6.4.7 CO2 storage project in Schwarze Pumpe – Schweinrich, Germany

For the assessment of the Schweinrich storage structure in Germany, a modified performance assessment (PA) methodology was used comprising the following steps: (i) definition of the assessment basis, (ii) FEP analysis, (iii) safety scenario formation, (iv) development of dedicated models for probabilistic simulation of safety scenarios, and (v) safety evaluation against HSE effects. The FEP database holds FEPs that may have a potential effect on the safety of the storage system. The latest version of the database contains a total number of 657 FEPs, extracted from various sources. All FEPs have a complete set of identification and classification attributes. These attributes have been assigned generically, and could be used to filter case-specific FEPs with respect to the assessment basis. Case specific FEPs for the Schweinrich case were identified according to the following criteria:

  1. FEPs should have a timescale of occurrence less than 1000 years,

  2. FEPs should lie within the spatial domains of reservoir, seal, overburden and fault,

  3. FEPs in the spatial domains shallow subsurface, ocean, atmosphere and underburdon are omitted,

  4. FEPs with respect to well integrity and engineering are not evaluated since the design and completion of future injection wells is unknown. EPs for the Schweinrich case will be divided into geochemical EPs acting on long timescales (about 1,000 years), and into geomechanical EPs valid for both short and long timescales of occurrence and duration.

Two EP groups were identified: a leaking fault EP group and a leaking seal EP group. The geomechanical EPs all relate to the leaking fault EP group. The geochemical EPs relate to both the leaking fault and the leaking seal EP group (Chadwick et al., 2008). Some limitations are applicable to the Schweinrich case study, as follows:

  • Time frame: The time frame for the FEP analysis was set to 1,000 years. Hazards that may occur as consequence of the identified safety factors were evaluated for 10,000 years, i.e., the simulation period was 10,000 years.

  • Spatial domain of the investigated storage system: The reservoir, seal, overburden, faults and wells compartments were evaluated. The shallow subsurface, ocean (not relevant for structure Schweinrich), atmosphere and underburden compartments were excluded. This selection process is related to the available input data and limitations in the model.

  • Probability of occurrence of evaluated scenarios: No attempt to quantify the probability of occurrence of the evaluated scenarios has been made. Instead, it was assumed that the scenarios will definitely occur, i.e., the probability of occurrence of the CO2 leakage scenarios is set to 1. These evaluated scenarios represent worst cases.

  • Input data: The study used input data that were gained from former geological surveys of the area.

  • Model limitations

Based on the FEP analysis and the scenario formation, the following "what if?" scenarios were identified for simulation:

  • Reference scenario assuming no failure of the containment zone occurs,

  • Leaking seal scenario assuming seal failure by geochemical processes and CO2 migration into the overburden,

  • Leaking well scenario.

Model software was used for the simulation of four scenarios. In the Schweinrich case, the scenarios present hypothetical future flow and fate of CO2 in the next 10,000 years. The potential impact of each scenario was expressed as the maximum concentration and flux of CO2 in the pore system in the shallowest overburden unit, Pleistocene sediments (which form the topmost subsurface layer in the simulation models). No outcome was simulated regarding groundwater deterioration and mobilisation of heavy metals, since no modelling of the flow and fate of CO2 in the unsaturated zone was conducted. In case of uncertainty on input parameters that were not varied stochastically, the worst-case scenario values were generally selected. Outcome distributions are consequently biased towards the worst-case scenarios. A 2D radial flow model was used to represent the reference scenario, the seal leakage scenario (Fig. 6-10a) and the well leakage scenario (Fig. 6-10b), while a 3D orthogonal model was used to represent the fault leakage scenario (Fig. 6-10c). Simulation was carried out with a 3D multiphase flow simulator called SIMED II. The amount of injected CO2, its lateral spread in time and the reservoir pressure were calibrated to the fine-scaled 3D SIMED II model over the injection period of 40 years.

E. Fig . 6-10

Fig. 6-10: Simulated CO2 saturations from the hypothetical leaking a) seal, b) well and c) fault scenario (probability of occurrence is set to 1) in the Schweinrich case (Chadwick et al., 2008).

This deterministic model represented the injection of CO2 on the flanks of the Schweinrich structure by 10 injection wells. In this model, the accumulation of CO2 was mainly in the topmost reservoir layer. The representation of the stochastic models was strongly simplified. Each 'what if?' scenario was evaluated with 1,000 model runs with varying stochastic parameters. Based on the results from these model runs concerning the safety of the reference scenario, no CO2 reaches the uppermost overburden sediments after 10,000 years. Regarding the leaking seal scenario, although CO2 passes through the seal in this scenario, the velocity of upward migration of CO2 is fairly small and therefore, no CO2 reaches the uppermost overburden. The leaking well scenario is easily the most significant in terms of modelled CO2 fluxes and CO2 concentrations in the shallow overburden. The probability that such a scenario will be valid depends on the existence of an old well, designed for a purpose other than CO2 storage, penetrating through the caprock. The critical safety factor in the leaking well scenario is the magnitude of the increase of the (vertical) permeability in the well zone, which would be improved by using a proper cement type. However, the best way to avoid the leaking well scenario is to design the injection wells in such a way that the scenario cannot occur, for example by designing the wells so that the caprock is not penetrated and that the wells enter the anticlinal structure from below the spill point. This can be done by the use of directionally drilled deviated Wells that inject the CO2 at the flanks of the reservoir. The leaking fault scenario indicates moderate CO2 fluxes and CO2 concentrations in the shallow overburden. Modelled maximum surface fluxes are comparable to observed leakage rates from natural CO2 accumulations in Europe and Australia. The maximum concentrations may lead to adverse effects in groundwater and freshwater ecosystems. The critical safety factor is the vertical permeability of the fault zone. It is important to stress that the probability of the above-mentioned four scenarios actually occurring has not been assessed. They represent hypothetical 'worst-case' situations that may well have a very low probability (Chadwick et al., 2008). It is obvious that the storage project in Schwarze Pumpe - Schweinrich tries to go quantitative.