MétaCan
Menu
Back to cohort
Record W2983689482 · doi:10.2118/197142-ms

Artificial-Intelligence-Based, Automated Decline Curve Analysis for Reservoir Performance Management: A Giant Sandstone Reservoir Case Study

2019· article· en· W2983689482 on OpenAlex

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.

affAt least one author lists a Canadian institution in the pinned OpenAlex snapshot.

Bibliographic record

Venuenot available
Typearticle
Languageen
FieldEngineering
TopicReservoir Engineering and Simulation Methods
Canadian institutionsImpact
Fundersnot available
KeywordsComputer scienceLift (data mining)Cluster analysisData miningArtificial intelligenceField (mathematics)Production (economics)Machine learning

Abstract

fetched live from OpenAlex

Abstract Decline curve analysis (DCA) is one of the most widely used forms of data analysis that evaluates well behavior and forecasts future well and field production and reserves. Usually, this practice is done manually, making analysis of assets with a large number of wells cumbersome and time-consuming. Moreover, results are subject to alternate interpretations, mostly as a function of experience and objectives of the evaluator. In this work, despite the common practice of the industry, i.e. manual DCA, we developed and deployed cutting-edge technologies that intelligently apply DCA methods to any number of wells in an unbiased, systematic, intelligent, and automated fashion. The tool reads production data, and multidisciplinary well information (e.g., drilling and completion data, geological data, artificial lift information, etc.). Then it performs cluster analysis using unsupervised machine learning and pattern recognition to partition the dataset into internally homogeneous and externally distinct groups. This cluster analysis is later used for type-curve generation for wells with short production history. For wells with long enough history, the tool first detects production events through a fully automated event detection algorithm without any human interference. Since production events are highly correlated with real-time events, it also cross-validates with the operating conditions. Next, the last event is selected, and a decline curve is fitted using advanced nonlinear optimization and minimization algorithms. This leads to a reliable and unbiased prediction. For each cluster, a type curve is computed that truly captures the underlying production behavior of the wells that belong to the same group or cluster, and then is applied to the wells with short production history within that cluster. To capture the probabilistic nature of such analysis and quantify the inherent uncertainty, we extended the method to a probabilistic DCA using quantile regression. We successfully deployed this technology/tool to a giant Middle Eastern reservoir, with more than 2,000 wells and 70 years of production. Our predicted aggregated field decline rate is in good agreement with the client's reservoir simulation results run under the "do-nothing" scenario. While performing traditional DCA for such a field would require several weeks and significant resources, our automated solution integrates all real-life events/information and provides a comprehensive analysis in field, cluster and well level. In addition, our results are "unbiased," as it is not subject to human errors or evaluator's interpretations. Our robust and intelligent DCA allows for exhaustive evaluation of production trends and opportunities in fields across time, production zones, well types, and any combinations of the above. The results demonstrate the effectiveness of the automated DCA to rapidly execute decline curve analysis for a large number of wells. The accuracy is improved significantly through automatic event detection, cross-validation of events, curve fitting optimization, quantile regression, and cluster-based type-curving.

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 imitation

Not 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.

metaresearch head score (Codex)0.002
metaresearch head score (Gemma)0.000
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesMeta-epidemiology (narrow)
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Simulation or modeling · Consensus signal: Simulation or modeling
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.283
Threshold uncertainty score1.000

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0020.000
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0010.000
Bibliometrics0.0010.002
Science and technology studies0.0000.000
Scholarly communication0.0000.000
Open science0.0000.000
Research integrity0.0000.000
Insufficient payload (model declined to judge)0.0000.000

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.

Opus teacher head0.046
GPT teacher head0.329
Teacher spread0.283 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it