Potential for Thin Bed and Lateral/Spatial Heterogeneity Analyses in Mudstones via Gamma Ray Logs, Drilling, and Geosteering: Union Springs Member of Marcellus

Potential for Thin Bed and Lateral/Spatial Reservoir Heterogeneity Analyses in Mudstones Via Gamma Ray Logs, Drilling, and Geosteering (Emphasizing the Union Springs Member of the Marcellus) - by Kent C. Stewart – Blue Dragon Geoscience, LLC (July 2015)

Introduction

This paper is an exploration of drilling and geosteering in the Marcellus Shale, particularly the organic-rich facies of the Union Springs member. Some structural, geochemical, and regional fracture system history are also discussed.  It is not meant for publication but simply to compare characteristics of deposition, structure, and rock properties (petrophysics) to drilling and geosteering.

If thin bed geology and stratigraphic reservoir variation could be better characterized, that knowledge could aid hydraulic fracture design, geo-targeting, knowledge of reservoir rock depositional and post-depositional variations, identification of possible frac barriers or boundaries that distribute frac energy, and a whole host of other processes.

Formation tops from well logs are typically called on inflection points between the extremes. However, with high resolution logs, and with multiple types of logs, one might decipher considerable thin bed geology and stratigraphic variation. With the current horizontally drilled wells, mainly shale wells, gamma ray is the tool of choice due to price and versatility. Gamma ray logs are probably the most correlatable log over long distances although other logs might do better with certain boundaries. Commercial mapping program suites like Geographix are able to delineate thin beds by log values separating them out into different colors. Horizontal wells drilled closely spaced on multi-well pads can give detailed mapping of local structural geology and depositional variation which can aid reservoir analyses. Structure and depositional variation along with analysis of reservoir accessed via geosteering can aid overall reservoir analysis when integrated with frac design and post-frac analysis.

Bedding Planes and Rock Contacts

Boundaries between beds are interesting places. These bedding planes as they are sometime called might host a number of geological processes. Bedding planes are zones of weakness that can transmit energy and sometimes fluids. Fracturing and fault movement are more likely along bedding planes than not. Bed boundaries also indicate a change of conditions such as deeper or shallower water, or higher or lower energy conditions that might enhance or occlude porosity. Induced fracture propagation is more likely along bedding planes. If available, upper and/or lower hard beds can confine induced fracs to a preferred zone by acting as barriers. Bed boundaries can confine the drill bit to a certain preferred zone in much the same way. However, they can also make it difficult to cut through to a different preferred zone. Bed boundaries seem to be more amenable to being picked on gamma ray logs mainly from direction changes rather than on inflection points as there may be several direction changes along an overall long gamma change in one direction. It can often be determined via drilling and geosteering where a bit is deflected off of a hard bed boundary. Such bed boundaries usually correspond to a direction change on the gamma ray log. A bed boundary where an impermeable rock such as shale or chert lies over porous reservoir rock can make a seal for the reservoir of fluids whether they are gas, oil, saltwater, freshwater, or sequestered CO2. A reservoir seal can also be a pressure seal. Contacts between organic-rich hot shales/mudstones and limestones (sequence boundaries) in the Marcellus can show quite a bit of variation in ROP due to significant differences in composition and rock properties. Even as I type the Mars Rover is climbing a Martian mountain in order to analyze an important rock contact. 

Anisotropy Variations at Different Scales More Common to Mudstones

In mudstones, which overall are highly anisotropic, there are variations in anisotropy at bed scale. Beds deposited by suspension settling tend to be anisotropic. Highly anisotropic rocks tend to respond better to induced hydraulic fracture treatments. Although alignment of clay minerals is responsible for much of the anisotropy of mudstones, many of the organic-rich source-rock mudstones are much lower in clay content than the typical shale/mudstone and much higher in biogenic silica or calcite. Also common to mudstones is early diagenesis, which reduced compaction and so preserved porosity. Some anisotropy in these mudstones is due to elongated grains (silica, calcite, or organic) due to compaction but the early diagenesis tends to reduce compaction and elongation of grains. The precipitated silica and carbonate cements favor brittleness over ductility when further compacted and when hydraulically fractured. Other interbedded units may be more isotropic. Mixing by bioturbation can prevent some anisotropic fabrics from developing or destroy them while they are developing. Shales/mudstones have significant vertical heterogeneity as they are often composed, in part, of many thin beds that have variable properties. The bottom line here is that each bed on whatever scale (m, cm, mm) can vary in porosity, clay content, brittleness, and ability to propagate induced fractures. It is also difficult to derive rock properties from seismic since there are these property changes at scales below seismic resolution. Some researchers have identified ductile-brittle cycles dependent on climate variations during the very long deposition time of shales/mudstones. Thus, geo-targeting in the most porous, brittle, and clay-poor zones is very important for ideal induced hydraulic fracture propagation and potential well production. As Engelder and Lash pointed out, it is thought to be coincidental that the “stress-induced permeability anisotropy” of the Marcellus is parallel to the higher joint density of the J1 fractures - so that wells drilled perpendicular to the J1s and the SHmax (maximum principal stress direction) encounter the most natural regional fractures AND are the most “fracable” due to the current stress regime.

