Ugjenkjennelig person som peker på et CO2 symbol. Foto.

28.09.2022

New method of assessing seal integrity early on in the screening phase

In a recently completed CLIMIT-Demo pilot project, Physiq.ai, represented by Torolf Wedberg, worked on a new method that can help to assess the integrity of sealing rocks, even in the exploration phase.

The preliminary results were positive, and a full CLIMIT-Demo application is being worked on to further develop these methods, which may also result in a direct contribution to the Norwegian Petroleum Directorate’s CO2 Storage Atlas.

The sealing capacity of caprocks is essential for carbon storage sites

The seal quality of a CO2 reservoir is a major question for the safety of a carbon storage site. There is a need to demonstrate sealing in an application for a storage licence. In general, it is difficult to show the quality of a seal in early phases, and if it then becomes clear that it is insufficient and that CO2 may leak out, significant investments and effort will have been wasted. That is why it is important to be able to screen the seal quality as best as possible in early stages in order to avoid disappointments down the line.

During the feasibility study phase, it is desirable to be able to assess pressure limits of potential carbon storage sites.

To summarise the pilot project, Wedberg said the following: “We tested an idea that can help to determine these pressure limits, and which can be specially adapted to areas relevant to carbon storage. An important part of our solution lies in utilising the mineralogical composition of the rocks. Physiq.ai has utilised rock cuttings to determine this in a reasonable and effective manner.”

The method makes use of existing data in a new way

The method is based on LoP (leak of pressure) data from the Norwegian Continental Shelf (NCS) which is accessible from npd.no. If you plot all this data against the depth, then it is possible to see that there are particular trends in the three main regions of the NCS, but also that there are large variations (Figure 1).

Figure 1: LoP from the North Sea (blue), the Norwegian Sea (orange) and the Barents Sea (green) plotted against depth. It shows that there are general trends from these plotted points, but also that there are significant variations at the same depth. This indicates that it is more than Sv (vertical stress) which affects Shmin (minimum horizontal stress).

Physiq.ai explored whether these variations could be explained by the quartz content of the rocks.

To test this hypothesis, wells from a geographically limited area were used to create a dataset with roughly the same geographic history. Figure 2 shows a plot for the Marflo Spur, where LoP is plotted against depth, but also includes the quartz concentration (size and colour of the plot markers).

Figure 2: Plot of LoT vs depth, with size and colour of the plot marker showing the quartz concentration at the test depth. The plotted points show a clear connection between LoT and quartz concentration in addition to the depth.

Based on the findings from Marflo Spur, three different models were developed to predict LoP for the selected area.

Only from Sv
Only from quartz concentration
From a combination of Sv and quartz concentration

Sv was chosen instead of depth (which was used in Figure 2) since there is a direct physical link to this size, and it was able to be used with greater success in shallower formations.

The three models were assessed with the help of R2 (a method for determining variation) and RMSE (root mean squared error). The results showed that method 3, a combination of Sv and quartz concentration gave an R2 of 0.82 and an RMSE of 0.06. This is a significant improvement in the results compared to methods 1 and 2. Other parameters were also tested but did not have as significant an impact as that shown by the quartz content.

The method was also tested at Tampen Spur and Halten Terrace to rule out that there was something specific going on at Marflo Spur. In these models, the combination of Sv and quartz concentration also led to significant improvement of R2 and RMSE.

The potential of this method is also shown in Figure 3 below.

*Figure 3: Measured well data vs predicted well data. From the left: lithostratigraphy, gamma ray (GR), density (DEN), deep resistivity (RDEP) and compressional sonic (AC), geomechanics and minerals.*

The second to last panel on the right, Geomechanics, is of the most interest here. The green line is the predicted Sv (from compaction trend) and the red is the predicted LoP (multilinear model). The blue dots are the measured LoP. As can be seen here, there is a strong correlation between the measured data and the predicted data.

– “The preliminary tests of this method are encouraging and provide us with motivation for further testing. If we are successful, it will contribute to a better understanding of the leakage dangers in the early exploration phase, something which reduces costs and thus makes carbon storage more commercially attractive. Hopefully it can be a small contribution to reducing the climate challenge facing our civilisation!” concludes Wedberg.