Case studies of predictive uncertainty quantification for geothermal models

Omagbon, Jericho; Doherty, John; Yeh, Angus; Colina, Racquel; O’Sullivan, John; McDowell, Julian; Nicholson, Ruanui; Maclaren, Oliver J.; O’Sullivan, Michael

doi:10.1016/j.geothermics.2021.102263

Cited by 9 publications

(5 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Similarly narrow permeability ranges were found by Omagbon et al. (2021), which focused on production history data and used linearized approaches to uncertainty. Previous studies using a fully Bayesian approach to uncertainty quantification have obtained similar uncertainty bounds as obtained in this study.…”

Section: Discussionsupporting

confidence: 60%

“…For example, McDowell et al (2018) used PEST along with used Monte Carlo methods to generate random models with rock permeabilities ranging over five orders of magnitude, but all accepted model realizations had permeability values with less than 1 order of magnitude of uncertainty. Similarly narrow permeability ranges were found by Omagbon et al (2021), which focused on production history data and used linearized approaches to uncertainty. Previous studies using a fully Bayesian approach to uncertainty quantification have obtained similar uncertainty bounds as obtained in this study.…”

Section: Calibration Of Natural State Reservoir Modelsmentioning

confidence: 68%

“…However, the natural state model could serve as a basis for the calibration of a production history model, for example, by setting the posterior parameter distributions obtained from this study as prior distributions for production history model calibration. Thus, the posterior samples obtained from calibration of the production history model could be used for quantifying uncertainty in predictions of the systems future response to production, potentially providing more realistic uncertainty bounds than results from assuming a linear relationship between the parameters and observables used for model calibration (Omagbon et al., 2021). Generally, the development of a natural state model should be seen as a dynamic, iterative process, that improves as additional data becomes available and the conceptual understanding of the system deepens.…”

Section: Discussionmentioning

confidence: 99%

See 2 more Smart Citations

Bayesian Calibration of a Natural State Geothermal Reservoir Model, Krafla, North Iceland

Scott

O’Sullivan

Maclaren

et al. 2022

Water Resources Research

View full text Add to dashboard Cite

The Krafla area in north Iceland hosts a high‐temperature geothermal system within a volcanic caldera. Temperature measurements from boreholes drilled for power generation reveal enigmatic contrasts throughout the drilled area. While wells in the western part of the production field indicate a 0.5–1 km thick near‐isothermal (∼210°C) liquid‐dominated reservoir underlain by a deeper boiling reservoir, wells in the east indicate boiling conditions extending from the surface to the maximum depth of drilled wells (∼2 km). Understanding these systematic temperature contrasts in terms of the subsurface permeability structure has remained challenging. Here, we present a new numerical model of the natural, pre‐exploitation state of the Krafla system, incorporating a new geologic/conceptual model and a version of TOUGH2 extending to supercritical conditions. The model shows how the characteristic temperature distribution results from structural partitioning of the system by a rift‐parallel eruptive fissure and an aquitard at the transition between deeper basement intrusions and high‐permeability extrusive volcanic rocks. As model calibration is performed using a Bayesian framework, the posterior results reveal significant uncertainty in the inferred permeability values for the different rock types, often exceeding two orders of magnitude. While the model shows how zones of single‐phase supercritical vapor develop above the deep intrusive heat source, more data from deep wells is needed to better constrain the extent and temperature of the deep supercritical zones. However, the model suggests the presence of a significant untapped resource at Krafla.

show abstract

Section: Discussionsupporting

confidence: 60%

Section: Calibration Of Natural State Reservoir Modelsmentioning

confidence: 68%

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Bayesian Calibration of a Natural State Geothermal Reservoir Model, Krafla, North Iceland

Scott

O’Sullivan

Maclaren

et al. 2022

Water Resources Research

View full text Add to dashboard Cite

show abstract

“…By far the best approach for addressing this problem is carrying out a formal uncertainty quantification of the geothermal model forecasts. This approach has been successfully applied to several commercial projects and described in several studies (e.g., de Beer et al, 2023;Dekkers et al, 2022;Omagbon et al, 2021). Producing forecasts of lithium production from the SS-GR that include uncertainty quantification would not only account for limitations in the measured data but would also provide much better support for long-term decision making.…”

