...

The steel mandrel was a 6" OD, 60" long, 1/8" thick steel tube. This mandrel was notably quite round without any machining, but it was extremely difficult to remove the cured parts from the mandrel. We think the steel was a structural steel tube since it has a weld.

Aluminum tube

To remove parts more easily, we tried using aluminum instead, because we hoped that cooling the aluminum mandrel would cause it to shrink. However, thermal separation never worked; we tried (water) ice, dry ice, and liquid nitrogen, in addition to heating the fiberglass before cooling the mandrela fiberglass composite has roughly the same coefficient of thermal expansion as aluminum, so this was ineffective.

The first of the aluminum mandrels was a 6" OD, 60" long, 1/8" thick aluminum tube. However, the mandrel was about 0.125" out of round (i.e. roughly an ellipse, where the major axis was about 0.125" longer than the minor axis). We hypothesize that this is because aluminum tubes are extruded at viscoplastic temperatures, to the aluminum deforms slightly after extrusion until it cools sufficientlythe mandrel was not shipped well, and was likely crushed during shipping.

The second of the aluminum mandrels was a 6" OD, 60" long, 1/4" thick aluminum tube. This mandrel was out of round by about 0.030", seemingly regardless of the supplierdue to the aluminum extrusion process.

Bidirectionally tapered tubes

Again, to remove parts more easily, we tried a tapered mandrel. We used two 30" long aluminum tubes to create a mandrel where the ends were 6.00" OD, and the center was 5.94" OD, again 60" long with a roughly 1/4" wall thickness. To keep them together, we glued epoxied a coupler section to the middle with Proline. The tubes were fairly easy to remove from this mandrel.

...