Means Field Residual Oil Zone Study – Extending the Life of a 75 Year Old Field
Richard J. Wachtman¹, Prabodh Pathak², Dale E. Fitz³, Jon Meissner¹, and Lizbeth Guijarro¹
¹ExxonMobil Production Company, Houston, Texas, USA
²XTO Energy, Fort Worth, Texas, USA
³ExxonMobil Exploration Company, Houston, Texas, USA
ExxonMobil recently conducted a thorough study of the residual oil zone (ROZ) of the Means field in Texas. In addition to analysis of pre-existing data, the study included in situ measurement of oil saturation, building a full-field geologic model and reservoir simulation models to predict flow streams from carbon dioxide (CO2) enhanced oil recovery, and design of a surveillance program for monitoring ROZ injection and production. The first phase of ROZ development is under way.
The Means oil field is located in West Texas on the northeastern edge of the Central Basin platform. Production to date has come from the top 200 to 300 ft of the Leonardian/Guadalupian (Middle Permian) age San Andres formation, a massive dolomite up to 1,400 ft thick that contains small amounts of sandstone, anhydrite, and shale. Means was discovered in 1934 and operated by primary production until 1963 when the field was unitized and a waterflood began. In 1983, after pilot testing, Exxon started a successful water-alternating-CO2 (WAG) flood, which included drilling infill wells that reduced well spacing to 10 acres. To date, approximately 261 million barrels of oil (MBO) has been produced.
In 2009 and 2010, ExxonMobil studied development of the ROZ, which is below the main pay zone (MPZ) and extends from the MPZ’s producing oil-water contact at 1,400 ft below sea level down to the original free-water level at 1,789 ft. The ROZ has had natural water influx over geologic time resulting in the oil saturation’s being reduced to values that are residual to water. Reservoir properties of the ROZ, with the exception of oil saturation, are similar to those of the MPZ—average porosity 9% and average permeability 20 mD. The ROZ has the potential to contain a very large resource, however, any estimate of the original oil in place is weakly constrained given uncertainty in oil saturation and limited well data in the ROZ. In addition, the oil cannot be produced by primary methods or waterflooding. The purpose of the study, therefore, was to reduce uncertainty associated with oil saturation and to determine whether it would be profitable to extend the WAG flood from the MPZ into the ROZ.
One key goal of the study was to understand the oil saturation in the ROZ. Before the study, ROZ oil saturation was uncertain due to the scarcity of data. Few wells were drilled down into the ROZ and what little data was available varied considerably and was difficult to explain. Residual oil saturation to water measured on MPZ cores varied from 11 to 54% and averaged 31%. Sponge core from the entire ROZ section from a well drilled in 2001 had oil saturations that varied from less than 5% to greater than 60%; no convincing explanation or correlation could be found for that data other than a trend of decreasing oil saturation with depth. For comparison, oil saturations of 30 to 38% are reported at analogue ROZ fields near Means—the Wasson Denver Unit and the Seminole Field. To reduce uncertainty in ROZ oil saturation, the study team selected three wells to test oil saturation using Single Well Chemical Tracer (SWT) testing and Log-Inject-Log (LIL) testing. The wells were chosen to evaluate geographical variations and multiple zones were tested in each well to understand vertical variations in saturation. Test results were then compared to core saturations collected over corresponding intervals.
The tests confirmed that oil saturation in the ROZ is variable, both geographically and vertically, and varies from 10% to 35%. An additional learning was that although SWT tests can be readily performed with equipment available in the field (inject a tracer solution into the reservoir, allow the solution to react in the formation for two days, and then produce the solution back to measure the tracer concentration), the tests are difficult to interpret in carbonates given reservoir-heterogeneity and material-balance challenges. Interpretation required a multi-layer model with unequal injection and production ratios for each layer. The LIL tests have more challenging equipment requirements but yield a more reliable interpretation. It is also worth noting that it is difficult to reconcile oil saturation measured at different scales such as core, layer, and interval.
