Principal Component Analysis
of
Seismic Data and Reservoir
Sequence
Stratigraphy: Powerful Allies in Reservoir
Modelling and Waterflood Management, Gbokoda Field, Nigeria
By
Stan P. Franklin1, Steve Jenkins1
(1) Chevron Nigeria Ltd, Lagos, Nigeria
Principal component analysis
(PCA) of seismic data, combined with reservoir
sequence
stratigraphy, creates a powerful method for mapping and modelling
pressure compartmentalization. Understanding compartmentalization during
waterflooding is key to locating water injectors and producers to maximize sweep
efficiency.
Gbokoda Field is located onshore Niger Delta and contains over 850 mmb oil-in-place. The main A-05 reservoir is a shoreface sandstone and was placed on peripheral waterflood during primary development due to anticipated weak aquifer support.
After 3 years of waterflooding, our pressure surveillance program indicated
pressure compartmentalization with ~350 psi of differential depletion across the
field. Several western wells ceased production due to low tubing pressure,
despite good injection from a horizontal injector downdip. In contrast, the
eastern portion of the field showed minimal depletion. Analysis
of the
log
and
seismic data within the western area showed a poorly expressed incised channel.
Starting with Gbokoda 3D seismic data, we used PCA to perform seismic curve shape comparison to group the different curve shapes into mappable clusters. Amplitude and phase form the largest contribution to the variance of the clusters. Clusters represent different rocktypes, facies, and fluids creating high-resolution mappable units.
By combining the PCA-based mapping with log
and core-based
sequence
stratigraphy, we determined the producers were located within the shoreface
sequence
, whereas the injection
well
was located within a disconnected incised
valley fill
sequence
.
After updating our simulation model, we restored production and reversed the
pressure depletion by reactivating the producers and completing a new horizontal
injector within the shoreface sequence
.