GeothermicsVol. 24, No. 5/6, pp.

695-705, 1995
Pergamon Copyright© 1995 CNR
ElsevierScienceLtd
Printed in Great Britain. All rights reserved
0375-6505/95 $9.50+ 0.00

0375-6505(95)00028-3

ANALYSIS OF CASING FAILURES OF DEEP

GEOTHERMAL WELLS IN GREECE

E. CHIOTIS and G. VRELLIS

Institute of Geology and Mineral Exploration, 70 Messoghion Street, Athens 115 27, Greece
(Received April 1994; acceptedfor publication May 1995)

Abstract--The specific characteristics of the Greek geothermal fields on the volcanic islands of Milos and
Nisyros are described, with emphasis on the characteristics affecting well casing. Overpressure, high
temperatures and highly saline brines are typical features of the Greek fields. A new treatment of the
different load conditions along the casing string is proposed that takes into consideration the possible
formation overpressure, the axial thermal stresses and their effect on burst and collapse resistance. This
treatment is further applied to some typical wells that have experienced casing failure.

Key words: casing failures, drilling, Greece.

INTRODUCTION
This paper describes a study of the casing problems and failures observed over a period of about
15 years in deep geothermal wells in Greece. The observations refer to five wells drilled on the
island of Milos (MZ-1, MA-1, M-l, M-2, M-3) and two on the island of Nisyros (N-1 and N-2),
details of which are shown in Fig. 1. Both islands belong to the active Aegean Volcanic Arc. An
update on the geology of Milos is given by Fytikas (1989). Cataldi et al. (1982) described the
geology of the geothermal field of Milos and the technical problems of the five wells drilled
there. Mendrinos (1988) studied the behaviour of the field of Milos for various exploitation
strategies.
The geology of Nisyros has been extensively studied by numerous volcanologists; geological
references are cited by Vougioukalakis (1993). Koutroupis (1985) commented on the drilling
problems of well Nisyros-1. A detailed description of the drilling operations on Nisyros is given
in two unpublished reports by Geotermica Italiana (1983, 1984).

CHARACTERISTICS OF THE GEOTHERMAL FIELDS OF INTEREST FOR

CASING DESIGN
The fields of Milos and Nisyros are hot-water systems with brines of high salinity. The
insulating caprocks of the Nisyros field have formed by self-sealing, through the deposition of
geothermal minerals in fractures and pore spaces. A similar self-sealing situation is very likely in
the overpressured field of Milos. The temperatures recorded exceed 300°C. The geological
characteristics of the fields causing self-sealing and overpressure, the prevailing high tempera-
tures and the highly saline nature of the geothermal brines constitute the main problems in
casing design.
695
696 E. Chiotis and G. Vrellis

L ~ ~. ~ ~ ~ ~, £ ~

.... ~ ++ ~" " ~ ='-" . . . . . . '. . . . . -"_ .... ~_~3_____~

~.~.~2~:~ . . . . . . . . . . . . . . . . . . ,~..__ ~,, ,r7 . . . . . . . . . . . .

E ~

........................ -*"---rE
:7 +.......... 1 £ 0

.i. 0

,E

. . . . . . . . . . . . . . . . + . . . . . . . . . . . . . . . . .
r
.
. . . . . . . . . . 0~

