How Uncertainty in Real Time Data is Managed in Hess
Luigi Saputelli, PhD, HESS Corporation
Field sensor data is a key component for optimizing production in real-time, however measurement uncertainty and data availability prevent operators from achieving such goals.
Field measurement performance is ususally the combined result of sensor health, tolerance, precision, drift of accuracy, calibration, compensation, aging and position. In addition process fluctuation, rate of sampling and environment temperature and noise will affect measurement performance.
Production model error sources are usually demonstrated by the lack of validation with field data (model vs actuals), driven by the lack of update with new field data or new field status. In addition model errors are exhacerbated by insufficient data, erroneous assumptions and invalid sensor data
As a result, errors induced by wrong sensor data are propagated through "history-matched" models (reservoir characterization, well models) leading to poor decision management (production/injection allocation, safe operating envelopes).
This paper discusses the implementation of a data validation and reconciliation study in an offshore field to improve real-time data performance. In this case, data, uncertainty and models are combined to minimize a global error function. Rigorous statistics are used to calculate new sensor estimates.
The project technical scope included calculating key well measurements, including tolerance of rate measurements and gauge readings, and identifying sources of unreliable measurements. Although the approach is not new in the petrochemical industry, the application is 'young' in the upstream.
Project benefits included less downtime due to well testing and early problem detection, 65% less time expended on validating well test and allocating individual well rates, and improved cost control due to calculating well-produced volumes hourly. These findings provide a better understanding of reservoir and well performance, which facilitates production optimization management.