skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: PoroTomo Final Technical Report: Poroelastic Tomography by Adjoint Inverse Modeling of Data from Seismology, Geodesy, and Hydrology

Abstract

This final technical report compiles the results of the PoroTomo project conducted between 1 October 2014 and 31 December 2018. The report cites articles published in the peer-reviewed literature and indicates data sets submitted to the Geothermal Data Repository (GDR). In the geothermal field at Brady Hot Springs, Nevada, subsidence occurs over an elliptical area that is ~4 km by ~1.5 km. Results from inverse modeling suggest that the deformation is a result of volumetric contraction in units with depth less than 600 m. Characterizing such structures in terms of their rock-mechanical properties is essential to successful operations of Enhanced Geothermal Systems (EGS). The goal of the PoroTomo project is to assess an integrated technology for characterizing and monitoring changes in the rock-mechanical properties of an EGS reservoir in three dimensions with fine spatial resolution. In March 2016, the PoroTomo team deployed the integrated technology in a 1500-by-500-by-400-meter volume at Brady Hot Springs. The integrated technology analyzes data from multiple arrays of sensors, including: active seismic sources, fiber-optic cables for Distributed Acoustic Sensing (DAS) and Distributed Temperature Sensing (DTS) arranged vertically in a borehole to ~400 m depth and horizontally in a trench 8700 m in length and 0.5 mmore » in depth, 246 three-component seismometers on the surface, three pressure sensors in observation wells, continuous geodetic measurements at three GPS stations, and seven satellite images using Synthetic Aperture Radar (SAR). The deployment consisted of four distinct time intervals ("stages"). During each measurement interval, the hydrological conditions were intentionally manipulated by modifying the rates of pumping in the injection and production wells. To account for the mechanical behavior of both the rock and the fluids, the PoroTomo team has developed numerical models for the 3-dimensional distribution of the material properties via inverse modeling of the three data sets (seismology, geodesy, and hydrology) individually. The estimated values of the material properties are registered on a three-dimensional grid with a spacing of 25 meters between nodes. The results agree on the following points. The material is unconsolidated and/or fractured, especially in the shallow layers. The structural trends follow the fault system in strike and dip. The geodetic measurements favor the hypothesis of thermal contraction. Temporal changes in pressure, subsidence rate, and seismic amplitude are associated with changes in pumping rates during the four stages of the deployment in 2016. The modeled hydraulic conductivity is high in the damage zones surrounding the faults. All the observations are consistent with a conceptual model of highly permeable conduits along faults channeling fluids from shallow aquifers to the deep geothermal reservoir tapped by the production wells. During the 51-month performance period, the PoroTomo project produced: 1 doctoral dissertation, 5 master's theses, 11 articles published in the international peer-reviewed literature, 48 presentations at conferences, and 73 data sets submitted to the GDR. Looking forward, the PoroTomo team makes the following recommendations that apply to other experiments, such as those planned for the Frontier Observatory for Research in Geothermal Energy (FORGE) site in Utah. A multi-disciplinary team, including researchers from universities, industry and national laboratories contributed to the success of the PoroTomo project at Brady Hot Springs, Nevada, and led to collaborations at other geothermal sites, including Raft River, Idaho; Coso California; San Emidio, Nevada; Fallon, Nevada; and Milford, Utah. Critical discussions, including fortnightly teleconferences and yearly reviews, as well as scientific conferences, were essential to integrating the three types of data. Regarding seismology, the PoroTomo learned several lessons that lead to the following recommendations. To compute material properties such as Young's modulus and Poisson's ratio from seismic data, the PoroTomo team took advantage of pre-existing 3-dimensional models of density that were inferred from a combination of gravimetric surveys and geologic models. For the active-source seismology, a repeatable, accelerated weight drop, such as the HH Seismic Hammer, would generate signals more like a simple impulse than traditional sources with time-varying ("sweeping") frequencies. Taking advantage of local