A Technical Feasibility Assessment Using Reservoir Simulation for CO₂ Storage in Sarmatian Formations of the Getic Platform, Romania

Abstract

Carbon capture and storage (CCS) represents a critical technology for achieving climate neutrality targets, particularly for regions with significant industrial CO₂ emissions. This study presents a comprehensive numerical simulation assessment of CO₂ geological storage potential in the Sarmatian formations of the Getic Platform, Romania, located near the Turceni power plant—one of Europe’s largest thermal power facilities. Using ECLIPSE 300 compositional simulator with the CO2STORE option, we developed reservoir dynamic models incorporating geological properties, fluid characteristics, and pressure–volume–temperature (PVT) data specific to the Sarmatian aquifer system. Multiple injection scenarios were evaluated, including configurations with 3, 4, and 5 injection wells at varying inter-well distances (2000–10,000 m). The simulations covered a 20-year injection period followed by 300 years of monitoring. While previous assessments have provided static capacity estimates for Sarmatian formations, this study presents the first dynamic simulation-based evaluation of multi-well injection scenarios and long-term CO₂ trapping behavior in this geological setting, directly linked to the Turceni Power Plant emissions profile. Results demonstrate that the study area (Zone V) can accommodate the target CO₂ injection rate of 2.07 × 10⁶ Sm³/day using five injection wells, with final reservoir pressure increasing only 7–9 bar above initial conditions, well below fracture pressure thresholds (~280 bar). Long-term simulations reveal favorable CO₂ trapping behavior, with significant portions immobilized through residual and dissolution trapping mechanisms. The static storage capacity was estimated at 2.44 × 10¹⁴ kg CO₂. These findings support the technical feasibility of large-scale CO₂ storage in Romanian Sarmatian formations, providing quantitative evidence for CCS implementation strategies in the region.

1. Introduction

The escalating atmospheric concentration of carbon dioxide (CO₂) and its contribution to global climate change represents one of humanity’s most pressing environmental challenges [1,2]. The Intergovernmental Panel on Climate Change (IPCC) has identified Carbon Capture, Utilization, and Storage (CCUS) as an essential technology for achieving climate neutrality, projecting it to contribute approximately 15% of cumulative emission reductions necessary for Net Zero objectives [3]. According to recent analyses, fossil fuels continue to account for approximately 80% of global energy needs, contributing roughly 89% of the 3.63 billion tons of annual CO₂ emissions [4]. This reality underscores the urgent need for effective carbon management strategies that can bridge the transition to renewable energy systems while maintaining energy security.

Romania faces particular challenges in its decarbonization pathway, with significant CO₂ emissions from thermal power generation concentrated in the Oltenia region. The Turceni Power Plant, located in Gorj County on the banks of the Jiu River, ranks among Europe’s largest lignite-fired thermal power facilities with historical installed capacity of 2310 MW [5]. In 2019, Turceni Power Plant ranked third in Romania for CO₂ emissions, generating approximately 3.3 MtCO₂ annually. The proximity of substantial CO₂ emission sources to potential geological storage formations creates favorable conditions for CCS implementation [6].

The Getic Platform, a major geological province in southern Romania, contains extensive Miocene–Pliocene sedimentary sequences including the Sarmatian formations that offer promising CO₂ storage potential [7]. These formations, characterized by thick aquifer intervals with favorable porosity and permeability characteristics, represent untapped capacity for geological carbon sequestration. The platform’s structural configuration, with sedimentary cover resting on Precambrian crystalline basement, provides multiple potential reservoir–seal combinations suitable for long-term CO₂ containment [8].

This study presents a comprehensive numerical simulation assessment of CO₂ storage feasibility in the Sarmatian formations of the Getic Platform (designated Zone V or Area V in the Romanian CCS Demo Project framework, as presented in Figure 1). Using the ECLIPSE 300 (Schlumberger, Slb presently, Houston, USA, version 2014.1) compositional simulator with the specialized CO2STORE option, we evaluate injection scenarios, storage capacity, and long-term CO₂ behavior to provide quantitative technical support for CCS implementation in Romania.

The geological storage of CO₂ represents a mature technology with demonstrated success across multiple large-scale projects worldwide. The history of CCUS dates back to the 1970s, with the first large-scale project beginning operation in 1972 at the Kelly Snyder oil field in West Texas, USA [4]. The 1996 Norway Sleipner gas field project marked a significant milestone in dedicated CO₂ storage, successfully storing nearly one million tons of CO₂ annually in deep saline formations [10]. As of 2022, there are 196 commercial CCUS infrastructure projects worldwide at varying stages of development, with an average CO₂ capture capacity of 243.9 Mt/a, although only 30 projects are currently operational [11].

CO₂ geological storage relies on multiple trapping mechanisms that evolve over different time scales. Structural and stratigraphic trapping provides immediate physical containment beneath impermeable cap rocks. Residual trapping immobilizes CO₂ in pore spaces through capillary forces. Solubility trapping occurs as CO₂ dissolves in formation fluids, while mineral trapping represents the most permanent form through conversion to stable carbonate minerals over geological time scales [3,12]. Volpi et al. [13] demonstrated the importance of comprehensive geophysical characterization for storage site assessment, integrating high-resolution seismic surveys, vertical seismic profiles, and electrical resistivity tomography to optimize injection well placement and identify potential leakage pathways.