c~

~[ ................... _ =:::::::::::~
 Casing Failures of Greek Geothermal Wells 697
Self-sealing and overpressure
In well Nisyros-1 the caprock in the depth interval from 1000 to 1500 m is composed of
crystalline limestone that has been isolated by self-sealing. In the same well but from 700 to 1000
m, this limestone is a reservoir rock. The inverse situation is met on Milos, where the geothermal
field is formed by a rock which is usually a caprock, namely the schists of the metamorphic
basement. This rock has been fractured due to the active extensive tectonism of the Aegean Sea
Region. This is a common phenomenon in this area and very often causes drilling fluid
circulation loss. Rock fracturing influences the quality of well cementing and the stability of the
casing.
The isolation and, thus, the overpressure resulting from self-sealing, can result in the
formation of phreatic craters, which are widespread on both islands. On Nisyros in particular,
there is evidence of a phreatic explosion that occurred during a seismic sequence in 1872-73.
Therefore, the possibility of blow-outs during drilling exists even at shallow depths. Overpres-
sure is also encountered at deeper levels. The wells on Miios were thus drilled using heavy mud
(specific weight 1.25-1.30 g/cm3). It is very likely that the deeper productive formation in well
Nisyros-1 is also overpressured. This is suggested by its chemical composition, which differs
significantly from that of the neighbouring upper productive horizon.

Temperatures
As revealed by Dench (1970) in New Zealand, the undisturbed temperature of the hot-water
geothermal systems follows closely the boiling water curve. This is true of the Greek fields as
well, as shown in Fig. 2, where the Horner plot estimations of the undisturbed formation
temperature are shown. Temperature estimates lie not far from the boiling curve for the brine
with 10% NaCI. As suggested by White et al. (1971), when the hot water rises the hydrostatic
pressure decreases, and eventually a level is attained where pressure is low enough for boiling to
begin. At this level there is a high potential for self-sealing by means of deposition of minerals in
outlet channels. This explains why the hot-water systems follow a water boiling curve with
depth.
Temperature measurements, taken a couple of hours after circulation ceased, are a valuable
source of information, although they can be considerably lower than the undisturbed formation
temperatures. Such measurements are shown in Fig. 3 for wells MZ-1, MA-1, N-1 and N-2. One
type of information derived from such graphs is the zones of cooling, temporarily and artificially
produced by cooler mud filtrate or water invasion into productive formations. Three such zones
can be distinguished in well MA-1, centred at 650, 900 and 1100 m respectively. Another two
zones each are clearly shown in wells MZ-1 (550 and 800 m) and N-2 (1100 and 1300 m).
Temperature measurements in well N-1 suggest a strong temperature depression in the depth
interval from 350 to 1000 m. Measurements taken long after drilling confirmed that this is a
permanent thermal anomaly. It is interesting to note that hydrothermal minerals in this zone
suggest an original temperature of more than 200°C, i.e. much higher than the 90°C measured by
Geotermica Italiana (1983). Apparently this is a zone of seawater invasion of seismo-tectonic
origin. Its influence can be significant on the thermal stresses that developed in the well casing
during production, as explained below.
Additionally, temperature plots such as that of Fig. 3 can serve as a basis for estimating the
casing neutral temperature, which is a critical parameter for evaluating the thermal stresses in
the casing.

The chemical characteristics of the geothermal fluids

The geothermal brines of Milos and Nisyros fields are acid and corrosive, abundant in CO 2
and H2S. Their pH is about 4. The average content of NaCl in the brine of the Zephyria field on
698 E. Chiotis and G. Vrellis
Temperoture

100 ZOO 300 oC

VAPOR

H~
t-
Q. 1000 LIQUID
4)
0

H~

m I

Fig. 2. Estimated undisturbed temperature of the productive formations of geothermal wells on Milos and Nisyros
compared with the water boiling curve calculated after Hass (1971) for 0, 10 and 25% NaC1.

Milos is 75,000 ppm (Mendrinos, 1988), although the content can be much higher on Nisyros.
During the first production test from the deeper production zone in well N-l, a vapour flow
deposited about 100 tons of solid salts, composed mainly of chlorides, in six hours. A second test
produced a brine with 100-300 g of chlorides per litre (Geotermica Italiana, 1983). The brine/
steam mass ratio is in general high; for example in well N-2 it was about 1:1 for a wellhead
pressure of between 6 and 15 bar.
The gas/steam ratio in well N-1 was 1.5-1.7%. Out of this gas, 2.5-4% was H2S and almost all
the rest CO2. The average gas/steam ratio in well N-2 was 4.8%, the gas having about the same
percentage of H2S as in well N-1. The brine produced in well N-2 contained about 20 ppm H2S
compared to 400 ppm at reservoir conditions.
The silica in the produced brine can be high enough to cause serious scaling problems. The
content of SiO2 was about 1000 ppm in the brine of well M-1 and 10 ppm in the steam
(Mendrinos, 1988).
From the above description it is concluded that conditions are worse in the water-dominated
Greek geothermal fields than in the typical vapour-dominated fields of Larderello in Italy and
The Geysers in California. These vapour-dominated fields produce dry or superheated steam
 Casing Failures of Greek Geothermal Wells 699
Temperat.u re