Middle Devonian Burial, Thermal, and Regional Fracture History

The rocks in the east of the play were buried deeper and over a longer time period and there the Middle Devonian sediments are much thicker. In addition it is thought that at least 12,500 ft. of sediment was eroded, mostly in the Permian, along the Allegheny front and eastern part of the basin. Such conditions allowed for generation of hydrocarbons through catagenesis. The dry gas, liquids, and oil windows are now well established. Marcellus is predominantly a dry gas reservoir with some natural gas liquids and little to no oil. High fluid-filled pore pressure from Devonian through Alleghanian time likely propagated the regional fracture systems through compaction disequilibrium. The Alleghenian J1 fractures are most numerous and favorably oriented so that drilling perpendicular to them (WNW-ESE) leads to the best production.

Present day pore pressure and degree of tectonic natural fracturing (due to Alleghenian compression) may aid in faster drilling and may possibly be associated with better hydraulic fracture propagation. Areas with more tectonic fracturing and lower pore pressure may drill faster and frac better, with the ability to accommodate higher pump rates during frac jobs. However, other factors such as BHA, bit design, bit jet configuration, mud properties, target selection, mud volume, and pump rates may also contribute to faster drilling.

It is now fairly obvious that targeting and drilling the lateral in the most organic-rich zones yields the best production in the Marcellus. This is not true of other “shale” plays such as the Bakken and the Utica/Point Pleasant. In the Marcellus, having the most direct contact via borehole perforations, with the organic-rich, gas-rich facies of the Union Springs is the key to the best wells. There may be some exceptions to this where a particularly brittle zone occurs adjacent to the hottest Union Springs.

    

Characterizing Lateral/Spatial Heterogeneity through Geosteering

Successful geosteering is dependent on correct correlation. Lateral/spatial heterogeneity can be determined if gamma ray between two clear correlations differs. Thickening and thinning of beds is common where there may be scour surfaces and other slight changes in sequence stratigraphy. Deposition may thin on pre-existing structural “highs” and thicken on pre-existing structural “lows.” This draping on pre-existing structure is a feature of deposition by suspension while bed-load and bottom current deposition (deposition of terrigenous sediment in deltas or submarine fans) tends to show clinoforms. A recent AAPG paper on sequence stratigraphy of the Union Springs member of the Marcellus and adjacent stratigraphic units (Kohl etal) shows that localized thinning of thin beds is common in various units along the western and northwestern parts of the basin and this may be due to draping over pre-existing structure. Sediments may move more in shallow water than in deeper water and there may be minor bottom currents in the Union Springs but most sedimentation is likely due to suspension settling.

In terms of correlation in geosteering, if there is a thinning of a bed or beds relative to the type log, then the apparent dip should change as that section is drilled through. If the section is then drilled through in the opposite direction the change in apparent dip should be in the opposite direction (up vs. down or vice versa). Such a scenario is the best geometrical indication of a thinning section. A thickening section relative to type log would similarly show changes in apparent dip but may be more difficult to determine as section would be apparently added rather than apparently missed. Added section is easier to integrate into an existing interpretation and so is less apparent as anomalous. The apparently missing sections in the bed thinnings are likely often interpreted as small faults due to apparent lost section but as stated above if there is an apparent dip change of similar magnitude but opposite orientation then the thinning section is more likely the correct interpretation. Thinning in beds in the Union Springs member of the Marcellus seem to be on the order of 1.5 ft or less. This could be significant in that a presumed 10 ft target window from the type log may be in actuality only 8.5 ft. 

