This page gives a problem description and the Eclipse 100 input decks for the simulations of CO2 storage in the paper:

Impact of relative permeability hysteresis on geological CO2 storage.
R. Juanes, E. J. Spiteri, F. M. Orr, Jr., and M. J. Blunt. Water Resources Research, 42, W12418 (2006), doi:10.1029/2005WR004806. (pdf)

Model description

We carried out simulations of CO2 injection in a synthetic but realistic model of a geologic formation, the PUNQ-S3 model. It is a geometrically complex and heterogeneous three-dimensional geologic model originally designed as a test case for oil production forecasting under uncertainty. The original PUNQ-S3 model is described in detail elsewhere [Floris et al., 2001], and the model data are publicly available for download (Netherlands Institute of Applied Geosciences, PUNQ case Studies).

PUNQ-S3 map of horizontal permeability

We modified the original model slightly to study hysteresis and trapping effects in a CO2 injection scenario. The modifications were limited to the well locations and flow rates, the fluid properties, the depth of the formation, and the relative permeability tables. The geometry of the model is characterized by a dome in the center and contains five layers of fluvial sand and shale. We set the top of the formation at a depth of 840 m. The average reservoir thickness is about 15 m. The formation is discretized into 19 x 28 x 5 grid blocks, of which 1761 blocks are active. The x and y dimension of each block is 180 m. The average porosity is 0.2, and the average horizontal permeability is 100 md. The permeability anisotropy ratio is about 3. A map of the horizontal permeability is shown here.

In order to model boundary conditions that are representative of an aquifer that extends beyond the simulation grid, we assign a very large pore volume to the boundary blocks. Specifically, we multiply the pore volume of these blocks by a factor of 1000. While this approach certainly cannot capture the flow dynamics associated with the surrounding aquifer, it has proved to be effective in practice. Upon CO2 injection, it allows that the brine leave the system, while reproducing a modest pressure buildup at the boundary.

Simulation runs and input decks

The grid geometry (.GEO) and material property (.PRP) files are common to all runs:

The input data (.DATA) for the different simulation runs:

  1. Gas injection, 10-yr injection, no hysteresis (CASE1.DATA)

  2. Gas injection, 10-yr injection, with hysteresis (CASE2.DATA)

  3. Gas injection, 50-yr injection, with hysteresis (CASE3.DATA)

  4. Water-alternating-gas (WAG) injection, 12-yr injection, with hysteresis (CASE4.DATA)

  5. Case 1, with refined grid (CASE1REF.DATA)


Massachusetts Institute of Technology · Department of Civil and Environmental Engineering

77 Massachusetts Avenue, Building 48 · Cambridge, MA 02139