Evaluating Fault and Cap Rock Seals

Edited by Peter Boult and John Kaldi

Foreword

Three AAPG Hedberg research conferences on seals have been held in the past 30 years. Each conference represented a quantum leap in the understanding and methodology of the subject of seals. This volume is a compendium of the proceedings of the 2002 meeting. The key driver for this meeting was the recognition that knowledge of risk (in the estimation of sealing capacity and faultseal potential) is important in making judgements at the exploration, appraisal, and development stages of the petroleum business. In addition, incorporating seal risk in the overall assessment of hydrocarbons in place can affect decisions to drill prospects, the location of appraisal and development wells, as well as reserve estimation. Improved methods to estimate seal capacity and fault integrity can lead to savings in well costs, improved recoveries through optimum placement of wells, and greater certainty of meeting contractual requirements through improved estimates of hydrocarbon in place.

The 2002 meeting was the first ever Hedberg held in Australia, and the venue, a glorious setting in the heart of the famous South Australia wine-growing region, may have had something to do with the attendance of 85 delegates, most of whom traveled halfway around the world to be there. The meeting consisted of 53 presentations over a period of two-and-a-half days with robust debate taking up almost half of the time. This volume is the result of the papers presented, debated, revised, and finally submitted to AAPG as part of a thematic state-of-the-art publication on seals.

The volume contains 18 chapters that reflect the spectrum of presentations at the conference. The knowledge imparted by these chapters will be a window on the state of knowledge at this juncture of time. It will be a lasting tribute to the efforts of the individuals and the synergy of the group, as a whole, that was established at the conference.

The plethora of new science presented at the Barossa meeting was obviously an evolutionary outgrowth of previous seals conferences. Threads connecting to the Crested Butte conference included the question of the importance of wettability in seals. At Crested Butte, the long held assumption that all rocks are water wet was questioned. Leith et al. (1993) showed that oil had penetrated seals in some North Sea traps, but whether this was by the process of hydraulic fracturing or capillary leakage through a water-wet seal was not resolved. Over the last 10 yr, very little work on seal wettability has been carried out or at least documented. Aplin and Larter’s chapter in this volume addresses some of the issues raised in 1993. Their conclusions are that “hydrophilic organic compounds in reservoirs, followed by diffusion into cap rock pores, may create oil-wet pathways into cap rocks and drive leakage.” This means that not only do petrophysicists and reservoir engineers have to consider wettability when estimating the saturation and flow properties of reservoirs, but explorationists also have to consider the phenomenon when risking entrapment. Aplin and Larter’s chapter not only probes the wettability question but also describes a methodology for determining, directly from logs, properties that control the flow of hydrocarbons through muddy seals that may undergo wettability change. They do this by investigating the key relationships between (1) porosity and effective stress, (2) porosity and permeability, and (3) porosity and threshold capillary entry pressure. The authors conclude that these properties are “strongly influenced by the grain-size distribution or clay content of the sediments.” These properties can be calibrated against cores and constrained by cuttings data.

Although Aplin and Larter do not consider the role of leakage via hydraulic fracturing of clay-rich cap rocks, Hermanrud et al. do. In their chapter, they address “three sets of such subsurface processes, and how they impact on hydrocarbon leakage: sediment compaction, fluid mobility in a two-phase (water-plushydrocarbons) reservoir, and relationships between stress and rock failure.” They describe the conditions for sediment compaction and remind us that stress-insensitive chemical compaction becomes increasingly important above 60°C. They state, conclusively, that “All pressure compartments leak fluids as long as [chemical] compaction occurs.” Hermanrud et al. argue that the relative permeability of seals and reservoirs to water never reaches zero on a geological timescale. This, they say, holds true despite the fact that water in the reservoir may be at a conventional irreducible saturation. Furthermore, they tell us that the “failure to recognize the mobility of this water may result in too-pessimistic seal capacity estimates.” They also describe the conditions for the creation of structural permeability that will allow the bleed-off of overpressure and the leakage of hydrocarbons.

