Faulting, Fault-Zone Processes & Hydrocarbon Flow in Three-Dimensional Basin Models

S.M. Clarke, S.D. Burley & G.D. Williams


Faults and their movement histories are key controlling elements in fluid flow systems in sedimentary basins.  When faults undergo displacement, they change their fluid transmissibility properties by:

  • juxtaposing varying lithologies (and lithological properties) across the fault;
  • smearing impermeable/semi-impermeable fault rocks in the fault zones;
  • cataclastic grain-size reduction resulting from abrasion during deformation;
  • the development of a damage zone of smaller faults adjacent to the main fault which may or may not have additional, associated sealing properties;
  • pumping or valving diagenetic fluids and hydrocarbons.

A complex fault zone will exhibit varying transmissibility values in three dimensions and these will change with displacement through time.

Current basin models treat faults in three different ways: 1) faults are manually opened and closed for a fixed time period to simulate periods of fluid flow and impermeability respectively, 2) faults are permitted to open at times of high pore pressure, or 3) a bulk fluid transmissibility property is assigned to the faults.

In this study, three-dimensional, sequential restoration and forward modelling of fault displacement is carried out using computer-modelling techniques in order to build evolutionary models of both artificial and real fault zones of varying three-dimensional geometries.  The inter-relationship between juxtaposition of fault blocks within these models and fault-rock development through time is studied.  Juxtaposition relationships, argillaceous smear and cataclasis, all varying in three-dimensional space, are combined to build fault-zone models that calculate, model and visualise fault-seal properties through time within a complex, three-dimensional fault setting. The resulting models are combined with invasion percolation-driven flow-pathway modelling techniques to analyse the three-dimensional models for possible fault-controlled hydrocarbon accumulations and migration pathways.

The application of the developed techniques is demonstrated using numerical models of a well-exposed, 22km section of the Moab Fault, Utah, U.S.A. and of a Southern Gas Basin prospect and of the Artemis Field, a highly-faulted, gas reservoir in the Southern North Sea, U.K. (data courtesy of BG Group). The Moab Fault represents an exposed analogue for subsurface faults within potential hydrocarbon reservoirs that cut stratigraphical sequences of interbedded clean sands and argillaceous units. It provides a well-constrained case study to validate the three-dimensional fault-seal techniques against reality and to compare three-dimensional methods to traditional two-dimensional analysis. By contrast, the Southern Gas Basin and Artemis Field examples demonstrate the application of the developed techniques to subsurface, commercial prospects based on industry data. The Southern Gas Basin example demonstrates the value of the developed modelling techniques in evaluating the effects of case-specific potential fault-seal problems on hydrocarbon migration and entrapment in exploration settings and the Artemis Field numerical model demonstrates the applicability of three-dimensional fault-seal techniques to understanding reservoir-scale compartmentalisation problems and related production considerations.

Data Structure & Modelling

The models used in all of the examples shown here are composed of ‘surfaces’ with each surface based on a triangular mesh of points specified in Cartesian space.  Each surface can be defined as a stratigraphical or lithological horizon, a fault, an unconformity or any other geological surface.  Each point within each surface (and therefore the surface itself) is assigned a series of attributes to represent the various lithological and petrophysical properties of the geological contact or the rock volume that the surface represents.  The properties can be defined globally - the entire surface takes the same value for a given property - or they can be assigned from well data and mathematically interpolated over the areas of the surface for which there are no data. Mathematical interpolation of geological data often results in regions of some surfaces to which no data pertain or for which is it is not mathematically possible or geologically plausible to interpolate a value. The most common example of this scenario is regions of a fault surface above and below the defined stratigraphical sequence of the faulted blocks. Regions such as these are assigned a null or ‘no data’ condition and represented in all figures by black (or dark grey where indicated). In all examples, larger images and further explanations are available by clicking on the image.

IMAGE1big The example LEFT shows a simple, synthetic wrench fault setting with one fault (shown in grey) and four lithological surfaces in the footwall and hanging-wall.  From this starting model, lithological properties can be assigned to the surfaces.
In the example shown RIGHT, the lithological horizons of the faulted blocks are coloured according to their ‘permeability’ (synthetic data).  These data have been derived from well logs pertaining to the wells shown, two in the footwall and two in the hanging-wall, and mathematically interpolated over the lithological surfaces. 

The properties of the surfaces that comprise the faulted blocks of the model are mapped to the fault on both the footwall and hanging-wall based on the premise that the surfaces of the fault blocks represent unit ‘tops’. One of two methodologies may be used to achieve this:

1) Data are interpolated to the fault directly from the fault-block surfaces down the fault to the surface below. This technique makes the assumption that the data interpolated to the fault-block surfaces are valid for the entire rock volume represented by that surface.