100 200 300 °C

! I !

.%.
".:.t~. ~,,~_
.

\ %\ \',,,.
500
I i '\ /
\ '\ ,C ,,
5O.
II)
r~ ~ 'l"~ . "-.. ~"~.
x\.. \\ -MZ-1
1000 ~"~<._/i / ,/" M A - 1

1500 """'~'- N-2 " ~ N-1

m

Fig. 3. Temperature measurements, 10-20 hours after circulation stopped in wells MZ-1 and MA-1 on Milos (after
Kouris et aL, 1976) and N-1 and N-2 on Nisyros (after Geotermica ltaliana, 1983, 1984).

with no associated liquid, their pressure is well below the hydrostatic and they tend to have an
initial temperature near 240°C (White, 1973).

CASING FAILURES

Some typical casing failures observed in the Greek wells are summarized below.
• Wellheads of MA-1 and MZ-1 rose by 0.5 m during production tests because of thermal
expansion and inadequate cementing of the casing.
• Casing joints decoupled in well M-1 (9~ in. csg at 69 m) because of thermal stresses.
• The 9~ in. (24.4 cm) casing in well N-1 was seriously damaged during the first production test
in the intervals 148-149.5 m, 305.7-307.2 m, 350-351 m, 992.7-994 m, 1208.6-1214.9 m and
1235.2-1236.5 m. Because of these buckled sections, a 7 in. (17.8 cm) tieback casing was set
down to 1258 m (Fig. 1).
• This tieback casing in N-1 failed later due to collapse and possible buckling.
• Well M-2, which has remained shut-in for 18 years, recently had serious leakages at the
wellhead associated with casing corrosion.

Obviously, these failures are at least partly associated with bad cementing. The question,
700 E. Chiotis and G. Vrellis
p ress u re pressure
0 C E Ps F Pf 0 C E Ps Pf

woter
"gradient
shoe
casing \ ~ l 'I,
well i well gradient
I botto m A B bottom
A B

depth depth
(a) (b)
Fig. 4. Casing internal pressure when ovcrpressurc, P~, is encountered at point A. CB: mud pressure when the wellhead
is closed and no m u d has been ejected. EDB: internal pressure when m u d is maintained only in the upper part of the well
above D. P~: steam pressure. Maximum wellhead pressure: (a) F in Fig. 4a, where P~ - Ph > P, and (b) P, in Fig. 4b,
where P~ - t' h < P,.

however, is whether better casing design could reduce or eliminate these problems. Therefore,
the characteristics of the casing strings of two typical wells were analysed.

CASING DESIGN
Casing strings for the Greek geothermal wells can be designed assuming that formation
temperature approaches the boiling curves for brines with 10-25% NaCI. Furthermore,
overpressure is expected, and cannot be predicted in advance. Another critical condition is that
high temperatures and large temperature changes are expected during production or well
cooling. Therefore, casing strength reduction at high temperatures and the development of
thermal stresses should be taken into account. For this reason, the neutral temperature of the
casing, i.e. the temperature at which the casing is axially unstressed, should be accurately
estimated. The approximation suggested by Dench (1970), that the neutral temperature can be
taken as half of the well-bottom temperature, is somewhat vague.

