Thomas Huxley (1825-1895), in an 1870 address to the British Association, famously said, “The great tragedy of science is the slaying of a beautiful hypothesis by an ugly fact.” One of the key roles of a geologist is to sort through well and rock information, and to test old-ideas with new-data or to find new-ways to think about old-data. It is no secret that most of the world’s new resource plays and giant conventional discoveries are often not made by the major oil companies, but smaller companies that have learned to think ‘out of the box’ with both old and new data.
But are we losing the ability to teach the next generation of geoscientists how to use existing data to get new ideas? Are we allowing 3D computer or seismic images to trump hard work analyzing well data?
Understanding how to interpret oil and gas shows in the context of migration and entrapment using well data is hard work. Well reports have to be read, logs and log analysis investigated, and mud logs, head space gas readings, biostratigraphical and petrography analysis must all be integrated to be fully understood. None of this is glamorous work, but it has to be done. Often, samples and cores have to be re-examined, inevitably located in dingy core repositories or warehouses. It is also imperative to capture shows, source rock, and other relevant data in ways that can be visualized on seismic, log cross-sections or in migration modeling software. For those committed to digging hard into details of key wells, new insights and plays often emerge.
Consider this case: Company A had a multi-disciplinary team screening its acreage to decide what to retain. A block acquired in a company merger has a dry hole on it. Seismic shows the dry hole is a small 4-way closure. Maturation maps show it is over 50 km away from mature source rock. Concluding that no oil had migrated into this area, and with no other visible structural traps, the acreage was dropped. Company B was thrilled, because the ‘dry hole’ actually had 3.5m of pay on logs in a porous Jurassic sandstone. They recognized a seismically defined up-dip stratigraphic trap - and subsequently drilled the 1.4 Bbo-in-place Buzzard Field discovery (Carstens, 2005; Dolson, 2016; Ray et al., 2010). The more careful analysis of the ‘dry well’ invalidated the pessimistic migration model and also de-risked reservoir issues.
Sound unusual? It isn’t. A major complaint I hear from all oil companies is that far too many young geoscientists are drawn immediately to 3D seismic or computer simulations without a firm understanding of how to use well and geochemical data to test models. That knowledge gap presents a world of opportunity for those who do, but a world of hurt for those who don’t.
Two of the ground-breaking papers on understanding oil and gas shows were those of Schowalter (1979) and Schowalter and Hess (1982). Few younger geoscientists are even aware of these papers or understand how to distinguish a show along a migration pathway from live oil within an accumulation, or how to use water saturation (Sw) to determine position in a trap. Vavra et al. (1992) and Hartmann and Beaumont (1999) are two additional papers that should be required reading for any geoscientist looking to understand how to interpret test data and shows in the context of capillarity and position in a column.
Consider Figure 1 (below), a carbonate shoreline trap with degrading reservoir quality due to pore throat reduction up-dip into waste zones. How to interpret the test results depends not just on the sequence in which the wells are drilled, but the ability to integrate the petrophysical data. If well 3 is drilled first, it may be declared a dry hole with high Sw that looks wet in porous but low permeability rock. Drill well 1 first, and optimism abounds, but results in disappointment on the up-dip offset. Well 2 is more problematic. The key to recognizing a potentially large field here is the pressure and rock data, which shows the different facies as part of the same column. The fact that oil is actually recovered in wells 2 and 3 shows a trap with a column. If knowledge exists of the pore-throat sizes or capillarity, an astute geoscientist might be able to quantify the elevation above the free water level and speculate on where better facies might exist that would produce water-free oil, potentially even down-dip of wells 2 and 3. Recommending that downdip location takes considerable knowledge of rock properties, relative permeability, facies distributions and how test and show data help define the size of the trap.
Learning to understand test and show data to seek out those elusive ‘NULFs’ requires re-thinking porosity and permeability in a way that quantifies pore throat distribution. In Figure 1, the porosities may be similar in wells 1-3, but with radically different pore throat sizes and saturations at any point in the trap. Pittman (1992) and Winland (1972) determined a way to calculate pore throat sizes statistically by adding calculated pore throat radii lines to a simple porosity and permeability plot. These lines are constructed from equations.
