). Since the high-density porosity reading in the shales is due primarily to organic matter rather than porosity, values of shale volume in these two members are considered underestimated. Vsh_GRn and Vsh_NPHI are taken as better representations of the relative volume fraction of shales in the formations of interest (red, yellow and green). (Modified from author's research results ).

Using porosity logs

Given the neutron porosity log and the densities of clay (2.68 g/cc) and shale (2.35 g/cc), the volume fraction of shale (Vsh_NPHI) can be calculated via determination of the clay-bound water by setting values representative of clean sand and pure shale that correspond to the maximum and minimum value of the neutron porosity log, respectively.

Another method for estimating the shale volume uses both the density and neutron porosity logs. The linear interpolation of the separation between these two logs corresponds to the following algebraic formula to solve for shale volume:

$Vsh_{\mathrm {PHI} }={\frac {NPHI-PHID}{NPHIsh-PHIDsh}}$

Where, Vsh_PHI = Vshale from porosity logs, NPHI = neutron log reading in zone of interest, PHID = density log porosity reading in zone of interest, NPHIsh = Neutron log reading in 100% shale, PHIDsh = apparent density porosity in shale. The density porosity and neutron porosity of pure shale are 0 and 0.4, respectively. The neutron porosity at pure shale is taken from the density neutron cross-plot at the depths of the shales of interest.

In general, the shale volumes calculated should follow the separation trend between NPHI and density (RHOZ) curves: the higher the separation, the higher the shale volume (Figure 2). In cases where the lithology is characterized by dolomites, the separation between the two curves is not a function of shale.

Fig. 3: Well A logs. From left to right, total GR and the normalized corrected GR (GRn), neutron porosity and density, P-wave sonic, and the superposition of the deep resistivity and the GRn curves. The cross-over between resistivity and gamma-ray readings (in yellow) is indicative of high TOC content. (Modified from author's research results ).

TOC estimation

Total organic carbon (TOC) is an important parameter in the evaluation of kerogen-rich unconventional reservoirs.

Heslop (2010) method

An initial identification of zones with high TOC content can be performed using the Heslop (2010) method. The increase in GR readings and deep resistivity (Rt) can be related to TOC within shales and there is a relationship between the curves associated with these two petrophysical properties. In a clean matrix, the GR typically decreases whereas the resistivity increases. On the other hand, in non-source shales (i.e. low TOC content), the GR increases while the Rt decreases. Since these two curves tend to “hour-glass” when plotted using conventional scales, reversing and selecting appropriate values for the Rt scale causes the GR and the Rt curves to track, except in source shales where both the GR and Rt values increase due to the TOC content.

The crossover between the GR and Rt curves is indicative of the TOC effect in the shale members in which these two properties increase, and the hydrocarbon potential is observed on logs as lower density relative to shale density, “hot” GR response, increased P-wave sonic (DT) relative to shale DT, and increased neutron porosity relative to shale neutron (Figure 3).

Fig. 4: Estimation of the TOC from the Passey (1990) Method for Well A. From left to right, total and normalized GR, deep resistivity, neutron porosity, P-wave sonic, density logs, and the comparison between their respective estimated values of TOC content. The inorganic formation in grey was taken as the baseline for ∆logR calculation in Passey’s formulation. Note how the curves approach to each other at the base of the grey formation in the second track. (Modified from author's research results ).

Passey (1990) method

Assuming that resistivity logs respond to fluids, while porosity logs (sonic, density, or neutron) respond to kerogen/matrix and fluids, the Passey (1990) method combines these two type of logs to estimate TOC in organic-rich rocks. Using either porosity curve, the method relies on porosity and deep resistivity readings separating from each other in organic-rich rocks, whereas in organic-lean rocks, the two curves overlie. The separation between the two curves or the scaled difference (∆logR) between them is related to the TOC content through the level of thermal maturation (LOM) by the following relation:

$TOC(wt\%)={\text{∆logR}}*10^{(2.297-0.1688*LOM)}$

Where, ∆logR = scaled difference between deep resistivity and density logs and LOM = level of organic maturity.

$\Delta logR={\text{log}}_{\mathrm {10} }{\frac {RT}{RT_{\mathrm {baseline} }}}-[ScalingFactor*(PHI-PHI_{\mathrm {baseline} })]$

Where, RT = deep resistivity log in ohm/m, RT_baseline = resistivity in the organic-lean zone in ohm/m, PHI = porosity log (i.e. sonic, density or neutron logs), PHI_baseline = porosity log reading in the organic-lean zone. The scaling factor is calculated after baselining the two curves in the organic-lean zone.

If the type of organic matter is known, the level of organic maturity (LOM) can be determined from a variety of measurements such as vitrinite reflectance, thermal alteration index (Tmax) or Rock-Eval pyrolysis. In over-mature shale reservoirs with LOM values greater than 10.5, the limit of calibration of maturity to TOC is reached. Figure 4 shows the estimated TOC logs using sonic, neutron porosity, density, and deep resistivity logs. Note the good agreement between the estimations calculated from the density and neutron porosity logs.

Fig.5: From left to right, total and normalized GR, deep resistivity (AT90), neutron porosity (NPHI), P-wave sonic (DT) and density (RHOZ) logs of well A, the comparison between the estimated values of TOC by the Passey (1990), Vernik and Landis (1996) and Schmoker and Hester (1983) methods, and the resistivity (AT90)-density (RHOZ) overlay for ∆logR calculation, where areas highlighted in purple correspond to high organic-matter zones. (Modified from author's research results ).

