Prediction of plume behavior and leakage

To complement numerical models, Mike Celia’s group has developed a fast semi-analytical model that is capable of quickly estimating leakage rates for hundreds of wells in a scenario with multiple geological layers (Figure 8). The speed of the model is an important advantage, as estimating total reservoir leakage in environments with large numbers of wells would be very time-consuming with more complex simulators.

After several years of development, the team now has a model that is more accurate and flexible than previous versions, can simulate a realistic injection and post-injection period, and can estimate leakage rates. This simulator now includes the following components:

  • Solution for injection dynamics.
  • Solution for multi-well leakage.
  • Solution for upconing along leaky wells.
  • Solution for post-injection plume evolution

For a realistic field site (see below) these calculations can be performed on a desktop or laptop computer in about 2 hours. The fast calculation times are allowing the group to use Monte Carlo techniques to develop rating curves that relate overall leakage rates to the underlying statistics of the well properties (Figure 9). In this way, quantitative measures like maximum leakage rates can be directly related to parametric values, such as maximum mean permeability, that characterize existing wells within the radius of influence of a proposed injection.

Figure 8. Schematic of the system that is modeled by the semi-analytical models. Any leakage along existing wells can form secondary plumes of CO2 in permeable layers above the injection zone, as shown in the figure.
Figure 9. A key advantage of the semi-analytical models is their computational speed and hence suitability for Monte-Carlo simulations. Here, the distribution of total leakage after 30 years of injection are shown based on assumed statistical distributions for properties of the existing wells.