The abundance of papers in the conference and in this volume containing elements of geomechanics attests to the importance of this area of investigation. Some of the presentations and papers also demonstrate that as a discipline, the investigation of hydrocarbon flow through seals has much to gain from other industries such as mining and construction. Whether leakage occurs through the cap rock as a result of perturbations in the stress field caused by fault discordance as suggested by Hunt and Boult or, as Mildren et al. suggest, the faults themselves are the mechanism for leakage as they dilate under the influence of the stress field, is still debatable. It would serve any explorationist well to keep a foot in both camps because both may, to some extent, be correct.

Two chapters are present in the volume that introduce, as Gartrell and Lisk put it, a “new layer of complexity to fault reactivation and associated hydrocarbon leakage.” Gartrell and Lisk’s chapter tests a new methodology for assessing paleostress regimes, and they conclude that, despite the potential errors inherent in the methodology they use, their estimation of the paleostress tensor at the southern end of the Timor Sea is “not in conflict with structural histories formulated by more traditional methods.” For the prediction of the paleostress tensor, they bring together, for the first time, a three-dimensional fault restoration technique, which is commonly performed on seismic data sets to verify fault and horizon interpretation, and a fault slip inversion technique, which had previously only been used in outcrop studies. They admit that this is very early work, and one data point does not mean the method is proven, but all such empirical methods have to start somewhere. Moreover, they conclude that without such an approach, paleostress estimations based on simple geometric analysis (such as fault orientations) are possibly prone to errors.

Dewhurst et al. however, definitely take the deterministic approach to fault-seal analysis. They examine the detail of the fault rock itself in terms of its response to geomechanical laboratory testing and relate results to petrological properties. Significantly, they find that in two study areas, phyllosilicate framework fault rocks are “stronger than their host reservoirs as a direct result of syn- and postdeformational physical and diagenetic processes.” They point out that with the trend toward using fault reactivation and consequent seal assessment methods derived from theoretical geomechanics, their observations of “strength recovery through healing” means that faults cannot be considered as being purely cohesionless. Thus, algorithms, such as dilation and slip tendency, which do not incorporate rock strength, may be misleading when assessing fault-seal risk. Furthermore, in a regional context, as Hunt and Boult point out, the regeneration of fault strength (via diagenesis) influences stress distribution.Within a regional top seal, this may result in fracturing (creation of pervasive structural permeability) and the concomitant loss of hydrocarbons.

Reynolds et al’s chapter is based on the application of geomechanical theory, but in this case, they explain what can be done to assess faults for leakage potential on a basinwide scale when little or no data are available for depth conversion. Their approach is a sensitivity analysis of fault assessment based on a geomechanical estimation of the relative risk of reactivation. They assume that risking reactivation of faults in Late Cretaceous sediments of the Great Australian Bight is equivalent to risking leakage. This chapter, along with those of Mildren et al. and Lyon et al., assumes that during reactivation, the rate of strain and the rheology of the fault rocks are such that when they interact, reactivation occurs by brittle failure, and dilation occurs to allow leakage. Although this may not be true for the more prolific hydrocarbon-producing basins of the world, where stress regimes are more stable and rocks are younger and weaker, it would appear to be the case for Australia, which, as described by Reynolds et al., currently has and historically has had a highly variable stress tensor in space and time relative to its continental margins.

Although not included in this volume, Fisher presented key data at the 2002 Hedberg Conference that helped clarify the debate on whether faults dilate or leak as they move (reactivate). Fisher described the controls on the geomechanical properties of sediments as they undergo burial and compaction and heating and diagenesis. Fisher illustrated the control of grain size, geothermal gradient, and burial rates on the depth of the ductile-to-brittle transition, which may also coincide with the top of the seismogenic zone. Although a useful guide for considering the impact of faults on production, Fisher’s data, like many other petrological studies, are largely confined to reservoir sandstones, because very little data are available from seal lithologies.

Couples’ chapter, “The role of geomechanics,” contributes to the debate associated with the assumption that faults that reactivate exhibit dilation. This chapter revisits some of the fundamental physical definitions of what a seal is and how geomechanics can describe its behavior. Couples provides an interesting definition of seals as “power transmission barriers,” which are places “where the rate of achievable energy dissipation (flow) is less than the rate at which potential energy can be imposed or renewed.” The keyword here is rate, and this will vary depending on whether one is dealing with the relatively high rates of stress change as experienced by Australia compared to the more stable stress regimes such as on the passive margins of the Atlantic. Couples’ chapter provides insights to understanding “the big picture” of geomechanics.