Injectivity considerations significantly impact CCS project feasibility. Yusof et al. [14] experimentally quantified CO₂ injectivity impairment induced by salt precipitation and fines migration, finding direct correlations between brine salinity and injectivity reduction—with impairment ranging from 6% at 6000 ppm to 27.3% at 100,000 ppm salinity. Zoeir et al. [15] developed pore-scale simulations demonstrating that internal media structure, pore-to-throat ratios, and contact angle variations critically influence residual CO₂ saturation and trapping efficiency.

Research on CO₂ storage in Eastern European geological contexts has expanded significantly. Studies in the Transylvanian Basin have identified depleted gas fields with total CO₂ storage capacity potential of approximately 2.30 Gt [6]. Multiple assessments have covered fields across South Romania, including sites in the Getic Basin (Bradu-Albota, Silistea, Babeni, Balteni), the Moesian Platform, and the Pannonian Basin [16]. Cormos and Dinca [17] evaluated post-combustion capture technologies for Romanian power plants, demonstrating technical feasibility of integrating CCS with existing thermal generation infrastructure. The Sarmatian formations, characterized by significant thickness variations (24–130 m for individual complexes) and porosity ranging from 2.3% to 24%, present favorable characteristics for CO₂ storage [7,18].

Numerical simulation methodologies for CO₂ storage assessment have evolved substantially. The ECLIPSE 300 compositional simulator with CO2STORE option represents an industry-standard tool specifically designed for modeling CO₂-brine-rock systems under typical storage conditions (12–100 °C and up to 600 bar) [19]. The CO2STORE module calculates mutual solubilities of CO₂ and H₂O following the Spycher and Pruess procedure, based on fugacity equilibration between phases, with water fugacity obtained by Henry’s law and CO₂ fugacity calculated using a modified Redlich–Kwong equation of state [20]. This approach enables accurate representation of phase behavior including CO₂-rich gas phase, H₂O-rich liquid phase, and potential solid phase precipitation involving salts (NaCl, CaCl₂, CaCO₃).

Li et al. [21] investigated CO₂ sequestration in low-permeability reservoirs through CO₂-EOR operations, finding storage capacities of 0.003–1.16 kg/m³ with positive net carbon storage achievable when reservoir oil gravity exceeds 35° API. Paltsev et al. [22] modeled CCUS deployment scenarios projecting 38–76 Gt CO₂ cumulative mitigation by 2100, with hard-to-abate industrial sectors representing primary near-term opportunities. These studies collectively inform the simulation approach and validation framework applied in the present investigation.

Policy frameworks significantly influence CCS deployment trajectories. Terjanika et al. [23] analyzed EU member states’ legislative frameworks for CCUS, identifying Sweden, Croatia, and Belgium as leaders with comprehensive policy scores. The European Union’s climate neutrality objectives and evolving carbon pricing mechanisms under the EU Emissions Trading System create increasingly favorable economic conditions for CCS implementation [24,25]. Romania’s strategic position, with over 30 potential storage sites in depleted hydrocarbon reservoirs offering combined capacity exceeding 514 Mt CO₂, positions the country advantageously for CCS development [6].

2. Materials and Methods

2.1. Study Area and Geological Setting

The study area (designated Area V) is located on the Getic Platform in southwestern Romania, approximately 50 km from the Turceni Power Plant. The Getic Platform, belonging to the larger Moesian Platform (Figure 2), consists of thick Mio-Pliocene, Mesozoic and Paleozoic sedimentary cover resting on the Precambrian crystalline schist basement [7,8]. The target formations comprise the Sarmatian aquifer system, which includes multiple reservoir horizons, representing vertically stacked storage opportunities within a thick sedimentary sequence.

Based on the analysis of available data, the deep-saline formations of Sarmatian age in the Getic Platform were selected for investigation and modeling the potential of CO₂ storage capacity. As per Figure 3 and Figure 4, this formation presents good reservoirs at depths ranging from 1000 to 3000 m, consisting of sandstones, conglomerates and sands alternating with clays. The reservoirs have good porosities and permeabilities and are protected by intra-Sarmatian and overlying Sarmatian clay formations.

In the Getic Platform, the Sarmatian formations (displaying thicknesses from 0 to 3000 m) generally maintain a conformable relationship with the underlying Badenian, though it locally overlies mid-Cretaceous tectonic units unconformably. Apart from the region situated west of the Timoc Fault, the upper boundary shows stratigraphic continuity with the Meotian in nearly all locations. In the Eastern sector, the formation either lies conformably upon the Badenian or rests discordantly over Upper Burdigalian deposits, while its contact with the overlying Meotian remains conformable everywhere. Proximal facies of Lower Sarmatian consist of coarse clastic rocks with occasional sandstone and mudstone layers to the north, along with localized skeletal carbonates. To the south, across the Central and Western sectors, this basal sequence grades laterally into a succession of sands and dark marlstones, reaching thicknesses of 1000 m, which were deposited without interruption over Badenian marls. The upper sequence (Upper Sarmatian) is characterized by laminated marlstones in the Western sector, transitioning into silt-rich arenites toward the east. In the Eastern sector, thick sandy units and paraconglomerates, interbedded with marlstone and tuffaceous beds, represent another set of strongly transgressive deposits. In the Bâlteni–Ţicleni subsurface area, south of the Pericarpathian Fault, the Sarmatian has been subdivided into four distinct members: (1) a basal unit of syntectonic sands; (2) a lower member composed of sandy marlstones; (3) a middle sandstone unit (up to 600 m thick) which is post-tectonic and becomes unconformable toward the south; and (4) an upper unit of dark shales (50–500 m) exhibiting source rock potential [8,27,28,29,30].

