Fast Calibration of Agbami Intelligent Oil Production Network using Structured Computer Assisted Workflow

Okpani, Olu (Chevron Nigeria Ltd) | Momoh, Victor (Chevron Nigeria Ltd)



Agbami field is a deep-water field located offshore Nigeria. The oil production system is a complex intelligent production network that consists of 22 subsea production wells (19 wells are dual zone completion while 3 wells are single zone completion), 8 subsea manifolds, 8 infield subsea flowlines and 8 subsea flowlines/risers. A production network model for Agbami production network was built in GAP production modeling tool. The model consists of 41 inflow elements, 177 pipe elements, 68 chokes and 200 nodes with real-time pressure/temperature (P/T) measurements. Due to the large number of elements and P/T nodes in the model, it was very daunting to calibrate the model by unstructured manual tuning of the model calibration parameters. In fact, it takes several days to manually calibrate the whole production network. A structured computer assisted calibration workflow was developed to aid in fast calibration of the Agbami production model.

The structured approach to the Agbami production network model calibration used in this work is to first break down the model into 8 independent riser subsystems. Each riser subsystem is then further broken down into segments. The segments are zones, wells, infield flowlines, and flowlines/risers; with each segment having several elements and P/T nodes. Each segment of the model is independently calibrated using the latest production test data corresponding to that segment. A computer guided wizard was developed to sequentially match the P/T at different nodes of the segment using the Secant root-finding algorithm1. The calibrated segments are then coupled together into riser subsystems and each riser subsystem is then calibrated to the latest riser tests by manual adjustment of few parameters. The riser subsystems are further calibrated to the current conditions. The use of the structured computer assisted workflow has resulted in the faster model calibration time of within a day.