Couples also warns us about some of the pitfalls of upscaling geomechanical laboratory measurements to large-scale models. He points out that “a scale dependency to the identification of localized vs. delocalized deformation is present,” which is “basically the same issue that arises in the way that many people use the terms brittle and ductile.” This is followed by a discussion of poroplasticity and its relation to volume loss or gain (dilation) during deformation. Although in many respects, this chapter is aimed at the geomechanically inclined, Couples is still concerned with getting his message across to the general practitioner. This message is that an understanding of “poro-plasticity provides the critical link to allow us to make realistic geomechanical predictions about seals.” He also provides some interesting new insights into everyday observations, such as “we can interpret dilational deformations as representing material that is ‘too solid’ to be able to undergo the imposed distortions, so extra porosity must be created to ‘weaken’ the material enough to allow cataclastic flow.” However, the main strength of this chapter is that he works through an example (without using mathematics) where he demonstrates how geomechanics explains most of what we know about the operation of seals and overpressured rocks. In addition, he describes how geomechanical modeling can be used to help understand the phenomena, such as the formation of shale gouge, the non-Andersonian behavior of stress around faults, and the prediction of petrophysical properties for input into migration simulation models.

In the previous Hedberg conferences, there was considerable interest in pressure compartments, and the term “pressure seal” was in common usage. From this, the term retention capacity (being the difference between the least principal stress and the pore pressure) was born, because seals were assumed to only fail either by hydraulic fracturing or capillary leakage (Watts, 1987). Thus, in water-wet shales with extremely high capillary entry pressures, high retention capacities were thought to be synonymous with good pressure seals. However, in this volume, Nordgard Bolas et al. show that in their study area, where differential stress is high, shear failure instead of tensile hydraulic fracturing occurs, and seals with high retention capacities may be breached. Shear fracturing is caused by very recent increases in differential stress attributable to ice loading and unloading and the advance of sedimentary wedges. Furthermore, their analysis indicates that in their case study, shear failure and associated leakage reduce pore pressure and, therefore, increase retention capacity.

Shear-induced failure and mixed-mode failure are mechanisms that have long been thought to be responsible for the creation of structural permeability through cap rocks in and around Australia (Hillis, 1998) because of the recent collision of that continent with New Guinea and Indonesia. In this volume, Nordgard Bolas et al. show that mechanisms other than continental collision and a resultant, highly variable stress tensor can cause rapid increases in differential stress, leading to shear failure. Those working in deep offshore basins where the postcharge advance of sedimentary wedges has occurred, should consider the findings of Nordgard Bolas et al. Nevertheless, situations in low differential stress environments where retention capacity (i.e., failure by tensile hydraulic fracturing) is still a good indicator of seal adequacy remain. Knowing where to apply this relationship is the main problem. Fortunately, this can be solved by use of the delta-P parameter or FAST approach, as described by Mildren et al. Delta-P is a measure of the increase in pore pressure that would be required to create fractures by either shear or tensile failure. The methodology considers not only the stress field and the orientation of planes of weakness and faults in that field, but also the strength of the fault rocks. The easiest way to picture this is in terms of the distance between poles to planes in a Mohr circle (representing stress state) and the Coulomb failure envelope (representing rock strength) of the rock that is perceived as the weak link in the sealing system. Whether this is a fault that cuts the seal or the seal itself depends on their relative strengths.

What of initial entrapment in fault-related traps? As previously mentioned, structural uncertainty has certainly been reduced in recent years by rapid advances in technology. Whether it is worth focusing on this kind of uncertainty at the exploration stage is still debatable, given all the other uncertainties of a petroleum system that may still need to be addressed. This lack of concern is reflected in the absence of papers presented on this subject at the conference. However, after a significant discovery has been made, the quantification of structural uncertainty becomes of greater significance. This is long overdue because, as Ottesen and Townsend point out, “the statistical treatment of tectonic heterogeneities in reservoir simulation is not as rigorous as that of their sedimentary counterparts.” However, the increasing tendency toward exploration and discovery of structurally complex traps demands that this aspect be risked in a similarly rigorous fashion. Their chapter describes a workflow for incorporating structural uncertainty into a geological model that can be used in a reservoir simulator. Ottesen and Townsend cite fault density and fault-relative permeability as the most influential structural factors that bear on ultimate recovery.