The Sarmatian formations in the study area exhibit favorable petrophysical properties for CO₂ storage. The effective porosity, determined as the product of total porosity and net-to-gross ratio (NTG = 0.7), ranges from 3% to 21% with an average of 12%. Horizontal permeability, derived from core data on the neighboring Piscuri-Hurezani structure, ranges from 0 to 50 mD, with typical values between 1.16 and 45 mD. The distributions of these properties are presented in Figure 5. The total thickness of the interest zone is approximately 300 m, providing substantial pore volume for CO₂ accommodation [24].

The petrophysical model was validated through history-matched porosity-permeability relationships from the neighboring Piscuri-Hurezani structure, which shares the same Sarmatian depositional system. Where direct core measurements were unavailable, porosity-permeability transforms derived from these analog data were applied. We acknowledge that the absence of core data from within the study area itself represents a source of uncertainty; sensitivity analyses on key parameters (injection pressure coefficients of 0.80, 0.85, and 0.90; well spacing of 2000–10,000 m) were therefore conducted to bound the expected performance envelope, as discussed in Section 3.

Formation water chemistry reflects typical saline aquifer conditions, with NaCl comprising 7.5% and CaCl₂ comprising 1.5% of dissolved salts. These salinity levels were incorporated into the simulation model through the CO2STORE SOLID option, which accounts for salt precipitation during CO₂ injection and its potential impact on injectivity. Initial water saturation throughout the formation is 100%, confirming the aquifer nature of the target storage complex [24].

2.2. Reservoir Pressure and Temperature Characterization

In the absence of direct pressure measurements in the Sarmatian formations (which have not been targets for hydrocarbon exploitation), reservoir pressure (Pp) was calculated using a hydrostatic gradient approach. The pressure gradient (Gp) was established at 0.11–0.12 bar/m based on correlation with adjacent oil structures, yielding the relationship Pp = Gp × H, where H represents depth. At the average reservoir depth of approximately 2000 m, this corresponds to reservoir pressures of 156–222 bar depending on specific well locations [24].

Temperature distribution was modeled using a geothermal gradient approach, with a surface temperature (T₀) of 10 °C and depth-dependent variation calibrated to measurements of 28 °C at 600 m and 130 °C at 4000 m depth, yielding a gradient of approximately 3 °C/100 m. Fracture pressure gradient was estimated at Gf = 0.18 bar/m using the relationship Gf = Gp + (Glit − Gp) × (ν/(1 − ν)), incorporating lithostatic gradient (Glit = 2.1–2.18 bar/10 m) and Poisson coefficient (ν = 0.4–0.44) data from local statistical compilations. Maximum injection pressure was constrained to 0.85 × Gf × H to maintain safe operating margins below fracture initiation thresholds, resulting in bottom-hole pressure limits of 217–309 bar depending on well depth as per

Table 1. Injection well parameters for the five-injector base case scenario.

Well	Top Perforation (m)	Injection Pressure (bar)	Reservoir Pressure (bar)
INJ1N	1916	293	211
INJ2N	1417	217	156
INJ4N	1774	271	195
INJ5N	2019	309	222

[24].

In

Table 2. Summary of essential geological and fluid model parameters: data sources, assumptions, and uncertainty ranges.

Parameter	Value(s) Used	Data Source	Key Assumption	Uncertainty Range/ Comment
Porosity (effective)	3–21% (avg. 12%)	Well logs from study area; core data from neighboring Piscuri-Hurezani structure	NTG = 0.711 applied uniformly	±2–3 porosity units; no core from within study area itself
Horizontal permeability	0–50 mD (typical 1.16–45 mD)	Core data from Piscuri-Hurezani structure	Porosity-permeability transform applied where core absent	Order-of-magnitude uncertainty where only log-derived
Reservoir pressure (Pp)	156–222 bar (at ~2000 m)	Calculated via hydrostatic gradient (Gp = 0.11–0.12 bar/m)	Gradient calibrated from adjacent oil structures	No direct pressure measurements in Sarmatian formations
Water compressibility (Cw)	4.5 × 10−5 bar−1	Standard brine correlation	Uniform value applied	±10%
Brine salinity	NaCl 7.5%; CaCl2 1.5%	Analog field data	Uniform composition assumed	No formation-specific samples
CO2 density (ρCO2)	710 kg/m3	PVT model (ECLIPSE 300)	At ~200 bar, ~70 °C	±5%

below, a summary of the essential geological and fluid model parameters is presented: data sources, assumptions, and uncertainty ranges.

2.3. Simulation Software and Modeling Approach

Dynamic reservoir simulations were performed using ECLIPSE 300, a compositional numerical simulator developed by Schlumberger, with the specialized CO2STORE option activated [18,19]. This configuration enables accurate modeling of CO₂-H₂O mutual solubilities under typical storage conditions, following the thermodynamic framework of Spycher and Pruess. The CO2STORE module considers three phases: a CO₂-rich phase (labeled gas), an H₂O-rich phase (labeled liquid/water), and a solid phase for salt precipitation. Fluid components specified using the COMPS keyword include CO₂, H₂O, NaCl, and CaCl₂.

