Journal of Petroleum Science and Engineering 107 (2013) 1830

Contents lists available at SciVerse ScienceDirect

Journal of Petroleum Science and Engineering

journal homepage: www.elsevier.com/locate/petrol

A radiofrequency/microwave heating method for thermal heavy oil

recovery based on a novel tight-shell conceptual design
Matteo Bientinesi a,n, Luigi Petarca b, Alessandro Cerutti c, Mauro Bandinelli c,
Michela De Simoni d, Matteo Manotti d, Giuseppe Maddinelli d
a

Consorzio Polo Tecnologico Magona, via Magona, Cecina (LI) 57023, Italy
Dipartimento di Ingegneria civile e industriale, Universit di Pisa, Largo Lucio Lazzarino, Pisa 56124, Italy
c
Ingegneria dei Sistemi, via Enrica Calabresi 24, Pisa 56121, Italy
d
Eni, Exploration and Production Division, via Emilia 1, San Donato Milanese (MI) 20097, Italy
b

art ic l e i nf o

a b s t r a c t

Article history:
Received 10 October 2011
Received in revised form
8 February 2013
Accepted 19 February 2013
Available online 30 April 2013

The ongoing depletion of light oil resources and the increasing global energy demand is driving oil&gas
companies towards the exploitation of unconventional oil resources. In order to extract crude oil from
these resources, a sufciently low oil viscosity must be achieved, for instance through temperature
increase. Electromagnetic irradiation through downhole antennae can be a suitable method for in situ
heating of reservoirs. Potential problems for this technique are the extremely high temperatures that can
be reached at the well containing the radiating element and the strong dependence of temperature
proles on local variation of reservoir material properties. These problems can be solved to a large extent
by inserting around the radiating well a tight shell made of a low loss dielectric material, and by selecting
the proper irradiation frequency.
The experimental work described in this paper aims to verify the effectiveness of a similar structure
during the electromagnetic heating of over 2000 kg of oil sand in a sandbox up to 200 1C, using a dipolar
antenna. Oil sand was irradiated at 2.45 GHz frequency with variable power (12 kW). The temperature
in the oil sand mass and on the boundary were recorded throughout the test in several specic points, in
order to estimate temperature proles along the distance from the antenna.
Experimental results conrmed that the presence of the low lossy material shell realized around the
antenna is extremely efcient in lowering the temperature in this critical zone and in better distributing
the irradiated energy in the oil sand mass.
& 2013 Elsevier B.V. All rights reserved.

Keywords:
enhanced oil recovery
thermal heavy oil recovery
electromagnetic heating
radiating antenna
oil sand
modeling

1. Introduction
The ongoing depletion of light oil resources and the increasing
global energy demand is driving oil&gas companies towards the
exploitation of unconventional oil resources. These include viscous
crude oil reserves (heavy and extra-heavy oils), oil sand and oil
shales. The International Energy Agency (IEA) estimates that there
are about 6 trillion barrels of such resources in place worldwide
(Clark, 2007).
Due to high viscosity, heavy oils generally require enhanced oil
recovery (EOR) techniques to be produced economically. Thermal
recovery methods, which introduce heat into the reservoir to
reduce oil viscosity, are effective techniques to enhance heavy oil
productivity. Steam injection processes namely cyclic steam
stimulation (CSS), steam ooding (SF) and steam assisted gravity
n

Corresponding author. Tel.: +39 586 632142; fax: +39 586 635445.
E-mail addresses: [email protected] (M. Bientinesi),
[email protected] (L. Petarca), [email protected] (A. Cerutti),
[email protected] (M. De Simoni).
0920-4105/$ - see front matter & 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.petrol.2013.02.014

drainage (SAGD) are the main thermal methods currently

employed commercially. However, steam assisted oil recovery
requires large amounts of fresh water, cannot be used for deep
or very shallow reservoirs, and its effectiveness depends on
reservoir geological properties. The economical protability or
even the technological feasibility of these methods can be greatly
reduced in thin payzones, in low permeability formations, or in
presence of a weak sealing cap rock (Clark, 2007).
Electromagnetic (EM) irradiation, at radiofrequency (RF) or microwaves (MW) frequencies, can be a sound alternative for in situ
heating of unconventional reserves. A RF/MW heating method, based
on a downhole radiating antenna, is less affected by formation
geology and is capable to distribute heat over a large reservoir
volume thanks to the propagation of electromagnetic energy through
the medium. Other advantages are equipment compactness (suitable
for off-shore elds), high efciency in the energy generationradiation process and the possibility to focus the energy on oil
bearing strata, reducing heat losses through the overburden.
Many industrial patents have been registered about this topic by
several companies in the last 60 years. Some examples are: Ritchey

M. Bientinesi et al. / Journal of Petroleum Science and Engineering 107 (2013) 1830

(1956), Haagensen (1965), Kasevich et al. (1979), Bridges et al.

(1979), Kiamanesh (1992), Kasevich (2008). Such patents describe
set-ups based on the introduction of an electromagnetic power
emitting apparatus in a well and differ from each other for
characteristics such as operating frequency, apparatus design and
well conguration.
In spite of the high number of patents, really few data were
published demonstrating the applicability of this technique and
assessing its efciency. In particular, three main lab-scale tests were
presented: Sresty et al. (1986) veried the heating up to temperature
between 150 1C and 350 1C of a 300 kg Utah oil sand sample,
irradiated through a triplate line system; Kasevich et al. (1994)
irradiated a small sample of heavy oil through an antenna, while Hu
et al. (1999) used two electrodes placed into a poly(methyl methacrylate) (PMMA) box to heat with radiofrequency (RF) a reconstructed oil
sand sample. However, none of these authors show a temperature
prole arising from electromagnetic irradiation, and little attention has
been paid to the temperature reached in the vicinity of the wellbore
during the irradiation process. This is one of the key issues of a RF/MW
reservoir heating technique. High reservoir heating rates require high
EM power irradiation by the downhole antenna. This can turn out in
an extremely high EM eld density located in the volume surrounding
the radiating element. Reliable RF/MW heating processes must therefore take into account the EM energy distribution through the
reservoir and must be designed to achieve a volume heating as
uniform as possible. This is a key factor to prevent the exposure of
well completion components to extreme temperatures, while irradiating high EM power rates into the reservoir.
In order to meet the requirements addressed above, in the
present work a new RF/MW method, which combines a downhole
antenna with an interface structure realized between radiating
well and reservoir (called tight shell), is proposed. A theoretical and
experimental study has been conducted to evaluate the potentiality of the proposed solution and to design a RF/MW method
capable to operate a high power, long term irradiation process,
required to heat a considerable volume of reservoir.
In this paper we describe the novel tight shell conceptual
design, assessed by means of a simple numerical model and an
experimental test. The objectives are to assess the effectiveness of
the novel tight shell conceptual design and to study the different
phenomena taking place in a reservoir during electromagnetic
irradiation and their effects on the application.

