234 (2024) 212620

Contents lists available at ScienceDirect

Geoenergy Science and Engineering

journal homepage: www.sciencedirect.com/journal/geoenergy-science-and-engineering

Experimental investigation of Al-Fe2O3 thermite reactions for thermal plug

and abandonment of oil wells
Kesiany M. De Souza a, Marcelo J.S. de Lemos a, *, Roberta dos R. Ribeiro a, Ana M.G. Marin b,
Paulo G.C. Martins c, Leonardo H. Gouvêa c
a
Departamento de Energia - IEME, Instituto Tecnológico de Aeronáutica - ITA, 12228-900, São José dos Campos, SP, Brazil
b
Departamento de Química, Instituto Tecnológico de Aeronáutica - ITA, 12228-900, São José dos Campos, SP, Brazil
c
Departamento de Aeronáutica, Instituto Tecnológico de Aeronáutica - ITA, 12228-900, São José dos Campos, SP, Brazil

A R T I C L E I N F O A B S T R A C T

Keywords: The use of thermite in the plug and abandonment of wellbores is a promising new method for sealing oil wells.
Aluminothermic reaction However, the reactants for thermite mixtures are usually in a powder state at ambient conditions, making a
Thermite perfect homogenization for high heat release a challenging task. In this sense, this study aims to investigate the
Fe2O3–2Al system
Fe2O3–Al thermite system prepared through a solvent-based method to maximize homogenization of the mixture
P&A process
and enhance energy release during the exothermic reaction. Tests were conducted to compare the burning ve­
Solvent-based mixing
locity, ejected mass percentage, and temperature profiles of the reaction through small steel-tubes, comparing a
dry-mixed, stoichiometric composition with a solvent-based mixture. The effect of additives such as Al2O3 and Al
were also evaluated. Results showed that the solvent-based process led to higher compactness, higher temper­
atures on the steel tube’s exterior, a more stable reaction, and a 40% decrease in ejected material. Also, Al-rich
mixtures had faster reactions, lower temperatures, and more ejected material when compared to the stoichio­
metric system, while Al2O3-diluted mixtures showed a linear decrease in burning velocity and mass ejection at
higher dilution levels, with no significant variation in temperature levels. Mixtures with 40% or higher dilution
did not self-propagate. Therefore, a thermite mixture prepared using solvent and diluted with Al2O3 at 20–30% is
recommended for the thermal plug and abandonment technology.

abandonment (P&A) of oil wellbores (Zhang et al., 2023) (De Andrade

1. Introduction et al., 2021) (De Andrade et al., 2023) (De Andrade et al., 2024a) (De
Andrade et al., 2024b) (De Andrade et al., 2024c). Portland cement has
Thermite is an energetic material in which a more active pure metal become the primary plugging material due to its low cost,
reacts with a metal or non-metal oxide, resulting in a more stable oxide, user-friendliness, and commercial availability. However, there are still
metal or non-metal, and releases a high amount of energy (Mei et al., issues such as chemical shrinkage, thermal instability, cracking, dura­
1999). Fisher and Grubelich (Fischer and Grubelich, 1998) presented an bility in corrosive environments, and degradation of properties due to
extensive list of thermite mixtures and their main theoretical charac­ contamination (Khalifeh et al., 2019). The next generation of plug and
teristics, such as adiabatic temperature and gaseous products. These abandonment operations will likely use new classes of cheap and effi­
energetic materials have a wide range of use, such as weld of rail tracks cient sealing materials (Achang et al., 2020). Salehi et al. proposed a
or electrical connections (Wang et al., 1993) (Lee, 2006) (Schroeder and geopolymer cement with good pumpability, low shrinkage, and high
Poirier, 1984), production of ceramic-lined pipes (Mu et al., 1997) compressive and shear bond strength (Salehi et al., 2017). Kristiansen
(Odawara and Ikeuchi, 1986), and synthetization of composites (Reddy et al. discussed the activation of shale (argillaceous rocks) to form well
et al., 2007). Harada et al. (2005) examined the potential use of the barriers (Kristiansen et al., 2018). Bismuth (Zhang et al., 2020) (Zhang
thermite reaction to the melting process of the municipal solid waste et al., 2021), resin (Rashid et al., 2019) (Retnanto et al., 2023), and
incineration ash. thermite (Levchenko et al., 2023) technologies are promising and sus­
Additionally, thermite systems seem to be an innovative solution to tainable alternatives that can reduce P&A time and costs with more
overcome the limitations of traditional methods of plug and efficient plugs and possibly shorter than traditional cementation plugs.

* Corresponding author.
E-mail address: [email protected] (M.J.S. de Lemos).

https://doi.org/10.1016/j.geoen.2023.212620
Received 25 August 2023; Received in revised form 28 November 2023; Accepted 27 December 2023
Available online 29 December 2023
2949-8910/© 2023 Elsevier B.V. All rights reserved.
K.M. De Souza et al. Geoenergy Science and Engineering 234 (2024) 212620

