ARTICLES

PUBLISHED ONLINE: 23 JANUARY 2017 | DOI: 10.1038/NGEO2879

Potentially exploitable supercritical geothermal

resources in the ductile crust
Noriaki Watanabe1*, Tatsuya Numakura1, Kiyotoshi Sakaguchi1, Hanae Saishu2, Atsushi Okamoto1,
Steven E. Ingebritsen3 and Noriyoshi Tsuchiya1

The hypothesis that the brittleductile transition (BDT) drastically reduces permeability implies that potentially exploitable
geothermal resources (permeability >1016 m2 ) consisting of supercritical water could occur only in rocks with unusually high
transition temperatures such as basalt. However, tensile fracturing is possible even in ductile rocks, and some permeability
depth relations proposed for the continental crust show no drastic permeability reduction at the BDT. Here we present
experimental results suggesting that the BDT is not the first-order control on rock permeability, and that potentially exploitable
resources may occur in rocks with much lower BDT temperatures, such as the granitic rocks that comprise the bulk of the
continental crust. We find that permeability behaviour for fractured granite samples at 350500 C under effective confining
stress is characterized by a transition from a weakly stress-dependent and reversible behaviour to a strongly stress-dependent
and irreversible behaviour at a specific, temperature-dependent effective confining stress level. This transition is induced by
onset of plastic normal deformation of the fracture surface (elasticplastic transition) and, importantly, causes no jump in
the permeability. Empirical equations for this permeability behaviour suggest that potentially exploitable resources exceeding
450 C may form at depths of 26 km even in the nominally ductile crust.

M
any magma bodies exist in the drillable parts of the Earths
upper crust, and from the 1970s to early 1990s there
was considerable interest in the possibility of extracting
geothermal energy directly from magma and adjacent very hot rock.
This interest gradually waned in light of the technical challenges of
drilling and extracting fluid at near-magmatic temperatures, and
in light of apparent evidence that very hot rocks are rarely (if
ever) permeable enough to support commercially viable flowrates.
Several wells in existing geothermal reservoirs had been drilled deep
enough to attain temperatures near the critical point of pure water
(350 C; the pure-water critical point is 374 C), and each of these
wells encountered high pressures, reactive fluids, and low mass flow
rates1 , implying low permeabilities. In 20082009 the IDDP-1 well
in Iceland unexpectedly encountered rhyolitic magma at a depth of
about 2 km (refs 24). Perhaps even more unexpectedly, subsequent
well tests achieved flowrates of up to 50 kg s1 at temperatures of up
to 440 C (ref. 5). At such high enthalpies (3,200 kJ kg1 ), sustained
flowrates of 30 kg s1 could generate 20 MWe. This discovery
renewed interest in the possibility of extracting energy from the
near-magma environment.
It has been proposed that permeability decreases markedly at
the brittleductile transition (BDT), which occurs at temperatures
of roughly 350400 C and depths of 1015 km in typical crustal
rocks along regional metamorphic geotherms, but can be much
shallower in areas of active magmatism. Assuming that the BDT
drastically reduces permeability, potentially exploitable supercritical
geothermal resources could occur only in rocks with unusually high
transition temperatures such as basalt (450 C)6 . However, even
rocks that are ductile in compression can be brittle in extension7,8 ,
and evidence of intensive tensile fracturing in metamorphic
and plutonic rocks in the middle crust beneath the BDT zone
has been found in outcrops in Antarctica9 . Furthermore, some
permeabilitydepth relations proposed for the continental crust
show no drastic permeability reduction at the BDT10 . Finally,
existence of high-permeability ductile crust, at least transiently,
has been suggested by the temporal expansion of earthquake
swarms induced by the 2011 Tohoku-Oki earthquake in northeast
Japan, attributed to fluid migration at 410 km depth and used to
estimate permeabilities of 1015 m2 (ref. 11). The fluid migration,
believed to be fed by underlying ductile crust, persisted for
more than two months. Thus, potentially exploitable supercritical
geothermal resources may be more widespread than has been
supposed. To explore the formation of potentially exploitable
resources in rocks with relatively low BDT temperatures, such as the
granitic rocks that comprise the bulk of the continental crust1,1214 ,
we performed a set of permeability measurements on fractured
granite at 350500 C and effective confining pressures of up
to 100 MPa.
Permeability behaviour influenced by elasticplastic transition.
We measured permeabilities of cylindrical granite samples
(diameter, 30 mm; length, 25 mm) containing either a mechanically
created single tensile fracture or thermally created multiple tensile
fractures at temperatures of 350, 380, 400, 450 and 500 C, and
effective (hydrostatic) confining stresses of up to approximately
100 MPa, using low-reactivity steam at 12 MPa as the permeant
fluid. The BDT temperature of granitic rocks is 360 C at
typical crustal strain rates1,1214 . Previous experiments on granite
at 300500 C (ref. 15) documented permeability decreases at a
constant effective confining stress of 50 MPa due to crack healing
and sealing by waterrock interaction. The stress dependence of
permeability of fractured rocks at various temperatures, which is
probably a primary control on crustal permeability structure, is
experimentally explored here for the first time.

