In the thermal management of spacecraft, space thermal radiators play a vital role
as heat sinks. The heat dissipation inside the spacecraft will be transported to space
thermal radiators through mechanically-pumped fluid loop (MPFL) or heat pipes. There
is significant published literature on the performance of heat pipe radiator
configuration, whereas only minimal data available on MPFL radiators. A serial
radiator with proven advantages in the ground applications is analysed for usage in
space applications. The parallel radiator configuration is also considered as a baseline
configuration to compare the performance of the serial radiator. This study aims to
bring out the detailed performance analysis of MPFL radiator configurations, parallel
and serial, as a function of various geometric and flow parameters. The analysis was
carried out through coupled thermo-fluid simulations using a commercial software.
Variations in the heat rejected, Radiator mass, and pumping power requirement with
the tube diameter, fin thickness, pitch of the tubes, and fluid mass flow rate are brought out. Design optimization of radiator is carried out using Taguchi-Neuro-Genetic
approach in three steps; 1) Design of Experiments using Taguchi Orthogonal arrays
to obtain the the smallest fractional factorials of coupled thermo-fluid simulations
covering the entire domain of design space and perform coupled thermo-fluid
simulations, 2) An artificial neural network is trained using the coupled thermo-fluid
analysis results to define the objective function by establishing a relation between the
input and output parameters, and 3) Genetic Algorithm to perform multi-parametric
design optimization. The optimization aims to obtain a configuration that provides the
least mass and least pumping power requirement for a given heat rejection. The
optimization results show that the proposed serial radiator configuration is lighter for
given heat rejection and requires less pumping power than the parallel radiator
configuration. By comparing the value of the composite objective function, the serial
radiator proves to be superior to parallel radiator in overall performance.