I am interested in quantifying ppm level
(i.e. 100 - 1000ppm) water content in hydrocarbon mixture on a TCD. Carbon number in mixture ranges from C1 through C12. I do not need to quantify anything but water. Does anyone have suggestions for column/columns and method?

If you can use a packed column, try 2m x 1/8" Porapak N, He carrier 25 mL/min, 140C. Water elutes in about 5.5 mins. TCD sensitivity should be OK for 100 ppm water but only just (at least for me). Calibration against known concentrations can be a problem. I used standard addition to methanol containing 1500ppm water, but perhaps you can find or make something drier.

I would suggest using Hayesep N as opposed to Poropak N. Poropak has been having tremendous shrinkage problems for the last several years. Also, you may be able to increase sensitivity by choice of carrier gas, choosing one with the maximum difference in thermal conductivity from water.

The measure of water from a sample of hydrocarbon is pretty tricky. I would really be skeptical of the GC injection/sampling technique and the uniformity of the sample. The differences in density will cause sampling problems. I would lean more along the lines of adsorbing the water and desorbing through Karl Fisher Titrations.

Instead of the Karl Fisher technique, take a look at the Meeco Instruments 'Iceman' portable analyser. It was designed for use with refrigerant gases and has a linear detection range from 1 to 1000 ppm w/w. Analysis times are typically less than 4 minutes. Meeco manufacture other moisture analysers and I'm sure they can meet your requirements.

I'd recommend Karl Fisher