1 Department of Environmental Studies for Advanced Society, Graduate School of Environmental Studies, Tohoku University, Sendai, Miyagi 9800845,
Japan. 2 Renewable Energy Research Center, National Institute of Advanced Industrial Science and Technology (AIST), Koriyama, Fukushima 9630298,
Japan. 3 US Geological Survey, Menlo Park, California 94025, USA. *e-mail: noriaki.watanabe.e6@tohoku.ac.jp

NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience 1

2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
ARTICLES NATURE GEOSCIENCE DOI: 10.1038/NGEO2879

a 1014 We also measured permeabilities of cylindrical basalt samples

Granite, multiple fractures, 350 C (same dimensions) containing a mechanically created single tensile
fracture. Permeability of the fractured basalt samples first decreases
1015 and then becomes almost constant with increasing effective
Elasticplastic
transition
confining stress at all temperatures (Fig. 1b and Supplementary
Permeability (m2)

Fig. 1), similar to the response observed for a basalt sample

1016
containing a single tensile fracture at room temperature18 . The
permeability behaviour of the fractured basalt samples may be a
1017 Recovery w consequence of the high thermal stability of mechanical properties
hen decrea
stress to th sing of basalts, which can have BDT temperatures as high as 800 C
e initial leve
l
(refs 19,20).
1018 Strongly stress-
Weakly stress dependent and dependent and Empirical equations for predicting the permeability behaviour.
reversible irreversible The permeability behaviour of the fractured granite samples may
1019 be more easily interpreted in loglog plots (Fig. 2) that reveal
0 20 40 60 80 100
two linear curves with different slopes intersecting at the effective
Effective conning stress (MPa) confining stress where the transition from elastic to plastic normal
b 1014
deformation of the fracture surface occurs (hereafter, elasticplastic
Basalt, single fracture, 350 C transition stress or transition stress). The elasticplastic transition
stresses are 70, 50, 3050, 25 and 15 MPa, respectively, at 350,
1015 380, 400, 450 and 500 C. Some variation (about 10 MPa) in
the transition stress at each temperature is expected, given the
range of 3050 MPa obtained from five different granite samples
Permeability (m2)

1016 containing either multiple fractures or a single fracture at 400 C

(Fig. 2cg). The transition stress is clearly temperature dependent,
and an empirical fit to the experimental data yields
1017
log10 eff,trans = 12.8 4.3 log10 T (1)
1018 No transition in the behaviour
where eff,trans is the elasticplastic transition stress (MPa) and T
is the temperature ( C). The slope of the linear curve (that is,
1019 stress dependence of permeability) during loading is 0.9 0.2
0 20 40 60 80 100 (mean s.d.) under elastic conditions and 3.8 0.5 under
Effective conning stress (MPa) plastic conditions, and these slopes appear to be independent of
temperature. Thus, the empirical equations for the average stress
Figure 1 | Relations between permeability and effective confining stress at dependence of permeability are
350 C for granite containing multiple fractures and basalt containing a
single fracture. a, For granite, the transition from weakly stress-dependent 1 log10 k = 0.91 log10 eff (eff eff,trans ) (2)
and reversible permeability changes to highly stress-dependent and
irreversible permeability changes occurs at a specific stress level (about
70 MPa in this case). b, No such transition occurs for the basalt samples.
1 log10 k = 3.81 log10 eff (eff > eff,trans ) (3)

