Shale Plays Need Seals Too

In an earlier post, I argued that there may be significant lateral migration within shale reservoirs that can lead to higher maturity fluids produced from lower maturity areas, and even occasionally dry gas production in the oil window. In this post, I would like to propose that shale reservoirs also need seals to work. 

Sedimentary rocks have a wide range of pore sizes. In a conventional reservoir, HC saturation builds up due to higher capillary pressure caused by the buoyancy of the column (Schowalter, 1979). Saturation is highest at the crest of the reservoir. 

In a shale reservoir, there may not be an effective column. The increase in saturation and capillary pressure is caused by generation of hydrocarbons. However, it will also require the presence of tight rock facies (above, below and laterally) to prevent migration out of the shale due to the increased capillary pressure. From MICP studies on shales, we see that shales have a wide range of displacement pressures (Pd), from 200 psi to >10,000 psi. The typical tight facies may have a Pd of ~6,000 psi mercury-air (~320 psi oil-water). After saturating the adsorptive kerogen, the generated HC fluid begins to fill the zones with larger pores (the reservoirs with low Pd) first. As saturation in the reservoirs builds up due to continued generation, capillary pressure increases, as hydrocarbons invade progressively smaller pores. Saturation may reach >50% when the capillary pressure exceeds the Pd of the seal and migration out of the shale begins. Obviously, without the sealing facies, a homogeneous rock cannot retain high saturation.

HC wet or partially HC wet pores may initially build up saturation without increasing capillary pressure.

 

From the above reasoning, higher Pd for the seals inter-bedded with the more porous zones at various scales ( millimeters, inches, to feet ) leads to higher saturation in reservoir intervals. Some shale plays may not work due to the lack of seals rather than the lack of porosity. Most studies on shale plays to date have focused on the porosity of the reservoirs, which ranges between 5 and 15% typically. If you agree with the above argument, perhaps it is also important to look at the seals (with less than 5% porosity) inter-bedded with, above and below the reservoir zones. 

Zhiyong He, ZetaWare, Inc.

Dry Gas, Wet Gas, Condensate and Condensables

At a recent industry conference a poster summarised aspects of the petroleum systems in a particular basin. The authors noted that some reservoirs contained "dry gas" while others contained "wet gas". The boundary between the two was not defined but it was clear from the context that the distinction reflected the condensate content: gases having more than about 20 bbls/MMscf  of condensate were classified as "wet".

Wet vs. dry gas definitions and terminology can be confusing so I thought it might be worth posting a summary here. Firstly, let's look at the composition of a typical gas condensate:

This one is from the textbook on the phase behaviour of reservoir fluids by Pedersen and Christensen (2007). We can define four groups of compounds: Methane (C1) being the only member of the first group then ethane (C2), propane (C3) and the butanes (normal and iso) making up the remaining "permanent" gases, the "condensate" range with compounds having from 5 to 14 carbon atoms and the "oil" range consisting of compounds with 15 or more carbon atoms. The "condensate" and "oil" ranges are labelled that way because fluids with most of their liquid mass in those carbon number ranges tend to be gas-condensates and oils in the sub-surface respectively. The "permanent" gases are in the gas state at standard surface conditions of 1 atm pressure and 15 C (60 F)

The ideal condensate-gas ratio or "CGR" of a fluid is the ratio of the liquids to the gas species, usually expressed as their respective volumes under standard surface conditions. In the US, "oil field" units are used so that CGR is barrels of condensate per million standard cubic feet of gas (bbls/MMscf). In Europe the units are more commonly cubic metre gas per cubic metre liquids (M3/M3).

When speaking of wet vs. dry gas in the conventional E and P realm, mostly we are referring to the CGR. What is a "significant" CGR depends on the context, particularly the value it may add to a gas development. For example, for an LNG development based on a 5 trillion cubic feet (TCF) resource, a CGR of 10 bbls/MMscf would yield 50 million barrels of condensate (in ideal circumstances).

There are few standard definitions in the literature but (a) the state of New Mexico defines a "gas" well as one producing a fluid with less than 10 bbls/MMscf of liquids  (see http://164.64.110.239/nmac/parts/title19/19.015.0002.htm. item G6) and (b) the Encylcopedia Brittanica on-line defines a wet gas as anything containing more than 2.5 bbls/MMscf.