and regional seismic activity would also enlarge the data set. Using a combination of recordings made by DAS and conventional geophones in a joint inversion would improve the spatial resolution of the resulting models of material properties. Horizontal DAS arrays would benefit from long, straight segments of cable that minimize corners. If future deployments include multiple wells with DAS and/or downhole geophone arrays, then cross-hole studies using 3-D vertical seismic profiling (VSP) with active sources and/or ambient noise tomography would provide additional information. Regarding geodesy, the PoroTomo learned several lessons that lead to the following recommendations. A GPS station in the actively deforming area is required to reference the InSAR data as well as to illuminate transient deformation signals on time scales on the order of a day that InSAR cannot recover. SAR images provide the required spatial resolution for deformation modeling, with X-band (e.g., TerraSAR-X) data providing the clearest interferograms. Although C-band (e.g., ERS-1, ERS-2) signals are adequate, SCANSAR images (e.g., Sentinel-1) degrade the spatial resolution too much to be useful for analyzing geothermal processes in the subsurface. It is also useful to have a satellite with a short revisit time to capture transient signals. In addition, a large dataset of interferometric pairs is useful to analyze trends over years. The "multi-cube" parameterization can be adapted to different spatial scales. Spatial correlation and prior models corresponding to the Bayesian, geostatistical inversion can also be adapted to reflect conditions at other sites. Regarding hydrology, the PoroTomo team learned several lessons that lead to the following recommendations. The hydraulic data set provides independent information on reservoir hydraulic and thermal properties as a baseline against which other geophysical inversion results can be assessed. The hydraulic data set allows investigation and explanation of "multi-physics" coupled processes that occur in geothermal reservoirs (e.g., thermal/hydraulic coupling associated with movement of heat in wellbores, and hydraulic/mechanical coupling associated with deformation). Indeed, any assessment of the value of geophysical information strongly benefits from detailed hydraulic characterization data be collected alongside to provide validation of geophysical imaging or predictions. In future investigations, the following actions are recommended in order to improve hydraulic characterization of geothermal reservoirs. Hydraulic monitoring of temporal changes would be improved by via continuously acquiring more numerous, more accurate and more detailed measurements, especially in observation wells that are not pumping for either production or injection. Similarly, deploying packers and DTS in boreholes would improve spatial resolution. Multi-frequency oscillating pumping tests would improve the resolution of hydraulic parameters. Regarding the fiber-optic technologies of DAS and DTS, the PoroTomo learned several lessons that lead to the following recommendations. The fiber-optic cable deployed vertically inside the casing in Brady Well 56-1 in March 2016 was a 1/8"-diameter (3.2 mm) bare 316 stainless steel double tube containing both single-mode and multi-mode, high temperature acrylate-coated fibers rated to 150*C. The cable was left in place following the deployment. More than two years later, in August 2018, the same fiber recorded temperatures up to 165*C with no loss of signal quality. Coupling is the most important consideration for DAS deployments. The particle motion needs to be coupled through the (rock or soil) formation, to the cable structure, and into the fiber itself. When installing new infrastructure, it is straightforward to clamp the cable to the casing as it goes into the well and then cement both the cable and the casing in place. This provides significantly improved coupling when compared to frictional coupling alone. The DAS technology has advanced since the PoroTomo deployment in 2016. For example, data from the Carina DAS system developed by Silixa in 2018 has a signal-to-noise ratio (SNR) that is two orders of magnitude greater than that deployed at Brady Hot Springs. The sensitivity allows for measuring strain on the picostrain scale and can be sensitive enough to measure the dilation and contraction of fractures directly in situ. To assess the usefulness of the various data sets, the PoroTomo Team used a mathematical theory known as the Value of Information (VOI). Applied to seismic imaging, this approach shows that forward modeling of wave fields in combination with classification by machine