analyze its effectiveness in the scenario of plugging and abandonment of

Nomenclature wells. In an earlier work, Souza and de Lemos (De Souza and de Lemos,
2021) (De Souza and de Lemos, 2023a) developed a one-dimensional
Latin Characters numerical model of the Fe2O3–2Al thermite reaction propagation in a
a Number of treatments radial domain based on several assumptions. Later, the model in (De
d Distance between steel-tube top surface and Souza and de Lemos, 2021) (De Souza and de Lemos, 2023a) was refined
thermocouple position were authors used various methods to analyze the differential scanning
EA Activation energy calorimetry (DSC) data obtained from a Fe2O3/Al reaction. Calorimetry
H Steel-tube height experiments were used to obtain the pre-exponential factor and the
H0 Null hypothesis activation energy, which were applied in a more complex model (De
H1 Alternative hypothesis Souza et al., 2023). Furthermore, in reference (De Souza and de Lemos,
N Total number of observations 2023b) authors found that the main factors in a thermite reaction are the
n Degrees of freedom specific heat and activation energy, having a total percent contribution
p Distance between steel-tube top surface and mixture top of more than 90% of all effect contributions in the burning velocity.
surface Further, initial studies involving thermal analysis of the heat propaga­
r Radius tion generated by the thermite reaction in a wellbore were conducted
S Sample variance considering a constant heat generation (Pena and de Lemos, 2021) (Pena
T Temperature and de Lemos, 2023a) (Pena and de Lemos, 2023b). Recently, the model
t0 Test statistic in (Pena and de Lemos, 2023a) (Pena and de Lemos, 2023b) was
v Burning velocity improved by considering a more realistic chemical kinetics in investi­
y Sample mean gating the thermal behavior or thermite reactions (Pena et al., 2023). In
addition, a previous study (De Souza et al., 2022) was carried out to
Greek Characters identify the most promising thermite reactants for P&A of wellbores
φ Equivalence ratio based on thermodynamic analysis. In this earlier study, the alumi­
μ Mean of a probability distribution nothermic reaction of aluminum (Al) and hematite (Fe2O3) showed to be
Acronyms an efficient solution because of the high-temperature levels and low
MS Mean Square gaseous products in comparison to other thermite mixtures. However,
P&A Plug and Abandonment even for thermite mixtures with the same reactants, their physico­
SS Sum of Squares chemical properties and kinetics can be different due to factors such as
the mixture preparation procedure, particle size, particle distributions,
mixture homogenization and the presence of additives. According to
Cheng et al. (2010), mixtures of nano-sized powders made via solvent
Among these options, thermite is noticeable because its final reaction mixing or self-assembly react at a higher impact energy threshold than
products (usually a heavy metal and a lighter oxide) form a solid barrier micron-sized thermite mixture systems. Also, Prentice et al. (2005)
that can restore the caprock formation through a ‘self-sintering’ process observed that burning of sol-gel-produced thermites gave around four
(Vralstad et al., 2019). Such a process forms an in-situ self-sintered times the reaction front velocities of physically mixed thermites.
ceramic plug that, by modification of the basic formula with additives, Although thermite mixtures are more easily prepared by simple physical
the final plug could exhibit structural, sealing, and corrosion properties mix, preparations carried with a solvent can produce a more homoge­
superior to hydrating cement (Lowry et al., 2015). Moreover, the ther­ neous mixture. In most cases, organic solvents are employed due to the
mite reaction reaches thousands of degrees centigrade which could melt poor reactivity between organic solvents and aluminum particles (Sui,
the well components, such as the production tube, casing, cables, and 2019). The preparation of some aluminothermic mixtures has been re­
cement. ported with the dispersion in either isopropanol, hexane, or other media
Thermite is an energetic substance that is highly exothermic, gen­ followed by complete evaporation of the solvent (Sanders et al., 2007)
erates a lot of heat through a self-propagating reaction, and produces (Walter et al., 2007) (Perry et al., 2004) (Prakash et al., 2005) (Wang
mostly condensed state products that can mimic a natural barrier and et al., 2018).
restore the rock formation that was present before the well was drilled. Also, additives can be useful to control the reaction temperature
The first developments in the thermal P&A technology with thermite levels and velocity propagation. Addition of the inert species alumina
were carried out by companies in the oil & gas industry. Patents (Al2O3) to the Fe2O3–Al system dilutes the overall thermite reaction,
involving this technology were first published by Skjold (2013) about 10 decreases the temperature levels reached and the burning velocity. This
years ago, followed by Lowry and Dunn (2015), and Hearn et al. (2020), is important to create a more stable reaction propagation, with less gas
applied by the companies Interwell Technology, Olympic Research, and products, and decrease the porosity of the final products (Wittberg,
Conoco Phillips Company, respectively. Interwell conducted success­ 2017) (Lowry et al., 2015) (Lowry and Nielson, 2018) (Merzhanov et al.,
fully the first-ever offshore deployment of thermite material, however, 1996) (Malchi et al., 2008). Also, the addition of aluminum beyond
the material test failed to achieve a fluid-tight seal, which requires stoichiometry seems to have the opposite effect of alumina on the
further studies and evaluations (Net Zero Technology Centre). The burning velocity, resulting in faster reaction propagations (De Souza and
mechanism involved in a thermite reaction is complex and comprises de Lemos, 2023a) (Durães et al., 2006). However, no experimental work
multiple physicochemical processes that make it difficult to predict its has yet been published in which the Fe2O3–2Al system, prepared via
performance in a P&A scenario. Further, as reported by (Durães et al., solvent-based mixing, is tested in different configurations in order to
2007), over-aluminized thermite mixtures present an enhanced mixture identify a promising candidate for the plug and abandonment thermal
reactivity and a faster reaction propagation. These characteristics are technology. In addition, Lowry et al. (2015) and Lowry and Nielson
advantageous when rapid and efficient ignition is desired. In the P&A (2018) conducted laboratory experiments to explore the advantages of
scenario, applying this kind of mixture to facilitate the ignition process diluting thermite systems with non-reactive oxides, aiming to create
would be interesting. controlled-porosity ceramic materials more efficient as plugging mate­
Therefore, a scientific foundation allied to experimental and nu­ rials. The effectiveness of a diluted thermite reaction can be attributed
merical studies is still required to characterize this type of reaction and mainly to a controlled reaction rate that allows for improved consistency
and control over the typical fast and violent thermite reaction. This

2
K.M. De Souza et al. Geoenergy Science and Engineering 234 (2024) 212620

Table 1 reliability are crucial, such as producing controlled-porosity ceramic

Formulations used in the experiments. materials. Also, diluting such materials can be more cost-effective than
Mixture Al [% w] Fe2O3 [% w] Al2O3 [% w] Mixing Method using a pure thermite mixture, as it reduces the amount of expensive
Code reactive materials needed. Additional information on the effects of
C1 25.26 74.74 0.0 Dry porosity in combustion in porous media can be found in (Coutinho and
C2 25.26 74.74 0.0 Solvent-based de Lemos, 2012).
C3 32.73 67.27 0.0 Solvent-based In this sense, the present contribution aims to experimentally
C4 22.73 67.27 10.0 Solvent-based compare the Fe2O3–Al system prepared by dry-mixing and solvent-based
C5 20.21 59.79 20.0 Solvent-based
C6 17.68 52.32 30.0 Solvent-based
mixing. Also, various Fe2O3–Al mixtures prepared via solvent-based
C7 15.16 44.84 40.0 Solvent-based mixing were investigated to observe the effects of alumina dilution
and aluminum-rich mixtures and identify an efficient composition to
apply in a Thermal P&A scenario. In addition, it is important to
consistency is important in the P&A scenario where precision and emphasize that this work was based in terms of macroscopic aspects,

Fig. 1. Mixture preparation: (a) dry mixing; (b) solvent-based mixing.

