Preventing fatigue failure of copper refrigerant lines that connect compressors to condenser coils is a critical aspect of designing a new scroll compressor chiller confi guration. Traditionally, R&D teams use a combination of physical testing and conventional finite element analysis to qualify the lines, especially to identify and correct resonances that could cause a reliability problem. But this approach is too slow to address chiller designs that have more than 100 refrigerant-line confi gurations.

Trane has developed a new automated workflow capable of developing robust designs. The methodology combines

Design of Experiments (DOE)

DOE samples the design space focusing on an optimal set of uniformly spread virtual experiments, maximizing the amount of information that can be obtained for a given simulation cost.

Response Surface Modeling (RSM)

RSM uses discrete information from DOE to create a best fit continuous model, allowing to predict result values for any combination of design variables without any simulation effort.

Numerical Optimization

Single or multi-objective optimization algorithms progressively adjust design variables of any simulation model to best match target response objectives, either directly on the simulation workflow or on a previously built RSM.