statistical modeling of subsurface temperatures in the ...€¦ · with the transfer of nw-trending...

7
PROCEEDINGS, 45 th Workshop on Geothermal Reservoir Engineering Stanford University, Stanford, California, February 10-12, 2020 SGP-TR-216 1 Statistical Modeling of Subsurface Temperatures in the Great Basin Cary R. Lindsey 1 , Whitney Trainor-Guitton 2 , Bridget Ayling 1 , and Bastien Poux 3 1 Great Basin Center for Geothermal Energy, Nevada Bureau of Mines and Geology, University of Nevada Reno, Reno NV 89557 2 Department of Geophysics, Colorado School of Mines, Golden, CO 80401 3 Seequent, Suite 300 – 860 Homer St., Vancouver, BC, Canada, V6B 2W2 [email protected] Keywords: geothermal, temperature, statistics, RBF, 3D ABSTRACT Robust subsurface temperature models in geothermal systems allow for more focused exploration, enhanced well targeting, and improved reservoir characterization. In the Great Basin region, existing maps of sub-surface temperatures at specified depths were created through synthesis and interpolation of bottom-hole temperature (BHT) data compiled by the Southern Methodist Geothermal Lab (e.g. Google.org map products). For this approach, corrected BHTs were used in combination with site specific models of vertical thermal conductivity to calculate heat flux. This in turn was used to extrapolate temperatures at various depth slices and interpolate between points. We explore an alternative approach to evaluating the BHT dataset, using geostatistical tools in combination with advances in modelling software. This approach may provide additional, novel insight into the spatial variability of the thermal regime in the Great Basin region. In contrast to previous methods used, geostatistical tools incorporate the spatial correlation structure of a dataset into the resulting model and can highlight other possible spatial relationships. In this pilot study, we use the Leapfrog Edge geostatistical software package to create a 3D statistical model of subsurface temperatures in an area of Nevada that was previously studied as part of a geothermal play fairway analysis and is home to multiple geothermal power plants. We use the same well BHT dataset from the Southern Methodist Geothermal Lab to facilitate direct comparison of our results with previous sub-surface temperature models. Variograms are used to define the spatial correlation structure and radial base functions used to estimate temperatures across the region. We ground-truth our results with available data from industry and other projects. 1. INTRODUCTION Geostatistical methods provide a means of producing 3D statistical temperature models of geothermal systems; however, most geostatistical temperature models produced to date have been 2D (Wittier et al., 2019) or quasi-3D. Many researchers have modeled temperatures in 2D using bottom-hole temperatures (BHTs) (Blackwell and Richards, 2004); shallow subsurface temperatures (Fairley et al., 2003, Price et al., 2017); 2-meter temperature and geo-probe data (Zehner et al., 2012; Coolbaugh et al., 2014); and heat flow and other derivatives of temperature measurements (Williams and DeAngelo, 2011). 3D models provide a means of focusing on multidirectional spatial correlation and producing a more constrained temperature distribution. Producing 3D temperature models of a world-class geothermal province such as the Great Basin could enhance exploration and further exploitation of known systems in the region and assist in the discovery of blind systems. Because 3D models such as we present here are predictive, they can reduce the amount of drilling required to define a system or allow for better drill targeting of TG (temperature gradient) wells. Estimates suggest that as many as 2/3 of the systems in the basin are blind systems (Faulds et al., 2015). In an effort to identify some of these blind systems, a Department of Energy-funded play fairway project was completed out in Nevada between 2015 and 2019. The project identified areas of high favorability for geothermal potential and saw success in the final stage with temperature gradient holes drilled in Gabbs Valley and Granite Springs with elevated temperature profiles. One TG well in Gabbs Valley registered temperatures of approximately 120 °C at 150 meters depth (Craig, 2018). While not as high, temperatures in Granite Springs were also elevated and further evaluation may better define the system for drilling of future TG wells. In part because of the success of the Nevada project, we decided to focus our study on this area though we did choose to focus only on the western half of the play fairway study. The study area in western Nevada (Figure 1A) has over 15 geothermal plants including Dixie Valley, Desert Peak, and Steamboat as well as several high-profile geothermal prospects such as Fallon (Ayling et al., 2018), Gabbs Valley (Craig et al., 2017), and Granite Springs Valley (Faulds et al., 2019) (Figure 2B). Because of the presence of geothermal plants and the potential availability of additional data points, this area is the best with which to validate the model results. The data used for the model were filtered from a larger data set compiled by Southern Methodist University (SMU, 2019). The entire dataset is available on the Southern Methodist node of the National Geothermal Data System. While this study includes only data available from the SMU, the Great Basin Center for Geothermal Energy continues to acquire and store geothermal datasets for the entire Great Basin. As these data are compiled, it will be possible to include more BHTs, 2-meter temperature data, geo-probe data, and any other relevant temperature data to refine the current model and expand the study across the Great Basin. Here, we present a preliminary, 3D statistical temperature module using the Leapfrog Edge.