2. Design of the system and preliminary modeling

The new system design, described in what follows, was rst
analyzed with preliminary numerical simulations.
In order to achieve short calculation times and a device as
exible as possible, a simple numerical model was developed (as
described in Appendix A) assuming static conditions (no ow is
described) and spherical symmetry. This allows us to perform a
wide spectrum of simulations in order to analyze:

reservoir up to temperatures high enough to mobilize hydrocarbons at relatively long distances from the well and, in order to
simplify the thermal resistance requirements for well completion
materials, limiting near wellbore temperature. In another way, the
system aims to obtain temperature proles versus the distance
from the radiating element as uniform as possible.
Fig. 1 shows one possible embodiment of the conceptual design
of the new RF/MW method, in which a radiating antenna is
coupled with a well-reservoir interface structure (tight shell)
whose scope is to limit temperature increase at the well. The
system is composed by:

a production well (whose completion scheme is specially

designed in order to host the RF/MW components and to allow

EM irradiation);
a high power RF/MW energy applicator (composed by a surface
unit with a high power RF/MW energy source, a downhole
transmission line and a bottom hole antenna);
a tight shell (a spherical or more likely cylindrical structure
interposed between oil well and reservoir, realized at the
antenna installation depth through drilling and completion
operations; the tight shell is made of a low loss dielectric
material and is impermeable to reservoir uids).

The same conceptual design can be applied also in congurations where the RF/MW irradiation well is separated from the
producer.
2.2. Preliminary numerical simulations
In order to be feasible, the electromagnetic irradiation technique must allow to heat large volumes of reservoir up to a
temperature at which oil viscosity is low enough to be produced.
Different heavy oils require different temperatures, varying from
70 1C to over 200 1C; in the preliminary study, we considered the
limit of 150 1C, over which the oil from the analyzed oil sand
samples has a viscosity lower than 100 mPa s. At the same time,
maximum temperature at the radiating well must be kept at a
level tolerated by well completion materials; we considered 350 1C
as the maximum temperature at the well for the present study.
Numerical simulations, performed varying a number of settings
such as reservoir material properties, irradiation frequency and
power, tight shell diameter, show that the operating system parameters have important effect on the thermal process, and the desired
uniform heating can be achieved only with a proper system design.
In particular, the following general observation can be stated.

Even changing reservoir material properties and system con-

(1) the optimal operating frequencies to achieve homogeneous

reservoir heating;
(2) the impact of reservoir dielectric and thermal properties and
of operating conditions;
(3) the effectiveness of the tight shell solution and its design
parameters.

2.1. Novel tight shell system design

In order to optimize the electromagnetic heating process, a
novel system was designed, with two main objectives: heating the

19

guration, best irradiation frequencies turn out to be those in

the 1020 MHz range. Higher frequencies are as well capable to
heat deep into the reservoir, but temperature at wellbore
results too high. This is shown clearly by the sensitivity analysis
in Fig. 2, which describe the temperature reached at two
different distances from the antenna (0.15 m and 10 m) after
1000 and 2000 days of irradiation at 200 kW for frequencies in
the range 103000 MHz. Two temperatures are outlined in
Fig. 2: 150 1C is indicatively the minimal temperature to be
reached at 10 m of distance from the antenna in order to
mobilize the oil, 350 1C is a limit for the temperature near the
wellbore in order to avoid structural problems. These limits can
change for different scenarios, in particular, several kinds of
heavy oil are mobile at much lower temperatures. Anyway, the
fact that lower frequencies give a better energy distribution in
the reservoir can be generalized.
The transient process of connate water evaporation in a volume
surrounding the radiating antenna can remarkably reduce

20

M. Bientinesi et al. / Journal of Petroleum Science and Engineering 107 (2013) 1830

Fig. 1. Conceptual design of a production/radiating well with a spherical tight shell.

10000
r = 0.15 m; t = 200 day
r = 0.15 m; t = 600 day
r = 0.15 m; t = 1000 day

r = 10 m; t = 200 day
r = 10 m; t = 600 day
r = 10 m; t = 1000 day

Temperature [C]

1000
350C

100

150C
400
350
300
250

10

200
150
100
10

15

20

25

30

35

40

45

50

1
10

100

1000

Frequency [MHz]
Fig. 2. Frequencytemperature analysis for 200 kW irradiated power, 3 m radius tight-shell (tg() 103). The diagram shows the temperatures reached at 10 m distance
from the radiating well (solid lines) and near the wellbore (circle spotted lines) for different irradiation times, as a function of the transmitted frequency (in the range
103000 MHz). In the window, the range 1050 MHz is zoomed.

energy dissipation close to the well. This is due to the change in

dielectric properties of reservoir material, once the water has
been removed, as shown as a way of example in Fig. 3 for an oil
sand sample with an original water and oil content respectively
of 4.2% and 10% by weight. Fig. 3 shows how, after a slight
initial increase both of the real and imaginary part of the
dielectric constant with increasing temperature, led by the

increase of salt water electrical conductivity (Hayashi, 2003),

there is a fall of both quantities once the boiling point of
connate water (about 100 1C at atmospheric pressure for the
analyzed sample) is reached and water is evacuated. This
situation is representative of reservoirs in which both the oil
and the inorganic matrix are non-dispersive. Anyway, several
oil sand, oil shale or heavy oil dry samples show higher values

M. Bientinesi et al. / Journal of Petroleum Science and Engineering 107 (2013) 1830

20

21

60
T = 20C

18

T = 50C

50

T = 100C

16

T = 125C

14

40

''

'

12
10

30

8
20
6
4

10

2
0

0
10

100

10

1000

100

1000

Frequency [MHz]

Frequency [MHz]

Fig. 3. Measurements of the real (left) and the imaginary (right) parts of dielectric permittivity of an oil sand sample with composition: 10% oil, 4.2% water, 85.8% quartz sand
by weight. The measurement is taken in the 101000 MHz range at 20 1C, 50 1C, 100 1C and 125 1C, according to the methodology described in Sarri et al. (2012).

