Analysis of Pressure Data As CO2/Brine Leak Diagnostic in Shallow Aquifers

Tuesday, 16 December 2014: 4:15 PM
Whitney Trainor Guitton1, Kayyum Mansoor2, Yunwei Sun3 and Susan Carroll3, (1)Lawrence Livermore National Laboratory, Livermore, CA, United States, (2)Lawrence Livermore National Lab, Livermore, CA, United States, (3)Lawrence Livermore Natl Lab, Livermore, CA, United States
Pressure is a promising signal for detecting CO2leakage from deep, geologic storage reservoirs to shallow groundwater sources. Pressure signals should faster than other physical indications (i.e. electrical or geochemical changes) thus allowing for a timely leak diagnosis and mitigation.

We explore the effectiveness of pressure as a detection tool. A simulation-based approach is used to diagnose a CO2/brine leak using pressure data from monitoring wells and to assess the influence of 3 principal uncertainties: distances between the source leak and the monitoring well, heterogeneity of the aquifer flow properties, and CO2 and brine leakage rates. Specifically, five parameters are sampled: the correlation lengths of the vertical and horizontal permeability for the aquifer (2), the sand proportion for each model (1), and the CO2 and brine leakage flux magnitude (2). Areal model dimensions and grid cell dimensions allow for sampling distances of 25 m to 990 m from the leaking well to the monitoring well. We generate 500 simulations by sampling each parameter within an appropriate range predefined by site-specific values.

Pressure transducers in monitoring wells will only be accurate at measuring changes on the order of 0.1 to 0.3 PSI. These pressure thresholds are used to establish which simulations are classified as leaks at the leaking location and which locations away from the leak would constitute a signal. We observe 3 conclusions from the results:

  1. vertical flow barriers (heterogeneity) creates complicated pressure signals by forcing convoluted flow paths
  2. false positives (Pr( No Leak | Signal)) do not occur with our sample simulations
  3. false negatives (Pr( Leak | No signal)) dominate after 200 days even when considering only potential monitoring wells within 100m of the leaking well.

We use these posteriors to calculate the value of information (VOI) from above zone pressure data. This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under contract DE-AC52-07NA27344. Lawrence Livermore National Security, LLC. LLNL-ABS-657971