Richard, I saw your presentation on expansion fittings and remembered that I had heard of a coefficient to use withh the PVC table for the expansion of steel conduit. This would have to be the type of conduit they had the problem with in Bakersfield, because the conduit served as a ground fault path (ie, it was conductive whereas PVC would not have been. I did a search and found this:
*A Fine Print Note in Section 300.7(B) of the NEC® refers the user to the Expansion Characteristics of PVC ,Table 352.44(A) for Rigid Nonmetallic Conduit
and suggests multiplying the lengths in that table by 0.20 in order to obtain a nominal number for steel conduit. Since the coefficient of steel conduit is
between 2-3 times less than that of PVC conduit you would need more expansion fittings for PVC conduit, for a given temperature and length than for
steel conduit. We have used the coefficient of expansion of steel, rather than the 0.20 multiplier, to calculate the exact length of change figures in Table 3.
Accessed from:Question 8 of http://www.steelconduit.org/qa_home.htm