Simulation study of critically loaded arrays of pillars

A failure in an overloaded system is a complex engineering problem. Arrays of pillars belong to a group of modern nanodevices composed of a large number of identical parts that function as a unit. This is because the pillars are fixed on a flat support and interactions among them emerge due to the support's rigidity. When a growing load is applied to the pillars it induces a sequence of failures among the pillars, decreases the device performance and eventually triggers a catastrophic avalanche of failures. A key aspect of how does such a critical destruction evolve is the so-called load transfer rule from destroyed pillars to the intact ones, where a particular transfer rule reflects a distance on which the pillars interact effectively. A common approach to study arrays-of-nanopillars-crushing processes is to employ an appropriate load transfer rule within computer simulations. In our simulations we use different load transfer rules to analyze amount of allocated memory (m) and CPU time (t) consumed to handle these rules for arrays with an increasing number of pillars (N). Specifically, we discuss distributions of m and t, referred to the employed rules as well as the corresponding mean values: m̅ ~ N and t̅ ~ N
^{b+c ln N}
.