The standard reservoir engineering definition of a "dry gas" is one that yields ZERO liquids at surface temperature and pressure. On the phase (P vs T) diagram for such a gas, the isotherm of surface temperature does not intersect the phase curve at any point. Another way of saying this is that the cricondotherm for this fluid is lower than surface temperature. The corresponding definition of a "wet gas" is one that will yield some liquids at surface temperature and pressure but there is no pressure at which liquids will begin to condense at reservoir temperature. For a wet gas, the cricondotherm lies somewhere between the surface and reservoir temperature. A gas-condensate is a fluid for which a reduction in pressure at reservoir temperature will, at some point between initial reservoir and surface pressure, cause liquids to drop out.

In the realm of unconventionals, a somewhat different terminology - one used by the natural gas industry -  is prevalent. Gas wetness refers to the content of C2+, i.e. everything except methane is the "wet" stuff. These are called the "natural gas liquids" or "NGL" even though ethane through to the butanes (C2 - C4) are not liquids under standard surface conditions. The condensate fraction (C5+) is a sub-fraction of the NGL and, just to be extra confusing, the NGL are also called the "condensables". This comes about because during natural gas processing methane - the ultimate "dry gas" -  is separated from all other compounds by either cryogenic cooling or by absorption. The condensate fraction of the condensables (the C5+ bit remember !) can also be called "plant condensate" or "natural gasoline". Just for completeness, let me add that the "liquefied petroleum gas (LPG)" that we use in our barbequeue, car or for cooking at the vacation house is propane, butane or a mixture of the two.

Confused yet ? Now enter the geochemists: Gas wetness to a geochemist is defined as the molar ratio or percentage of the "wet" gases to the total of C1 to C5 gases with no consideration of the condensate species at all. . Errr...except that the pentanes - liquids in a cool room but gas in a hot room (37 C/97 F) -  are generally included. Thus, we calculate the wetness of mud gas (gas while drilling) as the sum of C2-C5/ the sum of C1-C5 and express it as a percentage.

The wetness of a gas (geochemical definition) and condensate-gas ratio are clearly related. However, the relationship is specific to a particular petroleum system and also to processes which may have altered the fluid during movement from source to trap and/or in the reservoir. The figure below shows how gas wetness relates to CGR for several different gas-condensate fields, each hosting stacked accumulations. It is obvious that the gases in field 3 have a very different character to those in the other fields. In particular there is almost no change in gas wetness for fluids varying in CGR from ~ 30 to ~ 75 bbls/MMscf. This implies a decoupling of the gas and liquid fractions of the charge with, for example, a fixed amount of liquids being diluted to variable extent by a near fixed composition gas. This in turn might imply different source kitchens and migration routes or some migration fractionation process.

Finally, it is worth noting that the phase behaviour of a gas-condensate system also depends on the composition of the gas and liquids fractions. The figure below shows gas chromatograms for several condensates with different compositions and corresponding to fluids with different CGRs (nb: no chromatogram is available for fluid "F")

Note that condensate E is slightly contaminated with an olefin based synthetic drilling mud.

The pressure at which a gas condensate begins to separate into oil and gas in the subsurface is called the "dew point" or "saturation pressure (Psat)". This pressure is a function of the CGR but also of the mutual miscibility of the liquid and gas components. The most important factor is the composition of the liquids (condensate). If they are very light, they will more easily enter the vapour phase so that we have a higher CGR for a given dew point. Conversely, the heavier the liquids, the lower will be the CGR for the same saturation pressure. The dew point pressure vs. CGR data for the condensates A - G above are shown in this figure (along with some data from the UK North Sea petroleum system and a published emprical correlation for the same area. (nB; gas-liquid ratio - GLR - is displayed - CGR = 1/GLR *1,000,000)

Note the wide variation related to condensate composition. In particular, note that the very light condensates B and C (yellow dots) show low dew point pressures even though they have a high CGR (~ 95 bbls/MMscf, or GLR ~ 10,000). This reflects the high mutual miscibility of the liquids and gases for these fluids. It is often assumed that finding a high CGR gas in a shallow reservoir increases the likelihood of a finding oil in the system. In fact, the reverse is true: If the liquids are light enough to remain in the vapour phase even in high concentration, there are few oil range molecules present. Condensates of type "G" (43 bbls/MMscf) are much more likely to be found in association with oil.