500
800
700

450

600

400

500
400

Temperature [C]

350

300
200

300

100
0

250

10

20

30

40

50

200
Tb = 160C

150
'' (dry) = 0.1, 2000 d

100

'' (dry) = 0.1, 1000 d

'' (dry) = 0.01, 2000 d

50

'' (dry) = 0.01, 1000 d

0
0

10

15

20

25

30

Distance from antenna [m]

Fig. 4. Simulation results: comparison between thermal proles obtained varying the dielectric properties of reservoir material after water evaporation, after 1000 and 2000
days of irradiation, without the use of the tight shell. Irradiation frequency is 10 MHz and emitted power is 200 kW. The water dried reservoir material has imaginary
dielectric constant 0.1 or 0.01. In the window, the full range 050 m is shown.

of imaginary dielectric constant, depending on oil composition

(presence of polar compounds) or crystallographic composition
of the matrix (Friso et al., 1998; Kovalyova and Khaydar, 2004;
Fuji et al., 1999; Saraev et al., 2005, Epov et al., 2009). Thus, we
analyzed two different scenarios (see Fig. 4): for a non-lossy
reservoir (0.01), the temperature near the wellbore is
slightly above the boiling temperature of water in the reservoir
(about 160 1C for the analyzed scenario), since a volume of
reservoir where water has been evaporated and electromagnetic energy is poorly dissipated soon forms and increase its
radius during the irradiation operation; for lossy reservoir (
0.1), water evaporation is not sufcient to keep down the
temperature, since even in the dry volume dissipation takes
place, even if at a lower extent. We can then conclude that
water evaporation contribute to limit the temperature rise

nearby the antenna, but the extent of this reduction depends

on the reservoir pressure, which in turn determines the boiling
temperature of connate water, and on the dielectric loss of the
water dried reservoir material, which can vary signicantly for
different reservoirs.
The heating process, realized without a tight shell interface, is
then very sensitive to possible local variation of the reservoir
dielectric properties. In favorable conditions (low lossy solid
matrix and heavy oil, and shallow reservoirs with depth lower
than 1000 m) water vaporization can be an effective natural
method to produce a uniform heating. Nevertheless the robustness of a RF/MW method that relies only on connate water
vaporization appears to be dependent on dielectric properties
of materials, which, in real eld operation, can be unknown or
affected by reservoir heterogeneity.

22

M. Bientinesi et al. / Journal of Petroleum Science and Engineering 107 (2013) 1830

of the experimental set-up, as described in Appendix A, and was

used to interpolate the data.

Simulation then focused on evaluating the effectiveness of the

new tight shell conceptual design in achieving a more uniform
volume heating, limiting extreme temperatures nearby the
antenna, and to make the RF/MW method less dependent on
reservoir material properties.
In Fig. 5, we compare temperature proles obtained with and
without the use of the tight shell (with 3 m radius) in the case of a
lossy reservoir ( 0.1 for the dried material). The irradiation
frequency is 10 MHz and the emitted power 200 kW. It is evident
that a few meters radius shell would be sufcient to keep down
well temperatures, also in presence of a lossy reservoir medium
(  0.1), in which case the temperature reduction caused by the
shell is dramatic.
Several simulations were performed, varying the loss tangent
(tg()) of the tight shell material between 104 and 103; in any
case, the tight shell turns out to be extremely successful in
lowering well temperature. Therefore, an effective tight shell
completion does not required a very transparent material (for
instance, quartz sand and several proppants used in oil well
applications, measured by Sarri et al. (2012), have tg() 103).

3. Experimental methodology and material

The lab test was performed irradiating a sand box, properly
lled with oil sand, by an antenna and recording temperature data
along predened sections. In what follows, the experimental setup is described.
3.1. Lab-scale experimental set-up
The experiment was performed with a 2.45 GHz radiation. This
experimental choice, having a shorter wavelength and a smaller
penetration depth with respect to lower frequencies, allowed to
reduce the oil sand amount needed, however providing reliable
experimental results.

2.3. Design of the experimental test

According to preliminary numerical results, the tight shell
turned out to be extremely effective in lowering the temperature
at the radiating well and in allowing to heat larger volumes of
reservoir. This conclusion led us to project and perform an
experimental test, with the major aim of evaluating the effective
decrease of the temperature near the wellbore arising from the
insertion of a low lossy material (tight shell) and the effect of
water evaporation on the development of temperature proles.
Even though the optimal irradiation frequency was located
between 10 MHz and 20 MHz, an experimentation at these frequencies would be hard to be performed with an antenna-like
emitting apparatus, if not directly on eld, in a pilot well. As a
consequence, we chose to operate in the microwave range, at
2.45 GHz. The smaller wave length and penetration depth allowed
us to scale down the volume of interest, while still observing the
dielectric heating process and the temperature proles developed
in the oil sand mass. The numerical model was adapted to the
radiation pattern of the experimental antenna and to the geometry

Fig. 6. Steel containment tank: the inner side of the walls are covered with radarabsorbent material sheets.

500
1000
No tight shell, 1000 d

450
800
400

No tight shell, 2000 d

3 m tight shell, 1000 d

600

Temperature (C)

3 m tight shell, 2000 d

350

400

300

200

250

0
0

10

20

30

40

50

200

Tb= 160C
150

100
50
r=3m

0
0

10

15

20

25

30

Distance from antenna (r)

Fig. 5. Simulation results: comparison between thermal proles obtained without the tight shell or using a 3 m radius spherical shell (constituted by a material with tg()
103) after 1000 and 2000 days of irradiation. Irradiation frequency is 10 MHz and emitted power is 200 kW. The water dried reservoir material has imaginary dielectric
constant 0.1. In the window, the full range 050 m is shown.

