A geopressure interpretation technique known as the seismic velocity method is a common workflow in which shale compaction functions are characterized at offset control wells, matched to interval seismic velocities, and then used to predictively calculate geopressure away from well control. The seismic velocity method is used to interpret the expected geopressure profile at the Deep Blue subsalt exploration well in Green Canyon 723 in the deep water Gulf of Mexico. The Deep Blue prospect is distinct from other prospects in the play fairway in that the prospective section is overlain by a salt withdrawal minibasin, whereas the offsetting fields are positioned either along the flanks of minibasins or under a thick allochthonous salt canopy. Predrill geopressure interpretations using numerous tomographic imaging velocity data sets shows a large degree of consistency with the magnitude of geopressure encountered in offsetting supra salt and subsalt fields. Results from the Deep Blue 1 exploration well indicate the predrill geopressure interpretation from interval seismic velocities failed to anticipate the extreme degree overpressure encountered in the subsalt section of the well due to poor deep velocity resolution and an “unloaded” compaction signature. The magnitude of overpressure in the primary section is attributed to the emplacement of an unconformable halokinetic sequence over the primary subsalt basin. An interpretive paradigm is described in which the Deep Blue pressure cell is created through two halokinetic episodes: (1) rapid progradation of a salt canopy followed by (2) subsequent salt withdrawal and emplacement of an overlying minibasin. The linkage between halokinetic sequences, burial history, and the development of overpressure can be used to predictively characterize subsalt geopressure environments.
By their nature, exploration geophysics and geopressure prediction methods are reductive techniques. Processes are broken down into their components and addressed sequentially. Reflection data is processed in sequential workflows where parameters can be isolated and tested. Likewise, geopressure prediction workflows typically are comprised of sequential tasks such as log conditioning, trendline calibration, residual velocity picking, and error term quantification. A detailed description of the procedure and interpretive value of seismic geopressure prediction in the deepwater Gulf of Mexico is given by Kan et al. (1999). Chopra and Huffman (2006) offer an extensive overview of velocity determination for geopressure prediction, and Sayers et al. (2002) demonstrate the superior resolution of tomographic imaging velocities and consequently improved geopressure interpretation. Descriptive statistics (Sayers et al., 2006) can be used to quantify uncertainty in compaction trend parameters and interval seismic velocity; however, the statistical distribution functions are biased to the data quality and geologic controls at the calibration locations. Moreover, descriptive statistics alone do not reveal the causal mechanism responsible for the expected uncertainty range. The pitfall of the reductive workflow is that maximizing optima of each component of the process can result in a precisely quantified, but grossly inaccurate prediction, whereas an intuitive viewpoint can provide valuable insight in the face of uncertainty.
Pilcher et al. (2011) extensively characterize different types of primary basin salt welds in the salt canopy province of the deepwater Gulf of Mexico in the context of hydrocarbon trapping styles; however, geopressure compartmentalization is not explicitly addressed. The intent of this paper is to present an interpretive paradigm for the expected geopressure environment in a primary salt basin overlain by a welded minibasin halokinetic sequence and to discuss the predictive failing of seismic velocity based geopressure interpretation. The genetic linkage between welded minibasin halokinetic sequences and overpressure development will be supported by integrating formation pressure measurements, petrophysical, and chronostratigraphic interpretations from the Deep Blue 1 exploration well. The pitfall of using surface seismic to predict geopressure in deep, complex structural environments will be demonstrated by comparison of the predrill pressure prediction to the actual formation pressures measured in the well. The scope of discussion will be limited to notable learnings in the context of geopressure interpretation and not to characterize reservoir properties. However, a brief description of the geologic components of the prospect is critical to gain insight as to what elements of the geopressure story could have been anticipated predrill and whether the learnings can be applied elsewhere.
The Deep Blue prospect
The Deep Blue prospect is located in the salt canopy province of Central Green Canyon, situated in block 723 at a water depth of 5040 ft. Block 723 is located at the structural culmination of the Deep Blue primary basin which is interpreted to be a continuous stratigraphic sequence between the underlying Louann Salt and the overlying salt weld. The Deep Blue primary basin forms a structural turtle with an area of roughly 15 OCS blocks (Figure 1). Overlying the salt weld is an elongate minibasin, known as the Constitution minibasin, which dips to the north and is ponded onto salt that surrounds it. The minibasin contains productive horizons along its southern and northern flanks at the Ticonderoga (GC 768) and Constitution (GC 680) fields. Offsetting the Deep Blue primary basin to the north and east are the subsalt Tonga (GC 727), Caesar (GC 683), and Tahiti (GC 640) fields. Deep Blue 1 spud 24 November 2009 and reached TD on 3 May 2010 at a depth of 32,685 ft; pay was found in the objective section.