Upload: others

Post on 14-Jul-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Statistical Modeling of Subsurface Temperatures in the ...€¦ · with the transfer of NW-trending dextral shear of the Walker Lane to the WNW extension of the Great Basin (Faulds

PROCEEDINGS, 45th Workshop on Geothermal Reservoir Engineering

Stanford University, Stanford, California, February 10-12, 2020

SGP-TR-216

1

Statistical Modeling of Subsurface Temperatures in the Great Basin

Cary R. Lindsey1, Whitney Trainor-Guitton

2, Bridget Ayling

1, and Bastien Poux

3

1Great Basin Center for Geothermal Energy, Nevada Bureau of Mines and Geology, University of Nevada Reno, Reno NV 89557

2Department of Geophysics, Colorado School of Mines, Golden, CO 80401

3Seequent, Suite 300 – 860 Homer St., Vancouver, BC, Canada, V6B 2W2

[email protected]

Keywords: geothermal, temperature, statistics, RBF, 3D

ABSTRACT

Robust subsurface temperature models in geothermal systems allow for more focused exploration, enhanced well targeting, and

improved reservoir characterization. In the Great Basin region, existing maps of sub-surface temperatures at specified depths were

created through synthesis and interpolation of bottom-hole temperature (BHT) data compiled by the Southern Methodist Geothermal

Lab (e.g. Google.org map products). For this approach, corrected BHTs were used in combination with site specific models of vertical

thermal conductivity to calculate heat flux. This in turn was used to extrapolate temperatures at various depth slices and interpolate

between points. We explore an alternative approach to evaluating the BHT dataset, using geostatistical tools in combination with

advances in modelling software. This approach may provide additional, novel insight into the spatial variability of the thermal regime in

the Great Basin region. In contrast to previous methods used, geostatistical tools incorporate the spatial correlation structure of a dataset

into the resulting model and can highlight other possible spatial relationships. In this pilot study, we use the Leapfrog Edge geostatistical

software package to create a 3D statistical model of subsurface temperatures in an area of Nevada that was previously studied as part of

a geothermal play fairway analysis and is home to multiple geothermal power plants. We use the same well BHT dataset from the

Southern Methodist Geothermal Lab to facilitate direct comparison of our results with previous sub-surface temperature models.

Variograms are used to define the spatial correlation structure and radial base functions used to estimate temperatures across the region.

We ground-truth our results with available data from industry and other projects.

1. INTRODUCTION

Geostatistical methods provide a means of producing 3D statistical temperature models of geothermal systems; however, most

geostatistical temperature models produced to date have been 2D (Wittier et al., 2019) or quasi-3D. Many researchers have modeled

temperatures in 2D using bottom-hole temperatures (BHTs) (Blackwell and Richards, 2004); shallow subsurface temperatures (Fairley

et al., 2003, Price et al., 2017); 2-meter temperature and geo-probe data (Zehner et al., 2012; Coolbaugh et al., 2014); and heat flow and

other derivatives of temperature measurements (Williams and DeAngelo, 2011). 3D models provide a means of focusing on

multidirectional spatial correlation and producing a more constrained temperature distribution. Producing 3D temperature models of a

world-class geothermal province such as the Great Basin could enhance exploration and further exploitation of known systems in the

region and assist in the discovery of blind systems. Because 3D models such as we present here are predictive, they can reduce the

amount of drilling required to define a system or allow for better drill targeting of TG (temperature gradient) wells.