To assess the recovery efficiency and develop an understanding of the potential flowstreams, the team constructed a full field geologic model which was used to prepare two types of reservoir simulation models. The geologic model was constructed using data from an extensive re-evaluation of core data, in which 45 cores from throughout the field were described, correlated, and characterized into 22 different reservoir facies. These were later grouped into 8 reservoir rock types (RRT) based on similarities in lithology and reservoir quality parameters. That data was used to construct a stratigraphic framework, map environments of deposition (EOD), and then populate the model with RRT’s using variograms and an understanding of the percent occurrence of each RRT within the EOD’s. Porosity was populated using the porosity distribution for each RRT measured from core analysis and permeability was modeled using porosity-permeability transforms.
An oil saturation model based on drainage and imbibition capillary pressure was developed to account for the ROZ’s highly variable oil saturation and history of water influx. The first step was to calculate oil saturation for each cell using drainage capillary pressure relations taken from special core analysis and the original free-water level at 1,789 ft below sea level. That step represents the initial phase of oil’s filling the structure. The second step was to modify each grid-block’s saturation using an imbibition capillary pressure relation and producing water-oil contact at 1,400 ft to represent water influx over geologic time. Imbibition measurements were not available on ROZ core samples, as such the imbibition curve’s shape and endpoint residual oil saturation were adjusted to calibrate the model saturation distribution to that observed in core data.
The purpose of the simulation models was to predict and optimize recovery efficiency and flow streams for a WAG flood in the ROZ. The first type of model was a compositional reservoir simulation model of one 40-acre pattern. Simulation type curves relating oil recovery factor to percent pore volume of injection for various pattern locations, well configurations, and pattern sizes were analyzed and compared to performance data from the Means MPZ as well as from ROZ developments at Wasson Denver Unit and Seminole Field. As expected, one of the key factors influencing the simulation results was permeability heterogeneity.
Initial attempts to model permeability used up-scaled porosity-permeability transforms. Based on these transforms, the element simulation models predicted uniform sweep of the oil by the injection fluid. Years of production and injection data suggest that uniform sweep had not occurred in the reservoir. The permeability property was re-constructed using the entire permeability dataset in order to capture the full range in heterogeneity. This approach resulted in a simulation that more closely matched production data. However, even with this adjustment the simulation results continued to predict optimistic recovery relative to the MPZ and analog type curves. A subsequent study found that to match production history in a MPZ sector model, additional features needed to be included in the model: 1) enhanced injector well-reservoir connections to reflect the presence of fractures (calibrated based on injector data), 2) a high permeability leached zone occurring at the boundary between the MPZ and ROZ (mapped based on core and injection profile logs), 3) enhanced connectivity between some producers and injectors (calibrated based on producer rates), 4) a higher ratio of vertical to horizontal permeability.
Given the uncertainty associated with the element simulation model, field-wide flow streams of a WAG flood in the ROZ were calculated using an alternate model. This model consisted of a spreadsheet that aggregates flow streams for individual patterns into a full-field flow stream. Pattern flow streams are based on type curves from analogue ROZ fields and take into account the oil in place and permeability-thickness product of the pattern. The spreadsheet includes constraints for maximum drilling rate, CO2 purchase rate, gas production rate, and water production and injection rates and allows for testing of various WAG operation strategies. Pattern development order is prioritized mainly by a pattern’s oil in place. Based on this analysis, development of the Means ROZ could yield large recoverable volumes and extend field life by 20 years. The development is expected to be less efficient with lower incremental oil recoveries per well and higher gas and water handling requirements as compared to the original development in the MPZ.
To manage remaining project uncertainty and risk, ExxonMobil elected to pursue a phased development. The first phase began in early 2011 with the drilling of 17 new wells and deepening of nine existing wells to create nine 40-acre 5-spot patterns. To understand and monitor flood performance, the study team designed a comprehensive surveillance program. Core was collected from four wells over the entire ROZ to understand the variation in reservoir quality over the small development area. Pressure interference and pressure transient testing was done to understand the connectivity between injectors and producers and to estimate permeability thickness. Analysis on this data along with early production results is on-going.
AAPG Search and Discovery Article #120034©2012 AAPG Hedberg Conference Fundamental Controls on Flow in Carbonates, Saint-Cyr Sur Mer, Provence, France, July 8-13, 2012