This paper presents a case study showing that traditional pore pressure prediction techniques can be adapted to predict pore pressure in an unconventional play by using a newly defined Pressure Reference Trend (PRT) in-lieu of a Normal Compaction Trend (NCT) as used in conventional, or traditional, pore pressure prediction. The PRT is not linked to the expected compaction behavior of the rock (as inferred from an NCT) but it is simply an empirical depth trend from which the pore pressure can be predicted using industry standard formulae. Rather than constraining the surface and matrix value for an NCT using sensible geological parameters, the final position of the PRT in velocity-depth space is a function of the measured pressure, to which the trend is calibrated, combined with a lateral shift towards higher velocity/density due to tectonic uplift, secondary compaction, and chemical diagenesis that has occurred over the geological history of the basin.
Traditional pore pressure prediction typically assumes that all the shales are geologically young with low temperatures, are at their maximum burial depth, and have a demonstrable porosity/effective stress relationship where disequilibrium compaction is the mechanism of pressure generation (e.g., Osborne & Swarbrick, 1997). The critical assumption underpinning traditional pore pressure prediction is that the variations observed in specific wireline data (Vp, Vs, Rho, Neutron, Resistivity) are varying solely due to changes in porosity, and that the porosity is controlled by the pore pressure. High pore fluid pressure results in a high porosity rock as the pore fluid was unable to escape during further sedimentation and compaction; hence the fluid is trapped within the rock preserving high porosity. The follow-on assumption is that the wireline log response can be converted into pore pressure using traditional methods (Eaton Ratio method (Eaton, 1975); Equivalent Depth Method (Foster & Whalen, 1966); Vp-Effective Stress method (Bowers, 1994)) into a magnitude of pore pressure via a relationship between vertical stress (overburden), pore pressure, and the vertical effective stress (Terzaghi, 1943).
Shales in unconventional plays are (or were) at high temperature leading to diagenetic alteration of the mineralogy, are often dramatically uplifted, and have been affected by chemical processes in addition to mechanical compaction such that porosity may not be solely relatable to effective stress. Furthermore, the link between pore pressure and log response may be further disrupted by the presence of organic material (high TOC) or free gas in the pore space, and whilst these challenges are not unique to unconventional plays they are more likely to be present. An increase in TOC has been shown to significantly lower the magnitudes of velocity and density (Passey et al., 1990). Gas in the pore space has a similar slowing effect on the Vp data, although whilst the TOC effect has a linear relationship with log response the gas effect is not as straight forward to correct for due to the non-linear effect of gas saturation. Slow velocity (either due to TOC or to free gas in the pores) and low density are typically attributed to an increase in pore pressure, as they imply high porosity, so this effect needs to be removed from log data in order to correctly predict pore pressure. The actual presence of higher pore pressure can lead to “sweet spot” preservation and even natural fracturing that can enhance production without the need for artificial stimulation hence the need for an accurate pore pressure prediction method/workflow. It is because of the disconnect between the measured elastic response (porosity) from pore pressure as per a simple mechanical compaction model that has resulted in the presumption that pore pressure prediction cannot be performed in unconventional plays.
Overview of Pore Pressure Prediction in Unconventional Plays
In spite of the issues described above, it can still be possible to derive a relationship between wireline data and the pressure magnitude; a brief summary of prior work is given here. Ebrom et al. (2003) demonstrated a relationship between Vs and pore pressure which circumvents the effect of gas on the Vp data. The work was undertaken in a conventional play (offshore Trinidad) but within a gas chimney and used the Eaton Ratio approach with a modified exponent and a traditionally defined Normal Compaction Trend (Figure 1).
More recently, Zhang and Wieseneck (2011) presented a case study from the Haynesville and Bossier plays in the southern United States where they were able to link direct pressure data (kicks and/or DFIT-type data) to velocity. In this example, a pseudo-Vp was computed from measured Vs using a calibrated Castagna approach to avoid the slowing effect of gas on measured Vp. This approach was further modified by Couzens-Shultz et al. (2013) who showed that Vs data could be used to predict pore pressure in the same unconventional plays without needing to use the Castagna approach to derive a pseudo-Vp.
