Geosolution-Based Optimization of Geothermal Energy Resources from field to development scale: Case study of Hydrothermal-Volcanic Systems on Mt. Kuju

Chanmaly Chhun^1,2,3, Takeshi Tsuji^2,3, Tatsunori Ikeda², Yasuhiro Fujimitsu², Jun Nishijima², Keigo Kitamura²

​

1. Department of Earth & Atmospheric Sciences, Cornell University

2. Department of Earth Resources Engineering, Kyushu University

3. School of Engineering, the University of Tokyo

​

​

2

​

Introduction – Geothermal Structure

High resolution geothermal structure (velocity, 

resistivity, anisotropic structure, seismicity, etc.)

​

•  reduce the cost and risks of exploration and

development 

​

​

Abnormal reduction in velocity/resistivity/density caused by 

fracture or fluid filled rocks/chambers

3

• Reduce risk or the number of geothermal exploration/development boreholes.
• Different geo-methods, different resolution, thus our geophysical (seismic) imaging and Machine Learning can bridge the gaps and obtain high-resolution geothermal structures.
• Our work provides insights into how exploration and development techniques could be improved.

(Craig Hartline, 2022, Calpine)

Seismic Monitoring at Geysers Geothermal Producing Field

Mt. Kuju geothermal field (~ 50 boreholes)

​

Introduction – Objectives

4

​

Data and Methods

• Azimuthal anisotropic ambient noise tomography for 3D S-wave velocity and 3D seismic anisotropy (Azimuth, Amplitude) (Chhun et al., 2023).
• 3D Resistivity (Aizawa et al., 2022), seismicity catalog (JMA, Andajani et al., 2023).
• 8 of 13 Borehole data (Kitamura et al., 2023) available in the study area, 10km x 10km.

5

​

Key Analysis– Surface wave dispersion measurement & 3D S-wave inversion

(Ekström et al., 2009; Sadeghisorkhani et al., 2018;

Jiang and Denolle, 2020)

(Nimiya et al., 2020; Yao et al., 2006; Fang et al., 2015; direct 3D Vs inversion)

6

​

Key Analysis – Geophysical imaging, boreholes, and machine learning models

1. Training data: 8 kuju geothermal wells (depth, Vs, ani amplitude, ani azimuth, resistivity, and temperature).

2. Cross-Fold validation: R-square= ~0.83 based on the Fine Tree Model.

3. Test data: 3D data (depth, Vs, ani amplitude, ani azimuth, resistivity).

Exploration Scale�≥ 35 x 35 km²

8

​

Results – 3D Vs checker

9

​

Results – 3D Vs

Heat Source

Heat Source

Reservoir scale study

(Chhun et al., 2024)

10

​

Discussions – Heat Source/Reservoir Model System

This Study

Mt. Kuju (conceptual model), NEDO

Development Scale�≥ 10 x 10 km²

12

​

Results - 3D Vs & Anisotropic Structure Checker

13

​

Results - 3D Vs & Anisotropic Structure

(>4km)

Shallow Seismicity (<5km)

from JMA and

Andajani et al. (2023)

Low velocity anomalies >> potential geothermal pathways >> Geothermal fluid accumulations

14

​

Results - 3D Vs & Anisotropic Structure

(Boness and

Zoback, 2006)

Reservoir scale study

Power Stations (Hatchobaru-Otake)

Available Temperature Wells

Existing Wells such as Well-A, B, C, D, and WELL-Z (but no temperature data)

​

Results – geophysical (seismic) results and Machine Learning Models

1. Training data: 8 kuju geothermal wells (depth, Vs, ani amplitude, ani azimuth, resistivity, and temperature).

​

2. Cross-Fold validation: R-square= ~0.83 based on the Fine Tree Model.

​

3. Test data: 3D data (depth, Vs, ani amplitude, ani azimuth, resistivity).

16

​

Results – 8 wells (6 wells for train/validation & 2 wells for testing)

Temperature Prediction

17

​

Results – 8 wells (6 wells for train/validation & 2 wells for testing)

Temperature Prediction

​

Results – 3D Temperature Model

The results can be affected by:

​

• Different geophysical resolutions.
• Boreholes are not well distributed (e.g., all in low velocity zones).

(Chhun et al., 2024, work in progress)

Drilling Scale�km - m

Play fairway Analysis (Taverna et al.,

2024) and/or Machine Learning

Low Class / Score / Weight as 0

0: Seismicity/Earthquakes

0: No Heat Source

0: None/seismogenic fault/fracture

0: No fluid/permeability

0: low temperature

0: High velocity/density/resistivity

0: No anisotropy

0: No surface structure

0: No permit area

0: No safe, economic, environmental factors

1: Minor or no seismicity

1: Heat source

1: Individual fault/fracture

1: Fluid/permeability

1: High temperature

1: Low velocity/density/resistivity

1: High anisotropic structure

1: Surface structure (soil gas, thermal, altered strata)

1: Permit area

1: Safe, economic, environmental factors

​

(Chhun et al., 2024, work in progress)

Where to drill

High Class / Score / Weight as 1

21

​

Summary

(modified from Yasukawa, 2011)

• Our methods can locate and characterize magmatic and geothermal reservoirs with high resolution.
• Bridging the gap for geothermal characterization and siting.
• A role model analysis to both volcanic and non-volcanic types.
• We advance geothermal development technologies, contributing to a global sustainable energy transition.

​