The static geological model was constructed in Petrel 2014.2 software and exported directly to ECLIPSE format, including grid coordinates (COORD), corner-point depths (ZCORN), effective porosity (PORO), horizontal permeability (PERMX/PERMY), and active cell indicators (ACTNUM). The dynamic model comprises 74 simulation layers across three major vertical zones, with vertical resolution varying from 0.25 m to 15.24 m. Global grid dimensions of 2000 × 2000 m were employed, with local grid refinement (LGR) ranging from 100 × 100 m to 500 × 500 m near injection wells to capture detailed CO₂ plume behavior [24].

Relative permeability functions for the water–CO₂ system were derived from literature correlations appropriate for sandstone formations, as per

Table 3. Relative permeability data for water–CO₂ system.

Sw	Krw	Sg	Krg
0.30	0.0000	0.00	0.0000
0.46	0.0024	0.16	0.0004
0.61	0.0390	0.31	0.0241
0.77	0.1975	0.47	0.1406
0.92	0.6243	0.62	0.4848
1.00	1.0000	0.70	1.0000

. The water relative permeability curve spans from Krw = 0 at connate water saturation (Sw = 0.30) to Krw = 1.0 at Sw = 1.0, while gas relative permeability increases from Krg = 0 at Sg = 0 to Krg = 1.0 at Sg = 0.70. Rock compressibility was determined from Hall’s and van der Knaap’s correlations for unconsolidated sandstones, yielding Cr = 5 × 10⁻⁵ bar⁻¹ at 120 bar reference pressure based on mean porosity of 22.7% and burial depth of 800 m. Water compressibility was set at Cw = 4.5 × 10⁻⁵ bar⁻¹ [24].

2.4. Simulation Scenarios

Multiple injection scenarios were designed to evaluate system performance under varying configurations. As per Figure 6, the primary scenarios include (1) CO2_INJ_3_AREA_V with 3 injection wells; (2) CO2_INJ_4_AREA_V with 4 injection wells; (3) CO2_INJ_5_AREA_V with 5 injection wells at approximately 10,000 m spacing; (4) CO2_INJ_5_AREA_V_5K with 5 wells at 5000 m spacing; (5) CO2_INJ_5_AREA_V_2K with 5 wells at 2500 m spacing; and (6) CO2_INJ_5_AREA_V_PSEUDO_OB incorporating a horizontal pseudo-production well to simulate open boundary conditions [24].

All scenarios employed a target CO₂ injection rate of 2.07 × 10⁶ Sm³/day (equivalent to approximately 3.3 Mt/year), corresponding to the annual CO₂ emissions from Turceni Power Plant. The simulation timeline comprises injection starting 1 January 2025, continuing for 20 years through 2045, followed by a 300-year monitoring period extending to 2345. This extended timeframe enables assessment of long-term CO₂ fate including pressure dissipation, plume stabilization, and trapping mechanism evolution [24].

3. Results

3.1. Injection Performance and Well Behavior

The simulation results demonstrate that the target CO₂ injection rate can be successfully achieved using 5 injection wells (scenarios CO2_INJ_5_AREA_V and variants), while configurations with 3 or 4 wells proved insufficient to accommodate the full injection volume.

Table 4. Summary of dynamic simulation results for five-injector scenarios.

Scenario	Year	Injection Pressure (bar)	Field Pressure (bar)	Injection Rate (106 Sm3/d)
CO2_INJ_5_AREA_V_2K	2025	254–282	228.0	0.22–0.64
CO2_INJ_5_AREA_V_2K	2045	259–292	235.9	0.15–0.75
CO2_INJ_5_AREA_V_2K	2345	0	235.7	0
CO2_INJ_5_AREA_V_5K	2025	214–263	228	0.22–0.59
CO2_INJ_5_AREA_V_5K	2045	231–284	236.9	0.15–0.68
CO2_INJ_5_AREA_V_5K	2345	0	236.6	0
CO2_INJ_5_AREA_V	2025	180–258	228.0	0.23–0.69
CO2_INJ_5_AREA_V	2045	173–244	237.1	0.14–0.85
CO2_INJ_5_AREA_V	2345	0	236.9	0

and Figure 7 summarize the dynamic parameters across the primary five-well scenarios at key time points during the injection (2025, 2045) and post-injection monitoring (2345) periods.

The field pressure evolution shows a modest increase of approximately 7–9 bar over the 20-year injection period, from initial values of 227–228 bar to final injection-phase pressures of 235–237 bar.

This pressure buildup remains well below the estimated minimum fracture pressure of 280 bar (calculated for the shallowest perforation at 1635 m depth), confirming safe operating conditions throughout the injection program.

Moreover, it can be noticed that between years 2045 and 2345 the field pressure drops slightly due to the different trapping mechanisms (solubility and mineral) which are still active. The reservoir’s large pore volume (total pore volume at reference pressure: 44 × 10⁶ RM³) provides substantial accommodation capacity for the injected CO₂ volume [24].

Injectivity indices varied among wells depending on their structural position and local reservoir quality. Well INJ2N, located at the shallowest depth (top perforation 1417 m), exhibited the lowest injectivity indices across multiple scenarios, while wells in structurally deeper positions (INJ4N at 2019 m) showed superior injectivity.