In both the Zhang and Wieseneck (2011) and Couzens-Shultz et al. (2013) studies the method of pore pressure prediction was the velocity-effective stress cross-plot method (Bowers, 1994) as the authors had access to large datasets which suit that technique. The theory that supports the Bowers approach is a normal compaction approach but as it is a direct correlation of measured velocity and measured pore pressure it can be used to derive an empirical pore pressure relationship independent of the intrinsic geological constraint of the Eaton Ratio and Equivalent Depth methods and, therefore, is more suitable to unconventional plays (e.g., Rauch-Davies et al. 2018). In the case study presented here the data are more sparse hence the need to use the Eaton Ratio and Equivalent Depth methods which are dependent on some form of a normal compaction trend.
Pressure Reference Trend (PRT) Method
As stated previously, two of the most common methods for pore pressure prediction in conventional plays are the Eaton Ratio method and the Equivalent Depth method (EDM). The two methods both attempt to predict the vertical effective stress, which can then be subtracted from the vertical stress (overburden) to estimate the pore pressure. The vertical effective stress is the grain-to-grain contact stress, the component of the total vertical stress that the rock framework supports as opposed the to the pore fluid pressure. Both methods require a normal compaction trend (NCT) to be derived before a pressure prediction can be made.
Normally compacted shales are by definition shales that are normally pressured. The common shape of the NCT is a typical porosity-depth profile where the rate of porosity decline with increasing depth (and effective stress) is most rapid in the shallow section and slows down as the porosity declines and the rock becomes more resistant to grain collapse, grain rearrangement, and other processes during mechanical compaction (Figure 1; upper images). Where the porosity reduction with depth no longer follows the NCT then the process is termed disequilibrium compaction, or undercompaction, (Figure 1; lower images). The NCT can be constrained geologically using realistic surface and matrix parameters, for example a velocity close to that of water at the surface and a matrix velocity that is 90-95% of the grain velocity of the dominant clay mineralogy. However, the same constraints cannot be applied in unconventional plays as the porosity is not solely linked to mechanical compaction.
By combining the depth-trend approach of the Eaton Ratio and Equivalent Depth methods with the empirical relationship intrinsic to the velocity-effective stress method this study presents a new concept for pore pressure prediction in unconventional plays; the pressure reference trend (PRT) method. The PRT is not linked to the expected compaction behavior of the rock (as inferred from an NCT) but it is simply an empirical depth trend from which the pore pressure can be predicted using industry standard formulae. Rather than constraining the surface and matrix value for the depth trend (NCT) using sensible geological parameters the PRT is iterated until the predicted pressures match a series of data from offset wells. Furthermore, a few wells are excluded to be used as blind test wells to verify the applicability of the PRT as a predictive tool. It would not be possible to exclude some wells from the velocity-effective stress method in this case study as the minimal well coverage would result in too few data from which to derive a significant best-fit line.
Figure 1: A series of cartoons demonstrating normal compaction behavior (upper images) and disequilibrium compaction behavior (lower images). From left to right the images show the pressure-depth plot, the porosity-depth, and the porosity-effective stress plots for each compaction type.
Figure 2 shows an example of a multi-layer Pressure Reference Trend (PRT) in which the PRT for each stratigraphic package has been calibrated to offset wells. By comparison to Figure 1, it is clear that the shape of the PRT is not the same as an NCT and from Figure 2 the PRT clearly does not intercept the surface at velocity similar to water. In fact, the shape of the PRT is steep with minimal curvature and is closer to the shape of the NCT at depth. The shape of the PRT is due to the uplift the basin has undergone, whereby the shallow part of the NCT that would show significant curvature is no longer present due to erosion, and what is left is similar to the deep portion of the NCT (e.g. Ware & Turner, 2002). The final difference between the PRT and NCT is that the PRT will have been shifted laterally towards higher velocity/density, and possibly rotated, from where the uplifted NCT would plot due to the secondary compaction and chemical diagenesis that has occurred disconnecting the porosity from the effective stress. The final position of the PRT in velocity-depth space is a function of the measured pressure, to which the trend is calibrated, over the zone(s) of interest. The trend is then extrapolated from surface to a matrix intercept which may not be geologically sensible, i.e. may be outside the bounds of the elastic properties of the rock, but allow the PRT to be effective over the zone of interest whilst conforming to an typical decay function which is standard to most software that perform pore pressure analysis.
