Adrian,
There's a wiki page on one way to automate at
http://topas.dur.ac.uk/topaswiki/doku.php?id=automating_multiple_refinements.
The sorts of equations you can use to extract temperature are in (sorry for self-citation) Stinton and Evans, J. Appl. Cryst. (2007). 40, 87–95.
We've done exactly this type of calibration for our pHeniX cryostat using Al. Obviously thermal expansion is close to zero at low T (below ~50 K) so you have to do something else at these temperatures.
You could use something like Si which has well known negative then positive thermal expansion to help, or ZrW2O8 which shows nte from very low (~4 K) temperature. You could also use various materials with low T phase transitions.
For the Al cell there are data by Wang and Reeber or by Kroeger (thermal expansion). We've used an expression as in paper eq 3 but with coefficients:
a0 4.03255
c1 1.9833E-05
th1 249.97393
c2 7.7112E-06
th2 776.94283
Hope that's some help.
John