M. Bientinesi et al. / Journal of Petroleum Science and Engineering 107 (2013) 1830

The mass of oil sand used in the experiment has a base

dimension 1.25  1.25 m and a height of 0.75 m. The choice of
the dimensions depended on preliminary simulations results and
on the antenna design; the latter has in fact a dipole-like, nonspherical radiation pattern which extends more in the horizontal
direction than in the vertical one. To properly ll the space, the
samples were shaped in cubic blocks pressed into a mold of proper
shape. In order to press the blocks, oil sand material was heated up
to 70/80 1C, turning it malleable; each produced block weighed
between 28.5 kg and 32 kg, resulting in a void grade of 210%, with
an average value of 3%; 73 blocks were used, for a total amount of
over 2200 kg of oil sand material. The oil sand assembly was
conned by a steel containment tank (Fig. 6).
On the base of the tank, a 100 mm thick layer of quartz sand
was poured. On this layer, oil sand blocks were then assembled,
leaving between each block a 10 mm thick gap, lled again with
quartz sand. A vacant position was left in the center for the
insertion of the antenna (Fig. 7) and a cylindrical hole was realized,
in which the antenna was placed (Fig. 8). In the cylindrical hole,
two concentric polytetrauoroethylene (PTFE) tubes, closed at the
bottom, were installed. Outside the outer tube, a low lossy quartz
sand was poured, realizing a cylindrical shell. The gap between
outer and inner tube was lled with the silicone oil Rhodorsil Oil
47 V20 (Bluestar silicones). The inner tube contained a rigid
coaxial line, terminating with the antenna (Fig. 9).
The internal face and the base of the containment tank were
covered with sheets of radar-absorbent material (RAM, Eccosorb
SF2.5), glued to the metal with a room temperature vulcanizing
(RTV) silicone adhesive. The RAM absorbs all the electromagnetic
radiation reaching the tank walls, converting it into thermal
energy and avoiding reection of microwaves by the steel walls.
Fig. 10 shows the schematic of the experimental set-up. The
microwaves are generated by a water cooled magnetron with a
maximum output of 2 kW, fed by a switching power generator,
commanded via PC. Downstream the magnetron, an isolator
avoids the reected microwaves to come back to the magnetron;
the reected power is measured through a linear power sensor.
Microwaves travel through a waveguide from the control room
to the experiment room. Through an adapter, microwaves are
transferred into a rigid coaxial line, terminating in the antenna
situated in the exact center of the oil sand mass (Fig. 11).
The experiment room is thoroughly ventilated in order to
remove any gas possibly exhaled by the heated oil sand. However
operators were not allowed into the room throughout the
experiment.

23

Fig. 8. Realization of the cylindrical hole for the introduction of the antenna and
optical ber positioning.

Several ber optic temperature sensors were inserted in

selected oil sand blocks, in the quartz sand shell and in the silicone
oil between the two PTFE tubes. In addition, several K-type
thermocouples were used to monitor the temperature of the outer
surface of the steel walls.
The position of the various sensors is illustrated in Fig. 12,
where a section of the experimental set-up, at the height of the
center of the antenna, is depicted. Fiber optic sensors (FO##,
where ## is a number) and thermocouples (TC##) recorded the
temperature continuously. TCx (x being a letter) indicates the
thermocouples located in the gap between different oil sand
blocks, which periodically measured the temperature at different
depths, after temporarily stopping the irradiation.
The position of the ber optic sensors was dened in order to
record the temperature proles versus the distance from the
antenna, at the height of the antenna itself, where the radiating
power is at its maximum. The ber optic sensors provided the
temperature prole in the main direction A and allowed to
conrm the individuated trend on two other directions (B and C,
as dened in Fig. 12).
In order to obtain a good contact between the probes and the
oil sand material, once a sensor had been inserted in the predrilled block, oil (originating from the same oil sand samples) was
poured (at a temperature of 80 1C) in the hole in order to assure
the immobilization of the probe and the thermal equilibrium.
The thermocouples attached to the outer side of the tank walls
were used to qualitatively evaluate the thermal energy dissipated
by natural convection towards the surrounding environment.
Table 1 denes the position of each temperature probe,
assuming (Fig. 12) an xyz coordinate system with the origin at
the center of the antenna.
3.2. Antenna design and testing
The project of the antenna was developed through four stages:
requirements analysis, antenna type selection, antenna simulation
and individuation of mechanical devices for impedance adjustment and return loss minimization.
Requirements analysis lead to the individuation of the following specications:

Fig. 7. Assembling of the oil sand cubic blocks. Noteworthy, the bottom layer is
complete while the central layer, as well as the not shown top layer, has a vacant
position in the center for the introduction of the antenna.

geometrical specications (maximum diameter 60 mm);

operating frequency 2.45 GHz;
return loss lower than 10 dB;
radiation diagram optimized for power radial distribution.

24

M. Bientinesi et al. / Journal of Petroleum Science and Engineering 107 (2013) 1830

The antenna was designed through simulations with a modeling

tool developed by IDS (Ingegneria dei Sistemi), considering the
environment in which the antenna is inserted for the experiment
(PTFE tubes, silicone oil, low lossy sand shell, oil sand). The
projected antenna (Fig. 13) is constituted by:

a rigid coaxial line with the inner conductor opportunely longer

than the external conductor;

several circular elements terminating the inner conductor,

which are useful for impedance adjustment;

a mobile choke, which is used to decrease return currents on

the outer conductor and to adjust antenna impedance.
Once the antenna was realized, the return loss was measured
using a Network Analyzer (Agilents PNA-X-N5242A) in several
congurations, and specically in air, sand, wet sand, as well as in
the experimental set-up before and after lling the tank with oil
sand. The return loss at 2.45 GHz, in the denitive set-up, turned
out to be 29 dB. The antenna is moreover extremely adaptable to
different parameters of the surrounding environment.
3.3. Reservoir and tight shell materials
Two different bulk materials were used to ll the sand box in
the experimentation, namely:

oil sand samples, furnished by Eni;

low lossy quartz sand.

