0 Replies Latest reply on May 26, 2015 3:00 PM by James Sayre

    Thermal contraction simulation failures

    James Sayre

      Hi all,

       

      I've been struggling for awhile now trying to perform what seem to me to be a conceptually simple simulation.  I want to simulate the effect of cooliing an assembly (shown in attached image) from room temperature (say, 280 K) to 4K.  The base ring is aluminum, onto which are bolted a series of copper tabs, which have top copper clamp pieces bolted to them to hold the lip of an alumina lens.  There is considerable differential CTE between the materials, and I want to investigate the stresses it induces as the assembly is cooled.

       

      I use bolt contacts for all the holes in the copper pieces (brown in the image), and no penetration contacts everywhere else that two faces would mate.  I can successfully simulate the effects of gravity and the stresses associated with pre-loading the bolts with no problem.  I've manually entered material property curves (particularly CTE) as a function of temperature going from 4K - 300K.  Some properties I don't have as a function of temperature are left static, but still defined at the same points (4, 20, 77, 100, 150, 200, 250, and 300 K) as the ones that I do have vs. T data for.  Initially I ran sims with only the @300K material properties, and had many of the same problems I'm still facing, only with an upside of "at best, the answer will be very wrong".

       

      The problem in short, is that the solver fails every time. 

       

      I've tried tweaking parameters of the study (admittedly, with little direction in terms of what might help, but I feel like I'm looking for a magical combination of options that may not exist), and nothing has helped.  What few vaguely useful error message I get tend to point to the fixtures, but a) the fixtures are fine if I run the simulation without cooling and b) I have to be careful what I fix since everything shrinks during the cooling and artificially restraining elements of the assembly would introduce spurious stresses.

       

      Note: I've tried taking a "pie slice" of the assembly to include just one copper tab assembly and used circular symmetry fixtures to ease the computation load, to no avail.  I've also tried making a simplified test assembly with the same materials, but that doesn't work any better.

       

      Anyway, I feel like either this seemingly simple simulation is simply beyond the capabilities of the SW engine or else I am making a fundamental mistake in how I define the problem to the solver.  I'm happy to answer any questions about what I've tried, but I'm hoping perhaps someone with more experience (and success) with such thermal analysis can offer some guidance.