The test was performed in my shop using the work bench as the test jig. I was planning to rig up some 5/8" thick fiberglass and bolting it to the bench using some big backing plates. After running some numbers it appeared the 3/4" plywood should handle the loads.
I used a calibrated digital load cell to measure the load being gradually applied to the seacock. The load was measured in pounds and set to capture the moment of failure or "peak hold". To apply the load I used an old Lewmar self tailing winch attached to the end of the work bench.
The thru-hull was fitted to my bench and tightened down well using a spud wrench. The valve was then threaded onto it and the elbow threaded into the valve. I was able to get 4 full turns on the valve to thru-hull.
The ball valve to thru-hull is a known thread mismatch, NPS to NPT, but this is what is very often done by builders, boat yards and DIY's. I wanted to mimic a "real world" type installation and most, at least in bronze, are of mixed parts, not just one brand. It was not the thread mismatch that failed.
This image shows a close up of the failure. It's not a surprise that it failed where it did.