Fig. 9. Insertion of the PTFE tubes and of the radiating element in the low lossy
sand shell.

Fig. 11. Experimental set-up.

Fig. 10. Schematic of the experimental set-up and identication of the main equipment.

M. Bientinesi et al. / Journal of Petroleum Science and Engineering 107 (2013) 1830

25

Fig. 12. Position of ber optic sensors and of thermocouples for temperature
measurements.

Table 1
Position of temperature sensors during oil sand electromagnetic heating experimental testing. The coordinate system used has origin in the center of the antenna,
and is depicted in Fig. 8.
FO## x
y
z
TC## x
y
z
TCx x
y
[mm] [mm] [mm]
[mm] [mm] [mm]
[mm] [mm]
FO1
FO2
FO3
FO4
FO5
FO6
FO7
FO8
FO9
FO10

200
320
460
0
580
40
60
520
260
120

0
0
0
260
0
0
0
0
0
0

0
0
0
0
0
0
0
0
0
0

TC2
TC3
TC4
TC5
TC7
TC9
TC14
TC16

0
0
650
650
650
650
650
650

650
650
0
319
0
319
0
0

0
0
0
0
0
0
250
250

TCa 390
TCb 390
TCe
0
TCf
0
TCt
85

0
0
390
390
0

These materials were characterized through a series of techniques described in Sarri et al. (2012). The results of interest are
reported in Table 2 for oil sand and in Table 3 for quartz sand.
About the dielectric properties of oil sand samples, they remain
practically stable with increasing temperature as long as connate
water evaporation is limited while, after water evaporation,
imaginary permittivity decrease signicantly (Table 2).
The type of quartz sand used was selected for its particularly
low imaginary permittivity, which makes it a really low lossy
material, thus particularly suitable for the tight shell simulation
(though the shell is not actually tight in this specic case).
3.4. Electromagnetic heating experiment
A long term experiment was conducted in order to:
(1) evaluate the time evolution of temperature proles at different
radiating power levels;
(2) compare experimental data with the numerical model developed and described in Appendix A;
(3) evaluate the degree of mobilization of the bitumen in the
temperature range 150200 1C.

Fig. 13. MW radiating antenna.

Temperature data from ber optic sensors and thermocouples

were recorded with a frequency of 60 records per hour. Periodically, power was briey switched off in order to measure the
temperature at different depths.
In Table 4, the journal of the experiment is reported. Noteworthy, radiating power was rst set at 1000 W and then
increased to 1500 W and 2000 W once a quasi-stationary temperature prole was achieved. Finally, in order to further increase
the temperature in the oil sand, the outer side of the metallic walls
and the top surface were covered with 5 cm thick rockwool
insulating panels. The experiment lasted over 13 days.

4. Results and discussion

4.1. Experimental results
Fig. 14 shows temperature data from all the sensors placed in
the A direction (see Fig. 12), for the entire duration of the
experiment. In particular, seven ber optic sensors (FO6, placed
in the silicone oil between the two concentric PTFE tubes, FO7 and
FO10, inserted in the quartz sand shell, and FO1, FO2, FO3, FO5, in
the oil sand) and one thermocouple (TC7) are considered. These
data have to be analyzed taking into account the actions occurred
during the experiment, as reported in Table 4.
Figs. 15 and 16 report the time evolution of the temperature
prole along the x coordinate (see Fig. 12), during the rst part of
the experiment and after the thermal insulation of the set-up
respectively.
Fig. 17 shows the time evolution of the temperature prole
along the z coordinate in the point TCa (see Fig. 8), at a distance of
390 mm from the antenna.

26

M. Bientinesi et al. / Journal of Petroleum Science and Engineering 107 (2013) 1830

(Fig. 15); this arises from the evaporation of the small amount
of connate water present and the subsequent decrease of oil
sand loss tangent. The water evaporation clearly contributes to

Some main observations can be made based on these results:

oil sand temperature increases monotonically during electro-

220
200
180
160

Temperature [C]

magnetic irradiation due to the dielectric heating phenomenon

(Figs. 1417);
notwithstanding the higher electromagnetic eld present as a
consequence of the smaller distance from the antenna, the
temperature of the quartz sand shell (FO7 and FO10 sensors in
Fig. 14 and xo 200 mm in Figs. 15 and 16) is always lower than
the maximum temperature achieved in the oil sand material
(FO1 sensor in Fig. 14), thus conrming the efciency of the
shell design in decreasing the temperature around the well and
the antenna;
once the temperature of the oil sand material overcomes
100 1C, a sensible decrease in the heating rate can be observed

140
120
100

FO6, 40 mm
FO7, 60 mm

80

FO10, 120 mm
FO1, 200 mm

60

FO2, 320 mm

40

Wall
insulation

20

Table 2
Average properties of oil sand material used in the experimentation.

48

72

96

120 144 168 192 216 240 264 288 312 336

Table 3
Properties of quartz sand used in the experimentation.

Fig. 14. Temperature measured by the sensors placed along direction A (see Fig. 12)
for the entire duration of the experiment. In the legend, the distance of each probe
from the antenna is reported.

140
0h
5h

120

Temperature [C]

40%
0.25 wt%
13 wt%
3.8
0.1
2.5
0.01
2100 kg/m3
930 J/kg K
0.93 W/mK

Value

24

Time [h]

Sand void grade

Water content
Oil content
Dielectric constant
(@ 2.45 GHz, 25 1C)
Dielectric constant
(@ 2.45 GHz, 4100 1C)
Density
Specic heat (@ 25 1C)
Thermal conductivity (@ 25 1C)

Dielectric constant
(@ 2.45 GHz, 25 1C)
Bulk density
Void grade
Specic heat (@ 25 1C)
Thermal conductivity (@ 25 1C)

TC7, 650 mm

Value

Property

FO5, 580 mm

Property

FO3, 460 mm

10 h
15 h
20 h

100

25 h
30 h

80

35 h
40 h

60
40
20

2.9
0.001
1620 kg/m3
39.1%
800 J/kg K
0.24 W/mK

0
0

50 100 150 200 250 300 350 400 450 500 550 600 650

x [mm]
Fig. 15. Temperature proles in the direction A (see Fig. 12) versus the distance
from the antenna, before set-up thermal insulation.