     

In some plays the gamma ray is not very distinctive or variable in reservoir zones. Some companies are utilizing chemo-stratigraphic techniques or geochemical geosteering via XRD/XRF from drill cuttings in those sections. This could be quite useful but is limited to where it is applicable since gamma ray provides excellent correlatable curves in most cases. If used widely, it could also help map out bed and sequence boundaries. The Point Pleasant Calcareous Mudstone target zone may be amenable to this technique due to limited gamma variability. The Marcellus, however, seems to show significant gamma variability in the best targets through most of the fairway. There could be developed an XRD/XRF regional and local databases of multiple wells where XRD/XRF responses could be compared, heterogeneity determined, and compared with that expected by depositional models. Presumably, the same could be done with mass specrometers (MS). In both cases, newer, much cheaper, hand-held models could be used on wells past and present to develop such databases in order to make such research economically feasible. 

Effects of Structure and Natural Fracturing on Geosteering and Well Production in the Marcellus Play

In some areas of the Marcellus play there is significant structure, including large faults, steep dips, and multiple tight folds.  It is often difficult to delineate the structure in great detail on seismic due to seismic resolution limitations. Old 2D seismic may be difficult to interpret. In those areas, it is likely that laterals will be drilled out of zone for significant amounts of the lateral, with resulting lower well production. The presence of large faults may also preferentially dissipate the energy of fracs, resulting in poor hydraulic fracture propagation in zones interfacing the faults. Folding may help or hinder fracturing. It may increase tectonic fractures (such as so-called type III fractures at fold crests and type II fractures along fold flanks) but it may also lead to less lateral in zone. Some engineers think that a greater density of natural fractures due to tectonics (folding and faulting) may lead to higher initial production but lower sustained production as greater density of natural fractures can lead to wider fractures near the wellbore but less frac length, so less overall stimulated reservoir volume (SRV). Wider fractures near the wellbore have been shown to increase initial production. Other types of fractures such as microcracking during catagenesis may increase effective permeability and aid in regionally updip migration of fluids and hydrocarbons in generally impermeable source rocks, especially when interfacing regional vertical joint systems.

Depositional Models for Marcellus Shale

Prograding Mahantango Delta System with Clastic-Starved Distal Deep to the West where Hemipelagic(?) Basinal Facies was Dominant

Kohl, etal, favor this model where the prograding Mahantango delta system brought terrigenous sediment from the east-southeast evidenced by the areas adjacent to Harrisburg, PA being rich in clastics. The Marcellus overall is thought to have been deposited at the toe of a clinothem prograding west-northwest. The deeper part of the basin seems to be coincident with the Rome Trough associated with the opening of the Iapetus Ocean in Cambrian time. This deep is distal to the delta system and starved of clastics. The observation here is that a transgressive systems tract (TST) led to first deposition of the hot gamma organic-rich Union Springs with the rest of the Union Springs being part of a high stand systems tract (HST) and falling stage systems tract (FSST) as it got younger. The TST and HST units were likely deposited as pelagic or hemipelagic basinal facies by sediment suspension in low energy deep water. The FSST may have involved some progradation of clastics from the east. This model assumes deep water (>330 ft) and continuous anoxia over hundreds of thousands of years. The Union Springs was deposited on the widespread Onondaga Limestone which was the result of a lowstand systems tract (LST). It is thicker in the west and north due to development of a carbonate bank and reefs in shallower water. These banks set the stage for future onlapping of sediments filling the basin from the east. Kohl, etal noted that there was no bioturbation found in the deeper part of Union Springs deposition in one core – which favors deeper and anoxic water. They also note that the height of Onondaga pinnacle reefs show that the carbonates grew to a height of 220 ft. This also argues for deeper water deposition. Other researchers favor shallower water deposition of Marcellus. Kohl, etal also note that following deposition of the Union Springs the Purcell/Cherry Valley was deposited on bathymetric highs but they acknowledge that the bathymetry of these intervals is not well established. They note that subsequent base level rise after Union Spring deposition sequestered sediment from the east and allowed widespread limestone deposition of the Cherry Valley. They also note that some limestone may have been detritus removed by erosion from the northwest which may have been exposed during low stands.

The presence of volcanic ash beds (bentonites) below and within the Union Springs member allow for chronostratigraphic correlation where present and there are also some bio-markers which can be correlated in the Oatka Creek member of the Marcellus, above the Purcell-Cherry Valley limestone. These can aid in establishing sedimentation rates which can be suggestive of water depths and variation of water depths in different areas.

