Statistical Ranking of Stochastic Geomodels Using Streamline Simulation: A Field Application
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Abstract Streamline-based flow simulation for the purpose of ranking large-scale geologic realizations continues to receive significant attention. However, the procedures and the analyses for ranking are not straightforward and therefore actual case examples are very limited. This paper describes a field example showing a very practical process for dynamically ranking various geologic realizations using uniform well patterns. This mature field has a 60-year primary recovery history but still has potential for additional development. The ranking process is further complicated by the presence of a gas cap and a water zone. A major difficulty with dynamic ranking of geological models is that the recovery may be as much a function of the flow-physics as the geologic variability. Accounting for gravity, fluid contacts, changing streamlines, and fractional flow effects may be important to the ranking study. Even the choice of well locations, rates, boundary conditions, and patterns will affect the ranking. The uniform patterns used in this study are not representative of actual well patterns or injected fluids rates. The waterflood efficiency, however, can still be used as a basis of comparison. A novel map based presentation of the ranking simulations provides valuable understanding of the effect of the geologic model on recovery uncertainty. The use of regular well patterns is different from the common approach of using existing wells with pseudo boundary conditions. The uniform spacing ensures complete coverage of the area-of-interest and not just the areas where the model is already conditioned to existing data. This method tests the variability of the models away from existing wells as these areas will have longer-term effect on performance and affect the decision regarding future infill wells and recovery methods. Another important aspect of this paper is a demonstration of how modern tools and analysis techniques are greatly improving the ability to understand complex reservoirs and thus make improved decisions regarding optimum development. Efficient analysis and visualization of the data and interpretations is important for a detailed understanding of the reservoir. Motivation for Study The methodologies described here resulted from several major considerations:evaluate the impact of geologic uncertainties on production performance within a one month window during which a conventional history match is performed;use existing commercial software to prevent long delay time in project completion,present the results in a manner which visually relay the results to a wide audience, anddevelop a methodology which provides more information than a simple cumulative distribution of field recovery. Anyone involved in reservoir simulation realizes there are several potential sources of errors or uncertainties when doing a reservoir study:numerical error (from the approximate solution of non-linear partial differential equations),error from the approximations in the underlying equations (e.g. 3-phase approximation of Darcy's law)errors or uncertainties in data interpretation (e.g. converting log signals to reservoir properties),ignored data (e.g. not using the seismic data in reservoir property distribution),unknown or uncertain data (e.g. only a small portion of the reservoir is sampled) andincorrect averaging of data (e.g. averaging log measurements over a flow unit). All of these errors or uncertainties lead to uncertainties in forecasts of future production. Recognition of these uncertainties has lead to a desire to incorporate the resulting uncertain rate and recovery forecasts into a corporate risk analysis methodology1–9.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it