xpertsystems/oil016-sample

OIL-016是一个合成油井递减曲线数据集样本，由XpertSystems.ai发布，用于机器学习在石油和天然气上游储层工程中的应用。该数据集是一个免费预览版本，基于企业级数据集，专注于Arps递减曲线分析（DCA）、EUR（估计最终采收量）预测、储量分类和长期生产分析。样本包含500口油井，覆盖6个全球盆地（如二叠纪、鹰福特、巴肯等），模拟了360个月（30年）的生产历史，总计361,809行数据，包括180,000个月度生产记录，分布在7个关联表格中。数据表格包括井主表、生产时间序列、递减参数、EUR预测、压力衰减、人工举升事件和生产中断表。数据集经过行业标准校准，使用10个指标验证，参考了Arps（1945）、SPE PEH Vol V、SPE 167242、SEC 2008、SPE PRMS 2018等标准，验证结果显示整体得分为100.0/100（A+级）。数据集支持多种机器学习用例，如递减曲线回归、分类、时间序列预测等，并提供了加载方法和可重复性保证。但样本存在一些限制，例如均匀采样递减模式、无终端递减转换、线性压力衰减等，这些在诚实披露部分有详细说明。数据集适用于研究和开发，但不适用于实际储量决策。OIL-016 is a synthetic well decline curve dataset sample released by XpertSystems.ai for machine learning applications in upstream petroleum and natural gas reservoir engineering. This dataset is a free preview version built upon an enterprise-level dataset, focusing on Arps decline curve analysis (DCA), EUR (Estimated Ultimate Recovery) forecasting, reserve classification, and long-term production analysis. The sample contains 500 oil wells covering 6 global basins (e.g., Permian, Eagle Ford, Bakken, etc.), simulates 360-month (30-year) production histories, and totals 361,809 data rows including 180,000 monthly production records, distributed across 7 associated tables. The data tables include the well master table, production time series, decline parameters, EUR forecasting, pressure depletion, artificial lift events, and production downtime tables. The dataset has been calibrated against industry standards and validated using 10 metrics, referencing standards such as Arps (1945), SPE PEH Vol V, SPE 167242, SEC 2008, SPE PRMS 2018, etc., with validation results showing an overall score of 100.0/100 (A+ rating). The dataset supports a variety of machine learning use cases, such as decline curve regression, classification, time series forecasting, etc., and provides loading methods and reproducibility guarantees. However, the sample has certain limitations, including uniformly sampled decline patterns, no terminal decline transitions, linear pressure depletion, etc., which are detailed in the honest disclosure section. The dataset is intended for research and development purposes only, and is not applicable to actual reserve decision-making.