Offset well control
Although many wells had been drilled in the play fairway, Deep Blue is a rank wildcat in that there are no penetrations in the interpreted to be structurally isolated Deep Blue primary basin. Extensive well control, in conjunction with detailed stratigraphic mapping helped constrain the expected geopressure profile within the Constitution minibasin. Analysis of pressure measurements from the Conquest, Ticonderoga, and Constitution fields along the flanks of the minibasin demonstrated a high degree of lateral permeability in that aquifers could be mapped across the center of the minibasin. One such aquifer is plotted on the predrill geopressure interpretation (Figure 2). Sequence boundaries and disconformities create vertical seals that drive the compaction disequilibrium process deep into the minibasin. Offset well and seismic velocity analyses suggest the fluid retention depth (i.e., top of overpressure, Swarbrick et al. 2002) in the minibasin is between 2500 ft (in the updip position) and 3500 ft (in the down-dip position) below mudline. In the center of the minibasin at the Deep Blue location, the fluid retention depth is interpreted to be 3000 ft below mudline on a gradient of .
The subsalt section is constrained by the offset discoveries at Tonga, Caesar, and Tahiti. Although the expected stratigraphy of the primary section is thought to be similar to the offset wells, the offset pays are also overlain by a massive salt canopy. Needless to say, detailed structural mapping is challenging due to the quality of the subsalt seismic image and it is not known to what degree the offset fields are in communication with the Deep Blue prospect. Typical overpressure of the aquifers in the primary basins to the east is , assuming a fluid gradient of .
Predrill seismic velocities
The Deep Blue prospect is mapped on numerous 3D data sets of different vintage: an older narrow azimuth (NAZ) isotropic PSDM and different variations of wide azimuth (WAZ) anisotropic PSDM models. From a geopressure prediction standpoint, the NAZ imaging velocities capture gross compaction features in the minibasin, but fail to resolve important nuances across faults and unconformities. In addition, the isotropic imaging depth is significantly deeper than actual true vertical depth (TVD) as shown by the well ties in the minibasin. Track 2 in Figure 2 shows a comparison between the NAZ seismic velocity and the normal compaction trend.
Many iterations of VTI anisotropic velocity model building have been performed on the WAZ data set. Imaged depths matched well control in the minibasin with great accuracy; however, there are still fundamental problems with the velocity field for geopressure-related compaction analysis. Even the preferred velocity field contained undulations (green line in track 2, Figure 2) and what is interpreted to be smoothing artifacts associated with tomographically picked moveout velocities. Lastly, the image quality of the deep section is almost completely invariant when the data are migrated with significantly different velocity models. Representative velocity profiles are shown in Figure 3. An inspection of the image point gathers (Figure 4) demonstrates the lack of far offsets in the subsalt section, and hence no meaningful criteria for picking moveout velocities.
A seismically opaque section at the base of the minibasin is interpreted to be evacuating salt. The predrill migration velocity model included a 600-ft-section of salt, although it is not known for certain whether salt even existed in the weld because the image quality is invariant to the presence of salt in the migration velocity model.
Key operational events
Three primary operational challenges were encountered during the drilling of the well. The first event occurred after taking a kick at 10,770 ft from pressured sand below an unconformity. The well stabilized at a surface mudweight of 10.8# and subsequently a liner was run. Predrill geopressure estimates were 10.2# and 10.5# based on the velocity trendline and projection of mudweight down from the Conquest well, respectively.
There were no other pressure-related surprises until 25,984 ft, when another kick was encountered at the base of the minibasin. Consequently, the well had to be weighted up from 14.7# to 15.4#, which represented a kick of roughly 1000 psi. Predrill analysis anticipated the minibasin to follow a basic undercompaction profile down through the salt weld. The sealing properties of the salt weld were unknown, largely due to fundamental uncertainties as to the geologic manifestation of the interpreted salt weld on the seismic images. Predrill pressure from seismic in conjunction with the observation that the minibasin had numerous oil pools suggest that the interpreted salt weld may have been hydraulically conductive, and hence unlikely to retain anomalous geopressure. During wireline logging, formation pressure measurements were taken in the kick zone and affirm the pressure to be 21,240 psi or 15.73# equivalent mud weight (emw).
