VND iteratively explores neighbourhoods Nk; k D 1; : : : ; K for the descent and, in our case, applies a best improvement
strategy to each. VND can be successful because a local optimum within one neighbourhood is not necessarily a local
optimum for a different one. Thus, changing the neighbourhood can result in better local optima, but the performance of the
VND can depend on the sequence in which the local search operators are employed. We now describe the operators in more
detail. For all operators, all neighbouring solutions that follow from employing the given operator on the current solution
are investigated, unless stated otherwise.