Figure 2 provides an example of a Winland plot from Pennsylvanian carbonates of the Four Corners area in Colorado.
Note that the best porosity is a meso-porous limestone which will act as a baffle or seal, depending on the column height in the trap. A 15% porosity cutoff, in the absence of understanding rocks, can easily identify this facies as the best, when, in fact, an 8 % porosity algal boundstone should actually be the target for exploration.
For development purposes, flow units (Ebanks et al., 1992 and Gunter et al., 1997) will parallel the common pore unit lines, also helping to explain performance while testing.
What is not common knowledge to many geoscientists, and is illustrated in Figure 3 (below), is the control that pore throat radius exerts on water saturation. A 10% mega-porosity rock with 10 md perm will test much more oil at an equivalent position in a trap than a 23% meso- porosity rock with 10 md perm.
These graphs derived pore-throat calculations from porosity and permeability data alone (Pittman (1992) or Hawkins et al., 1993) - using reasonable assumptions as to subsurface salinities, API gravities, interfacial tension (IFT) and wettability - and can be very insightful. Figure 3 provides an example, where a supratidal limestone has 50% Sw testing 1-3 bopd and some water, indicating a column and a trap. Using the curves in Figure 3, a reasonable assumption is that the test is about 50-70m above free water, well within a trap. Offset wells may actually be drilled up-dip into the seal, when the best location might be downdip or on strike into the algal mound facies.
Without these skills as part of their fundamental ‘tool kit’, young geoscientists are doomed to miss a lot of plays, inappropriately plug a lot of wells and leave oil behind pipe for others to find. Most young geoscientists I work with have only a poor or no working knowledge of capillary pressure basics and how to calculate height above free water (base of the trap where buoyancy pressure is zero). Worse, they confuse oil or gas-water contacts with free water levels, not understanding that micro-porous rocks may have 100% Sw well above spill point of the trap.
Understanding pressures is another key skill. Integration of pressure data is critical to understanding connectivity and position in a trap. Figure 4 illustrates seal recognition from pressure data and how to use the slopes of the curves to calculate fluid density. However, in the absence of other data the pattern shown in the image could also be interpreted as the result of hydrodynamic flow (Figure 5).
In 37 years of consulting, I have seldom seen geoscientists make potentiometric surface maps to quantify the impact of hydrodynamic flow in tilting oil and gas contacts, despite the landmark work of Hubbert (1953), Dahlberg (1982, 1995) and England (1994). Many geoscientists do not recognize the presence of tilted oil and gas columns caused by an upward flow of expelled waters from over-pressured areas towards the basin flanks. Numerous recent examples of tilted contacts in deep basin settings show the potential to underestimate trap size due to tilting (Ferrero et al., 2012; Muggeridge and Mahmode, 2012; O'Connor and Swarbrick, 2008; Riley, 2009; Robertson et al., 2013).
A case history using the Temsah Field in the offshore Nile Delta (Dolson, 2016), is shown in Figure 6 (below). The Discover well, Temsah-1, was drilled in 1977, and tested some gas with high water rates at the crest of the structural closure. It took over 20 years to recognize the tilted nature of the gas water contact, which dips to the north-east as a result of deep basin water flow from the south-west. Trinity software (He and Berkman, 1999) provide examples and tutorials of how to construct potentiometric or pressure differential maps to test migration and entrapment with hydrodynamic flow.
FIS and Migration Modeling
Many advances have been made in capturing new information in old wells, particularly Fluid Inclusion Stratigraphy (FIS) (Dolson, 2016 and Hall, 2008). Mud log, mud gas and cuttings data provide the first step in detecting hydrocarbons but in some cases shows are suppressed and hydrocarbons missed. Fluid inclusion data often picks up subtle shows, or key information like temperature of emplacement of fluids, API gravity, salinities, proximity to pays, migration pathways, seals and even biomarkers useful for source to oil correlation. Classical single sample thin-section studies have been supplemented by the FIS technique, which can be relatively inexpensive to run and be done on cuttings from very old wells where mud log data alone has long since been lost.