Internal pressure
The maximum expected internal pressure on the casing depends on:
• the pressure of the overpressured formation, Pr
• the hydrostatic pressure, Ph, of a free column of geothermal water equal to the well depth
• the steam pressure, P~.
Two alternative cases are possible. If Pe - Ph > ~ , as in Fig. 4a, at the moment that the
overpressured horizon is drilled and the well is closed, the internal pressure versus depth will
jump from line OA to line CB. If part of the mud column is then ejected and replaced by
geothermal water up to the point D, the internal casing pressure will be that along the line EDB.
Obviously, the most severe situation will be when all the mud is replaced by geothermal water,
since the internal pressure follows the line FB in this case.
If Pr - Ph < P~ then, as soon as the overpressured formation is drilled and the wellhead is
closed with no loss of mud, the internal pressure increases from line O A to CB. If the mud is now
fully replaced by geothermal water, the upper part of the well above point F will be full of steam
 Casing Failures o f Greek Geothermal Wells 701
and the remaining part full of geothermal water. Internal pressure will then be that along the line
PsFB which is the most severe possible situation.
It is noted that the steam pressure in the well was previously assumed constant versus depth.
This is an acceptable simplification, since the gap between the wellhead pressure of a steam
column and the saturation pressure at depth, as given by Dench (1970), is small.

Neutral temperature
The casing neutral temperature, Tn, depends on two factors:
• the well temperature, Tc, during the hydration period of the cement as it solidifies
• the axial stress, (7, in the casing due to its weight plus any mechanical tension applied at the
wellhead.
The temperature increase that would release the axial tension, (7, is ATo
(7
ATo - (1)

where fl is the linear thermal expansion coefficient of steel, and E is the elasticity modulus of
steel.
Thus
(7
T,, = T~ + ATo = Tc + fl--~ (2)

The well temperature, Tc, can be approximated by the sum of the well temperature, T,
measured a couple of hours before cementing plus the temperature change during the cement
hydration, ATc, which is about 38°C (100°F) (Shryock, 1982). Finally
(7
T, = T + ATc + fl---~ (3)

The effect o f thermal stresses on casing radial strength

If the temperature of the cemented casing differs from the neutral temperature by AT,
thermal stress ot is induced
0 t = flgAT

For the commonly accepted values of

fl = 12.4 x 10-6/°C and E = 2 x 106 bar (4)
it is
o, = 24.8 a T (5)

where AT is in °C and at in bars.

This thermal stress, at, is compressive if the casing temperature exceeds the neutral one, and
tensional if the casing temperature is lower. Compared to the thermal stresses, the remaining
axial stress, such as the tension due to the casing weight, buoyancy and mechanical tension, can
be neglected. It is well known that axial tension reduces the normal (zero axial stress) collapse
resistance Pc to/'ca. It is (Dwiggins, 1983)
Pc =Pca[ X/-0.75(o/Yp) 2 - 0.5a/Yp] (6)
702 E. Chiotis and G. Vrellis
where Pc is nominal collapse resistance, Pea is reduced collapse resistance under axial tension, o
is axial tension, and Yp is yield point.
This relationship also holds for the reduction of the burst resistance under axial compression.
In this case the reduced burst resistance is calculated as a function of the nominal burst resistance
and the ratio o/Yp.
The effect of axial loading on the burst and collapse resistance is numerically very significant,
as will be shown by specific examples.
In addition, the axial thermal stresses may cause plastic deformation, which should be
avoided. This limitation implies a definite temperature range from T m i n = Tn - ATmin t o T m a x =
Tn + ATmax, outside of which the casing will be plastically deformed. Thus, when the casing is
cooled, the minimum temperature at which deformation is elastic is (T n - ATmi,), where

ATmin = ~ (7)
fie
It is clarified that Tmi,, as calculated above, provides the lowest temperature at which the
deformation due to thermal tension is maintained elastic. In casing design, however, a safety
factor must be taken into account in the denominator in equation (7), which normally varies
from 1.6 to 1.8 for tension.
Similarly, when casing is heated, the maximum temperature increase, ATmax, above the
neutral temperature without exceeding the yield point is ATmax