Schmoker and Hester (1983) method

Assuming that the change in density of the formation is due to the presence or absence of low-density organic matter, an empirical approach was developed by Schmoker (1979) to quantitatively estimate TOC in Devonian shales from log data. The methodology was then refined for the Bakken black shales which were treated as a four-component system consisting of rock matrix, interstitial pores, pyrite, and organic matter. High-density minerals other, than pyrite, are assumed to comprise a fixed (but unknown) percentage of the rock matrix. The formulation for TOC calculation with the Schmoker method is shown as follows:

$TOC(wt\%)={\frac {154.497}{\rho }}-57.261$

Where, ρ = bulk density in g/cc, and the constants were specifically calculated for the upper and lower shale members of the Bakken Formation based on an organic matter density of 1.01 g/cc, a matrix density of 2.68 g/cc, and a ratio between weight percent of organic matter and organic carbon of 1.3. The study reported an average of organic-carbon content in the upper and lower shale members of 12.1 wt.% and 11.5 wt.%, respectively, calculated at 159 locations in North Dakota and 107 in Montana.

Vernik and Landis (1996) method

Vernik and Landis (1996) also proposed an empirical formulation for TOC estimation using the density log as follows:

$TOC(wt\%)=67*{\frac {RHO_{\mathrm {k} }(RHO_{\mathrm {s} }-RHOB)}{RHOB(RHO_{\mathrm {s} }-RHO_{\mathrm {k} })}}$

Where, RHOB = density log, RHO_s = clay density and RHO_k = kerogen density.

Pitfalls in TOC estimation

• Common TOC from density logs methods are highly susceptible to borehole effects (i.e. washouts) and tend to ignore lithology changes. However, if heavy minerals (e.g., pyrite) are present as trace minerals, there is practically no variation in porosity, and there are good borehole conditions, using density logs could be a good method to estimate TOC.
• Since the Passey formulation utilizes resistivity and porosity logs, and Vernik’s method is based on density logs, their associated TOC estimations are conditioned by their response, readings, and implications. Porosity logs incorporate a lithology response in addition to porosity, the noisy resistivity readings might cause underestimation of TOC, and the complication of borehole related errors associated with the utilization of density logs may influence the final calculations (Figure 5).

Fig. 6: Poisson’s ratio vs. Eρ cross-plot of the formations of interest colored by unit in the well A. The domain Eρ/ν can be used as a brittleness indicator for the shale members. Brittle-shale zones and high-ductile zones in the red and blue rectangles, respectively. The lower shale member is generally less brittle than the upper shale member and the middle member shows points with moderate brittleness. (Modified from author's research results ).

Brittleness estimation

Finding areas in the shale play that are brittle is important in the development of a fracture fairway large enough to connect the highest amount of rock volume during the hydraulic fracturing process (Perez, 2013). Based on the amount of plastic deformation that the rock undergoes before a fracture occurs, its response to stress can be differentiated between brittle and ductile. Brittle rocks can potentially generate microfractures that could remain open during hydraulic fracturing and therefore they tend to break easily. On the other hand, ductile rocks absorb a high amount of energy before fracturing and as a consequence, they deform plastically.

The measure of stored energy before failure is a function of rock strength, lithology, texture, effective stress, temperature, fluid type, diagenesis and TOC and the analysis to differentiate brittle from ductile rocks has been key to stimulation process in shale oil and gas reservoirs where brittleness is mainly controlled by quartz content and ductility is controlled by clay, calcite, and total organic content.

Since the mineralogy of the shales is often defined by quartz and clay, the response to stress could be difficult to associate directly to brittleness or ductility. Hence, the brittleness needs to be more accurately estimated by a renormalization of Young’s modulus (E) and Poisson's ratio (ν) over the zone of interest according to the Rickman et al., (2008) relations:

$E_{\mathrm {BRIT} }={\frac {E-E_{\mathrm {min} }}{E_{\mathrm {max} }-E_{\mathrm {min} }}}$, $\nu _{\mathrm {BRIT} }={\frac {\nu -\nu _{\mathrm {min} }}{\nu _{\mathrm {max} }-\nu _{\mathrm {min} }}}$, $BI_{\mathrm {ave} }={\frac {E_{\mathrm {BRIT} }-\nu _{\mathrm {BRIT} }}{2}}$

Where, E_BRIT = normalized Young’s modulus, ν_BRIT = normalized Poisson’s ratio, and BI_ave = average brittleness index.

Since the average brittleness is defined by a subjective and arbitrary renormalization process with defined upper and lower E and ν bounds, it only provides relative information and it should be compared and correlated with an estimation using fractions of brittle minerals or other attributes such the product of Young’s modulus and density Eρ (Figure 6). Sharma and Chopra (2015) suggest that Eρ accentuates lithology detection in terms of brittleness and could be useful for the determination of E from seismic data by way of inversion. In this sense, impedance inversion of seismic data takes place using the resulting P and S-impedances to determine Eρ without the requirement of density which is difficult to derive from seismic data unless long offset information is available.

External links

• Crain's Petrophysical Handbook - [1]
• Super Petrophysics Cheatsheets - [2]