Monday, May 10, 2010

Time-temperature Superposition and going nuts

Just got a bollus of new samples that need their viscoelasticity measured. The measuring is hardly the problem - it's when I have to do the time-temperature superpositon that I start getting blurry-eyed. I always do them by hand. Given the software package I am working with, the automatic superposition is passable at best, most often useless and sometimes you have to wonder how any self-respecting machine could crank some of the results out just that way. I suspect [1] that the software has improved some, so that if this old machine ever gives up the ghost [2] then my life will be simpler.

Let me show you an example. Here's what the raw data: Then here's what the automatic software shiften things too:
Look at that. There all sorts of slack that can be taken up. It's disgusting.

While the final plot that I made by hand isn't clean, (the sample was getting too soft at the highest temperatures and things got noisy) it is so much better at the higher shears. As you can tell, these plots were made in Excel which certainly is less than desirable but much better than the instrument's software, but that also then requires that I cut-and-paste the data between the different programs. So that's my day. After a running a dozen of these, I'm pretty much brain-dead.

[1] Hope and pray?
[2] Not likely at all. Rheometrics built the RDA II to survive a direct ICBM hit. It will be something for the cockroaches to play with in the post nuclear era.


Materialist said...

This sounds like an excellent little issue to outsource to underemployed coders. I would guess that for $100 (payment or prize) or less someone would write an Excel macro that would shift an example set of inputs so they overlap as desired.
I think many organizations would benefit from having a staff person dedicated to data manipulation macros.

John said...

Personally I think that the solution is part technical, but also part art/experience. For some materials (acrylates in particular regardless of the moiety), the system works very well and then it is just a matter of reducing the residual.

But for other systems, especially with fillers, you can't just go for a residual minimization. Instead you have to go with select portions of the curves. Maybe the right side, maybe the middle, maybe... It's something that you just have an eye for. something picked up over time.

Anonymous said...

"Any self respecting machine" Ha Ha!!! No no, these machines are stupid. I agree with Materialist that many organizations would benefit from somebody dedicated to data manipulation macros. Look on the bright side, at least you didn't run a series of samples only to realize residual surfactant from the emulsion polymerization renders all the data useless....


John said...

That does sound nasty.

Was that surfactant then used by your mother to wash out your mouth after all the cursing and swearing?