i have what look like a tiny crack in the input pipe of my water block for the cpu, it is only very small like a hairline and about 3 mm in length and as far as i can see no water has been leaking, so my question is is there anything i can do to repair this as a precautionary measure like glue or something
looks like a tiny crack in the water block shall i repair it ?
- Thread starter west-wot
- Start date
Depends what made it crack. If it's a manufacturing defect or the result of one-off abuse then lathering it with epoxy will be very effective at waterproofing it. If it has occurred as a result of repeated or periodic stress (not been pulling on the tubes I trust?) then this will only be a temporary measure.
Impressed that Kae mentioned drilling a hole at the end, gold star to that man. Blunting the end of the crack in this fashion much reduces the stress concentration at the end and makes crack propagation much less likely. Still, sticking the sides of the crack together with epoxy also makes the crack spreading pretty unlikely given how low the pressures in watercooling systems are.
I'd patch it. I've got a radiator with a bleed screw removed and the hole filled with epoxy, and a gpu block with stripped threads that I epoxied barbs into. Neither are giving me any trouble at all.
Impressed that Kae mentioned drilling a hole at the end, gold star to that man. Blunting the end of the crack in this fashion much reduces the stress concentration at the end and makes crack propagation much less likely. Still, sticking the sides of the crack together with epoxy also makes the crack spreading pretty unlikely given how low the pressures in watercooling systems are.
I'd patch it. I've got a radiator with a bleed screw removed and the hole filled with epoxy, and a gpu block with stripped threads that I epoxied barbs into. Neither are giving me any trouble at all.
Just to add to the crack propogation theory discussion. In order for a fatigue crack to propogate you need a cyclic stress field. The dominant cyclic stress these acrylic tops see is almost totally due to thermal expansion of the block as it heats up after switch on. This cyclic stress will sit on top of whatever static stress is in the block top, two of the major sources of static stress are the bending stresses due to the block hold down mechanism, and the stresses around the threads of the barb holes, the tighter these are the higher the local stresses. The static stress due to water presure is very very small. The combination of high mean stress and a cyclic stress are what dictates the fatigue life of a component. For this reason overtightening the block might give you better contact but at the expense of potentially reducing the life of the acrylic top. Same goes for overtightening barbs.
Temperature variation will be from ambient to at most ten degrees over ambient, and while cyclic the period of the cycle will be long. I think you're saying the stress arises from the different thermal expansion coefficients of copper and acrylic. Which direction do you have in mind?
Parallel to the board stresses will result from the differing strains and that the top and base are firmly bolted together (one wonders why the holes in the acrylic aren't drilled 1/2mm bigger in light of this). Perpendicular to the board the stresses would be carried by the threads in the acrylic, possibly explaining why ek have clear drilled these holes. Since they are not threaded, the expansion of the acrylic leads to local compression around the screw heads. I imagine this is the source of the cyclic stress you have in mind?
This suggests my plans of epoxying milled pieces of acetal onto a copper plate for use as waterblocks will be changed to include screws and O rings.
Parallel to the board stresses will result from the differing strains and that the top and base are firmly bolted together (one wonders why the holes in the acrylic aren't drilled 1/2mm bigger in light of this). Perpendicular to the board the stresses would be carried by the threads in the acrylic, possibly explaining why ek have clear drilled these holes. Since they are not threaded, the expansion of the acrylic leads to local compression around the screw heads. I imagine this is the source of the cyclic stress you have in mind?
This suggests my plans of epoxying milled pieces of acetal onto a copper plate for use as waterblocks will be changed to include screws and O rings.
It may only be one cycle per day but it depnds entirely on what the mean stress is as to how long takes for a fatigue crack to start. Fatigue is a function of stress range and number of cycles as you probably know. So it all depends on what the static stresses are in an assembled block when it warms up, initially assembled whilst cold I might add. Differential thermal expansion is the cheif culprit for fatigue cracks in these water blocks as far as I see it.
As far as the directions of stress goes it is hard to predict, it will just go where it wants to go. I have thought about modelling a water block in Abaqus sometime, but I don't have properties for acrylic and accurately modelling the hold down forces is a bit of an unknown really.
As far as the directions of stress goes it is hard to predict, it will just go where it wants to go. I have thought about modelling a water block in Abaqus sometime, but I don't have properties for acrylic and accurately modelling the hold down forces is a bit of an unknown really.