Section: Summary and Recommendationsmentioning

confidence: 99%

Characterizing the Geothermal Lithium Resource at the Salton Sea

Dobson,

Araya,

Brounce

et al. 2023

View full text Add to dashboard Cite

The energy transition towards a more sustainable and renewable future is a pivotal global endeavor. Central to this shift for the United States is the critical role of domestically sourced lithium, a key mineral in the production of high-performance batteries essential for electric vehicles and renewable energy storage systems. This has driven the United States to invest heavily in a domestic supply chain for batterygrade lithium to enhance energy security, reduce supply chain vulnerabilities, and foster economic growth by tapping into local resources. A notable example is the Biden Administration's "American Battery Materials Initiative," which was included in the $2.8-billion Bipartisan Infrastructure Law (The White House, 2022).The "Salton Sea Known Geothermal Resource Area" in Imperial County, California has been identified as a potential domestic U.S. resource of lithium due to the brine-hosted lithium in the deep subsurface geothermal reservoir. An analysis funded by the U.S. Department of Energy provides an overview of opportunities and challenges associated with developing the lithium resource in the Salton Sea geothermal reservoir, as well as potential environmental and societal impacts to the county and surrounding region.The geologic history of the region suggests that lithium in the subsurface brines could have come from multiple sources, including water and sediments from the Colorado River, which have been periodically deposited over the past several million years; rocks from the mountain ranges surrounding the Imperial Valley; and lithium-bearing volcanic rocks and igneous intrusions from past geologic events. Further, several processes may have concentrated lithium in the brine over time, including evaporative concentration of lithium-bearing water that flowed into the basin and leaching of lithium from the sediments and rocks by the circulating geothermal brines.Geothermal brine production at the Salton Sea Geothermal Field, the area with existing geothermal power plants, has averaged just over 120 million metric tons per year since 2004. Using an approximate lithium brine concentration of 198 parts per million (ppm), the amount of dissolved lithium contained in these produced brines is estimated to be 127,000 metric tons of lithium carbonate equivalent (LCE) per year. The total dissolved lithium content in the well-characterized portion of the Salton Sea Geothermal Reservoir is estimated at 4.1 million metric tons of LCE, and the estimated total resource increases to 18 million metric tons of LCE if assumptions for porosity and total reservoir size are increased to reflect the probable resource extent.Analysts measured lithium concentrations in the reservoir rocks, which were shown to vary with depth and mineralogy. These data were used to help refine conceptual and computer models of the reservoir; specifically, two complementary computer models of the reservoir were developed. Analysts used the first model to simulate the approximate 30-year history of geothermal power production in the area u...

show abstract

“…While Bayesian calibration methods are used in several fields, for example, engineering mechanics (Rappel et al, 2020; Pepi et al, 2020), natural hazards engineering (Zheng et al, 2021), building energy modeling (Hou et al, 2021), and ecological and environmental modeling (Speich et al, 2021), they are not widely utilized in the context of large-scale subsurface models for shallow geothermal applications, and their performance, when applied to field data, has not been tested fully. Case studies utilizing such methods on groundwater models and deep geothermal reservoirs exist (Cui et al, 2019; Omagbon et al, 2021; Scott et al, 2022), however, the complexities present in the shallow subsurface, such as thermal influence from the surface as well as from anthropogenic infrastructure and activity, as well as the contrast in scale and magnitude make the problem context significantly different. Moreover, compared to synthetic data (i.e., data generated numerically or under controlled conditions) calibration on field data poses a number of non-trivial challenges: (a) observations are expensive to obtain and hence scarce, (b) while the spread of observations should capture meaningful variations for robust inference, these are often not easily tractable (even when data is available), leading to a lower content of information compared to synthetic data, and (c) observation noise and measurement errors introduce additional elements of uncertainty to the calibration process.…”

Section: Introductionmentioning

confidence: 99%

Bayesian parameter inference for shallow subsurface modeling using field data and impacts on geothermal planning

Kreitmair

Makasis²,

Menberg³

et al. 2022

DCE

View full text Add to dashboard Cite

Understanding the subsurface is crucial in building a sustainable future, particularly for urban centers. Importantly, the thermal effects that anthropogenic infrastructure, such as buildings, tunnels, and ground heat exchangers, can have on this shared resource need to be well understood to avoid issues, such as overheating the ground, and to identify opportunities, such as extracting and utilizing excess heat. However, obtaining data for the subsurface can be costly, typically requiring the drilling of boreholes. Bayesian statistical methodologies can be used towards overcoming this, by inferring information about the ground by combining field data and numerical modeling, while quantifying associated uncertainties. This work utilizes data obtained in the city of Cardiff, UK, to evaluate the applicability of a Bayesian calibration (using GP surrogates) approach to measured data and associated challenges (previously not tested) and to obtain insights on the subsurface of the area. The importance of the data set size is analyzed, showing that more data are required in realistic (field data), compared to controlled conditions (numerically-generated data), highlighting the importance of identifying data points that contain the most information. Heterogeneity of the ground (i.e., input parameters), which can be particularly prominent in large-scale subsurface domains, is also investigated, showing that the calibration methodology can still yield reasonably accurate results under heterogeneous conditions. Finally, the impact of considering uncertainty in subsurface properties is demonstrated in an existing shallow geothermal system in the area, showing a higher than utilized ground capacity, and the potential for a larger scale system given sufficient demand.

show abstract

Case studies of predictive uncertainty quantification for geothermal models

Cited by 9 publications

References 26 publications

Bayesian Calibration of a Natural State Geothermal Reservoir Model, Krafla, North Iceland

Bayesian Calibration of a Natural State Geothermal Reservoir Model, Krafla, North Iceland

Characterizing the Geothermal Lithium Resource at the Salton Sea

Bayesian parameter inference for shallow subsurface modeling using field data and impacts on geothermal planning

Contact Info

Product

Resources

About