Hi Minh Thanh.
As you still seem confused as to what has happened to your crank: (Maybe you think we are the cranks, and you may be right...? - But we think after decades of working in industry and having met these problems before, then we have an opinion that is worth presenting). Your choice - of course - as to what you do with it. But please be patient and learn from those who have really had this problem and know what they did to fix it?). And perhaps we can only answer with hypothesise, or opinions or even just open ideas, as not all the information is available for us to check the design, material, machining process, running conditions, etc. (e.g. if the engine was pre-igniting the forces would not be as "designed" - but we cannot know that unless you tell us).
You want direct answers, but if we say "You made it wrong" or "you used the wrong material", or "There was another fault in the engine", that does not help. You are asking "WHY? (did it fail)", an entirely different question, which needs us to explain the possible causes of failure. We cannot give exact cause of failure because we do not have all the information to do so. So we express our ideas
from our experience. Take it or leave it ("Be patient and learn"?). We are not forcing you to do anything.
Did you examine the cracks under a good lens or microscope to see if the cracks propagated from a tool mark or sharp corner? (40 x magnification of the picture loses resolution - a cheap toy microscope can do 100x magnification to help, if you can borrow one from a child?).
I can't see clearly from just this picture. But the amount of (inconsequential) scoring on the crank-web suggests a sharp-cornered tool has been used there. If the crank-pin is 5mm diameter, then I GUESS the sharp tool has a "radius" of less than 0.2 mm, but as I cannot see the side view of the journal I am guessing at what I am trying to measure. The radius could be as small as 0.06 - at a guess - from what I can see and try and measure off the picture.
But a radius in the corner of 0.2 mm for a 5mm pin would give a stress concentration factor of more than 1.9.
AND a radius of 0.06mm would have a stress concentration factor of in excess of 3: - That means an increase of >50% of the real stress seen by the pin, at some point where there is also a small tool mark that can propagate cracks easily. A real killer for fatigue failures of this type.
In fact, due to the stiffness of the crank web varying in different directions, these SCFs may be way more than "my guess" - even by as much as a factor of 10.
I think that others, as well as I, think the fatigue failure you have is that you are working in the "sloping" zone of the fatigue and endurance limit file picture: I.E. the higher stressed zone where fatigue occurs. You need to be working in the lower stressed zone (below the horizontal line).
My hypothesis is that the "normal" stress on this design of crank is OK (from what you intimate that it has worked before: "
I made 4 similar crankshafts, but this is the first time it broke "?) but that a sharp corner, machining mark, or other stress raiser has raised the stress
by more than 3 times to induce the fatigue failure. Reducing this stress raiser is not hard, just a small change to tooling and process.
So again I encourage you to find a way of getting a smooth transition by a larger radius from the crank-pin to web with a larger radius on the cutting tool (0.4mm or more?) - and NO SCORING.
And a more fatigue resistant material will be a benefit as well. A steel with twice the tensile strength will have something like twice the resistance to fatigue - perhaps moving you from the sloping zone to below the flat line?
I hope you find some patience and learning from us?
K2