Estimates suggest that as many as 2/3 of the systems in the basin are blind systems (Faulds et al., 2015). In an effort to identify some of

these blind systems, a Department of Energy-funded play fairway project was completed out in Nevada between 2015 and 2019. The

project identified areas of high favorability for geothermal potential and saw success in the final stage with temperature gradient holes

drilled in Gabbs Valley and Granite Springs with elevated temperature profiles. One TG well in Gabbs Valley registered temperatures of

approximately 120 °C at 150 meters depth (Craig, 2018). While not as high, temperatures in Granite Springs were also elevated and

further evaluation may better define the system for drilling of future TG wells.

In part because of the success of the Nevada project, we decided to focus our study on this area though we did choose to focus only on

the western half of the play fairway study. The study area in western Nevada (Figure 1A) has over 15 geothermal plants including Dixie

Valley, Desert Peak, and Steamboat as well as several high-profile geothermal prospects such as Fallon (Ayling et al., 2018), Gabbs

Valley (Craig et al., 2017), and Granite Springs Valley (Faulds et al., 2019) (Figure 2B). Because of the presence of geothermal plants

and the potential availability of additional data points, this area is the best with which to validate the model results.

The data used for the model were filtered from a larger data set compiled by Southern Methodist University (SMU, 2019). The entire

dataset is available on the Southern Methodist node of the National Geothermal Data System. While this study includes only data

available from the SMU, the Great Basin Center for Geothermal Energy continues to acquire and store geothermal datasets for the entire

Great Basin. As these data are compiled, it will be possible to include more BHTs, 2-meter temperature data, geo-probe data, and any

other relevant temperature data to refine the current model and expand the study across the Great Basin. Here, we present a preliminary,

3D statistical temperature module using the Leapfrog Edge.

Page 2: Statistical Modeling of Subsurface Temperatures in the ...€¦ · with the transfer of NW-trending dextral shear of the Walker Lane to the WNW extension of the Great Basin (Faulds

Lindsey et al.

2

2. STUDY AREA

The Great Basin spans approximately 500,000 km2 and includes much of Nevada and Utah as well as parts of Oregon, Idaho, and

California and is part of the Basin and Range Province in the western United States (Figure 1). This extensional province of horsts and

grabens has been active since the Miocene (Faulds et al., 2017). High geothermal gradients exist in the Basin and Range due to this

extension and subsequent thinning of the crust. This gradient along with the northeast-oriented range front faults and complex

Quaternary faults settings make for a world-class geothermal system with over 20 operating power plants. The Basin currently has over

~1,100 MWe installed geothermal power generation nameplate capacity and could have as much as ~10,000 MWe of undiscovered

potential.

The previous play fairway project in Nevada focused on a 96,000 km2 area in central Nevada (Figure 1A) with multiple power plants.

For this pilot project, we focused on the western half of the play fairway area as indicated by the BHT locations in Figure 1. This area

includes the majority of the power plants in Nevada and allows for easier validation of the model results.

The study area is dominated by NNE-trending ranges of the Great Basin and is representative of westward-increasing strain rates

associated with the Basin and Range (Faulds et al., 2016). The geothermal activity in this area is primarily structurally-controlled by

normal faults (Faulds et al., 2011). Many power plants in the region are associated with this typical Basin and Range setting including

Dixie Valley and McGinness Hills.

A second geologic feature of the study area associated with geothermal activity is the Walker Lane, a system of dextral strike-slip faults

in the southwestern corner of the study area. Geothermal activity in this area is presumed to be a result of extensional dilation associated

with the transfer of NW-trending dextral shear of the Walker Lane to the WNW extension of the Great Basin (Faulds et al., 2006).

Power plants in this area include Don Campbell, Wabuska, Soda Lake, and others.

Figure 1: A. The Great Basin with the Nevada play fairway study area and BHTs from the current study area defined, B. Power

plants in the Nevada Play Fairway study area.

3. METHODS

3.1 Data compilation

BHT data were acquired from the SMU node of the National Geothermal Data System (http://geothermal.smu.edu). The master file

includes data such as location (latitude and longitude), API numbers when available and applicable, BHT data, depth, thermal

conductivity, and heat flow among others. For this project, we filtered the data to include only locations within the western half of the

Nevada play fairway area (Figure 1A) with depth and BHT. After the data were filtered, we were left with approximately 1200 data

points.