After the primary reservoir section was drilled and evaluated, it was decided to deepen the well to test deeper objectives. A liner was set at 29,905 ft and the final hole size of 8.5” inches was drilled to TD at 32,684 ft. Drilling margin proved to be thin in the 8.5” hole section. A surface mudweight of 16.3# resulted in a circulating density of 17.03# on bottom. A drastic lithology change occurred at 31,800 ft. Cuttings were described as calcareous and it is difficult to discern whether LWD resistivity anomalies are related to pressure or lithology. The decision to call TD was made because of lost returns on bottom. Although there is some debate as to whether the lost returns indicate the possibility of a pressure regression in the Lower Tertiary, lower horizontal stress in the calcareous lithology, or ECD surges, the fluid pressure at TD is interpreted to be 27,665 psi or 16.3# emw (Figure 5).
Key observations and interpretations
By all accounts, the drilling of Deep Blue was an enlightening discovery process. The first key discovery is the nature of the interpreted salt weld at the base of the minibasin. Paleo data indicate the age of the first depositional sequence in the minibasin to be a condensed section that contains fauna of Serravallian (11.6–13.65 mya) to Langhian (13.65–15.97 mya) in age. Depositionally, this section has been interpreted as a stratal carapace (lithology related to a paleobathymetric high, Hart et al. 2004), which is comprised of tight, impermeable marls. Sonic and resistivity values of this interval are .87 ohmm and , respectively (Track 2 and Track 3 in Figure 5). Both logging measurements demonstrate a significant shift from the established compaction trend lines. The sonic method resulted in a mild pressure increase, whereas the dropping resistivity values suggest a sharp pressure transition at the top of the carapace. Interpreted pressure in the carapace is derived from drilling indicators, mud weights, and the kick taken at the salt weld.
A notable observation is the lack of halite above the salt weld at 25,984 ft. The seismic marker known as the “salt weld” is in fact bounded by thick sand deposits of hyper saline pore fluid and indubitably no salt! The pressure profile in Track 4 of Figure 5 demonstrates the effect of the unusual petrophysical characteristics of the carapace on the pressure interpretation. Pore pressure at the salt weld was 2100 psi higher than the predrill prediction. During drilling operations, conflicting sonic and resistivity trends made it difficult to discern whether the magnitude of pressure experienced at the salt weld was an isolated occurrence or whether it is representative of the primary section at large. Below the salt weld, resistivity values dropped back to a monotonically decreasing trend of 0.8 to 0.7 ohmm, whereas sonic values dropped back to an average of and remained mostly constant for the rest of the Miocene section, suggesting a constant state of undercompaction. Paleo data indicate the well drilled into an in situ section of Middle Miocene immediately underlying the salt weld (17.85–10.97 mya), the implication of which will be discussed later in this paper.
Numerous sands were encountered in the prospective subweld section. Extensive wireline pressure measurements indicate an overburden-parallel pressure profile that exhibits a constant compaction state of about 2200 psi vertical effective stress. Three thick sand deposits comprise a modest pressure regression between 28,000 and 29,000 ft. When compared to the encasing shale pressures, the sand fairway appears to be underpressured relative to the expected centroid, even though the prospect was drilled near the crestal position of the Deep Blue subsalt basin (Figure 6), implying that formation water is likely bleeding off pressure as it flows updip via a hydraulic conduit. Below the massive sand interval, geopressure continues to increase at a rate of until the TD of the well. Similar to the hydraulic properties of the overlying Constitution minibasin, permeability within the subsalt pressure cell is highly anisotropic. Water flow along sand conduits enables pressure to equilibrate parallel to bedding surfaces, whereas sequence boundaries and shale intervals impede vertical fluid flow up section. Boundaries of the Deep Blue pressure cell are defined by a weld surface and interpreted surrounding halokinetic features such as salt walls and diapirs.
