Dynaflow development and simulations

If cements in existing wells were exposed to large volumes of CO2-rich fluids, degradation like that seen in laboratory experiments would be a serious drawback for underground carbon storage. However, carbonated brines might become saturated and less agressive by the time they travel through a formation and come into contact with cements. To assess fluid compositions at wells, Jean Prevost’s model Dynaflow has been adapted this year to include pH. Experiments are now being initiated to simulate brine/cement interactions in the batch reactor experiments described above, and to simulate the composition of brines that might contact abandoned wells.

The Dynaflow model is also being used to simulate underground CO2 injection at American Electric Power’s Mountaineer site in New Haven, West Virginia. Working with colleagues from Battelle, Jean Prevost and Mohammad Piri are gathering information on subsurface conditions to design a simulation of the planned injection of CO2 into a saline aquifer.


Improved upscaling in numerical models

Simulating the fate of injected CO2 requires numerical models with large spatial scales, typically with domains 10 to 50 km across. On this scale, grid cells will necessarily be tens to hundreds of meters in size. Incorporating the impacts of multiple leaky wells in this type of simulation is challenging, since well dimensions are so much smaller than the gridscale. Fine-grid simulations can be used to differentiate flow within a well from that in the surrounding rock, but are computationally expensive and impractical for modeling large numbers of wells. To get around this problem, standard petroleum reservoir pseudo functions can be used to “upscale” properties in coarse-grid models, yielding a single value that accounts for both well and surrounding rock matrix in one grid cell.

However, even for a simple system in which a high-permeability well is embedded in a homogeneous rock (at the grid scale), use of the standard pseudo function for upscaling can overestimate flow by hundreds of percent. Sarah Gasda and Mike Celia have developed a hybrid approach for upscaling, using information on well and matrix flow from a fine-grid simulation to correct a standard pseudo function for use in a coarse-grid model. The corrected pseudo function dramatically reduces the inflated flow rates produced by the standard method, bringing leakage estimates into agreement with fine-grid predictions.