Notably, as per Figure 7, the sensitivity analysis revealed that inter-well distance (ranging from 2000 m to 10,000 m) did not significantly influence individual well injectivity indices, although it affected injection pressure distributions and CO₂ plume overlap patterns as in Figure 8 and Figure 9 [24].

Figure 8 shows the spatial CO₂ saturation distribution for the primary five-well scenarios of wells spacing 2 km, well spacing 5 km and well spacing 10 km at key time points during the injection (2025, 2045) and post-injection monitoring (2345). The view presented has a more areal focus.

If the cells fully saturated with water are filtered out (cells colored in magenta), with a condition such as if CO₂ saturation > 0.01, then the true spatial CO₂ saturation distribution, or the CO₂ plume, can be presented, and this is what is shown in Figure 9 for the same five-well scenarios and key time points of the injection (2025, 2045 and 2345).

One can notice that there are notable differences between the distributions in lines 2 and 3 of Figure 9 above, respectively between years 2045 and 2345. There are two main significant reasons for this to happen. The first is gravitational segregation: CO₂ component is lighter than water and it will rise spatially. The latter is the solubility of the CO₂ component in water.

Analysis of Well Configuration Performance: The superior performance of the 5-well configuration—and the failure of 3- and 4-well arrangements to accommodate the full target injection rate—can be understood through the interplay of three physical mechanisms: injectivity limits, pressure interference, and plume coverage.

Injectivity constraints in the 3- and 4-well configurations: The target injection rate of 2.07 × 10⁶ Sm³/day corresponds to approximately 3.3 MtCO₂/year. With only 3 or 4 wells, the per-well injection rate required to meet this target exceeds the local injectivity capacity of the Sarmatian formations at the available pressure differential (i.e., the margin between reservoir pressure and the fracture pressure limit). This is particularly acute for wells in structurally shallower positions (e.g., INJ2N at 1417 m top perforation), where the lower reservoir pressure (156 bar) provides a narrower safe injection window. The five-well configuration distributes the total target rate across more injection points, reducing the per-well rate to within the formation’s injectivity envelope and keeping bottom-hole pressures below the 0.85 × Gf × H safety threshold at all locations.

Well spacing and pressure interference: The sensitivity analysis across inter-well distances of 2000 m, 5000 m, and 10,000 m revealed a notable result: individual well injectivity indices did not vary significantly with spacing. This counterintuitive finding reflects the large pore volume of the Sarmatian aquifer (44 × 10⁶ RM³ at reference pressure) and its high lateral connectivity, which together provide sufficient pressure dissipation even at closer well spacing. However, spacing did influence the spatial distribution of injection pressures and the degree of CO₂ plume overlap. At 2000 m spacing, pressure interference between adjacent wells produced locally elevated pressures (up to 292 bar at individual wells, compared to 244 bar at 10,000 m spacing), though these remained below fracture thresholds. At 10,000 m spacing, plumes developed largely independently during the injection phase, maximizing the contact area between CO₂ and formation water—a factor that likely contributes to the higher dissolved CO₂ fraction observed in the wider-spacing scenario (1.18 × 10⁸ kg-mol dissolved for 5 K spacing vs. 0.95 × 10⁸ kg-mol for 2 K spacing;

Table 5. CO₂ distribution among trapping mechanisms after 300-year monitoring period.

Scenario	Dissolved (kg-mol)	Trapped Gas (kg-mol)	Mobile Gas (kg-mol)
CO2_INJ_5_AREA_V	1.01 × 108	2.88 × 108	2.48 × 108
CO2_INJ_5_AREA_V_2K	0.95 × 108	2.55 × 108	2.14 × 108
CO2_INJ_5_AREA_V_5K	1.18 × 108	2.85 × 108	2.27 × 108

).

Plume migration and practical implications: The plume evolution patterns (Figure 8 and Figure 9) illustrate that wider well spacing promotes more distributed CO₂ placement, which is beneficial for both pressure management and trapping efficiency: greater CO₂–water contact area enhances dissolution trapping, while reduced plume overlap minimizes the risk of a connected mobile CO₂ phase accumulating at structural highs. From a project design perspective, these results suggest that well spacing should be optimized to balance three competing objectives: (i) minimizing surface footprint and pipeline infrastructure costs (favoring tighter spacing), (ii) maximizing per-well injectivity and minimizing pressure interference (favoring wider spacing), and (iii) promoting dissolution trapping through increased CO₂–water contact (favoring wider spacing). For the Sarmatian formations in Zone V, the 5000–10,000 m range appears to offer a favorable compromise, though site-specific factors—including surface access, existing infrastructure, and detailed structural mapping—would ultimately determine optimal well placement. More broadly, the integration of storage-side optimization (well count, spacing, and placement) with capture-side technology selection—such as the comparative evaluation of CO₂ separation pathways (e.g., adsorption-based vs. absorption-based capture) recently demonstrated by Mohamad et al. [31] for biomass-to-biogas systems—represents an important direction for holistic CCUS project design, ensuring that capture and storage subsystems are co-optimized rather than developed in isolation.