learning algorithms would optimize the locations of seismic sources and receivers in a proposed deployment. Overall, the PoroTomo project achieved its objective of assessing "an integrated technology for characterizing and monitoring changes in an EGS reservoir in three dimensions". The technology performance metric of (spatial) resolution was assessed using checkerboard tests. For the material property of P-wave velocity estimated using body-wave, travel-time tomography, the spatial resolution of the model is 100 m over most of the study area at a depth of 200 m. The P-wave and S-wave velocity models estimated from sweep interferometry were combined with a model of density to calculate the material properties of Poisson's ratio and Young's modulus with the same level of resolution. The 100-meter resolution achieved using the data from seismology represents a 2-fold improvement over the models existing at Brady before the PoroTomo project began and meets the minimum requirement specified in the Statement of Project Objectives (SOPO). Using data from geodesy, the PoroTomo team was able to estimate the volumetric strain rate in the shallow aquifer with a spatial resolution of 100 m in a layer at depths of 50-150 m, thus going beyond the target resolution of 250 m specified in the SOPO. The rate of volumetric strain rate d\0x03B5/dt can be used to infer the rate of cooling dT/dt and the rate of change in thermal energy dE/dt in the shallow parts of the field. Using data from hydrology, the PoroTomo team was able to estimate the hydraulic conductivity K with a spatial resolution of 500 m, thus meeting the minimum requirement specified in the SOPO. These estimates of hydraulic conductivity K have been used to simulate the flow paths within the geothermal field at Brady. If estimates of porosity were available, then they could be combined with the hydraulic conductivity K to calculate permeability and thus further constrain models of fluid flow. Using the information from seismology, geodesy, and hydrology, the PoroTomo team achieved a spatial resolution of 100 m, thus meeting the minimum requirement specified in the SOPO. Before starting the PoroTomo project, the methods for analyzing each of the three types of data individually were at Technology Readiness Level TRL 2 ("technology concept formulated") or TRL 3 ("proof of concept"), as stated in the SOPO. Following the deployment of a prototype in the field at Brady Hot Springs in 2016, each of these methods has increased to TRL 5 ("component validation in relevant environment"). The integrated technology, analyzing all three data types, is now at TRL 3 ("proof of concept").« less

Authors:
ORCiD logo [1];  [1];
  1. Department of Geoscience, University of Wisconsin-Madison
Publication Date:
Research Org.:
Univ. of Wisconsin, Madison, WI (United States)
Sponsoring Org.:
USDOE Office of Energy Efficiency and Renewable Energy (EERE), Geothermal Technologies Office (EE-4G)
Contributing Org.:
University of Wisconsin-Madison (UW) Dept. of Geoscience http://geoscience.wisc.edu Temple University (TU) http://astro.temple.edu/~davatzes University of Nevada-Reno (UNR) http://geodesy.unr.edu Ormat Technologies, Inc. http://www.ormat.com/ Silixa Ltd. http://www.silixa.com/ Lawrence Berkeley National Laboratory (LBL) http://www.lbl.gov/ Lawrence Livermore National Laboratory (LLNL) https://www.llnl.gov/
OSTI Identifier:
1499141
Report Number(s):
3.1
DOE Contract Number:  
EE0006760
Resource Type:
Technical Report
Resource Relation:
Related Information: This final technical report compiles the results of the PoroTomo project conducted between 1 October 2014 and 31 December 2018. The report cites articles published in the peer-reviewed literature and indicates data sets submitted to the Geothermal Data Repository (GDR).
Country of Publication:
United States
Language:
English
Subject:
15 GEOTHERMAL ENERGY; InSAR GPS DAS DTS Brady Hot Springs Nevada PoroTomo

Citation Formats

Feigl, Kurt L., Parker, Lesley M., and Team, PoroTomo. PoroTomo Final Technical Report: Poroelastic Tomography by Adjoint Inverse Modeling of Data from Seismology, Geodesy, and Hydrology. United States: N. p., 2019. Web. doi:10.2172/1499141.
Feigl, Kurt L., Parker, Lesley M., & Team, PoroTomo. PoroTomo Final Technical Report: Poroelastic Tomography by Adjoint Inverse Modeling of Data from Seismology, Geodesy, and Hydrology. United States. doi:10.2172/1499141.
Feigl, Kurt L., Parker, Lesley M., and Team, PoroTomo. Wed . "PoroTomo Final Technical Report: Poroelastic Tomography by Adjoint Inverse Modeling of Data from Seismology, Geodesy, and Hydrology". United States. doi:10.2172/1499141. https://www.osti.gov/servlets/purl/1499141.
