Surface-Gravity Monitoring of the Gas Cap Water Injection Project, Prudhoe Bay, Alaska

Abstract

Summary Modeling and field-test surveys indicate that time-differenced surface gravity can be used to successfully monitor the Gas Cap Water Injection (GCWI) Project at Prudhoe Bay, Alaska. Simulation results have shown that more than 95% of the injected water can be accounted for, including reasonable assumptions concerning the noise level in the measurements. The flood front can be reliably detected within 2,000 to 3,000 ft for the Prudhoe Bay reservoir that is buried at 8,200 to 8,800 ft true vertical depth (TVD). The world's first 4D surface-gravity surveillance of a water-flood is being implemented at Prudhoe Bay, Alaska. This monitoring technique is an essential component of the surveillance program for the approved Gas Cap Water Injection Project (GCWI) at Prudhoe Bay. A major factor in the approval of the waterflood was to show that water movement could be monitored economically with a very limited number of wells. Conventional monitoring techniques would have been cost-prohibitive because of the requirement for drilling numerous new surveillance wells. Modeling studies indicate that density changes associated with water replacing gas can be detected with high-resolution surface-gravity measurements. Field tests at Prudhoe Bay have demonstrated that sufficiently accurate gravity data can be obtained. This paper will discuss both inverse modeling of time-differenced gravity maps and four test surveys that have been completed to perfect the gravity-measurement technique. Forward and inverse gravity modeling were carried out on a suite of reservoir simulations of the GCWI. Differences in the gravity field with time reflect changes in the reservoir densities, which in turn reflect pore-fluid content. A constrained least-squares method was used to invert synthetic gravity data for the subsurface density distribution. The modeling procedure has been formulated to allow testing for sensitivity to gravity-sampling patterns, noise, and various constraints on model parameters such as density, total mass, and moment of inertia. These simulation results have shown that more than 95% of the injected water can be accounted for in the inverse model. These estimates include reasonable assumptions concerning the noise level in the measurements of both the gravity data and the location data using the Global Positioning System (GPS). The average flood front can be reliably detected within 2,000 to 3,000 ft for the Prudhoe Bay reservoir that is buried at 8,200 to 8,800 ft TVD. Time-differenced surface-gravity and GPS data were gathered over the Arctic Ocean in typical winter conditions (−40°F) accurately enough to monitor the waterflood (gravity =±10 μGal and GPS = ±1 cm elevation; a μGal is approximately one-billionth of the Earth's normal gravity).