Table 4
Journal of the experiment.
Time

Action

0
17 h 15 min
17 h 22 min
28 h 46 min
28 h 56 min
36 h 16 min
36 h 30 min
61 h 42 min
61 h 54 min
68 h 59 min
69 h 02 min
86 h 33 min
86 h 36 min
165 h 51 min
166 h 02 min
183 h 21 min
183 h 56 min
205 h 55 min
205 h 57 min
232 h 29 min
232 h 38 min
326 h 06 min

Magnetron
Magnetron
Magnetron
Magnetron
Magnetron
Magnetron
Magnetron
Magnetron
Magnetron
Magnetron
Magnetron
Magnetron
Magnetron
Magnetron
Magnetron
Magnetron
Magnetron
Magnetron
Magnetron
Magnetron
Magnetron
Magnetron

switched
switched
switched
switched
switched
switched
switched
switched
switched
switched
switched
switched
switched
switched
switched
switched
switched
switched
switched
switched
switched
switched

on,
off,
on,
off,
on,
off,
on,
off,
on,
off
on,
off
on,
off,
on,
off,
on,
off
on,
off
on,
off,

power 1000 W. Start recording temperature data

measure with thermocouples (MTC1)
power 1000 W
measure with thermocouples (MTC2)
power 1500 W
measure with thermocouples (MTC3)
power 2000 W
measure with thermocouples (MTC4)
power 2000 W
power 2000 W
power 2000 W
measure with thermocouples (MTC5)
power 2000 W
steel wall and top surface covered with berglass insulation
power 2000 W
power 2000 W
power 2000 W
measure with thermocouples (MTC6), insulation removed

M. Bientinesi et al. / Journal of Petroleum Science and Engineering 107 (2013) 1830

limit temperature rising near the well and favors the penetration of electromagnetic energy;
the silicone oil temperature (FO6 in Fig. 14) increases initially up
to 70 1C; once this temperature is reached, convective motions
establish, favoring the dissipation of heat towards the environment through the upper surface of the uid; this heat dissipated
is the cause of the qualitative difference between the curves for
the tight shell scenario in Fig. 5, where the temperature
maximum is always at the antenna, and the curves in Fig. 15;
the temperature of the external metallic walls (curve TC7 in
Fig. 14) is determined by two concurrent phenomena: on one
side, the RAM sheets absorb practically all the residual electromagnetic energy reaching the borders, on the other side, heat is

220
200

Temperature [C]

180
160

removed from the walls through natural convection of the

surrounding air. This leads to avoid heat accumulation into the
oil sand thus providing boundary conditions that can be
roughly representative of a larger oil sand mass (obviously this
holds as long as the set-up is not thermally insulated);
the measured vertical temperature proles show a maximum
for z 0 (Fig. 17), i.e. on the horizontal plane intersecting the
center of the antenna, decreasing quasi-symmetrically upwards
(zo 0) and downwards (z 40), in agreement with the predicted
pattern of the antenna;
the nal phase of the experiment, after the insulation (see
Fig. 16), is used exclusively in order to increase the temperature
over to 100 1C in the whole mass of oil sand, allowing to evaluate
the possible mobilization of the oil at higher temperatures. Once
the set-up was dismounted, oil sand blocks were visually
analyzed and several points, where the oil partially had drained,
were detected (Fig. 18). It is clear that for this specic oil sand
sample gravity drainage is not sufcient, even at high temperature, to obtain a signicant oil mobilization.

140

4.2. Comparison with the numerical model

120
100
80

40 h
180 h
200 h
230 h
260 h
320 h

60
40
20
0
0

50

100 150 200 250 300 350 400 450 500 550 600 650

x [mm]
Fig. 16. Temperature proles in the direction A (see Fig. 12) versus the distance
from the antenna, after set-up thermal insulation.

200

17 h
29 h
36 h
62 h
166 h
326 h

180
160

Temperature [C]

27

140

The numerical model was adapted, as described in Appendix A,

to the geometry and radiation pattern of the experimental set-up,
in order to compare simulation results with experimental results.
In the model, material properties are time-independent, thus
the model is adequate to experimental data description only for
temperature below the boiling point of water (100 1C). Above this
temperature, water vaporizes and escapes from the oil sand mass,
dielectric permittivity decreases and electromagnetic energy dissipation decreases as well.
In Figs. 19 and 20, experimental temperature proles in the r
and z directions are compared with simulation results. Noteworthy
the agreement is excellent as long as the temperature of 100 1C is
not overcome; once the water contained in the oil sand is
evaporated, the model is no longer valid and thus the prediction
of temperature proles becomes largely inaccurate.

5. Conclusion

120
100
80
60
40
20
0
-350 -300 -250 -200 -150 -100 -50

50

100 150 200 250 300 350

z [mm]
Fig. 17. Temperature proles along the vertical coordinate in the point TCa (at a
distance of 390 mm from the antenna along the direction A, see Fig. 12).

This paper presents a novel design for the downhole electromagnetic heating of oil sand/heavy oil reservoir, including a tight
shell made of low lossy material inserted in the surrounding of the
well containing the radiating antenna. Preliminary numerical
simulation performed with a simplied model showed that the
presence of a tight shell could help to reach uniform heating of
large volumes of reservoir avoiding the risk of high temperatures
at the wellbore, and making the recovery method much less
sensitive to local dielectric properties of the reservoir materials.
An experimental test was set up to conrm these claims.
Experimental results showed that electromagnetic irradiation is

Fig. 18. Oil draining from oil sand at the end of the test.

28

M. Bientinesi et al. / Journal of Petroleum Science and Engineering 107 (2013) 1830

180
160
140

Temperature [C]

Appendix A

1h
3h
5h
10 h
15 h
20 h
25 h
28 h

120
100

The numerical model describes the coupled electromagnetic

and thermal problem related to RF/MW irradiation of a heavy oil
reservoir through an antenna installed inside a wellbore, at an oil
bearing level. Fluid ow due to production is not considered.
The problem is governed by the equation of thermal energy
conservation

80

eff C eff

60
40
20
0
0

100

200

300

400

500

600

x [mm]
Fig. 19. Comparison between experimental (symbols) temperature prole along x
and simulation results (continuum lines).