3.2 Data analysis

Summary statistics were calculated for both the temperature and depth data (Table 1). Point data for temperature and depth were also

plotted in ArcGIS to view both the spatial distribution and data range (Figures 2A and 2B).

Page 3: Statistical Modeling of Subsurface Temperatures in the ...€¦ · with the transfer of NW-trending dextral shear of the Walker Lane to the WNW extension of the Great Basin (Faulds

Lindsey et al.

3

Table 3. Summary statistics of temperature and depth data

Min 1st Quartile Median Mean 3rd Quartile Max StDev

Temperatures

(℃)

5.50 19.80 27.30 46.54 57.90 283.70 44.62024

Depth (m) -359.200 -15.240 -9.750 -26.539 -6.707 -0.460 54.80242

Figure 2: A. BHT locations by temperature, and B. BHT locations by depth.

3.2 Model generation

To create a 3D statistical model using the BHTS, we used Leapfrog Edge with Leapfrog Geothermal. The process involved several steps

including defining the spatial correlation structure of the data, choosing an appropriate interpolation method, and applying parameters

derived from the spatial correlation structure to the chosen interpolation method.

To define the spatial correlation of the data, we created and reviewed experimental variograms. An experimental variogram is a discrete

function calculated using a measure of variability between pairs of points at various locations and is defined as:

𝛾(𝑟) =1

2𝑛∑ [𝑧(𝑥) − 𝑧(𝑥 + ∆𝑥]2𝑁

𝑛=1 . Equation 1

The radial variogram shows a decrease in spatial correlation from the east to west as would be expected in the Great Basin (Figure 3).

The north-south trending ranges of the basin represent a clear divide between the spatial correlation that is evident across the basins. The

correlation distance (range) is dramatically reduced in the east-west direction (Figures 4A and 4B).

Page 4: Statistical Modeling of Subsurface Temperatures in the ...€¦ · with the transfer of NW-trending dextral shear of the Walker Lane to the WNW extension of the Great Basin (Faulds

Lindsey et al.

4

Figure 3: Radial variogram of BHT data.

Figure 4: A. Major axis variogram and B. Minor axis variogram.

Once the parameters were used with the radial base function, several versions of the 3D model were produced. A smoother contour

model typical of temperature models associated with geothermal heat flow was then sliced at 3 x,y,z locations for analysis (Figure 7A,

7B, and 7C). An alternate representation is a gridded contour (Figures 8A, 8B, and 8C), which perhaps visually provides a better

representation of the heat volume as do the 3D contours for the 100, 150, and 200 ºC isotherms (Figures 9A, 9B, and 9C). Figures 10A,

B, and C provide alternative representations such as a block diagram with added transparency, allowing the visualization of the

temperature contours and the BHT locations.

Page 5: Statistical Modeling of Subsurface Temperatures in the ...€¦ · with the transfer of NW-trending dextral shear of the Walker Lane to the WNW extension of the Great Basin (Faulds

Lindsey et al.

5

Figure 5: Contoured slices: A. Slice 1 at x=4396600, y=345400, z=438.86; B. Slice 2 at x=4396600, y=425400, z=438.86; C. Slice 3

at x=4396600, y=256400, z=438.86, locations are in UTM Zone 11.

Figure 6: Gridded Slices: A: Slice 4 at x=4361600, y=365400, z=438.86; B. Slice 5 at x=4361600, y=430400, z=438.86; C. Slice 6:

x=4367600, y=353400, z=438.86.

Figure 7: Isotherms: A. 100 °C, B. 150 °C, and C. 200 °C.

Page 6: Statistical Modeling of Subsurface Temperatures in the ...€¦ · with the transfer of NW-trending dextral shear of the Walker Lane to the WNW extension of the Great Basin (Faulds

Lindsey et al.

6

Figure 8: A. Temperature contours, B. Block model, C. Block model with temperature contours.

5. CONCLUSION

While the Great Basin has seen extensive geothermal development, it is evident that there is more hidden geothermal potential.

Understanding the structure of the heat flow in the basins can aid in exploration. Visualizing those heat structures as we have shown

here can allow researchers look for similar patterns across the basin and also understand controls on the favorability or unfavorability of

an area for geothermal development. Future work will continue to explore and develop these geostatistical approaches for building

improved thermal models in the Great Basin region, including incorporation of different datasets (e.g. geophysics).