Alternative Depositional/Anoxia Model for Organic-Rich Source-Rock Mudstones

Recent analysis by Taury Smith in the Marcellus and Utica/Point Pleasant suggests that these shales were not mainly deposited at the base of down-lapping clinoforms in a deep sea as in the suggested analog of present-day anoxia in the deep Black Sea. He suggests that it is possible that there was a seasonal anoxia at the base of a much shallower sea, with sediments deposited not at the base of clinoforms in deep water but in an onlapping sequence in the western part of the Appalachian Basin where mapping definitely shows successive thinning and pinching out of units in the Middle Devonian formations: Marcellus, Hamilton, Tully and the Upper Devonian Burkett/Geneseo. My own mapping of these units throughout the basin confirms this configuration. He notes that the shallow water anoxic zone of highest TOC seems to occur on the western flank of the basin rather than at basin center – on the craton-ward side with increasing interbedded limestones. One reason for this is lower dilution of the organic matter by influx of terrigenous material. Another may be faster settling and burial due to shallower water. Smith also notes that using a datum above the shale of interest will show the true structural architecture of that shale. The shallow water anoxia model is questionable as the level of seasonal organic matter for the shales would have to have been quite high but over the long periods of time that the shales were deposited this is quite possible. Decaying algal blooms, possibly some fed nutrients by volcanic activity, could have been a source of much of the organic matter. The seasonal anoxia and shallow water model also explains the presence of benthic fossils such as brachiopods, bioturbation, and scour surfaces found in Marcellus thin sections. This may be true of other organic-rich source-rock mudstones as well. Bottom water oxygen levels can be estimated by the amount of bioturbation. Smith also favors a shallow water model for the Upper Ordovician Utica Shale. The model there is more solid as carbonate contents are quite high compared to Marcellus and there are also significant benthic fossils and bioturbation. His model suggests that the anoxic environment could even have been in sludgy zones just beneath the sea bed.

Comparison of the Two Depositional Models

It is not thought that these two models are entirely mutually exclusive. The use of the term “hemipelagic” in the first model (Kohl etal) is a bit unclear. If the lower Union Springs organic-rich deposition is both distal and in the deepest part of the basin then one would think it would be pelagic (slow deposition in deepest water) rather than hemipelagic (faster deposition than pelagic and in shallower water). If by hemipelagic the authors mean deposition along the distal carbonate bank then that would imply shallower water than the basin center. The main difference between the two models is water depth. Another difference based on water depth is continuous anoxia vs. interrupted seasonal anoxia. Kohl etal seem to indicate that since 220 ft. pinnacle reefs occur in the Onondaga then that would be a minimum water depth during that inundation and since that occurred updip and up-basin the water would be presumed to be deeper toward basin center. Both models acknowledge onlapping topographic highs to the west and north (craton-ward, which is the common basin architecture). The organic-rich facies of the lower Union Springs is developed both near basin center and along the distal margin where it onlaps the carbonate banks to the north and west, possibly also inter-fingering with limestones sourced from those areas.

Some Observations and Conclusions

Geo-targeting in the Marcellus can be difficult in that interbedded limestone is significant in the generally thin hot gamma units. The question arises – does one try to stay in a thinner interval of hot shale or an overall thicker zone to access other hot shale units on the other side of lime stringers? While it may not be possible to stay in thin zones where there is significant structural geology, it is possible where the dip is milder and less variable mainly on the western margins of the play. One approach here might be to compare hydraulic fracture propagation and well production in wells that are able to stay in the thin lower Union Springs and those that skirt both the lower and upper part of the Union Springs. Production decline should also be considered and compared. One could also attempt to measure reservoir access by measuring hot shale access vs. limestone or less organic shale access and try to quantify and cross-compare the results. One possible method is Relative Optimized Reservoir Access – given elsewhere on this website.

Another geological situation often invoked when there is are unexpected discrepancies in correlation are facies changes. Kohl, etal. define the fairway for each 3^rd order stratigraphic sequence on the basis of mean gamma ray for that interval. It is unclear whether the gamma ray logs were corrected or normalized/equalized against gamma of a known stable position such as the Onondaga top or simply taken as recorded. These maps of mean gamma ray give a facies fairway for each sequence. The lowest Union Springs organic-rich TST sequence would be clastic rich further east (source-ward) and thinner westward as it onlapped the carbonate bank. It would also likely have more very thin interbeds of limestone and possibly as a result be more brittle overall. Sometimes very thin (~ 6 inch thick or less) limestone “stringers” can be seen in this interval these western areas when geosteering. They appear to be discontinuous and often not found on type logs. Kohl etal note that limestones were often deposited on bathymetric highs but it is unclear if these are associated with sea level drop or pelagic material, possibly sourced from the north and west.