Lescoffit and Townsend add to the contribution of Ottesen and Townsend by investigating the relative importance and effect of structural uncertainty on production predicted from models built into a reservoir simulator. To simplify their experiment, they did not vary fault location or geometry and only considered fault connectivity, fault displacement, and parameters related to displacement that affect transmissibility (e.g., fault thickness and gouge-smear prediction). Their models are based on a typical North Sea tilted fault block, and the authors conclude that whereas the sequence-stratigraphic models still dominate, this is not true for all simulation cases. Of the structural parameters that they did vary, the fault-seal algorithm and the subseismic fault pattern used had greatest impact on production prediction.

Bretan and Yielding have done considerable work in the past toward empirically constraining fault-seal capacity in the static case and transmissibility in the production case. Their chapter in this volume very succinctly describes the use of buoyancy pressure profiles for calibrating and incorporating uncertainty in the Vclay parameter, which, in turn, is used along with fault throw to generate fault algorithm parameters, such as shale gouge ratio (SGR) and shale smear factor. The use of the SGR factor is demonstrated in papers by Lyon et al. and Ottesen et al. Lyon et al. use it to help understand the original hydrocarbontrapping mechanism in the Zema structure in the Otway Basin, which has subsequently undergone leakage because of shear failure either along or associated with the major bounding fault. Ottesen et al. use SGR in combination with a clay smear algorithm to determine fault transmissibility.

Lothe et al. use the methodologies learned from reservoir simulations and apply these to the province-scale linked pressure and stress simulator. Their chapter describes a validated, time-stepping model for predicting overpressure and hydrocarbon leakage from linked pressure compartments. This, in turn, is based on the interaction of across-fault leakage and hydraulic fracturing of the cap rock in each compartment. Through an ingenious method of altering permeability in faults and modeling stress through time in the cap rocks, they create multiple realizations of leakage. The authors compare their results with actual pressure data in wells and zero in on the most likely hydrocarbon migration route across an area. This, of  course, is potentially invaluable in pinpointing prospective areas to reduce exploration uncertainty.

Lothe et al. bring faults and cap rock-sealing mechanisms together in a province-scale linked pressure and stress simulator. Hunt and Boult, however, bring these aspects together in an overall assessment of trap integrity. In the Otway Basin where Dewhurst et al. recognize that some faults are stronger than surrounding rocks, Hunt and Boult have observed stress perturbations that corroborate this observation and use discrete element stress modeling to investigate this phenomena. They base their model on data from multiple fault-bound traps, some of which have leaked. They assign three different fault sets and realistic geomechanical parameters of cap rocks to run their model forward until zones of high and low differential stress appear. They conclude that a positive observable relationship exists between zones of modeled high shear stress and trap leakage, and that in this case where observations indicate that faults are stronger than the surrounding rock, it is fracturing and the development of structural permeability in the cap rock that has caused leakage.

Lyon et al.’s chapter is a case study on the Zema structure, which occurs in the data set used by Hunt and Boult. This structure contains a well-documented paleo-oil-gas column in the Otway Basin. A core from the seal was analyzed to determine the original trapping mechanism, and Lyon et al. attribute the weak link to shale gouge that developed on the main bounding fault. They then consider the mechanism for failure. The seal over the Zema structure provides the rare opportunity to investigate structural permeability in the form of a deviation on the SP log, which correlates with an interpretation of a fault on the dipmeter log. An in-depth structural interpretation of the Zema structure provides an insight into how this structural permeability developed. The observed structural permeability is not associated with the main bounding fault but appears to be linked to younger faults, which formed under the current stress regime that propagate through the top seal and into the reservoir near its crest.