Discussion and predictive implications
The failure of the seismic velocity method to anticipate the magnitude of geopressure in the primary section can be characterized as a Type II data analysis error, i.e., failure to reject an erroneous regression. Figure 7 shows the best-fit Bower’s virgin compaction regression based on control points in the Constitution minibasin. Standard error of the regression for each of the control points is with an of 0.974 indicating a robust fit to the data over the expected vertical effective stress range. Moreover, alignment of the predrill model with formation pressure measurements from the nearest subsalt basin provided further corroboration, or at least confirmation bias, to support the quality of the Deep Blue predrill seismic geopressure model. However, both inputs to the predrill pressure model proved to be in error. The migration velocities did not have the moveout sensitivity to robustly anticipate the velocity regression in the primary section, and secondly, the velocity to effective stress function defined in the minibasin is significantly outside of the velocity to effective stress relationship observed in the subsalt primary section. Numerous pressure measurements (green points in Figure 5) taken in thin sands (limited centroid effect) between 26,000 and 30,000 ft provide control points to calibrate the relationship between velocity and effective stress in the shales of the primary section. Red bars in Figure 7 show that the velocity and effective stress ranges for Middle Miocene shales in the primary section are significantly different than the relationship defined in the overlying minibasin. In the framework of Bowers (1995) and Lahann (2002), the velocity to effective stress relationship demonstrates an “unloaded signature” relative to the minibasin trend. Plausible explanations for the unloaded signature of the primary section include clay diagenesis (Lahann and Swarbrick, 2011) or perhaps inelastic effects related to compressive salt stress (Bowers, 2007).
Arguably, the most novel aspect of this conclusion is the interpreted halokinetic linkage between the impermeable condensed section at the base of the minibasin, salt weld, repeat section, and extreme overpressure in the subsalt, primary section. Presented in Figure 8a–8d is a cartoon reconstruction of the events leading up the extreme degree of overpressure observed in subsalt section of Deep Blue. A kinematic summary of the events is presented below.
The carapace section is contemporaneously deposited at the top of a paleo salt canopy while the Deep Blue sand fairway is a bathymetric low flanked by salt diapirs (Figure 8a).
The carapace section is interpreted to be perched atop a translating salt canopy as it prograded over the Deep Blue primary section. Flanks of the primary basin deflate, creating the Deep Blue structural turtle (Figure 8b).
The salt canopy deflates as the incipient minibasin forms over the Deep Blue primary section. Formation water trapped in the primary section is unable to escape up section through the salt canopy (Figure 8c).
Salt is eventually completely evacuated at the base of the Constitution minibasin resulting in the impermeable condensed section directly overlying the overpressured primary basin. The extreme overpressure in the Deep Blue basin is explained herein by the interpreted interplay of two mechanisms. First, is the rapid subsidence of the minibasin, which replaced the low density salt canopy with a dense clastic minibasin comprised of compacting sediments. Secondly, the paleo salt canopy and later the condensed section acted as superseals, preventing the advection of pore water from the primary section into the overlying minibasin (Figure 8d).
From a geopressure prediction standpoint, what are the lessons learned? For one, undercompaction is the primary driver of overpressure in the Gulf of Mexico, although arguably one could infer about 1400 psi of “unloading pressure” based on the effective stress plot in Figure 7. Given the low geothermal gradient in deepwater Gulf of Mexico, nearly all overpressure is related to undercompaction and consequently the interplay between burial rate and hydraulic conductivity. In hindsight, the expected difference in magnitude between the Deep Blue pressure cell and the Tahiti and Tonga pressure cells could have been partly ameliorated by recognizing the density effect of the salt canopy and presence of an impermeable stratal carapace at the base of the minibasin. The cumulative density effect of a 14,000 ft salt canopy versus minibasin is 1750 psi, which would consequently result in a pore pressure increase of 1750 psi in a comparably permeable undercompaction environment. Nevertheless, density alone accounts for neither the magnitude of overpressure in the primary section nor the failure of the predrill model, particularly because the denser overburden of the minibasin was explicitly used in the predrill geopressure calculation. The coincidence of the predrill seismic geopressure prediction and the magnitude of formation pressures in the adjacent primary basin falsely led us to infer lateral pressure communication between the two primary basins. We now know to consider unloaded effective stress transforms in isolated subsalt compartments and to be wary of deep velocity resolution.
