Petroleum Geoscience - Volume 22, Issue 1, 2016

Reservoir modelling studies are widespread and are often built into a formal gated process used for decision-making. Now ubiquitous, it is easy for the models to simply become tools for verification of a decision that has partially or wholly been made – ‘modelling for comfort’. This is particularly the case in mature fields, when the presence of an inherited model already anchors the view of the field, and the volume of production data discourages the practitioner from exploring uncertainties with multiple models. It is proposed that reservoir modelling offers most value when used to create some discomfort – a stress test for decision-making that can identify upsides and secure against loss. This requires an awareness of the biases at work in model design and a conscious choice to move away from the default of a single, detailed, full-field model. This ideally means moving away from base-case-led modelling altogether, and typically involves multi-scale model design and multiple models for uncertainty handling, based either on stochastic modelling or multi-deterministic, scenario-based approaches.

When assessing a reservoir’s Hydrocarbon Initially In Place (HIIP) volumes there are often many uncertain input factors. If they are discretized into low, mid and high case scenarios, this can result in tens of thousands of possible combinations of factors and it is often necessary to reduce the amount of uncertain factors to a manageable number. This can reduce the range of model outcomes, and the confidence in the predicted volumes will be overestimated. Experimental design allows a reduced representative set of models to be constructed for uncertainty analysis when it would take too long to construct all possible realizations. This paper uses a case study to demonstrate a refined application of experimental design, whereby uncertain factors were first identified and experts were consulted to specify high, mid and low values with associated probabilities. The analysis was completed in two parts: a screening phase, whereby non-significant factors were eliminated; and an optimization phase, whereby a limited number of model runs were used to train a proxy model for use in a Monte Carlo procedure to generate a probability distribution of HIIP. The key advantage of the method is the comprehensive treatment of uncertainty with a reduced amount of modelling effort.

Block 22, located offshore in the North Coast Marine Area (NCMA) of Trinidad and Tobago, contains the Cassra gas accumulation discovered by the Cassra-1 and Cassra-2 wells. Both wells encountered dry, biogenic gas in early Pliocene, shoreface–shelfal sandstone reservoirs.

A best-practice modelling methodology is documented to capture a range of subsurface uncertainties for use in reservoir simulation to generate production profiles in support of pre-development project planning. Calibration of a sequence stratigraphic interpretation from seismic with core data was instrumental in generating a depositional model, within which sedimentary facies could be stochastically distributed and used to constrain the population of petrophysical properties.

Static geological uncertainties were modelled using an uncertainty workflow methodology in commercial three-dimensional (3D) geomodelling software, resulting in multiple, static model realizations and probabilistic GIIP (Gas Initially In Place) distributions for Cassra and some nearby prospects. The multi-regional model required careful selection of realizations for reservoir simulation, based on a ranking scheme that combined GIIP, GRV (gross rock volume) and net:gross (net to gross ratio), and not simply P90–P50–P10 percentiles from the global distribution.

The reservoir simulation phase included a dynamic uncertainty workflow using commercial experimental design modelling software. By including dynamic uncertainties such as horizontal/vertical permeability and well productivity in the workflow, a more objectively defined suite of production profile predictions was achieved.

The Lower Cretaceous Goldeneye gas field lies in the Captain turbidite fairway of the Moray Firth and has recently ceased production. Its situation and dimensions have made it an excellent candidate for CO2 sequestration. The field was extensively modelled for the original development planning when uncertainty was assessed from the perspective of volumetrics and field behaviour. The subsequent need to assess its suitability as a CO2 store has given the opportunity for a look back at an uncertainty analysis with the benefit of full-field performance, and to perform a new analysis aimed at different issues concerning behaviour during CO2 injection.

Both sets of analyses required coupled static–dynamic modelling runs in which the key static parameter ranges of the field were varied, including depth conversion, internal geometries and aquifer properties. For the field development work the parameter ranges were explored to assess in-place volumes and field behaviour under natural aquifer and depletion drive; for the CO2 uncertainty work, however, parameter ranges were explored to demonstrate storage capacity and CO2 containment. The look-back showed that the field volumes indicated by production data landed in the upper part of the original uncertainty range and that there was definitive spare capacity in the field relative to the planned injection volume.

A fully integrated reservoir modelling approach aiming for the best conditioned static model for an underground gas storage facility (UGSF) in a complex structural and depositional setting is presented.

The Nussdorf UGSF is a depleted gas field characterized by typical deep-water depositional environment settings including sediment mass-flow systems being shed off the emerging Alpine thrust front during the Neogene. The key challenge in assessing this specific storage performance is the communication within the individual stacked sandstone layers, as well as determining the existing cross-flow between such layers through wells and due to juxtaposition across faults.

Highly heterogeneous reservoir facies, representing thin-layered, stacked sandy fans embedded in marly shale, were realized by joining object-based and Gaussian simulations constrained by a gross depositional environment model. Modelling known pressure communication across intrareservoir faults required fault throws to be adjusted at scales below the limit of seismic resolution. Scoping simulation runs on a best-guess model led to a full back-loop of the geological modelling. Several loops revealed that iterations limited to property realizations were insufficient, requiring additional modifications of the structural model. Only via this expensive approach could a geologically consistent and ‘fit for purpose’ reservoir model for the UGSF be achieved.

