The Middle Magdalena Valley Basin, located in the central part of Colombia, has evolved through many stages, resulting in a highly complex structural and stratigraphic framework. Exploration in this region was initiated in 1955 and, until recently, the available data have never provided an accurate image of the reservoir for developing successful drilling plans for various reservoirs. After many years of oil production, some results of recent drilling campaigns have demonstrated that various reservoirs in the basin have not been accurately mapped, leading to sub-optimal exploitation. The main challenge of current planning is to understand the geometry and kinematics of the basin, provide new insight into the tectonic setting and gain a clearer picture of the migration and trapping of hydrocarbons in the various reservoirs. To accomplish this, geoscientists need to build an accurate geologic model to add information regarding the location of potential resources and help to perform a valid volumetric evaluation.

Setting
The Middle Magdalena Valley Basin is an intermountain basin located in northwestern Colombia between the Central and Eastern Cordilleras of the Andes Mountains. It is structurally bounded by the Palestina Fault to the west (dextral strike slip system) and the Bucaramanga Fault to the east (sinistral strike slip system). The basin is part of the Magdalena Valley, which includes the Upper Magdalena Basin to the south and Lower Magdalena Basin to the northwest of the Middle Magdalena Basin.

The basin is elongated. It is only about 80 km (50 miles) wide but extends to the north about 450 km (280 miles), where it terminates against the Santander Massif and Cesar Valley. To the south it terminates against the Upper Magdalena Basin, which consists of the Girardot and Nieva sub-basins where the Central Cordillera and Eastern Cordillera converge. Faulting in the Middle Magdalena Basin is primarily reverse and thrust faulting. Reverse faulting is high-angle in the west and low-angle in the eastern and central areas of the basin, with normal faults also developing along the eastern margin. These thrust faults formed from thrusting from the eastern margin of the Central Cordillera in the Eocene and the western margin of the Eastern Cordillera in the Miocene. Folds, which can be described as a series of asymmetric syncline against the hanging wall of the fault next to an inclined anticline, are key structures for hydrocarbon explorations. The structure from this case study is an asymmetric anticline limited to the east and west by major reverse faults.

Methodology
Previous studies had not succeeded in associating information from the 185 wells (22 stratigraphic units) and seismic interpretation data (reverse faulting and main horizons) in a unique 3-D geologic model. Based on the complexity of the structure and quality of the seismic data, only five horizons representing the main formations and 10 faults were interpreted. Conventional modeling solutions and technologies are mostly 2-D-based. Triangulated surfaces used to represent fault or horizons are difficult to bring in perfect contact with one another. Pillar grids represent horizons and faults together and do not suffer from the triangulated surface contact issues, but the fault network needs to be fully represented by a coherent set of pillars. This is not possible when fault contacts intersect one another or become horizontal.

Building a precise stratigraphy column from well markers, seismic interpretation and the deposition mode paired with a unique pillar-less full 3-D approach provided a true representation of the complexity of the reservoir. This was done through an accurate static model, which in turn provided a more reliable model for reservoir properties modeling and simulation workflows.

This pillar-less approach is based on the concept of a space/time mathematical framework introduced by Mallet (2004, 2014). In this approach, any subsurface is curvilinearly parameterized by a uvt-transform: it maps every point, defined by x, y, z in the geological space into the paleo-space by its u, v, t coordinates. The uvt-transform is computed so an iso-t surface corresponds to a stratigraphic horizon, and an iso-t is discontinuous across the faults. If the seismic reflections, and therefore associated well markers, are assumed to be consistent with the chronostratigraphy (time-stratigraphic correlation), then seismic interpretation (even incomplete) and well markers are considered as iso-t surfaces of the uvt-transform. The regular way of building the uvt-transform is to assign a relative geologic time to each interpreted event (seismic and defined only by well markers) through the definition of a stratigraphic column and to interpolate the values across the volume of interest.

The uvt-transform removed the usual bottlenecks of traditional pillar-based technologies by representing complex fault networks without having to simplify the original information. Without pillars, geoscientists are able to create a more accurate model of the subsurface (Figure 1).

The algorithm delivered a consistent representation without “dumbing down” the data so that there was no need to take into account all the available data (seismic interpretation, well markers) to ensure an accurate image of the subsurface. This process allowed 22 stratigraphic levels to be created from five interpreted horizons and well information. The final model was a full, perfectly sealed structural model. The next stage, associated with the structural and stratigraphic model (static), was to build the geologic grid, which will be used to propagate the petrophysical properties of the reservoir (Figure 2). Geological grids for geostatistical simulation of rock properties were computed directly from the uvtmodel without any additional user interaction. These grids can be used for velocity modeling or geological modeling.

In this project, the next challenge was to understand the internal distribution of the facies within the reservoir. Reservoir facies determination is one of the uncertainties in reservoir modeling studies. It will affect the reservoir properties distribution, and inappropriate determination of the facies distribution may give unrealistic reservoir behavior. Integrating the petrophysical data from the wells combined with a conceptual deposition model from sedimentology maps enabled the creation of a 3-D facies proportion cube. This provided the geological background for all the property models.

Three-dimensional reservoir models play an essential role in the assessment of hydrocarbon resources. They not only are used to estimate in-place volumes but are also the primary input to flow simulation for optimizing and forecasting how much resource can actually be recovered. Uncertainty in hydrocarbon volumes is typically assessed through multiple realizations of the reservoir model’s petrophysical content.

Using an accurate subsurface geologic model, it was possible to quantify and model the uncertainty related to each property and come up with a wide range of geologically feasible stochastic scenarios to define the impact of uncertainty when estimating the reserves, ranking the degree of influence of each property in the reservoir volume calculation through spider and tornado charts. This workflow enabled the quality control and validation of the static model by integrating the dynamic data and by comparing the production history with the model prediction before performing simulation.

The first result of this project was a validated, accurate structural and stratigraphic model containing 22 stratigraphic units that honored both the seismic interpretation and well markers. The geologic grid created from the final structural model honored both the geology and stratigraphy. A solid statistical analysis of the petrophysical well properties was delivered, as were property models honoring sedimentology and petrophysical data.

Comments