A Tmax = RIRaYp (8)

fie
where the coefficient R 1(RI > 1) accounts for the increase of the yield point under compression
and Rz(R 2 < 1) for the yield point reduction at high temperatures. As a first approximation it is
suggested that the following coefficients can be taken:
RI ~ 1.1 and R2 ~-0.85 (9)

APPLICATION FOR TYPICAL WELLS

Well MZ-1 (7 in. casing)
A number of temperature curves are shown in Fig. 5 for the 7 in. (17.8 cm) casing of well MZ-I:
• temperature measurements shortly before cementing, T
• temperature increase, ATe, during cement hydration
• temperature increase, z~To, which relaxes the tension due to casing weight
• calculated neutral temperature, Tn
• marginal temperatures Tminand Tmax for elastic deformation.
The main component of the neutral temperature, T,, is obviously the well temperature before
cementing. The effect of the casing weight (J-55, 26 lb/ft; 39 kg/m) is marginal.
As shown in Fig. 5, the 7 in. casing of well MZ-1 is sensitive to cooling, whereas plastic
deformation might be expected only at the wellhead zone during production.
Plastic deformation during cooling can be avoided through well cooling before casing
cementing and by using the higher casing grade C-75, which corresponds to A T m i n = 208°C.
The effect of compression during cooling on the collapse resistance can be very crucial in the
case of steam production immediately after cooling. Assuming cooling at 100°C at the casing
shoe (758 m)
AT= T,- T = 195°C
 Casing Failures of Greek Geothermal Wells 703
Temperature

100 200 300 ,~O0 oC

xX zx T o ' '

250
.IT.

CJ
500

Trniff ~ T
750

f~

Fig. 5. Well MZ-1. Minimum (Train) and maximum (Tmax) permissible temperature for elastic thermal deformation of
the casing. T: well temperature a few hours after circulation stopped. ATc: temperature increase during casing
cementing. AT,,: temperature increase required for compensation of initial axial tension due to casing weight. 7",:
neutral temperature.

at = 24.8 AT = 4836 bar = 70,160 psi

o/Yp = 70,160 : 55,000 = 1.28
For these data, according to equation (6), the collapse resistance of the J-55 casing is practically
reduced to zero. Even the C-75 grade is inadequate to resist the expected external pressure of 93
bar, corresponding to a 758 m mud column of specific gravity 1.25 g/cm3; only the C-95 casing (23
lb/ft; 34 kg/m); would be satisfactory. It is noted that the reduction of collapse resistance during
cementing, due to the tension caused by the casing string weight, is much less than that resulting
from cooling and is mainly restricted to the upper part of the string.
At least an indicative value of formation pressure is required to estimate the expected internal
pressure. Assuming an overpressured formation balanced by a mud column of specific weight
1.25 g/cm 3, it is found that
P f - Ph < Ps
Consequently, there can be a steam column in the well and the maximum internal pressure
expected at the wellhead is Ps ~ 1150 psi (79 bar) for a well that is 1100 m deep. For a wellhead
flowing temperature of T ~ 240°C this is
AT = T- Tn ~ 140°C
O
- 1.08
RzYp
Thus, according to equation (6) applied this time for burst, the burst resistance of the J-55 casing
is practically zero and grade C-75 casing with 23 lb/ft (34 kg/m) is required.
704 E. Chiotis and G. Vrellis

Temperoture

0 I00 200 300 400 oc

'

¢'1

121
1000

1500r
fn
Fig. 6. Well N-1. Explanationsas in Fig. 5. Tf : well flowingtemperature.

