5 ms·

It's cool that something good can still come out of this "disaster" - it's a shame that it wasn't kept up, but if this sort of forensic analysis reveals ways to

It's cool that something good can still come out of this "disaster" - it's a shame that it wasn't kept up, but if this sort of forensic analysis reveals ways to improve cable manufacturing or maintenance routine improvements, that feels like a pretty good win

adrian_b 3y ago
It is well known that zinc, like all metals with low melting temperature, flows slowly under high stress. So this failure was easily predictable. I assume that the cables have been designed for a much shorter lifetime, so it was expected that either the cables would be replaced or the radiotelescope would be decommissioned many years ago. Nevertheless, the choice of pure zinc for the cable sockets is somewhat weird, because it guarantees a short lifetime. Had a zinc alloy been used, like ZAMAK (Zn-Al alloy), the flowing speed would have been much less and the lifetime of the cables would have been greater. Alloys are much more resistant to plastic deformation and flowing than pure metals, because the atoms with different sizes cause defects in the crystal structure that prevent the easy slipping of the atom planes over each other.
- roberthahn 3y ago
  Could you cite sources? Specifically, when was it discovered that zinc flows under high stresses? By whom? (Not a materials engineer so I’m not sure what to search for)
  - adrian_b 3y ago
    Any handbook about the strength of materials has a chapter about creep a.k.a. cold flow. The handbooks from immediately after WW2 already included such a chapter, but I believe that the first studies of this problem must be much older. When any metallic structure is designed, it must be verified that it will not fail in any of the possible modes, including due to flow over the intended lifetime. See in: https://en.wikipedia.org/wiki/Creep_(deformation) https://en.wikipedia.org/wiki/Creep_(deformation) at "Temperature dependence". By the approximate rule mentioned there, zinc begins to have non-negligible creep already above minus thirty Celsius degrees, so at normal ambient temperatures you must always compute the creep of zinc for any structural design. On the other hand, metals like iron or copper have negligible creep at room temperature, even when pure. Aluminum and magnesium begin to have non-negligible creep at temperatures only a little above normal ambient temperatures. Hard alloys can have much lower flowing speeds than the metals included in their composition. In integrated circuits, the metal connections are affected by electromigration, which is the flowing of the metal due to electrical current instead of mechanical stress. The electromigration properties and creep properties of a metal are closely related. In the beginning, the ICs used pure aluminum for interconnections, but when their size was reduced, the connections began to fail after a too short lifetime. The first solution for this problem was the replacement of pure aluminum with harder aluminum alloys, including small quantities of copper and/or silicon. When the ICs became even smaller, the aluminum alloys had to be replaced with a metal having a higher melting temperature, i.e. copper, which fortunately also has a lower resistivity.
    - roberthahn 3y ago
      Thank you for the information! Sometimes it’s worth pausing a moment when seeing a “It’s well known that…” to check the timeline because the construction of Arecibo might well have taken place (or planned) before it was known (let alone well known). Edit: a bit more searching suggests that this was studied in 1947 (Andrade’s Creep Law and the Flow of Zinc Crystals, by AH Cottrell)
    - RajT88 3y ago
      This feels like the facility was in operation for so long, that the people who knew about this potential problem all retired and with them went that knowledge.
  - sheepshear 3y ago
    Andrade is the namesake of Andrade creep due to his work in the early 1900s, but the existence of creep had been known for a long time by then. I'd imagine smiths have been aware of creep throughout the history of metallurgy. https://royalsocietypublishing.org/doi/10.1098/rspa.1910.0050 https://royalsocietypublishing.org/doi/10.1098/rspa.1910.005...
- nullc 3y ago
  https://www.nsf.gov/news/special_reports/arecibo/Arecibo-Telescope-Collapse-Forensic-Investigation-508c.pdf https://www.nsf.gov/news/special_reports/arecibo/Arecibo-Tel... Go to pdf page 53-ish. They show the zinc doesn't continue to creep if the load is below a threshold-- technically the zinc doesn't creep if it's not subject to persistent shear forces, which is won't be if the cable load is below a threshold as the wires take up the load. This essentially results in a long term capacity which is different from the short term capacity. Unfortunately the ratio between the two depends on the wire splaying geometry which isn't well controlled, resulting in wide differences. The report concludes that if it had been built to a safety factor of 3 rather than 2, the issues wouldn't have been experienced. Alternatively, the failure could have been avoided by reacting to flow over some threshold (which was noticed well in advance, but not reacted to) Presumably we don't see similar failures in sockets in suspension bridges because they're built at a safety factor of 5+ and usually that's a SF over their rarely reached maximum load rather than a factor over their 24/7 load.