2) Data are interpolated directly from the well logs that penetrate the fault blocks but constrained by the geometry of the fault-block surface cut-offs at the fault. This method allows high-resolution well data to be interpolated to the fault without the computational load of many surfaces within the model. In this case, the fault-block surfaces represent major successions or stratigraphical sequence boundaries and the number required is simply that which accurately defines the structural geometry of the geology over the extent of the model and therefore constrains the mathematical interpolation of data in a geologically sound manner. In the example above, synthetic permeability data have been interpolated to the fault directly from the relevant well logs using this method.

These data modelling techniques are used to produce sound geological models of structure and stratigraphy with assigned petrophysical properties. These models represent an interpretation of the geology and provide a framework with which to deterministically derive properties of fault seal.

Cross-fault Juxtaposition & Seal

Movement on a fault surface displaces the stratigraphical units of the hanging-wall relative to the footwall. In the absence of any entrained material within the fault zone, this can seal faults to cross-fault hydrocarbon flow and hence control hydrocarbon trapping if displacement on the fault juxtaposes impermeable units (seals) against permeable (potential reservoir) units (Allan, 1989). It follows that an analysis of the lithological juxtaposition (or the petrophysical properties of the lithologies) across a fault surface for a given amount of deformation can be used to determine sealing capacity to cross-fault migration resulting from fault-block juxtaposition alone.

Allan (1989) developed a graphical technique of mapping the relative positions of lithological cutoffs in both the footwall and the hanging-wall. This methodology is commonly termed Allan Mapping or fault-plane-separation modelling and generates relationships between footwall and hanging-wall stratigraphy. The relationships are visualised on a two-dimensional section parallel to the strike of the fault surface, termed a fault-surface-section (below). Using this method, the sealing capacity of a fault and potential hydrocarbon flow pathways across it can be determined from the cross-fault juxtaposition of lithologies, based on the assumptions that the fault itself has no sealing properties and it is not an open conduit to flow.


The Allan Mapping technique produces a two-dimensional model of the three-dimensional fault surface. It is a static model that demonstrates the relationship of the lithological cutoffs across the fault surface for a given temporal stage (usually Present Day) in the deformational history of the fault.

In three dimensions, comparisons of lithological or petrophysical data across the fault surface can be determined from data interpolated to both sides of the fault.


Cross-fault comparisons of fault-block lithology can be examined to understand the juxtaposition of potential reservoir and sealing lithologies. The simple, synthetic example shown LEFTdemonstrates a model of interbedded sand and shale units. The tops of these units are represented by surfaces coloured yellow and grey respectively. The sequence is faulted by one listric fault (blue) that displaces the hanging-wall by varying amounts along its length.

Using the modelling techniques described above, data can be interpolated from the fault-block surfaces to both the footwall and hanging-wall of the fault surface. In the image shown RIGHT a cross-fault comparison of juxtaposed lithological data highlights areas of sand - sand (leak) contacts shown in yellow, shale-shale (non-reservoir) contacts shown in green and potential fault-sealed footwall and hanging-wall reservoir strata (dark and light blue respectively).

Similarly, using geophysically-derived data, it is possible to analyse the cross-fault relationships in a mathematical manner. For example (BELOW) minimum juxtaposed 'permeability' (synthetic data) can be determined representing the maximum resistance to the cross-fault migration of hydrocarbons afforded by the fault at any spatial point.


Why 3D?

Traditional Allan mapping reduces the three-dimensional geometry of the fault surface to a two-dimensional representation (a fault surface section). This introduces errors in interpretation in regions of tight fault-surface curvature, the geometry of which can be misleading on the two-dimensional fault-surface section. Furthermore, juxtaposed fault-surface areas are not correct in the fault surface section and therefore the volumes of potential traps that are dependent on juxtaposed areas on the fault surface cannot be accurately determined. In the example shown BELOW, the simple, synthetic model of interbedded sands and shales shown above is resolved into a two-dimensional section (Allan Mapping). The geometries of juxtaposed properties in the section are noticeably different but, significantly, the magnitude of juxtaposed areas varies by as much as twenty percent between the three-dimensional model and its two-dimensional representative. This fault shows small variations in strike along its length, this error will increase in complexly faulted regimes with tight fault-surface curvature.


