The "natural" protective oxide layer is formed extremely fast (some hundreds of pico seconds) and is very thin (4 nano meter).
So it appears almost impossible to do something on the "bare" / native alu. I think I once read about using some acid treatment to avoid that natural layer being formed, in order to treat the bare material.
From what I remember from a work, an anodisation layer is some hundredfolds thicker.
About the U cross section, the video from Hambini or so, shows two U shapes, one narrower, one wider, shifted inside eachter, forming a closed rectangle with two double sides the overlap), the mating surfaces glued, although the glue appears everywhere there, even on the inside where no surfaces mate.
One may wonder, there has to be some tolerance, between the narrower and the wider, to give the glue some space to be there / some thickness for strength of bond. Now, the glue has probably different mechanical properties than the aluminium U profiles, causing torsion forces to be able to act differently on one U shape than on the other, causing the U profiles to undergo stress between eachother, with the glue having to absorb the torsion difference, which may cause it to detach gradually, especially, and first, at an end of the double profile, the crank basis due to the highest leverage there, as also sketched by Hambini.
About glue and water, and maybe eventual galvanic corrosion (but that would dissolve and I don't see powder in the video), normally, the glue should keep water out. Glue can withstand water when it can dry out again, it's when long time wet that
glue fails and I don't think that water can permanently stay in a crank since it goes up and down / inverted position, in the case of an entrance for water.
It would be interesting to see the internals of a replacement crank/spider, to compare these with the faulty, the differences may give a clue on the cause of the losening.
So it appears almost impossible to do something on the "bare" / native alu. I think I once read about using some acid treatment to avoid that natural layer being formed, in order to treat the bare material.
From what I remember from a work, an anodisation layer is some hundredfolds thicker.
About the U cross section, the video from Hambini or so, shows two U shapes, one narrower, one wider, shifted inside eachter, forming a closed rectangle with two double sides the overlap), the mating surfaces glued, although the glue appears everywhere there, even on the inside where no surfaces mate.
One may wonder, there has to be some tolerance, between the narrower and the wider, to give the glue some space to be there / some thickness for strength of bond. Now, the glue has probably different mechanical properties than the aluminium U profiles, causing torsion forces to be able to act differently on one U shape than on the other, causing the U profiles to undergo stress between eachother, with the glue having to absorb the torsion difference, which may cause it to detach gradually, especially, and first, at an end of the double profile, the crank basis due to the highest leverage there, as also sketched by Hambini.
About glue and water, and maybe eventual galvanic corrosion (but that would dissolve and I don't see powder in the video), normally, the glue should keep water out. Glue can withstand water when it can dry out again, it's when long time wet that
glue fails and I don't think that water can permanently stay in a crank since it goes up and down / inverted position, in the case of an entrance for water.
It would be interesting to see the internals of a replacement crank/spider, to compare these with the faulty, the differences may give a clue on the cause of the losening.