Fig. 2. Experimental setup: a) schematic of the measuring system, b) steel-tube; c) internal setup.

3
K.M. De Souza et al. Geoenergy Science and Engineering 234 (2024) 212620

Table 2 25.26% micro-Al powder with particles size 38–125 μm and purity
Dimensions of steel tubes used in experiments. 99.7+%, and 74.74% Fe2O3 powder with particles size <5 μm and purity
Tube H [mm] p [mm] rin rex d1 d2 96%+. The Al-rich mixtures were prepared by adding 10% Al to a
type [mm] [mm] [mm] [mm] stoichiometric system, while the Al2O3-diluted mixtures were prepared
1 4.8 ± 0.8 ± 13.0 16.5 18 ± 2 19 ± 2 by adding 10–40% Al2O3 powder with a nominal particle size of 60 μm.
2 0.1 0.1 12.5 17.0 Each mixture was combined in glass vials (see Fig. 1a) and, for the
solvent preparation, slurred in ~30 mL of solvent 99% (see Fig. 1b) and
left to dry at ambient conditions for about 2–3 days.
Table 3
Mixture and setup parameters of experimental tests. 2.2. Procedure
Test Tube Mixture Thermite dry Solvent in ρ [g/
configuration type code mass [g] dry mass [g] cm3] A schematic of the measuring system is presented in Fig. 2a. The
t1 1 C1 30.0 ± 2.2 0.0 1.4 ± mixtures were placed in steel tubes shown in Fig. 2b with dimensions in
0.1 Fig. 2c and Table 2. A glass stirring rod was used to compact the mix­
t2 C2 51.0 ± 2.2 6.8 ± 0.1 2.4 ± tures. The final density was estimated by dividing the mixture mass
0.1
placed inside the steel tube by the approximate volume occupied. In
t3 2 C2 37.5 ± 1.2 0.9 ± 0.6 1.9 ±
t4 C3 0.1 each test, two type-K thermocouple joints were welded at the steel tube’s
t5 C4 axial positions T1 and T2, as illustrated in Fig. 2c. In addition to
t6 C5 providing the temperature profile and peak temperature at the steel tube
t7 C6
surface, the data collected by each thermocouple also allowed for the
t8 C7
estimation of the burning velocity from the signal delay between
thermocouples.
such as average reaction wave propagation, temperature levels, and Overall, 16 tests were conducted by replicating the 8 experimental
relative expelled mass during the reaction, and not on additional studies configurations shown in Table 3. In order to compare the resulting
involving emissions caused by the generated products, for example. densities of the first two configurations, thermite mixes were compacted
Furthermore, microscopy analyses including studies such as field emis­ as much as possible with a glass stirring rod. Observe that the mixes
sion and scanning electron microscope were not carried out in the pre­ produced by the solvent-mixing approach (test configuration 2) revealed
sent work. higher density levels than dry-mixing (test configuration 1), which is to
be expected given that the solvent dilution increases mixture uniformity
2. Experimental methodology and appears to enhance the engagement between solid particles. In the
others 3–8 configurations, the density of the thermite mixtures were
The experimental setup consists of a carbon-steel tube filled with maintained at similar levels in order to compare the effect of alumina
thermite mixtures prepared with different formulations, a hypergolic and aluminum addition to the combustion process without taking into
mixture of KMnO4 + glycerin for ignition, two type-K thermocouple account the density effect.
joints welded at different longitudinal positions of the tube external Each test was ignited by a hypergolic mixture of permanganate and
surface, and a NOVUS Field Logger acquisition system to capture the glycerin. In every test, approximately 1g of permanganate was placed
thermocouples data. Overall, 7 formulations were tested in duplicates. over the thermite mixture, followed by 2–3 drops of glycerin to induce
These experiments mainly aimed to compare the average reaction ignition in approximately 20–60 s, depending on ambient conditions.
propagation, temperature profiles, and total expelled mass among the For the Al2O3-diluted mixtures (C4–C7) the hypergolic ignition was not
formulations tested. sufficient to start the self-propagating reaction. Therefore, approxi­
mately 5g of a stoichiometric Fe2O3–2Al non-diluted mixture was first
placed over the diluted system, and then permanganate and glycerin
2.1. Sample preparation were used to ignite the stoichiometric mixture, which then ignited the
diluted mixture.
The reaction investigated takes the following stoichiometric form
Fe2 O3 + 2Al → 2Fe + Al2 O3 + 856 kJ / mol (1) 2.3. Statistical analysis

where one mol of hematite (Fe2O3) reacts with two mols of aluminum Some statistical tools were applied to the experimental data for
and forms pure iron (Fe) and alumina (Al2O3). Table 1 shows seven analysis. The results obtained to compare dry-mixing versus solved-
different Fe2O3–Al thermite mixtures prepared to examine the effects of based mixing (t1 and t2 test configurations) and to compare stoichio­
the preparation methods (C1 and C2), as well as the effects of increasing metric versus Al-rich mixture (t3 and t4 test configurations) were
Al amounts above stoichiometry (C3) and adding Al2O3 as a diluent analyzed by applying the Welch two sample t-test, given by the
(C4–C7). The stochiometric mixtures were prepared with a ratio of following equation (Montgomery, 2013)

Table 4
The analysis of variance (ANOVA) table for the single factor model applied (Montgomery, 2013).
Source of variation Sum of Squares Degrees of Freedom Mean Square F0

Between treatments SSTreatments = a− 1 SSTreatments MSTreatments

a MSTreatments = F0 =
∑ a− 1 MSE
n (yi. − y.. )2
i=1
Error (within treatments) SSE = N− a SSE
MSE =
SST − SSTreatments N− a
Total SST = N− 1
∑a ∑ n
(yij − y.. )2
i=1 j=1

4
K.M. De Souza et al. Geoenergy Science and Engineering 234 (2024) 212620

Fig. 3. Mean-temperature profiles and standard deviations measured in the experimental configuration (a) t1 (dry mixing), and (b) t2 (solvent-based mixing).