3.2. Storage Capacity Assessment

Storage capacity was evaluated at two distinct levels: theoretical static capacity and practically achievable dynamic capacity. The analytical static capacity calculation employed the volumetric approach: Static Capacity = VB × NTG × Φ × (1 − Swirr) × ρCO₂, where VB = 4.9 × 10¹² m³ (bulk volume from simulation), NTG = 0.711, Φ = 0.1232 (average porosity), Swirr = 0.2 (irreducible water saturation), and ρCO₂ = 710 kg/m³ (CO₂ density at reservoir conditions). This calculation yields a theoretical static storage capacity of 2.4 × 10¹⁴ kg CO₂ (244 Gt) [24]. This value represents the total pore volume available for CO₂ storage under ideal conditions, i.e., if the entire connected pore space within the modeled domain could be accessed and filled. It is best understood as an upper-bound resource estimate rather than a development-planning figure.

Practically achievable dynamic capacity was assessed through dynamic simulation over 100 years of continuous injection (extended scenario) and resulted in cumulative storage of 5.13 × 10¹⁰ Sm³ (equivalent to approximately 10.16 × 10¹⁰ kg CO₂, 102 Mt), representing a fraction (0.01–0.04%) of the theoretical static capacity. This difference reflects practical constraints including well placement, injection rate limitations, and pressure buildup considerations. For the 20-year base case injection period, the cumulative injected volume reached approximately 1.51 × 10¹⁰ Sm³ (roughly 3.0 × 10¹⁰ kg CO₂), demonstrating that the storage complex can accommodate decades of emissions from the Turceni Power Plant [24].

To contextualize these results against industrial emission volumes, the Turceni Power Plant emits approximately 3.3 MtCO₂ per year [5]. The 20-year dynamic simulation therefore accommodates the equivalent of approximately 9 years of Turceni’s current emissions, while the 100-year extended scenario accommodates approximately 31 years. It should be noted that these figures correspond to a single storage zone (Zone V); the Getic Platform contains six additional prospective zones identified in the Romanian CCS Demo Project framework (Figure 1), suggesting that the aggregate practical storage capacity across the platform could accommodate Turceni-scale emissions for well over a century.

At the national level, Romania’s estimated total CO₂ storage capacity in depleted hydrocarbon reservoirs and saline aquifers exceeds 514 Mt [6]. The practical dynamic capacity demonstrated in this study for Zone V alone (~30 Mt over 20 years, ~102 Mt over 100 years) represents a meaningful contribution to this national portfolio—approximately 6–20% of the total estimated national capacity from a single storage zone. These comparisons underscore the industrial relevance of the Sarmatian formations within Romania’s broader decarbonization strategy and highlight the value of a multi-zone, hub-and-cluster approach to CCS deployment in the Oltenia region.

3.3. Long-Term CO₂ Behavior and Trapping Mechanisms

The 300-year monitoring simulation provides insights into long-term CO₂ fate within the storage formation. Table 5 presents the distribution of CO₂ among different trapping mechanisms at the end of the monitoring period (year 2335). The results show significant portions of injected CO₂ immobilized through residual (trapped gas) and dissolution (dissolved in water) mechanisms, with remaining mobile gas phase CO₂ contained within the structural closure [24].

As per Figure 9, the CO₂ saturation plume evolution shows characteristic behavior consistent with literature observations [11,12]. During the injection phase, CO₂ migrates radially from injection wells, with plume geometry influenced by formation heterogeneity and structural dip. Post-injection, buoyancy-driven migration continues until the plume reaches structural highs or becomes immobilized through residual trapping. The scenario with wider well spacing (CO2_INJ_5_AREA_V_5K) showed greater dissolved CO₂ fractions (1.18 × 10⁸ kg-mol versus 0.95 × 10⁸ kg-mol for 2K spacing), likely reflecting increased contact area between CO₂ and formation water during migration [24].

Pressure dissipation during the monitoring period proceeded gradually, with field pressure stabilizing at 236–237 bar by year 2335, only marginally above initial conditions. This behavior indicates effective pressure communication with the broader aquifer system, supporting long-term containment without excessive pressure accumulation that could threaten caprock integrity. The pseudo-production well scenario (PSEUDO_OB) confirmed that even with open boundary conditions simulating pressure relief pathways, CO₂ plume migration remained contained within the modeled domain [24].

4. Discussion

4.1. General

This study provides a comprehensive quantitative assessment of the potential for permanent CO₂ storage in deep saline reservoirs of Sarmatian age located in the Getic Platform, Southern Romania. The assessment based on published geological information on the Getic Platform and numerical simulations includes (1) the stratigraphic and structural framework of the Sarmatian interval; (2) petrophysical characteristics of Sarmatian reservoirs in the region; (3) regional seal effectiveness and tectonic setting; (4) static and dynamic reservoir modeling; (5) determination of storage capacity; (6) analysis of the long-term CO₂ behavior and trapping mechanisms; and (7) injection simulation scenarios.

In the studied area, which includes a part of the western sector of the Moesian Platform, Sarmatian subsidence and clastic input were particularly strong, resulting in thick stacked deltaic and shallow marine successions, widespread development of high-quality arenitic reservoirs, and a high concentration of medium to large hydrocarbon accumulations. Moreover, this geological major unit accounts for a dominant share of the Getic Oil Province discovered gas and a significant proportion of its oil resources, which indirectly confirms the efficiency of the reservoir–seal systems.