@article{osti_1499141,
title = {PoroTomo Final Technical Report: Poroelastic Tomography by Adjoint Inverse Modeling of Data from Seismology, Geodesy, and Hydrology},
author = {Feigl, Kurt L. and Parker, Lesley M. and Team, PoroTomo},
abstractNote = {This final technical report compiles the results of the PoroTomo project conducted between 1 October 2014 and 31 December 2018. The report cites articles published in the peer-reviewed literature and indicates data sets submitted to the Geothermal Data Repository (GDR). In the geothermal field at Brady Hot Springs, Nevada, subsidence occurs over an elliptical area that is ~4 km by ~1.5 km. Results from inverse modeling suggest that the deformation is a result of volumetric contraction in units with depth less than 600 m. Characterizing such structures in terms of their rock-mechanical properties is essential to successful operations of Enhanced Geothermal Systems (EGS). The goal of the PoroTomo project is to assess an integrated technology for characterizing and monitoring changes in the rock-mechanical properties of an EGS reservoir in three dimensions with fine spatial resolution. In March 2016, the PoroTomo team deployed the integrated technology in a 1500-by-500-by-400-meter volume at Brady Hot Springs. The integrated technology analyzes data from multiple arrays of sensors, including: active seismic sources, fiber-optic cables for Distributed Acoustic Sensing (DAS) and Distributed Temperature Sensing (DTS) arranged vertically in a borehole to ~400 m depth and horizontally in a trench 8700 m in length and 0.5 m in depth, 246 three-component seismometers on the surface, three pressure sensors in observation wells, continuous geodetic measurements at three GPS stations, and seven satellite images using Synthetic Aperture Radar (SAR). The deployment consisted of four distinct time intervals ("stages"). During each measurement interval, the hydrological conditions were intentionally manipulated by modifying the rates of pumping in the injection and production wells. To account for the mechanical behavior of both the rock and the fluids, the PoroTomo team has developed numerical models for the 3-dimensional distribution of the material properties via inverse modeling of the three data sets (seismology, geodesy, and hydrology) individually. The estimated values of the material properties are registered on a three-dimensional grid with a spacing of 25 meters between nodes. The results agree on the following points. The material is unconsolidated and/or fractured, especially in the shallow layers. The structural trends follow the fault system in strike and dip. The geodetic measurements favor the hypothesis of thermal contraction. Temporal changes in pressure, subsidence rate, and seismic amplitude are associated with changes in pumping rates during the four stages of the deployment in 2016. The modeled hydraulic conductivity is high in the damage zones surrounding the faults. All the observations are consistent with a conceptual model of highly permeable conduits along faults channeling fluids from shallow aquifers to the deep geothermal reservoir tapped by the production wells. During the 51-month performance period, the PoroTomo project produced: 1 doctoral dissertation, 5 master's theses, 11 articles published in the international peer-reviewed literature, 48 presentations at conferences, and 73 data sets submitted to the GDR. Looking forward, the PoroTomo team makes the following recommendations that apply to other experiments, such as those planned for the Frontier Observatory for Research in Geothermal Energy (FORGE) site in Utah. A multi-disciplinary team, including researchers from universities, industry and national laboratories contributed to the success of the PoroTomo project at Brady Hot Springs, Nevada, and led to collaborations at other geothermal sites, including Raft River, Idaho; Coso California; San Emidio, Nevada; Fallon, Nevada; and Milford, Utah. Critical discussions, including fortnightly teleconferences and yearly reviews, as well as scientific conferences, were essential to integrating the three types of data. Regarding seismology, the PoroTomo learned several lessons that lead to the following recommendations. To compute material properties such as Young's modulus and Poisson's ratio from seismic data, the PoroTomo team took advantage of pre-existing 3-dimensional models of density that were inferred from a combination of gravimetric surveys and geologic models. For the active-source seismology, a repeatable, accelerated weight drop, such as the HH Seismic Hammer, would generate signals more like a simple impulse than traditional sources with time-varying ("sweeping") frequencies. Taking advantage of local and regional seismic activity would also enlarge the data set. Using a combination of recordings made by DAS and conventional geophones in a joint inversion would improve the spatial resolution of the resulting models of material properties. Horizontal DAS arrays would benefit from long, straight segments of cable that minimize corners. If future deployments include multiple wells with DAS and/or downhole geophone arrays, then cross-hole studies using 3-D vertical seismic profiling (VSP) with active sources and/or ambient noise tomography would provide additional information. Regarding geodesy, the PoroTomo learned several lessons that lead to the following recommendations. A GPS station in the actively deforming area is required to reference the InSAR data as well as to illuminate transient deformation signals on time scales on the order of a day that InSAR cannot recover. SAR images provide the required spatial resolution for deformation modeling, with X-band (e.g., TerraSAR-X) data providing the clearest interferograms. Although C-band (e.g., ERS-1, ERS-2) signals are adequate, SCANSAR images (e.g., Sentinel-1) degrade the spatial resolution too much to be useful for analyzing geothermal processes in the subsurface. It is also useful to have a satellite with a short revisit time to capture transient signals. In addition, a large dataset of