Recent work in the Permian Basin suggests that the carbonate ramps there had significant Sediment Gravity Flows (SGF) such as carbonate-rich aprons, fans, sheets, and linear flows confined to previous topography. Cores have shown routine basal scouring and mud clasts (rip-up clasts) indicative of such structures. Sharp basal contacts are likely seen on logs as well with gradational contacts on the upper boundaries being common in core. In some of the Permian Basin reservoirs it has been demonstrated that carbonate content (via these SGFs) is associated with lower porosity and permeability so that the SGFs should be avoided as much as possible. This is in contrast to other mudstone reservoirs with carbonate interbeds like the Point Pleasant Shale where carbonate content is associated with lower clay content and better fracability when interbedded with high-TOC shale. In the Marcellus, it may be that some of the limestone beds were deposited as sediment gravity flows in the same manner in areas near the shelf margins. Times of shallow water and possible minor subaerial erosion could have increased sedimentation rates. Basin bathymetry and storms likely also distributed carbonates in the Marcellus. As stated above Kohl, etal noted that basin bathymetry is not well established. Localized and regional mapping of Marcellus limestones, particularly the thin one just above the Onondaga could aid in determining depositional patterns. In the Permian Basin plays the margins of the SGFs appear to be the best places to drill. Since the brittleness of the Marcellus is based on quartz content there would be increased quartz and decreased carbonate content on the margins and in between such possible SGFs. Basically, there would be more reservoir available, however slightly more, where the limestone beds are thinner. Where they are thicker, there would possibly be thinner hot shales deposited on top of them.    

           

References: 

Taury Smith – AAPG Distinguished Lecturer – talk given at Appalachian Geological Society Meeting – April 2015 – Alternative Shallow Water Anoxic Model for Organic-Rich Source-Rock Mudstone Deposition

Bruce S. Hart, Joe Macquaker, and Kevin Taylor – Mudstone (“shale”) Depositional and Diagenetic Processes: Implications for Seismic Analyses of Source-Rock Reservoirs – in Interpretation: A Journal of Subsurface Characterization – A Joint Publication of Society of Exploration Geophysics (SEG) and AAPG – Vol. 1, No. 1, August 2013.

Daniel Kohl, Rudy Slingerland, Mike Arthur, Reed Bracht, and Terry Engelder - Sequence Stratigraphy and Depositional Environments of the Shamokin (Union Springs) Member, Marcellus Formation, and Associated Strata in the Middle Appalachian Basin in AAPG Bulletin Vol. 98, No. 3, (March 2014), p. 483-513

DRAFT – Terry Engelder and Gary Lash – Systematic Joints in Devonian Black Shale: A Target for Horizontal Drilling in the Appalachian Basin (2008?)

Elizabeth L. Rowan – Burial and Thermal History of the Central Appalachian Basin, Based on Three 2-D Models of Ohio, Pennsylvania, and West Virginia - USGS Open-File Report 2006-1019

David T. McConaughy and Terry Engelder – Joint Interaction with Embedded Concretions: Joint Loading Configurations Inferred from Propagation Paths, in Journal of Structural Geology 21 (1999) p. 1637-1652

Gary G. Lash and Terry Engelder – An Analysis of Horizontal Microcracking During Catagenesis: Example from the Catskill Delta Complex – in AAPG Bulletin, V. 89, No. 11 (November, 2005), pp. 1433 - 1449

Digging the Rocks of Avalon: New Ways of Evaluating Sweet Spots Take Hold - news story by Heather Saucier in AAPG Explorer, Vol. 36, No. 7,  July 2015, p.10-14

John E. Repetski, Robert T. Ryder, David J. Weary, Anita G. Harris, and Michael H. Trippi - Thermal Maturity Patterns (CAI and %Ro) in Upper Ordovician and Devonian Rocks of the Appalachian Basin: A Major Revision of USGS Map I-917-E Using New Subsurface Collections - Scientific Investigations Map 3006 - 2008