u Introduction

u Figures

u Aquifers

u Target formation

u Model

u Conclusions

u References

u Introduction

u Figures

u Aquifers

u Target formation

u Model

u Conclusions

u References

u Introduction

u Figures

u Aquifers

u Target formation

u Model

u Conclusions

u References

u Introduction

u Figures

u Aquifers

u Target formation

u Model

u Conclusions

u References

u Introduction

u Figures

u Aquifers

u Target formation

u Model

u Conclusions

u References

u Introduction

u Figures

u Aquifers

u Target formation

u Model

u Conclusions

u References

u Introduction

u Figures

u Aquifers

u Target formation

u Model

u Conclusions

u References

  Figure and Table Captions

Screening of the Aquifer Systems

The initial step in the Washburn study consisted of collecting available data relevant to the hydrogoelogical characteristics of regional saline aquifer systems; oil, gas and water well data; and existing geographic information system (GIS) map data. In this way, evaluation of three major aquifer systems, which were identified as being potentially suitable for CO₂ storage, was conducted. The systems are the Lower Cretaceous aquifer system, the Permian-Pennsylvanian Minnelusa Group, and the Mississippian Madison Group. All of the identified systems have high potential to store CO₂ because of their significant areal extent, competent cap rocks and substantial porosity and permeability. Thus, reconnaissance level evaluation of the systems was conducted (Sorensen et. el., 2005). The evaluation consisted of collecting geological, hydrological and petrophysical data. Digital maps of the aquifers, specifically structural and thickness-porosity maps were created and allowed for better understanding of the most promising injection zones. However, the initial reconnaissance screening did not include creation of a detailed geologic model of the systems.

Choosing the Primary Target Formation

The reconnaissance level study has indicated that the Broom Creek Formation within the Minnelusa Group has the highest thickness of porous reservoir rock of all the studied aquifer systems in the Washburn area with relatively few interbeds of lower permeability rocks. This property of the formation combined with favorable petrophysical properties, appropriate formation water chemistry, and a competent seal provided by the directly overlying Opeche Shale, makes the Broom Creek Formation the best target horizon for large-scale saline aquifer storage of CO₂ in the central part of the Williston Basin. Thus, a more detailed study was conducted for selected parts of the aquifer. The study followed the best practices for creating a petrophysical model of a geologic system commonly used in the oil and gas industry for hydrocarbon exploration andproduction planning. The exact locations for the case studies are chosen based on the proximity to the major sources of CO₂ and favorable reservoir properties. Figure 1 shows the locations of the sources and the study area.

Creating a Petrophysical Model of the System

Several areas within the Broom Creek Formation were studied in detail. The methodology used is illustrated by an example characterizing the Washburn study area. As indicated by Figure 1 major sources of CO₂ are located in the area. The sources include six power generating and three fuel processing facilities. After reviewing geophysical data available for the area the study focused on two sites, a northwestern (NW) and a southeastern (SE), as being most appropriate for CO₂ injection. The sites are shown in Figure 2 . They were selected because of the following characteristics:

· combination of thick reservoir with good porosity

· relatively dense grid of wells serving as sources of geophysical log data

· availability of core data.

For the NW site an area of approximately 182 square miles was characterized using log data from 10 wells. The following types of log data were available: gamma ray, resistivity, porosity and compensated neutron density. Figure 3 presents a stratigraphic sequence of rocks within the Minnelusa Group based on logs. As illustrated in Figure 3 , the reservoir rock consists of Upper and Lower Broom Creek sandstones. The reservoir rock is capped by low permeability shales, siltstones and limestones of the Spearfish and Opeche formations. Nearly impermeable anhydrite of the Amsden Formation underlies the targeted injection zone. Wire-line logs from the 10 wells in the study area were correlated. Petrel software provided by Schlumberger was used in all steps of the model building process. Structure maps for the top of each zone shown in Figure 3 and for the base of the Lower Amsden unit were created and then utilized for building the frame of the 3D geologic model of the Broom Creek aquifer system. Internal layers were then added to capture the variability in reservoir properties evidenced in the well logs. The complete model had about 410,000 cells with dimensions of 1000 feet in the two horizontal directions. Average cell thickness was 12 feet and ranged from 5 to 23 feet.

The next stage of model development was populating the geologic model with petrophysical properties. Given the sparse well data (ten wells in 182 square miles), to ensure that uncertainties in reservoir properties were adequately captured, the decision was made to populate the model with low, mid, and high case estimates of porosity and permeability. Broom Creek core data was available from one well, the ANG #1, in the Washburn study area. Log data on rock porosity was correlated to measured core porosity data (Figure 4 ) from this well. The correlation allowed for creating an empirical relationship (1) between the porosities.