where eff is the effective confining stress (MPa) and k is the

At all temperatures, permeability of the fractured granite samples
first decreases, then becomes almost constant with increasing
effective confining stress, and then decreases again at a relatively
high stress level (for example, Fig. 1a). The stress dependence of
permeability seen at lower stress levels is similar to that observed for
tensile fractures in granite at room temperature16,17 . It reflects elastic
normal deformation of the fracture surface, whereas the renewed
stress dependence of permeability at high stress levels reflects onset
of plastic normal deformation of the fracture surface. Indeed, as
shown in Fig. 1a, decreasing the effective confining stress below
the elasticplastic transition results in almost perfect recovery of
the initial permeability, whereas decreasing the effective confining
stress from points above the elasticplastic transition to its initial
value shows significant permeability hysteresis. The change in stress
dependence of permeability under hydrostatically applied pressure
(that is, very small differential stress) does not represent the BDT;
rather, it represents a newly discovered fracture closing and opening
behaviour with normal stress changes. The BDT is perhaps related to
this phenomenon, but was originally defined in terms of a transition
from frictional sliding to plastic flow in shearing of a rock under high
differential stress7,8 . The observed hysteresis is unlikely to represent
crack healing and sealing15 , because the experiments used low-
pressure steam of low reactivity and were completed within one to
two days.
permeability (m2 ). The slope of the unloading curve is similar to
that for elastic loading (Fig. 2e). Our experiments involved samples
with initial permeabilities ranging widely from 1017 m2 (Fig. 2b) to
1014 m2 (Fig. 2g), so that we suggest that equations (2) and (3) may
reasonably be applicable at a wide range of initial permeabilities.
The permeability behaviour represented by equations (1)(3) is
consistent with suggested permeabilitydepth relations for the con-
tinental crust10 , assuming a zero-confining-pressure (zero-depth)
intercept of 1015 m2 (Fig. 3). This suggests that the temperature-
and stress-dependent permeability behaviour seen in the present
study may be widespread in nature. It is also consistent with obser-
vations in deep geothermal wells that have reached temperatures of
350 C, namely WD-1a in Kakkonda, Japan2123 , Wilson 1 in The
Geysers, United States1,24 , San Vito 1 in Campi Flegrei Fields near
Naples, Italy1,25 , San Pompeo 2 in Larderello, Italy2628 , and NJ-11
in Nesjavellir, Iceland1,29 (Fig. 4). In these deep geothermal wells,
the base of a productive high-temperature geothermal reservoir
was marked by the onset of a conductive geothermal gradient
(WD-1a) or occurrence of elevated pore pressures (the other wells),
either of which suggest a transition from a higher-permeability
regime to a lower-permeability regime. Conditions at the base of
these productive reservoirs are within the plastic conditions (San
Pompeo 2) or near the conditions of the elasticplastic transition
(the other wells).

2 NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience

2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE GEOSCIENCE DOI: 10.1038/NGEO2879 ARTICLES
a 1014 b 1014 c 1014
Granite, multiple fractures, 350 C Granite, multiple fractures, 380 C Granite, multiple fractures, 400 C
1015 Elasticplastic 1015 Elasticplastic 1015 Elasticplastic
transition stress transition stress transition stress
Permeability (m2)

Permeability (m2)

Permeability (m2)
1016 70 MPa 1016 50 MPa 1016 40 MPa

1017 1017 1017

1018 1018 1018

1019 1019 1019

1 10 100 1,000 1 10 100 1,000 1 10 100 1,000
Effective conning stress (MPa) Effective conning stress (MPa) Effective conning stress (MPa)

d 1014 e 1014 f 1014

Granite, multiple fractures, 400 C Granite, multiple fractures, 400 C Granite, single fracture, 400 C
1015 Elasticplastic 1015 Pe 1015
rm
transition stress eab
Permeability (m2)

Permeability (m2)

Permeability (m2)
50 MPa ilit
1016 1016 y-s 1016
tre
ss
pa
1017 1017 th 1017
Elasticplastic Elasticplastic
1018 1018 transition stress 1018 transition stress
30 MPa 40 MPa
1019 1019 1019
1 10 100 1,000 1 10 100 1,000 1 10 100 1,000
Effective conning stress (MPa) Effective conning stress (MPa) Effective conning stress (MPa)

g 1013 h 1014 i 1014

Granite, single fracture, 400 C Granite, multiple fractures, 450 C Granite, multiple fractures, 500 C
1014 1015 1015
Permeability (m2)

Permeability (m2)

Permeability (m2)
1015 1016 1016

1016 1017 1017

Elasticplastic Elasticplastic
1017 transition stress 1018 Elasticplastic 1018 transition stress
40 MPa transition stress
15 MPa
25 MPa
1018 1019 1019
1 10 100 1,000 1 10 100 1,000 1 10 100 1,000
Effective conning stress (MPa) Effective conning stress (MPa) Effective conning stress (MPa)

Figure 2 | Loglog diagrams of the relation between permeability and effective confining stress for granite samples containing either multiple fractures
or a single fracture. The stress level at the intersection of the two linear curves is the elasticplastic transition stress.

