Abstract

Within static geologic models of unconventional plays, the levels of uncertainty in porosity, saturation, and permeability may be as large as the values modeled. Property uncertainty has many sources, including original log measurements, the petrophysical model, and the method of property population. When static reservoir models are used as input for numerical simulation for field development planning, these uncertainties can significantly affect development strategies, economics, and estimated hydrocarbon recovery.

Uncertainty analyses are useful in quantifying the risk associated with static reservoir properties prior to dynamic modeling and for ranking potential acreage positions. Multiple realizations of static models are typically ranked by hydrocarbon pore volume (HCPV) during uncertainty analysis. These realizations are generated by shifting petrophysical input parameters (typically porosity and water saturation) commensurate with the total influence of the input uncertainties. A suite of probability cases are traditionally passed to reservoir simulation for history matching, forecasting, and field development planning. A drawback of this approach is that all cases going into the HCPV distribution are considered equally probable. Imposing no restrictions ensures that reasonable combinations of porosity and water saturation are used and that the property uncertainty space is adequately sampled. However, the nonuniqueness of HCPV can allow cases with extreme property combinations to be selected for reservoir simulation. Static models with extreme property values will ultimately fail during dynamic modeling, requiring the static model to be revisited, which costs time and delays results.

A methodology is proposed that enables systematic selection of property models by identifying the probability distribution of HCPV as a function of porosity and saturation shifts. Knowing the impact of each petrophysical variable's shift on the HCPV allows selecting specific cases that sample the range of petrophysical uncertainty space that also represents the range of HCPV. The resulting static models can be used for the purposes of further uncertainty and sensitivity analyses during dynamic modeling.

This content is only available via PDF.
You can access this article if you purchase or spend a download.