[stamford]

I must admit to not knowing of a block test dye kit, I do know about crack testing using a special dye and UV light. I did this many moons ago when I used to work in a foundry. Only problem is this partucluar method required a cleaning tank to take the dye and then washed off allowing it to settle in any hairline cracks, the UV highlighted this if found. Quite how this would work if the fault was internal and not visable is another matter.

I get the feeling that you hit the nail on the head and the current design cannot take the pressures generated and finds the weakest exit point. I think you are going to have trouble pinpointing the cause without spending alot of time and money monitoring the engine and finding any hotspots in the operating range you see it happening.