(1)   φ_core = 0.3427φ_log + 13.696

This correlation indicates that for log derived porosity φ_log greater than 20%, log porosity is more optimistic than measured core porosity φ_core. For the Broom Creek Formation in the base case model, porosity was first distributed in the model by SGS (sequential Gaussian simulation). A second step was then undertaken, when the resulting porosity in the model exceeded 20%, equation (1) was used to recalculate the porosity in that cell. Two other models were created: low and high case porosity models. On average low case porosity was 15% lower and high case porosity was 7% higher than that of the base case. Probability distribution function for the porosities was obtained empirically from geophysical logs and core data.

Core porosity and permeability data for the Spearfish and Broom Creek formations were analyzed to derive the relationship between rock porosity φ_core and permeability k. The cross plots of porosity and permeability along with the suggested trend lines are shown in Figure 5a,b.

The trend lines shown in Figures 5a and 5b represent high case (blue lines), mid case (black lines) and low case (green lines) permeability approximations. The empirical formulae for the trend lines are given in Table 1 . No core data is available for the Opeche or Amsden formations. Therefore permeability of the Opeche Formation was assumed to be similar to that of the Spearfish Formation and the low, mid and high case permeability transforms developed for the Spearfish Formation were also used for the Opeche. For the upper anhydrate zone within the Amsden Formation low, mid and high permeability values of 0.001md, 0.1md 0.5 md respectively were assigned. Permeability in the remaining Amsden layers was assigned using the transforms derived for the Spearfish Formation. Permeability distribution within the model wasspecified by applying the transforms to the values of porosity which have been prescribed earlier. Figure 6 illustrates the resulting mid case permeability distribution within the model.

Broom Creek Formation pressure p (psi) and temperature T (F) distributions within the model were assigned by calculating the measured depth D of each cell and applying standard gradients in the area. The formulae for pressure and temperature are:

p = 14.7 + 0.4616D,

T = 43.5 + 0.0123D

The final step in the creation of the model was the determination of water saturations and salinity from resistivity logs. The following commonly used formula (e.g. [Hilchie, 1982]) was used to estimate water resistivity R_w from resistivity R_t measured by wire-line logging.

R_w = 1.23R_tφ²

Estimated resistivity of water was further used to estimate irreducible water saturation S_wirr and the salinity s_w of water. Usually the salinity of water is determined from graphs (e.g. [Hilchie, 1982]). However, the numerical model required an analytical relationship. Analysis of graphs presented in Hilchie (1982) yielded the following relationship

(2) S_w = 4*10⁵T^-1.0145R_w^-1.1

The following formula modified from (Haro, 2004) served for prescribing irreducible water saturation.

(3) S_wirr = 1/φ√(R/R_tT

To populate the grid with salinity and irreducible water saturation, the values for each well were calculated using formulae (2, 3) and then interpolated throughout the grid.

The resulting static model was then used to estimate the Broom Creek pore volume available for CO₂ storage and the volume of CO₂ that can be injected into the Broom Creek Formation in the NW site of the Washburn study area. Table 2 presents the results for the three considered cases: low, mid and high.

Thickness-porosity map for the whole Washburn area (Figure 2 ) served for the estimation of the amount of CO₂ that might be stored in the area. Pressure and temperature data allowed for calculating CO₂ density ρ at reservoir conditions and the total mass Q_M of stored CO₂

Q_M = V_pore*ρ_CO2E

where E = 4% is the efficiency factor (RCSP, 2006). Integration of Q over the area has yielded the total amount of CO₂ to be 5.2Gt. Figure 7 illustrates the storage potential in the Washburn area.

Conclusions

• Rocks of the Minnelusa Group in the central part of the Williston Basin provide environment appropriate for storage of big amounts of CO₂. The amount of CO₂ that can be stored in the Washburn area is estimated to be 5.2 billion tons.
• The experience with creating the model indicated that data from existing oil and gas wells can be used for the thorough characterization of deep saline aquifers.
• The proposed approach allows for bottom-up analysis of the pore system; estimation of the pore volume available for CO₂ storage and of the amount of CO₂ that can be stored in the volume.
• The geologic model created using the described approach can be modified and serve for reservoir simulations and predicting the fate of the injected CO₂.

References

Bachu, S., and J.J. Adams, 2003, Sequestration of CO₂ in geological media in response to climate change: Capacity of deep saline aquifers to sequester CO₂ in solution: Energy Conversion and Management, v, 44, p.3151-3175.

IPCC, 2005, Special report on carbon dioxide capture and storage: Chapter 5, Underground Geological Storage, p. 5-1 - 5-134.

Sorensen, J. A., M.D. Jensen, S.A. Smith, D.W. Fisher, E.N. Steadman, and J.A. Harju, 2005, Geologic sequestration potential of the PCOR partnership region: PCOR Report, 22 p.

Haro, C.F., 2004, The perfect permeability transform using logs and cores: SPE paper 89516.

Hilchie, D.W., 1982, Advanced Well Log Interpretation: Douglas W. Hilchie, Inc.

Regional Carbon Sequestration Program Geologic Working Group, 2006, Draft of National Geological Carbon Sequestration Capacity Assessment.

 Return to top.