Fig. 4. Boxplots comparing dry mixing and solvent-based mixing process: (A) Thermocouples peak temperature; (B) Reaction burning velocity; (C) Dispersed mass in
% of the initial dry mass.

y1 − y2 Al2O3-dilution was investigated at various levels to establish a regres­

t0 = √̅̅̅̅̅̅̅̅̅̅̅̅
̅ (2)
S21 S2 sion model. Therefore, the results from test configurations with stoi­
+ n22
n1
chiometric (t3) and Al2O3-diluted (t5-t8) mixtures were analyzed by an
analysis of variance (ANOVA) to determine if the dilution effects
where t0 is the test statistic, y, S, and n are the mean, variance, and
observed are statistically significant or not. The ANOVA analysis follows
degrees of freedom of each sample, respectively. The t-test evaluates the
the equations shown in Table 4, where a is the number of treatments
means of two samples using the following hypothesis test
(levels of Al2O3 dilution), n is the number of replicates, N=an is the total
H 0 : μ1 = μ2 (3) number of observations, yi. is the average of the observations under the
ith treatment, and y.. is the grand average of all the observations. Then,
H 1 : μ1 ∕
= μ2 (4) using linear regression, mathematical models were generated for re­
sponses that showed significant variations. The Shapiro-Wilk and stu­
where H0 is the null hypothesis, H1 is the alternative hypothesis, and μ1 dentized Breusch-Pagan tests were applied to check the normality and
and μ2 are the means being compared. Then, the null hypothesis is homogeneity of the residuals.
rejected or accepted based on the computed value of the p-value for the
two-sample t-test, given a significance level of 5%.
Different from the two-sample comparative studies, the effect of

5
K.M. De Souza et al. Geoenergy Science and Engineering 234 (2024) 212620

Table 5
Two sample t-test results for dry vs solvent-based mix (95% confidence, alter­
native hypothesis: true difference in means is not equal to 0).
Table 6
Response Mean ± std. dev. t- df p- 95% Conf. Two sample t-test results for stoichiometric vs Al-rich mixture (95% confidence,
value value interval alternative hypothesis: true difference in means is not equal to 0).
Dry Solvent-
mix based mix Response Mean ± std. dev. t- df p- 95% Conf.
Tpeak [oC] 568 ± 690 ± 43 − 2.8 4.7 0.042 (-237.3, value value interval
Stoich. Al-rich
77 − 6.8)
v [mm/s] 22.8 ± 1.8 ± 0.2 44.0 1.1 0.009 (16.5, 25.5) Tpeak [oC] 554 ± 416 ± 2.4 3.6 0.079 (-26.4, 303.4)
0.6 35 108
Disp. mass 83.6 ± 50.9 ± 1.5 13.7 1.5 0.017 (17.6, 47.8) v [mm/s] 3.2 ± 11.6 ± − 28.3 1.4 0.007 (-10.4, − 6.4)
[%] 3.0 0.2 0.4
Disp. mass 62.0 ± 95.0 ± − 7.9 2.0 0.016 (-51.2,
[%] 4.2 4.3 − 14.9)

Fig. 5. Mean-temperature profiles and standard deviations measured in the experimental configuration (a) t3 (stoichiometric), and (b) t4 (Al-rich).

Fig. 6. Boxplots comparing stoichiometric and Al-rich reactions: (A) Peak temperature; (B) Burning velocity; (C) Dispersed mass in % of the initial dry mass.

6
K.M. De Souza et al. Geoenergy Science and Engineering 234 (2024) 212620

Fig. 7. Mean-temperature profiles and standard deviations measured in the Al2O3-diluted systems with (a) 0% dilution; (b) 10% dilution; (c) 20% dilution; and (d)
30% dilution.

expelled from the measurement of the initial and final system mass
Table 7
enabled us to identify possible candidates for efficient plug and aban­
Analysis of variance, normality, and homogeneity tests carried for the data with
donment of oil wells.
dilution effects.
Response ANOVA ANOVA p- SW test p-value BP test p-value
F0 value value 3.1. Dry mixing versus solvent-based mixing
Tpeak [oC] 0.197 0.664 0.384 0.146
v [mm/s] 42.17 6.34e-4 0.860 0.467 In the initial tests, we compared the performance of the Fe2O3–Al
Disp. mass [%] 100 5.79e-5 0.120 0.838 system prepared by simple physical mixing with that made by mixing
with an organic solvent (test configurations t1 and t2). Fig. 3 shows the
3. Results and discussion mean-temperature profiles and standard deviations captured by ther­
mocouples at positions T1 and T2 (see Fig. 2). The simple dry-mixing
Before present the results and accompanying discussions, it is thermite mixtures presented a smaller delay between the mean-
important to clarify that this work intended to compare burning veloc­ temperature profiles at each position, which indicates a faster reaction
ities, ejected mass percentages and temperature profiles of the reaction propagation when compared to the thermites prepared with a solvent.
through small steel-tubes, comparing a dry-mixed stoichiometric Moreover, the mean-peak temperature reached by the dry-mixing
composition with a solvent-based mixture. These data are expected to thermites appear at lower levels and present a higher data variability.
contribute to the understanding and development of more reliable ma­ The comparative boxplots and two sample t-test findings for the two
terials and, in the future, to the improvement of kinetic modes. There­ mixture formulations are shown in Fig. 4 and Table 5, respectively. The
fore, the effectiveness of the chemical materials in plugging wellbores in peak temperature at the external surface of the inox-tube (Tpeak), the
real word applications was not investigated here as an additional burning velocity (v), and the percentage of dispersed mass during the
experimental campaign would be needed, for example for checking the reaction are compared. As all p-values of the t-tests are lower than 5%,
leakage after cooldown under an imposed differential pressure through we can assume that the observed differences are significant with 95%
the cooled sample. confidence, even though a high variance is observed in the peak tem­
It is also important to emphasize that this scientific investigation was perature estimation. These results indicate that the solvent-based
undertaken with the primary aim of conducting a comparative analysis mixture would be more efficient for a P&A process because of the
of various thermite systems in terms of possible candidates to be applied higher peak temperature reached at the inox-tube external surface and
in a plug and abandonment process to melt, in a field operation, the well lower mass-dispersion to the environment. That is, it dispersed only
tubular and surroundings. As such, the experimental strategy adopted ~50% of the initial mass, while the mixture prepared by simple dry-
herein focused on utilizing a thick and short reaction vessel to prevent mixing dispersed ~83%. Since a solvent can favor the formation of
melting during exothermic reactions while enabling precise temperature more gaseous species, this result was unforeseen. However, a steadier
measurements through thermocouples welded to the tube. This flame was seen in the solvent-based combination as it burned more
approach was crucial in ensuring the integrity of the experimental setup slowly, which can be attributed to the greater uniformity and
and in facilitating the acquisition of reliable data for the purpose of compactness of the mixture. For a P&A process, a slower reaction is
system comparison. In this way, temperature levels, estimated burning required in order to guarantee an adequate heat transfer through the
speed from the thermocouples signal delays, and the estimated mass wellbore. As a result, the solvent-based thermite mixture appears to be
more effective for producing heat in the P&A situation, as well as being