Almon et al.’s chapter is one of few dealing exclusively with top seal, reflecting the change in the industry’s focus since the previous Hedberg conference. This chapter continues to push the knowledge boundary on top seal risk reduction by describing a methodology for seismic-based predrill evaluation of top seal capacity. The authors describe the facies variation seen in outcrop and in shallow drill holes of the Lewis Shale in sufficient petrological detail to enable their separation using discriminant function analysis. They then measured the petrophysical properties for each facies and note that where calcareous laminated shale (from highstand systems tracts) lies directly above organic laminated shales (from transgressive systems tracts), a strong seismic reflection and an associated amplitude-vs.-offset anomaly is generated. These seismic responses could be confused with reservoired hydrocarbons.

Heggeland’s chapter is a reminder of how far seismic display and interpretation have come in the last decade. He shows examples of mobile hydrocarbons in gas chimneys from the North Sea, Gulf of Mexico, Nigeria, and the Caspian Sea and classifies them into two types. Type 1 chimneys are chimneys associated with faults. These commonly have circular cross sections with diameter in the order of 100 m (330 ft) and create pockmarks where they erupt on the seafloor. They are interpreted to have a high flux rate and, depending on fault location, can significantly increase exploration risk. Type 2 chimneys are not associated with faults, and their lateral extent can be on the order of several hundred meters. These are interpreted to have a low flux rate and have proven to be good indicators of commercial hydrocarbons.

Undershultz et al.’s chapter encourages the incorporation of hydrodynamics in seal studies. They comment that hydrodynamics is a commonly underused methodology for assessing fault-sealing characteristics by stating that “hydrodynamic analysis of aquifers cut by faults can be used as an indirect indicator of the fault zone hydraulic properties.” They then use “case studies from the foothills of western Canada and the North West Shelf of Australia to define a workflow for hydrodynamic analysis in faulted strata” and provide us with a checklist for identifying which faults, or parts of faults, are sealing to aquifer flow. Faults that show medium to strong aquifer flow, either in an upfault or across-fault sense, are unlikely to be sealing to hydrocarbons.

Lowry makes a very important contribution by discussing various prospect-risking methodologies and how to risk seal, especially top seal, and how this should be incorporated into expected monetary value investment calculations. He extols the virtues of valuing prospects by varying the chance of success with the reserve distribution and states that “If seal capacity is a continuously varying function of column height, using a single Fill-Factor becomes a blunt instrument likely to lead to a distorted evaluation.” Being a realist, Lowry does not wish to overcomplicate the risking procedure and provides a hint as to when the more complicated method, which he has proposed, should be used by stating that “A common early warning sign is when a group of explorationists focused on risking a prospect start to ask ‘What are we risking?’”

Clearly, not all leakage through top seals is caused by buoyancy pressure of the hydrocarbon phase exceeding threshold capillary entry pressures, perhaps aided by a change of wettability, and much debate still exists over the role of interconnected fractures as a mechanism for leakage. A marked contrast between the Hedberg conference on seals in 1993 and that of 2002 was the increase in the number of presentations on seals related to fault traps in 2002. This probably reflects the maturation of the industry in terms of other disciplines especially geophysical interpretation, which has increased the certainty of structural modeling that allows previously problematic fault-related traps to be viable targets. The interaction of seal rocks with the Earth's paleo- and contemporary stress field is inevitably more complicated in faulted terrain, where discordance abounds, than in gently undulating plays that rely on four-way dip closures.The importance of incorporating the risk element into all such studies is a marked improvement and brings such work on seals directly in congruence with the bottom-line of industry needs.

Peter Boult and John Kaldi
Conference co-convenors and editors

REFERENCES CITED

Hillis, R. R., 1998, Mechanisms of dynamic seal failure in the Timor Sea and central North Sea, in P.G. Purcell and R. R. Purcell, eds., The sedimentary basins of Western Australia 2: Proceedings of the Petroleum Exploration Society of Australia Symposium, Perth,Western Australia: p. 313–324.

Leith, T. L., I. Karstadd, J. Connan, J. Pierron, and G. Caillet, 1993, Recognition of cap rock leakage in the Snorre field, Norwegian North Sea: Marine and Petroleum Geology, v. 10, p. 29–41.