Figure 2: An example of a Pressure Reference Trend (PRT) for the Liard Basin. In this example several PRTs have been developed for different stratigraphic sections.
The data in this study were taken from two wells in the Liard Basin, straddling the boundaries of the North-West Territories (NWT), the Yukon, and northeastern British Columbia, Canada; within the Western Canada Sedimentary Basin (WCSB). The WCSB is a vast sedimentary basin underlying western Canada from southwestern Manitoba, through southern Saskatchewan, Alberta and northeastern British Columbia. It consists of a significant wedge of sedimentary rock that extends from in front of the Rocky Mountains (~6 kilometres or 3.7 miles thick) in the west and thins to zero at the Canadian Shield in the east (Mossop and Shetsen, 1994).
The Liard Basin itself is structurally complex; located in a recess along the Cordilleran Deformation Front and bounded to the east by the Bovie Structure. The recess exists where the orientation of deformation in Cordillera shifts and changes character from a narrow fold and thrust belt with significant shortening in the south to much broader deformation to the North in the Mackenzie Mountains (personal communication Southwest Research Institute research team including D.A. Ferrill, 2015). The basin is characterized by a thick sedimentary wedge and the development of extremely high pressures and significant overpressure. The sedimentary wedge from Devonian to surface (Figure 3) is 4 to 6 kilometres thick, thinning to the east, with multiple significant unconformities. These include the present day Pre-Quaternary surface (~2 kilometres erosion since 60 Ma) and the Pre-Cretaceous Unconformity (which may have had as much as 1 to 3 kilometres of erosion from 145 Ma to about 119 Ma; personal communication with M.E. McMechan, 2015) (Figure 4).
Figure 3: Liard Basin Stratigraphic Cartoon, derived from Mossop and Shetsen, 1994; Monahan, 2000; CoreLabs Stratigraphic Chart and internal correlation.
While there has been limited exploration in the Liard Basin, significant shale gas discoveries from 2012 to 2016 have been made in the Exshaw Formation and the upper member of the Besa River Formation (Figure 1). The Exshaw Formation is composed of black, biogenic silica-rich shale and the upper member of the Besa River Formation is composed of interlaminated black, biogenic silica-rich shale with fine carbonate material. The reservoir package is contained by clay rich shales of the Banff Formation above and undifferentiated Besa River Formation below. According to the Energy Briefing Note produced for the National Energy Board in March 2016, the unconventional potential in the Liard Basin for the Exshaw/Besa River Formations is estimated at 1213 Tcf (34365 billion m3) in place with the potential for 219 Tcf (6196 billion m3) of marketable gas (reasonable confidence expected numbers).
Figure 4: Burial and tectonic overview of the Liard Basin from key well A-67-D/94-O-13, modified after Wright et al., 1994 and personal communication with M. E. McMechan, 2015.
Despite both basins containing similar stratigraphy, that in a traditional setting would allow the use of a common NCT, the significant variation in present day depth below surface and the varying pressure regimes between the two areas means separate PRTs are required. The final model is a set of Pressure Reference Trends (PRTs) where each stratigraphic package required a bespoke trend line although occasionally one PRT was suitable to more than one stratigraphic package. Both observations above are an illustration of the variation to conventional workflows that is needed when undertaking pore pressure prediction in unconventional plays, i.e. depth trends for pressure prediction are calibrated empirically rather than relying on a geological framework. As with building an NCT model a Vshale cut-off was applied to focus on the clay-rich material; in this study the cut-off was 0.7.
