A Hybrid Solution Method Using ThermalDesktop and TMG to Solve LISA and other Large Thermal Models 2005-01-3055
LISA (Laser Interferometer Space Antenna) is a constellation mission designed to detect gravitational waves using laser interferometry by measuring the laser path length between two free floating proof masses in opposing spacecraft separated by ∼5 million kilometers. As such, the LISA mission requires unprecedented thermal stability and, due to difficulties with ground testing, relies heavily on analysis.
A Finite Element Model (FEM) was selected as the baseline modeling technique, using an identical mesh for each phase of STOPG (Structural-Thermal-Optics-Performance-Gravity) analyses. Two thermal codes were evaluated (TMG® and ThermalDesktop®) that both interfaced well with FEM software. TMG calculates element mid-side and centroid temperatures and then interpolates/extrapolates to derive nodal temperatures, which could introduce potential errors. ThermalDesktop provides a better element formulation for conduction for our particular model (which had a number of nodal connections through bar elements), directly calculating nodal temperatures with no extrapolation. Therefore, the final recommendation was to use ThermalDesktop® for subsequent thermal analyses.
Running ThermalDesktop on a PC with the 32-bit Windows operating system resulted in a maximum of 2 GB of addressable memory. Consequently, a finite number of conductors (∼25 million) may be included in the generated SINDA model, requiring significant filtering of the radiation output. Furthermore, the solution scheme employed by SINDA is not well suited for models of this size and complexity.
TMG provides a more robust and faster solution algorithm than SINDA, but may include fewer conductors (∼8 million). A 64-bit version of TMG is currently available for workstations, and should soon be available for Linux and 64-bit Windows. This would allow for the inclusion of far more radiation conductors. This paper describes a proposed method for using ThermalDesktop to generate the radiation and conduction network, but using TMG to solve for the temperatures, reducing the analysis time necessary to generate temperatures for thermal distortion.