xpertsystems/enr004-sample

ENR004是一个合成上游石油和天然气生产数据集（样本预览），包含四个物理校准的表：生产数据、井元数据、油田元数据和设施数据。数据集覆盖美国五个盆地（二叠纪、鹰福特、巴肯、阿巴拉契亚/马塞勒斯、墨西哥湾海上），提供日分辨率的井级生产时间序列数据，集成Arps递减曲线物理、Standing（1947）和Vasquez-Beggs（1980）PVT相关性、Ornstein-Uhlenbeck商品价格以及EPA Subpart W甲烷强度。校准优先参考SPE石油资源管理系统、API生产报告、EPA Subpart W和EIA钻井生产力报告等行业标准。样本包含50口井、8个油田、2年日数据（约15,699条生产记录），适用于类型曲线/EUR预测、生产速率预测、井经济学ML、储层压力下降建模、GOR/WOR进展建模、ESP/抽油杆泵异常检测、甲烷排放强度基准测试、盆地分类、举升类型选择、商品价格压力测试、封堵和废弃预测以及多表关系ML等用例。ENR004 is a synthetic upstream oil and gas production dataset (sample preview) comprising four physics-calibrated tables: production data, wells metadata, fields metadata, and facilities data. It spans 5 US basins (Permian, Eagle Ford, Bakken, Appalachian/Marcellus, GOM Offshore) with daily-resolution well-level production timeseries, incorporating Arps decline curve physics, Standing (1947) and Vasquez-Beggs (1980) PVT correlations, Ornstein-Uhlenbeck commodity prices, and EPA Subpart W methane intensity. Calibration is benchmark-first against SPE Petroleum Resources Management System, API production reporting, EPA Subpart W, and EIA Drilling Productivity Reports. The sample includes 50 wells, 8 fields, 2 years of daily data (~15,699 production records), and supports use cases such as type curve/EUR forecasting, production rate forecasting, well economics ML, reservoir pressure decline modeling, GOR/WOR progression modeling, ESP/rod pump anomaly detection, methane emissions intensity benchmarking, basin classification, lift type selection, commodity price stress testing, plug-and-abandon prediction, and multi-table relational ML.