7
K.M. De Souza et al. Geoenergy Science and Engineering 234 (2024) 212620

Fig. 8. Experimental data obtained in alumina (Al2O3) diluted mixtures: (a) Peak temperature; (b) Burning velocity; (c) Dispersed mass in % of the initial dry mass.

aluminum-rich (φ = 1.6) mixtures. These findings provide 95% confi­

Table 8 dence that the aluminum-rich combinations burn more quickly and
Linear regression data of burning velocity and dispersed mass models. release more mass into the atmosphere than stoichiometric mixtures. As
Burning velocity (mm/s) Dispersed mass (%) aluminum is the first species expected to evaporate from the system,
Equation y = a + bx y = a + bx adding more aluminum increases the total dispersed mass to the envi­
Intercept (a) 3.27600 ± 0.16176 60.685 ± 3.636 ronment during the combustion process. Also, aluminum works as a
Slope (b) − 0.05615 ± 0.00860 − 1.944 ± 0.194 catalyst and increases the burning velocity, as also observed in (Durães
Adj. R-square 0.85469 0.93396 et al., 2006). A 10% aluminum addition increased the burning velocity
SW test p-value 0.8599 0.1195
BP test p-value 0.4674 0.8376
in ~260%. There was no significant variation in peak temperature
(p-value >5%) that could be detected with 95% confidence due to the
considerable variability in temperature measurement. However, if one
more suited for transportation inside the wellbore. accepts a 10% significance level, the Al-rich mixtures did result in a
lower peak temperature at the outer surface of the steel tube. In
conclusion, aluminum addition is not indicated for P&A because it in­
3.2. Stoichiometric versus Al-rich mixture creases generation of gaseous species and accelerates the reaction pro­
cess, which might result in a less efficient heat transfer to the wellbore
The second set of experiments looked at what would happen if components.
aluminum was added to the Fe2O3–Al thermite system in excess of the Therefore, adding aluminum to the Fe2O3–2Al system beyond stoi­
stoichiometric ratio (test configurations t3 and t4). The mean- chiometry enhanced reactivity through a faster reaction propagation.
temperature profiles and standard deviations at thermocouple posi­ However, although aluminum-rich mixtures would benefit the ignition
tions T1 and T2 are shown in Fig. 5 for both test configurations carried process, the higher particle dispersion to the environment and violent
out. In case of stoichiometric system, the mean peak-temperature reaction propagation could compromise the integrity of the wellbore or
reached higher levels than the Al-rich mixtures. However, a high vari­ the plugging material.
ance was captured in the Al-rich tests for the thermocouple at T1.
Moreover, the temperatures reached at position T1 are overall higher
than the ones measured at T2, different from the previous set of results 3.3. Alumina-dilution effects
shown in Fig. 3. This different is attributed to the thicker tube in which
the test configurations t3-t8 were conducted, as detailed in Table 2. The final set of experiments examines the impact of adding alumina
Fig. 6 and Table 6 shows the boxplots and paired t-tests results, at different levels in the Fe2O3–2Al thermite system. The mean-
respectively, for comparison of peak temperature at inox tube outer- temperature profiles and standard deviations measured at thermocou­
surface, reaction burning velocity, and dispersed mass percentage dur­ ples positions T1 and T2 are shown in Fig. 7. Observe that, as the dilution
ing the reaction process, observed for stoichiometric (φ = 1.0) and level increased, the observed variances became greater. Also, different

8
K.M. De Souza et al. Geoenergy Science and Engineering 234 (2024) 212620

Fig. 9. Photos of the steel-tubes after reaction completion at different dilution levels.