Below are two examples from wells in North East British Columbia, the two wells (Well A and Well B) show a calibration well and a blind test well from the Liard Basin. In all the examples, there are three pore pressure predictions shown; two are based on the Vp log using the Eaton Ratio method (purple) and the Equivalent Depth method (blue) and the other is based on the density data using the Equivalent Depth method (red). The reason there is no Eaton Ratio result from density is due to the lack of quantified Eaton exponent for density as that was never presented in the original research (Eaton, 1975). The use of multiple methods and log types is crucial to understanding the uncertainty in the magnitude of the predicted pressure and to make an assessment of which log type presents the most viable proxy for pore pressure. Resistivity is another commonly used log type for pressure prediction but the log signal was too noisy to observe viable pressure trends.
The first well (Well A; Figure 5) is an offset well used as part of the PRT calibration process. The results shown in Figure 5 and Figure 6 show a close degree of agreement between each pore pressure curve and match the static mudweight used while drilling the well as well as the production test taken at ~3700m. This well shows that the wireline data (both Vp and Rho) can be easily translated into a meaningful pore pressure curve with minimal uncertainty and with a clear depth trend predicted for the deep, overpressured intervals.
The next example (Well B; Figure 7 & Figure 8) is a well that was excluded from the PRT calibration process and was used to blind test the model. There is overpressure (pressure above hydrostatic) predicted from the density log in the shallow section but this is due to detrimental effects on density log from washout which can be seen on the caliper log. The Vp log did not suffer from the same hole problems. This well is interesting because the mudweight used to drill the well indicates a much lower overpressure than in Well A, implying that the in-situ pore pressure was much lower. However, the production test taken at the base of this well reveals that the overpressure is much higher than the mudweight suggests. This shows that wells can be drilled underbalanced without experiencing overly detrimental wellbore stability problems, especially in the vertical wellbores which all three examples are. More importantly, mudweight cannot be assumed as a proxy for pore pressure as it can severely underpredict the magnitude, which will have a significant impact on the accuracy of any subsequent geomechanical model.
The predicted pore pressure from the PRT for Well B is shown in the same blue, red, and purple curves as shown for Well A. The critical observation is that the predicted pressures are on trend with the production test taken in the well, i.e., the PRT accurately predicts the in-situ pore pressure as measured in the well (Figure 8). The accuracy of the PRT is critical in this area as it gives confidence in the magnitude of the predicted pore pressure and allows doe a more accurate hydraulic fracturing plan (geomechanical model) to be derived. This leads to more efficient fracturing and more cost effective wells. Interestingly, other wells close by did experience higher mudweights and drilling events (gases, flows) which were on trend for the deep pore pressure data point (Figure 9). The additional data from these offset wells (Wells X & Y) show that the mudweight in Well B is actually anomalous and that the predicted pressures are more accurate in the context of all the data.
This paper presents a case study showing that traditional pore pressure prediction techniques (Eaton Ratio and Equivalent Depth methods) can be adapted to predict pore pressure in an unconventional play, such as the Devonian Black Shales of North East British Columbia, by using a newly defined Pressure Reference Trend (PRT) in-lieu of a Normal Compaction Trend (NCT) as used in conventional, or traditional, pore pressure prediction.
The PRT is not linked to the expected compaction behavior of the rock (as inferred from an NCT) but it is simply an empirical depth trend from which the pore pressure can be predicted using industry standard formulae. Rather than constraining the surface and matrix value for an NCT using sensible geological parameters, the final position of the PRT in velocity-depth space is a function of the measured pressure, to which the trend is calibrated, combined with a lateral shift towards higher velocity/density due to tectonic uplift, secondary compaction, and chemical diagenesis that has occurred over the geological history of the basin.
The advantages of using traditional pore pressure prediction techniques is that they are standard to most industry software, they work better than other techniques (e.g., Bowers) in areas with less data, they work with industry standard logs (e.g., Vp, Rho. Although not shown in this paper, the PRT model was also used to generate a regional 2D line of pore pressure that can be used to characterize inter-well regions for future exploration and exploitation of the resources in the basin.
We would like to thank Nexen Energy ULC, CNOOC Limited, and IGBC for supporting this work through provision of the data and permission to present these results.