Comparison of CO2 trapping in highly heterogeneous reservoirs with Brooks-Corey and van Genuchten type capillary pressure curves

Geological heterogeneities essentially affect the dynamics of a CO2 plume in subsurface environments. Previously we showed how the dynamics of a CO2 plume is influenced by the multi-scale stratal architecture in deep saline reservoirs. The results strongly suggest that representing small-scale features is critical to understanding capillary trapping processes. Here we present the result of simulation of CO2 trapping using two different conventional approaches, i.e. Brooks-Corey and van Genuchten, for the capillary pressure curves. We showed that capillary trapping and dissolution rates are very different for the Brooks-Corey and van Genuchten approaches when heterogeneity and hysteresis are both represented.

💡 Research Summary

The paper investigates how the choice of capillary pressure–CO₂ saturation relationship influences CO₂ trapping and dissolution in heterogeneous deep saline reservoirs. Two widely used empirical models, Brooks‑Corey (BC) and van Genuchten (vG), are compared through numerical simulations that incorporate realistic geological heterogeneity consisting of sandstone (76 % of the volume) and open‑framework conglomerate (OFC, 24 %). The reservoir model spans 100 m × 100 m × 5 m with a fine discretization (250 000 cells of 2 m × 2 m × 0.05 m). CO₂ is injected at 3.6 standard m³ day⁻¹ for ten days, delivering a total of 594 kg at a depth of 2360 m. All boundaries are set to no‑flow, and twelve characteristic curves (six drainage, six imbibition) are assigned to the two facies.

For the BC model the drainage capillary pressure follows the classic power‑law form 𝑃c = Pe (S − Swi)^(‑1/λ), where λ is the pore‑size distribution index, Pe the entry pressure, and Swi the irreducible water saturation. The vG model uses the same functional form for most of the saturation range but replaces it in the low‑saturation “tail” (1 − S ≤ S_nt) with a smoother expression that introduces an additional parameter S_nt, controlling the width of the entry‑slope region. Both models share the same λ = 0.55, Swi values (0.20 for sandstone, 0.14 for OFC), and Pe values (4.6 bar for sandstone, 0.72 bar for OFC). Imbibition curves are identical for BC and vG, so differences arise solely from the drainage portion.

Results are presented for homogeneous (single‑facies) and heterogeneous (two‑facies) reservoirs. In the homogeneous case, capillary trapping is virtually identical for BC and vG, confirming earlier findings that trapping is dominated by relative‑permeability hysteresis rather than the exact shape of the low‑saturation capillary curve. Dissolution, however, is about 20 % higher with the vG model because the smoother tail widens the CO₂‑brine interface, increasing the effective contact area.

In the heterogeneous reservoir the differences become pronounced. Overall capillary trapping is roughly 12 % larger with the vG model. The effect is facies‑specific: sandstone traps more than 60 % more CO₂ under vG, whereas OFC traps about 20 % less. This asymmetry is explained by the “secondary‑seal effect”: CO₂ migrates preferentially through high‑permeability OFC clusters but encounters capillary barriers at sandstone boundaries. The vG tail produces a broader CO₂‑brine transition zone, allowing more CO₂ to spread into sandstone and less to remain in OFC. Consequently, the total trapped volume increases.

Dissolution in the heterogeneous case is dramatically amplified. Using vG, dissolved CO₂ is nearly double that obtained with BC, and roughly three times the amount seen in a homogeneous reservoir with BC. The larger interface area at the OFC–sandstone contacts, combined with the wider low‑saturation tail, drives this increase. Mobile CO₂ in the gas phase becomes negligible for the vG case, while the BC case retains a substantial mobile fraction (the BC/ vG mobile CO₂ ratio exceeds 50 at the end of the simulation).

A sensitivity test on the tail width parameter S_nt shows that reducing S_nt from 0.1 to 0.04 slightly raises the mobile gas fraction and reduces trapped gas, but the vG model still outperforms BC in both trapping and dissolution. This confirms that even modest adjustments to the low‑saturation capillary curve can materially affect predictions.

The discussion emphasizes that capillary pressure curves at low CO₂ saturation are difficult to measure experimentally, yet the choice between BC and vG can lead to large uncertainties in storage performance estimates, especially in heterogeneous formations. The authors argue that reservoir‑scale modeling must incorporate facies‑specific capillary pressure functions and account for the secondary‑seal mechanism, rather than relying on averaged properties. Accurate characterization of the entry‑slope region (S_nt) is crucial for reliable forecasts of both capillary trapping and solubility trapping.

In conclusion, the study demonstrates that (1) capillary trapping is significantly sensitive to the capillary pressure model in heterogeneous reservoirs, with vG generally yielding higher overall trapping; (2) CO₂ dissolution is markedly enhanced when a vG tail is used, particularly because the broadened CO₂‑brine interface coincides with facies boundaries; and (3) proper representation of low‑saturation capillary behavior and facies‑specific hysteresis is essential for realistic predictions of CO₂ sequestration efficiency.