High-resolution stable carbon isotope monitoring indicates variable flow dynamic patterns in a deep saline aquifer at the Ketzin pilot site (Germany)

• Isotopically very distinct kinds of CO2 are injected at the Ketzin pilot site.
• We used these isotope variances at the injection well as a tracer input signal.
• We found none of these signals at a monitored observation well.
• This indicates that CO2 is immobilised at this point in the reservoir.
• After immobilization, the δ13CCO2 value was altered by fractionation processes.

Stable isotopes of injected CO2 act as useful tracers in carbon capture and storage (CCS) because the CO2 itself is the carrier of the tracer signal and remains unaffected by sorption or partitioning effects. At the Ketzin pilot site (Germany), carbon stable isotope composition (δ13C) of injected CO2 at the injection well was analyzed over a time period of 4 months. Occurring isotope variances resulted from the injection of CO2 from two different sources (an oil refinery and a natural gas-reservoir). The two gases differed in their carbon isotope composition by more than 27‰. In order to find identifiable patterns of these variances in the reservoir, more than 250 CO2-samples were collected and analyzed for their carbon isotope ratios at an observation well 100 m distant from the injection well. An isotope ratio mass spectrometer connected to a modified Thermo Gasbench system allowed quick and cost effective isotope analyses of a high number of CO2 gas specimens. CO2 gas from the oil refinery (δ13C = −30.9‰, source A) was most frequently injected and dominated the reservoir δ13C values at the injection site. Sporadic injection of the CO2 from the natural gas-reservoir (δ13C = −3.5‰, source B) caused isotope shifts of up to +5‰ at the injection well. These variances provided a potential ideal tracer for CO2 migration behavior. Based on these findings, tracer input signals that were injected during the last 2 years of injection could be reconstructed with the aid of an isotope mixing model and CO2 delivery schedules. However, in contrast to the injection well, δ13C values at the observation well showed no variances and a constant value of −28.5‰ was measured at 600 m depth. This is in disagreement with signals that would be expected if the input signals from the injection would arrive at the observation well. The lack of isotope signals at the observation well suggests that parts of the reservoir are filled with CO2 that is immobilized.