0 effective confining stress (or equivalent depth) for potentially

exploitable supercritical geothermal resources may be predicted
This study
5 for granite and other silicic rocks, as shown in Fig. 4. This
Elasticplastic transition prediction assumes that the maximum feasible stress is that at
10 which permeability is significantly (an order of magnitude) less
Brittleductile transition zone than the permeability at the transition stress, based on the fact
Depth (km)

15
that the onset of the transition does not represent complete loss of
permeability (Fig. 2), and that the stress at the base of the productive
20
reservoir in San Pompeo 2 is significantly higher than the transition
Permeability range from stress. Thus, the yellow triangle in the figurebounded by the
25
ref. 10 maximum feasible stress, the critical temperature of water, and the
30
depth corresponding to the critical pressure of waterrepresents
conditions for the exploitable resources.
35 The formation of exploitable supercritical geothermal resources
20 18 16 14 12 (permeability >1016 m2 ) is a transient process caused by magma
log10(Permeability (m2)) intrusion into host rocks6 , and such resources may occur both in the
host rocks and in recently solidified magma even at temperatures
Figure 3 | Comparison between the experimental results summarized in higher than the BDT temperature. The allowable stress will depend
equations (1)(3) and a permeabilitydepth relation from the literature10 . on the initial permeabilities of host rocks and solidified magma right
The permeabilitydepth relation in this study is consistent with those after formation of tensile fractures (that is, at near-zero effective
suggested in the literature, which indicates no abrupt decrease in normal stress on fracture surfaces), where the initial permeabilities
permeability at the brittleductile transition. The elasticplastic transition subsequently decrease with increasing effective normal stress to a
occurs at a depth of approximately 11 km (temperature, 285 C; effective depth-dependent stress level. The initial permeabilities may vary
confining stress, 177 MPa). depending on intensity of fracturing, with values of 1014 to 1015 m2
possible even for tensile fracturing, as demonstrated in the present
study (Fig. 2). For the higher initial permeability of 1014 m2 ,
Potentially exploitable supercritical geothermal resources. On the allowable stress may be between the transition stress and the
the basis of the postulated linkage between the elasticplastic maximum feasible stress, as observed in San Pompeo 2 (or Fig. 2g),
transition and the base of productive reservoirs, a maximum whereas in the case of lower initial permeability, the allowable stress

NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience 3

2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
 ARTICLES
200
tran
Normal deformation of the fracture surface
Elast stress c
sitio
icp
n
Effective conning stress (MPa)
irreversible permeability of the fracture
150
lastic rve
u
sup
erc Poten
100
Wilson 1
50 Elastic
WD-1a
Normal deformation of
the fracture surface
NJ-11
Weakly stress dependent and
reversible permeability of the fracture
0 0
200 300
Temperature (C)
Figure 4 | Temperature and effective confining stress (depth) conditions
for the base of high-temperature geothermal reservoirs, and potentially
exploitable supercritical geothermal resources. Given that the
elasticplastic transition causes a gradual change from a
higher-permeability regime to a lower-permeability regime in the crust,
productive supercritical geothermal resources from 375 C to
approximately 460 C may form at effective confining stresses of
approximately 30100 MPa or corresponding depths of approximately
26 km, depending on temperature, even if the rocks are nominally ductile.
may be close to the transition stress, as observed in the other wells
(or Fig. 2f).
Nearly all previous models for crustal permeability have
invoked stress-independent, temperature-dependent permeability
behaviour as a key first-order control, with permeability decreasing
log linearly to a negligibly small value (for example, 1022 m2 )13,3032
above the BDT temperature. Such conditions preclude potentially
exploitable resources at temperatures higher than the BDT
temperature. However, the experimentally observed elasticplastic
transition suggests that the BDT is not a key first-order control on
permeability, and that potentially exploitable resources may occur
at temperatures from 375 C to approximately 460 C even in the
nominally ductile granitic crust at depths of approximately 26 km
(Fig. 4). This finding provides a scientific basis for planning the
ongoing Japanese national geothermal research and development
project, undertaken in light of the COP21 Paris agreement, by
expanding the possible environments for geothermal-resource
exploration into the high-temperature ductile crust.
The field-scale spatiotemporal evolution of permeability will be
a key determinant of whether a potentially exploitable supercritical
geothermal resource is indeed exploitable, and it is always
challenging to assess the implications of short-term lab-scale
experiments for longer-term field-scale behaviour. At elevated
temperatures and pressures, the field-scale permeability structure
will reflect a highly dynamic competition between permeability-
creating processes such as tensile fracturing (and subsequent
shear offset in some cases) and permeability-destroying processes
related mainly to waterrock interaction. Although adequate
porosity and permeability can be thermally created in granite
at high temperature and atmospheric pressure33,34 , the extent
of porosity and permeability creation by thermal and hydraulic
fracturing at in situ conditions remains unclear. Moreover, field-
scale permeability loss is difficult to predict. The nature and extent
of waterrock interaction depends on in situ pressure, temperature,
and fluid and mineral chemistry. The resulting net permeability
change will reflect complicated competition between permeability
loss and gain by processes such as mineral precipitation, removal of
bridging asperities and etching within fractures15,35,36 .
4 NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience
NATURE GEOSCIENCE DOI: 10.1038/NGEO2879
12 Methods
Plastic Methods, including statements of data availability and any
Strongly stress dependent and
associated accession codes and references, are available in the
online version of this paper.
9
Received 27 September 2016; accepted 13 December 2016;
published online 23 January 2017
riti
Depth (km)
t
cal ially
ge exp
oth lo 6
Th erm itab
eb al r le References
a
ge se of eso
urc
oth
erm high- es 1. Fournier, R. O. The transition from hydrostatic to greater than hydrostatic fluid
al r te m pressure in presently active continental hydrothermal systems in crystalline
ese er p
San Vito 1 rvo atu rock. Geophys. Res. Lett. 18, 955958 (1991).
irs re 3
2. Elders, W. A. et al. Rhyolite magma drilled in Iceland could power the worlds
San Pompeo 2 1 ord
reduc er of magn hottest enhanced geothermal system. Geotherm. Res. Trans. 34, 765770 (2010).
tion in itu
perm de
eabilit 3. Elders, W. A., Frileifsson, G. . & Albertsson, A. Drilling into magma and the
y
implications of the Iceland Deep Drilling Project (IDDP) for high-temperature
geothermal systems worldwide. Geothermics 49, 111118 (2014).
400 500 600
4. Frileifsson, G. ., Elders, W. A. & Albertsson, A. The concept of the Iceland
deep drilling project. Geothermics 49, 28 (2014).
5. Ingason, K., Kristjnsson, V. & Einarsson, K. Design and development of the
discharge system of IDDP-1. Geothermics 49, 5865 (2014).
6. Scott, S., Driesner, T. & Weis, P. Geologic controls on supercritical geothermal
