Diatomites are high-porosity, low-permeability reservoirs with elastoplastic properties and high geo-mechanical responsiveness. They have a great potential for oil recovery despite these drawbacks. Withdrawal of fluids from the reservoir rock leads to subsidence causing compaction and shear stresses. This disturbed stress distribution results in well failures that causes loss of millions of dollars. Successful maintenance of pressure support through optimum injection/production is key to preventing subsidence to mitigate the risk of well failure and achieve better sweep efficiency for recovery.

There have been different approaches to tackle subsidence and well failures in diatomites, including the use of ‘backpressure method’, coupled with a neural network to optimize injection-production to ‘balance’ the rock in terms of stress-distribution and thus decrease well failure due to shearing. However, using such methods may mask other problems the well is experiencing including several mechanical issues that influence production. Another existing approach, satellite-imaging (InSAR) cannot be used to take real-time actions that is crucial in diatomites.

Surface tiltmeter data is collected to undertsand the relationship between injection/production and resulting surface deformation, which also provides information about well-to-well connectivity. A neural network-based approach is followed to determine the nonlinear relationship between surface subsidence/dilation and injection-production. This is then used to build an objective function that works to minimize the differences between well-to-well subsidence/dilation measured by the tiltmeters, by adjusting injection-production for the wells.

In this paper, a method that harnesses real-time surface tiltmeter data to adjust injection-production distribution in diatomites to decrease well failures is used beyond the existing applications of surface tiltmeter, for instance, in the areas of detection of early steam breach to surface in steam operations and fracture orientation. This method also provides real-time data for robust reservoir management of such reservoirs where satellite imaging is not effective.

You can access this article if you purchase or spend a download.