Three-dimensional analysis is not limited to one fault. Many faults within the model can be analysed for juxtaposition relationships in the same manner, not only as separate entities but also in relation to each other. In the examples BELOW, the effects of the juxtaposition relationships across more that one fault on the sealing of a fault-bounded block are clearly shown in both synthetic and real cases. In the first example, potential reservoir units may leak at one fault only to be sealed at a synthetic splay creating a fault-block trap. Examples of this scenario can be seen in the Moab Fault and Southern Gas Basin models. In the second example, the juxtaposition relationships at branch points within faulted basins may create fault-bounded, sealed compartments. The second example is taken from the Artemis Field in the southern North Sea. For examples of how such compartmentalisation by cross-fault juxtaposition can form small traps leading to associated production considerations see the Artemis Field example.

Structural Restoration and Forward Modelling

Three-dimensional juxtaposition analysis clearly has a number of advantages over its two-dimensional predecessor but it is still static and applicable only to the temporal stage in the evolution of the basin contemporary with the deformation indicated. Using three-dimensional structural deformation algorithms, models can be structurally restored and then deformed through time.  Repeated juxtaposition analysis allows investigation of changes in juxtaposition with evolution of the fault. In the example shown BELOW, the synthetic model of interbedded sands and shales demonstrated above has been restored and deformed sequentially using a heave magnitude that varies laterally across the fault surface and a heave azimuth that is perpendicular to the average fault strike (dip-slip).  The evolution of different regions of seal and leak and potential traps is clearly visible.


Fault Zone Processes – Argillaceous Smear Modelling

Generally, the relative cross-fault juxtaposition of potential reservoir and non-reservoir units across the fault determines fault-seal potential. However, there are many examples of faulted, hydrocarbon-bearing, sedimentary basins in which faults can seal hydrocarbons even in the presence of juxtaposed reservoir units. In these cases, fault seal is the result of one or a number of geological processes that operate within the fault zone.

Processes that operate within the fault zone during deformation are collectively referred to as fault-zone processes, and generate a gouge of material between faulted blocks referred to as fault rock or fault-related rock (Peacock et al., 2000). Such material can have lithological and petrophysical properties very different from those of the faulted blocks between which it resides and hence act as a further influence on the migration of hydrocarbons between faulted blocks.

Argillaceous smearing (the entrainment of clay-rich lithologies into the fault zone) is the primary example of a fault-zone process that contributes to fault-seal potential. This process, above others, has attracted much research and attempts at numerical quantification due to its particular relevance to mixed arenaceous and argillaceous sequences.

Factors that govern the development of an argillaceous smear within the fault zone are:

  • the quantity, mineralogy and distribution of argillaceous source units within the faulted sequence,
  • their thicknesses,
  • the throw on the fault.

Using these input parameters and the dynamics of smear emplacement, many workers have produced numerical models to quantify the quality of a resultant seal in terms of a dimensionless number that, when calibrated with appropriate analogue examples, can be used to express the likely sealing capacity of a fault cutting a sequence of known lithology. Of these numerical models, the Shale Gouge Ratio or SGR (Freeman et al., 1998; Fristad et al., 1997) - a measure of the proportion of argillaceous material within the fault zone - is commonly employed for the analysis of fault seal using the expression:


where Dz is interval or unit thickness of unit i, fcl is the argillaceous fraction of unit i, t is fault throw and I is the total number of units that have passed a given point on the fault surface.

The shale gouge ratio is a one-dimensional solution based on one-dimensional parameters (thickness and throw) but it may be repeated for many points on a fault surface to generate a pseudo two-dimensional solution.

Forward modelling of structural deformation as described above, allows pre and post deformational positions of hanging-wall surface and amounts of movement to be tracked with time. These can be combined with a true, three-dimensional derivation of the shale gouge ratio to generate an analysis performed within the three-dimensional model and based on the true, three-dimensional variation in geometry and properties of the fault and faulted sequence. In the example shown BELOW, the synthetic model of interbedded sands and shales demonstrated above is redefined in terms of percentage shale of the lithological units. Surfaces representing shale units are coloured red (100% shale) and surfaces representing sandstone units are coloured blue (0% shale). A three-dimensional SGR has been evaluated based on the deformation regime of the model and shale content of the faulted lithologies.


Why 3D?

A one-dimensional SGR calculation does not allow for three-dimensional variations in deformation or variations in fault and fault-block geometry and properties. This introduces inherent inaccuracies when applied to scenarios in which there is a component of strike-slip to the deformation and in which the geometry of the faulted blocks or their properties vary along strike. Using a three-dimensional SGR a more accurate prediction of the magnitude of argillaceous material within the fault zone can be achieved. The example below shows part of a major reservoir-controlling fault from Southern Gas Basin Model.The coloured line shows the displacement path followed by a specific point in the hanging-wall (indicated by the blue arrow) as the hanging-wall moves down the fault surface under an applied deformational regime. The colour variations along the path show the changes in shale content of the footwall stratigraphy over which the point moves. The main image shows the path followed under a dip-slip regime. The insert shows the path followed under an oblique-slip regime. The amount of throw is identical in both cases. Clearly the path is significantly different in length in each case but the hanging-wall point also passes over different stratigraphy with different volumetric shale contents. A three-dimensional SGR calculation can incorporate the true three-dimensional displacement and variation in shale content into the analysis of shale gouge on the fault surface.