from the 0–20% dilution tests, in the 30% dilution there is an inversion dispersed mass can decay extensively. Also, the impact on the average
of the mean-temperature peaks captured at T1 and T2. Moreover, the peak temperature at the tube’s outer surface seems small, possibly
higher diluted mixtures were more difficult to ignite until a point of no resulting in higher temperature levels. This may be justified as, although
ignition was noticed at 40% in the present test settings. a diluted thermite mixture generates less energy, the slower reaction
To ascertain whether the observed variations in these responses were propagation with less gaseous formations and the insulating character­
statistically significant, an analysis of variance (ANOVA) was performed istics of alumina may improve the efficiency in heat transfer through the
for each response, as shown in Table 7. The ANOVA results indicate that, wellbore. However, at 40% dilution, the heat generated was insufficient
with 95% confidence, the alumina dilution had a considerable impact on to self-propagate the thermite reaction, probably due to heat dissipation
the observed burning velocity and dispersed mass percentage (ANOVA to the surrounding before it can sustain the reaction propagation added
p-value <0.05), but that the peak temperature response was unaffected to the reduced reactivity of a diluted mixture.
(ANOVA p-value >0.05). The normality and homogeneity of the re­ Moreover, the final plug formed inside the steel tubes could be
siduals were confirmed by the p-values of Shapiro-Wilk normality test compared for each alumina dilution test, as shown in Fig. 9. Observe
(SW) and the studentized Breusch-Pagan (BP) test, respectively, with that, at 0% dilution, most reactants and products mass was dispersed to
values higher than 5%. the environment (see Fig. 9c), some stayed attached to the tube walls,
Plots showing peak temperature at the tube’s outer surface, mixture and only a small amount formed a plug at the bottom, which got loose on
burning speed, and total mass dispersed as a function of alumina dilution one test. As alumina was added in the experiments, the final plug
are shown in Fig. 8. Although no statistically significant differences in became larger, apparently less porous, and more attached to the walls.
peak temperature were found, the alumina-dilution of the Fe2O3–Al In view of all these results, alumina dilution of 20–30% is recommended
thermite system did reduce the burning velocity and the dispersed mass for P&A due to the improved efficiency of the heat diffusion process with
percentage. The mean burning velocity was reduced by 50% at a 30% lower burning velocity, fewer gaseous species formed, consequently
dilution when compared to the non-diluted mixture, and the mean lower mass dispersion, and larger final plug formed.
dispersed mass percentage was reduced from 62% to 4.7%. Linear No tube melting was observed in these small-scale tests conducted
regression models were retrieved for burning velocity and percentage of with short tubes. However, the tube coating was severely degraded, as
dispersed mass as a function of the dilution level, as shown in Fig. 8b and shown in Fig. 9. Considering the potential application of this material in
c. The model parameters of each regression model are shown in Table 8 the plug and abandonment scenario, potential drawbacks would be
together with the p-values higher than 5% of the Shapiro-Wilk normality damage to the cement and/or rock around the formed plug due to
test (SW) and the studentized Breusch-Pagan (BP) test, which confirms thermal stress (Khalifeh and Saasen, 2020), leading to potential leak
the adequacy of the models. paths. Also, if the thermite mixture is applied only to melt the produc­
Therefore, these results indicate that an alumina-dilution of the tion tube and open a window for traditional cementation, damage to the
Fe2O3–2Al thermite system in 10%–30% would benefit a plugging and casing could compromise the cement plug adhesion. To overcome these
abandonment process as the reaction propagates slower and the total challenges, additives can be employed to control the reaction rate and