180
17 h 15 min

160

A1

where effCeff and Keff are respectively the effective heat capacity
and the effective thermal conductivity of the reservoir material,
and q is the heat source term which accounts for the energy
released per unit time and unit volume by the electromagnetic
eld into the reservoir.
In order to calculate the q term, the following equation, which
governs the attenuation of the irradiated energy ux (Von Hippel,
1966), is rst solved
dPr
2Pr
dr

A2

28 h 46 min
36 h 13 min

140

Temperature [C]

T
 K eff T q
t

62 h 00 min

120
100
80
60

v
0s 1
u


u
2f umix @
mix 2 A
n
t
f ; mix
1
1
c
2
mix

40
20
0
-350 -300 -250 -200 -150 -100 -50

where P is the electromagnetic wave power per unit solid angle

irradiated by the antenna, r is the coordinate along the wave
propagation direction, and is the electromagnetic wave attenuation coefcient.
The attenuation coefcient is a function of the wave
frequency and of the complex dielectric permittivity of the
material

50

A3

100 150 200 250 300 350

z [mm]
Fig. 20. Comparison between experimental (symbols) temperature prole along z
in the point TCa (r 390 mm, see Fig. 12) and simulation results (continuum lines).

capable of heating oil sand, even above the boiling temperature of

connate water. The low lossy material shell realized around the
antenna turned out to be efcient in lowering the temperature in
this critical zone, as demonstrated by the fact that the maximum
temperature is reached outside the shell.
In a real oil well, from a technical point of view a 3 m radius
tight shell is hardly feasible with current technologies but future
studies will be performed in order to asses if less stringent
temperature limits and the effect of convective cooling due to
uids ow could lead to signicant shell radius reduction.
It was moreover shown that water vaporization has a signicant impact on temperature proles and contributes to limit the
temperature rise near the wellbore region and to better distribute
the irradiated energy in the reservoir.
Current ongoing activities are aiming to develop a more
realistic reservoir model that includes a 3D geometry, the convective term in the thermal equation, as well as oil and water ow,
in order to assess the effective benets of the designed system in
terms of oil recovery and productivity index improvement.

where f is the wave frequency, c is the speed of light in vacuum,

mix and mix are the real and imaginary part of the complex
dielectric permittivity of the reservoir medium, which is a mixture
of different components (namely solid matrix, oil and water).
Hence, is not actually constant but varies with time and with the
spatial coordinate, as a consequence of material compositional
changes (water vaporization).
Once provided the solution of the EM power eld P(r), the heat
source term can be calculated applying the following equations
! !
q  F
Fr

Pr
r2

A4
A5

where F(r) is the electromagnetic energy ux (i.e. the energy

carried by the electromagnetic wave crossing the unit area per
unit time).
In order to study the effect on the reservoir heating process, the
model was congured to take into account water vaporization.
A strong assumption we make is that the produced steam can
escape the reservoir and does not cause a pressure increase. This
assumption is hardly realistic, but it signicantly simplify the
model. The reservoir is ideally divided in three zones, with
different dielectric and thermal properties:

water saturation zone (with T oTboil): dielectric and thermal

Acknowledgements
Eni, e&p division, is gratefully acknowledged for funding this
research activity and providing reservoir materials for the experimental activities.

properties of the reservoir are those of the rock-heavy oil

water system (dielectric properties are assumed equal to the
values measured at 20 1C and shown in Fig. 3);
boiling layer zone (T  Tboil): transition phase for a little interval
around Tboil; the reservoir thermal capacity is calculated so as
to include the water evaporation latent heat;

M. Bientinesi et al. / Journal of Petroleum Science and Engineering 107 (2013) 1830

29

Table A1
Setting for numerical simulation for the comparison with experimental results.
Parameter

Value

Observation

Transmitted power, P0

1000 W (for
0o t o29 h)
47.5 mm
125 mm
650 mm
16 1C
T0+0.714 t [min]

External PTFE tube radius

Low lossy quartz sand shell radius
Minimum distance between the antenna and the lateral walls of containment tank
Measured
From the interpolation of data measured with thermocouples on the external walls

T0

Equal to the temperature of the room

dT/dr0.5 K/mm

This condition arises from the average value of heat dissipated through the silicone oil
throughout the trial
Measured (Sarri et al., 2012)
Measured (Sarri et al., 2012)
Measured
Measured (average value in the range 20200 1C)
Measured (average value in the range 20200 1C)
Calculated from the measured values of density of the components bitumen, connate water,
sand, and from the measured composition
Measured (average value in the range 20200 1C)
Measured (average value in the range 20200 1C)

Well radius, r0
Shell radius, rshell
Tank radius, rcont
Initial temperature, T0
Temperature of the outer walls (boundary
condition), Text
Temperature of the upper and bottom walls
(boundary conditions), Text
Boundary condition at the well
Sand shell permittivity, shell
Oil sand samples permittivity, OS
Shell sand density, shell
Shell sand specic heat, Cshell
Shell sand thermal conductivity, Kshell
Oil sand samples density, OS

2.9i0.001
3.8i0.1
1620 kg/m3
900 J/kg/K
0.1 W/m/K
2000 kg/m3

Oil sand samples specic heat, COS

Oil sand samples thermal conductivity, KOS

1020 J/kg K
0.8 W/m/K

dried zone (for T 4Tboil): dielectric and thermal properties of

the reservoir are those of the rock-heavy oil system, from which
water has been removed (dielectric properties are assumed
equal to the values measured over 100 1C and shown in Fig. 3).
These three zones evolve with time, determining the material
properties to be used in Eq. (A1) and Eq. (A2) at a certain time for a
given value of the spatial coordinate.
The described physical model is implemented using Comsol
Multiphysics, in a 1D spherical geometry. Initial and boundary
conditions are:

thermal equation:

uniform temperature (25 1C) for t 0 through the entire
domain;

adiabatic boundary condition at the well, imposed by the

symmetry of the problem;

constant temperature at the outer boundary, equal to the initial

temperature (this condition is justied by the fact that the

considered domain is much larger compared with the heating
radius;
electromagnetic wave attenuation equation (since this is a rst
order, stationary equation, it requires a single boundary
condition):
boundary condition at the well, where P0 is the total power
irradiated by the antenna.

