Assessing Geological Uncertainty: Counting the Costs
It is a common misconception that inherent subsurface geological complexity prevents us from accurately estimating key reservoir parameters. The understanding developed over many decades of research combined with extensive operational experience means that general reservoir geology is no longer a mystery. The real challenges are our lack of understanding and data coverage and quality for a given target reservoir interval.
What Happens Today?
Current technologies support the generation of suites of three-dimensional (3D) models; this has become best practice for simulating the physical properties of oil and gas fields. They are a fundamental part of the reservoir characterization workflow, used to forecast field production, and are the basis for assessing economic viability and the associated risks.
Uncertainty is present in all aspects of the reservoir modeling workflow, from obtaining the raw data, to data processing and interpretations, through to model creation. As we cannot peer into the subsurface directly, our non-unique geological interpretations are based on expert knowledge/assumptions combined with a mixture of locally available hard and soft data.
Multiple versions of the same model are often built. There are a variety of approaches to generate these model suites, but regardless of the approach, multiple models can help geoscientists identify, assess and evaluate the geological uncertainties, as well as quantify their cumulative economic impact.
Emerging data collection and analysis approaches, combined with the scalability offered by cloud architectures, provide opportunities to increase complexity in our model representations. However, generation of larger and more complex models without appropriate geological constraints is not valuable. This potential for misdirected effort is unacceptable, especially when many teams are currently understaffed. Geoscientists are often left without the time or the technology to generate either multiple plausible scenarios for a single reservoir and/or multiple models for individual aspects of their simulations, such as facies, petrophysical and structural models. The representations of these three areas of uncertainty are susceptible to ambiguous data and subjective interpretation.
The State of Play for Structural Modeling
Geoscientists recognize that in the vast majority of geological settings, a reasonably accurate and plausible structural model is crucial in terms of estimating the gross value and development cost of an oil or gas field. They also recognize that anchoring any subsequent uncertainty workflows on a single structural scenario is undesirable, but often challenging to realize, given time and technology constraints.
The number, location and connectivity of faults throughout the reservoir are the primary factors that determine the number and size of reservoir compartments, impacting reservoir efficiency, sweep patterns and ultimate recoverable volumes. Failure to express the potential for structural compartmentalization is risky and is the subject of many after-action well reviews.
To fully characterize structural uncertainty in a reservoir, objective assessment of fault interpretation quality is needed, testing it with basic structural geology rules. A detailed understanding of the fault network and connectivity allows geoscientists to develop and probe plausible alternative structural models that are representative of the range of uncertainty, and its impact on the number and size of reservoir compartments. Importantly, they can act as proxies to development cost.
Links Between Structural and Saturation Modeling?
Increased confidence in investment decisions can result from understanding structural uncertainty in detail, in particular its interplay with saturation modeling.
Unfortunately, the derivation and then application of a saturation height function to a reservoir model with multiple fault blocks can be a time-consuming, arduous process. This is because building a saturation model involves many calculation steps, the inputs to which have their own specific uncertainties (e.g. porosity depends on the logging tool, calculation method and input parameters).
Frequently, a single saturation height function is applied to the reservoir model with uncertainties relating to the calculation method and its impact on reservoir volumes not considered due to a lack of time. The relationship between permeability distribution and saturations is often a single scenario and uncertainties in free water level depth are not always considered when deriving a saturation-height function.
However, not exploring multiple scenarios can have a significant impact on the distribution and volume of hydrocarbons in a reservoir, which can in turn have important economic implications.
How About Facies Modeling?
In terms of facies distributions in the inter-well space, although it’s possible to investigate aspects of potential depositional settings through experimental equivalents in the lab and numerical simulations, the processes cannot be replicated to scale. To fill this knowledge gap, we have to rely on other information to augment our understanding.
Previously, geological and expert knowledge of similar assets would be relied upon to define facies properties in the inter-well space. Though valuable and still necessary, there are inherent shortcomings to this approach: it is subjective and lacks the rigor of evidence-based decision-making. Ample data acquisition efforts now mean that incorporating analogs in reservoir modeling workflows is considered best-practice However, it remains a time-consuming process that is still somewhat subjective and highly variable, often depending on available domain expertise.
To fill some of these gaps there is a requirement for high quality sedimentological databases, which encapsulate scientific rigor in the data collection stage and which incorporate a breadth of information from both modern and ancient successions. They must have the potential to represent analogs to present-day hydrocarbon reservoirs by including quantitative analog data distributions, which can be used to inform workflows such as facies modeling.
By incorporating appropriate analog information, an improved and defensible distribution of reservoir facies through the model can be achieved. Crucially, effective investigation of depositional uncertainties is enabled by the selection of suitable analog datasets to define a range of contrasting, but geologically realistic, scenarios.
An Uncertain Future?
The multifaceted nature of geological uncertainty in 3D modeling makes its appropriate characterization one of the biggest challenges that the oil and gas industry faces. Despite the difficulties in defining reasonable ranges for these key parameters, the economic costs of not doing so are significant. As the industry is forced to adapt to the ongoing demographic transformation and the exploitation of resources in increasingly marginal geological environments, the benefits of dedicating more to the appropriate assessment of uncertainty in 3D models far outweigh the costs.
There are many exciting new technologies coming into the market that help assess these key geological parameters. The integration of established approaches, such as databasing, with more recent innovations such as the application of machine learning, create new opportunities for geoscientists to explore the financial impact of critical geological variables on the economics of their asset, in an efficient, rigorous and repeatable way.