A novel cooling geometry for subsea variable speed drives

• Published in: Applied thermal engineering Vol. 185; p. 116483
• Main Authors: Militão, Lucas A., Fernandes, Caio D., dos Santos, Diego, Machado, Douglas M., Heldwein, Marcelo L., Rambo, Carlos R., da Silva, Alexandre K., Barbosa Jr, Jader R.
• Format: Journal Article
• Language: English
• Published: Oxford Elsevier Ltd 25.02.2021; Elsevier BV
• Subjects: Boards; Computational Fluid Dynamics; Cooling; Cooling rate; Deep sea; Deep-sea oil exploration; Deviation; Enclosures; Equivalent thermal network; Fluid flow; Frequency inverter; Genetic algorithms; Geometry; Heat sinks; Heat transfer augmentation; Low temperature; Mathematical models; Oil exploration; Seawater; Studies; Thermal management; Variable speed drives; Velocity distribution; Deep-sea oil exploration; Thermal management; Equivalent thermal network; Computational Fluid Dynamics; Frequency inverter; Heat transfer augmentation
• ISSN: 1359-4311; 1873-5606
• DOI: 10.1016/j.applthermaleng.2020.116483

Experimental and theoretical analyses are conducted to evaluate the passive cooling performance of a novel geometry for subsea variable speed drives, a common piece of equipment in deep-sea oil exploration. Relying on the sea water as a low-temperature thermal reservoir, the new design forms an enclosed, annular space with centrally located modular boards that compose the power electronics inverter. Buoyancy-induced motion of a dielectric coolant conveys the heat dissipated by the electronic boards to the sea water through the outer and innermost walls of the annular enclosure. A thermal network model is implemented and used to optimize the enclosure geometry through a genetic algorithm, which served as a reference for a scaled experimental setup. A Computational Fluid Dynamics (CFD) simulation of the conjugate heat transfer yielded temperature distributions on the electronic boards and temperature and fluid velocity fields inside the enclosure. A comparison between the experimental data and the modeling results indicated a good agreement, with average RMS deviations of a modified Nusselt number of 7.0% and 8.5% for the thermal network and CFD analysis, respectively. For a 140-W operating point dissipation rate in the scaled test setup, the thermal network and the CFD models presented maximal deviations of 4°C and 2.3°C with respect to the heat sink temperature measurements. •Natural circulation cooling of a dielectric oil in an annular enclosure is studied.•Experiment was designed based on a thermal network (TN) model and genetic algorithms.•Oil velocities and circuit board temperature distribution were determined via CFD.•TN and CFD predicted the Nusselt number with RMS deviations of 7.0% and 8.5%•New geometry is a promising compact solution for subsea applications.