9
K.M. De Souza et al. Geoenergy Science and Engineering 234 (2024) 212620

amount of heat generated (Chukwuemeka et al., 2023), such as the Cheng, J.L., Hng, H.H., Lee, Y.W., Du, S.W., Thadhani, N.N., 2010. Kinetics study of
thermal- and impact-Initiated reactions in Al-Fe2O3 Nanothermite. Combust. Flame
alumina dilution here analyzed.
157, 2241–2249.
Chukwuemeka, A.O., Oluyemi, G., Mohammed, A.I., Njuguna, J., 2023. Plug and
4. Conclusions abandonment of oil and gas wells - a Comprehensive review of Regulations,
Practices, and related impact of materials Selection. Geoenergy Sci. Eng. 226,
211718.
The 16 experimental tests conducted in small steel tubes revealed a Coutinho, J.E.A., de Lemos, M.J.S., 2012. Laminar Flow with Combustion with in Inert
very promising mixture to be applied in the P&A process of wellbores: an Porous Media. Int. Commun. Heat Mass Tran. 39, 896–903.
De Andrade, G.S., de Lemos, M.J.S., Colombo, D., 2021. A new hybrid analytical/
alumina-diluted Fe2O3–Al mixture prepared in solvent. Solvent medium
numerical method for transient heat conduction on composite hollow cylinders
preparation improves homogeneity of the mixture and allows greater applied to plug and abandonment of oil wells. Int. J. Therm. Sci. 168, 106981.
compactness, increasing the overall temperature levels that can be De Andrade, G.S., Nascimento, E.J.G., de Lemos, M., 2023. Analytical solution of two-
reached. Addition of alumina is recommended because it increases dimensional unsteady heat conduction in Multi-Layered Cylindrical Sectors applied
to thermal plug and abandonment of oil wells. Int. J. Therm. Sci. 188, 108180.
combustion stability in the Fe2O3–Al system, decreases material ejec­ De Andrade, G.S., Nascimento, E.J.G., de Lemos, M.J.S., 2024a. Finite integral transform
tions and, consequently, increases thermal efficiency. However, alumina solution of unsteady heat conduction applied to thermal plug and abandonment of
dilutions higher than 40% is not recommended because it can prevent oil wells. Appl. Therm. Eng. 236 (B), 121576.
De Andrade, G.S., Nascimento, E.J.G., de Lemos, M.J.S., 2024b. Transient heat transfer in
ignition and formation of a self-sustained reaction. Also, in the context of concentric cylinders using periodic boundary condition and asymmetric heat
plugging and abandonment scenarios, the use of additives to regulate generation applied to thermal plug and abandonment of oil wells. Int. J. Therm. Sci.
reaction rate and heat generation, such as the alumina dilution exam­ 196C, 108689.
De Andrade, G.S., Nascimento, E.J.G., de Lemos, M.J.S., 2024c. Finite integral transform
ined in this study, may address thermite plugging challenges such as with homogenized boundary conditions solution of transient heat conduction
damage to cement and rock surrounding the plug due to thermal stress. applied to thermal plug and abandonment of oil wells. Geoenergy Sci. Eng. 233,
Adding aluminum beyond stoichiometry to the Fe2O3–2Al system 212530.
De Souza, K.M., de Lemos, M.J.S., 2021. Detailed numerical modeling and Simulation of
increases reactivity, facilitating faster reaction propagation. However, Fe2O3-2Al thermite reaction. Propellants, Explos. Pyrotech. 46, 806–824.
an excess of aluminum, while advantageous for ignition, raises concerns De Souza, K.M., de Lemos, M.J.S., 2023a. Advanced one-dimensional modeling of
due to higher particle dispersion and the risk of violent reaction prop­ thermite reaction for thermal plug and abandonment of oil wells. Int. J. Heat Mass
Tran. 205, 123913.
agation, potentially jeopardizing wellbore and plugging material
De Souza, K.M., de Lemos, M.J.S., 2023b. Full Factorial Design Analysis of
integrity. Thermodynamic and Kinetic Parameters in Simulated Thermite Reaction
Propagation. Continuum Mechanics and Thermodynamics.
De Souza, K.M., de Lemos, M.J.S., Kawachi, E.Y., 2022. Thermodynamics of thermite
CRediT authorship contribution statement reactions for a new thermal plug and abandonment process. Continuum Mech.
Therm. 34, 259–271.
Kesiany M. De Souza: Conceptualization, Data curation, Formal De Souza, K.M., de Lemos, M.J.S., Ribeiro, R.R., Marin, A.M.G., 2023. Advanced
isoconversional kinetic analysis of Fe2O3-2Al thermite reaction for plug and
analysis, Funding acquisition, Investigation, Methodology, Project
abanonment of oil wells. Chem. Eng. J. 455, 140725.
administration, Resources, Software, Supervision, Validation, Visuali­ Durães, L., Campos, J., Portugal, A., 2006. Radial combustion propagation in iron(iii)
zation, Writing - original draft, Writing - review & editing. Marcelo J.S. oxide/aluminum thermite mixtures. Propell. Explos. Pyrot. 31, 42–49. https://doi.
de Lemos: Conceptualization, Data curation, Formal analysis, Funding org/10.1002/prep.200600006.
Durães, L., Costa, B.F.O., Santos, R., Correia, C., Campos, J., Portugal, A., 2007. Fe2O3/
acquisition, Investigation, Methodology, Project administration, Re­ Aluminum thermite reaction Intermediate and final products Characterization.
sources, Software, Supervision, Validation, Visualization, Writing - Mater. Sci. Eng., A 465, 199–210.
original draft, Writing - review & editing. Roberta dos R. Ribeiro: Data Fischer, S.H., Grubelich, M.C., 1998. Theoretical energy release of thermites,
Intermetallics, and combustible metals. In: 24th International Pyrotechnics Seminar.
curation, Resources, Validation. Ana M.G. Marin: Data curation, Monterey, California.
Methodology. Paulo G.C. Martins: Conceptualization, Data curation, Harada, H., Takaoka, M., Oshita, K., Takeda, N., Matsumoto, T., 2005. Ash-melting
Formal analysis, Funding acquisition, Investigation, Methodology, process utilizing thermite reaction between Chrominum Electroplating Sludge and
aluminum Dross. Environ. Eng. Sci. 22, 716–724.
Project administration, Resources, Software, Supervision, Validation, Hearn, D.D., Hirschmann, C., Shafer, R.S., 2020. Nano-Thermite Well Plug, vol. 10.
Visualization. Leonardo H. Gouvêa: Data curation, Investigation, United States of America Patent US, 871,050 B2.
Methodology, Resources. Khalifeh, M., Saasen, A., 2020. Types of permanent plugging materials. In: Introduction
to Permanent Plug and Abandonment of Wells. Ocean Engineering & Oceanography,
pp. 97–136.
Declaration of competing interest Khalifeh, M., Saasen, A., Hodne, H., Motra, H.B., 2019. Laboratory evaluation of rock-
based geopolymers for Zonal Isolation and permanent P&A applications. J. Petrol.
Sci. Eng. 175, 352–362.
The authors declare that they have no known competing financial Kristiansen, T.G., Dyngeland, T., Kinn, S., Flatebo, R., Aarseth, N.A., Bp, A., 2018.
interests or personal relationships that could have appeared to influence Activating Shale to Form Well Barriers: Theory and Field Examples. Society of
Petroleum Engineers, 191607.
the work reported in this paper. Lee, F.T., 2006. Managing thermite weld Quality for Railroads controlling the Key
Variables will assure the Quality of thermite Weldments. Welding Journal 85, 24–29.
Data availability Levchenko, A., Clark, B.R., Carragher, P.J., 2023. Eliminate Section Miling in plug &
abandonment applications with Liquid thermite. In: Offshore Technology
Conference. Houston.
No data was used for the research described in the article. Lowry, W.E., Dunn, S.D., 2015. Well Sealing via Thermite Reactions. United States of
America Patent US 2015/0211327 A1.
Lowry, B., Nielson, D., 2018. Application of controlled-porosity material in geothermal
Acknowledgments drilling. In: 43rd Workshop on Geothermal Reservoir Engineering. Stanford,
California.
The authors would also like to express their gratitude to CAPES, Lowry, W., Dunn, S., Coates, K., Duguid, A., Wohletz, K., 2015. High performance
ceramic plugs for borehole sealing. In: IHLRWM 2015. Charleston.
CNPq (407979/2022-0, 405649/2022-3)and FAPESP (2022/03698-8), Malchi, J.Y., Foley, T.J., Son, S.F., Yetter, R.A., 2008. The effect of added al2o3 on the
Brazil, for their support during the course of this work. The authors also propagation behavior of an Al/CuO nanoscale thermite. Combust. Sci. Technol. 180,
acknowledge the early support of PETROBRAS, Brazil, during the first 1278–1294.
Mei, J., Halldearn, R., Xiao, P., 1999. Mechanisms of the aluminum-iron oxide thermite
stage of this research endeavor.
reaction. Scripta Mater. 41, 541–548.
Merzhanov, A.G., Gordopolov, Y.A., Tnofimov, V.S., 1996. On the possibility of gasless
References detonation in condensed systems. Shock Waves 6, 158–159.
Montgomery, D.C., 2013. Design and Analysis of Experiments, eighth ed. John Wiley &
Sons, Danvers.
Achang, M., Yanyao, L., Radonjic, M., 2020. A review of Past, present, and future
Mu, L., Yin, S., Yanping, W., Heyi, L., 1997. The characteristics of combustion in a
technologies for permanent plugging and abandonment of wellbores and Restoration
Centrifugal-thermite process. J. Mater. Sci. 32, 4711–4713.
of Subsurface Geologic barriers. Environ. Eng. Sci. 37.