When are Rift Models relevant for the Petroleum System ?

There are currently contrasting views on the way strain is distributed within the lithosphere during rifting and the formation of passive continental margins, with direct implications for the subsidence and heat flow histories of the overlying sedimentary basins, and potentially also for the timing and degree of source rock maturity in these systems.

According to some authors, the asymmetry observed between most conjugate margin pairs (e.g. West Iberia-Newfoundland, East Coast USA-NW Africa, NE Brazil-West Africa and the Southern Australia-Antarctica) results from the activity of low-angle normal faults (detachments), which shift the region of pervasive upper crustal thinning and normal faulting (lower plate) from that of intense lower crust and mantle lithosphere thinning (upper plate; see Rosenbaum et al., 2008 and references therein). A paradoxical observation, nevertheless, is that in most margins the extension measured from normal fault throws appears to be much smaller than that inferred from subsidence and gravity modelling, thus implying ubiquitous upper-plate rift margin settings (the “Upper Plate Paradox”; Driscoll & Karner, 1998; Davis & Kusznir, 2004). 

Pervasive depth-dependent stretching (DDS) is also implied in dynamic models of rifting to explain features such as the deposition of salt over extremely thinned crust (e.g. off western Angola) and the exhumation of continental mantle prior to breakup in magma-poor margins (e.g. the West Iberia Margin; Lavier & Manatschal, 2006; Huismans & Beaumont, 2011). In contrast, results from a recently published kinematic rift model suggest that the crustal structure and subsidence along most passive continental margins can be explained assuming an essentially depth-uniform strain distribution through time (Crosby et al., 2011). Alternative models have also been put forward to explain the apparent deficit of extension in the brittle upper crust, namely by Reston (2005) and Ranero & Perez-Gussinyé (2010), who argue that the amount of extension accommodated in normal faults may have been largely underestimated in earlier studies.

The figures below illustrate the results from two simple experiments in actively explored rift settings: (Figure 1) the North Sea; and (Figure 2) the Angola passive continental margin. The pseudo wells were built from published seismic data and assume a simplified stratigraphy, where the thin black layers correspond to the location of two hypothetical source rocks in each setting. For simplicity all models assume a constant temperature at the base lithosphere of 1330⁰C, and the source rocks use a Type II, marine shale kerogen facies, with an initial TOC of 5% wt and HI of 500 mg/g TOC. The impact of varying the rift model assumptions is then evaluated in terms of the SR’s maturity.

 

The North Sea comprises a series of rift basins that formed over a sequence of extensional pulses between the Permian-Triassic and Early Cretaceous, interspersed with periods of thermal quiescence, volcanic activity and doming (Ziegler and Cloetingh, 2003). The Pseudo-well in this experiment was built from a NW-SE seismic constrained transect redrawn from Bell et al. (2014), at a location where the inferred total stretching factor (β) is 2; i.e. the crust, or the whole lithosphere, have been stretched to half their initial thickness during rifting (if the rift is assumed instantaneous; McKenzie, 1978). For the purposes of the experiment it is assumed that all extensional deformation took place during the Late Jurassic (160-150 Ma), except in the last scenario, where most thinning occurs during an earlier rift stage, in the Permian (260-250 Ma), in agreement with the published profile (see Bell et al., 2014 and references therein). 

The models show that changing the amount of lithosphere thinning within a reasonable range (black lines), imposing significant differential stretching between crust and mantle, has some impact on the timing/degree of maturity of the deeper, pre-rift SR. This results from differences in the post-rift thermal structure of the basin combined with rapid sediment burial. However, a similar effect is obtained by varying the steady state thickness of the lithosphere by only ±10 km, often beyond realistic model constraints, and a greater impact is even predicted when distributing the extensional deformation over several rift events (or varying the duration of rifting). The maturity of the shallower SR is independent of the rift model, although some differences are noticed for variations in the thickness of the steady state lithosphere.