In the Getic Platform, the Sarmatian is dominated by deltaic systems (upper and lower delta plain, delta front), shoreface and littoral facies, and locally shelf edge and shallow marine sand bodies. Reservoir rocks/formations occur as stacked sandstone and sandy conglomerate bodies (often grouped into numbered pay zones, Sa I to Sa X), channelized sandstones and distributary channels, mouth bars, shoreface bars and sheet-like bodies with variable lateral continuity. This depositional context results in laterally extensive but internally heterogeneous sandstone packages and vertical alternations of high-quality sand with marls and silty mudstones. Such architecture is generally favorable for CO₂ storage, allowing large, connected volumes for injection, multiple levels of internal baffling, and partial compartmentalization, which may help in pressure management and plume control.

Published regional data for Sarmatian reservoirs in the assessed region indicate the following:

○

Porosity: typically in the range 20–33% for Sarmatian sandstones, with local values up to 45–47% in high quality, weakly cemented sands.

○

Permeability: ranges from 23 mD to 2500 mD, with common values in the tens to hundreds of mD range. Very high permeabilities (hundreds–thousands of mD) are reported in certain delta front and shoreface sand bodies.

○

Productive thickness: individual Sarmatian pay zones commonly show 20–100 m cumulative net sandstone thickness at field scale (including multi-layer stacks).

These petrophysical ranges are highly favorable for CO₂ injection because high porosities ensure substantial storage capacity per unit rock volume, and high permeabilities and net thickness provide excellent injectivity, supporting industrial-scale CO₂ rates with less risk of excessive pressure build up.

In the Getic Platform, the primary seals for Sarmatian reservoirs are intra-Sarmatian marls and claystones, Pontian and Pliocene regional mudstone seals (up to 500–600 m), and additional Miocene seals. The combination of intraformational and regional seals results in a robust multi-seal system, which is a strong positive factor for long-term CO₂ containment. Regarding structural sealing and faults, these features rely heavily on faulted anticlines and combined structural–stratigraphic traps (e.g., erosional truncation, pinch outs against marls).

A key aspect relevant to CO₂ storage in the region is the existence of long-lived hydrocarbon accumulations, which shows that many of the bounding faults behave as effective sealing faults (at least under historical pressure regimes). However, CO₂ injection will increase pore pressure, so the reactivation potential of these faults must be assessed geomechanically by determination of the in situ stress field and fault orientation and calculation of fault slip tendency and fracture pressure thresholds.

The findings of this study align with observations from other CO₂ storage assessments in similar geological settings. The achieved storage efficiency (ratio of stored CO₂ volume to pore volume) falls within ranges reported for saline aquifer storage projects globally [10,21]. The modest pressure buildup relative to fracture thresholds compares favorably with experience at operational sites including Sleipner, where careful pressure management has maintained safe injection conditions over decades [9].

The trapping mechanism distribution observed—with significant residual and dissolution trapping contributions—reflects the physics-based modeling incorporated in the CO2STORE simulation framework. These results are consistent with laboratory and field observations indicating that 20–40% of injected CO₂ can become residually trapped during post-injection redistribution [13,14]. The contribution of dissolution trapping (approximately 15–20% of total stored CO₂) aligns with theoretical predictions for saline formations with similar salinity levels [4,11].

Compared to recent assessments of Romanian storage sites, including work on the Bibești-Bulbuceni structure in the Meotian formations [6], the Sarmatian formations evaluated here offer complementary storage capacity within the broader regional CCS infrastructure framework. The technical feasibility demonstrated supports consideration of multiple geological horizons for optimized storage network development, potentially enabling hub-and-cluster approaches connecting multiple emission sources to distributed storage complexes [22,23].

Several sources of uncertainty affect the simulation results and should inform interpretation. The primary data quality limitations include (1) petrophysical properties derived from well logs with limited calibration data, particularly the absence of core analysis from within the study area requiring reliance on analog data from neighboring structures; (2) absence of direct pressure measurements in the Sarmatian formations necessitating calculated pressure profiles; (3) relative permeability functions based on literature correlations rather than special core analysis; and (4) fluid property characterization from analog fields rather than formation-specific samples [24].

Sensitivity analyses conducted on injection pressure coefficients (0.80, 0.85, 0.90) and well spacing (2000–10,000 m) provide bounds on expected performance variability. The reservoir pressure response remained below fracture thresholds across all scenarios evaluated, suggesting robust safety margins despite input parameter uncertainties. For improved confidence in project planning, acquisition of site-specific data through appraisal drilling, including pressure measurements, core analysis, and fluid sampling, would be recommended as standard practice [24].

4.2. Long-Term Storage Security Considerations

While the simulation results demonstrate favorable pressure behavior and trapping efficiency, a critical assessment of long-term containment security must address four interrelated factors: caprock integrity, potential leakage pathways, geomechanical stability, and the inherent uncertainty of multi-century predictions.

Caprock integrity: The Sarmatian storage complex benefits from a robust multi-seal system comprising intra-Sarmatian marls and claystones, Pontian and Pliocene regional mudstone seals reaching 500–600 m thickness, and additional Miocene seals at low depth. The maximum simulated pressure increase of 7–9 bar above initial conditions (final field pressure 235–237 bar) remains well below the estimated minimum fracture pressure of ~280 bar, corresponding to a safety margin of approximately 43–45 bar. This margin is maintained across all scenarios tested, including the tighter well-spacing configurations (2000 m) that produced the highest local injection pressures (up to 292 bar at individual wells). However, we note that the fracture pressure estimate relies on a calculated gradient (Gf = 0.18 bar/m) rather than direct in situ stress measurements (e.g., leak-off tests or mini-frac data), which represents a source of uncertainty that should be addressed through appraisal well testing prior to any operational commitment.