The setting of reservoir material parameters in the model is

based on dielectric and physical characterization, carried out on oil
sand samples. Detailed description of the experimental laboratory
set-up and of the methods for the RF/MW dielectric characterization and for the measure of thermal properties of reservoir
materials is provided by Sarri et al. (2012).
Other relevant settings used in the preliminary numerical
simulations are:

reservoir average depth 80 m;

reservoir pressure 6 bar;
boiling temperature of connate water 160 1C;
sand porosity 25%;
initial water saturation 23%;
initial oil saturation 77%;
latent heat of vaporization of water 2080 kJ/kg.

In order to compare simulation results with experimental

results, the model was adapted to the geometry and to the
radiation pattern of the experimental set-up. Two major modications were made:

a 2D axisymmetric geometry was used in substitution of the

spherical geometry;

material properties dependence on temperature was neglected

for simplicity.
The choice of a 2D geometry is determined by the fact that the
radiation pattern of the experimental antenna is not isotropic but
dipole-like, so focused in the horizontal plane and symmetric with
respect to the vertical axis, with a 3 dB gain in the direction of
maximum irradiation. In order to simulate the distribution of
energy in the material, a 2D axisymmetric geometry was adopted
and an analytical expression was used for the approximate
calculation of the radiation diagram:
4

f pattern 2  sin

A6

where is the polar angle.

Considering material properties independent of temperature, it
was possible to write the analytical solution of the heat source
term in the thermal balance
8
< r 0 o r rshell qr;

2shell P 0 f pattern
exp2shell rr 0 
4r2
2OS P 0 f pattern
exp2shell r shell r 0 exp2OS rr shell 
4r 2

: r shell o r o r cont qr;

A7
where P0 is the transmitted power, shell and OS are the attenuation coefcient in the shell and in the oil sand respectively, r0 is the
well radius, rshell is the radius of the shell and rcont is the minimum
distance of the containment tank wall from the axis of the
antenna.
Attenuation coefcients, shell and OS, were calculated from
experimental data on the dielectric permittivity of the two
materials at the frequency of 2.45 GHz
v
1
u 0s
 2
u
2f

u
t @ 1
A8
f ; n
1A
c
2

Since material properties are considered independent of temperature, the model is adequate as long as these properties do not

30

M. Bientinesi et al. / Journal of Petroleum Science and Engineering 107 (2013) 1830

differ signicantly from their values at room temperature. This is

what happens experimentally once the connate water in oil sand
samples begins to evaporate, leading to a signicant decrease of
both the real and the imaginary part of the oil sands permittivity.
Thus, we expect the model to be accurate in interpolating experimental data only till the temperature does not overcome 100 1C.
Table A1 shows the values of the parameters used for the
numerical simulations.

References
Bridges, J., Taove, A., Snow, R., 1979. Method for in situ heat processing of
hydrocarbonaceous formations. US Patent 4140180.
Clark, B., 2007. Heavy oil. Working Document of the NPC Global Oil and Gas Study.
Epov, M.I., Mironov, V.L., Bobrov, P.P., Savin, I.V., Repin, A.V., 2009. Dielectric
spectroscopy of oil-bearing rocks at 0.0516 GHz. Russ. Geol. Geophys. 50,
470474.
Friso, T., Schildberg, Y., Rambeau, O., Tjomsland, T., Fordedal, H., Sjoblom, J., 1998.
Complex permittivity of crude oils and solutions of heavy crude oil fractions. J.
Dispersion Sci. Technol. 19 (1), 93126.
Fuji, M., Yoneda, T., Sanada, Y., 1999. Study on the micelle size of asphaltenes in
vacuum residue by dielectric loss. ACS Fuels 44 (4), 738742.
Haagensen, D.B., 1965. Oil well microwave tools. US Patent 3170519.

Hayashi, M., 2003. Temperature-electrical conductivity relation of water for

environmental monitoring and geophysical data inversion. Environ. Monit.
Assess. 96, 119128.
Hu, Y., Jha, K.N., Chakma, A., 1999. Heavy-oil recovery from thin pay-zones by
electromagnetic heating. Energy Source 21, 6373.
Kasevich, R.S., Kolker, M., Dwyer, A.S., 1979. In situ radio frequency selective heating
process. US Patent 4140179.
Kasevich, R.S., 2008. Method and apparatus for in situ radiofrequency assisted
gravity drainage of oil (RAGD). US Patent 7441597 B2.
Kasevich, R.S., Price, S.L., Faust, D.L., 1994. Pilot testing of a radio frequency heating
system for enhanced oil recovery from diatomaceous earth. In: SPE Annual
Technical Conference and Exhibition, 2528 September 1994, New Orleans, LA.
Kiamanesh, A.I., 1992. In situ tuned microwave oil extraction process. US Patent
5082054.
Kovalyova, L.A., Khaydar, A.M., 2004. Physical and rheological properties of
petroleum uids under the radio-frequency electromagnetic eld effect and
perspectives of technological solutions. Appl. Surf. Sci. 238, 475479.
Ritchey, H.W., 1956. Radiation heating. US Patent 2757738.
Saraev, D.V., Lunev, I.V., Yusupova, T.N.., Tagirzyanov, M.I., Yakubov, M.R., Gusev, Y.A.,
Romanov, G.V., 2005. Dielectric spectroscopy in studying mechanisms of
structure forming oils. Oil Gas Bus. 2005.
Sarri, A., Batisti, M., Bientinesi, M., 2012. Measurements of complex permittivity of
geological materials mixtures at RF frequencies. In: Sixth European Conference
on Antennas and Propagation (EUCAP), 2630 March 2012, Prague, Czech
Republic.
Sresty, G., Dev, H., Snow, R.H., Bridges, J.E., 1986. Recovery of bitumen from tar sand
deposits with the radio frequency process. SPE Reservoir Eng. 1 (1), 8594.
Von Hippel, R., 1966. Dielectrics and Waves. MIT Press, Cambridge.