REFERENCES

Ayling, B., and Blankenship, D.: Phase 2 update for the Fallon FORGE site, Nevada, USA. Proceedings, 43rd Workshop on Geothermal

Reservoir Engineering, Stanford University, (2018). SGP-TR-213.

Blackwell, D.D., and M.C. Richards, 2004, Geothermal Map of North America, American Association of Petroleum Geologists.

Boyd, D.L., Walton, G., and Trainor-Guitton, W., 2018, Improving geological models through statistical integration of borehole data

and geologists’ cross-sections, in 52nd U.S. Rock Mechanics/Geomechanics Symposium.

Coolbaugh, M., Sladek, R., Zehner, and Kratt, C., 2014, Shallow Temperature Surveys for Geothermal Exploration in the Great Basin,

USA and Estimation of Shallow Aquifer Heat Loss. Transactions, Geothermal Resources Council, 38.

Craig, J.W., Faulds, J.E., Shevenell, L.A., and Hinz, N.H., 2017, Discovery and Analysis of a Potential Blind Geothermal System in

Southern Gabbs Valley, Western Nevada., Transactions, Geothermal Resources Council, 41.

Craig, Jason: Discovery and Analysis of a Blind Geothermal System in Southeastern Gabbs Valley, Western Nevada (Master’s thesis).

University of Nevada Reno, (2018).

Deutsch, C. V., 2002, Geostatistical Reservoir Modeling: New York, Osford University Press, 124–152 p. Deutsch, C. V., and Journel,

A.G., 1998, GSLIB: Geostatistical Software Library and User’s Guide.: New York, Oxford University Press.

Fairley, J., Heffner, J., and Hinds, J., 2003, Geostatistical evaluation of permeability in an active fault zone: Geophysical Research

Letters, v. 30, p. 1962, doi: 10.1029/2003GL018064.

Faulds, J., Coolbaugh, M., Hinz, N., Sadowski, A., Shevenell, L., McConville, E., Craig, J., and Siler, D., 2017, Progress Report on the

Nevada Play Fairway Project: Integrated Geological, Geochemical, and Geophysical Analyses of Possible New Geothermal

Systems in the Great Basin Region, Transactions, Geothermal Resources Council, 42.

Isaaks, E., and Srivastava, R., 1989, An introduction to applied geostatistics: Oxford Univ. Press, New York, 561 p. Lindsey et al. 11

Kelkar, M., and Perez, G., 2002, Applied Geostatistics for Reservoir Characterization: Society of Petroleum Engineers, 264 p.

Lubenow, B.L., Fairley, J.P., Lindsey, C.R., and Larson, P.B., 2016, Influences on shallow ground temperatures in high flux thermal

systems: Journal of Volcanology and Geothermal Research, v. 323, doi: 10.1016/j.jvolgeores.2016.04.039. Price, A.N., Lindsey,

C.R., and Fairley, J.P., 2017, Interpretation of Ground Temperature Anomalies in Hydrothermal Discharge Areas: Water Resources

Research, v. 53, doi: 10.1002/2017WR021077.

SMU Node of National Geothermal Data System, SMU Heat Flow Database of Equilibrium Log Data and Geothermal Wells. Retrieved

from http://geothermal.smu.edu/static/DownloadFilesButtonPage.htm? on January 10, 2019.

Page 7: Statistical Modeling of Subsurface Temperatures in the ...€¦ · with the transfer of NW-trending dextral shear of the Walker Lane to the WNW extension of the Great Basin (Faulds

Lindsey et al.

7

Williams, C.F., and Deangelo, J., 2011, Evaluation of Approaches and Associated Uncertainties in the Estimation of Temperatures in

the Upper Crust of the Western United States: Transactions - Geothermal Resources Council, v. 35.

Witter, J.B., Trainor-Guitton, W.J., and Siler, D.L., 2019, Uncertainty and risk evaluation during the exploration stage of geothermal

development: A review: Geothermics, v. 78, p. 233–242, doi: 10.1016/j.geothermics.2018.12.011. Zehner, R.E., Tullar, K.N., and

Rutledge, E., 2012, Effectiveness of 2-Meter and Geoprobe Shallow Temperature Surveys in Early Stage Geothermal Exploration:

GRC Transactions, v. 36, p. 835–842