Our team also hypothesizes that framework collapse of marly lithologies in the carapace section may have generated additional overpressure, although without core data this hypothesis cannot be tested. With a rich suite of wireline data, it can be inferred that the structurally isolated primary section is compacted to an equivalent burial of below mudline; however, it is difficult to envision a robust method to anticipate the timing of the effective hydraulic seal. The actionable learning is that isolated primary basins can be extremely undercompacted.
The other key learning from a prediction standpoint is not to trust seismic velocities if there is no moveout sensitivity at depth. A rigorous inspection of image point gathers is critical for determining whether there is a material amount of hyperbolic moveout to pick robust interval velocities on image point gathers. Figure 3 shows substantially different migration velocity models that had no material effect on image quality. It turns out the velocity model in Figure 3d is closest to the actual geopressure environment, but predrilling there was no diagnostic indication as to which model was most accurate. Recognizing the forecasting issue with subsalt seismic velocities, characterization of halokinetically bounded structural compartments may guide the interpretation of predrill pore pressure. For example, an encapsulated basin such as Deep Blue is likely to contain extreme overpressure, whereas a semiregional primary basin such as the Tahiti Basin is likely to be drained to some extent. Specifically, if one could infer the timing of a hydraulic encapsulation, then expected pore pressure could consequently be calculated through the use of basic compaction disequilibrium principles.
Regarding the impact of halokinetic features such as salt welds, it has been demonstrated that salt can completely evacuate from a weld zone. In this example, we infer the contemporary master seal between the primary section and the overlying minibasin to be the stratal carapace, not the halite-free weld surface. From a pressure cell standpoint, the carapace section is the master aquatard that prevents bulk water advection from the encapsulated primary basin. Likewise, one could infer the condensed carapace section could act as a highly effect membrane seal to two-phase flow in a water wet oil/water system. However, the presence of oil pools along the flanks of the overlying minibasin suggests a material degree of oil infused groundwater migration up section through some type of unidentified hydraulic pathway.
Suggestions for further study
We encourage the exploration community to come forward and share other examples of abnormal pressure associated with halokinetic features such as bucket welds, bowl welds, keels, and encased basins. Given the known limitations of seismic geopressure interpretation in the subsalt domain, it will be of great benefit for the industry to develop predictive methods to quantify the linkage between halokinetic sequences and overpressure development.
The seismic velocity technique failed to anticipate the extreme magnitude of overpressure in the primary section of Deep Blue because the inability to resolve moveout velocity below the salt weld and because the effective stress to velocity transform was not calibrated to the geologic conditions experienced in the primary basin. The actionable learning from a subsalt pressure prediction standpoint is to step back and holistically consider the effect of halokinesis on burial history and stratigraphic permeability. Based on our interpretation of the Deep Blue well, we presented an interpretive paradigm for the expected geopressure environment in an isolated primary salt basin overlain by a minibasin weld halokinetic sequence.
We would like to thank the management at Noble Energy Inc. for permission to publish, and to Martin Albertin, Charlie Garvey, and Charlie West, as well as two anonymous reviewers for constructive feedback.
Niven Shumaker studied geotechnical engineering at the University of Iceland, received an M.S. (2005) in geophysics from Virginia Tech, and will complete an M.B.A. from Rice University in 2014. He currently works in the Geoscience Technology Group, where he supports global drilling operations and the company’s global exploration portfolio. He has worked projects in the Black Sea, Middle East, North Sea, West Africa, China, Mediterranean, Gulf of Mexico, and onshore basins in the United States as a prospector, development geoscientist, and in well services.
Daniel Haymond received an M.S. (1980) in geology, and he has participated in graduate studies at Brigham Young University, the University of Utah, and Johns Hopkins University. He is a deepwater regional exploration geologist. He has more than 36 years of worldwide exploration experience with Noble Energy, Inc./Samedan International, Santa Fe Energy Resources, Inc., and Phillips Petroleum Company.
Joe Martin received a B.S. (1978) in geology with a minor in physics from the University of Texas at Arlington. He worked for 34 years in exploration geophysics, first at Texas Eastern Exploration Company, followed by Sonat Exploration Company, and the retired from Noble Energy. He generated prospects and supervised exploration teams primarily in U.S. offshore basins, including Alaska, California, and the Gulf Coast, with some additional time in the upper Gulf Coast onshore.