interferometric pairs is useful to analyze trends over years. The "multi-cube" parameterization can be adapted to different spatial scales. Spatial correlation and prior models corresponding to the Bayesian, geostatistical inversion can also be adapted to reflect conditions at other sites. Regarding hydrology, the PoroTomo team learned several lessons that lead to the following recommendations. The hydraulic data set provides independent information on reservoir hydraulic and thermal properties as a baseline against which other geophysical inversion results can be assessed. The hydraulic data set allows investigation and explanation of "multi-physics" coupled processes that occur in geothermal reservoirs (e.g., thermal/hydraulic coupling associated with movement of heat in wellbores, and hydraulic/mechanical coupling associated with deformation). Indeed, any assessment of the value of geophysical information strongly benefits from detailed hydraulic characterization data be collected alongside to provide validation of geophysical imaging or predictions. In future investigations, the following actions are recommended in order to improve hydraulic characterization of geothermal reservoirs. Hydraulic monitoring of temporal changes would be improved by via continuously acquiring more numerous, more accurate and more detailed measurements, especially in observation wells that are not pumping for either production or injection. Similarly, deploying packers and DTS in boreholes would improve spatial resolution. Multi-frequency oscillating pumping tests would improve the resolution of hydraulic parameters. Regarding the fiber-optic technologies of DAS and DTS, the PoroTomo learned several lessons that lead to the following recommendations. The fiber-optic cable deployed vertically inside the casing in Brady Well 56-1 in March 2016 was a 1/8"-diameter (3.2 mm) bare 316 stainless steel double tube containing both single-mode and multi-mode, high temperature acrylate-coated fibers rated to 150*C. The cable was left in place following the deployment. More than two years later, in August 2018, the same fiber recorded temperatures up to 165*C with no loss of signal quality. Coupling is the most important consideration for DAS deployments. The particle motion needs to be coupled through the (rock or soil) formation, to the cable structure, and into the fiber itself. When installing new infrastructure, it is straightforward to clamp the cable to the casing as it goes into the well and then cement both the cable and the casing in place. This provides significantly improved coupling when compared to frictional coupling alone. The DAS technology has advanced since the PoroTomo deployment in 2016. For example, data from the Carina DAS system developed by Silixa in 2018 has a signal-to-noise ratio (SNR) that is two orders of magnitude greater than that deployed at Brady Hot Springs. The sensitivity allows for measuring strain on the picostrain scale and can be sensitive enough to measure the dilation and contraction of fractures directly in situ. To assess the usefulness of the various data sets, the PoroTomo Team used a mathematical theory known as the Value of Information (VOI). Applied to seismic imaging, this approach shows that forward modeling of wave fields in combination with classification by machine learning algorithms would optimize the locations of seismic sources and receivers in a proposed deployment. Overall, the PoroTomo project achieved its objective of assessing "an integrated technology for characterizing and monitoring changes in an EGS reservoir in three dimensions". The technology performance metric of (spatial) resolution was assessed using checkerboard tests. For the material property of P-wave velocity estimated using body-wave, travel-time tomography, the spatial resolution of the model is 100 m over most of the study area at a depth of 200 m. The P-wave and S-wave velocity models estimated from sweep interferometry were combined with a model of density to calculate the material properties of Poisson's ratio and Young's modulus with the same level of resolution. The 100-meter resolution achieved using the data from seismology represents a 2-fold improvement over the models existing at Brady before the PoroTomo project began and meets the minimum requirement specified in the Statement of Project Objectives (SOPO). Using data from geodesy, the PoroTomo team was able to estimate the volumetric strain rate in the shallow aquifer with a spatial resolution of 100 m in a layer at depths of 50-150 m, thus going beyond the target resolution of 250 m specified in the SOPO. The rate of volumetric strain rate d\0x03B5/dt can be used to infer the rate of cooling dT/dt and the rate of change in thermal energy dE/dt in the shallow parts of the field. Using data from hydrology, the PoroTomo team was able to estimate the hydraulic conductivity K with a spatial resolution of 500 m, thus meeting the minimum requirement specified in the SOPO. These estimates of hydraulic conductivity K have been used to simulate the flow paths within the geothermal field at Brady. If estimates of porosity were available, then they could be combined with the hydraulic conductivity K to calculate permeability and thus further constrain models of fluid flow. Using the information from seismology, geodesy, and hydrology, the PoroTomo team achieved a spatial resolution of 100 m, thus meeting the minimum requirement specified in the SOPO. Before starting the PoroTomo project, the methods for analyzing each of the three types of data individually were at Technology Readiness Level TRL 2 ("technology concept formulated") or TRL 3 ("proof of concept"), as stated in the SOPO. Following the deployment of a prototype in the field at Brady Hot Springs in 2016, each of these methods has increased to TRL 5 ("component validation in relevant environment"). The integrated technology, analyzing all three data types, is now at TRL 3 ("proof of concept").},
doi = {10.2172/1499141},
journal = {},
number = ,
volume = ,
place = {United States},
year = {2019},
month = {3}
}