Well N-1 (9~ in. casing)

The estimated temperature limits for elastic axial deformation, Tmi, and Tmax, for this casing
are shown in Fig. 6. It is inferred that the temperature inversion between 350 m and 1000 m
aggravates the thermal stress, in particular during production. The 9~ in. (24.4 cm) casing (J-55,
40 lb/ft; 60 kg/m) is found to be inadequate to withstand axial thermal stress as well as internal
and external pressure. As a matter of fact, the numerous casing failures already mentioned were
caused initially by sudden heating during the production test, undertaken without any prep-
aration for gradual heating, and later while killing the well after the first test by injecting 32.5 m 3
of seawater within 25 minutes.

CONCLUSIONS

The major casing failures observed in the Greek geothermal wells were caused by thermal
stress while producing or cooling the well. They could have been avoided through a combination
of slow preheating before production and proper casing design.
Estimates of the neutral temperature are very critical for calculation of the thermal stress and
should be based on well temperature measurements shortly before or after cementing. This
would ensure that the estimated temperature range for elastic deformation is reliable. Casing
should be designed so that no plastic deformation takes place.
Burst and collapse strength are severely reduced by axial thermal stress, which should be
taken into account in the selection of the appropriate casing grade.
The probability of drilling into overpressured formations is another factor that must be
considered in casing design for geothermal wells in Greece.

Acknowledgements--The authors would like to thank two anonymousrefereeswhosesuggestionsimprovedthe original

manuscript significantly.
 Casing Failures o f G r e e k G e o t h e r m a l Wells 705

REFERENCES
Cataldi, R., Del Gaudio, P., Neff, G., Rossi, U., Toneatti, R., Koutroupis, N., Michailakis, C. and Vondichakis, M.
(1982) Milos geothermal project (Greece) : exploration, drilling and preliminary production data. In Proc. Int. Conf.
Geoth. Energy, Florence, pp. 97-111.
Dench, N. D. (1970) Casing string design for geothermal wells. U.N. Symp. Devel. Util. Geoth. Res., Pisa. Geothermics,
Spec. Iss. 2, 2, 1485--1496.
Dwiggins, L. J. (1983) Casing and drill pipe. Workshop on high temperature geothermal wells: planning, drilling and
completion, Geothermal Resources Council, pp. 180-240, Reno, NV.
Fytikas, M. (1989) Updating of the geological and geothermal research on Milos Island. Geothermics 18, 485--496.
Geotermica Italiana (1983) Nisyros 1 geothermal well. Final (unpublished) report, Pisa.
Geotermica Italiana (1984) Nisyros 2 geothermal well. Final (unpublished) report, Pisa.
Hass, J. L. Jr. (1971) The effect of salinity on the maximum thermal gradient of a hydrothermal system at the hydrostatic
pressure. Econ. Geol. 66,940-946.
Kouris, D., Vondikakis, E. and Ghinis, A. (1976) Geothermal wellson Milos. Unpublished report, Public Power Corp.,
Athens.
Koutroupis, N. (1985) Geothermal exploration in the island of Nisyros. Nisyros I geothermal well. In European
Geothermal Update (Edited by Strub, A. and Ungemach, P.), pp. 440 ~5. Reidel, Dordrecht.
Mendrinos, N. (1988) Modelling of Milos geothermal field. M.Sc. Thesis, Univ. of Auckland, 196 pp.
Shryock, S. H. (1982) Geothermal cementing. In Handbook of Geothermal Energy (Edited by Edwards, L. M.,
Chilingar, G. V. et al.), pp. 310-325. Gulf, Houston, TX..
Vougioukalakis, G. (1993) Volcanic stratigraphy and evolution of Nisyros Island. Bull. Geol. Soc. Greece 28(2), 239-
258.
White, D. E. (1973) Characteristics of geothermal resources. In Geothermal Energy (Edited by Kruger, P. and Otte,
C.), pp. 69-94. Stanford University Press, Stanford, CA.
White, D. E., Muffler, L. J. P. and Truesdeil, A. H. (1971) vapor-dominated hydrothermal systems compared with hot-
water systems. Econ. Geol. 66, 75-97.