The continuous exploration of oil wells has increased the demand for plug and abandonment procedures. Current techniques used for well plugging involve the cementing process, which is extremely expensive and difficult to execute. In addition, it has been reported that gaps might be created by cement cracking or shrinking. Aiming to overcome these challenges, a new approach in this area has been investigated. This novel technology, named as Thermal Plug and Abandonment (TP&A), proposes an exothermic chemical reaction that would be introduced through the production tube or the steel casing. Then, the reaction would generate enough heat to melt the surrounding materials and, after the cooling stage, the solidified mass would seal the wellbore. Considering the foregoing, this work aims to investigate this new technology assuming a thermite mixture that would be introduced through the steel casing. For that, a numerical analysis was employed to investigate the heat conduction through the oil well structure. The energy equation was discretized through a finite volume approach and solutions were calculated through the commercial software ANSYS FLUENT. Prior to TP&A simulations, the energy generated by the thermite reaction was approached as a transient heat flux profile, which was calculated and validated based on experimental data available in open literature. With respect to the new technology simulations, the calculated heat fluxes were applied in a multilayered domain containing the steel casing, the cement and cap rock. A one-dimensional analysis was carried out through this domain and results were compared with an analytical solution. Firstly, simulations considered only the heat conduction model and a phase change model was applied later. After that, the wellbore was approached as a two-dimensional axisymmetric domain, which made use of both constant and space-varying heat fluxes profiles. Overall, results indicated that temperatures were high enough to completely melt the steel casing and partially the cement. However, desirable temperatures could not reach the cap rock domain. Also, temperatures heavily dropped through the cement layer, pointing out the thermal barrier behavior of such layer. Comparisons were made between simulations without and with phase change, demonstrating that the last one produced smaller temperatures. Moreover, two-dimensional analysis have shown greater heat losses through the longitudinal direction. Also, comparisons were made between both heat fluxes models and have shown a bigger melted area for the one that was constant in space. Finally, the analysis developed in this work might contribute to advancements regarding this novel P&A technology.