resources above magmatic intrusions. Nat. Commun. 6, 7837 (2015).
7. Heard, H. C. Transition from brittle fracture to ductile flow in Solenhofen
limestone as a function of temperature, confining pressure, and interstitial fluid
pressure Rock Deformation. Geol. Soc. Am. Mem. 79, 193226 (1960).
8. Byerlee, J. D. Brittle-ductile transition in rocks. J. Geophys. Res. 73,
47414750 (1968).
9. Engvik, A. K. et al. Magma-driven hydraulic fracturing and infiltration of
fluids into the damaged host rock, an example from Dronning Maud Land,
Antarctica. J. Struct. Geol. 27, 839854 (2005).
10. Manning, C. E. & Ingebritsen, S. E. Permeability of the continental crust:
implications of geothermal data and metamorphic system. Rev. Geophys. 37,
127150 (1999).
11. Okada, T. et al. Hypocenter migration and crustal seismic velocity distribution
observed for the inland earthquake swarms induced by the 2011 Tohoku-Oki
earthquake in NE Japan: implications for crustal fluid distribution and crustal
permeability. Geofluids 15, 293309 (2015).
12. Fournier, R. O. Hydrothermal processes related to movement of fluid from
plastic into brittle rock in the magmatic-epithermal environment. Econ. Geol.
94, 11931211 (1999).
13. Hayba, D. O. & Ingebritsen, S. E. Multiphase groundwater flow near cooling
plutons. J. Geophys. Res. 102, 1223512252 (1997).
14. Tsuchiya, N., Yamada, R. & Uno, M. Supercritical geothermal reservoir
revealed by a granite-porphyry system. Geothermics 63, 182194 (2016).
15. Moore, D. E., Lockner, D. A. & Byerlee, J. D. Reduction of permeability in
granite at elevated temperatures. Science 265, 15581561 (1994).
16. Watanabe, N., Hirano, N. & Tsuchiya, N. Determination of aperture structure
and fluid flow in a rock fracture by a high-resolution numerical modeling on
the basis of a flow-through experiment under confining pressure. Wat. Resour.
Res. 44, W06412 (2008).
17. Watanabe, N. et al. New -type relative permeability curves for two-phase
flows through subsurface fractures. Wat. Resour. Res. 51, 28072824 (2015).
18. Nara, Y., Meredith, P. G., Yoneda, T. & Kaneko, K. Influence of macro-fractures
and micro-fractures on permeability and elastic wave velocities in basalt at
elevated pressure. Tectonophysics 503, 5259 (2011).
19. Violay, M. et al. An experimental study of the brittleductile transition of basalt
at oceanic crust pressure and temperature conditions. J. Geophys. Res. 117,
B03213 (2012).
20. Violay, M., Gilbert, B., Mainprice, D. & Burg, J.-P. Brittle versus ductile
deformation as the main control of the deep fluid circulation in oceanic crust.
Geophys. Res. Lett. 42, 27672773 (2015).
21. Doi, N. et al. Genesis of the plutonic-hydrothermal system around quaternary
granite in the Kakkonda geothermal system, Japan. Geothermics 27,
663690 (1998).
22. Ikeuchi, K. et al. High temperature measurements in well WD-1a and the
thermal structure of the Kakkonda geothermal system, Japan. Geothermics 27,
591607 (1998).
23. Saishu, H., Okamoto, A. & Tsuchiya, N. The significance of silica precipitation
on the formation of the permeable-impermeable boundary within Earths crust.
Terra Nova 26, 253259 (2014).
24. Walters, M. & Combs, J. Heat flow regime in the Geysers-Clear Lake area of
Northern California, USA. Geotherm. Res. Trans. 13, 491502 (1989).
25. De Vivo, B. et al. The Campi Flegrei (Italy) geothermal system: a fluid inclusion
study of the Mofete and San Vito fields. J. Volcanol. Geotherm. Res. 36,
303326 (1989).
2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE GEOSCIENCE DOI: 10.1038/NGEO2879
26. Batini, F. et al. in European Geothermal Update: Third International Seminar of
E. C. Research and Demonstration Projects in the Field of Geothermal Energy
341353 (D. Reidel Publishing Company, 1983).
27. Cathelineau, M., Marignac, C., Boiron, M.-C., Gianelli, G. & Puxeddu, M.
Evidence for Li-rich brines and early magmatic fluid-rock interaction
in the Larderello geothermal system. Geochim. Cosmochim. Acta 58,
10831099 (1994).
28. Marini, L. & Manzella, A. Possible seismic signature of the - quartz
transition in the lithosphere of Southern Tuscany (Italy). J. Volcanol. Geotherm.
Res. 148, 8197 (2005).
29. Steingrmsson, B., Gumundsson, ., Franzson, H. & Gunnlaugsson, E.
Evidence of a supercritical fluid at depth in the Nesjavellir field. Proc. Fifteenth
Workshop Geotherm. Reservoir Eng. SGP-TR-130, 8188 (Stanford University,
Stanford, 1990).
30. Hasenclever, J. et al. Hybrid shallow on-axis and deep off-axis hydrothermal
circulation at fast- spreading ridges. Nature 508, 508512 (2014).
31. Weis, P., Driesner, T. & Heinrich, C. A. Porphyry-copper ore shells form at
stable pressuretemperature fronts within dynamic fluid plumes. Science 338,
16131616 (2012).
32. Weis, P. The dynamic interplay between saline fluid flow and rock permeability
in magmatic-hydrothermal systems. Geofluids 15, 350371 (2015).
33. Chaki, S., Takarli, M. & Agbodjan, W. P. Influence of thermal damage on
physical properties of a granite rock: porosity, permeability and ultrasonic wave
