Let's suppose that we use to identical length of coax, and on each
one, we connect the shield to the inner conductor at each of its ends.
If we use these two independant (?) sections as a substitute for an
open-wire line, could we use the usual formula Z=276*log(2S/D) to
compute the impedance of that line?

In that case, the characteristic impedance Zo of the coax could be
neglected. And what about the matched losses in dB/100' of the
resultant line? Twice the one of the coax?

73 de Pierre VE2PID