Reservoir simulation

Modeling of a Complex Reservoir Where the Normal Modeling Rules Do Not Apply

With the easy conventional oil in Argentina having been produced, one remaining way to find new oil in existing fields is to convert fields from primary or secondary production to secondary or tertiary production, respectively.

Image of a complex maze
Getty Images

With the easy conventional oil in Argentina having been produced, one remaining way to find new oil in existing fields is to convert fields from primary or secondary production to secondary or tertiary production, respectively. For the Cerro Fortunoso field, the high costs required to develop the field, combined with an ambiguous log response that suggested that sand was predominantly disconnected, resulted in secondary production not being implemented. To reduce the risk, a waterflood pilot was necessary to demonstrate that waterflooding has potential and to provide water-injection and production data to constrain the history-matching process.


Fig. 1 shows the location of Cerro Fortunoso in Mendoza province, approximately 1200 km southeast of Buenos Aires. Most large fields in Mendoza province were exploited by waterflood decades ago. For waterflooding not to be initiated, the nature of the reservoir had to be unclear or the capital expenditure required had to be high. In the case of Cerro Fortunoso, a substantial investment is required to cover the cost of drilling infill wells and for the construction of a pipeline to provide injection water from a river 15 km away.

Fig. 1—(a) The location of the Cerro Fortunoso field in Argentina, with productive and nonproductive basins. (b) Satellite image of its location.

To minimize the risk associated with converting the field to waterflooding, an integrated study was sanctioned that aimed to characterize the field initially and then to determine if waterflooding would, in fact, be economically viable.

Structural-Model Construction

Usually, the first step in the construction of a structural model is to extract a surface from a seismic marker. Unfortunately, in Cerro Fortunoso, the presence of a significant thickness of igneous rock and high-angle flanks causes high noise and attenuation, respectively, on existing 2D-seismic data and makes completing a 3D survey a fruitless exercise. Without 3D seismic, the structural model was constructed on the basis of data from both surface outcrops and subsurface well data. This was a challenge because Cerro Fortunoso is in a complex structural environment and the fold geometry was shown to vary significantly along the strike of the field.

The complex structural model was captured successfully in the static model. Specific challenges successfully met included:

  • Preserving the zone thickness within correlated cycles
  • Modeling of complex thrust faults and stratigraphic repetitions

The Simulation Model

Cerro Fortunoso posed a unique challenge for simulation because of the large number of cells needed in the vertical dimension to model the 1- to 2-m-thick sands that make up the productive intervals of the field. All 17 correlated zones were shown to contain potentially productive thin sand intervals, so it was not possible to increase the cell thickness in any of the productive zones. In addition, to history match the waterflood pilot, it was necessary to ensure there were at least two cells between each injector/producer pair. Because the well spacing in the field is 150 m in some cases, this meant that the horizontal dimension of the cells was restricted to an average of 30 m. A full-field static model, therefore, would be either very slow or impossible to simulate. Consequently, it was necessary to consider simulating sectors of the field only. Structural interpretation and a variation in deepest oil suggest that Cerro Fortunoso is separated into individual blocks that could be simulated separately. It was decided initially to simulate only the northeast block, which included the waterflood pilot.

Gas Cap. The reservoir has a 95% carbon dioxide gas cap from volcanic activity. However, it was decided to model the gas cap as a conventional hydrocarbon gas cap.

Simulation Time Optimization. To improve the running speed of the model, it was necessary to run the model constrained by reservoir volume rate initially. If run on the more typical liquid rate, wells unable to make their liquid-rate constraint reduced their flowing bottomhole pressure to the minimum allowable pressure of 1 bar. If these wells were located near the gas cap, this resulted in the subsequent production of large quantities of gas and, hence, simulation instability. By limiting the amount of gas produced by use of the reservoir-fluid history-matching constraint, the volume of gas produced in the simulation is restricted.

Relative Permeability. A number of experimental relative permeability curves were available. Some of these were regarded as being of doubtful quality, with endpoints inconsistent with the expected range for this type of oil viscosity and reservoir permeability. Directly using relative permeability based on experimental data, even the relative permeabilities that are regarded as being more reliable, resulted in high levels of simulated water production early in field life while, in fact, water production in the field was initially negligible. Applying the J-function saturation, with its initial water saturations that were lower than the resistivity indicated, still resulted in an excessive simulated water production. To solve this, the critical water saturation was increased so that water does not flow at the water saturations modeled initially by the averaged J-function in the dry oil sands. (J-function refers to the classical expression relating water saturation and capillary pressure.)

Saturation Modeling. The productive features on Cerro Fortunoso are thin sands dispersed stratigraphically between massive dense beds. The effect of this was that, even in the center of a reservoir bed, well logs would be reporting partly the log response from either the overlying or the underlying dense formations. For both porosity and resistivity, this meant that, if the log response was interpreted directly, the properties would be underestimated. For porosity, this would affect the material balance. However, the greatest effect on the history match came from the saturation modeling. When log-derived saturations were used in the model, a high level of water production was predicted even though production history had shown negligible water production away from the aquifer. This was interpreted as being caused by the saturations in the productive sands being estimated as lying in the transition zone of the relative permeability curve, resulting in simultaneous production of oil and water. In reality, the saturation in the reservoir rock was at initial water saturation, which obviously results in the production of dry oil.

J-function values were calculated for the  controlled-relative-permeability water-saturation points and were entered into the simulation oil/water relative permeability table. The surface tension for Cerro Fortunoso between the oil and water systems is 25 dynes/cm.

With these data and the permeability and porosity data obtained from the static model, the saturation in the field was populated.

Pressure Data. Because most of the field has been developed exclusively by primary depletion, there were limited water-breakthrough data to history match, and these were available only after the initiation of the waterflood pilot. However, there were significant repeat-formation-tester (RFT) data available. Because the field is in an area where the surface is significantly above sea level, the RFT data were critical in determining the equilibration conditions.

The first wells in each block showed equilibrium and constant gradients. Subsequent wells, as could be expected, showed varying levels of pressure depletion. This demonstrated that there was pressure communication between wells. In addition, pressure data demonstrated depletion of varying levels from all individual reservoir zones. This meant that no zones could be discarded for history matching, thus slowing run times.

Fig. 2—Simulation of water saturations in the northeast block in 2038 with the existing waterflood (left) and an expanded waterflood (right). 


As a result of the integrated modeling project, a development plan for expanding the existing waterflood has been proposed for the whole field. However, to date, forecasts have been limited to the northeast sector. Fig. 2 shows the simulated water saturation in the MK-150 zone in 2038. The left-hand side shows water saturation after the continuation of the existing pilot waterflood, and the right-hand side shows water saturation after an expansion of waterflooding to cover the entire block. The reduction in oil saturation with the expansion of the waterflood is especially apparent in the southern region of the northeast block.

This article, written by Special Publications Editor Adam Wilson, contains highlights of paper SPE 169369, “Integrated Modeling of a Highly Structural and Complex Reservoir Where the Normal Modeling Rules Do Not Apply—Cerro Fortunoso,” byAnthony Thompson, SPE, Abel Garriz, Gaston Manestar, SPE, Griselda Vocaturo, Vanesa Consoli, and Pablo Giampaoli, YPF, prepared for the 2014 SPE Latin American and Caribbean Petroleum Engineering Conference, Maracaibo, Venezuela, 21–23 May. The paper has not been peer reviewed.