New reservoir characterization technology benefits from a reality check.

Tien-When Lo, who was with Reservoir Characterization Research and Consulting Inc. (RC2) until its acquisition by Veritas, is vice president, integrated reservoir studies at VES.
Advanced reservoir characterization technology is breathing new life - and profits - into mature and marginal fields. By providing insights about what's really going on inside a reservoir, this technology leads to better, more cost-effective decisions about multimillion-dollar drilling investments and solutions to reservoir management problems.
For Texaco's Bryant G, a 30-year-old retrograde condensate field in Midland County, Texas, it meant a 2,500% increase in production. A horizontal drilling program designed and executed during a 3-year period using this technology added significant new reserves to the field.
Referred to as geostatistical reservoir characterization at Veritas DGC, the technology produces more accurate reservoir models and performance forecasts than are possible through traditional characterization techniques. Geostatistical reservoir characterization is a totally integrated process that incorporates expertise and data from all geoscience disciplines - geology, petrophysics, geophysics and reservoir engineering. Instead of disregarding or overriding the inherent conflicts between "soft" seismic data and "hard" well control data, geostatistical reservoir characterization "listens" to all the data, extracting everything they have to say about each other and the reservoir.
Geostatistics provides the common language that facilitates integration of seismic, well log, core and production data. Specialized tools and techniques bring these soft and hard data together through a systematic, repeatable procedure of interpreting and correlating these differences to interpolate reservoir properties. The result is a more detailed and reliable reservoir model than is possible with conventional methods.
Seismic data become more meaningful to reservoir characterization when geostatistical seismic inversion rigorously brings additional data from logs and transforms them into 5- to 10-ft vertical resolution, comparable to log-data resolution. By correlating seismic attributes with the geological model and reservoir properties indicated by well control, geostatistical techniques define the inter-well space, creating vertically and spatially high-resolution reservoir models. Such precise, realistic reflections of the geologic environment in which the reservoir was deposited also capture the complex heterogeneity in hydrocarbon reservoirs. As a result, these high-resolution, geostatistically derived models become a reliable basis for simulating and predicting reservoir and well performance, estimating recoverable reserves and identifying and deploying recovery methods that meet operators' needs to optimize production, minimize costs and mitigate risk.
The geostatistical reservoir characterization approach is also faster and more efficient than traditional reservoir technologies. An integrated team effort takes less time than individual disciplines working in a vacuum and allows discrepancies to be resolved as soon as they are identified. Model-building is an iterative process, seamlessly displayed in high-tech visualization facilities where multidisciplinary teams interactively manipulate and quickly collaborate on complex data and models.
Bryant G: Better than ever
Discovered in 1965 and on line since 1967, Bryant G's daily production from 19 wells peaked at 600 bbl of oil. Following a steady course of decline, daily production was down to about 2 MMcf of gas and 100 bbl by late 1995. Cumulative production at that time was about 28 Bcf of gas and 1.8 million bbl of oil. Although 3-D seismic data indicated significant reserves were not being tapped, 11 infill wells drilled between June 1994 and October 1995 didn't provide the desired boost to production. The question was how to get to the reserves and improve recovery efficiency.
Geostatistical reservoir characterization provided the answers. Data used in Texaco's geostatistical reservoir study included log and core samples from 55 wells and 30 sq miles (78 sq km) of 3-D seismic data acquired in November 1993. With geostatistical tools and expertise, these data were quantitatively integrated.
By using all the data, for example, correlation was established between porosity and impedance for each well rather than for all the wells mixed together as traditional reservoir characterization methods would do. Listening to the data meant recognizing there were significant differences from well to well, suggesting the field couldn't be drilled in a usual pattern.
Another geostatistical tool, "cloud transform," generated a permeability model that preserved the large variation of permeability values observed from core data. Because these extremely high- and low-permeability values dictate the flow characteristics of the reservoir, preserving the variations resulted in a much more realistic permeability model than standard (nongeostatistical) methods could produce.
From these data, a reservoir model was constructed that successfully predicted flow history and convincingly suggested optimum development choices. Well planning located and designed horizontal drilling to target the main producing level.
In February 1996, the first horizontal well came on line producing more than 3.5 MMcf/d of gas, and it remains one of the best horizontal wells in the field. Based on the same approach, during the next 3 years, Texaco drilled another 28 horizontal wells in the unit.
Recovery efficiency was improved. Daily production peaked at 56 MMcf of gas, more than 2,500 bbl of oil and 8,000 bbl of natural gas liquids, about a 25-fold increase. Field recovery for the 3-year period after the application of geostatistical reservoir characterization technology was about twice the recovery rate during the first 30 years of the field's life. The cost of about US $53 million for the 3-year drilling program paid off in the first year.