Oil Recovery Prediction for Polymer Flood Field Test of Heavy Oil on Alaska North Slope Via Machine Assisted Reservoir Simulation

Abstract

The first ever polymer flood field pilot to enhance the recovery of heavy oils on the Alaska North Slope is ongoing. This study constructs and calibrates a reservoir simulation model to predict the oil recovery performance of the pilot through machine-assisted reservoir simulation techniques. To replicate the early water breakthrough observed during waterflooding, transmissibility contrasts are introduced into the simulation model, forcing viscous fingering effects. In the ensuing polymer flood, these transmissibility contrasts are reduced to replicate the restoration of injection conformance during polymer flooding, as indicated by a significant decrease in water cut. Later, transmissibility contrasts are reinstated to replicate a water surge event observed in one of the producing wells during polymer flooding. This event may represent decreased injection conformance from fracture overextension; its anticipated occurrence in the other production well is included in the final forecast. The definition of polymer retention in the simulator incorporates the tailing effect reported in laboratory studies; this tailing effect is useful to the simultaneous history match of producing water cut and produced polymer concentration. The top 24 best-matched simulation models produced at each stage of the history matching process are used to forecast oil recovery. The final forecast clearly demonstrates that polymer flooding significantly increases the heavy oil production for this field pilot compared to waterflooding alone. This exercise displays that a simulation model is only valid for prediction if flow behavior in the reservoir remains consistent with that observed during the history matched period. Critically, this means that a simulation model calibrated for waterflooding may not fully capture the benefits of an enhanced oil recovery process such as polymer flooding. Therefore, caution is recommended in using basic waterflood simulation models to scope potential enhanced oil recovery projects.

Concern Condition for Applying Optimization Techniques with Reservoir Simulation Model for Searching Optimal Rule Curves

“…Additionally, the water deficit and excess release situations observed using the obtained rule curves are not significantly different. Based on these results, it is concluded that the minimum initial capacity for the simulation model should be set at 10% of the reservoir capacity [89,90]…”

EOR of heavy oil polymer on Alaska's North Slope: History matching challenges and solutions in a significant long-term water cut reduction

“…A previous study of this research explored modern history-matching methods, including machine learning techniques such as CMG/CMOST [11] . This study focused on achieving a historic match between the field-scale model and production data.…”

“…Only a few studies have published numerical simulation history matches specifically for heavy oil reservoirs using polymer flooding [5] [14][8] [9] . These few published papers highlight different approaches for achieving accurate history matches by adjusting various input parameters, such as the slope of the relative permeability curve, endpoint tuning, formation permeability corrections, and skin effect modifications, irrespective of the machine learning methods employed [11] .…”

Recent Advances in Polymer Flooding in China

“…Research on polymers that displace very viscous oils should be creatively conducted. For example, the combination of polymer flooding with horizontal wells could produce unexpectedly good performance, which has been well-practiced in the Pelican Lake [ 13 , 14 , 15 , 16 , 17 ] and Alaska North Slope [ 22 , 23 , 29 , 217 ]. Hybrid methods [ 20 , 82 ], such as the combination of polymer flooding with LSW, could also be viewed as a promising alternative.…”

“…Recent advances in CO 2 EOR were reviewed in [ 12 ]. Despite the progress that was made in applying polymer flooding to very viscous oils in the Pelican Lake in Canada [ 13 , 14 , 15 , 16 , 17 ] and the Alaska North Slope in the USA [ 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 ], the selection of polymer viscosity and slug size to recover very viscous oils [ 25 , 26 , 27 , 28 , 29 ] remains quite different from that in China because of the not well-understood mechanisms under an unfavorable mobility ratio. Polymers have been regarded as not having an effect on residual oil saturation (ROS) reduction [ 5 , 30 ].…”