Furthermore, using three dimensions it is possible to analyse the distribution of argillaceous material in addition to the magnitude. The example BELOW (taken from the Moab Fault model) shows the distribution of argillaceous smear resulting from deformation on the fault as predicted by a three-dimensional application of the SGR. For an in depth comparison of one- and three-dimensional SGR calculations see the Moab Fault example.


Fault Zone Processes – Cataclastic Fault Rock Modelling

In sequences dominated by arenaceous units deformation results in the grinding of the fault-block lithologies into a finer grained fault gouge. This cataclastic gouge has a porosity that can be several orders of magnitude lower than that of the fault-blocks between which it resides and hence it can contribute to sealing even in the presence of juxtaposed reservoir lithologies and the absence of argillaceous gouge.

The relationship between the parameters of deformation, the geometry of the faulted blocks and the lithology of the faulted stratigraphy with the sealing characteristics of the resultant cataclastic gouge is poorly understood. The majority of data come from laboratory-based experimentation rather than from field observation.

From laboratory experiments Scholz (1987) concluded that the amount (mass) of cataclastic fault rock produced by the abrasion of arenaceous lithologies against each other is related to:

  • The hardness of the intact rock
  • The magnitude of displacement.
  • Initial normal stress on the fault surface prior to movement.
  • The presence of cataclastic fault rocks prior to deformation.

Using these ideas he developed the relationship:


where t is the thickness of cataclastic fault rock produced, d is fault displacement, s is the normal stress on the fault surface, h is the hardness property of the fault-block rocks and k is a dimensionless parameter known as the wear coefficient.

This equation can be developed to its three-dimensional equivalent for use in the prediction of cataclastic gouge from the parameters of deformation and the properties of the faulted lithologies. In a synthetic example (BELOW) this can be modelled but in reality it is clearly difficult to define parameters such as wear coefficient and hardness for subsurface, faulted stratigraphy, especially as both parameters are, in reality, related to facies rather than lithology.

In the example BELOW, the synthetic model used in the examples above has been redefined as a sequence of interbedded sands of differing wear coefficients. By assuming a normal stress regime that corresponds to an extensional scenario, it is possible to calculate a map of the likely distribution of cataclastic gouge as a result of deformation. This result is very difficult to calibrate to fault seal capacity but can be used in a relativistic sense.

Without the large-scale analysis of cataclastic gouge in exposed faults in the field it is difficult to produce more practical three-dimensional predictions.


Combining Fault Processes – Full Fault Seal

Analyses of cross-fault juxtaposition and fault-zone processes can be combined to produce maps of fault-seal potential. This may be achieved in a mathematical manner (mathematically combining various maps) or in an abstract sense by looking for overlapping regions within a given value range.

In the example shown RIGHT, the model of interbedded sands and shales introduced above has been used to determine maximum cross-fault juxtaposed shale content and the argillaceous content of the fault-rock  (determined from a three-dimensional application of the SGR) resulting from deformation. These two fault properties have been combined mathematically and fault surface areas shown to leak on the juxtaposition map can be shown to have increased sealing capacity as a result of the fault-zone argillaceous smear.

It is clear from these examples that the interaction of fault-block juxtaposition and fault rock is of paramount importance in determining fault seal. The processes overprint one another and can produce complex variations in seal capacity over the surface of a fault in three dimensions. The interaction of this three-dimensional fault-seal with the geometry and petrophysical properties of the faulted blocks will determine possible migration pathways and traps.

Three Dimensional Hydrocarbon Flow Modelling

In the synthetic examples introduced above potential traps and leak points resulting from the sealing properties of the fault can usually be observed in the seal diagrams. In real, complex models it is often difficult to discern potential traps and leak points by visual inspection of the fault-seal diagrams on fault surfaces. Invasion Percolation (IP) techniques can be used to model hydrocarbon flow through three-dimensional basin models in order to detect possible traps and leak points.