adrian_b 3y ago

It is well known that zinc, like all metals with low melting temperature, flows slowly under high stress. So this failure was easily predictable. I assume that the cables have been designed for a much shorter lifetime, so it was expected that either the cables would be replaced or the radiotelescope would be decommissioned many years ago. Nevertheless, the choice of pure zinc for the cable sockets is somewhat weird, because it guarantees a short lifetime. Had a zinc alloy been used, like ZAMAK (Zn-Al alloy), the flowing speed would have been much less and the lifetime of the cables would have been greater. Alloys are much more resistant to plastic deformation and flowing than pure metals, because the atoms with different sizes cause defects in the crystal structure that prevent the easy slipping of the atom planes over each other.

roberthahn 3y ago
Could you cite sources? Specifically, when was it discovered that zinc flows under high stresses? By whom? (Not a materials engineer so I’m not sure what to search for)
- adrian_b 3y ago
  Any handbook about the strength of materials has a chapter about creep a.k.a. cold flow. The handbooks from immediately after WW2 already included such a chapter, but I believe that the first studies of this problem must be much older. When any metallic structure is designed, it must be verified that it will not fail in any of the possible modes, including due to flow over the intended lifetime. See in: https://en.wikipedia.org/wiki/Creep_(deformation) https://en.wikipedia.org/wiki/Creep_(deformation) at "Temperature dependence". By the approximate rule mentioned there, zinc begins to have non-negligible creep already above minus thirty Celsius degrees, so at normal ambient temperatures you must always compute the creep of zinc for any structural design. On the other hand, metals like iron or copper have negligible creep at room temperature, even when pure. Aluminum and magnesium begin to have non-negligible creep at temperatures only a little above normal ambient temperatures. Hard alloys can have much lower flowing speeds than the metals included in their composition. In integrated circuits, the metal connections are affected by electromigration, which is the flowing of the metal due to electrical current instead of mechanical stress. The electromigration properties and creep properties of a metal are closely related. In the beginning, the ICs used pure aluminum for interconnections, but when their size was reduced, the connections began to fail after a too short lifetime. The first solution for this problem was the replacement of pure aluminum with harder aluminum alloys, including small quantities of copper and/or silicon. When the ICs became even smaller, the aluminum alloys had to be replaced with a metal having a higher melting temperature, i.e. copper, which fortunately also has a lower resistivity.
  - roberthahn 3y ago
    Thank you for the information! Sometimes it’s worth pausing a moment when seeing a “It’s well known that…” to check the timeline because the construction of Arecibo might well have taken place (or planned) before it was known (let alone well known). Edit: a bit more searching suggests that this was studied in 1947 (Andrade’s Creep Law and the Flow of Zinc Crystals, by AH Cottrell)
  - RajT88 3y ago
    This feels like the facility was in operation for so long, that the people who knew about this potential problem all retired and with them went that knowledge.
- sheepshear 3y ago
  Andrade is the namesake of Andrade creep due to his work in the early 1900s, but the existence of creep had been known for a long time by then. I'd imagine smiths have been aware of creep throughout the history of metallurgy. https://royalsocietypublishing.org/doi/10.1098/rspa.1910.0050 https://royalsocietypublishing.org/doi/10.1098/rspa.1910.005...
nullc 3y ago
https://www.nsf.gov/news/special_reports/arecibo/Arecibo-Telescope-Collapse-Forensic-Investigation-508c.pdf https://www.nsf.gov/news/special_reports/arecibo/Arecibo-Tel... Go to pdf page 53-ish. They show the zinc doesn't continue to creep if the load is below a threshold-- technically the zinc doesn't creep if it's not subject to persistent shear forces, which is won't be if the cable load is below a threshold as the wires take up the load. This essentially results in a long term capacity which is different from the short term capacity. Unfortunately the ratio between the two depends on the wire splaying geometry which isn't well controlled, resulting in wide differences. The report concludes that if it had been built to a safety factor of 3 rather than 2, the issues wouldn't have been experienced. Alternatively, the failure could have been avoided by reacting to flow over some threshold (which was noticed well in advance, but not reacted to) Presumably we don't see similar failures in sockets in suspension bridges because they're built at a safety factor of 5+ and usually that's a SF over their rarely reached maximum load rather than a factor over their 24/7 load.