evolutions. Constr. Build. Mater. 22, 14561461 (2008).
34. Yang, S.-Q., Ranjith, P. G., Jin, H.-W., Tian, W.-L. & Ju, Y. An experimental
investigation on thermal damage and failure mechanical behavior of granite
after exposure to different high temperature treatments. Geothermics 65,
180197 (2017).
NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience
2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
ARTICLES
35. Polak, A., Elsworth, D., Yasuhara, H., Grader, A. S. & Halleck, P. M.
Permeability reduction of a natural fracture under net dissolution by
hydrothermal fluid. Geophys. Res. Lett. 30, 2020 (2003).
36. Yasuhara, H. et al. Evolution of fracture permeability through fluidrock
reaction under hydrothermal conditions. Earth Planet. Sci. Lett. 244,
186200 (2006).
Acknowledgements
The present study was supported in part by the Japan Society for the Promotion of
Science (JSPS) through a grant-in-aid for Specially Promoted Research (no. 25000009).
The authors would like to thank Toei Scientific Industrial Co., Ltd. for manufacturing the
experimental system.
Author contributions
N.W. conducted overall study design and N.T. supervised the study. N.W., T.N. and K.S.
conducted the experiments, analysed the results, and estimated the permeabilitydepth
relation. H.S., A.O. and N.T. summarized temperature and depth conditions at the base
of high-temperature geothermal reservoirs. N.W., H.S. and S.E.I. wrote the manuscript,
and all authors commented on the manuscript.
Additional information
Supplementary information is available in the online version of the paper. Reprints and
permissions information is available online at www.nature.com/reprints.
Correspondence and requests for materials should be addressed to N.W.
Competing financial interests
The authors declare no competing financial interests.
5
ARTICLES
Methods
Sample preparation. Cylindrical granite samples (diameter, 30 mm; length,
25 mm), containing either a single tensile fracture induced by the Brazilian test or
by splitting a 20-cm cubic notched granite block with a wedge, or multiple tensile
fractures induced by heating in an electric furnace at 570 C and atmospheric
pressure for 3 h, were prepared using Inada granite from Ibaraki prefecture, Japan
(Supplementary Fig. 2). Permeability of the sample containing the fracture induced
by a wedge was much higher those of the other samples because the rock was
completely split only when using the wedge. Cylindrical basalt samples (same
dimensions) containing a single tensile fracture were prepared from Genbudo
basalt from Hyogo prefecture, Japan. Before fracturing, Youngs modulus, porosity
and permeability at or near atmospheric pressure are, respectively, 5580 GPa,
0.660.70% and 1018 m2 for Inada granite, and 4050 GPa, 2.903.87% and
1017 m2 for Genbudo basalt. After fracturing, the porosity is increased by an
average of 0.63% for granite samples and 0.26% for basalt samples. Two of the
granite samples contained single fractures induced by the Brazilian test or by a
wedge, and the others contained multiple fractures. A preliminary experimental
investigation on granite samples with multiple tensile fractures revealed a relation
between fractional porosity () and permeability (k) in m2 represented by
log10 = 0.19 log10 k + 1.03. The two types of fractures are used to demonstrate that
the permeability behaviour observed for the granite samples is independent of
fracture type. The two types of rocks are used to demonstrate that the observed
permeability behaviour is not specific to our experimental system. The BDT
temperature for basalt is as high as 800 C19,20 , much higher than the temperature of
360 C for granite1,1214 , so that the basalt samples are mechanically stable at the
temperatures of the present experiments (350500 C). Thus, permeability
behaviour of the basalt samples should be consistent with the well-known
behaviour at room temperature.
Permeability measurement. An experimental system newly developed by the
authors was used for permeability measurement. The novelty of this system is the
use of a special tri-axial cell, which uses a high-viscosity plastic melt as a confining
fluid and a thin plastic film as a sleeve (Supplementary Fig. 3). The plastic melt is
composed of PEEK (polyether ether ketone), which has a melting point of 343 C
and a decomposition temperature greater than 538 C. Because of the high viscosity
of 350 Pa s even at 400 C, sealing of the PEEK melt is easy. The plastic film is
polyimide film (thickness: about 50 m), which has no melting point (that is, it
decomposes before melting), and a high decomposition temperature of >500 C.
Metallic sleeves are usually used in high-temperature fluid-flow experiments under
confining stress, because rubber sleevessuch as the silicone rubber sleeves that
are usually used in low-temperature fluid-flow experimentscannot be used due
to low decomposition temperatures. However, with a metallic sleeve it is difficult to
create a homogeneous distribution of confining pressure on the sample surface,