 

The Angola (deep) passive continental margin formed due to intense stretching during the Early Cretaceous (mostly Berriasian-Aptian) followed by a transition period of thick salt deposition (Aptian) and continental break-up (e.g. Teisserenc & Villemin 1990). The transect shown above is redrawn from Lentini et al. (2010), based on deep seismic reflection and refraction data. At the location of the Pseudo-well the present day thickness of the crust is 8 km, measured between the base of the sediments and the Moho. For the experiment it is assumed that all extensional deformation took place during the Early Cretaceous (145-135 Ma) and that the initial crustal thickness is 32 km (i.e. β_crust = 4).

In the margins, where the lithosphere stretches to infinity prior to break-up, depth dependent stretching (DDS) may have a greater impact on the distribution of heat during and following rifting, and thus in the maturity of SR’s. In the experiment above this is observed when varying the amount of stretching in the mantle (β_mantle) between a factor of 3 and 4. For higher stretching factors, in this particular setting, the increase in heat flow converges asymptotically. The models also show, however, that similar magnitude effects, or even more pronounced, are produced when varying the thickness of the lithosphere and/or the duration of the rifting events. As in the case of the North Sea experiment the maturity of the shallower SR is independent of the rift model.

In summary, the experiments discussed here show that the implications of assuming conceptually different rift models for the timing and degree of source rock maturity in these settings may be of the same order of magnitude, and thus indistinguishable, from those inherent to the uncertainty in the parameterization of the rift model, such as the thickness of the underlying lithosphere and the age and duration of the rift events. Moreover, it is likely that the maturity of most syn- and post-rift source rocks does not depend significantly on the rift model, but mostly on the rate of post-rift burial. As good practice, these effects should be tested in order to identify the key sensitivities of the basin model, at least within a first order approximation.

 

References:

Bell, R. E., C. A.-L. Jackson, P. S. Whipp, and B. Clements (2014), Strain migration during multiphase extension: Observations from the northern North Sea, Tectonics, 33, doi:10.1002/2014TC003551.

Crosby, A. G.,  N. J. White, G. R. H. Edwards, M. Thompson, R. Corfield, and L. Mackay (2011). Evolution of deep‐water rifted margins: Testing depth‐dependent extensional models, Tectonics, 30, doi:10.1029/2010TC002687.

Davis, M., and N. Kusznir (2004), Depth-dependent lithospheric stretching at rifted margins, in Karner, G. D., Taylor, B., Driscol, N. W., & Kohlstedt, D. L (eds), Rheology and Deformation of the Lithosphere at Continental Margins, pp 92-137 Columbia University Press.

Driscoll, N. W., and G.D. Karner (1998), Lower crustal extension across the Northern Carnarvon Basin, Autralia: Evidence for an eastward dipping detachment, Journal of Geophysical Research, 103, 4975-4992.

Huismans, R., and C. Beaumont (2011), Depth-dependent extension, two-stage breakup and cratonic underplating at rifted margins, Nature, doi:10.1038/nature09988.

Lavier, L.L., and  Manatschal, G. (2006) A mechanism to thin the continental lithosphere at magma-poor margins, Nature, 440, doi:10.1038/nature04608.

Lentini, M.R., S. I. Fraser, H. S. Sumner and R. J. Davies (2010), Geodynamics of the central South Atlantic conjugate margins: implications for hydrocarbon potential, Petroleum Geoscience, 16, 217-229, DOI 10.1144/1354-079309-909.

McKenzie, D (1978), Some remarks on the development of sedimentary basins. Earth and Planetary Science Letters, 40, 25-32.

Ranero, C.R. and M. Perez-Gussinyé (2010), Sequential faulting explains the asymmetry and extension discrepancy of conjugate margins, Nature, doi:10.1038/nature09520.

Reston, T.J. (2005), Polyphase faulting during the development of the west Galicia rifted margin, Earth Planetary Science Letters, 237, 561-576, doi:10.1016/j.epsl.2005.06.019.

Rosenbaum, G., R. F. Weinberg and K. Regenauer-Lieb (2008), The geodynamics of lithospheric extension, Tectonophysics, 458, 1-8.

Teisserenc, P. and J. Villemin (1990), Sedimentary basin of Gabon; geology and oil systems, in Divergent/passive Margin Basins, AAPG Memoir 48, 117–199.

Ziegler P. A. and S. Cloetingh (2003), Dynamic processes controlling evolution of rifted basins, Earth-Science Reviews , 1-50, doi:10.1016/S0012-8252(03)00041-2.