An experiment is performed to measure the thermomagnetic convection effect in a prototype immersed coil. A magnetic fluid (Midel vegetable oil - CoFe 2 O 4 Cobalt ferrite magnetic fluid) with nanoparticles volume fraction of 5% is used as a coolant for this purpose. The test cell consists of a copper coil, in the form of a two-wire conductor, chosen to obtain two identical coaxial resistors. The coil entirely immersed in ferrofluid is set into an Aluminium tank closed at the top by a PVC cap. When the directions of the currents flowing in the two resistors of the coil are opposite, the generated magnetic fields are compensated in the device and the thermomagnetic convection effect is deactivated. On the other hand, if the current directions are the same, a magnetic field is applied in the test cell and the thermomagnetic convection effect is activated. In both configurations, the same amount of heat is dissipated by the Joule effect at the solenoid. Two sensors are inserted in the fluid and at the coil respectively, to show temperature evolution during experiment progress. The current direction in each resistor is initially the same, then reversed every 60 minutes to activate/deactivate the magnetoconvection. The results corresponding to the time evolution of the coil temperature show reproducible crenelations with equal amplitude of 2,2 ◦ C approximately. Hence, when the thermomagnetic convection is active, the coil temperature decreases, revealing the impact of the thermomagnetic convection in the cooling operation. The same process is numerically tested, with a 2D axisymmetric model of the setup, using the finite element method. We use the magnetic fluid properties deduced from the experiment to improve the cooling efficiency of the coil. Due to the presence of an additional magnetic force, the fluid flow around the coil is modified. As a result, a new convection cell arises in the ferrofluid, and cools down the heated coil. Experimental results in agreement with numerical ones prove that the proposed model and the applied boundary conditions are appropriate for the modelling of the actual setup.