and therefore to prevent short-circuit flow between the sleeve and sample surface.
Use of the thin plastic film avoids such difficulties. This new type of tri-axial cell
can work at lower temperatures when a plastic with a lower melting point, such as
polyethylene, is used as a confining fluid.
The sample wrapped with the polyimide film is first placed inside a PEEK
cylinder within a tri-axial cell. A graphite gasket having a hole for fluid flow is
attached at both ends of the sample. A tube containing a PEEK rod, which is used
to inject PEEK melt, is attached to the upper part of the cell. The cell is then placed
within an electric furnace. At a small axial pressure of about 2 MPa maintained by a
hydraulic jack, the temperature is increased to a prescribed value, melting both the
PEEK cylinder and the rod. Temperatures are measured both inside and outside the
cell. A prescribed confining stress is then applied to the sample by adjusting the
2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE GEOSCIENCE DOI: 10.1038/NGEO2879
confining pressure through the PEEK melt; the axial pressure is slightly higher than
the confining pressure (that is, differential stress of about 2 MPa). The confining
pressure is controlled by injecting PEEK melt at a prescribed pressure through the
upper tube, using a metallic piston that is displaced by pumping silicone oil at
constant pressure. At the prescribed temperature and confining stress, steam is
injected into the sample at a constant pressure of 2 MPa using a syringe pump,
and exits the sample at atmospheric pressure, with the flow rate measured using the
syringe pump. The steam is generated from pumped liquid water and preheated to
the prescribed temperature before entering the sample. Sample permeability is
calculated using Darcys law based on the pressure, flow rate, steam viscosity, and
the cross-sectional area and length of the sample. Low-pressure steam of low
reactivity is used to minimize dissolution and precipitation effects on the
rock permeability.
Estimation of the permeabilitydepth relation for the continental crust (Fig. 3).
We calculate the permeability decrease from a near-surface value with increasing
depth (that is, a combination of effective confining stress and temperature), where
the effective confining stress is a difference between lithostatic stress (rock density:
2.7 103 kg m3 ) and hydrostatic pressure (fluid density: 1.0 103 kg m3 ), and
the temperature gradient is 25 C km1 (ground surface temperature: 20 C). The
near-surface permeability is 1 1015 m2 , an intermediate value from the literature
(1 1017 1 1012 m2 )10 . Permeability decreases based on equation (2) until the
depth at which effective confining stress is equal to the elasticplastic transition
stress, then decreases based on equation (3); the elasticplastic transition stress is
given by equation (1).
Temperature and depth conditions at the base of high-temperature geothermal
reservoirs (Fig. 4). The temperature and depth conditions at the base of selected
high-temperature geothermal reservoirs shown in Fig. 4 were determined based on
deep drilling data. Well WD-1a at the Kakkonda geothermal field, Japan,
penetrated the boundary between a hydrothermal convection zone and a heat
conduction zone at 3,100 m, and reached 3,780 m total depth21,23 . The temperature
at 3,100 m depth in WD-1a was 380 C based on logging, the homogenization
temperature of liquid-rich fluid inclusions, and Horner plots22 . For the other wells,
the bottom of the well essentially corresponds to the base of the high-temperature
geothermal reservoir. The total depth of the Wilson 1 well in The Geysers,
United States, is 3,673 m, and analysis of fluid inclusions in cuttings from 2,980 m
and below 3,350 m indicates a temperature range of 350400 C (ref. 1). A
temperature gradient of 92.6 C km1 and a measured temperature of 326 C at
3,353 m depth24 yields a bottomhole temperature of 356 C. Thus, the bottomhole
temperature of Wilson 1 seems to be in the range of 350400 C. For the well San
Vito 1 in Campi Flegrei Fields, Italy, one temperature estimate is >420 C at the
total depth of 3,000 m (ref. 1). However, extrapolation of well temperatures from
cores at 8032,865 m (ref. 25) results in a bottomhole temperature of 399 C. Thus,
the bottomhole temperature of San Vito 1 may be 399 C. The San Pompeo 2 well
at Larderello, Italy, was drilled to 2,962 m (ref. 27). The temperature was 370 C at
2,400 m depth and 400 C at 2,600 m depth26,28 , which suggests that the deep
temperature gradient and the bottomhole temperature may be 150 C km1 and
450 C, respectively; the bottomhole temperature of San Pompeo 2 may therefore
be in the range of 400450 C. The total depth of well NJ-11 in Nesjavellir, Iceland
is at 2,265 m, and bottomhole temperature is estimated to be >381 C by
temperature logging1,29 .
Data availability. The data that support the findings of this study are available
from the corresponding author upon request.
NATURE GEOSCIENCE | www.nature.com/naturegeoscience