Potential leakage pathways: Three categories of potential leakage pathways warrant consideration: (i) Fault-related leakage—The Getic Platform is characterized by faulted anticlines and combined structural–stratigraphic traps. The presence of long-lived hydrocarbon accumulations in the region provides indirect evidence that many bounding faults act as effective seals under natural pressure regimes. However, CO₂ injection will elevate pore pressure, and the reactivation potential of these faults has not been quantitatively assessed in this study. Determination of the in situ stress field (magnitude and orientation of principal stresses), fault orientation relative to the stress tensor, and calculation of slip tendency and fracture pressure thresholds would be required for a complete geomechanical risk assessment. (ii) Wellbore leakage—The injection wells themselves represent potential leakage pathways if not properly completed and abandoned. While outside the scope of this reservoir simulation study, well integrity—including casing design, cement placement, and post-injection abandonment procedures—constitutes a critical component of storage security that should be addressed in subsequent engineering studies. (iii) Caprock discontinuities—The lateral continuity of the Pontian–Pliocene regional seal has been inferred from seismic data, but sub-seismic-scale features (e.g., small-offset faults, fracture swarms) could provide undetected migration pathways. High-resolution seismic acquisition and attribute analysis would reduce this uncertainty.

Geomechanical stability: Beyond caprock fracturing, injection-induced pressure changes can trigger fault reactivation or induced seismicity. The modest pressure buildup observed in our simulations (ΔP < 10 bar field-wide) suggests a low risk of large-scale geomechanical failure, consistent with operational experience at sites such as Sleipner where similar pressure management has been maintained over decades [9]. Nevertheless, the absence of a coupled geomechanical model in this study means that stress arching, poroelastic effects, and potential shear failure along pre-existing faults have not been explicitly evaluated. We recommend that future work incorporate coupled reservoir–geomechanical simulation using the in situ stress data that would be acquired during appraisal drilling.

Uncertainty of 300-year predictions: The 300-year monitoring period simulated in this study extends well beyond the temporal scope over which reservoir models are typically validated. The primary sources of prediction uncertainty include (i) relative permeability functions derived from literature correlations rather than site-specific special core analysis (SCAL); (ii) the absence of direct pressure and fluid samples from the Sarmatian formations; (iii) the assumption of uniform brine chemistry, which affects dissolution trapping rates; and (iv) the omission of mineral trapping kinetics, which become increasingly significant over centennial-to-millennial timescales. The CO2STORE module captures residual and solubility trapping but does not simulate geochemical reactions (mineralization), meaning that the long-term immobilization predicted here may be conservative—mineral trapping would provide additional permanent storage capacity over geological time. Conversely, the assumption of an intact multi-seal system throughout the 300-year period is optimistic in the absence of a coupled geomechanical assessment. We therefore present these results as a technically grounded feasibility assessment rather than a definitive prediction, and we emphasize that progressive reduction of these uncertainties through site-specific data acquisition is a prerequisite for advancing to the storage development phase.

5. Conclusions

This study presents a comprehensive numerical simulation assessment of CO₂ geological storage feasibility in the Sarmatian formations of the Getic Platform, Romania. Using ECLIPSE 300 compositional simulator with the CO2STORE option, multiple injection scenarios were evaluated covering 20 years of injection and 300 years of post-injection monitoring. The key findings support the technical viability of large-scale CO₂ storage in this geological setting and provide quantitative parameters for project planning.

The principal conclusions are as follows:

• Storage Capacity: The Sarmatian formations offer substantial storage potential with static capacity estimated at 2.44 × 10¹⁴ kg CO₂. Dynamic simulations confirm that the 20-year target injection volume (approximately 3.0 × 10¹⁰ kg CO₂ at 2.07 × 10⁶ Sm³/day) can be successfully accommodated using five injection wells.
• Safe Operating Conditions: Reservoir pressure increase during injection remains modest (7–9 bar), well below the fracture pressure threshold of approximately 280 bar. This safety margin is maintained across all evaluated scenarios, supporting confidence in long-term containment integrity.
• Favorable Trapping Behavior: Long-term simulations demonstrate effective CO₂ immobilization through multiple trapping mechanisms. At the end of the 300-year monitoring period, significant fractions are immobilized as residually trapped gas (approximately 45%) and dissolved CO₂ (approximately 15–20%), with remaining mobile phase contained within structural closures.
• Well Configuration Flexibility: While five injection wells are required to achieve the target rate, spacing between wells (2000–10,000 m) can be optimized based on site-specific considerations without significantly impacting overall storage performance.
• Regional CCS Potential: The results support Romania’s potential as a significant player in European CCS development, with the Sarmatian formations representing one component of the country’s estimated 514 Mt CO₂ storage capacity in depleted reservoirs and saline aquifers.

Future work should focus on reducing key uncertainties through acquisition of site-specific data, particularly direct pressure measurements, core samples for petrophysical and geomechanical characterization, and formation fluid samples for accurate brine chemistry modeling. Integration with surface infrastructure planning, including capture technology selection and pipeline routing from Turceni Power Plant, would advance project development toward implementation readiness. The demonstrated technical feasibility positions this storage complex as a candidate for detailed appraisal in Romania’s decarbonization strategy.