Hydrocarbon allocation becomes essential when facilities are shared for various contributing wells or reservoirs. Uncertainty in the reassignment factor can significantly impact operators’ daily decisions, reservoir health, and ultimately the organization's financial investment. Historically, most oil companies rely on the latest production test done at the actual choke to calculate the produced volume per string, regardless of changes in well performance over time. But this simplistic routine leaves a gap between estimated and actual rates that require more user interventions for manual inputs or corrections.
This paper illustrates an example of hydrocarbon allocation process from a smart field. An intelligent optimization system that receives real time data instantly automatically back allocates the field rate based on continuously-tuned well performance curves at wellhead node. This system is capable of managing large amount of data as well as combining analytical models, data mining with predictive analytics, bringing significant benefits to reservoir management, and surveillance process. However, it was found that the field correction factor fluctuated about 25%, when some production optimization and well operations were performed due to inconsistencies in well performance curves.
The traditional methodology to calibrate well models starts by choosing the nearest multiphase flow vertical correlation to the measured pressure gradient, reducing the error between calculated and measured points by tuning the "L factor." This procedure cannot ensure that the resulting model will satisfy other production tests at different flowing conditions, which reveals the difference between accuracy and consistency. In contrast, the present effort suggests a practical approach for better estimation of daily well rates as well as improving allocation process, since it attempts to minimize the error between representative production tests and well performance curve rather than reproduce a high accurate bottom hole pressure for a particular flowing condition. Additionally, a more representative productivity index is derived from this change.
This novel approach showed better well rate estimation for different flowing conditions reducing reassignment factor deviation below 5%. Consequently, it proved that having a more consistent well model tuned by representative production tests and controlled by real time data, enhances oil back allocation process.