Each model is filled with volume nodes distributed on an orthogonal grid. Each volume node represents an orthogonal rock volume equal in linear dimensions to the grid spacing. Points within one grid spacing of a fault represent the fault zone rather than bulk rock volume. The points are seeded with permeability values from the surfaces or faults and transmissibility can be calculated between any two neighbouring nodes and used to determine hydrocarbon flow pathways through the model.

In the synthetic example BELOW - representing an extensional margin with pre- and post-rift sequences and a synthetic fault block - 'permeability' values have been interpolated from the surfaces to the volume nodes and describe regions of high permeability representing possible traps (reds and oranges – post-rift sequence and fault block) separated by low permeable units representing potential seals (blues and greens). Within each unit there is also a variation in permeability but this is not evident from the surface colouring since the variation between units is several orders of magnitude greater than that within units.



In the example LEFT, the faults are not open pathways to hydrocarbon flow (no fault-parallel flow) and there are no fault rocks. The juxtaposition of permeability is the only controlling factor on hydrocarbon flow across the fault zone. Starting at the base of the footwall, flow can be seen to percolate upwards until trapped by the juxtaposition of permeability across the fault. Flow proceeds by repeated cross-fault leakage and up-dip flow to fill the potential traps in both the fault block and the post-rift sequence.

Argillaceous smear can be calculated for both faults within the model and incorporated into the flow algorithm. In the model shown BELOW flow pathways are affected by both the juxtaposition of fault blocks and fault zone argillaceous smear. In this example the higher fault-block potential traps and the post-rift are sealed by the fault-zone smear.


Alternatively (RIGHT), the fault can be considered as an open pathway to hydrocarbon flow and hydrocarbons percolate into the fault zone, filling it and spilling to the top post-rift potential trap but not filling potential traps in the pre-rift or the fault block. IMAG18big

Invasion Percolation-driven hydrocarbon flow pathway modelling is a very fast way of discerning flow pathways through faulted structures. Properties of flow are not calculated at each spatial and temporal step therefore models run many times faster than FD or FE driven methods.

The four, down-loadable movie clips BELOW are real-time screen captures of the fault juxtaposition simulation and the juxtaposition combined with argillaceous smear simulation shown above. In these examples the model surfaces are shown coloured by permeability and the predicted pathway depicted in blue (for clarity). In all videos the fault is not open to fault parallel flow.

 Movie 1: IP Flow controlled by lithological juxtaposition alone

Movie 2: IP Flow controlled by lithological juxtaposition alone (reverse angle)

Movie 3: IP Flow controlled by juxtaposition and argillaceous fault rocks.

Movie 4: IP Flow controlled by juxtaposition and argillaceous fault rocks (reverse angle)

The speed of an IP approach to flow modelling makes it an effective method for use in risk-driven analysis. Many realisations of the same model can be performed with different inputs and the effects of poorly constrained input data on the range of possible outcomes explored. Case-specific geological uncertainties can be explored by simply manually changing properties of the model and re-simulating flow. Alternatively, the effects of generalised uncertainties, such as those associated with the fine detail of the stratigraphy or structural geometry, can be explored using a stochastic modelling approach.

For further examples of the use of IP modelling in discerning flow pathways through faulted structures see the Moab Fault and Artemis Field examples. The exploitation of IP modelling to examine the effects of stratigraphical uncertainties on hydrocarbon accumulations in faulted structures is demonstrated by the Southern Gas Basin example.

Further Information

The three-dimensional modelling presented here was performed using in-house modelling software developed by the Basin Dynamics Research Group at Keele. This software has now been developed into a commercial three-dimensional fault seal analysis and flow pathway modelling software package called Qfault. The software combines all of the three-dimensional modelling techniques presented here with a powerful advanced visualisation and model-building tool. Images produced specifically using Qfault have the Qfault logo on them.

For further information on Qfault or any of the techniques demonstrated here, please contact the corresponding author:

Stuart Clarke


Allan, U.S. 1989 Model for hydrocarbon migration and entrapment within faulted structures. AAPG Bul. 73 803-811
Freeman, B., Yielding, G., Needham, D.T. & Badley, M.E. 1998 Fault seal prediction: the gouge ratio method. In: Coward, M.P., Daltaban, T.S. & Johnson, H. (eds.) Structural Geology In Reservoir Characterisation. Geol. Soc. Ldn. Spec. Pub. 127 19-25
Fristad, T., Goth, A., Yeilding, G. & Freeman, B. 1997 Quantitative fault seal prediction: a case study from Oseberg Sud. In Moller, P. & Koestler, A.G. (eds.) Hydrocarbon Seals: Importance for Exploration & Production. NPF Special Pub. 7 107-124
Scholz, C.H. 1987. Wear & gouge formation in brittle faulting. Geology 15 493-495tial.