The example shown in Figure 7 is from a well in the Barmer Basin where, given a total lack of shows on the mud-log, this dry hole was attributed to lack of charge. FIS data, however, showed adequate charge and migration, but strong evidence of seepage at the top of the well, indicating top and lateral seal failure. These data provide new insight plays and prospects.
Modern petroleum systems software performs very powerful 4D and 3D migration and maturation modeling - but how rigorously are the models tested against well data? The models often are beautiful to behold, like a stunning seismic 3D section, but are they right? Can your staff pick apart older wells and use the information from them and fields to calibrate the models? Or do they stop once the model is done and proclaim the model to be the answer?
Figure 8 illustrates one result of a 3D migration simulation in the Barmer Basin of India (Naidu et al., in press), where shows data are captured in a spreadsheet and then visualized in the model. Two levels of migration with varying fault and seal capacity are shown with the oil shows and types.
In this example, over 20 different scenarios of seals have been used to test various models and this figure represents only one of those solutions. While none of the models achieved a 100% predictive level of the accumulations and dry holes, running multiple models helped clarify risk on charge for hydrocarbons migrated vertically to the ultimate seal at the Eocene Akli coal (AK).
Another useful technique is to quickly and qualitatively convert seismic depth images into reservoir-seal pairs (in meters of seal capacity) based on amplitude variations. Figure 9 models migration off the west coast of Africa: a simple image of the seismic amplitudes has been qualitatively converted to seal values in meters of column height and then used as a grid in a migration model. While only a crude representation of the real subsurface data, these solutions which can be developed quickly, give a good feel for how migration might work.
While much more sophisticated models can be run using more quantitative rock property modeling, pressures, hydro-dynamics, and other data, they are time-consuming and expensive to build, particularly if in a 3D seismic volume. Worse, all models involve multiple assumptions on fluid phase, seal capacity, fault leakage, pressures, etc., adding complexity to the model but not necessarily insight. A good discussion of the main benefits and pitfalls to petroleum systems migration modeling is that of He (2016).
The Young Geologist Skill Set
In essence, geoscientists need to learn to ‘think like a hydrocarbon molecule’. They need to be able to describe oil shows in a trap and how the oil got from the source to the reservoirs. Just some of the skill sets needed to enable young geoscientists to do this are summarized below:
- Understand the rocks: go look at cores and cuttings.
- Think of rocks in terms of pore throats, capillarity, Sw and position in a trap. What does the saturation mean in terms of column height, fluid phase and trap positon relative to the free water level?
- Understand pitfalls in log analysis or formation damage due to unusual minerals and/or shaliness.
- Build and refine oil and gas show databases that can be mapped and visualized quantitatively. Learn how to determine if a show is in a trap, a source rock or along a migration pathway or in residual shows of a breached paleo-accumulation.
- Question whether a ‘dry’ hole is actually dry, and challenge the prevalent explanation for it.
- Simulate migration scenarios, first conceptually, and then with appropriate software.
- Build appropriate seal and pressure maps and integrate them into migration models.
- Supplement mud log shows with fluid inclusion, thin sections or FIS data to help ground-truth migration and charge models, track migration pathways, visualize seals, and identify possible proximity to pay or by-passed pay
- Systematically use geochemical analysis of source rocks and reservoired hydrocarbons to try to understand migration pathways and source to oil correlations.
There are many more fundamental skills required of geologists that supplement 3D seismic or petroleum systems models with computers. Today’s geoscientist needs to be well-versed in petroleum geochemistry and petrophysics, and have the desire to dig through cuttings, core, well reports and subtle shows to find those ‘NULF’s’ that lead to recognition of new potential or new plays.
Most importantly, young geoscientists should be taught to become skeptics who search for data that does not fit existing paradigms. They need to be rewarded for the hard work of integrating data properly to test their models and play concepts. They must go back to basics, challenge conventional wisdom and continuously scour for data that yields the ‘NULFs’ that lead to new plays.