A unique approach has been developed to optimize injection for managing excess produced water at oil and gas fields. A simulation case study is presented demonstrating how the approach helped ChevronTexaco to quantify ultimate injection volume, number of injectors, injection rate per well, and well location for two sandstone aquifers downgradient and fault separated from their associated oil reservoirs. The approach combines stochastic, deterministic, and optimization models and is flexible to honor all the available data and also incorporate uncertainty in the field parameters (e.g. permeability distributions, boundary and initial conditions, and geologic barriers). A stochastic version of the model is best used to assess uncertainty in the geologic and pressure regimes. These runs are preceded by traditional deterministic simulation techniques, including history matching to the available pressure transient data. Once the scenarios have been selected, optimization simulations quantify the optimum number of injection wells and location, and the total volume of water that can be injected over the period of interest. A particle tracking technique is also used to define the pathways of the injected water and assess its ultimate fate and potential impacts. This approach can be used cost effectively to optimize produced water management under conditions of uncertainty with a sparse data set.
Typically, subsurface data at potential or existing injection fields is limited. Generally, prospective wells are sited based on available geological, geophysical, and injectivity testing but rarely on a comprehensive assessment of pressure buildup throughout the injection zone and optimization of well location based on pressure buildup constraints. The reason that this step is not generally undertaken is what is perceived to be a lack of necessary data to perform such an analysis.
A unique approach has been developed to optimize the required number of injectors, injection rate per well and well location at the lowest cost and using any number of pressure constraints. The overall approach is summarized on Figure 1.
During the model setup stage a database of all available location, geologic, geophysical, pressure, production and injection data is constructed. Geological interpretation is then performed and is used to construct a three dimensional block model. The block model is then converted to a finite-difference numerical simulation model.