10
K.M. De Souza et al. Geoenergy Science and Engineering 234 (2024) 212620

Net Zero Technology Centre, "Cutting the cost of well P&A with alternative well Retnanto, A., Yrac, R., Cardena, N., Khan, S., Al-Abssi, A., Retnanto, A., 2023.
barriers," [Online]. Available: https://www.netzerotc.com/case-studies/interwell-th Assessment of resin systems as solution for well integrity challenge in carbonate
ermite/. [Accessed 29 January 2023].. Reservoirs. J. Pet. Explor. Prod. Technol. 13, 1345–1359.
Odawara, O., Ikeuchi, J., 1986. Vacuum Centrifugal-thermite process for producing Salehi, S., Ezeakacha, C.P., Khattak, M.J., 2017. Geopolymer Cements: How Can You
ceramic-lined pipes. J. Am. Ceram. Soc. 69, 85–86. Plug and Abandon a Well with New Class of Cheap Efficient Sealing Materials.
Pena, F.J.C., de Lemos, M.J.S., 2021. Unsteady heat conduction with phase Change Society of Petroleum Engineers, 185106.
applied to a Novel thermal plug and abandonment process. Int. J. Thermal Sci. 170, Sanders, V.E., Asay, B.W., Foley, T.J., Tappan, B.C., Pacheco, A.N., Son, S.F., 2007.
107155. Reaction propagation of four nanoscale energetic composites (Al/MoO3, Al/WO3,
Pena, F.J.C., de Lemos, M.J.S., 2023a. Simulation of multidimensional unsteady heat Al/CuO, and Bi2O3). J. Propul. Power 23, 707–714.
transfer with aluminothermic reaction and phase transition. Int. J. Heat Mass Tran. Schroeder, L.C., Poirier, D.R., 1984. The Mechanical properties of thermite welds in
145 (11), 110802. Premium Alloy rails. Mater. Sci. Eng. 63, 1–21.
Pena, F.J.C., de Lemos, M.J.S., 2024. Numerical Solution of Coupled Heat Transfer with Skjold, M., 2013. Method of well operation. WO Patent WO 2013/135583 A2.
Phase-Change and Thermite Reaction. International Journal of Energy for a Clean Sui, H., 2019. Investigation on Reaction Mechanisms of Nano-Energetic Materials and
Environment 25 (4), 53–65. Application in Joining. PhD. thesis, University of Waterloo, Waterloo, Ontario.
Pena, F.J.C., De Souza, K.M., delemos, M.J.S., 2023. Thermal behavior of Vralstad, T., Saasen, A., Fjaer, E., Oia, T., Ytrehus, J.D., 2019. Plug & abandonment of
aluminothermic thermite reaction for application in thermal sealing of oil wells. Int. offshore wells: ensuring Long-Term well integrity and cost-efficiency. J. Petrol. Sci.
Commun. Heat Mass Tran. 149 (12), 107113. Eng. 173, 478–491.
Perry, W.L., Smith, C.J., Bulian, C.J., Busse, J.R., Macomber, C.S., Dye, R.C., Son, S.F., Walter, K.C., Pesiri, D.R., Wilson, D.E., 2007. Manufactoring and performance of
2004. Nano-scale Tungsten oxides for Metastable Intermolecular composites. Nanometric Al/MoO3 energetic materials. J. Propul. Power 23, 645–650.
Propellants, Explos. Pyrotech. 29, 99–105. Wang, L.L., Munir, Z.A., Maximov, Y.M., 1993. Thermite reactions: their Utilization in
Prakash, A., McCormick, A.V., Zachariah, M.R., 2005. Synthesis and reactivity of a the Synthesis and processing of materials. J. Mater. Sci. 28, 3693–3708.
Super-reactive Metastable Intermolecular composite formulation of Al/KMnO4. Adv. Wang, H., Fang, X., Guo, T., Yao, M., Song, J., Ding, W., 2018. Experimental research
Mater. 17, 900–903. about the effect of particle sizes on the melt-through effects of Al-MnO2 flame Jet
Prentice, D., Pantoya, M.L., Clapsaddle, B.J., 2005. Effect of Nanocomposite Synthesis on thermite. IOP Conf. Ser. Earth Environ. Sci. 186, 012048.
the combustion performance of a Ternary thermite. J. Phys. Chem. B 109, Wittberg, S.A., 2017. Expanding the well Intervention Scope for an effective P&A
20180–20185. operation. In: Master Thesis. Faculty of Science and Technology, University of
Rashid, K., Ahmed, H., Ahmad, M., Ali, S.D., Khan, M., Anjum, A.M., 2019. Successful Stavanger, Norway.
Implementation of resin based Isolation plug RBIP in Annulus to restore Lost well Zhang, H., Ramakrishnam, T.S., Elias, Q.K., Perez Jr., A., 2020. Evaluation of Bismuth-
integrity. A case study from South of Pakistan. In: Society of Petroleum Engineers. Tin Alloy for Well Plug and Abandonment, vol. 35. Society of Petroleum Engineers -
PAPG Pakistan Section Annual Technical Symposimum and Exhibition - paper SPE- Production & Operations, pp. 111–124.
201174-MS, Islamabad. Zhang, H., Ramakrishnam, T.S., Elkady, Y.M., Feng, Y., Elias, Q.K., 2021. Comparative
Reddy, B.S.B., Das, K., Das, S., 2007. A review on the Synthesis of in situ aluminum based Evaluation of Bismuth-Silver and Bismuth-Tin Alloys for Plug and Abandonment,
composites by thermal, Mechanical and Mechanical-thermal activation of chemical vol. 36. Society of Petroleum Engineers - Drilling & Completion, pp. 368–382.
reactions. J. Mater. Sci. 42, 9366–9378. Zhang, J., Sun, R., Li, Z., Li, X., Tao, H., Wang, Z., Yu, S., Zhang, W., 2023. Control model
and Optimization study of temperature distribution applied in thermite plugging and
abandonment technology. J. Energy Resour. Technol. 145, 032601.

11