The bathymetric datum with respect to global sea level for Aptian salt deposition in the South Atlantic is hotly debated. Some models propose that the salt was deposited in an isolated ocean basin in which local sea level was between 2 and 3 km below the global level. In this study, we use reverse post-break-up subsidence modelling to determine the palaeo-bathymetry of base Aptian salt deposition on the Angolan rifted continental margin. The reverse post-break-up subsidence modelling consists of the sequential flexural isostatic back-stripping of the post-break-up sedimentary sequences, decompaction of remaining sedimentary units and reverse modelling of post-break-up lithosphere thermal subsidence. The reverse modelling of post-break-up lithosphere thermal subsidence is carried out in 2D and requires knowledge of the continental lithosphere stretching factor (β), which is determined from gravity anomaly inversion. The analysis has been applied to the ION-GXT CS1-2400 deep long-offset seismic reflection profile, and two seismic cross-sections (P3 and P7+11) from offshore northern Angola. Reverse post-break-up subsidence modelling restores the proximal autochthonous base salt to between 0.2 and 0.6 km below global sea level at the time of break-up. In contrast, the predicted water-loaded bathymetries of the more distal base salt restored to break-up time are much greater between 2 and 3 km. The predicted bathymetries of the first unequivocal oceanic crust at break-up are approximately 2.5 km, as expected for newly formed oceanic crust of ‘normal’ thickness. Several interpretations of these results are possible. Our preferred interpretation is that all Aptian salt on the northern Angola rifted continental margin was deposited between 0.2 and 0.6 km beneath global sea level, and that the proximal salt subsided by post-rift (post-tectonic) thermal subsidence alone; while the distal salt formed during late syn-rift, when the underlying crust was actively thinning, resulting in additional tectonic subsidence (followed by post-rift thermal subsidence). An alternative interpretation is that the distal salt is para-autochthonous and moved downslope into much deeper water during and just after break-up. We do not believe that a deep isolated ocean basin, with a local sea level 2–3 km beneath that of the global sea level, as has been proposed, is required to explain the Aptian salt deposition on the northern Angolan rifted continental margin.

Albian carbonate reservoirs are prominent in the subsurface of the South Atlantic. Equivalent exposures in southern Angola (Benguela and Namibe basins) have received relatively little detailed sedimentological work. In the Namibe Basin, carbonates form metre-thick beds interbedded with shallow-marine and continental alluvial fan siliciclastics. Characteristic carbonate mounds (≤5 m high, 1–2 m in diameter) rise above a basal carbonate bed, which consists of oncoid–peloidal rud-grainstones with oysters and echinoderms. Thrombolite mound microfacies include red algae and microbial–algal crusts. The microfacies are marine, and compare with documented occurrences of algal–microbial–oncoidal textures in Albian carbonates of the Congo and Angola. Burial processes dominated diagenesis and have reset carbonate geochemistry from marine values, with the probable exception of Mg concentrations. Up to 22% of primary (intergranular) and secondary (microporosity, mouldic, vuggy, fracture) porosity developed as a consequence of important dissolution and partial cementation.

Two depositional models for the localized mound occurrence are discussed: (1) marine ingression into a coastal embayment and the formation of shallow-water microbial bioherms; and (2) a submarine groundwater spring discharging in coastal areas downdip from alluvial siliciclastics. Marine fauna, similarity with marine Albian strata elsewhere and a partly preserved marine Mg geochemical signature favour a marine ingression. Environmental conditions were likely to have been stressed on account of the siliciclastic input, variable salinity and elevated nutrients, all of which are consistent with the observed microfacies. A submarine spring is conceptually feasible, but is considered to be less likely owing to the absence of a clear meteoric signature and the low likelihood of bicarbonate-rich groundwater in the region.

Using the discussion of depositional models for the studied outcrop, and incorporating a literature review, the study proposes a set of criteria to distinguish various marine and non-marine carbonate mounds in the subsurface. The most diagnostic criteria are: (1) marine or continental fauna; (2) sediment geochemistry, in particular Mg, Sr, and δ13C and δ18O isotopes where preserved through diagenesis; and (3) carbonate fabrics, such as crystalline shrubs, that are diagnostic of thermogenic continental mounds. The scale of geobodies and the mineralogy of mounds can sometimes be used as additional criteria. This set of criteria can help exploration and production geologists who need to devise exploration and development strategies in unconventional carbonate reservoirs of the South Atlantic and other rift basins.

Permeability in fractured carbonate reservoirs is very heterogeneous due to fracturing at different scales superimposed on inherent textures from deposition and diagenesis. Observations of fractures in core and outcrop indicate that flow in open fractures in carbonate rock tends to be channelled rather than through fissures. Most of the flow takes place along a few dominating channels in the fracture plane, whereas most of the fracture plane is not effective for fluid flow. The formation of flow channels is caused by a combination of mechanical and, in particular, diagenetic processes. Single extension fractures occur as partly open or vuggy fractures, and their hydraulic properties are controlled by dissolution and cementation. Single shear fractures are typically open at local steps in the fault plane controlled by shearing along irregular fracture surfaces. Fault damage zones tend to be concentrated at fault tips, intersections, pull-aparts and overlap zones that represent areas of dilation. These damage zones represent elongated features in three dimensions with a high fracture density that will result in channelled flow at reservoir scales. The effect of channelled flow should be taken into account during evaluation of fractured carbonate reservoirs and when building dynamic flow models.