So, after verifying my leak-fix on three assemblies, I was ready to apply it to the rest of my heads. JB Weld's temperature spec of 500F continuous and up to 600F intermittent sounded plenty adequate for my needs. Then, Dickiebird asked if I had considered a possible issue with temperature cycling the epoxy. After some more thought and web research, I could imagine JB Weld's loosely worded high temperature spec might not actually include an ability to be temperature cycled. And, I ran across lots of stories of users' repairs gone bad that were blamed on temperature cycling.
I decided to run my own test before repairing any more of my heads. I made a test bed by drilling and tapping a 3/8" aluminum plate for five spark plugs. Stainless washers were epoxied into three of the positions and temporarily secured with spark plugs tightened with a clamping force of about 100 psi while they cured just as was done on the first three head assemblies. For comparison, I also threaded in two additional plugs with stainless washers sealed against the plate using gaskets I made from commonly available 1/64th inch thick brown automotive gasket material. In fact, it's the same material I'm using for my exhaust flange gaskets. I decided, at the last minute to include these gaskets in my testing as another potential solution. After considering some of the other readers' comments; and, depending upon my test results, I might find myself machining a batch of ridged compression washers. Finally, I sprayed portions of the plate with a Rustoleum 500F high heat paint that I also want to test. The black painted surfaces will also provide nice targets for my non-contact thermometer.
As a control, I JB Welded a similarly cured sixth washer to a separate plate that will not be heated. This control might be used later to quantitatively compare the forces needed to push a cycled vs. non-cycled epoxied joint to failure.
After a 15 hour cure and with the plugs torqued to 75 in.-lbs., I used a torch to heat the back side the plate for 2-3 minutes until the target test temperature was reached and then held for about 30 seconds on the washer side of the plate. After 45 minutes or so when the plate had cooled back to room temperature, I repeated the cycle using the same test temperature. After the fifth cycle I un-torqued and re-torqued all the plugs a half dozen times each before removing them and attempting to dislodge the three epoxied washers from the plate by pressing hard against their edges with a wood block. After the tenth heat cycle, I repeated the torquing tests and then I also baked the entire plate of re-torqued plugs in an oven for three hours at the test temperature. The purpose of this constant temperature bake was to accelerate the aging of the resin. I performed the same ten cycle heat, torque/un-torque, and aging bake for test temperatures of 300F, 350F, and 400F.
I really don't expect the JB-Weld to even come close to 300F on my radial because of the huge finned heads and massive prop wash. The reason for the higher test temperatures is to accelerate the failure mechanisms created by temperature cycling the epoxy and to force the failures to occur as early as possible during testing. Also, it's an opportunity to learn more about JB Weld for a possible future application.
At the end of the 300F tests, I began having doubts about a gasket solution. Both gaskets had become stuck fast to their washers which was a good. But, the side of the gasket against the smooth aluminum plate did not stick; and the material had become very hard and its surface glazed from the heat. These gaskets would probably continue to seal until the plugs were removed the first time. But, it would be difficult to re-install them with the washers in their original orientations; and the gasket material no longer seemed to have the compliance needed to re-seal the surfaces against combustion pressures. I've dealt with this on old automobile engines. Some of their irregularly shaped gaskets, if not damaged during disassembly, can sometimes be returned to their original positions and re-used but only in low pressure sealing applications.
After successfully passing the 400F tests (total accumulated 30 temperature cycles, 36 torque/un-torque cycles, and 9 hours total bake time) I decided it was safe to continue on with the JB-Weld repair of my heads.
As one of the photos shows, the gaskets had by this time completely deteriorated. Since none of the epoxy bonds had yet failed, I continued on with the 450F test. At a 450F head temperature the Loctite 620 sealing my valve cages would decompose and the soft solder used on the exhaust flange would melt, and so further testing was just more for my curiosity.
I modified the 450F test slightly by leaving out one of the spark plugs to see if the clamping pressure had been playing a major role in maintaining the bonds on either side of the washer. I also stopped the gasket testing since there wasn't anything left to test. After five 450F heat cycles, all three washers were still holding tight; and so I baked the test plate containing two torqued spark plugs and one open washer at 500F for three hours. At the conclusion of that test the washers were still being held in place by the JB Weld, but using the same moderate force I was able to dislodge all three of them from the plate. Inspection of the washers showed it was the bond between the epoxy and stainless that released.
In any event, I came away from this little diversion with a healthy respect for JB Weld. One important manufacturer's application tip I ran across in my research is that cleaning the surfaces to be bonded with acetone or lacquer thinner is recommended, but use the use of alcohol is specifically discouraged. I previously assumed that JB Weld became hard and brittle when cured. But after recently mixing several batches of the stuff and monitoring the leftover I discovered that there is always a small degree of compliance left in the material that probably accounts for its ability to be temperature cycled. - Terry
